CARNEGIE
INSTITUTION
Annual Report of the Director
Geophysical Laboratory
2801 UPTON STREET, NORTHWEST, WASHINGTON, D.C. 20008-3898
1988-1989
For the year July 1, 1988-June 30, 1989
Issued December 1989
Papers from the Geophysical Laboratory
Carnegie Institution of Washington
NO. 2150
Digitized by the Internet Archive
in 2012 with funding from
LYRASIS Members and Sloan Foundation
http://www.archive.org/details/annualreportofd198889carn
Geophysical Laboratory
Washington, District of Columbia
Charles T. Prewitt
Director
Published by: Geophysical Laboratory
2801 Upton St., N.W.
Washington, D.C., 20008-3898
USA
ISSN 0576-792X
December 1989
When used in bibliographic citations, The Annual Report should be cited as follows:
Author, Title, Annual Report of the Director of the Geophysical Laboratory, Carnegie Instn. Wash-
ington, 1988-1989, Geophysical Laboratory, Washington, D.C., page range, 1989.
GEOPHYSICAL LABORATORY
Contents
Introduction 1
Igneous and Metamorphic Petrology - A. Field
Studies 3
Kaapvaal Spinel Peridotites: Evidence of Craton
Origin. Francis R. Boyd 3
Rare Earth Element Zoning in Pyrope-rich Gar-
nets From Mantle Xenoliths. Donald D.
Hickmott 6
The Earth's Convection Framework: Its Behavior
Since the Jurassic and Implications for the
Geomagnetic Field. T. Neil Irvine 1 1
Fracture-controlled Fluid Flow During Chlorite -
grade Metamorphism at Waterville, Maine.
Douglas Rumble, Nicholas H. S. Oliver, and
Thomas C. Hoering 20
The Reaction Progress Method: Quantitative Tests
of Petrologic Models on a Microscopic Scale.
Craig M. Schiffries 26
Liquid- Absent Aqueous Inclusions.
Craig M. Schiffries 30
Igneous and Metamorphic Petrology - B.
Experimental Studies 33
Oxygen Fugacity and Evaporation Phase Rela-
tions in the Solar Nebula. Bjorn O. My sen and
Ikuo Kushiro 33
Experimental Determination of Element Parti-
tioning and Calculated Phase relations in the
Mg-Fe-Si-O System at High Pressure and High
Temperature. Yingwei Fei, Ho-kwang Mao,
and Bjorn O. Mysen 37
Partitioning of High Field Strength Elements
Among Olivine, Pyroxenes, Garnet and Calc
Alkaline Picrobasalt: Experimental Results and
An Application. Peter Ulmer 42
Relationships Between Composition, Pressure and
Structure of Depolymerized, Peralkaline Alu-
minosilicate Melts. Bjorn O. Mysen 47
Igneous and Metamorphic Facies of Potassium-
rich Rocks: Coexisting assemblages in
Kalsilite-Forsterite-Larnite-Quartz at 950°C and
2 kbar With and Without H20.
Hatten S. Yoder, Jr 54
Techniques for Experimentally Loading and
Analyzing Gases and Their Application to
Synthetic Fluid Inclusions. John D. Frantz, Yi-
gang Zhang, Donald D. Hickmott, and
Thomas C. Hoering 59
Investigations of Fluid Properties in the C02-CH4-
H20 System using Synthetic Fluid Inclusions.
Yi-gang Zhang and John D. Frantz 65
A Laser-based Carbon Reduction Technique For
Oxygen Isotope Analysis of Silicates and
Oxides. Zachary D. Sharp and
James R. O'Neil 72
Crystallography - Mineral Physics 79
Isotope Effects in Dense Solid Hydrogen: Phase
Transition in Deuterium at 190 (±20) GPa.
Russell J. Hemley and Ho-kwang Mao 79
The Effect of Pressure, Temperature, and Compo-
sition on the Lattice Parameters and Density of
(Fe,Mg)Si03 - Perovskites to 30 GPa. Ho-kwang
Mao, Russell J. Hemley, Jinfu Shu, Liang-chen
Chen, Andrew P. Jephcoat, Yan Wu, and
William A. Bassett 82
Single Crystal X-ray Diffraction Study of A New
Hydrous Silicate, Phase E. Yasuhiro Kudoh,
Larry W. Finger, Robert M. Hazen,
Charles T. Prewitt, and Masami Kanzaki...S9
Spectroscopic Evidence for a new New High-
pressure Magnesium Silicate Phase.
James D. Kubicki and Russell J . Hemley... .91
Compression and Polymorphism of CaSi03 at
High Pressures and Temperatures.
Liang-chen Chen, Ho-kwang Mao, and
Russell J. Hemley 94
The Polarized Raman Spectra of Tourmaline.
Mingsheng Peng, Ho-kwang Mao, Liang-chen
Chen, and Edward C. T. Chan 99
New Optical Transitions in Type la Diamonds at
Very High Stresses. Russell J. Hemley and
Ho-kwang Mao 105
CARNEGIE INSTITUTION
Premonitory Twinning in the High-Pressure Phase
Transition of ZrOr Yasuhiro Kudoh,
Charles T. Prewitt, andHaruo Arashi 108
BlOGEOCHEMISTRY Ill
Nitrogen Isotope Tracers of Human Lactation in
Modern and Archeological Populations. Mari-
lyn L. Fogel, Noreen Tuross,
and Douglas W. Owsley Ill
Nitrogen Isotope Fractionation in the Uptake of
Ammonium by a Marine Bacterium.
Matthew P. Hoch, David L. Kirchman,
and Marilyn L. Fogel 117
Dissolved Nitrogen Isotopic Distribution in the
Black Sea. David J. Velinsky, Marilyn L. Fogel,
and Bradley M. Tebo 123
Mineralogical and Oxygen Isotope Analyses of
Manganese Oxides Precipitated by Spores of a
Marine Bacterium. Kevin W. Mandernack,
Marilyn L. Fogel, Bradley M. Tebo,
and Jeffrey Post 130
Separation and Purification of Phosphates for
Oxygen Isotope Analysis. Ellen K. Wright
and Thomas C. Hoering 137
Scientific Highlights of the Geophysical
Laboratory, 1905 - 1989
Hatten S. Yoder, Jr 143
Publications 199
Personnel 204
GEOPHYSICAL LABORATORY
Introduction
This is the second year in which the
Annual Report of the Director is published
by the Geophysical Laboratory using desk-
top publishing techniques. I would like to
thank the many people who wrote in re-
sponse to last year's Report telling us that
you were glad to see it revived after an
absence of several years. We continue to
feel that the Report is a very important part
of our communication with the scientific
community and take pleasure in preparing
the scientific articles and sending the Re-
port to those on our mailing list. I again
emphasize that the articles in these issues
are only progress reports and that you should
expect to see the final results published in
the refereed scientific literature.
I am pleased to report that the Geo-
physical Laboratory had a very successful
year during 1988-1989, with a wide range
of research activity by staff members,
postdoctoral fellows, and visitors. Next
year, 1990, will be one of great change for
the Geophysical Laboratory staff and also
for our colleagues at the Department of
Terrestrial Magnetism. A new laboratory
building is now under construction at the
DTM site and we plan to move to this new
building during the summer or early fall of
1990. A resolution was passed by the
Carnegie Institution of Washington Trus-
tees in May 1986 to "Authorize the Presi-
dent to commission the development of an
architectural schematic plan for new con-
struction and renovation of existing struc-
tures on the Broad Branch Road site in
Washington, DC, appropriate to co-loca-
tion of the Geophysical Laboratory and the
Department of Terrestrial Magnetism."
Since that time, we have been involved in
devising the general requirements for the
building, working with the architectural
firm of Peirce, Pierce & Kramer of Cambr-
idge, Massachusetts, to produce a detailed
design, and in planning for the actual trans-
fer of equipment and personnel to the new
facility. The New Research Building will
have about 72,000 gross square feet of
floor space, and the present main DTM
building will be renovated to contain
administrative, library, and auditorium
facilities. The present Cyclotron Building
will be expanded and refurbished, and will
include clean laboratory facilities for use
by the geochemists of the two departments.
Many of us are sad at having to leave the
Upton Street location that has served us so
well for 85 years, but the prospect of having
an increased amount of modem laboratory
space and the great opportunity of expand-
ing our horizons through collaboration with
the DTM staff are very appealing.
As this will be the last Annual Report
published before the move to DTM takes
place, I asked Director Emeritus Hatten S.
Yoder, Jr. to write a section for this Report
on the history of the Geophysical Labora-
tory and he agreed to do so. As a relative
newcomer to the Lab, I am always im-
pressed when I read or hear about its his-
tory and the people who have worked here
during the 85 years of its existence. Rather
small as research institutions go, the Lab
usually has had about fifteen or sixteen
CARNEGIE INSTITUTION
research staff members at any one time
along with ten to fifteen Post- and Predoc-
toral Fellows and visitors.
The Laboratory building itself was
constructed for only $75,940 in 1905 and
research budgets through the years have
been relatively modest. However, the Lab's
impact on geoscience has been enormous;
I believe this is the result of the Carnegie
Institution's policy of letting its scientists
pursue their own goals with a minimum of
interference. In a time when national re-
search budgets are tight and industrial labo-
ratory managements are requiring more
applied research, I think the Carnegie ideal
is a very important concept to maintain as
an example of what can be done with lim-
ited financial support coupled with flexi-
bility and freedom in choosing research
topics.
A development of great interest to the
Geophysical Laboratory is the recent es-
tablishment of the Bayersiches
Forschungsinstitut fur Experimentelle
Geochemie und Geophysik at the Univer-
sitat Bayreuth. The founding of this insti-
tute was influenced greatly by the Geo-
physical Laboratory experience and its
Director, Friedrich Seifert, was a Postdoc-
toral Fellow and Staff Member here in the
1970s and early 1980s. This initiative has
created worldwide interest among geos-
cientists and is considered to be one of the
more exciting developments in the basic
earth sciences in recent years. It is interest-
ing to note that when the Geophysical
Laboratory was founded early in this cen-
tury, it was based to a great extent on the
experience of German laboratories and
institutes. Now it appears that the circle has
been completed.
It is with regret that I report that re-
search staff member Peter Bell decided to
take early retirement at the end of June
1989. During his 25 years at the Lab, Peter
made many contributions in geophysics,
mineral physics, and petrology. The col-
laboration between Bell and Ho-kwang
Mao was especially productive and they
became known as the world's leading pro-
ponents and users of the diamond-anvil cell
for ultra-high-pressure research. Peter was
a valuable member of our research staff and
we will miss his involvement in day-to-day
research activities. However, he will retain
an official connection with the Lab and will
continue to advise and consult with us in
his areas of expertise and interest.
GEOPHYSICAL LABORATORY
Igneous and Metamorphic Petrology -
A. Field Studies
Kaapvaal Spinel Peridotites: Evidence
of Craton Origin
Francis R. Boyd
Continental cratons have mantle roots
that extend to depths of at least 200 km,
giving cratonic lithosphere a thickness that
is two or more times that of oceanic plates.
Cratonic lithosphere differs from oceanic
in composition as well as in thickness. The
garnet peridotites that are the principal
components of the Kaapvaal lithosphere,
southern Africa, have markedly lower Mg/
Si, lower Ca/Al and higher Mg/Fe than do
residual oceanic peridotites and their com-
positional relations are believed to be rep-
resentative of other cratons (Boyd, 1989).
The origin of these differences in structure
and composition between cratons and oce-
anic plates is an important chapter in Earth
history.
There are few constraints on the origin
of craton roots other than their Archaean
age (Richardson et al., 1984). At least two
scenarios seem possible. Cratonic nuclei
might have developed at oceanic spreading
centers and subsequently been underplated
by peridotites having compositions similar
to Kaapvaal garnet peridotites. If that ori-
gin occurred, peridotites now forming the
shallow portions of craton roots should
have composition typical of oceanic peri-
dotites. Alternatively, the cratonic litho-
sphere in its entirety may be uniquely dif-
ferent in composition and origin from
oceanic lithosphere. In that event, perido-
tites forming the top of a craton root would
have compositions similar to the underly-
ing garnet peridotites.
Evidence required to distinguish be-
tween these possible processes of craton
development can be obtained by study of
the compositional relations of the spinel-
facies peridotites that occupy the upper
portion of the craton lithosphere. Spinel
peridotites are of widespread occurrence in
xenolith suites of the Kaapvaal craton,
having been collected at over a dozen lo-
calities in South Africa and Lesotho. Their
proportions in individual xenolith suites
vary widely from small to amounts ap-
proaching those of garnet peridotites. The
relative abundance of spinel peridotites led
Cars well et al. (1984) to propose that they
form a continuous layer at the top of the
craton lithosphere, overlying garnet peri-
dotites that are the principal rock type at
greater depth.
The depth at which the assemblage Mg-
rich garnet + olivine replaces aluminous
pyroxene + spinel is the boundary dividing
the spinel- and garnet-facies peridotites.
This depth is difficult to estimate by ther-
mobarometric methods because many of
the peridotites that occur near the top of the
lithosphere have failed to equilibrate at
relatively low ambient temperatures and
CARNEGIE INSTITUTION
Table 1. Mineral and Bulk Analyses for Spinel Peridotite PHN 5254, Premier mine, R. S. A., wt%.
Bulk*
Olivine
Enstatite
Diopside
Spinel
Si02
43.14
40.6
56.7
54.8
0.06
Ti02
0.05
<0.03
0.05
0.12
0.09
MA
1.36
<0.03
2.57
3.34
41.6
Cr203
0.43
<0.03
0.41
1.21
26.8
Fe203
2.78
-
-
-
-
FeO
4.21
7.78
5.11
1.40
12.0
MnO
0.12
0.11
0.14
0.07
0.16
MgO
42.47
50.9
35.6
16.4
18.4
CaO
1.53
<0.03
0.65
22.4
<0.03
Na,0
0.08
n.d.
0.06
1.34
n.d.
K,0
0.10
n.d.
n.d.
n.d.
n.d.
PA
0.00
n.d.
n.d.
n.d.
n.d.
NiO
0.27
0.40
0.09
0.06
0.18
LOI
4.53
-
-
-
-
Totals
101.07
99.8
101.4
101.1
99.3
Mg/(Mg + Fe)**
0.919
0.921
0.925
0.954
0.731
Ca/(Ca + Mg)**
-
-
0.013
0.496
-
MODES
Olivine
66.6
Enstatite
25.2
*S. A. Mertzman,
analyst
Diopside
6.3
**mole fractions
Spinel
1.3
their pyroxenes are chemically inhomo-
geneous. Nevertheless, the transition depth
can be estimated to be near 50-55 km on the
basis of experimental data (Wood and
Holloway, 1984). If the base of the crust in
the Kaapvaal craton is 40 km, the thickness
of a spinel peridotite layer may then be 10-
15 km.
A chemical feature that effectively
characterizes most Kaapvaal spinel peri-
dotites (Table 1 ) is that the alumina content
of the orthopyroxene is in the range 1 .5 - 4
wt %, contrasting with values of 0.7 - 1 .0 wt
% for the underlying garnet peridotites.
Amphibole (pargasite)-bearing spinel peri-
dotites form an exception to this generali-
zation, however, because their alumina
contents of orthopyroxene are low, 0.5 - 0.6
wt % (e.g. Boyd, 1971). Spinels are more
aluminous than the chromites in garnet
facies rocks and there is a positive correla-
tion for the concentration of A1203 in coex-
isting spinel and orthopyroxene (Carswell
et al., 1984). Diopsides in the spinel peri-
GEOPHYSICAL LABORATORY
dotites contain somewhat less Cr203 (aver-
age 1.0 wt.%) than is commonly found in
diopsides in garnet lherzolites.
Distinctive textural characteristics of
the spinel peridotites include an unusually
coarse grain size, commonly 1-2 cm. The
enstatite in almost all specimens have thin
exsolution lamellae, presumed to be clino-
pyroxene. The igneous age of the spinel
peridotites, like that of the garnet perido-
tites, may be Archaean, but the age of the
exsolution is not known. The spinel charac-
teristically forms symplectites (Dawson and
SPINEL PERIDOTITES
KAAPVAAL CRATON
o>
c
*>
*5
94r
93-
92-
91-
90-
89-
o
J L
J L
100 90 80 70 60 50
Modal olivine. wt%
40
Fig. 1. Compositional data for Kaapvaal spinel
peridotites (solid circles) and transitional rocks
(half solid circles) compared with data for low-
temperature, Kaapvaal garnet peridotites (open
circles), oceanic residues (open squares), and
pyrolite (open diamond). Most of the spinel peri-
dotites are from the Premier mine but samples
from Kimberley, Frank Smith and Letseng are
included. Sources of data for the oceanic residues
and garnet peridotites are listed in Boyd (1989)
and the pyrolite composition is from Ringwood
(1979). Modes for the Kaapvaal rocks were cal-
culated from the mineral and bulk analyses. Ana-
lytical samples for the bulk analyses were ap-
proximately 0.5 kg and were analyzed by S. A.
Mertzman.
Smith, 1975) with olivine and pyroxene as
well as amphibole and mica; this habit
contrasts markedly with the bleb-like tex-
ture of primary spinels in peridotite xeno-
liths in basaltic volcanics. Two specimens
from the Premier mine have the textural
characteristics of spinel peridotites but
contain small amounts of fme-grained in-
terstitial garnet; these are taken to be tran-
sitional between the spinel and garnet fa-
des.
Insights to the origin of the spinel peri-
dotites can be obtained by comparing their
bulk compositions with those of cratonic
garnet peridotites and oceanic peridotites.
A plot of the modal abundance of olivine
against mg number [mg number = Mg/(Mg
+ Fe)] for either olivine or the whole rock
does an excellent job of discriminating
oceanic residues and craton-forming, low-
temperature garnet peridotites (Fig. 1). The
oceanic harzburgites represented by abys-
sal peridotites and ophiolite tectonites have
70-80% olivine and mg numbers predomi-
nantly in the range 0.905-0.915. The cra-
tonic peridotites from the Kaapvaal are
enstatite-rich with 20-45 % modal enstatite,
40-80% modal olivine and mg numbers of
0.920 - 0.935.
Points for the Kaapvaal spinel perido-
tites superimposed on plots for low-tem-
perature garnet peridotites and oceanic
peridotites clearly overlap the garnet-fa-
cies rocks (Fig. 1). These data suggest a
common origin for the Kaapvaal perido-
tites, whether of spinel or garnet facies. The
compositional differences between these
cratonic peridotites and oceanic residues
make it appear unlikely that any large part
of the Kaapvaal craton originated as an
CARNEGIE INSTITUTION
oceanic plate. Cratonic peridotites may have
formed as bouyant residues of segregated
ultramaf ic liquids at depths of 300-400 km
(Boyd, 1989). The circumstances in which
these residues floated in a denser, more
fertile and largely crystalline mantle and
coalesced to form proto-cratons are diffi-
cult to clarify. Underplating may have
played a role in this process, however, and
the spinel peridotites at the top of the cra-
tonic lithosphere might be the oldest rocks
in these ancient tectonic blocks.
References
Boyd, F. R., Compositional distinction between
oceanic and cratonic lithosphere, Earth Planet
Sci. Lett., in press, 1989.
Boyd, F. R., Pargasite - spinel peridotite xenolith
from the Wesselton Mine, Carnegie Instn.
Washington Year Book, 70, 138-142, 1971.
Carswell, D. A., W. L. Griffin, and P. Kresten,
Peridotite nodules from the Ngopetsoeu and
Lipelaneng kimberlites, Lesotho: a crustal or
mantle origin?, in Kimberlites - II: The Mantle
and Crust-Mantle Relationships, J. Kornprobst,
ed., Elsevier, New York, pp. 229-243, 1984.
Dawson, J. B. and J. V. Smith, Chromite - silicate
intergrowths in upper-mantle peridotites, Phys.
Chem. Earth, 9, 339-350, 1975.
Richardson, S. H., J. J. Gurney, A. J. Erlank, and
J. W. Harris, Origin of diamonds in old enriched
mantle, Nature, 310, 198-202, 1984.
Ringwood, A. E., Origin of the Earth and the
Moon, Springer- Verlag, New York, 1979.
Wood, B. J., and J. R. Holloway, A thermody-
namic model for subsolidus equilibria in the
system CaO - MgO - A1203 - Si02, Geochim.
Cosmochim.Acta,48, 159-176, 1984.
Rare Earth Element Zoning in Pyrope-
rich Garnets From Mantle Xenoliths
Donald D. Hickmott
Studies of mantle xenoliths in alkali
basalts and kimberlites provide direct evi-
dence concerning the chemical constitution
and evolution of the sub-crustal lithosphere
and upper asthenosphere. The abundances
and isotopic compositions of the rare earth
elements have been particularly important
in determining the geochemical evolution
of these rocks.
Garnet peridotites from the Archeaen
Kaapvaal craton can be divided into two
texturally and chemically distinct classes,
the granular lherzolites and the sheared
lherzolites (Boyd and Nixon, 1973). Ther-
mobarometric determinations suggest that
the latter equilibrated at higher tempera-
tures (> 1100°C vs. <1100°C) and pres-
sures (> 55 kbar vs. < 55 kbar) than the
former (Finnerty and Boyd, 1987). The
major elements Na, Al, Ca and Ti are gen-
erally enriched in the higher temperature
peridotites (Nixon and Boyd, 1973; Boyd
and Mertzman, 1 987), as are the heavy rare
earth elements (HREE) (Shimizu, 1975).
Sr and Sc may be enriched in the low-P-r
suite lherzolites relative to those of the
high-P-7 suite, based on reconstitution of
bulk-rock abundances from element con-
centrations in garnet and clinopyroxene
(Shimizu and Allegre, 1978). Low-P-7
nodules contain more modal enstatite and
less olivine than those from the high-P-J
suite (Boyd and Mertzman, 1987; Boyd,
1989).
GEOPHYSICAL LABORATORY
100
<D
■C10
"D
C
o
O
LU
m 1
tr 1
0.1
t — i — i — i — i — i — i — i — i — i — i — i — i — r
FRB76
J I L
J I I I I I I I I L
i — i — i — i — i — i — i — i — i — i — i — r
FRB 450
Ce Nd SmEu Dy Er
J I I I I L
J L
Ce Nd SmEu Dy Er
Fig. 2. Rare earth element abundances in two pyrope-rich garnets from the high-P-T suite. See Griffin
etal. (1989) for a description of the petrography and chemistry of these samples. Open symbols - near-
rim points; closed symbols - interior points.
The mineral zoning observed in mantle
nodules provides constraints on the nature
and timing of processes that occurred within
the upper mantle prior to eruption of the
host magma of a nodule. Smith and co-
workers have documented zoning of Ti, Cr,
Fe, Na and P in garnets from selected
xenoliths from the Kaapvaal craton, South
Africa and The Thumb, Colorado Plateau
(Smith and Ehrenburg, 1984; Smith and
Boyd, 1987; Smith, 1988). Griffin et al.
(1989) determined that garnets from two
samples from Frank Smith kimberlite were
characterized by Fe, Ti, Y, Zr and Ga en-
richments in their rims relative to their
cores. They suggested: 1) that the rim en-
richments could be explained by infiltra-
tion of an alkalic melt into the mantle
rocks; 2) that the enrichments reflected re-
equilibration of overgrowths on pre-exist-
ing garnet rather than simple diffusional re-
equilibration of such garnets; and 3) that
the enrichments occurred within a few
hundred years prior to eruption of the host
kimberlite.
Rare earth element abundances were
measured in rims and cores of garnets from
the two high-P-T suite samples (FRB 76
and FRB 450) previously investigated by
Griffin et al. (1989) and one low-P-7 suite
garnet (PHN 2302A). The studied garnets
from the high-P-T samples are from the
same nodules, but are not the same crystals
as those analyzed by Griffin et al. (1989).
Analyses were made with a Cameca-IMS-
3F ion microprobe with previously ana-
lyzed pyrope-rich garnets as standards.
Analytical techniques are similar to those
described by Shimizu and Richardson
(1987). Estimated accuracy of the REE
8
CARNEGIE INSTITUTION
analyses is ±10%. Yb abundances are not
presented due to GdO molecular ion inter-
ferences on the Yb mass spectra.
All of the REE investigated in the high-
P-T samples are enriched in garnet rims
relative to garnet cores (Fig. 2). Cerium
rises by a small amount (less than 20%),
whereas Er rises up to a factor of two; thus
HREE/LREE ratios increase from core to
rim. In contrast, garnets from lherzolite
PHN 2302A from the low-P-7 suite are
HREE enriched and LREE depleted in their
rims relative to cores (Fig. 3). Cerium drops
up to 70% from roughly 1.7 ppm to 0.4
ppm, whereas Er rises up to a factor of three
(0.3 ppm to 1.1 ppm).
The trace element signature of the gar-
net cores may provide constraints on the
nature and evolution of the mantle in the
source region of the kimberlites prior to
final interaction with a metasomatizing
agent, whereas that of the rim may help
identify that metasomatic agent. The high
LREE abundances of the cores of the high-
P-T suite garnets relative to chondrites are
incompatible with the hypothesis that the
rocks are directly residual to magma gen-
eration from a chondritic mantle. These
samples may have experienced an earlier
episode of fluid or melt infiltration prior to
their most recent metasomatism. The low
abundance of HREE in the core of the low-
P-T suite garnet suggests that the granular
nodules interacted with a LREE-enriched
fluid or melt in the absence of garnet prior
to subsolidus growth of the garnet cores
(Shimizu, 1975). The relatively high mo-
dal abundance of garnet in many low-P-T
nodules may require an episode of meta-
morphic differentiation subsequent to gar-
net nucleation.
100F
CO
CD
-■£10
"O
c:
o
O
LU
cc 1
0.1
7— i — i — i — i — i — i — i — i — i — i — r
PHN 2302A
J I l I l l l l l l l I l L
CeNdSmEuDy Er
Fig. 3. Rare earth element abundances in a pyrope-
rich garnet from the low P-T suite. See Boyd
(1974), Shimizu (1975) and Shimizu and Allegre
( 1 978) for petrographic and chemical data pertain-
ing to this sample. Open symbols - near-rim points;
closed symbols, interior points.
Rare earth element patterns for hypo-
thetical magmatic liquids calculated to be
in equilibrium with the rims of the three
samples are strongly LREE enriched rela-
tive to chondrites (Fig. 4). Selection of
alternate sets of partition coefficients for
this calculation (i.e., Irving and Frey, 1978)
leads to varying absolute REE abundances
in calculated liquids, but yield roughly
similar REE patterns to those shown. The
calculated patterns for both the low-P-7
and high-P-T suite samples are roughly
similar in element abundance and slope to
each other, although the low-P-7 garnet
may have interacted with a fluid with lower
HREE abundance and higher MREE/HREE
and MREE/LREE ratios than the high-P-T
samples. Absolute abundances of the REE
in the hypothetical liquids are in approxi-
mate agreement with REE abundances of
GEOPHYSICAL LABORATORY
1000
CD
•c100
c
O
O
LU
LU 10
DC
t — i — i — i — ri — i — i — i — i — i — i — rr
Calculated
Liquids
J I I I I I I I I I I I 1_L
CeNdSmEu DyEr
Fig. 4. Magmatic liquids calculated to be in equi-
librium with select rim points of zoned, pyrope-
rich garnets FRB 76, FRB 450 and PHN 2302A.
Partition coefficients of Fujimaki et al. (1984)
were used in the calculation. Diamond = PHN
2302A; square = FRB 450; open square = FRB 76.
Shaded area shows upper- and lower-bounds of
kimberlite data reported by Kramers et al. ( 1 98 1 ).
Solid symbols within shaded area is Kramers et al.
(1981) "average kimberlite".
South African kimberlites (Fig. 4), except
for Ce and the HREE, which are depleted
relative to most kimberlites.
The simplest explanation for the major
and trace element enrichments observed is
that the nodules interacted with a melt with
kimberlitic affinities. Alkali basalts gener-
ally have higher HREE abundances than
kimberlites (Menzies et al, 1987); thus,
interaction with alkali basalt is unlikely.
Hydrous fluid probably does not exist at the
depth of origin of these nodules (Smith and
Boyd, 1987) and mantle hydrous fluids are
generally Ti-poor (Menzies et al, 1987); a
hydrous fluid is an unlikely metasomatic
agent for the high-P-r rocks. The discrep-
ancies between the calculated REE pat-
terns and those observed in kimberlites
may be due to 1 ) inaccurate knowledge of
partition coefficients between garnet and
melt at the depth in the mantle where these
nodules were derived (Kramers et al. , 1 98 1 ),
or 2) interaction between the proto-kim-
berlite and minerals in the mantle prior to
interaction with the studied garnets. The
large grain-size (> 5 mm diameter) and
small diffusion coefficients (Cygan and
Lasaga, 1986) of garnets enhance preser-
vation of chemical gradients in this mineral
relative to clinopyroxene, olivine and
orthopyroxene. It is possible that other
minerals in the studied nodules have also
interacted with the melt that caused the
garnet zoning, even though they are un-
zoned (Griffin et al, 1989).
Neodymium isotopic studies of lher-
zolite nodules in kimberlites have been
hampered by contamination problems
(Richardson et al, 1985; Zindler and Ja-
goutz, 1988). The most primitive mantle
isotopic compositions are frequently deter-
mined on the most painstakingly hand-
picked and acid washed mineral separates
(Richardson et al, 1985; Zindler and Ja-
goutz, 1988). The present study of REE
zoning in mantle garnets suggests that garnet
isotopic compositions may also be sensi-
tive to the relative percentages of rim-
material and core-material in a given min-
eral separate. The intermineral isotopic
disequilibrium observed in some studies
(Allegre et al, 1982) may reflect incom-
plete infiltrative re-equilibration of the
isotopic signatures of garnets relative to
co-existing minerals (Smith and Boyd,
1987).
The presence of REE zoning in these
garnets suggests that both the trace element
10
CARNEGIE INSTITUTION
and isotopic signatures of the mantle de-
rived from kimberlitic nodules must be
interpreted cautiously. Even in cases where
there is no evidence for modal metasomatic
effects, recent cryptic metasomatism can
be imprinted on older trace element and
isotopic signatures. Accurate characteriza-
tion of mantle chemistry by xenolith stud-
ies requires an understanding of the proc-
esses that produce element zoning in mantle
minerals.
References
Allegre, C. J., Shimizu, N., and D. Rousseau,
History of the continental lithosphere recorded
by ultramafic xenoliths, Nature, 296, 732-735,
1982.
Boyd, F. R., Ultramafic nodules from the Frank
Smith Kimberlite Pipe, South Africa, Carnegie
Instn.Washington Year Book, 73, 285-294, 1974.
Boyd, F. R., Compositional distinction between
oceanic and cratonic lithosphere, Earth Planet.
Sci. Lett., in press, 1989.
Boyd, F. R., and S. A. Mertzman, Composition
and structure of the Kaapvaal lithosphere, south-
ern Africa, in Magmatic Processes: Physico-
chemical Principles, B.O. Mysen, ed., Spec.
Pub. No. 1, The Geochemical Society, Univer-
sity Park, PA, pp. 13-24, 1987.
Boyd, F. R., and P. H. Nixon, Structure of the
upper mantle beneath Lesotho, Carnegie Instn.
Washington Year Book, 72, 431-446, 1973.
Cygan, R., and A. C. Lasaga, Self-diffusion of
magnesium in garnet at 750° to 900°C, Am. J.
Sci., 285, 328-350, 1985.
Finnerty, A. A., and F. R. Boyd, Thermobarom-
etry for garnet peridotites: basis for the determi-
nation of thermal and compositional structure of
the upper mantle, in Mantle Xenoliths, P. H.
Nixon, ed., John Wiley & Sons, New York, pp.
381-402, 1987.
Fujimaki, H., M. Tatsumoto and K. Aoki, Partition
coefficients of Hf, Zr, and REE between phe-
nocrysts and groundmasses, in Proceedings of
the 14th Lunar & Planetary Science Confer-
ence, Part 2, J. Geophys. Res., 89 Supplement,
pp. B662-B672, 1984.
Griffin, W. L.,D. Smith, F. R. Boyd, D.R. Cousens,
C. G. Ryan, S. S. Sie, and G. F. Suter, Trace
element zoning in garnets from sheared mantle
xenoliths, Geochim. Cosmochim. Acta, 53,561-
569, 1989.
Irving, A. J., and F. A. Frey, Distribution of trace
elements between garnet megacrysts and host
volcanic liquids of kimberlitic to rhyolitic
composition, Geochim. Cosmochim. Acta, 42,
771-787, 1978.
Kramers, J. D., C. B. Smith, N. P. Lock, R. S.
Harmon, and F. R. Boyd, Can kimberlites be
generated from an ordinary mantle?, Nature,
291, 53-65, 1981.
Menzies, M., N. Rogers, A. Tindle, and C. Hawkes-
worth, Metasomatic and enrichment processes
in lithospheric peridotites, an effect of astheno-
sphere - lithosphere interaction, in Mantle
Metasomatism, M.A. Menzies & CJ. Hawkes-
worth, eds., Academic Press, New York, pp.
313-365, 1987.
Richardson, S. H., A. J. Erlank, and S. R. Hart,
Kimberlite-borne garnet peridotite xenoliths
from old enriched subcontinental lithosphere,
Earth Planet. Sci. Lett., 75, 116-128, 1985.
Shimizu, N., Rare earth elements in garnets and
clinopyroxenes from garnet lherzolite nodules
in kimberlites, Earth Planet. Sci. Lett., 25, 26-
32, 1975
Shimizu, N., and C. J. Allegre, Geochemistry of
transition elements in garnet lherzolite nodules
in kimberlites, Contrib. Mineral. Petrol., 67,41-
50, 1978.
Shimizu, N., and S. H. Richardson, Trace element
abundance patterns of garnet inclusions in peri-
dotite - suite diamonds, Geochim. Cosmochim.
Acta, 51, 755-758, 1987.
Smith, D., Implications of zoned garnets for the
evolution of sheared lherzolites: Examples from
Northern Lesotho and the Colorado Plateau, /.
Geophys. Res., 93, 4895-4905, 1988.
Smith, D. and F. R. Boyd, Compositional hetero-
geneities in a high-temperature lherzolite nod-
ule and implications for mantle processes, in
Mantle Xenoliths, P.H. Nixon, ed., John Wiley
& Sons, New York, pp. 551-561, 1987.
Smith, D., and S. N. Ehrenberg, Zoned minerals in
GEOPHYSICAL LABORATORY
11
garnet peridotite nodules from the Colorado
Plateau: implications for mantle metasomatism
and kinetics, Contrib. Mineral. Petrol., 86, 274-
285, 1984.
Zindler, A., and Jagoutz, E., Mantle cryptology,
Geochim. Cosmochim.Acta, 52, 319-333, 1988.
The Earth's Convection Framework:
Its Behavior Since the Jurassic and
Implications for the Geomagnetic Field
T. Neil Irvine
In last year's Report, a "global convec-
tion framework" was defined for the Earth,
comprising six "convection centers" at the
intersections of three mutually perpendicu-
lar great circles. Four of the centers, distrib-
uted at 90° -intervals on one great circle,
appear to mark axes of upwelling in the
Earth's convection system. Three of them
are the major volcanic hotspots of Hawaii,
Iceland, and the Balleny Islands (the last
being affiliated with the McMurdo vol-
canic province of Antarctica); and the fourth,
located next to the Okavango Delta in
Botswana, is a seismically active locality
that effectively defines the southwest,
"geophysical end" of the East African rift
valley system. The other two centers are on
either side of this circle of four, one in Peru,
the second on the edge of Vietnam. Both
these regions feature major subduction
systems, so the two centers are presumed to
represent axes of down- welling. The char-
acteristics of the framework resemble
considerably the tomography of the lower
mantle, but the most intriguing correlations
are with the geomagnetic field (Fig. 5), a
matching that suggests that the framework
reflects convection in the liquid part of the
core as well as the mantle.
When the framework was first con-
ceived, it was thought that its position had
probably been constant for some extended
period of geological time because the three
hotspots traditionally have been included
in a global hotspot system that has been
relatively stationary for millions of years.
Further research has shown, however, that
the Earth's magnetic history correlates
closely with events of crustal rifting and
continental flood basalt magmatism (Fig.
6). In combination with observations that
the convection framework appears to be
linked to the geomagnetic field, that the
Iceland and Balleny centers adjoin mid-
ocean rift zones, and that the Okavango
center is situated amidst the enormous
Karoo basalt floods, this finding suggested
that the framework might have moved with
time in ways that can be defined by rifting
events and flood basalt eruptions. This
possibility has been explored, and results
extending back to the Middle Jurassic are
described below. Through them a convec-
tion structure is suggested for the core that
may be of interest to specialists in that field.
Movement of the Convection Framework
Since the Jurassic
The process of locating the convection
framework in the geological past involves
two assumptions deriving from its present-
day relationships. One is that the arrange-
ment of the six convection centers was
always orthogonal; the other is that the
Peru and Vietnam centers were always on
12
CARNEGIE INSTITUTION
,60E
MAGNETIC \-"
DECLINATION \ toe
IN DEGREES
Fig. 5. Map showing the global convection framework superimposed on the Earth's magnetic field
(contoured data for 1980 from Fabiano et al., 1983). It is seen that the Peru and Vietnam centers (of down-
welling) are almost coincident with the main intersections of the magnetic equator with the lines of zero
declination extending from the North and South Magnetic Poles (NMP, SMP). Also, these two centers
lie on the meridian (light dotted line) through the North and South Geomagnetic Poles (NGP, SGP; poles
of the best-fit dipole field), and the great circle containing the four centers of upwelling (Hawaii, Iceland,
Okavango, and Balleny) passes within a few degrees of all four magnetic poles. It might be noted that
the large, double-hairpin loop of the line of zero declination extending across Asia is a transitory feature
that has shifted widely in historic times because of secular magnetic field variations(see Courtillot and
LeMouel, 1988, Fig. 35). Despite this shifting, the intersection of the line with the magnetic equator near
Vietnam moved very little.
the magnetic equator. At the present time,
the magnetic equator is not coincident with
the true equator, but such coincidence is
generally assumed in paleomagnetic stud-
ies, at least as a time-averaged condition.
When it obtains, the combination of the
above assumptions additionally implies that
the four centers of upwelling are on the
same meridian.
Beyond these assumptions, the posi-
tioning of the framework requires identifi-
cation of the centers of upwelling based on
events of rifting and volcanism. In this
search, a valuable test or criterion derives
GEOPHYSICAL LABORATORY
13
DATE
MA
0
iOO-
120-
140-
160-
180-
200-
220-
240—
260-
280-
300-
AGE
Pleistocene
Late
Cretaceous
Early
Cretaceous
Middle
Jurassic
Early
Jurassic
Early
Permian
REVERSALS/ MA
1 2 3
VOLCANISM, RIFTING
6
Cretaceous
normal
superchron
]
-h
Manicouagan
Permian
reverse
superchron
N<ursk, other?
-Columbia River eruptions
J-Main Ethiopian eruptions
^Ethiopian start; Hawaii- Emperor bend
East Greenland floods; N. Atlantic opens
Brito-W. Greenland magmatism
Main Deccan eruptions; Eurasian Basin
India starts north
opens
•Benue rifting; Makarov Basin opens
2
Pacific Cretaceous hotspot track bend
S.Atlantic opens; Rajmahal eruptions
'-Karoo, Parana eruptions end;
Canada Basin begins to open
Parana eruptions begin-, minor Karoo
Minor Karoo magmatism
_^Minor Karoo; India leaves Africa
-L-Central Atlantic opens, Ferrar magmas
-^-Major Karoo eruptions; Rhine graben
J-Major Karoo eruptions; Atlantic dikes
Karoo magmatism begins (minor)
-t Eastern U.S.A. volcanism
Eastern U.S.A. rifting
Gulf of Mexico opens
J Scoresbyland rifting; Tethys opens
>-Main Tunguskan eruptions
-— -Tunguskan eruptions begin
CRYPTOEXPLOSION
OR IMPACT STRUCTURES
Fig. 6. Correlation of the polarity reversal-frequency histogram for the Earth's magnetic field (Permian
to present) with events of crustal rifting and flood basalt magmatism. The histogram is from Creer and
Pal (1986); the other information is from many sources, but most notably Erlank (1984), MacDougall
(1988), and Vogt and Tucholke (1986). It is seen that all the major changes in reversal frequency are
matched by majorrifting events or continental flood basalt magmatism. Thus: ( 1 ) the episode of frequent
reversals through the Triassic began with the eruption of the Tunguskan basalts in Siberia and continued
with North American rifting events extending from East Greenland to the Gulf of Mexico; (2) the
numerous reversals in the Jurassic and Early Cretaceous seem clearly related to the Karoo and Parana
magmatism and the concurrent opening of the South Atlantic; and (3) the steplike increases of reversal
frequency from Late Cretaceous through the Miocene successively match (a) the movements of India
and the Deccan eruptions, (b) the opening of the North Atlantic and Arctic ocean basins and the related
Brito- Arctic magmatism, (c) the rifting of central and eastern Africa and the Ethiopian basalt floods, and
(d) the Columbia River basalt eruptions. A few major shock metamorphic (impact or crypto-explosion)
structures are indicated for comparison (data from Grieve, 1 987) because some investigators believe that
geomagnetic reversals are caused by meteorite impacts.
from the orthogonal arrangement of these then the other three would be expected also
centers. If one of them is properly located, to be within regions of rifting or volcanism.
14
CARNEGIE INSTITUTION
110 MA
LATE EARLY CRETACEOUS
^sD> Convection center path ^Q Basaltic magma source sujpply
GEOPHYSICAL LABORATORY
15
ICELAND
Canada Basin continues
to open
West -dipping
subduction
\\ accompanies -?
\\ Peru Center
Makarov Basin opens
mn ./
w-v^Abu Gabra rift
80 MA
_£^-^|MS LATE CRETACEOUS
Bfito-Arctic
magmatis'm
60-54Ma^ h/( .
\ Approximate
Y* present-day
\ framework
galleys ;/ /y Basalt erupts
S/ ^' when its supply
/ /Center passes
i / / south of the
. />; w / / eruption site
^South Sandwich arc / -y then moves rap
Ms _^^^ 65 Ma northward.
CRETACEOUS- TERTIARY
BOUNDARY
PALEOCENE-EOCENE
BOUNDARY
Fig. 7. Paleo-reconstructions on which an attempt is made to position the convection framework with
time over the past 175 Ma during the breakup of Gondawana and the opening of the Atlantic. Assump-
tions are explained in the text. The reconstructions are slightly modified from Sclater et al. (1977),
principally by the additions of India and Madagascar in maps F and G, and by slight shifting of the
magnetic pole positions in maps G and H to bring them into better accord with more recent data from
Irving and Irving (1982). Note how this latter change affects the configuration of the magnetic equator
in the projection. Other control data are largely from the same sources as Fig. 6. Stars are hotspots. A
feature of particular interest in the map sequence is that the convection framework was apparently
aligned with the magnetic poles — and presumably, therefore, with the Earth ' s spin axis — during the time
of the Cretaceous superchron (from about 115 to 90 Ma ; cf. Fig. 6) when there were no magnetic
reversals. For further discussion, see text.
16
CARNEGIE INSTITUTION
In the present documentation (Fig. 7), the
use of this criterion is limited to a compari-
son of Iceland and Okavango, because the
maps cover only the half of the Earth on
which the continents have been mostly
concentrated since the Jurassic. A related
effect is that the paths of the centers for any
interval of time should be mutually com-
patible within the restrictions of the above
assumptions. This effect shows in Fig. 7 in
the similarities of the Iceland and Okavango
path segments on several maps.
By the Jurassic, the structural bonds
between North America and Africa had
already been broken, as evidenced by the
Late Triassic rift basins of the eastern U.S. A.
The concern in Fig. 7, therefore, is primar-
ily with events in the southern and extreme
northern parts of the Atlantic. In map A, it
is inferred that the Okavango center has
just left Antarctica, where the Ferrar doler-
ite and the large Dufek layered intrusion
formed about 178-172 Ma ago. The center
then moves westward around the tip of
Africa, where Middle Karoo basalts erupted
at 178 and 165 Ma, and past Patagonia,
where volcanism was extensive between
165 and 155 Ma. During the same period,
the Iceland center moves northward from
central Europe along the North Sea, where
Jurassic volcanism has been reported, and
then turns southwest past the Rockall
Trough, which was the site of the earliest
ocean-floor spreading in the North Atlan-
tic. From there, the center loops back to the
northwest past the tip of Greenland, where
Jurassic basaltic sills and dikes are found.
During the transition to the Cretaceous
in map B, the Okavango center continues
northward along the developing rift be-
tween South America and Africa as the
Parana volcanics formed in Brazil and the
Late Karoo and Etendeka volcanics erupted
in Africa. The Iceland center concurrently
moves up the rift between Canada and
Greenland, then hooks westward around
Ellesmere Island. From there, it continues
in map C past Axel Heiberg Island, where
basalt was erupted in the Cretaceous, on to
where the Canada basin was soon to open.
Meanwhile, the Okavango center has moved
north and west to the point where South
America and Africa last were joined. A
major implication of these paths, of course,
is that the centers of upwelling were carv-
ing out the future continental blocks.
During the time period of maps D and E,
the Iceland center is stationed at the North
Pole, where it presumably controlled the
opening of the Canada basin. The Okavango
center accordingly is positioned on the
equator, arbitrarily at a place where the
Atlantic could broaden around it. It is no-
table that the Canada basin has no magnetic
stripes (Sweeney, 1983), which implies
that it formed during the Cretaceous super-
chron (see Fig. 6). We thus have an indica-
tion that the core convection system was
aligned with the Earth's spin axis during
this extended period when the geomag-
netic field did not reverse. Such a relation-
ship would seem particularly significant to
dynamo models for the origin of the geo-
magnetic field.
The opening of the Canada basin was
completed by map F, and the Iceland center
then moves successively to the neighbor-
ing Makarov and Eurasian basins, which
opened in that order — and which are both
magnetically striped. To be in accord with
GEOPHYSICAL LABORATORY
17
these movements, the Okavango center has
to shift rapidly eastward across Africa at
latitudes just below the equator. The Benue
trough was developing in this region at that
time (Browne and Fairhead, 1983). It re-
ceived marine sediments from about 1 10 to
65 Ma, and at about 85 Ma (during the
Santonian), the existing deposits under-
went relatively pronounced deformation.
The Okavango center is presumed, there-
fore, to have passed along the trough at
about 85 Ma, ultimately inducing the for-
mation of all three rift zones identified on
mapF.
Map F also shows the Okavango center
turning southward at the East Africa rift
valleys. This change of trend is obviously
attractive, because it leads along the rift
system to the present-day site of Okavango.
It might, however, also be justified on other
grounds. As the Okavango center reached
the rift valleys, the Peru center, following
behind 90° to the west, would just have
reached South America. The proposition is
that its progress was stopped by the conti-
nent, hence the Okavango center (and the
rest of the convection framework) had to
change direction. By the illustrated inter-
pretation, the Peru center was manifest as a
west-dipping subduction zone that con-
sumed ocean floor to its east as it moved
across the Pacific. But the continent was
too buoyant to subduct, hence when it was
reached, the subduction system had to "flip"
to its present east-dipping configuration,
and the Peru center became locked to the
continental margin. But if the flipping only
occurred where the subduction encoun-
tered continental blocks, then as illustrated
in map G, two parts of the system that were
exceptional could have continued on to the
east to become the west-dipping Caribbean
and South Sandwich Islands Benioff zones.
In maps G and H, the Iceland and
Okavango centers both move southward
toward their present locations. Through
this period, the Okavango upwelling had to
yield, first the Deccan basalts, then the
Ethiopian floods, and the Iceland center
had to yield the Tertiary magmas of Britain
and Greenland. The possibilities relating to
these complex events cannot be discussed
here, but it is interesting to note that the
Deccan eruptions appear to have occurred
just when the southward-moving Okavango
center passed northward-moving India.
If the analysis presented here has valid-
ity, then it demonstrates that the convection
framework is potentially a very useful
device for relating tectonic and magmatic
events in different parts of the world at
various times in the geological past. But
there obviously are many points on which
the analysis might be challenged. Only one
question can be addressed here, but it has
been the most frequently asked: What about
all the other hotspots? Two observations
are noted. One is that, among the four
centers of upwelling, only Hawaii has a
hotspot track that is paralleled by other
hotspot tracks. As discussed in last year's
report, however, Hawaii apparently plays
an extraordinary role in Earth convection
in that it appears largely to control the
motion of the Pacific Plate — which is where
the parallel tracks occur. The other obser-
vation is that many (if not all) of the other
hotspots might have formed in consequence
of the movements of the main centers of
upwelling. Thus, for example, as the
18
CARNEGIE INSTITUTION
VIETNAM
PERU
NORMAL/REVERSE
0> SUPERCHRONS C5
Fig. 8. A suggested convection structure for the Earth's core, based on the convection framework and
a fluid dynamics model by Veronis (1959). See text for explanation and discussion.
GEOPHYSICAL LABORATORY
19
Okavango center opened the South Atlan-
tic (Fig. 7, A-D), it might also have initiated
the volcanic centers at Bouvet, Tristan da
Cunha, St. Helena, Ascension, and so on
(see Fig. 7).
Core Convection
The suggested core structure (Fig. 8)
combines the orthogonal axes of the con-
vection framework with a fluid dynamics
model by Veronis (1959) describing the
effect of the Coriolis force on a fluid layer
heated from below and cooled above. The
key feature is that the liquid of the outer
core rises and descends at the framework
axes by way of spool-shaped vortex struc-
tures in which it spirals inward with one
sense of rotation to about mid-level in the
layer, then switches and spirals outward
with the opposite rotation as it completes
its transit. When these structures are viewed
individually from above, their directions of
rotation appear opposite for upwelling and
down-welling in the same (northern or
southern) hemisphere, and they also appear
opposed for either upwelling or down-
welling structures in opposite hemispheres.
But when the whole core structure is viewed
from a single point, as in each diagram in
Fig. 8, then the upper (or lower) parts of
vortices on opposite sides of the solid inner
core are seen actually to have the same
rotation. A suggestion that arises here is
that, if there were oscillations of the depth
at which the flow spiraling reverses, then at
any particular time (as illustrated sche-
matically in a relatively extreme way in
Fig. 8C), most of the liquid along axial
lines through the whole core could be rotat-
ing either one way or the other. The propo-
sition is that such oscillatory flow reversals
might underlie the polarity reversals of the
geomagnetic field.
A further feature of the proposed core
structure, portrayed in Fig. 8C (c and d), is
that the spiral vortex flow should not occur
on convection axes lying in the plane of the
equator where the Coriolis force vanishes.
Thus, in the analysis of Fig. 7, there should
not have been spiraling on the Peru-Viet-
nam axis (at least until mid-Cenozoic times),
nor should it have occurred on the
Okavango-Hawaii axis during the Creta-
ceous superchron. A possible implication
of the latter condition is that it had the effect
of damping the oscillatory flow reversals
suggested above, thereby detering polarity
reversals and causing the superchron.
How these various processes might be
tied to rifting and volcanism in terms of
cause/effect relationships is debatable, but
the physical connection is presumably
plume activity in the mantle. It would seem
also that, if the analysis in Fig. 7 is on the
right track, then the inclination of the con-
vection framework relative to the Earth's
spin axis is especially critical, perhaps
because it influences the planet's rotation
characteristics. But this inclination in turn
might be controlled by factors such as the
interaction or coupling of the convection
framework with the continents (as in the
case postulated above for the Peru center),
by the relative freedom of the lithospheric
plates to move without interfering with one
another, and by the ease with which the
continental blocks can be rifted. A general
implication is that the crystallization, cool-
20
CARNEGIE INSTITUTION
ing, and convection of the core liquid is
probably strongly linked to the major tec-
tonic and magmatic events observed at the
surface.
References
Browne, S. E, and J. D. Fairhead, Gravity study of
the central African rift system: A model of
continental disruption, Tectonophysics, 94, 1 87-
203, 1983.
Creer, K. M, and P. C. Pal, Geomagnetic reversal
spurts and episodes of extraterrestrial
catastrophism, Nature, 320, 148-150, 1986.
Courtillot, V, and J. L. Le Mouel, Time variations
of the Earth's magnetic field: from daily to
secular, Ann. Rev. Earth Planet. Sci., 16, 389-
476, 1988.
Erlank, A. J, Pedogenesis of the Volcanic Rocks
of the Karoo Province, Geol. Soc. S. Africa,
Spec. Paper 13, 1984.
Fabiano, E. B, N. W. Peddie, and A. K. Zunde, The
magnetic field of the Earth, U. S. Geol. Surv.,
Misc. Invest. Series, Map 1-1457, 1983.
Grieve, R. A. F., Terrestrial impact structures,
Episodes, 10, 86, 1987.
Irving, E., and G. A. Irving, Apparent polar wan-
der paths, Carboniferous through Cenozoic , and
the assembly of Gondwana, Geophys. Surveys,
5, 141-188, 1982.
MacDougall, J. D., Continental Flood Basalts,
Kluwer, Dordrecht, Netherlands, 1988.
Sweeney, J. F, Arctic seafloor structure and tec-
tonic evolution, Am. Geophys Un., Geodynam-
ics Series, 2, 55-64, 1981.
Veronis, G, Cellular convection with finite ampli-
tude in a rotating fluid, J. Fluid Mechanics, 5,
311-324,1959.
Vogt, P. R, and B. E. Tucholke, The western North
Atlantic region, The Geology of North America,
v. M, Geological Society of America, Boulder,
Colorado, 696 p., 1986.
Fracture-controlled Fluid Flow Dur-
ing Chlorite-grade Metamorphism at
Waterville, Maine
Douglas Rumble, Nicholas H. S. Oliver,*
and Thomas C. Hoering
A current controversy in studies of
metamorphic rocks concerns the nature
and extent of fluid flow during metamor-
phism. Some researchers have found fluid-
rock ratios as high as 17.1 in bedded car-
bonate metasediments (Hoisch, 1987).
Other workers point out, however, that if
large ratios are characteristic of an entire
metamorphic complex rather than merely
pertaining to specific acquifers, difficult
questions arise about a feasible source for
such vast amounts of fluid (Wood and
Walther, 1986). Significant caveats have
been issued regarding uncertainties in the
magnitude of fluid-rock ratios measured
with the reaction progress method (Wood
and Graham, 1986). Resolving the contro-
versy is important because of the ramifying
effects the putative fluids would have on
metamorphic belts. Among the effects that
have been claimed are (1) regional alkali
metasomatism (Ferry, 1982); (2) regional
stable isotope metasomatism (Wickham and
Taylor, 1985); (3) advective heat transfer
(Chamberlain and Rumble, 1988); and (4)
removal of fluid reaction products allow-
ing devolatilization to proceed to comple-
tion (Ferry, 1986a).
We are addressing the controversy by
testing the hypothesis of fluid flow in the
Waterville-Augusta area, Maine. The re-
* Monash University, Dept. of Earth Sciences,
Clayton, 3168, Victoria, Australia
GEOPHYSICAL LABORATORY
21
gion was chosen for study because it is here
that the reaction progress method of esti-
mating fluid-rock ratios was developed and
widely applied (Ferry, 1980). The rocks of
the area are the focus of a debate about the
magnitude of fluid-rock ratios (Wood and
Graham, 1986; Ferry, 1986). A practical
advantage of the locality is that stratigra-
phic units strike perpendicularly across
metamorphic isograds from chlorite to sil-
limanite zones (Osberg, 1968). Thus, it is
possible to measure metasomatic changes
caused by metamorphism with minimal
ambiguity.
The reaction progress method of esti-
mating fluid flow developed by Ferry ( 1 980)
is best suited to study of metasediments and
meta-igneous rocks containing mineral
assemblages with low phase rule variance.
The method cannot be readily adapted to
veins containing assemblages of one or two
minerals (e.g. quartz-calcite), however.
Ferry's (1980) mapping of fluid flow gives
a detailed view of intergranular infiltration
of H20. But little is known of the role of
fracture permeability during metamorphism
as recorded by veins. This is a significant
lack of knowledge in view of the likelihood
that fluid flow through fractures may greatly
exceed intergranular infiltration. We de-
cided to investigate fracture permeability
by measuring the stable isotope composi-
tion of veins and their wall rocks from the
Waterville limestone, a member of the Si-
lurian Waterville Formation.
Methodology
Large specimens of 5 to 10 kg are col-
lected showing structural relations of veins
and their wall rocks. Hand specimens are
slabbed with a rock saw, polished, and
stained with Alizarin Red-S and Potassium
Ferricyanide. Staining the polished slabs
helps to locate and identify carbonate
minerals, chiefly calcite and dolomite.
Samples weighing 5 to 60 mg (larger
amounts for rocks with lower abundances
of carbonates) are drilled out with either a
1 or 2 mm diameter, diamond tipped drill.
Sample powders are loaded immediately
into two-legged reaction vessels for reac-
tion with 100% phosphoric acid at 25°C
(McCrea, 1950).
In many chlorite, biotite, and some
garnet-grade wall rock samples, calcite and
dolomite-ankerite solid solutions are so
intimately intergrown that they cannot be
physically separated. The simultaneous
reaction of both carbonates with phospho-
ric acid leads to cross-contamination of
evolved C02 The problem of cross-con-
tamination is usually dealt with by taking a
first aliquot of C02 released early during
reaction as representative of faster reacting
calcite; a later aliquot samples slower re-
acting dolomite (Epstein et al., 1964).
Cross-contamination is not a major diffi-
culty in the Waterville samples because the
wall rocks are usually either predominantly
calcite or dolomite. Samples with both
carbonates equally abundant are uncom-
mon. We have found that an aliquot taken at
5 minutes into the reaction gives a reliable
value for the #80 of calcite from calcite
rich rocks. Aliquots taken at 24 and 48
hours give results for dolomite reproduc-
ible to ±0. 1 %o for dolomite-rich rocks. We
do not yet have accurate data on the iso-
22
CARNEGIE INSTITUTION
" ■ r- 1 "■ i" -~i" ? "r -" i' r r
topic compositions of co-existing ealcite
and dolomite from the same powdered
sample. The data presented below on both
carbonates from the same hand specimen
refers to ealcite and dolomite from separate
lithologic layers.
Results
We have found evidence of fluid-rock
interaction in the outcrops of the Waterville
limestone along the east bank of the Ken-
nebec River in Waterville, Maine. Analysis
of ealcite and dolomite for #80 and 5*3C
shows that there are cryptic alteration halos
around certain veins in which wall rock
values have been depleted by 1-2 %o in
S^O and by similar amounts in #3C. The
halos are termed "cryptic" because no min-
eralogical features have been recognized
apart from proximity to veins. In what
follows we demonstrate the pre-metamor-
phic isotopic composition of the limestone
and establish its range of isotopic heteroge-
neity. An account is given of the relative
chronology of vein emplacement and the
evolution of isotopic values in the veins.
These results are used to evaluate the sig-
nificance of the isotopic alteration halos.
The pre-metamorphic isotopic compo-
sition of the Waterville limestone was
measured by analyzing vein and wall rock
samples taken from localities remote from
Fig. 9. Sample locations and geologic sketch map
at Waterville, Maine. Waterville limestone shown
in blank with line sketches of refolded isoclinal
folds. Graded bedded phyllite is stippled, SE cor-
ner of map. Metamorphosed granitic dikes are
heavily stippled at localities EE and PP. "WW"
denotes Waterville- Winslow bridge.
GEOPHYSICAL LABORATORY
23
-1
o
CO
I/O
-3
-4
B
a
□
WALLCC VEINCC WALL DO L VEINDOL
■ D * A
17 18 19 20
5180
Fig. 10. Plot of ff^C vs. #80. Box A shows compositions of pre-metamorphic limestone at localities H,
I, J, and K (Fig. 9). Box B outlines limestone samples taken within 0.1 m of contact with granitic dikes
at localities EE and PP. Box C gives analyses of limestone from 0.5 to 7 m from contact with graded
bedded phyllite at GG and II. Area D denotes limestones inside alteration halo at localities AA and BB
(Fig. 9).
contacts with the metamorphosed granitic
dikes and the graded bedded phyllite (Fig.
9, samples H, I, J, K). The #80 and #3C
values are consistently higher than from
any other samples analyzed but the &*0
values are depleted by some 4.0 %o in
relation to pristine marine limestones (Fig.
10A). There is abundant evidence of fluid
flow prior to metamorphism in the form of
stylolites and the replacement of current-
bedded, clastic calcite grains by fine-grained
dolomite. Thus, it is possible that the initial
depletion of the limestones in &*0 oc-
curred during diagenesis. Analysis of indi-
vidual beds 1-2 cm thick from single hand
specimens shows variation of no more than
0.3 %o in both #80 and #3C from bed-to-
bed. The effects of proximity to contacts
with metamorphosed granitic dikes are
depletion in ^80 by a maximum of 1.0 %o
but little change in &3C (Fig. 10B, samples
EE, PP). Samples collected within 1-5
meters of the contact with the graded bed-
ded phyllite are lower in ^80 by 1 .0 %obut
unchanged in #3C (Fig. 10C, samples GG,
II). It is concluded that pre-metamorphic
values of the limestone ranged from 19.2 to
20.0 %o in ^80 and from -0.9 to +0.3 %oin
#3C. Isotopic exchange between limestone
and dike rocks or phyllite led to depletion
in #80 by 1.0 %obut little change in #3C.
The relative chronology of vein em-
placement has been established by field
observations of the structural relationships
24
CARNEGIE INSTITUTION
between veins, re-folded isoclinal folds,
and granite dikes. The oldest structures are
bedding and stratigraphic contacts. These
are followed in age by stylolites and dolo-
mitized limestone beds. Beds, contacts,
and diagenetic features were subsequently
isoclinally folded and intruded by granitic
dikes. Both the isoclinal folds and dikes are
pre-metamorphic in age (Osberg, 1968).
Finally, the area was metamorphosed un-
der greenschist fades conditions and folded
along NNE axes.
The oldest (and rarest) vein (VI) dis-
covered is composed of sigmoid black fibres
of calcite extending perpendicular from the
vein's wall. The vein is isoclinally folded
and, therefore, is pre-metamorphic in age.
The &*0 and #3C values for the pre-meta-
morphic vein are 19.2%o, and +0.5, respec-
tively. The next youngest group of veins
(V2) consist of quartz and calcite. The V2
veins are irregular in shape and size (up to
20 cm in width) and contain isolated frag-
ments of wall rock. Their &*0 and #3C
values are typically in the range +19.2 to
19.8 %o and -0.8 to +0.2 %o, respectively,
and are closely similar to immediately
adjacent wall rock (Fig. 10A). A third
group of veins (V3) is made of fibrous
calcite with a fringe of muscovite, quartz,
and pyrite along the walls. The veins are 0. 1
to 1.0 cm thick. They strike NNE and are
parallel to the axial surfaces of the minor
folds that refold the older, isoclinal folds.
The V3 veins have values of 1 8. 1 to 1 8.5 %o
in 9*0 and -0.5 to +0.2 %o in #3C (Fig.
10C). Preliminary results of thin section
study suggest that the growth of metamor-
phic porphyroblasts began after the em-
placement of VI and extended throughout
the formation of V3. The youngest veins
(V4) contain quartz, calcite, and rare ga-
lena, and cross-cut the older veins. The
isotopic values of V4 calcites range from
+17.4 to 17.6 %o (#80) and from -3.2 to -
2.0%o(#3C)(Fig.lOD).
There is a systematic trend of depletion
in both 180 and 13C from oldest to youngest
veins. The values of the V4 veins are simi-
lar to some of the wall rocks and veins of
the Waterville limestone collected in the
staurolite and sillimanite zones. Veins
showing values characteristic of V4 have
been found in the graded-bedded phyllite,
as well.
Cryptic alteration halos have been rec-
ognized around V4 veins at the N end of the
outcrop (sample locations A A and BB, Fig.
9). The wall rocks consist of current-bed-
ded clastic limestones that have been
"dolomitized" parallel to bedding. Calcite-
rich beds alternate with dolomitized layers
on a scale of 1 to 2 cm. These rocks are
indistinguishable in appearance from dolo-
mitized limestones that preserve "pre-
metamorphic" isotope values. Both wall
rocks and older veins (V2 and V3) have
been depleted in &*0 and #3C in halos
surrounding V4 veins (Fig. 10D). Isotopic
exchange between V4 and its wall rocks
appears to be more complete in regard to
&*0 than it is in #3C. The vein calcites at
AA and BB (Fig. 9) with #3C values of -1.2
to -0.5 %o belong to the V2 and V3 genera-
tions (Fig. 10D). Wall rock calcite and
dolomite have #3C values between -2.0
and -1.2 %o. The calcite of V4, itself varies
in #3C from -3.2 to -2.0 %o (Fig. 10D).
Values of #80 in veins and wall rocks,
however, extend over a narrow interval
from 17.4 to 18.1 %o.
Calculation of single pass fluid/rock
GEOPHYSICAL LABORATORY
25
ratios needed to convert pre-metamorphic
wall rock #80 and 513C values to those
observed in the alteration halos gives 2.4
(molar ratio) for 180 and 2.9 (molar ratio)
for 13C. In these calculations the fluid
composition is assumed to be H20-C02
[X(C02) = 0.07 Ferry, 1987] with #80 (H20)
= 14.0 %o and &3C (C02) = -0.6 %o (e.g.,
fluid in equilibrium with most depleted V4
vein calcite at 390°C). The initial and final
values of #80 for wall rocks are 19.5 and
18.0, respectively, and for &3C these val-
ues are -0.5 and -2.0 %o. The fractionation
of 180/160 between calcite and H20 is +3.5
and that of 13C/12C between calcite and C02
is -2.6 at 390°C (Friedman and O'Neil,
1977). The calculations were made with
the equations of Rye and Bradbury (1988,
p. 214).
Our results demonstrate that chlorite-
grade limestones were penetrated by fluid-
filled fractures during the peak-to-waning
stages of metamorphism. Fluids infiltrated
into wall rocks to a depth of at least 4 cm
from vein walls. The fluid/rock ratios cal-
culated from isotopic data for wall rock
infiltration are approximately 3.0 by vol-
ume, some 8 times greater than the largest
ratios estimated by Ferry (1987, outcrop 7),
for intergfanular infiltration at the same
outcrop.
References
Chamberlain, C. P., and D. Rumble, Thermal
anomalies in a regional metamorphic terrane: an
isotopic study of the role of fluids, /. Petrol. , 29,
1215-1232, 1988.
Epstein, S., D. L. Graf, and E. T. Degens, Oxygen
isotope studies on the origin of dolomite, in
Isotopic and Cosmic Chemistry, H. Craig, S. L.
Miller, and G. Wasserburg, eds., North-Hol-
land, New York, pp. 169-180, 1963.
Ferry, J. M., A case study of the amount and
distribution of heat and fluid during
metamorphism, Contrib. Mineral. Petrol., 71,
373-385, 1980.
Ferry, J. M., Mineral reactions and element migra-
tion during metamorphism of calcareous sedi-
ments from the Vassalboro Formation, south-
central Maine, Am. Mineral. 68, 334-354, 1983.
Ferry, J. M., Infiltration of aqueous fluid and high
fluid-rock ratios during greenschist facies meta-
morphism: a reply, J. Petrol., 27, 695-714,
1986.
Ferry, J. M., Metamorphic hydrology at 13-km
depth and 400-550°C, Am Mineral. , 72, 39-58,
1987.
Friedman, I., and J. R. O'Neil, Compilation of
stable isotope fractionation factors of geochemi-
cal interest, U. S. Geol. Surv. Prof. Paper 440-K,
12 pp., 1977.
Hoisch, T. D., Heat transport by fluids during Late
Cretaceous regional metamorphism in the Big
Maria Mountains, southeastern California, Geol.
Soc. Am. Bull. 98, 549-553, 1987.
McCrea, J. M., On the isotopic chemistry of car-
bonates and a paleo-temperature scale, /. Chem.
Phys., 18, 849-857, 1950.
Osberg, O. H., Stratigraphy, structural geology
and metamorphism of the Waterville- Vassalboro
area, Maine, Maine Geol. Surv. Bull., 20, 1968.
Rye, D. M., and H. J. Bradbury, Fluid flow in the
crust: an example from a Pyrenean thrust ramp,
Am. J. Sci., 288, 197-235, 1988.
Wickham, S. M., and H. P. Taylor, Jr., Stable
isotope evidence for large scale seawater infil-
tration in a regional metamorphic terrane: the
Trois Seigneurs Massif, Pyrenees, France,
Contrib. Mineral. Petrol., 91, 122-137, 1985.
Wood, B. J., and C. M. Graham, Infiltration of
aqueous fluid and high fluid: rock ratios during
greenschist facies metamorphism,/. Petrol., 27,
751-761, 1986.
Wood, B. J., and J. V. Walther, Fluid flow during
metamorphism and its implications for fluid-
rock ratios, in Fluid Rock Interactions during
Metamorphism, J. V. Walther and B. J. Wood,
eds., Springer- Verlag, New York, pp. 89-108,
1986.
26
CARNEGIE INSTITUTION
The Reaction Progress Method: Quan-
titative Tests of Petrologic Models on
a Microscopic Scale
Craig M. Schiffries
A central problem in petrology is to
understand the physical and chemical con-
trols that govern the progress of mineral
reactions. It is generally assumed that meta-
morphic mineral reactions are driven by
variations in temperature and pressure, and
that metamorphism is nearly isochemical
except for the loss or gain of volatiles.
According to a growing school of thought,
however, fluid infiltration commonly plays
an essential role in driving metamorphic
mineral reactions (Newton et al.9 1980;
Rumble et al., 1982; Ferry, 1986). In con-
trast to the traditional view of metamor-
phism, fluid infiltration can cause large-
scale chemical mass transfer of non-vola-
tile components. The dichotomy of opinion
over the relative importance of the phys-
icochemical variables that govern the prog-
ress of metamorphic mineral reactions il-
lustrates the need to develop quantitative
tests that can distinguish between alterna-
tive petrologic models. The principal pur-
pose of this paper is to demonstrate that the
reaction progress method (Thompson et
a/., 1982; Thompson, 1982; Ferry, 1986)
can be used to distinguish between iso-
chemical and metasomatic models for the
origin of myrmekite (Fig. 11). The iso-
chemical model (Schwantke, 1909) is
analogous to the traditional view of meta-
morphism because the mineralogical
changes occur in response to variations in
temperature and pressure at constant bulk
Fig. 1 1. BSE image of calcic myrmekite from the
Bushveld Complex. The vermicular intergrowth
consists of quartz (black) and calcic plagioclase
(grey).
composition. The metasomatic replacement
model (Becke, 1908; Sederholm, 1916) is
analogous to fluid infiltration models of
metamorphism because the mineral reac-
tions occur in response to fluid-rock inter-
actions.
Myrmekite has been studied by penolo-
gists since the 19th century (Michel-Levy,
1874-75), and the extensive literature is
reviewed by Ash worth (1972) and Phillips
(1974). Myrmekite is commonly defined
as a descriptive, nongenetic term that refers
to vermicular intergrowths of quartz and
plagioclase. The intergrowths discussed in
this paper consist of quartz and plagio-
clase, but they are referred to as * calcic
myrmekite' because they are distinctly
different from the type of myrmekite that
has been widely discussed in the literature
(Table 2). It appears that calcic myrmekite
GEOPHYSICAL LABORATORY
Table 2. Comparison between calcic myrmekite and typical myrmekite
27
Calcic Myrmekite
Typical Myrmekite
Occurrence
Alkali Feldspar
Plagioclase
Composition
anorthositic and
gabbroic rocks
generally absent
An > An
myr host
granitic rocks and quartzo-
feldspathic gneisses
generally present
An < An.
myr
host
has been frequently overlooked and is a
common accessory feature in anorthositic
and gabbroic rocks (Schiffries and Dymek,
1985; Dymek and Schiffries, 1987). Calcic
myrmekite comprises minerals that are
intergrown on a length-scale of approxi-
mately 10 fim (Fig. 11), and precise meas-
urements of modal composition cannot be
obtained by standard point-counting tech-
niques. Micro-modal data were obtained
by inversion of broad-beam electron micro-
probe analyses and by image processing of
digitized BSE photomicrographs. Mineral
compositions were determined by conven-
tional electron microprobe techniques.
The metasomatic model for the origin
of calcic myrmekite involves an open sys-
tem reaction in which the host plagioclase
is replaced by an intergrowth of quartz and
relatively calcic plagioclase. The replace-
ment process can be represented by a single
net-transfer reaction:
2NaAlSi308+CaNa2
albite
fluid
=CaAl2Si208+ 4Si02.
(1)
anorthite
qtz
The endmember reaction (1 ) can be rewrit-
ten in terms of arbitrary values for the
initial and final plagioclase composition:
(l^ajCa^SiJ), + (f-/)CaNa 2 =
sodic plag fluid
(l+/)Na,./Ca/All4/Si3./08 + A(f-i)S\Ov (2)
calcic plag qtz
where i = [nr Knr +nv )]. . . ., / = [nr I
L Ca' x Ca Na'-1 initial7 •> L Ca'
("c+nJkai' and / > *• An advantage of
writing the net-transfer reaction in this form
is that the stoichiometric reaction coeffi-
cients are expressed in terms of the initial
(host) and final (myrmekite) plagioclase
composition. The production of quartz by
reaction (2) is coupled with a progressive
increase in the anorthite content of the
plagioclase.
The replacement model can be tested by
comparing measurements of reaction prog-
ress that monitor changes in mineral com-
position and modal proportions. Reaction
progress, ^, is defined for an arbitrary unit
of rock as:
I = An Iv
^ XX
(3)
28
CARNEGIE INSTITUTION
1.0
0.8
£ 0.6
CO
i
§? 0.4
a Bushveld Complex
0 St-Urbain anorthosite
a Other anorthosites
1.0
n,
Fig. 12. A
model.
qtz/(nqtz + nplag)
test of the metasomatic replacement
where Anx is the change in the number of
moles of species, x , per unit of rock that
results from the progress of a given chemi-
cal reaction, v is the stoichiometric reac-
tion coefficient of species, jc, and the sign of
v is positive for the reaction products and
negative for the reactants. Two expressions
for reaction progress of equation (2) are:
S=AnJ4(f-i)
S = AnJ(l+i)
(3a)
(3b)
modal abundance). Fig. 12 shows there is
excellent agreement between the empirical
data and the relationship predicted by equa-
tion (4), providing strong support for the
replacement model for the origin of calcic
myrmekite.
The isochemical model is based on the
assumption that myrmekite forms by exso-
lution of quartz from the "Schwantke mole-
cule" (Ca05 □ 05AlSi3Og), which is essen-
tially a feldspar that contains excess silica
in solid solution. The isochemical model
can be represented by the following end-
member reaction:
2Ca05q,5AlSi3O8= CaAl2Si208 + 4Si02,(5)
Schwantke anorthite qtz
where myrmekite consists of a vermicular
intergrowth of the product minerals. Reac-
tion (5) can be rewritten for an arbitrary ini-
tial composition:
Na1/Ca/ n/A1+/Si3+yOg+gf
Schwantke-albite solid solution
= Na,/Ca/Al„/Si3,0, + (4/)SiO,
plagioclase qtz
(6)
where n . refers to the number of moles of
pi»g
calcic plagioclase on the right hand side of
reaction (2). The initial assemblage does
not contain either quartz or calcic plagio-
clase and therefore An = n and An . =
qtz qtz plag
n . . Rearrangement of equations (3a) and
(3b) yields:
nj(na+nj = 4(f-i)/(4f-3i+l) (4)
In contrast to the replacement model, the
composition of plagioclase produced by
the isochemical process is independent of
the composition of plagioclase in the host
rock. Two expressions for the progress of
reaction (6) are:
qtz
9"
plag
I = «, •
~ plag.
(7a)
(7b)
Equation (4) provides a test of the replace-
ment model that is based on two independ-
ent sets of data (mineral composition and
Combination of equations (7a) and (7b)
yields:
GEOPHYSICAL LABORATORY
29
100
80
?60
**—
$ 40
20
0
* Q qbAB^***
n Bushveld Complex
♦ St-Urbain anorthosite
a Other anorthosites
20 40 60 80
nqtz/(nqtz + nplag)
100
Fig. 13. A test of the isochemical exsolution model
nJ(n,+nJ = Afl{Af+l).
qti « qtz plag
(8)
Equation (8) provides a test of the iso-
chemical exsolution model and Fig. 13
shows that the empirical data are inconsis-
tent with this model.
Alternative models for the origin of
calcic myrmekite have been tested through
the coordinated application of two micro-
analytical techniques that monitor changes
in mineral composition and modal abun-
dance on a sub-millimeter scale. The data
are in agreement with the metasomatic
replacement model but are not consistent
with the isochemical exsolution model.
Similar applications of the reaction prog-
ress method may provide a powerful tool
for testing petrologic models on a micro-
scopic scale.
References
Ash worth, J. R., Myrmekites of exsolution and
replacement origins, Geol. Mag., 109, 45-62,
1972.
Becke, F., Uber Myrmekit, Tschermaks Mineral.
Petrogr. Mitt., 27, 377-390, 1908.
Dymek, R. F., and C. M. Schif fries, Calcic myrme-
kite: Possible evidence for the involvement of
water during the evolution of andesine anorthosite
from St-Urbain, Quebec, Can. Mineral., 25,
291-319, 1987.
Ferry, J. M., Reaction progress: A monitor of
fluid-rock interaction during metamorphic and
hydrothermal events, in Fluid-Rock Inter actions
During Metamorphism, J. V. Walther and B. J.
Wood, eds., Springer- Verlag, New York, 89-
108, 1986.
Michel-Levy, A., De quelques characteres micro-
scopiques des roches anciennes acides, consid-
eres dans leurs relations avec l'age des eruptions,
Soc. Geol. France Bull., 3rd series, 3, 199-236,
1874-75.
Newton, R. C, J. V. Smith, and B. F. Windley,
Carbonic metasomatism, granulites, andcrustal
growth, Nature, 288, 45-50, 1980.
Phillips, E. R., Myrmekite — one hundred years
later, Lithos, 7, 181-194, 1974.
Rumble, D., J. M. Ferry, T. C. Hoering, and A. J.
Boucot, Fluid flow during metamorphism at the
Beaver Brook fossil locality, New Hampshire,
Am. J. ScL, 282, 886-919, 1982.
Schiffries, C. M., and R. F. Dymek, Calcic myrme-
kite in gabbroic and anorthositic rocks, Geol.
Soc. Am. Abstr. Program., 17, 709, 1985.
Schwantke, A., Die Beimischung von Ca im
Kalifeldspat und die Mymrekitbildung, Cen-
tralbl. Mineral, 311-316, 1909.
Sederholm, J. J., On synantectic minerals and
related phenomena, Comm. Geol. FinlandeBull.,
153, 1-148, 1916.
Thompson, J. B., Reaction space: An algebraic
and geometric approach, Rev. Mineral., 10, 33-
51,1982.
Thompson, J. B., Jr., J. Laird, and A. B. Th-
ompson, Reactions in amphibolite, greenschist,
and blueschist, /. Petrol., 23, 1-27, 1982.
30
CARNEGIE INSTITUTION
Liquid- Absent Aqueous
Fluid Inclusions
Craig M. Schiffries
A wide variety of geological phenom-
ena are governed by interactions between
fluids and rocks at elevated temperatures
and pressures. Studies of fluid inclusions
provide important constraints on the chemi-
cal composition of crustal fluids and the
physical conditions of fluid-rock interac-
tions. Raman spectra and microthermom-
etric data are reported here for a new class
of aqueous fluid inclusions that is charac-
terized by the absence of a liquid phase at
20°C. The inclusions hold special interest
because they display the following proper-
ties: (1) Although they do not contain a
liquid phase at 20°C, the inclusions ho-
mogenize to an aqueous liquid at elevated
temperatures; (2) Initial melting occurs at a
reaction point (+29°C), rather than a eutec-
tic point as commonly assumed (TE = -52°C
for the system CaCl,- NaCl - H20); (3) Ice
is absent in the subsolidus assemblage
despite the high-H20 contents of the inclu-
sions. At room temperature, most of the
water occurs as structurally bound H20 in
hydrate minerals and a relatively small
amount occurs in a low-density vapor phase;
(4) The most abundant daughter minerals
in the subsolidus assemblage are antarc-
ticite (CaCl2»6H20) and a second hydrate
that may be a new mineral; (5) The fluid
compositions fall outside the compositional
limits defined by previous studies of natu-
ral fluid inclusions. The inclusions de-
scribed here occur in quartz from a mafic
pegmatoid in the Bushveld Complex.
Similar inclusions probably occur in other
geological environments, but they may be
overlooked or misinterpreted because their
solidus temperature is above 20°C.
At room temperature, the liquid-absent
aqueous inclusions consist of birefringent
hydrates (-60 to 70 volume percent), halite
(less than 10 volume percent), and a low
density vapor (-20 to 30 volume percent).
Raman spectroscopy was used to identify
hydrate daughter minerals that are difficult
to distinguish by optical microscopy
(Dubessy et aL, 1982). Experimental con-
ditions of the micro-Raman optical system
are similar to those described by Hemley et
al. (1987). The most abundant phase in the
subsolidus assemblage has asymmetric
Raman bands at 1660 and 3430 cm1 that are
indicative of the bending and stretching
modes, respectively, of structural water in
antarcticite. The presence of a second
hydrate daughter mineral in the subsolidus
assemblage is indicated by an additional
peak at 1620 cm1 (Fig. 14). The second
hydrate has not been positively identified,
but the most likely possibility is a poly-
morph of CaCL/4H20. A preliminary study
of synthetic compounds indicates that at
least one polymorph of CaCl^HjO has a
Raman peak at approximately 1620 cm1.
None of the CaCL^Hp polymorphs is
currently recognized as a mineral species.
Microthermometric measurements were
performed with a USGS-type, gas-flow,
heating-cooling stage (Fluid Inc.) mounted
on a petrographic microscope. Initial melt-
ing occurs at approximately +29°C and a
large fraction of liquid (more than 25 vol-
ume percent) is present at 30°C in some
inclusions. It appears that antarcticite melts
GEOPHYSICAL LABORATORY
31
Ice (H20)
Hydrohalite + L
1300 1500 1700 1900
A Wavenumber (cm1)
Fig. 14. Micro-Raman spectra (OH-bending mode
region) of hydrate daughter minerals at 20°C. The
position and shape of the Raman band at 1660
cm1 in the bottom spectrum are consistent with the
spectrum of antarcticite. The presence of a second
hydrate daughter mineral in the subsolidus assem-
blage is suggested by an additional Raman band at
1620 cm"1 (top spectrum). The mineral responsible
for the peak at 1620 cm1 has not been positively
identified, but the most likely possibility is a
polymorph of CaCl2»4H20. The Raman band at
1660 cm1 in the top spectrum may reflect contri-
butions from both antarcticite and CaCl2»4H20.
incongruently at the initial melting tem-
perature and the inclusions subsequently
consist of CaCl^H/X?) and vapor. With
increasing temperature, the final hydrate
[CaCl2«4H20(?)] rapidly diminishes in size
and melts at +32° to +38°C. Halite is pres-
ent in some inclusions [Tm(halite) =
199±26°C] and it is probably metastably
absent in other inclusions. Further heating
causes the vapor bubble to shrink and the
inclusions homogenize to an aqueous liq-
uid.
The phase equilibria can be interpreted
in terms of the vapor-saturated, liquidus
diagram for the system CaClj- NaCl - H20
(Fig. 15; see also Brass, 1980; Crawford,
Hydrohalite
(NaCI-2H20)
Halite (NaCl)
E~
-52°C
Pf
- -23°C
P2"
•+29°C
P3-
- +45°C
Antarcticite + L
ntarcticite (CaCl2-6Hs>0)
' 4H26 + L
4H20 .
aCI,-2H,0
CaCI,
Fig. 15. Provisional vapor-saturated, liquidus
diagram for the system CaCL-NaCl-Hp (after
Brass, 1980; Crawford, 1981; Oakes etaU 1988;
Vanko et al, 1988; Zhang and Frantz, 1989).
Boundaries of the CaCl2#4P^O liquidus are based
on extrapolations that are consistent with the
Alkemade theorem, Schreinemakers rules and the
binary CaClj-I^O phase diagram. The liquidus
fields for antarcticite and CaCl2*4H20 have been
enlarged for clarity. Subsolidus assemblages are
indicated by dashed lines. The location of reaction
point P2 relative to the subsoludus join between
antarcticite and hydrohalite is uncertain, but the
point lies above the join between CaCl^HjO and
hydrohalite. For the assemblage ice + hydrohalite
+ antarcticite, the first equilibrium melt forms at
-52°C at the ternary eutectic (E). For the assem-
blage halite + hydrohalite + antarcticite, the first
equilibrium melt forms at approximately -22°C at
Px. For the assemblage halite + antarciticte +
CaCl2*4H20, the first equilibrium melt forms at
approximately +29°C at P2.
1981; Oaks etal., 1988; Vanko etai, 1988;
Zhang and Frantz, 1989). For the subsoli-
dus assemblage antarcticite + CaCL^F^O
+ halite, the first equilibrium melt forms at
approximately +29°C at reaction point P2.
At P2 antarcticite melts incongruently, to
CaCl^HjO and an aqueous liquid. With
increasing temperature, the composition of
the liquid evolves along the cotectic bound-
ary between halite and CaCl2«4H20 until
the latter phase melts completely at 32° to
38°C. The composition of the liquid subse-
quently migrates across the halite liquidus
until 7m(halite). Previous studies (e.g.
Crawford, 1981; Vanko et al.y 1989) indi-
cate that most fluid inclusions in the system
32
CARNEGIE INSTITUTION
Ice (H20)
Hydrohalite + L
Hydrohalite
(NaCI-2H20)
Halite (NaCI)
E ~ -52°C
Pi - -23°C
P2 - +29°C
P3 ~ +45°C
Antarcticite + L
ntarcticite (CaCl2-6H20)
"aCI2.4H26 + L
aCI2.4H20
aCI2-2H20
CaCI.
Fig. 16. The bulk composition of the fluid lies at
the intersection of the ^(halite) isotherm, and the
tie-line between halite and the point on the halite-
CaCl2*4H O cotectic that corresponds to
rm(CaCl24Hp).
CaC^- NaCI - H20 exhibit the following
melting behavior: the subsolidus assem-
blage consists of ice + hydrohalite + antarc-
ticite; initial melting occurs at the eutectic
point (£); and antarcticite is consumed at
the eutectic temperature (-52°C). Given
the topology of the phase diagram (Fig.
15), small variations in bulk composition
may result in large differences in the sub-
solidus mineral assemblage, the initial
melting temperature, the equilibrium melt-
ing sequence, and the phase ratios at 20°C
(Schiffries, in preparation).
In theory, the bulk composition of the
inclusions discussed here can be deter-
mined from two measurements: (1)
TJCaCl^HjO) in the presence of a cotec-
tic liquid plus halite and vapor and (2)
TJhalite) in the presence of liquid and
vapor. By neglecting the mass of the vapor,
the bulk composition lies at the intersection
of the halite dissolution isotherm, and the
tie-line between halite and the point on the
halite - CaCl2*4H20 cotectic correspond-
ing to 7m(CaCl2«4H20) (Fig. 16). In prac-
tice, the bulk composition cannot be deter-
mined precisely by this technique because
there are uncertainties in the location of the
halite - CaCl2«4H20 cotectic and the halite
dissolution isotherms in the relevant part of
the phase diagram. Despite uncertainties in
the liquidus diagram, the subsolidus as-
semblage indicates that the compositions
of these inclusions fall outside the com-
positional limits defined by previous stud-
ies of natural fluid inclusions. The fluids
are characterized by a high Ca/Na ratio and
a very high concentration (greater than 52
wt %) of total dissolved solids.
References
Brass, G. W., Stability of brines on Mars, Icarus,
42, 20-28, 1980.
Crawford, M. L., Phase equilibria in aqueous fluid
inclusions, in Short Course in Fluid Inclusions:
Applications to Petrology, L. S. Hollister, and
M. L. Crawford, eds., Short Course Handbook 6,
Mineral Assoc. Canada, Ottawa, pp. 75-100,
1981.
Dubessy, J. D., D. Audeoud, R. Wilkins, and C.
Kosztilanyi, The use of the Raman micro-probe
MOLE in the determination of the electrolytes
dissolved in the aqueous phase of fluid inclusions,
Chem. Geol, 37, 137-150, 1982.
Hemley, R. J., P. M. Bell, and H. K. Mao, Laser
techniques in high-pressure geophysics, Sci-
ence, 237, 605-612, 1987.
Oakes, C.S., RJ. Bodnar, and J. M. Simonson,
Phase equilibria in the system NaCI - CaC^ -
H20: The ice liquidus, Geol. Soc. Am. Abstr.
Progm., 20, A390, 1988.
Vanko, D. A., R. J. Bodnar, and S. M. Sterner,
Synthetic fluid inclusions: VIII. Vapor- satu-
rated halite solubility in part of the system NaCl-
CaCl2-H20, with application to fluid inclusions
from oceanic hydrothermal systems, Geochim.
Cosmochim. Acta, 52, 2451-2456, 1988.
Zhang, Y. G., and J. D. Frantz, Experimental
determination of the compositional limits of
immiscibility in the system CaC^-F^O-COj at
high temperatures and pressures using synthetic
fluid inclusions, Chem. Geol, 74,289-308, 1989.
GEOPHYSICAL LABORATORY
33
Igneous and Metamorphic Petrology -
B. Experimental Studies
Oxygen Fugacity and Evaporation
Phase Relations in the Solar Nebula
Bjorn O. My sen and Ikuo Kushiro*
Characterization of the pressure (P) -
temperature (T) - oxygen fugacity \f(02)]
phase relations that govern evaporation,
condensation and melting relations in the
system CaO - MgO - A1203 - Ti02 - Si02 is
of interest because this system contains
most of the refractory minerals (perovskite,
corundum, spinel, hibonite and calcium
dialuminate) expected to form at the high-
est temperatures during rock-forming proc-
esses in the early solar nebula. Principal
evidence for this suggestion is the phase
assemblages in Ca-, Al-rich inclusions
(CAFs) in carbonaceous chondrites (see
MacPherson et ai, 1988 for review).
The suggested/(02) range during rock-
forming processes in the early solar nebula
is from about 5 orders of magnitude below
to perhaps 2 orders of magnitude above
that defined by the iron-wustite (IW) oxy-
gen fugacity buffer (e.g., Fegley, 1985;
Brett and Sato, 1984; KozuletaL, 1986). It
is possible that partial or complete reduc-
tion of metal cations (Si4+, Al3+, Ca2+, Ti4+,
and Mg2+) in the system may take place in
the lowest portion of this/fO^ range. Thus,
the phase relations in the early solar nebula
* Address: Geological Institute, University of
Tokyo, Hongo-Tokyo 113, Japan
will be affected by the oxygen fugacity. Ex-
perimental determination of phase rela-
tions as a function of/(02) is reported here.
The Knudsen Cell technique (Knudsen,
1909) has been used to measure vapor
pressures of the phase assemblages (with
modifications and calibrations described
by My sen and Kushiro, 1988). The samples
were contained in Mo and C Knudsen cells.
Through interaction between the sample
containers (Mo and C) and the oxide start-
ing materials, /(02)- values at or near the
Mo - Mo02 and C - CO - C02 buffers were
defined This/(02) range covers that from
near the highest oxygen fugacities recorded
from intrinsic oxygen fugacity measure-
ments of chondrites (1-2 orders of magni-
tude above the IW buffer; Brett and Sato,
1984; Kozul et al, 1986) to about two
orders of magnitude above the /(02) sug-
gested for the primordial solar nebula (5-6
orders of magnitude below the IW buffer;
Fegley, 1985).
The melting and vaporous phase rela-
tions of the high-temperature refractory
aluminates and perovskite are distinctly
dependent on the/(02) (Fig. 17). Not only
do the P - T coordinates of the vaporous
boundaries change as the/(02) is lowered,
but in the case of the most important phases
such as spinel (MgAl204), hibonite
(CaO6Al203) and perovskite (CaTi03), the
vaporous phase also changes as a function
of /(Oj) at pressures below the triple point
34
CARNEGIE INSTITUTION
Spinel
1400 1600 1800 2000
Temperature, °C
£ 10"2
CD 10
-4
3 10-6
8 io*
qI 1400
1600 1800 2000
Temperature, °C
Hibonite
g10"5
^10"7
CO lu
CD10"9
Mo - Mo02
Hib+V
_g 10"^C-CO-CO2
CD 10
-5 .
1400 1600 1800 2000
Temperature, °C
sho-7
W Q
£ 10"9
CL 1400
Cor+
L+V
• Hib+V
L+V -
1600 1800 2000
Temperature, °C
Perovskite
§ 10-8
CL
1400 1600 1800
Temperature, °C
c5 10"2
13 10"6
co lu
0) 1Q-8
CL 1400
c-
CO-
co0
.Rt
,L-
!+>-
l+v
Rt +
V
^
. Pv
+ V
V
1600 1800 2000
Temperature, °C
Fig. 17. Pressure - temperature trajectories of phase relations in the systems MgAl204, CaO»6Al203 and
CaTiOj at the/(02) of the Mo - Mo02 and C - CO - C02 oxygen fugacity buffers. Abbreviations: Sp -
spinel, Cor - corundum, Hib - hibonite, Pv - perovskite, Rt - rutile, L - liquid, V - Vapor. Closed symbols
represent experimental points along a vaporous with congruent evaporation, grey symbols show the
vaporous of liquid-bearing assemblages, and open symbols represent incongruent evaporation phase
relations.
(marked a in Fig. 17). Whereas spinel,
hibonite and perovskite are the vaporous
phases at the highftO,) (Mo - Mo02), at the
lower/(02) (C - CO - C02), corundum (for
the aluminates) or a Ti02 phase (for
perovskite starting material, probably ru-
tile) becomes the vaporous phase.
These changes in vaporous phase rela-
tions result from partial reduction of Ca2+
and Mg2+ in the vapor. From the slopes of
the vapor pressure (In Pv) versus absolute
temperature (1/T) for the CaO and MgO
GEOPHYSICAL LABORATORY
35
«.60 5.00 5.40
1/Tx104(K"1)
5.80
•c-co-co2
■ Mo - Mo02
AH AS
AH AS
AIA 466 110
405 80
MgO 411 115
332 67
CaO 530 159
349 54
05
>
GL
4.80 5.20 5.60 6.00
1/Tx104(K'1)
4.60 5.00
l/TxKTOC1)
5.40
4/lS-V
5.80
Fig. 18. Vapor pressure (natural logarithm, In Py) versus temperature (1/T) for the systems AljOj, CaO
and MgO at the/(02) of the Mo - Mo02 and C - CO - C02 oxygen fugacity buffers. Thermodynamic
parameters pertinent to evaporation are also shown.
system (Fig. 18) it is evident that both the
entropy, AS, and enthalpy, AH, of evapora-
tion increases with the reduced /(02). The
effect of /(Oj) on the corundum vaporous
relations is significantly less.
The increased AH and AS values of
evaporation of CaO and MgO with
decreasing fiQ2) (Fig. 18) can be rational-
ized by suggesting increased dissociation
in the vapor by lowering the^Oj) (e. g., Ca,
Mg, 02 and O). At the /(02) of the Mo-
Mo02 buffer, the evaporation reaction is:
MgO(s) <=> MgO(v),
(1)
with a AH of evaporation of 331 KJ/mol
(Fig 18). From the data of Do wart et al.
(1964) about 50% of the MgO(v) will be in
the elemental state at the log f{Q^) of the C-
CO-C02 buffer. The resulting reduction of
the activity of Mg2+ most likely is the expla-
nation of incongruent evaporation of spinel
at the latter fiQ2), whereas spinel evapo-
rates congruently at the^OJ of the Mo -
Mo02 buffer.
Similar reasoning can be applied to
evaporation of lime (CaO) at the two differ-
ent oxygen fugacities. Analogous species
(Ca, CaO, O and 02) exist in the vapor from
this system as in the Mg-0 system . The AH
for the reaction:
CaO(v) + O(v) <=> Ca(v) + 02(v), (2)
is 117 KJ/mol (Dowart etai, 1964). From
the data in Fig. 18, this enthalpy would
imply essentially complete dissociation of
CaO(v) to Ca(v) and 02(v) as the oxygen
fugacity is reduced from that of the Mo -
Mo02 buffer to that of the C - CO - C02
buffer. The consequent reduction in Ca2+
activity in the vapor (probably can explain
36
CARNEGIE INSTITUTION
~1450°C
~1445°C
~1375°C
~1380°C
Z
MaO
Ptot = 103bar
mol %
CaO
C-CO-CO.
MgO
~1250°C
~1200°C
~1210°C
Ptot= 10* bar
mol %
CaO2AL0,\AI,0, CaO
CaO6AI203
CaO-2AI203\AI203
CaO-6AI203
S: ~1 500°C
i,: ~1475°C
i2: ~1485°C
l3: ~1485°C
Z
MaO
CaO
Ptot=10-3bar
mol %
Mo-MoO,
Hib
Cor
CaO-2AI203\AI203
CaO-6AI203
~1290°C
~1275°C
~1290°C
~1295°C
7ZZ
MgO
Ptot = 10-5bar
mol %
Hib
Cor
CaO
CaO-2AL03\AI20
CaO-6AI203
Fig. 19. Vaporous surfaces in the system CaO - MgO - A^ - SiO, at 103 bar (A,C) and 105 bar (B,
D) total pressure. S denotes solar CaO/MgO/Al203 (from Anders and Ebihara, 1982). Si02 content is mat
of the solar abundance. Temperatures of various invariant points and the vaporous surface temperatures
of S are indicated in the inserts. The overall dilution in the gas is about l& (solar dilution). Abbrevia-
tions: Cor - corundum, Sp spinel, Pe - periclase, CA2 - calcium dialuminate, Hib - hibonite, Fo - forsterite.
Oxygen fugacity is at that of the C-CO-C02 (A,B) and Mo - Mo02 (C,D) buffer.
the change from congruent to incongruent
evaporation ofhibonite and perovskite (Ca-
Ti03) as the^Oj) is lowered.
From the evaporation data reported here
and elsewhere (Mysen, 1988; Mysen and
Kushiro, 1988), vaporous surfaces in the
system CaO - MgO - Al^ - Si02 with Mg/
Si = 1 have been calculated (Fig. 19) under
the assumption of ideal mixing in the gas
phase and with the proportion of oxide
GEOPHYSICAL LABORATORY
37
components relative to an inert gas diluent
equal to that of the solar nebula (-104;
Anders and Ebihara, 1982). The condensa-
tion and evaporation sequences are strongly
affected by /(02). For example, spinel is
only stable at high pressure, or high/(02),
orboth. The occurrence of spinel-rich CAI's
may indicate that such conditions existed
in the solar nebula. Corundum andhibonite-
rich phase assemblages probably required
either low total pressure (less than -10 5
bar), or that the/(02) was at or below that of
the C - CO - C02 oxygen buffer during their
formation. For example, reheating and
partial evaporation of materials rich in
organic carbon will take place with/(02)
near that of the C - CO - C02 buffer. With
this low oxygen fugacity, the evaporation
process yields a residue enriched in alumi-
num relative to that of the starting material.
In the early solar nebula where tempera-
tures and pressures probably increased
toward its center, the experimentally deter-
mined phase relations lead to the sugges-
tion of an overall increase in (Mg + Ca)/Al
in the condensates with decreasing dis-
tance to the Sun. This trend may be retained
in the bulk composition of the planets to-
day.
References
Anders, E., and M. Ebihara, Solar system abun-
dances of the elements, Geochim.Cosmochim.
Acta, 46, 2363-2380, 1982.
Brett, R., and M. Sato, Intrinsic oxygen fugacity
measurements on seven chondrites, a pallasite
and a tektite and the redox state of meteorite
parent bodies, Geochim. Cosmochim. Acta, 48,
111-120,1984
Dowart, J., G. Exsteen, and G. Verhaegen, Mass
spectrometric determination of the dissociation
energy of thermocouples MgO, CaO, SrO and
Sr20*, Trans, Faraday Soc, 60, 1920-1933,
1964.
Fegley, B., Oxidation state indicators of the solar
nebula, Lunar Planet. Sci. XVI, 232-233, 1985.
Kozul, J., G. C. Ulmer, and R. Hewins, Allende
inclusions are oxidized!, EOS, 67, 300, 1986.
Mysen, B. O., Rock - forming processes in the
early solar nebula: Phase relations in the system
CaO - MgO - Aip3 - Si02 to 2000°C and 1 0 ■» bar,
Annual Report of the Director of the Geophysi-
cal Laboratory, Carnegie Instn. Washington,
1987-1988, Geophysical Laboratory, Washing-
ton, D. C. , 69-76, 1988.
Mysen, B. O., and I. Kushiro, Condensation,
evaporation, melting and crystallization in the
primitive solar nebula: Experimental data in the
system MgO - Si02 - H2 to 10x1 a9 bar and
1870°C with variable oxygen fugacity, Am.
Mineral., 73 1 1-19, 1988.
Knudsen, M., Die Molekularstromung der Gase
durch Offnungen und die Effusion, Ann. Phys.,
25,999-1016,1909.
MacPherson, G. J., D. A. Wark, and J. T. Arm-
strong, Primitive materials surviving in chon-
drites: refractory inclusions, in Meteorites and
the Early Solar System, J. F. Kerridge and M. S.
Matthews, eds., The University of Arizona Press,
Tucson, pp. 746-807, 1988.
Experimental Determination of Ele-
ment Partitioning and Calculated
Phase relations in the Mg-Fe-Si-0
System at High Pressure and High
Temperature
Yingwei Fei, Ho-kwang Mao, and
Bjorn O. Mysen
The principal components of the Earth's
mantle are MgO, FeO and Si(\ Phase rela-
tions in this system are of great interest to
geochemists and geophysicists because both
the 400 km and the 670 km seismic discon-
38
CARNEGIE INSTITUTION
tinuities may reflect the phase transforma-
tions of olivine (a) to p-phase and of spinel
(y) to perovskite plus magnesiowiistite,
respectively.
The phase relations can be established
in two ways. One is to determine phase
boundaries in P-T-X space (e.g. Akimoto,
1987; Katsura and Ito, 1989; Ito and Taka-
hashi, 1989). The other is to determine
precisely the phase boundaries of pure
phases and the mixing properties of each
solid solution. The phase relations in bi-
nary or multi-component systems can then
be calculated. Solid solutions in the Mg-
Fe-Si-0 system include (Mg,Fe)0 (mag-
nesiowiistite), (Mg,Fe)2Si04 (olivine, p-
phase, spinel) and (Mg,Fe)Si03 (pyroxene,
ilmenite, perovskite). These solid solutions
form four pairs of coexisting phases with
magnesiowiistite (Mw), Mw-olivine, Mw-
p, Mw-spinel and Mw-perovskite. The
solution properties of the individual phase
may be derived from the element distribu-
tion data by considering the distribution of
an element between two solid solutions as
an exchangeable reaction (Fei and Saxena,
1986). The purpose of this study is to derive
the mixing properties of each solution by
determining experimentally the distribu-
tion of Mg and Fe between coexisting solid
solutions at various pressure and tempera-
ture conditions and to establish phase rela-
tions in the system.
The experiments are conducted with the
piston-cylinder apparatus (up to 50 kbar)
(Boyd and England, 1960, 1963), the multi-
anvil device (up to 300 kbar) (Liebermann
et al. , 1 986 and Remsberg et al. , 1 988), and
the diamond-anvil cell device (Mao and
Bell, 1978). Chemical and structural ana-
lytical techniques such as x-ray diffraction,
microprobe or SEM and Raman spectros-
copy were used to characterize the struc-
ture and chemical composition of the
samples.
Synthetic olivine and magnesiowiistite
solid solutions with different iron contents
were used as starting materials. Magne-
siowiistite solid solutions were those syn-
thesized by Rosenhauer et al. (1976). Oli-
vine solid solutions were synthesized in the
piston-cylinder apparatus, with oxide mix-
tures as starting material, for 48 hours at
1273K and 15 kbar. The synthetic olivines
were examined optically and with x-ray
diffraction; no oxide remainder was pres-
ent. The compositions were checked by
electron microprobe.
For the Mg-Fe partition experiments
with magnesiowiistite and olivine, magne-
siowiistite and olivine of suitable composi-
tions were mixed in appropriate proportions
(e.g., 2 magnesiowiistite to 3 olivine in
most cases). The mixtures were ground to
a grain size of less than 3 jim and well ho-
mogenized. Two different types of cap-
sules, platinum capsule sealed inside with
graphite capsule and molybdenum cap-
sule, were used to test if there was iron loss
in the runs. No evidence for iron-loss was
found.
The piston-cylinder apparatus was used
for experiments below 50 kbar. The multi-
anvil device was used for determining dis-
tribution coefficients between magne-
siowiistite and olivine at 90 kbar and 1 723K,
between magnesiowiistite and p-phase and
between magnesiowiistite and spinel at 1 50
kbar and 1773K. The multi-cell sample
chambers described by Mao et al. (1989)
were used in the experiments.
GEOPHYSICAL LABORATORY
39
1.0
0.8
1 £ 0.6 [
X
0.4 -
0.2 -
0.0
a 20 kb, 1473 K
. 20 kb, 1723 K
0.0 0.2 0.4 0.6 0.8 1.0
X
oi
Fe
Fig. 20. Distribution of Mg and Fe between coex-
isting magnesiowiistite (Mw) and olivine (Ol) at a
pressure of 20 kbar and at temperatures of 1473K
(upper curve) and 1723K (lower curve). Curves
are calculated results and symbols represent the
experimental data.
Figs. 20 and 21 show the temperature
and pressure dependences of the Mg-Fe
distribution coefficients between magne-
siowiistite and olivine. The experimental
results show systematic variations of the
distribution data with temperature and
pressure.
I.U
• i '
~i — • — i — ■—J'-'- **
^*~*~
f
0.8
/
-
■ /
■
<: 0.6
-
5 ©
2 u.
f
•
X 0.4
J
P = 150kb '
0.2
T=1773K -
C- *. i i
-i — i i l,..i i
0.0 0.2 0.4 0.6 0.8 1.0
X
P
Fe
Fig. 22. Distribution of Mg and Fe between coex-
isting magnesiowiistite (Mw) and P-phase (p) at a
pressure of 150 kbar and at a temperature of
1773K. Curves are calculated results and symbols
represent the experimental data.
+ 90kb, 1723K.
a 40 kb, 1723K
■ 20 kb, 1723 K
0.0 0.2
0.4 0.6
-Ol
Fe
0.8 1.0
X
Fig.21. Distribution of Mg and Fe between coex-
isting magnesiowiistite (Mw) and olivine (Ol) at a
temperature of 1723K and at pressures of 20 kbar
(lower curve), 40 kbar (middle curve) and 90 kbar.
The Mg-Fe distribution coefficients
between magnesiowiistite and p-phase or
spinel are shown in Figs. 22 and 23. The
results by Ito et al. (1984) and Yagi et al.
(1979) are also shown for comparison.
1.0
0.8"
| 0)
X
0.6"
0.4
0.2
0.0
I » ! ' I ' ■—rjpii^'
—
&yA
-
kw*
-
JTa This work
/ + Yagiefa/, 1979
"tr
a Ito, 1984
i i i i i i * j
0.0 0.2 0.4 0.6 0.8 1.0
X
Sp
Fe
Fig. 23. Distribution of Mg and Fe between coex-
isting magnesiowiistite (Mw) and spinel (Sp) at a
temperature of 1773K and pressures of 150 kbar
(lower curve), and at temperature of 1873K and
pressure of 200 kbar (upper curve). Curves are
calculated results and symbols represent the ex-
perimental data.
40 CARNEGIE INSTITUTION
Table 3. Nonideal mixing parameters of solid solutions in the Mg-Fe-Si-O system
Solid solutions
Wr (J/mol)
MgFe
Wr (J/mol)
Magnesiowiistite
Olivine (a)
p-phase
Spinel (y)
Perovskite
16100
4500 + 0.013P
1000
3900-1. 107
4130- 1.377 + 0.01 IP
26300 - 5.567
6500 + 0.013/>
2000
3900
-4050- 2.457 + 0.015P
To determine the distribution of Mg and
Fe between magnesiowiistite and
perovskite, olivines (Fo85, Fo80 and Fo73)
were used as starting materials in the dia-
mond-anvil cell. Olivine with small ruby
grains was imbedded in a 250 jim hole
drilled in a gasket. The sample was pressur-
ized in the diamond-anvil cell and heated
by YAG laser beam. The pressure was
measured with the fluorescence technique.
The products were examined optically and
by x-ray diffraction. The composition of
■ This work
a Yagi etal, 1979
+ Ito, 1984
J • L
0.0 0.2 0.4 0.6 0.8 1.0
X
Pv
Fe
Fig. 24. Distribution of Mg and Fe between coex-
isting magnesiowiistite (Mw) and perovskite (Pv)
at a temperature of 1873K and pressures of 260
kbar (lower curve), and 300 kbar (upper curve).
Curves are calculated results and symbols repre-
sent the experimental data.
each phase in the assemblage was deter-
mined both by x-ray diffraction (the rela-
tion between volume and composition us-
ing the calibrations of Yagi et ah, 1979;
Rosenhauer et al.9 1976) and by electron
microprobe. The lattice parameters of each
phase were determined by x-ray diffrac-
tion, with gold as internal standard for
calibration. The results accord with those
determined with the electron microprobe.
The results on the Mg-Fe partitioning be-
tween magnesiowiistite and perovskite are
shown in Fig. 24. These are in agreement
with those obtained by Ito etal. (1984) and
Yagi et al. (1979).
Solution parameters (W.) for five solid
solutions have been obtained by fitting the
experimental data simultaneously using the
Margules formulation (Table 3). Various
calculated Roozeboom diagrams for the
exchange of Mg2+ and Fe2+ between coexist-
ing solid solutions are shown in Figs. 20 -
24.
The solution parameters listed in Table
3 with an internally consistent thermody-
namic data set on phases in the system (Fei,
1989) can be used to compute phase rela-
tions. Fig. 25 shows computed isothermal
phase relations in the binary system Mg2Si04
- Fe2Si04 at temperatures of 1473K and
GEOPHYSICAL LABORATORY
41
300
0.0 0.2 0.4 0.6 0.8 1.0
Mg2Si04 XFe Fe2Si04
Fig. 25. Calculated isothermal phase relations in
the binary system Mg2Si04 and Fe2Si04 at tem-
peratures of 1473K (the heavy lines) and 1873K
(the light lines) and pressures to 300 kbar.
1873K and pressures to 300 kbar.
The calculated diagram is in good agree-
ment with those determined by Katsura and
Ito (1989) and Ito and Takahashi (1989) in
the Mg-rich region and by Akimoto (1987)
in the Fe-rich region. The thermodynamic
data base evaluated from experimental data
can be used not only for reproducing the
existing experimental results, but also for
interpolation and even extrapolation with
caution. It allows us to explore phase rela-
tions in the system in various ways. For in-
stance, one may construct phase relations
in pressure-temperature space or in pres-
sure-composition space. In the mantle
model of peridotitic composition where
olivine (Mg088Fe012)2SiO4 is the major
component, the phase diagram of this oli-
vine indicates that the depth and width of
the phase transformations of olivine to p-
phase and of spinel to perovskite plus
magnesiowiistite are compatible with the
seismic observation of the 400 km and the
670 km discontinuity, respectively. The
density profile of the mantle can be simu-
lated by varying chemical composition
along the assumed geotherm. However, to
make the comparison between the calcu-
lated and observed profiles, at least Ca and
Al should be included in the system. Ex-
perimental determination and computer
simulation of phase relations in the ex-
tended system can provide critical con-
straints for models of the Earth's mantle.
References
Akimoto, S., High-pressure research in geophys-
ics: past, present and future, in High Pressure
Research in Mineral Physics, M. H. Manghnani
and Y. Syono, eds., Terra Scientific Publishing
Company (TERRAPUB), Tokyo/American
Geophysical Union, Washington, D. C, pp. 1-
13, 1987.
Boyd, F. R., and J. L. England, Apparatus for
phase-equilibrium measurements at pressures
up to 50kilobars and temperatures up to 1750°C,
/. Geophys. Res., 65, 741-748, 1960.
Boyd, F. R., and J. L. England, Effect of pressure
on the melting of diopside, CaMgS^Og, and
albite, NaAlSi3Og, in the range up to 50 kilobars,
/. Geophys. Res., 68, 311-323, 1963.
Fei, Y., Thermochemical - thermophysical data on
phases in the Mg-Fe-Si-O system: a synthesis of
theory and experimental data and computation
of phase equilibrium, Ph.D. dissertation, City
University of New York, 1989.
Fei, Y., and S. K. Saxena, A thermochemical data
base for phase equilibria in the system Fe-Mg-
Si-0 at high pressure and temperature, Phys.
Chem. Minerals, 13, 311-324, 1986.
Ito, E., and E. Takahashi, Post- spinel transforma-
tions in the system Mg2Si04-Fe2Si04 and some
geophysical implications, /. Geophys. Res., in
press, 1989.
Ito, E., E. Takahashi, and Y. Matsui, The mineral-
ogy and chemistry of the lower mantle: an impli-
cation of the ultrahigh-pressure phase relations
42
CARNEGIE INSTITUTION
in the system MgO - FeO - Si02, Earth Planet.
Sci. Lett., 67, 238-248, 1984.
Katsura, T., and E. Ito, The system Mg2Si04-
Fe2Si04 at high pressures and temperatures:
precise determination of stabilities of olivine,
modified spinel and spinel, /. Geophys. Res., in
press, 1989.
Liebermann, R. C, H. Watanabe, T. Gasparik, C.
T. Prewitt, and D. J. Weidner, High pressure
mineral synthesis in USSA-2000, EOS, 67, 361,
1986.
Mao, H. K., and P. M. Bell, Design and varieties
of the megabar cell, Carnegie Inst. Washington
Year Book, 77,904-908, 1978.
Mao, H. K., Y. Fei, and T. Gasparik, Multi-cell
sample chamber in multi-anvil apparatus - a tool
for high-resolution and high-efficiency experi-
mental studies, EOS, 70, 471, 1989.
Remsberg, A. R., J. N. Boland, T. Gasparik, and R.
C. Liebermann, Mechanism of the olivine-spi-
nel transformation in Co2Si04, Phys. Chem.
Minerals, 15, 498-506, 1988.
Rosenhauer, M., H. K. Mao, and E. Woermann,
Compressibility of (Fe04Mg06)O magne-
siowiistite to 264 kbar, Carnegie
Instn.Washington Year Book, 75, 5 1 3-5 1 5, 1 976.
Yagi, T., P. M. Bell, and H. K. Mao, Phase rela-
tions in the system MgO - FeO - Si02 between
150 and 700 kbar at 1000°C, Carnegie Instn. of
Washington Year Book, 78, 614-618, 1979.
Partitioning of High Field Strength
Elements Among Olivine, Pyroxenes,
Garnet and Calc Alkaline Picroba-
salt: Experimental Results and An
Application
Peter Ulmer
On chondrite- and MORB -normalized
trace element plots, the calc-alkaline mag-
mas of convergent plate margins character-
istically show depletion of high field
strength elements (HFSE) such as Ta, Nb,
Zr, Ti, Hf, and Y relative to the alkaline
earth and rare earth (REE) elements. This
characteristic is widely interpreted as an
imprint of a depleted MORB -type source,
a possibility that is also consistent with Sr,
Nd, and Hf isotopic relationships (e.g.,
Morris and Hart, 1 983; White and Patchett,
1984). On the other hand, the magmas
show relatively strong enrichments of
highly incompatible elements (notably Ba,
K, Rb, Sr, Th) and REE, and these charac-
teristics are frequently attributed to metaso-
matism of the mantle source by fluids re-
leased through dehydration or low-percent-
age melting of subducting oceanic crustal
slabs (Green and Ringwood, 1968; Lam-
bert and Wyllie, 1970; Arculus and Curran,
1972; Green, 1980). The need for metaso-
matism is perhaps most strongly suggested
by modelling results (e.g., Lopez-Escobar
et ai, 1977) indicating that, for a primor-
dial or chondritic mantle source, simple
batch or fractional melting can only yield
the observed incompatible element con-
centrations at degrees of melting too low
(2-5%) to be consistent with major element
constraints established by experimental
studies (e.g., Green, 1973).
The basis of this concept requiring first
depletion, then metasomatism of the mantle
source is weak, however, in that reliable
HFSE partition coefficients have not been
available for minerals and liquids of appro-
priate compositions at appropriate pres-
sure-temperature conditions. In the study
described here, the crystal/liquid partition
coefficients of Nb, Ta, P, Zr, Hf , Ti, Y, Sc,
and V have been determined experimen-
tally for olivine, clinopyroxene (cpx),
orthopyroxene (opx), and garnet in equilib-
rium with a calc-alkaline picrobasaltic melt.
GEOPHYSICAL LABORATORY
43
Table 4. Summary of the Experimentally Determined Crystal/Liquid Partition Coefficients for High
Field Strength Elements.
Element
Garnet
Cpx
Opx
Olivine
Nb
0.07
0.02
<0.01
<0.01
Ta
0.04
0.02
<0.01
<0.01
P
0.10
0.03
0.03
<0.01
Zr
0.32
0.03
0.03
0.01
Hf
0.44
0.22
0.14
<0.01
Ti
0.28
0.18
0.10
0.02
Y
2.11
0.20
<0.01
<0.01
Sc
2.27
0.51
0.33
0.16
V
1.57
1.31
0.90
0.06
Partition coefficients were determined experimentally using powders of picrobasalt sample RC158c
(Ulmer, 1988) doped with 1 wt % of oxide of the element. The melting experiments for garnet, cpx, and
opx were run at 1330°-1340°C at 28 kbar; those for olivine, at 1330°-1350°C at 1 atm with oxygen
fugacity close to the NNO buffer.
The picrobasalt (sample RC158c) comes
from late dikes in the Tertiary calc-alkaline
Adamello batholith of northern Italy. Pre-
vious phase-equilibria work showed that
the rock probably represents primary
magma because its melt is in equilibrium
with garnet lherzolite at an upper mantle
pressure of 28 kbar (Ulmer, 1988).
Experimental Study
For garnet and the pyroxenes, the HFSE
partition coefficients were determined at
28 kbar and 1330°C, conditions approxi-
mating the liquidus of the picrobasalt in the
upper mantle. Doped powders of the rock
were melted in graphite containers sealed
in Pt capsules in solid-media, high-pres-
sure apparatus at the Geophysical Labora-
tory and ETH-Zurich. Standard 12.5 mm
and 14 mm talc-.Pyrejc™ assemblies were
used. Run products typically consisted of
60-70% melt, as represented by glass and
quench products, plus stable, unzoned
crystals of cpx, opx, and garnet.
For olivine, the HFSE partitioning had
to be determined from 1-atm experiments,
because the mineral is not readily stabi-
lized at liquidus conditions at high pres-
sures owing to Fe losses from the melt to
the Pt capsules by way of cracks in graphite
container and to gains of C02 through reac-
tion of the melt with the graphite. The 1-
atm experiments were done by the Pt wire
loop technique at oxygen fugacities close
to the NNO buffer at 1330°C.
Run products from all experiments were
analyzed by electron microprobe for major
elements and HFSE. No reversals were
made, but the approach of the HFSE parti-
tioning to equilibrium was tested by com-
paring results for runs of 6 and 12 hours.
For Zr, P, and Ti, test runs of 5 1 hours were
44
CARNEGIE INSTITUTION
100
c/)
c
o
o
o
c
tr
0$
Q.
1 F
2 .1
.01
.001
Zr-Partitioning
JT*
\
Garnet/Liquid
Cpx/Liquid
10 20 30 40 50
Run duration, hours
60
Fig. 26. Variation of the crystal/liquid partition
coefficients of Zr for garnet and cpx in picrobasal-
tic liquid as a function of run duration. The error
bars indicate uncertainties for single runs.
also made. With the possible exception of
Zr, the partition coefficients were stable in
runs longer than 6 hours. The coefficient
for Zr showed a slight increase from 6 to 1 2
hours but was the same after 51 hours as at
12 (Fig. 26).
The determined partition coefficients
are summarized in Table 4 and illustrated in
Fig. 27. Garnet has the highest partition
coefficients, followed by cpx and opx in
that order. Solubilities of the HFSE in oli-
vine were barely detectable by the electron
microprobe analytical method, so its coef-
ficients are all low; in fact, only those for Sc
10
1 1
x
Q .1
.01
1 1 r
-•— Garnet
-*— Cpx
-■-- Opx
—•• Olivine
T 1 1 I 1 — 3
Nb Ta
Sc V
Fig. 27. Experimentally measured coefficients
describing the crystal/liquid partitioning of HFSE
for garnet, cpx, opx, and olivine in picrobasalt
melt.
Nb Ta
Zr Hf
Fig. 28. (A) Variation diagram for natural HFSE
abundances in picrobasalt RC158c. Concentra-
tions are chondrite normalized for all elements
except Sc and V. Primitive mantle values are from
Sun and Nesbitt (1977), Sun (1982), Thompson et
al. (1982). (B) Crystal/liquid partition coefficients
for the HFSE for garnet in picrobasaltic melt.
and V are significant. It is evident that, in
the melting of garnet lherzolite, garnet will
be the dominant control for all HFSE, but
cpx could have a substantial subordinate
influence, especially for Hf, Ti, Sc, and V.
The opx coefficients are almost as large as
the cpx coefficients for P, Zr, Hf , Ti , Sc and
V, but the opx value for Y is lower by an
order of magnitude.
The partition coefficients for garnet are
compared with the chondrite-normalized
natural HFSE concentrations of the picro-
basalt in Fig. 28. An inverse relationship is
revealed, implying strongly that garnet
largely controlled the HFSE composition
of the picrobasalt magma during partial
melting of a garnet lherzolite source.
GEOPHYSICAL LABORATORY
45
100
wjlj jlj ni m w w. u c_ *— -J h—
COCC *Z I- _lO CO 2 Q. (ON UJ I
>££3<8:
Fig. 29 . The natural trace element pattern of
picrobasalt RC158c compared with patterns cal-
culated for melts of a garnet lherzolite source
using the simple batch-melting equation of Shaw
(1970) and experimental partition coefficients
(including those obtained in the present study).
Application
Trace element patterns have been calcu-
lated for partial melts of a garnet lherzolite
source using the measured HFSE coeffi-
cients in combination with published coef-
ficients for other elements (Harrison, 1981;
Shimizu and Kushiro, 1974; Irving, 1978;
Hanson, 1980;TerakodaandMasuda, 1979;
Irving, 1 978). Two sets of calculations were
made, as illustrated in Figs. 29 and 30.
For Fig. 29, the following conditions
were assumed: (1) the source rock was a
fertile garnet lherzolite composed of 57%
(by weight) olivine, 19% opx, 12% cpx,
and 12% garnet; (2) the melting equation
was that derived in the experimental study
by Ulmer (1988): 0.582 cpx + 0.388 garnet
+ 0.031 opx <=> 0.075 olivine + 0.925
liquid; and (3) after extraction of the melt,
0.2% liquid remained trapped in the source.
The simple, non-modal batch melting
equation of Shaw (1970) was used, and
partial melting percentages of 5, 10, and
20% were investigated. The cpx content of
10 r
— •- PicriteRC158c
.
— ♦-•- 20% batch melting \
J>
— •— 1 6% batch melting ■
^^W^V, i
*Vi
**
» ' » I -1 1 1 1 1 i- 1 1 1 1 1 11 11 1 1 1 1
0)
*c
c
o
o
^»
c
E
o
lli
0
O
CO
C0.O -O (0 (0 ® ,_ -Q Ei_=Jh-"D._.q E.O 3 O
CO0C *Z I--JO CO Z O.CONUJIO \-\->\-> -J CO >
Fig. 30. The natural trace element pattern of picro-
basalt RC158c compared with patterns calculated
for aggregated liquid from a garnet lherzolite
mantle plume using the batch plume melting
equation of O'Hara (1985). The curve labeled
20% melting represents a maximum of 20% melt-
ing in the center of a plume composed of the same
fertile peridotite assumed in the calculations in
Fig. 29. The pattern for 16% melting pertains
similarly to a plume of less fertile peridotite. For
both calculated patterns, the melting maximum
represents the stage at which the cpx-content of the
source rock was exhausted.
the source is just exhausted at 20% melting.
The model patterns show that the picroba-
salt composition can be approximated by
the simple batch melting equation only if
the percentage of melting is low, of the
order of 5% or less. The heavy REE con-
centrations derived by this equation are too
low, and the depletion of the HFSE is also
poorly reproduced. This type of melting
would indeed require metasomatism of .the
source to account for the enrichment of the
highly incompatible elements and REE.
It is unlikely, however, that the simple
batch partial melting equation is appropri-
ate. The mantle source was probably a
plumelike structure in which temperature
varied both horizontally and vertically;
hence the amount of melting probably was
not constant. O'Hara (1985) has derived al-
ternative partial melting equations that
46
CARNEGIE INSTITUTION
describe trace element abundances for
aggregated melt from such a regime. The
equations are denoted CAPEPM and
CAPFPM, acronyms for Complex Aggre-
gated Perfect Equilibrium (or Fractional)
Partial Melting. They were modified for
present purposes in such a way that melt
percentages varied with temperature in
accordance with 30 kbar experiments on
the picrobasalt (Ulmer, 1988). The melt
distribution was approximated by a simple
analytical function, and the equation was
integrated numerically from the edge of the
melting region to its center. The important
consequence is that, in the aggregated liq-
uid, the melt from the high-percent-melt-
ing central part of the plume mostly con-
trols major element abundances, whereas
that from the low-percent-melting fringe
mainly controls the trace elements.
Trace element patterns from the O ' Hara-
type equilibrium (batch) calculations are
compared with the natural picrobasalt pat-
tern in Fig. 30. The dotted-line pattern
derives from the same parameters used in
Fig. 29 for 20% melting at the center of the
plume. The broken-line pattern represents
a less fertile source containing only 11%
cpx and 7% garnet. With it, cpx and garnet
are both consumed at approximately 16%
melting. The two calculated patterns both
match the natural pattern reasonably well,
but Zr and, to a lesser extent, Hf, Nb, and
Ta, are not closely reproduced. Possible
explanations of the discrepancies are:
(1) The measured distribution coefficients
for Zr, Hf, Nb, and Ta are too low in
consequence of the charge doping. The
doped concentrations are 2-4 orders of
magnitude larger than the natural abun-
dances of the picrobasalt; thus the distribu-
tion coefficients could be outside the range
of Henry's law (cf. Mysen, 1978; Harrison;
1981). The reported D-values should be
regarded as minimum values.
(2) Some additional mineral might retain
these elements in the melting range of the
picrobasalt (cf. Sauders etal., 1980; Green,
1981). For example, rutile was observed in
picrobasalt charges melted at 35 - 40 kbar
at temperatures up to 1150°C.
The O'Hara model does, however, re-
produce most of the trace element pattern
of the natural calc-alkaline picrobasalt from
a chondritic to primitive mantle source
without requiring that it be metasomatized.
Most particularly, it reproduces the enrich-
ment of LLLE and REE and the depletion of
the HFSE. The possibility of metasoma-
tism is not ruled out, but much less is
necessary than has commonly been con-
tended.
References
Arculus, R. J., and E. B. Curran, The genesis of the
calc-alkaline rock suite, Earth Planet. Sci. Lett.,
75,255-262,1972.
Green, D. H., Experimental testing of "equilib-
rium" partial melting of peridotite under water
saturated, high-pressure conditions, Can. Min-
eral., 14, 255-268, 1976.
Green, T. H. , Experimental evidence for the role of
accessory phases in magma genesis, J. Vol-
canol. Geotherm. Res., 10, 405-422, 1981.
Green, T. H., and A. E. Ringwood, Genesis of the
calc-alkaline igneous rock suite, Contrib. Min-
eral. Petrol, 18, 269-385, 1968.
Green, T. H., Island arc and continent-building
magmatism - A review of petrogenic models
based on experimental petrology and
geochemistry, Tectonophysics, 63, 367-385,
1980.
GEOPHYSICAL LABORATORY
47
Irving, A. J., A review of experimental studies of
crystal/liquid trace element partitioning, Geo-
chim. Cosmochim. Acta, 42, 743-770, 1978.
Hanson, G. N., Rare earth elements in petrogen-
etic studies of igneous systems, Ann. Rev. Earth
Planet. Sci. Lett., 8, 371-406, 1980.
Harrison, W. J., and B. J. Wood, An experimental
investigation of the partitioning of REE between
garnet and liquid with reference to the role of
defect equilibria, Contrib. Mineral. Petrol., 72,
145-155,1980.
Harrison, W. J., Partition coefficients for REE
between garnets and liquids: implications of
non-Henry's law behavior for models of basalt
origin and evolution, Geochim. Cosmochim.
Acta, 45, 1529-1544, 1981.
Lambert, J. B., and P. J. Wyllie, Melting in the
deep crust and upper mantle and the nature of
low- velocity layer, Phys. Earth Planet. Inter., 3,
316-322.
Leeman, W. P., The influence of crustal structures
on compositions of subduction-related magmas,
J. Volcanol. Geotherm.Res., 18, 561-598, 1983.
Lopez-Escobar, L. L., A. F. Frey, and M. Vergara,
Andesites and High-Alumina basalts from the
central-south Chile high Andes: Geochemical
evidence bearing on their pedogenesis, Contrib.
Mineral. Petrol., 63, 199-228, 1977.
Morris, J. D., and S. R. Hart, Isotopic and incom-
patible element constraints on the genesis of
island arc volcanics from Cold Bay and Amak
Island, Aleutians, and implications for mantle
structure, Geochim. Cosmochim. Acta, 47, 2015-
2033, 1983.
Mysen, B. O., Experimental determination of nickel
partition coefficients between liquid, pargasite,
and garnet peridotite minerals and concentration
limits of behavior according to Henry's law at
high pressure and temperature, Am. J. Sci., 278,
217-243, 1978.
O'Hara, M. J., Importance of the 'shape' of the
melting regime during partial melting of the
mantle, Nature, 314, 58-62, 1985.
Shaw, D. M., Trace element fractionation during
anatexis, Geochim. Cosmochim. Acta, 34, 237-
243, 1970.
Shimizu, N., and I. Kushiro, The partitioning of
REE between garnet and liquid at high pres-
sures: Preliminary experiments, Geophys. Res.
Lett., 2, 413-416, 1974.
Sun, S. S., and R. W. Nesbitt, Chemical heteroge-
neity of the Archean mantle, composition of the
Earth and mantle evolution. Earth Planet. Sci.
Lett., 35, 429-448, 1977.
Sun, S. S., Chemical composition and origin of the
Earth's primitive mantle, Geochim. Cosmochim.
Acta, 46, 179-192, 1982.
Terakoda, Y. and A. Masuda, Experimental study
of REE partitioning between diopside and melt
under athmospheric pressure, Geochem. J,, 13,
121-129, 1979.
Thompson, R. N., A. P. Dickin, L. L. Gibson, and
M. A. Morrison, Elemental fingerprints of iso-
topic contamination of Hebridean Palaeocene
mantle-derived magmas by Archean Sial, Con-
trib. Mineral. Petrol, 79, 159-168, 1982.
Ulmer, P., High pressure phase equilibria of a
calc-alkaline picro-basalt: Implication for the
genesis of calc-alkaline magmas, Annual Report
of the Director of the Geophysical Laboratory,
Carnegie Instn. Washington, 1987-1988, Geo-
physical Laboratory, Washington, D. C, 28-35,
1988.
White, W. M., and J. Patchett, Hf-Nd-Sr isotopes
and incompatible element abundances in island
arcs: implications for magma origins and crust-
mantle evolution, Earth Planet. Sci. Lett., 67,
167-185, 1984.
Relationships Between Composition,
Pressure and Structure of Depolymer-
ized, Peralkaline Aluminosilicate
Melts
Bjorn O. Mysen
The structure of silicate liquids at high
temperature and high pressure, and rela-
tionships between structure and properties
are important to characterize natural mag-
matic processes. Bulk compositions be-
tween depolymerized (nonbridging oxy-
gen per silicon; NBO/Si > 0) alka&or alka-
line earth silicates and fully polymerized
(NBO/Si = 0) silica - aluminates represent
48
CARNEGIE INSTITUTION
mol %
rhyolitic
andesitic J°,
basaltic ^29
30
40
komatiitic
Fig. 36. Approximate fields of major groups natu-
ral magmatic liquids in in the system MO - A1203
- Si02 (M = K + Na + Ca + Mg + Fe2+) (data from
Mysen, 1988).
the structural environment of natural mag-
matic liquids (see Fig. 36). Their A1/(A1 +
Si) typically is between 0.2 and 0.3. Alkali
metals are the principal cations for charge-
balance of tetrahedrally-coordinated Al3+ at
1 atmosphere (Mysen, 1988).The excep-
tions are picrite andkomatiite. These melts
are less polymerized and often have lower
A1/(A1 + Si) than other major groups of
magmatic melts.
Silicate melts in this composition range
(Fig. 36) consist of coexisting structural
units characterized by their average num-
ber of nonbridging oxygen per tetrahe-
drally-coordinated cations (NBO/T) equal
to 0 (T02, or (^-species), 1( T205, or Q3-
species), and 2 (T03, or (^-species; see, for
example, Virgo et al., 1980). The equilib-
rium in such melts is described with the
equation (Virgo et al., 1980);
T205(Q3)^T03(Q2) + T02(Q<), (1)
where T = Al + Si. The Q4, Q3 and Q2
notations are sometimes used in place of
the stoichiometric designations. The su-
perscript in the Q-notation denotes the
number of bridging oxygen in the unit.
Physical and chemical properties of
natural magmas depend on the properties,
detailed structure, and proportions of these
individual structural units. The relative
abundance of the different structural units
is likely to depend on melt polymerization,
the ionization potential of charge-balanc-
ing and network-modifying cations, A1/(A1
+ Si), pressure, and temperature (e.g.,
Mysen et al., 1985; Brandriss and Steb-
bins, 1988). In order to provide a structural
framework for quantitative characteriza-
tion of the properties of natural magma, it
is necessary to characterize the relation-
ships between pressure, temperature, and
the compositional variables. In view of the
fact that most magmatic processes take
place at pressures above 1 atmosphere, it is
particularly important to determine these
relationships at high pressure.
Raman spectroscopy (Fig. 37) has been
used to determine the abundances of indi-
vidual structural units in the quenched melts.
All compositions in this study are on the
join M2Si409 - M2(MA1)409 (M = K, Na and
Li; denoted KS4 - KA4, NS4 - NA4 and
LS4 - LA4), with, therefore, NBO/T = 0.5
(see Fig. 36). The unit distributions in
quenched melts probably reflect those near
the glass transition temperature (T ). One-
bar samples were formed by quenching at
~500°C/s in vertical quench-furnaces,
whereas high-pressure samples (to 30kbar)
were quenched at ~100°C/s in the solid-
media, high-pressure apparatus. The abun-
dances have been determined from calibra-
tion of area ratios of (Si,Al) - O stretch
bands in the high-frequency envelopes of
GEOPHYSICAL LABORATORY
49
1 bar
0.80 -
T3 °-60
g 0.40 t-
0.30
0.00
0.10 0.20
AI/(AI+Si)
0.30
0.00 0.10 0.20
AI/(AI+Si)
0.30
1UU
/\ AI/(AI+Si)=0.15
^
o^
/ \ NBO/T=0.5
>* „,.
■t= 75
co
c
CD
A4+A5
■- 50
T5
CD
N
| 25
A2 J
/ U3
o
A1 ->V
■ ••^' >/\
2 0
mrf^V' "
'■\''0->. v
850 975 1100 1225 1350
Wavenumber, cm'1
Fig. 37. Relationship between area ratios (as indicated) from Raman spectra of quenched melts on the
join KjS^O, - K2(KA1)409 at 1 bar, 12 and 30 kbar. The relevant portion of the high-frequency envelope
of a typical Raman spectrum with compositional variables as indicated is also shown.
the spectra (Fig. 37). In order to obtain the
exact frequencies and areas of individual
Raman bands, the spectra were fitted statis-
tically with the method described by My sen
et al. (1982). The Al is the area of the (Si,Al)
- O antisymmetric stretch band from T03
units and A3 from T205 units. The remaining
bands, against which these areas are nor-
malized, are from bridging oxygen bonds.
The relative errors (from the fitting proce-
dure) in these determinations are between
10 and 20%.
The 1-bar relationship between unit
abundances and A1/(A1 + Si) is shown in
Fig. 38. Substitution of charge-balanced
Al3* for Si4* results in a systematic lowering
of the abundance of T205 units together
with a concomitant increase in the more po-
50
CARNEGIE INSTITUTION
0.8
2 0.6 TOiJQl
T205 (Q3) 1 A; LS4-LA4
T03 (Q2)
0.1 0.2
AI/(AI+Si)
0.8
.2 0.6
+- •
o
0$ 0.4
O 0.2
0,0
TO- (Q4)
i —
(Q3)
i
B
NS4-NA4 •
'TOajQfL-^
0.1 0.2 0.3 0.4
AI/(AI+Si)
0.5
0.1 0.2
AI/(AI+Si)
0.3
Fig. 38. Mol fraction of structural units in 1-bar,
temperature-quenched melts as a function of Al/
(Al + Si). A - The join LLSi409 - LL(LiAl)409 (LS4
- LA4), B - The join Na2Si409 - Na^aAlJA (NS4
- NA4), C - The join ICSi409 - IC(KA1)409 (KS4 -
KA4).
lymerized tinit, T02 and the less polymer-
ized one, T03. The overall degree of po-
lymerization of the melts remains constant
(NBO/T = 0.5) in this process. Thus, quali-
tatively, increasing A1/(A1 + Si) shifts equa-
tion (1) to the right. It is notable, however,
that even in the absence of Al3+ the larger
the ionization potential of the network-
modifying cation (Li > Na > K), the mol
fractions, X(T02) and X(T03), are greater,
and the X(T205) is smaller. This observa-
50
40
CD
O
CD
Q.
30
c
o
CO
20
CD
Q.
CO
b
ion
T02;Q4
0.0 0.1 0.2 0.3 0.4 0.5
AI/(AI+Si)
20 r
T205;Q3
8 -20
w.
CD
CL
x\ \
- -40
c
o
Li
o -60
\
CL
CO
Q -80
N. Na
-mn
I ■
— i
0.0 0.1 0.2 0.3
AI/(AI+Si)
0.4
-200
0.0 0.1
0.2 0.3 0.4
AI/(AI+Si)
0.5
Fig. 39. Dispersion of structural units (as indicated
on the panels) as a function of A1/(A1 + Si) and the
type of network-modifying and charge-balancing
cation.
GEOPHYSICAL LABORATORY
51
0.80 r
O 0.60-
10.0 r
n^^s
AI/(AI+Si)
0.80 r
g 0.60
"«£■?
o
£: 0.40 r
0.20 -
12kbar
T02 (Q4)
TCb(tf)
0.00
0.0
0.80
§ 0.60
cO « ,A
£: 0.40 -
0.20 -
0.00
0.1 0.2
AI/(AI+Si)
0.3
30 kbar
T02 (Q4)
T03 (Q2)
T205 (OP)
o.o
0.1 0.2
AI/(AI+Si)
0.3
Fig. 40. Mol fraction of structural units in K^S^O,
- K^KAl^melts as a function of A1/(A1 +Si) at
the pressures indicated.
tion accords with results from 29Si NMR
spectroscopy of analogous melt composi-
tions (Stebbins, 1987).
For the compositions studied here, the
alkali metals also serve to charge-balance
Al3+ in tetrahedral coordination. It is evi-
-40.0
0.00
0.10 0.20
AI/(AI+Si)
0.30
10.0r
-40.0
0.00 0.10 0.20 0.30
AI/(AI+Si)
Fig. 41 . Free energy (AG) for equilibrium ( 1) at the
glass transition temperature (AGTg) and at 273K
{AG™) for melts on the join K^O, - K^KAl)^
as a function of pressure as indicated.
dent that the influence of Al/Al + Si) on
equation (1) at 1 bar also depends on the
charge-balancing cation (Fig. 39). The
increase (or decrease) in mol fractions of
the coexisting structural units, relative to
the abundance in the Al-free end-member
silicate melts is termed dispersion. The dis-
persion is more pronounced (K > Na > Li)
the smaller the ionization potential of the
cation.
The distribution of structural units as a
function of A1/(A1 + Si) follows the same
general trends at high pressure (Fig. 40) as
52
CARNEGIE INSTITUTION
at 1 bar (Fig. 38) at least for the potassium
tetra-aluminosilicate quenched melts.
Qualitatively, from the abundance infor-
mation in Fig. 40, at a given A1/(A1 + Si),
the X(T205) tends to be lower at higher
pressure, thus indicating a shift of equation
(1) to the right. Such a trend has also been
suggested for Al-free KS4 quenched melts
reported by Dickinson et al. (1985).
Under the assumption of ideal mixing
of the structural units, the free energy of
reaction (1) can be calculated (Fig. 41).
Although the viscosity of these melts proba-
bly decreases with increasing pressure
(Kushiro, 1976), and, therefore, the glass
transition temperature probably also de-
creases (e.g., Rosenhauer et al., 1979).
This effect has not been taken into account
as the exact relationship between pressure
and T is not known for these melts. Also
shown in Fig. 41 are the free energy data at
273K calculated with the AS-value (5 J/mol
K) suggested by Stebbins (1989). In light
of available high-temperature calorimetric
data (see Richet and Bottinga, 1986, for
summary), temperature-dependence of ACp
for equation (1) is not considered.
It is clear (Fig. 41) that not only does the
AG decrease with increasing A1/(A1 + Si) at
constant pressure, but increasing pressure
generally results in a further decrease in AG
for all aluminum contents studied. The
pressure effect on AG for reaction (1) is
further enhanced as a function of increas-
ing aluminum content. Thus, as suggested
from density measurements on fully po-
lymerized (NBO/T = 0) alkali aluminosili-
cate melts (e.g., Kushiro, 1980), depolym-
erized (NBO/T > 0) aluminosilicate melts
become increasingly compressible as Al
20 r
10
o
F
c^
0
E
o
-10
>
<
-20
-30
30 kbar
J i I i L
J i I
0.0
0.1
0.2
3+
AI/(AI+Si)
Fig. 42. Volume change, AV, for equilibrium (1) at
30 kbar as a function of A1/(A1 + Si) for melts on
the join K^i fi9 - ^(KAl^O,.
is substituted for Si4+ (Fig. 42) at least for
the potassium tetra-aluminosilicate com-
positions.
The shift of reaction (1) to the right with
increasing ionization potential of the net-
work-modifying cation probably is a result
of local electric charge constraints. That is,
geometric constraints increasingly tend to
prevent neutralization of the negative charge
in the anionic network the smaller the net-
work-modifying metal cation. This prob-
lem compounds as theNBO/Si of a particu-
lar structural unit decreases. Notably, among
crystalline alkali silicates, only potassium
forms a stable tetrasilicate. Potassium and
sodium form stable disilicate crystalline
materials, whereas no lithium disilicate is
known. In melts where more than one de-
polymerized structural unit is present, the
smaller cations will prefer the least polym-
erized unit, and, thus, drive, reactions such
as (1) to the right.
The aluminum distribution between the
structural units is governed by the differ-
ences in intertetrahedral angles (a) be-
GEOPHYSICAL LABORATORY
53
tween the various units, and will favor the
unit with the smallest intertetrahedral angle
(Mysen et al., 1985). As a(T02) < a(T205)
< a(T03) (Furukawaef a/., 1981),Al3+will
then substitute preferentially for Si4+ in the
T02 units, thus driving equation (1) to the
right.
Finally, because the intertetrahedral
angle in fully polymerized structural units
depends on pressure (Seifertet al., 1983), it
is suggested that the compressibility of the
individual coexisting units in melts govern
the influence of pressure on reaction (1).
Among the coexisting units in tetrasilicate
and tetra-alumino silicate melts, the T02
structure is much more compressible than
either the T205 or the T03 (Bockris and
Kojonen, 1960). These compressibility
relations have two consequences. First,
because (dV/dP)T < 0 even for Al-free sili-
cate melts, equation (1) shifts to the right.
Second, by substituting charge-balanced
Al3+ for Si4+, the X(T02) increases, thus
increasing the bulk compressibility. Fur-
thermore, the A1/(A1 + Si) in the T02 units
increases. This increase lengthens the
(Si,Al)-0 bridging bonds, and the oc(T02)
becomes smaller. This angle is expected to
be more compressible. All these factors
serve to enhance the bulk melt compressi-
bility. Thus, the observed enhancement of
bulk melt compressibility with increasing
A1/(A1 + Si) would be expected.
The pressure-composition relationships
provides a rationale to understand the be-
havior of natural magmatic liquids under
pressure, (i) As indicated in Fig. 36, mag-
matic liquids become more polymerized
with increasing silica, or alumina, or both.
Thus, the relative abundance of T02 units
in the melts increases. As a result, the molar
volume and the melt compressibility of
magmatic liquids increase in the order
komatiite < basalt < andesite < rhyolite. (ii)
For basaltic melts, high-aluminum basalt is
more compressible than tholeiite even
though their NBOITdst practically the same
(Mysen, 1988). This behavior is the result
of the higher A1/(A1 + Si) in the high-
alumina basalt as compared with tholeiite.
This higher A1/(A1 + Si) shifts reaction (1)
to the right thus enhancing the melt com-
pressibility, (iii) The bulk composition of
alkali basalt and tholeiite differs princi-
pally in the values of the average ionization
potential (Z/r) of the charge-balancing metal
cations. The (Z/r)a]kalibasalt < (Z/r\olciisc with the
result that AG of reaction (1) for alkali
basalt is smaller (more negative) resulting
in greater compressibility. Relationships
similar to (i-iii) would also be expected for
all other melt properties that depend on the
abundance and character of bridging oxy-
gen bonds in T02 units in the magmatic
liquids.
References
Bockris, J. O., and F. Kojonen, The compressibil-
ity of certain molten alkali silicates and borates,
/. Am. Chem. Soc., 82, 4493-4497, 1960.
Brandriss, M. E., and J. F. Stebbins, Effects of
temperature on the structures of silicate liquids:
29Si NMR results, Geochim. Cosmochim. Acta,
52, 2659-2669, 1988.
Dickinson, J. E., and C. M. Scarfe, Pressure-
induced structural changes in K^S^O, silicate
melt, EOS, 66, 395, 1985.
Furukawa, T., K. E. Fox, and W. B. White, Raman
spectroscopic investigation of the structure of
54
CARNEGIE INSTITUTION
silicate glasses. IE. Raman intensities and struc-
tural units in sodium silicate glasses, /. Chem.
Phys., 153, 3226-3237, 1981.
Kushiro, I., Changes in viscosity and structure of
melt of NaAlS^Og composition at high pressures,
/. Geophys. Res., 81, 6347-6350, 1976.
Kushiro, I., Viscosity, density, and structure of
silicate melts at high pressures, and their petrol-
ogical implications, in Physics of Magmatic
Processes, R. B. Hargraves, ed., Princeton Uni-
versity Press, Princeton, Ch. 3, 1980.
Mysen, B. O., Structure and Properties of Silicate
Melts, Elsevier, Amsterdam, 354 pp., 1988.
Mysen, B. O., L. W. Finger, F. A. Seifert, and D.
Virgo, Curve-fitting of Raman spectra of amor-
phous materials, Am. Mineral., 67, 686-696,
1982.
Mysen, B. O., D. Virgo, and F. A. Seifert, Rela-
tionships between properties and structure of
aluminosilicate melts, Am. Mineral, 70, 834-
847, 1985.
Richet, P., and Y. Bottinga, Thermochemical
properties of silicate glasses and liquids: A
review, Rev. Geophys., 24, 1-26, 1986.
Rosenhauer, M., C. M. Scarfe, and D. Virgo,
Pressure dependence of the glass transition
temperature in glasses of diopside, albite and
sodium trisilicate composition, Carnegie Instn.
Washington, Year Book, 78, 556-559, 1979.
950°C
2kbar
En
Mol Percent
Fo
Fig. 31. Potassium analogue of the extended nor-
mative, basalt tetrahedron Ks - La - Fo - Qz
exhibiting stable joins determined experimentally
at 950°C and 2 kbar.
Ak
akermanite
Lc
leucite
Di
diopside
Mer
merwinite
En
enstatite
Mo
monticellite
Fo
forsterite
Ra
rankinite
K-mel
potassium melilite
Sa
sanidine
Ks
kalsilite
Qz
quartz
La
larnite
Wo
wollastonite
Igneous and Metamorphic Facies of
Potassium-rich Rocks: Coexisting as-
semblages in Kalsilite-Forsterite-
Larnite-Quartz at 950°C and 2 kbar
With and Without H20.
Hatten S. Yoder, Jr.
The wide variety of potassium-rich
igneous and metamorphic rocks appears to
result primarily from the great diversity of
bulk composition and from the large num-
ber of reactions between phases (Yoder,
1986). The mineralogical complexity of
the final products probably results from
incomplete or failed reactions as the mag-
mas cool. For this reason an effort was
made to establish experimentally a stable
array of assemblages with which observed
and alternative assemblages could be
compared.
The principal minerals involved in an-
hydrous potassium-rich rocks are displayed
in Fig. 3 1 . The tetrahedron is the potassium
analogue of the extended basalt tetrahe-
dron (Schairer and Yoder, 1964) wherein
kalsilite replaces nepheline. The joins
connecting coexisting phases were estab-
lished by experiments at 950°C and P=2
kbar of 5 days duration. Stable assem-
blages were determined by reacting natural
GEOPHYSICAL LABORATORY
55
minerals close to endmember composition,
synthetic endmembers, or both, in compat-
ible and incompatible combinations. The
experiments were carried out in an inter-
nally-heated, gas-media, high-pressure
apparatus (Yoder, 1950). Over 60 mixtures
of compatible and incompatible phases were
reacted to establish the stable joins for both
the anhydrous and hydrous tetrahedrons at
isothermal, isobaric conditions. The condi-
tions 950°C and 2 kbar were chosen so that
there was adequate pressure to stabilize the
appropriate hydrous minerals, and at a
sufficiently high temperature to be at or
near the beginning of melting, yet above
the stability of the amphiboles.
To illustrate the strategy in determining
these joins under anhydrous and hydrous
conditions, the join Di (CaMgSi206) - Ks
(KAlSi04) will be used as an example.
(Abbreviations are defined in the caption
of Fig. 31).
The reactions investigated that prove
that Di-Ks is the stable join are:
Fo + 3Lc + 2Ak -> 4Di + 3Ks
(1)
H20
Fo + 3Lc + 2Ak -> 4Di + 3Ks (2)
Ph + 3Ak + 4Lc -> 6Di + 5Ks + HX)(3)
Mo + Lc -> Di + Ks
2Mo + Sa -> 2Di + Ks
h2o
2Mo + Sa -> 2Di + Ks
(4)
(5)
(6)
12Di + 5Y-ALP3 + 5[K202Si02]
-> 12Di + lOKs ' (7)
Dol + Sa -> Di + Ks + 2C02. (8)
The Di-Ks join is critical because it pene-
trates the common assemblage Mel+Fo+Lc
of the katungites. Note that the join is stable
under both hydrous and anhydrous condi-
tions. (The hydrous tetrahedron is presented
below.) Reactions 4-6 show that monticel-
lite is not stable with either Lc or Sa at these
conditions - it is cut by the planes Ak - Ks
- Fo and Di-Ks - Fo. Reaction 7 was a test
of the alkali-loss problem experienced in
the past. To avoid alkali loss, it was usually
necessary to use natural minerals close to
the end members or well characterized
synthetic phases. The last reaction is the
well known metamorphic reaction observed
at Brome Mt., Quebec, and Hendrickspla-
ats, RSA.
In order to appreciate the value of the
tetrahedron and clarify the relationships,
the various subtetrahedra are presented in
an exploded view in Fig. 32. The subtetra-
hedral volumes are labelled with the appro-
priate rock names underlined. The faces
and joins have also been labelled where
rock names have been assigned. (Alkali-
rich rock enthusiasts have, fortunately,
missed naming a few of the joins). Begin-
ning at the left, the kalsilite-rich end, the
principal mineral assemblage for the me-
lilite mafurites, Mel + Ks + Cpx + Fo, is
found. Passing through the mafurites, Ks +
Cpx + Fo, one comes to the leucite mafu-
56
CARNEGIE INSTITUTION
Wo Wo Wo
r"y Hy
Fo Fo Fo
Fig. 32. Exploded view of generalized subtetrahedra in Fig.3 1 with names of closely related rock types
inscribed. Names for four-phase volumes are underlined.
Cpx
clinopyroxene
Mel
melilite
Hy
hypersthene
Ol
olivine
Kf
potassium feldspar
rites. Then, through the ugandites, Lc +
Cpx + Fo, the absarokite assemblage is
next encountered. The adjoining assem-
blages are the cancalites and finally the
chamockites. Illustrated in the subtetrahe-
dra on the top row of Fig. 32 are those
assemblages of monticellite -bearing rocks
related to the kimberlites to be discussed
below. Potassium melilite (K-Mel) did not
form on the join Ks-Wo under either anhy-
drous or hydrous conditions at this pressure
and temperature. The vesbites, common in
the Roman region of Italy, are next exhib-
ited. The two upper subtetrahedra on the
right are very rare metamorphic assem-
blages.
The point to be emphasized is that all of
these rock types may be stable at the same
pressure and temperature under anhydrous
conditions. Thermodynamically, all com-
950°C
PH2o = 2kbar
En
Mol Percent
Fo
Fig. 33. Potassium analogue of the extended nor-
mative basalt tetrahedron Ks - La - Fo - Qz exhib-
iting stable joins determined experimentally at
950°C with PiUfi) = 2 kbar.
Mg-Cel magnesium celadonite
Ph phlogopite
Tr tremolite
See Fig. 31 for definition of abbreviations of
anhydrous phases.
GEOPHYSICAL LABORATORY
57
Hy Hy
Fig. 34. Exploded view of generalized subtetrahedra by Fig. 33 with names of closely related rock types
inscribed. Abbreviations given in Figs. 32 and 33.
positional space must be represented by a
stable or metastable assemblage at every
pressure and temperature. Displayed is the
entire array of stable assemblages, which
describe a consistent set of rock types and
show their compositional relationships.
The hydrous tetrahedron at the same
pressure and temperature is displayed in
Fig. 33. The anhydrous phases are the same.
Phlogopite is the key hydrous phase and
joins radiate out to all the principal phases
except Qz, Wo, and, contrary to natural
occurrences, Mo. Phlogopite is the phase
that appears to provide liquid lines of de-
scent through many of the thermal barriers
in the flow sheet developed for the anhy-
drous potassium-rich lavas (Yoder, 1986).
The theoretical mica, magnesium celadonite
(Mg-Cel), lies on the join Sa - En, but it did
not form under the applied conditions. The
amphibole tremolite (Tr) lies on the face Di
- En - Qz. It is stable only up to 890°C at 2
kbar, and was, therefore, not encountered.
On the other hand, tremolite was a useful
starting material because joins with other
phases penetrated a large number of other
planes.
The significance of these stable assem-
blages, all proven experimentally, can be
appreciated in an exploded view of the
individual subtetrahedra in the hydrous
tetrahedron in Fig. 34. Only six of the
previous subtetrahedra in the anhydrous
tetrahedron persist. Note again that Mo is
cut off from Lc and Sa, whereas Ak is stable
with Lc, but not Sa, as observed in natural
rocks. The principal mineral assemblages
are assigned the names of the most repre-
sentative rock types. The Italian and Afri-
can provinces are represented in the upper
left two subtetrahedra. One may begin a
review of the array of assemblages with the
okaites and katungites. These adjoin the
alnoites, a major member of the lampro-
58
CARNEGIE INSTITUTION
phyres. After passing through the verites,
one comes to the mica peridotites. These
rocks are contiguous with the lamproites,
represented by the jumillites and orendites.
The lamproites in turn adjoin sequentially
the basic and acid charnockites. The in-
credible array of rock types may be stable
at the same pressure and temperature. It is
not necessary, therefore, to assign special,
exotic pressure and temperature conditions
to explain the assemblages represented by
kimberlite, lamprophyres, lamproites or
charnockites. The critical question to be
resolved is how the large range of bulk
composition was generated.
Special note should be made of the
absence of the join Mo - Phi so critical to
the Group I variety of kimberlite. The planes
Ak-Ks-XandAk-Fo-XcuttheMo-Phl
join. The X represents a K - Mg-rich fluid,
not yet adequately characterized, that is
required to account for mass balance. The
composition X is of exceptional impor-
tance because there appears to be a fluid
available for metasomatism that is comple-
mentary to phlogopite. In other words,
phlogopite is an indicator that a metasoma-
tizing solution was produced and is not
itself a product of that solution. From this
viewpoint, phlogopite may form at the same
time the metasomatizing fluid is generated;
for example, by reaction of Fo+Ks with
H20. It will be illuminating to reexamine
the textural relations of phlogopite in kim-
berlites with this concept in mind.
These results are for exactly one pres-
sure, 2 kbar, and one temperature, 950°C. If
the temperature is raised or lowered 100°C,
another ensemble of assemblages will be
generated. In addition, a modest pressure
Ks
950°C
PH20=2kbar
Fo
En
Weight Percent
Fig. 35. The system Ks - Fo - Qz at 950°C and
PQlfi) = 1 kbar. Abbreviations as in Figs. 32 and
33. L = liquid.
increase of 5-10 kbar will eliminate me-
lilite, monticellite, and leucite — some of
the principal phases in these assemblages.
Melting was not observed in the anhy-
drous tetrahedron, but melting was ob-
served in the charnockite subtetrahedra
under hydrous conditions (Fig. 35). Only
the base plane of the tetrahedron is dis-
played to reduce the complexity. Melting
begins near the alaskite join previously
investigated by Shaw (1963). All of the
major phases except Fo and Ks are on the
liquidus. The significance of these obser-
vations is that the charnockites and related
migmatites may be partially melted while
all the other rock types will be in a crystal-
line state. These results suggest that mig-
matites, formed in several ways, do not
necessarily imply a hot spot — difficult to
explain thermally — but may indicate a
regional temperature rise where all the
surrounding related rock types may remain
stable in the solid state. For this reason it
GEOPHYSICAL LABORATORY
59
will be necessary in the future to relate
igneous facies with ultrametamorphic fa-
des for the entire array of compositions in
order to characterize the pressure and
temperature.
The results provide a consistent en-
semble of assemblages from which com-
patible and incompatible mineral assem-
blages in K-rich rocks can be determined.
The complexity of the alkaline rocks ap-
pears to result primarily from variations in
bulk composition, however the large num-
ber of potential reactions between phases
(Yoder, 1986) contribute to further com-
plexity when the successive reactions are
incomplete. Because of the large number of
phases in alkaline rocks, the textural rela-
tions (e.g., rimming, morphology, and
zoning) may be of great importance in
revealing the sequence of reactions.
References
Schairer, J. F., and H. S. Yoder, Jr., Crystal and
liquid trends in simplified alkali basalts, Carne-
gie Instn. Washington Year Book, 63, 65-74,
1964.
Shaw, H. R., The four-phase curve sanidine -
quartz -1 iquid - gas between 500 and 4000 bars,
Am. Mineral., 48, 883-896, 1963.
Yoder, H. S., Jr., High-low quartz inversion up to
10,000 bars, Trans. Am. Geophys. Union, 31,
827-835, 1950.
Yoder, H. S., Jr., Potassium-rich rocks: Phase
analyses and heteromorphic relations, J. Petrol.,
27, 1215-1228, 1986.
Techniques for Experimentally Loading
and Analyzing Gases and Their Appli-
cation to Synthetic Fluid Inclusions
John D. Frantz, Yi-gang Zhang,
Donald D. Hickmott, and
Thomas C. Hoering
Experimental studies of equilibria be-
tween minerals and mixed-volatiles under
hydrothermal conditions have long been a
research focus for many experimentalists.
One of the principal difficulties in experi-
mentally investigating reactions between
minerals and mixed volatiles under hy-
drothermal conditions is loading the ex-
perimental charges with gas mixtures of
known composition. In the past, gas load-
ing was accomplished primarily by the
addition of solid compounds that produced
volatiles in the experimental charge. For
example, silver oxalate was used as a source
of carbon dioxide (Holloway etal., 1968);
iridium carbonyl, to produce carbon mon-
oxide (Eggler et al., 1979); oxalic acid, as
a source of carbon dioxide, water, and
hydrogen (Holloway and Reese, 1974);
chromium nitride, to produce nitrogen or
ammonia if used in conjunction with an
external oxygen buffer (Hallam and Eug-
ster, 1976); acetic acid, as a source 6f
carbon dioxide and methane (Seitz et al,
1987; Palmer and Drummond, 1986); and
graphite to produce carbon dioxide, carbon
monoxide, and methane when used in
conjunction with an external oxygen buffer
(French, 1966; Eugster and Skippen, 1967).
The applications of these procedures have
been instrumental in many studies impor-
tant to our understanding of equilibria be-
60
CARNEGIE INSTITUTION
K
VACUUM
Details of I
Fig. 43. Gas pipette apparatus. See text for details.
tween metamorphic mineral assemblages
and mixed volatiles (Skippen, 1971; Jacobs
and Kerrick, 1981).
The use of these compounds, however,
has limitations. For example, at tempera-
tures below 700°C, graphite reacts slowly
with C-O-H gases (Ziegenbein and Johan-
nes, 1980). Below 600°C, the breakdown
of acetic acid is sluggish (Palmer and
Drummond, 1986). Chromium nitride does
not decompose completely under hydroth-
ermal conditions and thus cannot be used
quantitatively to add fixed amounts of gas
to an experimental charge. Compounds such
as oxalic acid, ammonium oxalate, and
ammonium nitrate decompose to yield
unwanted gas species such as hydrogen or
oxygen (Holloway and Reese, 1974).
Continuously variable gas compositions in
ternary and quaternary gas mixtures such
as the C-O-H and C-O-H-N systems are
difficult to achieve by the addition of stoi-
chiometric compounds. For research in-
volving mixed volatiles in synthetic fluid
inclusions, a relatively fixed ( and known)
distribution of gas species must be achieved
shortly after the experimental temperature
and pressure is reached because inclusions
form rapidly, trapping the current fluid
compositions. Seitz et al. (1987) demon-
strated a large variability in the carbon
dioxide-methane ratio for different inclu-
sions in the same sample. The synthetic
fluid inclusion sample contained carbon
dioxide, methane, and water generated by
the addition of acetic acid and water to the
charge. Due to these limitations, develop-
ment of a new technique for loading mixed
GEOPHYSICAL LABORATORY
61
volatiles in hydrothermal experiments was
desirable.
In order to load volatiles quantitatively
in experimental hydrothermal charges, a
gas pipetting apparatus was constructed of
Pyrex glass shown in Fig. 43. The system
consists of a gas storage section (separated
from the rest of the system by valves A, D,
and F), a measured gas aliquot section
(between valves F and //), a vacuum
manifold K with a vacuum gauge, and the
sample holder/. Gas is first introduced into
the gas storage section that consists of a
2000 ml flask C, a 100 ml flask £, and a
mercury manometer E. The small flask can
be used to purify gases, such as C02, which
have relatively high condensation tempera-
tures of gas impurities with lower conden-
sation temperatures such as methane, nitro-
gen, and oxygen. This is accomplished by
immersing the small flask in liquid nitro-
gen and evacuating the system by opening
valve D and exposing the storage section to
the vacuum manifold. In the case of load-
ing gases with low condensation tempera-
tures, gas impurities having high conden-
sation temperatures can be condensed into
the cold finger G by immersing it in liquid
nitrogen bath with valve F open. Valve F is
then closed, the liquid nitrogen bath is
removed, and the gas impurities in the gas
aliquot section are evacuated. With valves
A, D, and H closed and F opened, a small
quantity of gas can be introduced into the
gas aliquot section. By having previously
evacuated this section (using valves H and
7), the pressure of gas introduced into this
section can be determined with the second
manometer and a cathetometer. After evacu-
ating the sample holder section, the gas
aliquot section can then be opened to the
sample holder section by opening valve H
(valve / closed). The sample holder con-
sists of a 1/2" Swagelock™1 union (L) with
Teflon ferrules (to connect the holder to the
1/2" glass tubing used in the pipette appa-
ratus), a valve (M), a 1/4" Swagelock™
union with Teflon ferrules (N), and a noble-
metal capsule with the bottom end welded
(0). Electronic pressure transducers could
be used to replace the above-mentioned
mercury manometers.
The volume of the gas aliquot section
was calibrated by connecting a glass bulb
fitted with a stopcock. The volume of the
bulb has been determined accurately by
weighing it when filled with mercury and
subtracting the weight of the empty volu-
metric. Then, carbon dioxide gas is al-
lowed to enter both the gas aliquot and
sample holder sections with the valve on
the calibration volumetric open. After the
pressure is measured, the valve on the cali-
bration bulb is closed and both sections are
evacuated through the vacuum manifold.
Next, the valve on the calibrated volumet-
ric is opened and all the carbon dioxide is
transferred into the gas aliquot section by
immersing the cold finger (G) into liquid
nitrogen. After closing valve H and allow-
ing the carbon dioxide to warm to room
temperature, the pressure on the manome-
ter is again measured. Because we know
the volume and the gas pressure that ex-
isted in the calibrated volume and we know
the final gas pressure of the same quantity
of gas in the gas pipette section alone, the
volume and the number of moles of gas
contained in the gas pipette section can be
computed for that particular pressure using
62
CARNEGIE INSTITUTION
D)
X
E
E.
Z3
</)
o
Q.
(0
>
Moles (x10"4)
ono _
2
i
4 6 8
i i <
10
i
^uu
Ny
(a)
100 -
/ CH4.N2
"*"^*^'^7
^ CH4
COo
r •
0 10 20 30 40
Total Pressure (mm Hg)
Moles (x10"4)
o
Q_
05 0 20 40 60 80 10
Total Pressure (mm Hg)
Fig. 44. (a), (b) Vapor pressure of nitrogen, meth-
ane, carbon dioxide, and 50-50 mole% methane-
carbon dioxide mixture above 38 mg silica gel as
a function of the pressure and number of moles of
the gases in the measured gas aliquot section of the
gas loading apparatus.
the ideal gas law. This calibration was done
at a series of pressures because the volume
of the manometer contributes to the vol-
ume of the gas pipette section.
A quantity of gas in the gas aliquot
section is transferred to the experimental
capsule by opening valve H to the evacu-
ated sample holder section and drawing the
gas into the capsule by immersing it in
liquid nitrogen. In the case of carbon diox-
ide, the resultant vapor pressure above the
frozen gas is less than lxlO"4 mm mercury
at -195.8°C and the transfer is complete. In
the case of other gases, there may be sig-
nificant vaporpressure above the condensed
gas at liquid nitrogen temperatures. It is
well known that solid compounds with
pore structures absorb gases strongly at
low (liquid nitrogen) temperatures. When
dealing with gases having significant va-
por pressures at liquid nitrogen tempera-
tures, the adsorbtion by porous solids is
advantageous since substantial vapor pres-
sures above the condensed gases at liquid
nitrogen temperatures introduce appre-
ciable error. A gel, zeolite, or some other
compound that adsorbs gases at cryogenic
temperatures is placed in the capsule with
the charge. Due to the properties of adsorp-
tion of these compounds, the resultant vapor
pressures are quite reduced. In Fig. 44a, the
vapor pressures above 38 mg of silica gel
for nitrogen, methane, carbon dioxide, and
methane -nitrogen mixtures are plotted as a
function of both the total pressure and the
number of moles of the gas or gas mixture
contained in the measured gas aliquot sec-
tion of our apparatus (Fig. 43). In the case
of nitrogen, for example, the vapor pres-
sure of liquid nitrogen is 760 mm mercury
and quantitative transfer without the use of
a gas adsorbing compound is not possible.
When nitrogen is adsorbed, however, the
vapor pressure is only 6 mm Hg (using 38
mg gel) for 0.0001 mol of the gas (Fig.
44b). In the case of methane, the vapor
pressure is even lower. A vaporpressure of
less than 1 mm Hg occurs for 0.001 mol of
methane, allowing the addition of precisely
known quantities of the gas. Addition of
more gel will result in lower vapor pres-
GEOPHYSICAL LABORATORY
63
f~\
Au
Capsule
Silica
Gel
Quartz
Prism
Pt Capsule
Water
Fig. 45. Experimental capsule detail.
sures and the possibility of adding a larger
quantity of gas. This technique can be used
for any mixture such as carbon dioxide +
methane or carbon dioxide + methane +
nitrogen containing a non-condensed gas.
The use of an adsorbing compound is vital
in this case because significant partitioning
between the condensed and solidified gases
and the coexisting vapor would occur with
mixtures comprised of gases with varying
vapor pressures at liquid nitrogen tempera-
ture. With respect to C02 - CH4 mixtures,
best results were obtained by first loading
pure carbon dioxide into the capsule. By
keeping the capsule at liquid nitrogen
temperature, pure methane can then be
added. Carbon dioxide appears to freeze
before being adsorbed by the gel and thus
the capacity of the gel for methane is not
reduced by the presence of carbon dioxide.
In fact, the resultant vapor pressure is even
somewhat less than in the case of pure
methane possibly indicating adsorption of
methane on solid carbon dioxide.
Water, however, cannot be directly added
to the capsule as it may saturate the porous
solid. To overcome this problem, a length
of 2 or 3 mm platinum tubing is completely
filled with water and a section correspond-
ing to the length containing the desired
quantity of water is cut and cold welded on
both ends using a pinch-off device
(Komarneni et ai, 1979). This inner cap-
sule is placed in the larger capsule. After
the loading procedure is complete, the
portion of the capsule containing the ex-
perimental charge is cut and cold welded
using the pinch-off device and, while the
capsule remains in the liquid nitrogen, is
electrically welded. A vacuum-tight cold
weld is essential as a large amount of air
can be drawn into the capsule by the ab-
sorbing solid before the capsule is electri-
cally welded. The capsule is then removed
from the nitrogen, checked for leaks in
water, and placed in the hydrothermal pres-
sure vessel. Such a welded capsule can
safely hold 100 bar of pressure. Upon
heating, the absorbing solid releases the
gases and the inner platinum capsule con-
taining water ruptures, releasing water,
producing a mixed volatile fluid of known
composition.
A major application of this technique
lies in the experimental study of the prop-
erties of hydrothermal fluids containing
mixed volatiles using synthetic fluid inclu-
sions. In an effort to both test the loading
technique described above and to initiate a
thorough study of the C-O-H system (dis-
cussed in Zhang and Frantz, this Report),
synthetic fluid inclusions with fluids con-
taining mixtures of carbon dioxide, meth-
ane, and water were grown. Fractured quartz
64
CARNEGIE INSTITUTION
Table 5. Analyses by gas chromatography
Initial Gas Composition
Wt% Mol Mol Mol Mol Ratio
CH4 CH4 CQ2 Hp CH4/CQ2
Measured Gas Composition
Mol Mol Mol Mol Ratio
CH4 CQ2 Hp CiyC02
5 0.000098
10 0.000143
15 0.000145
20 0.000169
25 0.000152
0.000175 0.001221 0.560
0.000124 0.000843 1.153
0.000083 0.000537 1.747
0.000068 0.000442 2.485
0.000047 0.000298 3.234
0.000086 0.000169 -na- 0.509
0.000131 0.000122 0.0008995 1.074
0.000129 0.000079 0.0005386 1.633
0.000176 0.000066 -na- 2.667
0.000155 0.000047 0.0002778 3.298
prisms, cold welded platinum capsules
containing water, and approximately 75
mg silica gel were placed in thick-walled
gold capsules (4.75 mm O.D.; 4.0 mm I.D.)
(Fig. 45). The gas pipette was used to load
the methane and carbon dioxide. Five dif-
ferent bulk compositions containing 5,10,
15, 20, and 25 wt% methane were chosen
with wt% ratios for CO JYL£> of 0.357. The
experiments were placed in standard hy-
drothermal pressure vessels for four days at
600°C and 2000 bar. In order to assure
thorough mixing of water and the gases in
the tiny cracks of the quartz prism, the
pressure vessel was heated at 1000 bar.
After reaching the experimental tempera-
ture, the vessel was cycled four or five
times between 500 and 3000 bar. Further
discussion of the synthetic fluid inclusion
techniques and the hydrothermal proce-
dures employed in this study are described
in Bodnar and Sterner (1987) and Zhang
and Frantz (1987, 1989). An Accuspec™
Compac II gas chromatograph was modi-
fied to accept gases released from the gold
hydrothermal capsules. The results of these
measurements are tabulated in Table 5 . The
measured number of moles of methane and
carbon dioxide are computed based on the
measured area as referenced to the least-
squares fit of all the data and thus only
reflect the internal consistency of the
measurements. The mole ratios, however,
depend only on the ratio of the measured
areas and the instrument calibration using
the above mentioned gas standards. The
measured values for water were determined
by weighing the capsule before puncturing,
drying it in a vacuum oven (120°C) after
puncturing, re weighing the capsule, and
computing the number of moles of H20
after subtracting the measured weight of
the carbon dioxide and methane. The close
agreement of these mole ratios between the
initial composition and the measured
composition demonstrates the precision of
both our gas loading and the gas chroma-
tography techniques.
A new gas loading technique has been
described by which gases that are non-
condensible at room temperature and pres-
sure can be loaded into experimental
charges. The method has been successfully
demonstrated by growing synthetic fluid
GEOPHYSICAL LABORATORY
65
inclusions in C-O-H fluids at high tempera-
tures and pressures analyzing the gases in
the capsule by gas chromatography. Ex-
perimental studies of mineral-fluid equili-
bria and other studies involving mixed
volatiles at hydrothermal temperatures and
pressures should benefit from the develop-
ment of the gas pipette loading technique.
References
Bodnar, R. J., and S. M. Sterner, Synthetic fluid
inclusions, in Hydrothermal ExperimentalTech-
niques, G. C. Ulmer and H. L. Barnes, eds., John
Wiley & Son, New York, pp. 423-458, 1987.
Eggler, D. H., B. O. Mysen, T. C. Hoering, and J.
R. Holloway, The solubility of carbon monox-
ide in silicate melts at high pressures and its
effect on silicate phase relations, Earth and
Planetary Sci. Lett., 43, 321-330, 1979.
Eugster, H. P., and G. B. Skippen, Igneous and
metamorphic reactions involving gas equilibria,
in Researches in Geochemistry, 2, P. H. Abel-
son, ed., John Wiley and Sons, New York, pp.
492-520, 1967.
French, B. M., Some geological implications of
equilibrium between graphite and a C-H-O gas
phase at high temperatures and pressures, Rev.
Geophys., 4, 223-253, 1966.
Hallam, M., and H. P. Eugster, Ammonium sili-
cate stability relations, Contrib. Mineral. Pet-
rol, 57, 227-244, 1976.
Holloway, J. R., C. W. Burnham, and G. L. Mill-
hollen, Generation of I^O-CX^ mixtures for use
in hydrothermal experimentation, /. Geophys.
Res., 73, 6598-6600, 1968.
Holloway, J. R., and R. L. Reese, The generation
of N^CX^-Hp fluids for use in hydrothermal
experimentation I. Experimental method and
equilibrium calculations in the C-O-H-N system,
Am. Mineral., 59, 587-597, 1974.
Jacobs, G. K., and D. M. Kerrick, Devolatilization
equilibria in H20 - C02 and H^O - C02 - NaCl
fluids: An experimental and thermodynamic
evaluation at elevated pressures and
temperatures, Am. Mineral., 66, 1135-1153,
1981.
Komarneni, S., W. P. Freeborn, and C. A. Smith,
Simple cold- weld sealing of noble-metal tubes,
Am. Mineral., 64, 650-651, 1979.
Palmer, D. A., and S. E. Drummond, Thermal
decarboxylation of acetate. Part I. The kinetics
and mechanism of reaction in aqueous solution,
Geochim. Cosmochim.Acta,50, 813-823, 1986.
Seitz, J. C, J. D. Pasteris, and B. Wopenka, Char-
acterization of C02- CH4- I^O fluid inclusions
by microthermometry and laser Raman micro-
probe spectroscopy: Inferences for clathrate and
fluid equilibria, Geochim. Cosmochim. Acta,
51, 1651-1663, 1987.
Skippen, G. B., Experimental data for reactions in
siliceous marbles, /. Geol., 70, 457-481, 1971.
Zhang, Y. G., and J. D. Frantz, Determination of
the homogenization temperatures and densities
of supercritical fluids in the system NaCl - KC1
- CaC^ - H^O using synthetic fluid inclusions,
Chem. Geol, 64, 335-350, 1987.
Zhang, Y. G., and J. D. Frantz, Experimental
determination of the compositional limits of
immiscibility in the system CaC^ - C02 - Hfi at
high temperatures and pressures using synthetic
fluidinclusions,C/^m.Geo/.,74,289-308,1989.
Ziegenbein, D., and W. Johannes, Graphite in C-
O-H fluids: an unsuitable compound to buffer
fluid composition at temperatures up to 700°C,
N. 3b. Miner. Abh., 7, 289-305, 1980.
Investigations of Fluid Properties in
the CO,-CH -H,0 System using syn-
2 4 2
THETIC FLUID INCLUSIONS
Yi-gang Zhang and John D. Frantz
High-temperature, high-pressure inter-
granular fluids in the Earth's mantle and
crust have had a profound influence on the
evolution and resulting mineral pedogene-
sis of igneous and metamorphic rock suites.
These fluids, mixtures of gases, water, and
dissolved electrolytes, exist either as a
66
CARNEGIE INSTITUTION
supercritical phase or possibly as a mixture
of two immiscible phases (Zhang and
Frantz, 1989). Of particular importance is
the ubiquitous C-O-H system. Its presence
has a major influence on the genesis of
primary magma in the mantle and subse-
quent metasomatism resulting from ele-
ment partitioning between the melt and the
fluid (Green etal, 1987). Calculation of the
oxidation state of the mantle involves
multicomponent equilibria between the
mineral phases and C-O-H fluids (Saxena,
1989). In the Earth's crust, element parti-
tioning and material transport within C-O-
H fluids containing dissolved electrolytes
have resulted in the metasomatism of meta-
morphic mineral assemblages and the for-
mation of ore deposits (Hollister and Burrus,
1976; Wintsche et. al. 1981; Ramboz et.
al.y 1985). Vestiges of ancient C-O-H fluids
are commonly found in the incorporation
of these components in minerals and in
their presence in natural fluid inclusions.
The latter represent a unique opportunity to
1) detect the composition of the fugitive
volatile phase; 2) to determine the tempera-
ture-pressure conditions of the formation
of the surrounding mineral; and 3) to un-
derstand the evolution of the associated
rock suites. Petrologists routinely measure
properties of fluid inclusions such as the
homogenization temperature along the liq-
uid-vapor curve and, assuming that the
volume of the inclusion remains constant
with increasing temperature and pressure,
use these measurements to delineate the
possible temperature-pressure conditions
of formation. Measurements such as these
require knowledge of both the composi-
tions of the inclusions and the PVT proper-
ties of the fluids corresponding to those
particular compositions.
Due to its obvious importance, the C-O-
H has received considerable attention in
both experimental and theoretical investi-
gations. For the most part, the studies have
been concerned with one of the three bina-
ries: C02 - Hp, CH4 - H20, and C02 - CH4.
In the case of the C02 - H20 system, experi-
mental work at high temperatures and pres-
sures has been performed by Franck and
Todheide (1959), Todheide and Franck
(1963), Takenouchi and Kennedy (1964),
Greenwood (1969, 1973), and Schmulov-
ich (1980). The CH4 - H20 binary has
received less attention with the work of
Welsch (1973) being the principle high-
temperature, high-pressure study. The C02
- CH4 binary system, important to the chemi-
cal industry, has attracted geologists ' atten-
tion as fluids of this composition have been
discovered in natural fluid inclusions
(S wanenberg, 1 979). The system along with
previous experimental work is discussed in
Swanenberg (1979), Burrus (1981), and
H&ycnetal., (1982). No experimental work
has been done at high temperatures and
pressures for the ternary C02 - CH4 - H20
system.
With the development of accurate gas-
loading techniques (Frantz et al., this
Report) combined with synthetic fluid
inclusion techniques (Bodnar and Sterner,
1985; Zhang and Frantz, 1987), experi-
mental studies of the C-O-H system are
now possible. Experiments producing
synthetic fluid inclusions containing wa-
ter-rich CH4 - H20 binary and C02 - CH4 -
1^0 ternary fluids were performed at tem-
peratures from 400 to 600'C and at pres-
GEOPHYSICAL LABORATORY
67
^ 3000
CO
O 2000
O 1000 A
(a)
v>v
/
:x>
yy
/
/////s'sos'
■••' y ?' s s S-' s *s y ** a"
/' *»
. MO
250 350 450 550 650
Temperature (°C)
cd 3000
250 350 450 550
650
Temperature (°C)
-^3000-
co
0) 2000-
to
to
0) 1000-
CL
(c) /*3 / y&
s s s y
19» >• /\l*y ,•''15.1 ,s
< •' S S S «•■* «■*" ,-* T-*"*
Liquid
Vapor
250
350
450 550
Temperature (°C)
650
CO
©
Ik.
13
CO
CO
CD
* '• i»
3000-
(d)
20-3 .-X jSxi***
y y y »
20.9 /^/^ ,.
2000-
< ^,yi7,VX"^''' J«5^ ^^''(13.7) m
1000-
Liquid + \^-~"'^' ""
Vapor
1 1 1 1 1
250
350
450
550
650
Temperature (°C)
Fig. 46. Plot of the measured clathrate melting temperatures as a function of experimental temperature
and pressure for a) Hp, b) 5.5 mol% CH4-H20, c) 1 1.0 mol% CH^Hp, and d) 16.5 mol% CH4-H20.
The shaded lines represent lines of constant clathrate melting temperature (isochores). The non-italicized
numbers refer to the measured clathrate melting temperatures and the italicized numbers refer to the
clathrate melting temperatures represented by these lines. The curved lines represent the liquid- vapor
curve for that compositional section.
sures from 1000 to 3000 bar. These experi-
ments have resulted in the determination of
1) isochores; 2) liquid-vapor curves; and 3)
the melting relations of clathrate as a func-
tion of composition.
CH4-H20 binary compositions
Lines of constant homogenization tem-
perature and isochores have been extremely
useful in natural fluid inclusion studies for
the determination of the temperature-pres-
sure history of the host minerals. Zhang
and Frantz (1987) determined constant
homogenization temperature lines and
isochores for fluids in the NaCl - KC1 -
CaCl2 - H20 system and found that they
could be represented in temperature-pres-
sure space by a series of straight lines
described by a simple function of the form;
/?(bar)=A1+A2r(°C)
(1)
The coefficients At and A2 were functions
of the homogenization temperature (Th)y
the solute type, and the solute concentra-
tion. Using the regression parameters of
68
CARNEGIE INSTITUTION
Table 6. Isochore regression parameters
System
Mol%
CO,
Mol%
CR
a.
a.
a,
CH4-H20
—
5.5
3.274x10*
-9.974 x 102
4.317 x W
-7.073 x 10°
11.0
3.528 x 103
-6.123 x 102
1.760 xlO1
-2.661 x 10°
—
16.5
1.482x10*
-1.919 xlO3
5.634 x 10
-3.500 x 101
C02-CH4-H20
5.5
5.5
6.011 xlO3
-1.311 xlO3
5.135 x 10
-7.779 x 10°
11.0
5.5
-4.771 x 10*
6.277 x 103
-2.096 x 102
1.097 x 102
5.5
11.0
-6.906 x 103
9.524 x 102
-3.725 x 101
2.027 x W
Zhang and Frantz ( 1 987) isochores for pure
water for the unary H20 system are pre-
sented as a function of temperature and
pressure in Fig. 46a.
Clathrate melting temperatures ranging
from 7° to almost 2 1°C were measured with
temperature-cycling techniques and are
plotted as a function of the experimental
temperature and pressure in Figs. 46b, 46c,
and 46d for 5.5, 11.0, and 16.5 mol% CH4
respectively. The italicized numbers repre-
sent experiments in which the inclusions
homogenized to vapor rather than liquid.
Since the melting temperature of clathrate
in the pure CH4-H20 system depends only
on composition and total inclusion density,
they can be used in much the same way as
homogenization temperatures were in the
salt - H20 system to determine isochores.
The data were fit to the following function:
P = (a, + a7T +aj 2)
v 1 2 mc 3 mc '
+ (aA + a,T +al 2)7,
v 4 5 mc 6 mc ' '
(2)
in which Tm is the clathrate melting tem-
perature. The regression constants of these
fits are given in Table 6. Lines of constant
clathrate melting temperature or isochores
are shown as shaded lines in Figs. 46b, 46c
and 46d. The italicized numbers represent
the clathrate melting temperatures corre-
sponding to the lines. The slopes of the
isochores generally become shallower with
increasing concentration of the volatile,
but the major change in slope occurs below
5.5 mol%. The liquid-vapor curve for each
concentration was computed by solving
equation (1) for homogenization pressures
using the experimental homogenization
temperatures (letting T=Th) and clathrate
melting temperatures. The results are shown
as diamonds in Figs. 46b, 46c, and 46d. The
homogenization pressures are high, with
some high density inclusions being greater
than 2000 bar. In Fig. 47, our CH4 - H20
liquid vapor curve results plotted as solid
squares are compared to those of Welsch
(1973) shown as shaded curves. The agree-
ment is remarkable considering the differ-
ent techniques used in the two studies.
The volumetric properties of the CH4 -
C02 isochores can be computed using the
clathrate melting temperatures in conjunc-
tion with the bulk compositions. Bonham
GEOPHYSICAL LABORATORY
69
2500
2500
2000
CO
£, 1500 •
3
W
a> 1000 h
0.
500 •
250 300 350 400
Temperature (C)
11.0 Mol%
*
V
2500
2000 -
CO
* 1500 -|
fl>
k.
3
CO
CO
2 1000 .
Q.
500 -
250 300 350 400
Temperature (C)
16.5 Mol%.
f
t
350
250 300 350
Temperature (C)
Fig. 47. Comparison of the liquid-vapor curve data of this study (solid squares) with those of Welsh
(1973) for 5.5 mol% , 11.0 mol%, and 16.5 mol% CH^Kp.
(1978) demonstrates that the solubility of
methane in water at ambient temperatures
and pressures to 100 bar is less than 0.2
mol% . The ice melting point measurements
made on our methane-water experiments
were within 0. 1 or 0.2 of 0°C indicating low
methane solubility in the liquidphase. Based
on this, the vapor phase can be assumed to
be composed entirely of methane and the
liquid phase, of water and
V** =V(H20)/(l+l/X) +V(CH4)/(1 +X),(3)
where Vtoul is the molar volume of the entire
inclusion, V(CH4) is the molar volume of
the vapor phase, V(H20) is the molar vol-
ume of the liquid phase, andX is the ratio of
the number of moles of CH A over the number
of moles of H20. Values for X can be
computed from the relative amounts of
methane and water introduced into the
experimental charge. The clathrate melting
temperature can be used to compute the
internal pressure of the inclusion at the
clathrate melting temperature using the data
of Deaton and Frost (1946). Their data
appear to be linear when the logarithm of
the pressure is plotted against the melting
temperature and the computed pressures
ranged from 70 to over 200 bar. These
values of the internal pressure were used in
conjunction with the molar volume data for
methane (Angus etal., 1978) and the molar
volume data for pure water to compute the
total inclusion molar volumes correspond-
ing to our experiments. These are shown as
functions of the clathrate melting tempera-
ture for 5.5, 11.0, and 16.5 mol% CH4 in
Fig. 48.
o
E
CO
E
<D
E
3
O
>
hi
«
O
S
0 10 20 30
Clathrate Melting Temperature ( °C)
Fig. 48. Plot of the molarvolume of 5.5, 1 1.0, and
16.5 mol% CF^-t^O as a function of the clathrate
melting temperature.
70
CARNEGIE INSTITUTION
00
O
3
CO
(0
o
» '« is
3000 -
(a)
/' y" X /
y ' y y y
/Ha y y*.9 y ^tt.s
.y ■ y .y • s y ■ it
2000 -
y y y y y y
y y y y' y' y ^ "
y y y S s' **"
s ..' y y y^ *■*
1000 -
y y y y y y
y y y *y y ■
y y y y y
.
Liquid +
• y v .y •'
Y/'y>"
Vapor
^t y.-*'
0-
— i ' 1 ■ 1 '
250
350
450
550
650
Temperature (°C)
3000 -
(b)
CO
s y**
.a
2000 -
y y
y 1S.3 -•• t45
y ■ .y ■
a>
* y
i_
3
CO
y y
>•' y'
2>
1000 -
\ y^m y*
Q.
V" .y
Liquid
+ \ y^
Vapor
\*y
:tr
.***'
old
250 350 450 550 650
Temperature ( ° C)
3000 -
CO
5. 2000 -
CO
3
(/>
CO
0)
1000
(c)
Liquid +
Vapor
250 350 450 550 650
Temperature (°C)
Fig. 49 Plot of the measured clathrate melting
temperatures as a function of experimental tem-
perature and pressure for a) 5.5 mol% CO : 5.5
mol% CH4 , b) 11.0 mol% C(X : 5.5 mol% CH4,
and c) 5.5 mol% C02 :1 1 mol% CH The shaded
lines represent lines of constant clathrate melting
temperature (isochores). The non-italicized num-
bers refer to the measured clathrate melting tem-
peratures and the italicized numbers refer to the
clathrate melting temperatures represented by these
lines. The curved lines represent the liquid- vapor
curve for that compositional section.
C02-CH4-H20 ternary compositions
As in the case of the binary CH4-H20
system, the clathrate melting temperatures
for ternary C02-CH4-H20 compositions
were plotted as a function of the experi-
mental temperature and pressure in an ef-
fort to determine the location of lines of
constant Tme or isochores (Fig. 49a shows
the results for 5.5 mol% C02 : 5.5 mol%
CH4; Fig. 49b, for 11.0 mol% C02 : 5.5
mol% CH4; Fig. 49c, for 5.5 mol%C02 :
1 1 .0 mol% CH4). The numbers in parenthe-
ses indicate experiments in which the in-
clusions homogenized to vapor. These data
were fit with equation (2) yielding regres-
sion coefficients listed in Table 6. Lines of
constant clathrate melting temperature
calculated from these regressions are shown
in Figs. 49a, 49b, and 49c as shaded lines
labeled with italicized values of T . De-
mc
spite the limited number of data points for
the ternary compositions, it is clear that the
slopes of the isochores in Fig. 49a for the
5.5 mol% C02 : 11.0 mol% CH4are quite
similar to those of the 5.5 mol% methane-
water binary (Fig. 46b). The 5.5 mol% C02
: 11.0 mol% CH4 (Fig. 49c) has isochores
with slopes similar to those of the 11.0
mol% CH4- H20 binary (Fig. 46c). It ap-
pears from these observations that the slopes
of the isochores as determined from
clathrate melting temperatures tend to fol-
low the trends of the CH4 - H20. The loca-
tions of the liquid-vapor curves were
computed in the same manner as in the case
of the CH4-H20 binary using the measure-
ments of both the clathrate melting tem-
peratures and the homogenization tempera-
tures in conjunction with equation (2). The
GEOPHYSICAL LABORATORY
71
results, plotted as diamonds, are shown on
Figs. 49a, 49b, and 49c. Liquid-vapor curves
for the 5.5 mol% C02 : 5.5 mol% CH4 and
the 1 1 .0 mol% C02 : 5.5 mol% decompo-
sitions are both almost identical to that of
the 5.5 CH4- H20 binary. The liquid- vapor
curve for the 5.5 mol% C02 : 11.0 mol%
CH4 is very similar to that of the 1 1 .0 mol%
CH4 binary. Even more than in the case of
the slopes of the isochores, the positions of
the liquid-vapor curves in the portion of the
C02 - CH4 - H20 ternary studied in this
investigation closely follow the locations
of the CH4 - F^O liquid-vapor curves hav-
ing the same mol% methane.
The synthetic fluid inclusion method
combined with the gas-loading techniques
of Frantz et al. (this Report) have proven
quite effective in the determination of the
molar volumes, liquid-vapor curves, iso-
chores, and low-temperature phase rela-
tions of C-O-H gas mixtures. The fluids
produced using these techniques yielded
inclusions of extremely homogeneous
composition as evidenced by the low stan-
dard deviations of the microthermometric
measurements for clathrate melting tem-
perature (approximately ±0.1 °C). Regres-
sions considering measurements of the
clathrate melting temperature or the ho-
mogenization temperature as a function of
the experimental temperature and pressure
yielded sets of isochores which systemati-
cally varied as a function of concentration.
As the gas concentration increases, the
slopes of the sets of isochores tend to be-
come shallower. Molar volumes consid-
erably larger than those of pure water were
computed for the isochores using measure-
ments of the clathrate melting temperature
The quality of the data permitted the calcu-
lation of liquid-vapor curves for the CH4 -
Hf> binaries which are in reasonable agree-
ment with data obtained using large-vol-
ume hydrothermal pressure vessels. The
presence of methane appears to dominate
both the slopes of the isochores and the
position of the liquid-vapor curve in the
ternary compositions considered in this
study.
References
Angus, S., B. Armstrong, and K. M. deReuck,
InternationalThermodynamicTables of the Fluid
State-5. Methane. IUPAC chemical data series
16, Pergamon Press, New York 251pp, 1978.
Bodnar, R. J., and S. M. Sterner, Synthetic fluid
inclusions in natural quartz, n. Application to
PVT studies, Geochim. Cosmochim. Acta, 49,
1855-1859, 1985.
Burrus, R. C, Analysis of fluid inclusions: phase
equilibria at constant volume, Am. J. ScL, 281,
1104-1126,1981.
Deaton, W. M., and E. M. Frost, Gas hydrates and
theirrelation to the operation of natural pipelines,
U. S. Bur. Mines Monogr., 8, 103p, 1946.
Franck, E. U., and K. Todheide, ThermischeEigen-
schaften uberkritischer Mischungen von
Kohlendioxyd und Wasser bis zu 750 °C und
2000 atm, Chem. Neue. Folge., 22, 232-245,
1959.
Green, D. H., T. J. Falloon, and W. R. Taylor,
Mantle-derived magma-roles of variable source
peridotite and variable C-H-O fluid composi-
tions, in Magmatic Processes: Physicochemical
Principles, B. Mysen, ed., Spec. Pub. No. 1, The
Geochemical Society, University Park, Penn-
sylvania, pp. 139-154,1987.
Greenwood, H. J., The compressibility of gaseous
mixtures of carbon dioxide and water between 0
and 500 bars pressure and 4500 and 800°C, Am.
J. ScL, 267 A, 191-208, 1969.
Greenwood, H. J., Thermodynamic properties of
gaseous mixtures of H20 and C02 between
71
CARNEGIE INSTITUTION
450°and 800°C and 0 to 500 bars, Am. J. Sci.,
273,561-571,1973.
Heyen, G., C. Ramboz, and J. Dubessy, Simula-
tion des equilibres des phases dans le system
C02-CH4 en-dessous de 50°C et de 100 bars, C.
R. Acad. Sci. Paris, 294, 203-206, 1982.
Hollister, L., and R. C. Burrus, Phase equilibria in
fluid inclusions from the Khtada Lake metamor-
phic complex, Geochim. Cosmochim. Acta , 40,
163-176, 1976.
Ramboz, C, D. Schnapper, and J. Dubessy, The P-
V-T-X-f(02) evolution of Ufl - CO, - CH4-
bearing fluid in a wolframite vein: Reconstruc-
tion from fluid inclusion studies, Geochim.
Cosmochim. Acta, 49, 205-219, 1985.
Saxena, S. K., Oxidation state of the mantle,
Geochim. Cosmochim. Acta, 53, 89-95, 1989.
S wanenberg, H. E. C, Phase equilibria in carbonic
system, and their application to freezing studies
of fluid inclusions, Contr. Mineral. Petrol., 68,
303-306, 1979.
Takenouchi, S., and G. C. Kennedy, The binary
system Hfl - C02 at high temperatures and
pressures, Am. J. Sci., 262, 1055-1074, 1964.
Welsch, H., Die Systems Xenon-Wasser und
Methan-Wasser bei lohen Drucken und tem-
perature, Ph. D. dissertation, Inst, for Physical
Chem., Karlsruhe, 1973.
Wintsche, R. P., A. F. O'Connell, L. Ransom, and
M. J. Wiechmann, Evidence for the influence of
f(CH4) on the crystallinity of disseminated car-
bon in greenschist facies rocks, Rhode Island,
USA,Contr.Mineral.Petrol., 77,207-213, 1981.
Zhang, Y. G., and J. D. Frantz, Determination of
the homogenization temperatures and densities
of supercritical fluids in the system NaCl - KC1
- CaCl2 - I^O using synthetic fluid inclusions,
Chem. Geol., 64, 335-350, 1987.
Zhang, Y. G., and J. D. Frantz, Experimental
determination of the compositional limits of
immiscibility in the system CaCl2 - C02 - H20 at
high temperatures and pressures using synthetic
fluid inclusions, Chem. Geol., 74,289-308, 1989.
A Laser-based Carbon Reduction Tech-
nique For Oxygen Isotope Analysis of
Silicates and Oxides
Zachary D. Sharp and James R. O'Neil
Stable isotope analysis is one of the
most valuable geochemical techniques
available for constraining the conditions of
formation and alteration in most rock types.
However, the methods for extracting oxy-
gen from silicates and oxides have remained
relatively unchanged over the past 25 years.
As a result, large sample sizes are required
for analysis and nearly all spatial resolution
is lost. Fine-scale variations in the #80 of
oxides and silicates cannot normally be
determined. In comparison, other geo-
chemical techniques have been used suc-
cessfully to determine major, minor and
trace element concentrations, crystal struc-
tures, and isotopic compositions (i.e., U-
Th-Pb, ^ArrAr, #4S) on a Jim scale. This
report outlines a microanalytical method
for &*0 determinations of selected sili-
cates and oxides using a laser-heated car-
bon reduction technique.
The natural variations in the 5180 of
minerals is so small that analytical tech-
niques of very high precision are required
to distinguish them. Two analytical meth-
ods have been developed that are capable
of analyzing the Sl 80 of oxides and sili-
cates at the high level of precision neces-
sary to discern these small isotopic differ-
ences: 1) fluorination and 2) carbon reduc-
tion.
Fluorination of minerals at moderate
temperatures (200-650° C) liberates 02
* Dept. of Geological Sciences, University of
Michigan, Ann Arbor, Ml 48109-1063
GEOPHYSICAL LABORATORY
73
Stabilities of Common Oxides
h0
o
-20
-40
•60
i i
Mln. temp.
of system
£.'»»■
1
CO
, so2
A1203
MgO
CaO
FeO
t^O
1000 1500 2000 2500
Temperature (K)
3000
Fig. 50. T-f(0)2 diagram showing the stability of
common oxides relative to the elements. The C -
CO buffer (thick solid line) crosses below the
stability field of all common oxides above 2450K.
In the presence of graphite, all oxides will reduce
to the elements and carbon monoxide above this
temperature. The minimum temperature of the
present system (shaded area) is 2890K (melting
point of molybdenum). Thermodynamic data are
from Chase etal (1985).
which is converted to C02 and subsequently
analyzed by conventional isotope ratio mass
spectrometry. The fluorinati on method (e.g. ,
Baertschi and Silverman, 1951; Clayton
and Mayeda, 1963) is applicable to all but
the most refractory minerals, but relatively
large sample sizes of 5-30 mg are generally
required to obtain accurate and reproduc-
ible results. Samples as small as 1 mg can
be analyzed only with extreme care and
assigning a somewhat arbitrary blank cor-
rection (Lee et al., 1980).
The carbon reduction method involves
the high temperature (1000-2400°C) re-
duction of minerals to carbides or native
elements and CO with varying amounts of
C02. The evolved CO is converted to C02
and analyzed. This technique has met with
partial success (e. g., Schwander, 1953;
Clayton and Epstein, 1958), but has been
limited, in part, by the difficulty of attain-
ing the high temperatures required for reac-
tion. The traditional carbon reduction tech-
nique also requires large sample sizes, a
time consuming degassing procedure and
is claimed to be unsatisfactory for alkali-
bearing minerals (Clayton and Epstein,
1958). This is a result of the volatile alkali
metals being oxidized by the evolved CO,
which involves a large isotopic fractiona-
tion. The laser-based carbon reduction
technique allows for reduced sample sizes
of 1-3 mg or less, and because the heating
is so rapid, analyses can be made much
more quickly and cleanly than with either
the conventional fluorination or carbon
reduction method.
The carbon reduction method is based
on the fact that all common oxides and
silicates will be reduced in the presence of
carbon, if sufficient temperatures are
reached. For example, quartz and mag-
netite are reduced by the following reac-
tions, respectively (Clayton and Epstein,
1958):
Si02 + 3C = SiC + 2 CO,
(1)
Fe304 + 4 C = 3 Fe + 4 CO. (2)
The stabilities of selected common oxides
relative to the C - CO - C02 buffer are
shown as a function of temperature at 1 bar
in Fig. 50. Above 2400°C, all common
oxides in the presence of graphite should be
reduced to either elements or carbides.
Temperatures of 2617°C (melting point of
molybdenum) are easily achieved with the
Nd- YAG laser equipment at the Geophysi-
cal Laboratory.
In the present system, samples are pre-
pared by grinding the weighed mineral
74
CARNEGIE INSTITUTION
7\
EXTRACTION SYSTEM
m
%K
V
LASER
c
n^O1^
SAMPLE
CRUCIBLE
REACTION CHAMBER T
1 ZEOLITE
Itrap
CO-C02
CONVERTER
Sample
Bulb
Manometer
w
Fig. 51. Laser extraction system. The system consists of a Nd-YAG laser, a sample crucible, a reaction
chamber with zeolite trap and a CO - C02 converter. During reaction, the CO is collected on the zeolite
trap, which is removed from the reaction chamber and placed onto the vacuum line. The CO is then
desorbed and converted to C02 on the CO
C02 converter.
with excess graphite (15-50% over stoi-
chiometric). The mixture is pressed into a
pellet 3 mm in diameter to insure intimate
contact between the two phases. During
laser heating, the graphite absorbs the ra-
diation, heating the entire pellet to ex-
tremely high temperatures. To prevent
reaction with the sample chamber, the pel-
let is nested in a 5 mm O.D. graphite
crucible which in turn is placed in a plati-
num crucible holder. The platinum cup is
used to prevent direct contact between the
hot graphite crucible and the walls of the
sample chamber.
The extraction system comprises a la-
ser, a sample crucible, a reaction chamber
and a CO - C02 converter (Fig. 51). A
variable power, 18 watt maximum Nd-
YAG laser (A = 1.064 ^m) with a 6X air-
objective is used as a heating source. The
focal length of this lens is 20 mm with a
focused spot of -10 jum. Although most
minerals are transparent to radiation of this
wavelength, graphite strongly absorbs the
radiation and heats up. The admixed min-
eral is heated by conduction from the graph-
ite powder. The sample chamber consists
of a 20 mm O.D. quartz sample tube with a
ground glass joint to admit the sample and
a removable zeolite trap (cold finger)
equipped with a stopcock (Fig. 51). The
reaction chamber containing the sample
pellet is evacuated and degassed at 600°C
for 1 hour. The sample is gently rastered
under the focused laser beam until the
sample pellet is entirely reacted. The
evolved CO is collected on the cold finger
as the reaction proceeds. Thus a high vac-
GEOPHYSICAL LABORATORY
75
uum is maintained in the reaction chamber
at all times. By keeping the CO fugacity
extremely low, there is less chance for the
CO to react with reactive alkali metals that
may vaporize and plate out on the chamber
walls during reaction. The zeolite trap is
isolated from the reaction chamber and
heated to 300°C to desorb the CO. The CO
is then converted to C02 with a high volt-
age electric discharge between two parallel
platinum plates (Aggetera/., 1965; Rafter,
1967) by the reaction
2 CO = C02 + C.
(3)
The C02 yield is measured manometri-
cally, and the gas is fed to a sample tube.
Isotopic data obtained with the laser-
based carbon reduction method are pre-
sented in Table 7 and Fig. 52. Analyses of
all oxides and some of the silicate minerals
examined are in agreement with analyses
made with conventional fluorination meth-
ods. Refractory minerals such as olivine
and kyanite give erratic results and are not
amenable to the present carbon reduction
technique. Analyses of feldspars are con-
sistently low by 3.3 ± 0.3 %o. During laser
heating, tiny particles of unreacted mate-
rial are blown out of the graphite crucible
along with the evolving CO and conse-
quently yields are generally low, typically
40-60 per cent. Because magnetite reacts at
very low temperatures, yields of 100% are
achieved. The low yields obtained for the
other minerals do not affect the isotopic
results. There is no correlation between the
% yield of a mineral and the difference
between the measured and true isotopic
composition of that mineral, as long as
feldspar is not included in the correlation.
If the isotopic analyses of feldspars are
always low by a constant amount, a correc-
tion can be applied to the data in order to
bring them into accord with the actual
values as is done for carbonates analyzed
by the H3P04 technique (McCrea, 1950). It
is not clear why the #80 values are lower
than their accepted values. Clayton and
Epstein (1958) first reported a problem
with carbon reduction of feldspar. They
observed the formation of a metallic mirror
on the walls of the glass chamber as the
samples were heated. Such mirrors are
probably the result of the vaporization of
alkalis from the mineral and condensation
on the glass walls as metals. As their reac-
tions proceeded, the mirrors lost their
metallic finish presumably due to the oxi-
dation of the highly reactive metal by either
CO or C02. Because the oxidation of the
c
o
o
=J
ID
rr
c
o
-Q
t_
03
o
30.0
25.0
20.0
«M5.0
10.0
5.0
0.0
¥'
* ♦
ft' \k iv i i i i i
,-tj
,-A Monticellite a
Quartz a
Magnetite A
MnC>2 O
BiP04 +
Garnet x
Diopside a
Orthopyroxene ■
Wollastonite •
Olivine y
i F.eldjjpar, +
0.0 5.0 10.0 15.0 20.0 25.0 30.0
5180
Fluorination
Fig. 52. Comparison between &*0 values deter-
mined with the carbon reduction method and those
from the fluorination method. The diagonal line
represents perfect agreement between the two
methods. The #80 values of feldspar determined
with the carbon reduction method are consistently
3.3 %o light. Olivine and garnet may give erratic
results.
76
CARNEGIE INSTITUTION
Table 7. Isotopic composition, sample size and % recovery of various minerals determined with the
carbon reduction technique. The #80 actual represents either accepted values of isotopic standards or
newly measured values using the fluorination method. All data are reported in the %o notation relative
to SMOW.
Mineral
(TO
#8o
Sample Size
/zmole C02
%
actual
measured
(mg)
recovered
recovery
Quartz
7.2
7.2
3.6
n.d.
n.d.
7.2
7.1
3.0
12
23
7.2
7.1
3.9
25
38
16.2
15.9
3.2
24
45
16.2
15.7
2.8
20
42
16.2
16.5
3.0
7
13
16.2
16.5
3.3
35
63
25.3
24.4
2.3
23
59
25.3
25.4
1.5
n.d.
n.d.
Magnetite
0.6
0.6
1.9
16
100
0.6
0.8
2.3
20
100
MnOz
3.1
3.1
2.9
14
34
3.1
3.5
2.9
18
56
BiP04
11.7
11.5
3.6
13
55
11.7
11.7
2.7
8
42
Orthopyroxene
6.8
6.8
3.9
34
77
6.8
6.4
2.4
16
60
Diopside
19.0
18.0
2.7
18
49
19.0
19.4
6.6
41
45
19.0
19.5
4.1
32
56
Wollastonite
9.3
9.1
5.3
18
26
9.3
9.6
2.4
12
39
9.3
9.3
4.1
20
38
Garnet
6.4
5.7
4.1
23
46
6.4
5.8
2.7
14
44
6.4
6.3
3.0
16
44
Forsterite
5.7
6.0
3.0
n.d.
n.d.
5.7
7.2
3.2
n.d.
n.d.
5.7
1.0
1.9
4
13
5.7
6.4
1.2
4
25
GEOPHYSICAL LABORATORY
Table 7. Continued
77
Mineral
#*o
#8o
Sample Size
/miole C02
%
actual
measured
(mg)
recovered
recovery
Monticellite
22.0
21.5
4.8
20
33
22.0
21.6
4.5
18
31
22.0
21.4
4.8
19
31
Feldspar
12.5
9.1
2.7
16
42
12.5
9.5
2.5
16
46
3.4
0.3
3.2
13
29
3.4
-0.2
3.7
17
32
Kyanite
n.d.
7.5
3.6
19
34
n.d.
8.6
4.0
24
39
CaO
n.d.
25.6
3.3
11
37
n.d.
25.4
4.9
20
46
A1203
n.d.
14.4
4.7
24
35
n.d.
14.2
3.8
23
41
MgO
n.d.
n.d.
3.1
15
39
n.d.
n.d.
3.4
20
47
metallic mirror is a kinetic rather than
equilibrium reaction, carbon oxides con-
taining the light isotope of oxygen should
react preferentially. As a result, the residual
CO would have a &*0 value that is higher
than the actual value of the alkali-bearing
silicate. This expected elevation of the &%0
values of feldspars was observed by Clay-
ton and Epstein (1 958), but with the present
carbon reduction technique, the measured
&%0 of the feldspars is lower than the actual
value (Table 7). One explanation for the
low isotopic values is that the oxygen in the
Si-O-Al bonds reacts preferentially to the
oxygen in Si-O-Si bonds in the laser-based
carbon reduction procedure. The sign and
magnitude of this effect is consistent with
previous estimates based on relative bond
strength (Taylor and Epstein, 1962).
Isotopic analyses of the refractory min-
erals forsterite, kyanite, and to a lesser
extent, garnet are more erratic than those of
other minerals, and in the case of olivine,
the yields are substantially lower. The ox-
ides that comprise olivine and kyanite
(MgO, Si02 and A^Og) all react to a uni-
form extent with reproducible isotopic
analyses. The calculated /(02) required to
reduce forsterite to its elements is higher
than for the oxide components, so thermo-
78
CARNEGIE INSTITUTION
dynamics favors the reduction of olivine
over the oxides. Furthermore, olivine melts
when radiated by the Nd-YAG laser be-
cause the iron linkages absorb radiation at
1 .063 Jim. The reaction involves a thermo-
dynamically favorable reduction of a melt
with graphite, but very little takes place.
The erratic #80 observed for forsterite, but
not monticellite (CaMgSi04), cannot be
explained by preferential reaction of differ-
ently bound oxygen. All of the oxygen in
olivine is shared between a silicon and M2*
cation. There is no isotopic distinction
between sites, and therefore, no possible
reaction of one site relative to another.
The safety, rapidity and small sample
sizes are the primary benefits of the laser-
based carbon reduction technique over
conventional fluorination. Microvariations
in the &*0 of quartz veins, porphyroblasts
and phenocrysts can be measured with this
new method. Unfortunately, not all miner-
als are amenable to the carbon reduction
method at this time, but further investiga-
tions of the isotopic systematics involving
the laser-based carbon reduction technique
may lead to a better understanding of high-
temperature, rapid kinetic processes that
occur during laser heating.
Baertschi, P., and S. R. Silverman, The determina-
tion of relative abundances of the oxygen iso-
topes in silicate rocks, Geochim. Cosmochim.
Acta, 7,317-328, 1951.
Chase, M. W. Jr., C. A. Davis, J. R. Downey, Jr.,
D. J. Frurip, R. A. McDonald, and A. N. Syverud,
JANAF Thermochemical Tables, 3rd ed., J.
Phys. Chem. Ref. Data, 14, 1986.
Clayton, R. N., and S. Epstein, The relationship
between 018/016 ratios in coexisting quartz,
carbonate and iron oxides from various geologi-
cal deposits, /. Geol., 66, 352-371, 1958.
Clayton, R. N., and T. K. Mayeda, The use of
bromine pentafluoride in the extraction of oxy-
gen from oxides and silicates for isotopic analysis,
Geochim. Cosmochim. Acta, 27, 43-52, 1963.
Lee, T., T. K. Mayeda, and R. N. Clayton, Oxygen
isotopic anomalies in Allende inclusion HAL,
Geophys. Res. Lett., 7, 493-496, 1980.
McCrea, J. M., On the isotopic chemistry of car-
bonates and a paleotemperature scale, /. Chem.
Phys., 18, 849-857, 1950.
Rafter, T. A., Oxygen isotopic compositions of
sulphates - 1: A method for extraction of oxygen
and its quantitative conversion to carbon diox-
ide for isotope ratio measurements, N. Zealand
J.ScL, 70,493-510, 1967.
Schwander, H., Bestimmung des relativen Sauer-
stoffisotopen-Verhaltnisses in Silikatgesteinen
und -Mineralien, Geochim. Cosmochim. Acta,
4, 261-291, 1953.
Taylor, H. P., Jr., and S. Epstein, Relationship
between 018/016ratios in coexisting minerals of
igneous and metamorphic rocks, Geol. Soc. Am.
Bull., 73, 675-694, 1962.
References
Aggett, J., C. A. Burton, T. A. Lewis, D. R.
Llewellyn, C. O'Connor, and A. L. Odell, The
isotopic analysis of oxygen in organic com-
pounds and in coordination compounds contain-
ing organic hazards, /. Appl. Radiat. I sot., 16,
165-170, 1965.
GEOPHYSICAL LABORATORY
79
Crystallography - Mineral Physics
Isotope Effects in Dense Solid
Hydrogen: Phase Transition in
Deuterium at 190 (±20) GPa
Russell J. Hemley and Ho-kwang Mao
Once the exclusive domain of theory,
the behavior of hydrogen at ultrahigh pres-
sures is now amenable to direct experimen-
tal investigation using ultrahigh-pressure,
diamond-anvil techniques (Hemley and
Mao, 1988; Mao and Hemley, 1989). The
goal of this effort has been the characteriza-
tion of hydrogen above 100 GPa and test-
ing of theoretical predictions of the insula-
tor-metal transition pressure in this mate-
rial, currently predicted to occur between
150 and 300 GPa. Recently, we demon-
strated that solid hydrogen undergoes a
phase transition at 145 GPa and 77K
(Hemley and Mao, 1988). Changes in the
Raman spectra indicate that the transition
may be a structural one between insulating
molecular phases, possibly associated with
orientational ordering. Recently, we have
shown that on further increase in pressure
the optical properties of the high-pressure
phase change dramatically. Above 200 GPa
there is evidence for ground-state elec-
tronic excitations at visible wavelengths,
and at pressures in the 250 GPa range
hydrogen samples are opaque (Mao and
Hemley, 1989). These observations have
led to the possibility that the transition at
145 GPa may be associated with band
overlap. If so, the character of the optical
spectra indicate that the gap may be indi-
rect at closure. This interpretation is con-
sistent with theoretical predictions that band
overlap may occur below 200 GPa (Friedli
andAshcroft, 1977; MmetaL, 1986;Barbee
et al, 1989).
An important constraint on the nature of
phase transitions in solid hydrogen is pro-
vided by the study of isotope effects, which
produce a well-known effect on the orien-
tational ordering transformation at low
densities (Silvera, 1988). The ordering
transformation occurs at significantly lower
pressures in D2 than in H2 because of the
lower rotational constant of the heavier
isotope. There is a pronounced isotope effect
on the Raman-active vibron. In both hydro-
gen and deuterium, the frequency of the
vibron increases with pressure but then
decreases above a critical pressure. In H2
the critical pressure is 30 GPa whereas in
D2 it is shifted to 50 GPa (Sharma et al.
1980). The weaker negative pressure shift
of the D2 vibron is magnified at higher
pressures: at 125 GPa, for example, the
vibron frequency is 40 cm * above and 40
cm1 below the zero-pressure values for D2
and H2, respectively (Mao et al, 1985).
Although these results suggest that the
equations of state for the two solids are
significantly different at these very high
densities, direct measurements at low pres-
sures indicate that the compressibility of
the two solids are similar (Mao et al.,
1988).
In the present study we have pressur-
80
CARNEGIE INSTITUTION
ized deuterium to pressures above 250 GPa
to examine its vibron shift above the previ-
ous limit of 125 GPa (Mao et ai, 1985). In
particular, we wished to examine possible
phase transitions in the region of the transi-
tion observed in hydrogen. The experi-
ments were performed using techniques
described previously (Hemley and Mao,
1 988). The Raman active vibron was meas-
ured as a function of pressure, with pres-
sure determined using the ruby fluores-
cence technique. Because of the use of low-
fluorescence diamonds and the high sensi-
tivity of the optical system, time-resolved
techniques were not needed to measure the
ruby fluorescence spectra. The ruby spec-
tra were measured to pressures of about
240 GPa on the quasihydrostatic pressure
scale.
The pressure shift of the Raman-active
vibron of molecular deuterium to pressures
above 200 GPa is shown in Fig. 53. A
single, well-resolved band which decreases
in frequency above 50 GPa was observed
initially, as in previous studies (Sharma et
al, 1980; Mao et ai, 1985). Above 100
GPa, the negative pressure shift of the
vibron, measured here at 77 K, converged
with that measured previously at room
temperature to 125 GPa, and continued
smoothly to higher pressures. At 190 (±20)
GPa, however, a second, broader peak
appeared 130 (±5) cm * below that of the
first. The two peaks coexisted over a small
pressure interval with the intensity of the
new peak growing at the expense of the
first with increasing pressure. With further
increase in pressure, the higher frequency
peak completely disappeared. The second
peak continued to decrease in frequency
3200
§3100
_§3000
E
§2900
CD
03 2800
2700
i I I
Vibron
Pressure Shift
Deuterium
77 K
0 50 100 150 200 250 300
Pressure, GPa
Fig. 53. Pressure shift of the Raman-active
vibron of molecular deuterium to ultra-high
pressures. The second, lower frequency vibron
first appeared at a ruby pressure of 190 (± 20)
GPa. The two peaks coexisted over a pressure
interval of approximately 20 GPa, with an aver-
age separation of 130 cm1 between them. The
width of the mixed phase region is likely to
reflect the magnitude of pressure gradients in
the sample across the laser spot (~5 mm in
diameter). Similar effects were observed in
hydrogen.The error bars for the higher pressure
points include the uncertain-ties arising from
the broadness of the ruby Rj peaks and from
pressure calibration (hydrostatic versus qua-
sihydrostatic). The pressure could not be meas-
ured directly for the two highest pressure spec-
tra. At these pressures, the sample had com-
pletely transformed to the high pressure phase,
as evidenced by the single peak present in the
measured spectra.
with increasing pressure; at the highest
pressures attained, the frequency of the
peak was 2793 cm1. The pressures could
not be determined reliably above 240 GPa
as a result of the decrease in the ruby
fluorescence intensity. We estimate that
the maximum pressure reached on the
deuterium was 240-280 GPa (see Mao and
Hemley, 1989).
The appearance of the second, broader
vibron, and coexistence of the two peaks
over a small pressure interval, strongly
resembles that recently observed in hydro-
GEOPHYSICAL LABORATORY
81
gen. However, there are some notable dif-
ferences in the two isotopes that bear fur-
ther analysis. First, despite the uncertain-
ties in pressure, the data clearly indicate
that the transition in deuterium occurs at a
higher pressure than that in hydrogen (190
versus 145 GPa). Second, the vibron dis-
continuity is larger (130 versus 105 cm1 for
hydrogen). The weaker pressure depend-
ence of the vibron in deuterium has been
noted previously (Mao et al. 1985). The
transition in deuterium occurred when the
vibron reached a value of 2990 cm1, which
is close to its zero-pressure value; in con-
trast the transition in hydrogen occurred
when the vibron was approximately 100
cm1 below its zero-pressure value.
The more pronounced pressure depend-
ence of the vibron frequency for hydrogen
indicates that the molecules in this solid are
significantly more anharmonic than in
deuterium at a given pressure (density).
Ashcroft (1988) has argued that the differ-
ent pressure dependencies of the vibron
frequencies may arise from large differ-
ences in zero-point energy for the two iso-
topes. If the phase transitions in the two
isotopes involve the same crystal struc-
tures, it is possible that the differences in
zero-point energy may also be responsible
for the higher phase transition pressures for
the heavier isotope. Moreover, since the
difference in zero-point energy for the two
isotopes is expected to be comparable to
the energy differences between different
structures, it is also possible that the two
isotopes crystallize in different structures
at high pressures and have a different series
of phase transitions in the molecular solid
prior to metallization. Finally, recent theo-
retical calculations indicate that band over-
lap should occur in the molecular phase as
low as 150 GPa (see Mao and Hemley,
1989). At the point of band overlap, elec-
tron density will be removed from the
molecular bonds to conduction states,
causing a decrease in the frequency of the
molecular vibron. The magnitude of this
frequency shift could therefore provide
useful constraints on possible band overlap
in the molecular phase.
References
Ashcroft, N. W., Quantum liquid metals: the
physics of dense hydrogen, Z. Phys. Chemie,
756,41-51,1988.
Barbee, T. W., A. Garcia, M. L. Cohen, and J. L.
Martins, Theory of high-pressure phases of
hydrogen,Phys.Rev.Lett.,62, 1150-1153, 1989.
Friedli, C, and N. W. Ashcroft, Combined repre-
sentation method for use in band structure calcu-
lations: application to highly compressed
hydrogen, Phys. Rev. B, 16, 662-672, 1977.
Hemley, R. J., and H. K. Mao, Phase transition in
solid molecular hydrogen at ultra-high pressure,
Phys. Rev. Lett. 61, 857-860, 1988.
Mao, H. K., and R. J. Hemley, Optical studies of
hydrogen above 200 gigapascals: evidence for
metallization by band overlap, Science, 244,
1462-1465, 1989.
Mao, H. K., P. M. Bell, and R. J. Hemley, Ultra-
high pressures: optical observations and Raman
measurements of hydrogen and deuterium to
1.47 Mbar, Phys. Rev. Lett., 55, 99-102, 1985.
Mao, H. K., A. P. Jephcoat, R. J. Hemley, L. W.
Finger, C. S. Zha, R. M. Hazen, and D. E. Cox,
Synchrotron x-ray diffraction measurements of
single-crystal hydrogen to 26.5 gigapascals,
Science, 239, 1131-1134, 1988.
Min, B. I., H. J. F. Jansen, and A. J. Freeman,
Pressure-induced electronic and structural phase
transitions in solid hydrogen, Phys. Rev. B, 33,
82
CARNEGIE INSTITUTION
6383-6390, 1986.
Sharma, S. K., H. K. Mao, and P. M. Bell, Raman
measurements of deuterium in the pressure range
of 8-537 kbar at room temperature, Carnegie
Instn. Washington Year Book, 79, 358-364, 1980.
Silvera, I. F., The phase diagram and excitations in
solid hydrogen: prospects for metallization, in
Simple Molecular Systems at Very High
Densities, A. Polian, P. Loubeyre, and N. Boc-
cara, eds., Plenum, New York, pp. 33-46, 1988.
The Effect of Pressure, Temperature,
and Composition on the Lattice Para-
meters and Density of (Fe,Mg)Si03 -
Perovskites to 30 GPa
Ho-kwang Mao, Russell J. Hemley, Jinfu
Shu, Liang-chen Chen,
Andrew P. Jephcoat, Yan Wu, and
William A. Bassett*
Information on the physical properties
(density, bulk modulus, and lattice parame-
ters) of the MgSi03-perovskite as a func-
tion of pressure, temperature and Fe-Mg
composition is of fundamental importance
for a realistic model of the lower mantle.
Although there is a growing body of data
on these properties from high pressure
single-crystal and polycrystalline x-ray
diffraction (Yagi etal., 1982; Kudoh etal.,
1987; Knittle and Jeanloz, 1987; Ross and
Hazen, 1989), from Brillouin scattering
measurements (Yeganeh-Haeri and Weid-
ner, 1989) and from theoretical calcula-
tions (Hemley etal. , 1 987; Wolf and Buko w-
inski, 1987), little is known about the prop-
erties of perovskite at higher pressures.
* Department of Geological Sciences,
Cornell University, Ithaca, NY 14853
Indeed, the equation of state of the orthor-
hombic perovskite has not been studied
above 1 3 GPa under hydrostatic pressure
conditions; hydrostatic pressure is neces-
sary for distinguishing the compressibility
of individual lattice parameters. Previous
quasihydrostatic measurements on the elas-
ticity of the perovskite were made at ambi-
ent conditions or at pressures far below the
stability field of the perovskite, which could
cause samples to behave abnormally. Ther-
mal expansion data for (Fe,Mg)Si03-
perovskite were collected at ambient pres-
sure by Knittle et al. (1986). No measure-
ments have yet been carried out at simulta-
neous high-pressure and high-temperature
conditions. The effects of Fe/Mg ratio on
the equation of state or the perovskite,
which is crucial for the determination of the
iron content in the mantle, has also not been
studied experimentally nor theoretically.
Polycrystalline x-ray diffraction in the
megabar diamond-anvil cell with an exter-
nal resistance heater can reach the stability
field of the perovskite and cover the pres-
sure range of the lower mantle. A high
degree of hydrostaticity and high resolu-
tion in the diffraction measurements are
necessary for observing changes in orthor-
hombic distortion and for determining a
precise equation of state for the perovskites.
The present paper reports the development
of a new technique for such experiments .
The silicate perovskite samples were
synthesized from synthetic pyroxenes by
laser-heating at 40 GPa in diamond-anvil
cells (Mao et al., 1977). The perovskite
sample formed a disc-shaped polycrys-
talline aggregate, 150 mm diameter and 20
to 50 mm thick and gold was used as a high-
GEOPHYSICAL LABORATORY
83
Belleville Springs
TC1
III Lever
Diamond Anvil
Fig. 54. (a) The diamond-anvil cell with external
heater; (b) Sample configuration
temperature high-pressure calibrant
(Jamieson et al., 1982; Ming et al., 1983).
Several ruby grains, 2-5 mm size, were
placed on top of the perovskite sample as
an ambient-temperature high-pressure cali-
brant (Mao etal.y 1986). The sample cham-
ber was then filled with neon gas at 200
MPa and ambient temperature in a high-
pressure gas-loading device (Jephcoat et
al., 1987). In the present experiment, a
sleeve-shaped platinum-wire heater with a
shell of ceramic insulator was fit around the
extruded portion of piston-cylinder (Fig.
54). Temperatures were measured with two
chromel - alumel thermocouples, mounted
on the diamond anvils. The temperature
difference between the two thermocouples
was normally less than the uncertainty of
the thermocouple (~2 K) at 900K.
One of the major difficulties in studying
the (Fe,Mg)Si03-perovskite by polycrys-
talline x-ray diffraction techniques arises
from the fact that the diffraction pattern
consists primarily of many groups of multi-
plets (Yagi et al., 1977). The separation of
the peaks within each multiplet is typically
only ~ 1%. In order to obtain accurate
measurement of the lattice parameters, it is
necessary to resolve these multiplets by
high-resolution techniques. Monochro-
matic synchrotron x-ray radiation currently
provides the highest possible resolution for
polycrystalline diffraction. In the present
study, we used a wiggler beamline at Cor-
nell High Energy Synchrotron Source
(CHESS); a sagittal single-crystal mono-
chromator was used to provide focused
16.1 keV x-ray radiation. The diffraction
patterns were recorded using film tech-
niques. In order to increase the angular
resolution, we replaced the commonly used
50-mm radius film cassette for diamond
cell with a newly designed 100-mm radius
cassette. Example of diffraction patterns
are shown in Fig. 55.
The orthorhombic distortion of the
(Fe,Mg)Si03-perovskites is clearly revealed
by the splitting of single diffraction lines of
the cubic structure into doublets or triplets
in the orthorhombic structure. The most
intense triplet consists of the 020, 112, and
200 diffraction lines (Fig. 55), which are
equivalent to the 110 diffraction peaks of
the cubic perovskite. By monitoring the
splitting of these triplets at high pressure
and temperatures, the change of orthor-
84
CARNEGIE INSTITUTION
CO
— Perovskite
9.7 GPa, 298 K
30 31
32 33
29 (°)
34 35
— i 1 1 r~
F<b.iM9o.9Si<% Au
— Perovskite
4.2 GPa, 877 K
30 31
32 33
20 (°)
34 35
Fig. 55. (a,b) Diffraction patterns of (Fe,Mg)Si03-
perovskite at 9.7 GPa and 298 K and at 4.2 GPa
and 877 K.
hombic distortion can be obtained directly.
The lattice parameters and unit cell vol-
umes of the (Fe,Mg)Si03-perovskites de-
termined at high pressures and 298 K in this
manner are listed in Table 8.
Pressures measured on the gold and
ruby scales were in good agreement, and
were averaged (standard deviation, 0.2 GPa)
to yield the reported pressures in Table 8.
Pressures were also determined by the
volume compression and P-V equation of
state of neon (Hemley et al., 1989).
The compressibilities of the three
FexMg1 xSi03 perovskites with
x = 0, 0.1, and 0.2 determined from the
present study are indistinguishable (Table
8). The three sets of data were combined for
a least-squares fit of a second-order Mur-
naghan equation of state (Murnaghan,
1944);
V/Vo = /7 + (K'P/KJ]^
(1)
K = - (dP/dlnV)T = Ko + Ko'P, (2)
where V, K, and K' are molar volume, bulk
modulus, and pressure derivative of the
bulk modulus; the subscript zero denotes
the parameters at zero pressure. Two-pa-
rameter, least-squares fitting yielded Ko =
275(±8) GPa and K ' = 3.7 (±0.8). How-
300
280-
*°260
220
220
(Fe,Mg)Si03
Perovskite
4 6
K '
8 10
Fig. 56. Dependence of Ko on fixed K\ for the
298 K isotherm of (Fe,Mg)Si03 perovskites.
GEOPHYSICAL LABORATORY
85
Table 8. Lattice Parameters and Unit-Cell Volume of FexMg1.iSi03-Perovskite up to 30
GPa (298 K).
/>,GPa
a> A
b, A
c, A
V,A3
a/a
o
bib
0
clc
o
V/V
o
x=0.0
3.86
4.753
4.911
6.875
160.46
0.9947
0.9962
0.9966
0.9876
6.29
4.736
4.895
6.852
158.84
0.9911
0.9931
0.9932
0.9776
13.3
4.707
4.869
6.793
155.69
0.9851
0.9877
0.9847
0.9582
x=0.1
2.50
4.768
4.916
6.881
161.30
0.9972
0.9971
0.9965
0.9908
9.69
4.726
4.883
6.819
157.37
0.9885
0.9903
0.9875
0.9667
13.5
4.707
4.867
6.796
155.70
0.9844
0.9871
0.9842
0.9564
18.6
4.683
4.848
6.761
153.51
0.9795
0.9832
0.9792
0.9430
27.2
4.648
4.815
6.690
149.73
0.9721
0.9765
0.9689
0.9197
x=0.2
5.13
4.770
4.905
6.857
160.41
0.9944
0.9940
0.9925
0.9810
13.3
4.713
4.868
6.795
155.87
0.9826
0.9864
0.9835
0.9532
20.3
4.689
4.846
6.750
153.37
0.9776
0.9820
0.9769
0.9379
29.6
4.642
4.807
6.688
149.24
0.9679
0.9740
0.9680
0.9126
ever, since the total range of volume com-
pression in the present study was only 9%,
the data were better used to constrain only
1.02
1.00
0.98
I
1 X r-1 *,J „l J
Mg).xFexSi03 — Perovskite
877 K — 298KIK _
X- 0.0 +
-
V -820K «"
N. •773 K 01 * Hi*,TwnP
-
>s^ • 658 K
-
548KV544 K
I
382 KN» -
1 1 1 1 1 1
>°
^ 0.96
0.94
0.92
0.90
0 5 10 15 20 25 30
Pressure, GPa
Fig. 57. P-V-T relations for FexMg1]5Si03
perovskites. Open squares (x=0.2), open circles
(x=0.1) and crosses (x=0) are measurements at
298 K. Solid circles (x=0.1) are high-pressure data
at high temperatures as marked next to the data
points. The solid curve is the 298 K isotherm.
one parameter by fixing K\ The depend-
ence of Kq on the fixed Ko is shown in Fig.
56. The preferred value is Ko = 272.5(±2.4)
GPa when K = 4 is assumed. The curve and
o
data are plotted in Fig. 57. When a second-
order Birch equation of state (Birch, 1952)
is used instead of the Murnaghan equation
to fit the data, we obtained Ko = 273.4(±2.4)
GPa with fixed Ko = 4. Within the present
range of compression, the Murnaghan and
Birch equations are indistinguishable.
A similar least-squares fit for the iso-
thermal compression data of the three lat-
tice parameters yields:
bao = L291(0.02)GPa 7, K'ao= 11.4;
bbo = 1. 053(0.011 )GPa -1, K'b0= 13.9;
86 CARNEGIE INSTITUTION
Table 9. Lattice Parameters and Unit-Cell Volume of Fe^M^SiC^-Perovskite at High
Pressures and Temperatures.
/>,GPa
t,k
a> A
b, A
c, A
v,A3
alao
bib.
clcm
v/v#
7.2
298
4.648
4.815
6.690
149.73
0.9721
0.9765
0.9689
0.9197
26.8
382
4.661
4.802
6.696
150.46
0.9749
0.9777
0.9698
0.9243
23.9
544
4.665
4.830
6.730
151.63
0.9757
0.9795
0.9746
0.9315
22.5
548
4.668
4.829
6.735
151.81
0.9762
0.9793
0.9755
0.9325
15.1
658
4.737
4.872
6.790
156.72
0.9908
0.9881
0.9834
0.9627
11.4
773
4.745
4.894
6.829
158.56
0.9923
0.9925
0.9890
0.9740
8.38
820
4.770
4.908
6.863
160.66
0.9975
0.9954
0.9939
0.9869
4.19
877
4.794
4.937
6.916
163.72
1.0027
1.0014
1.0016
1.0057
Table 10. Zero-Pressure Bulk Modulus and Linear Compressibilities of Fe^Mg^SiC^
Perovskite at 298 K.
o
FT
o
b„
K
ba
P
max
rneda
Sample
X
GPa
TPa»
TPa-»
TPa1
GPa
2461
1.31
1.20
1.56
0
single xl.
0
2472
4
1.41
1.07
1.57
10
M-E-W
single xl.
0
2543
4
1.30
1.04
1.24
13
M-E, Ne
single xl.
0
258*
4
1.58
1.19
1.10
7
M-E
powder
0
2665
3.9
—
—
—
112
none
powder
0.12
272«
4
1.29
1.05
1.33
30
Ne
powder
0,0.1,0.2
(1) Adiabatic bulk modulus and compressibilities measured with Brillouin scattering tech-
nique,
(2)-(6) Isothermal moduli and compressibilities determined by fitting P-V-a-b-c data meas-
ured with x-ray diffraction techniques.
M-E: methanol ethanol mixture; -H: -water.
Reference
(1) Yeganeh-Haeri et al. (1989)
(2)Kudohefa/.(1987)
(3) Ross and Hazen (1989)
(4)Yagiera/.(1982)
(5) Knittle and Jeanloz (1987)
(6) Present study
GEOPHYSICAL LABORATORY
87
1.01
1.00
o
0.94
0.97
I
1 A J
Mg^Fe
xSi°3-
— Perovskite
•
877 K
• 820 k ;•;
^V . 773 K oi
+\~ * 658 K
— 298 K fit
D
o _
•High Temp
-
o\
548 K
v. .544 K
XsVo382j*
I
a
I I I
1
i r
0 5 10 15 20 25 30
Pressure, GPa
1.01
1.00
,QO0.99 -
0.94 -
0.97 -
1
1.01
1.00
o°0.99
o
0.94
0.97
Mg^^ejjSiOg — Perovskite
• 877K
— 298 K fit
X- 0.0 +
02 O
.820K 0.1 O
•770 1/ °-1 • HighTsmp
658 K
544 K _
\.382K
J L
0 5 10 15 20 25 30
Pressure, GPa
i i I i i r
Mg^xFexSiOj — Perovskite
• 877 K x. oo +
0 2 o
\ + ,820 K 0, . High Temp
548 KXV 544 K
382 K^^>
L _t
0 5 10 15 20 25 30
Pressure, GPa
Fig. 58. Compression of lattice parameters, (a)
a. (b) b. (c) c. See Fig. 57 for notation.
1.026
1.020
oo
8 1.015
^ 1.010
>°-
1.005 r-
1.000
MaJoFeJ SiO| — Perovskite
- 09 af^ 4.19 GPa %
8.38 GPa /6 _
15.1 GPa° X°
26.8 GPa ^
° ^^° 23.9 GPa
r-^""" ° 22.5 GPa
27.2 GPa
I I I I I I
0.995
280 380 480 580 680 780 880
Temperature, K
Fig. 59. Thermal expansion, V ^V m, as a func-
tion of temperature at various pressures. The
value of P is marked at each data point.
bco = 1.330(0. 0l8)GPa ■' K'c0= 11.0;
bx = -(dlnx/dlP)T
The three compression curves and data
points are plotted in Fig. 58A-C.
A study for simultaneous high P-T dif-
fraction measurements was also conducted
for Fe01Mg09SiO3 perovskite. At tempera-
tures above 500K, however, the gold stan-
dard tended to anneal to larger crystals. The
effect can be seen in Fig. 55, in which the
298 K diffraction peak of gold is broad due
to the small grain size, and the 877K peak
of gold is sharpened. The results of lattice
parameters and volume as a function of
pressure and temperature were listed in
Table 9 and plotted in Fig. 58. The thermal
expansion as a function of temperature at
various pressures is plotted in Fig. 59.
The orthorhombic (Fe,Mg)Si03
perovskite has the axial ratio a:b:c =
0.97: 1 : 1 .4 1 , which differs from the equiva-
lent ratio of the ideal cubic perovskite,
88
CARNEGIE INSTITUTION
1:1:1.41, mainly in that the a axis is 3 %
smaller. The present results clearly show
that the orthorhombic (Fe,Mg)Si03
perovskite is elastically anisotropic with
the b axis being the least compressible, ~ 25
% less compressible than a or c. The com-
pressibilities of a and c are similar, with c
slightly more compressible. With such
differential compressibilities, the differ-
ence between a and b becomes even larger
and the structure becomes more distorted at
higher pressures. Results from other stud-
ies of linear compressibilities of lattice
parameters of (Fe,Mg)Si03 perovskite are
listed in Table 10 for comparison.
Nonhydro static pressure has major ef-
fect on the relative compressibilities of the
lattice parameters. Pressure conditions in
the diffraction measurements of Yagi etal.
(1982) became clearly nonhydrostatic be-
tween 7.5 and 9 GPa, and thus only data
below 7.5 GPa were used for the calcula-
tions of the compressibility. The pressure
conditions in the measurements of Knittle
and Jeanloz (1987) were nonhydrostatic.
The relative axial compressibility would
be different from the hydrostatic one. In
addition, without the resolution to separate
the orthorhombic splitting of the equiva-
lent cubic diffraction peaks, relative com-
pressibilities of lattice parameters could
not be accurately determined.
In summary, the differences in Ko among
various studies are larger than the claimed
uncertainty in each study, but are much
smaller than those typically observed in
other materials when compared with meas-
urements from different laboratories. It is
also important to emphasize that although
the extrapolated zero-pressure parameters
are useful for comparisons with low-pres-
sure data, they do not carry any specific
significance in high-pressure experiments.
Since the purpose of studying the
(Fe,Mg)Si03 perovskite for solid-earth
geophysics is to assess its role in the lower
mantle, the more important parameters are
the density and bulk modulus of perovskite
with the appropriate composition above 20
GPa. These parameters were measured
directly in the present study.
References
Birch, R, Elasticity and constitution of the Earth's
interior, /. Geophys. Res., 57, 227-286, 1952.
Hemley, R. J., M. D. Jackson, and R. G. Gordon,
Theoretical study of the structure, lattice dy-
namics, and equations of state of perovskite-
type MgSi03 and CaSi03, Phys. Chem. Miner-
als, 74,2-12, 1987.
Hemley, R. J., C. S. Zha, A. P. Jephcoat, H. K.
Mao, L. W. Finger, and D. E. Cox, X-ray dif-
fraction and equation of state of solid neon to
U0GPsL,Phys.Rev.B,39, 11820-11827, 1989.
Ito, E., and D. J. Weidner, Crystal growth of
MgSi03 perovskite, Geophys. Res. Lett., 11,
464-466, 1986.
Jamieson, J. C, J. N. Fritz, and M. H. Manghnani,
Pressure measurement at high temperature in x-
ray diffraction studies: Gold as a primary stan-
dard, in High-Pressure Research in Geophys-
ics, S. Akimoto, and M. H. Manghnani, eds.,
Reidel Publ., Boston, pp. 27-48, 1982.
Jephcoat, A. P., H. K. Mao, and P. M. Bell,
Operation of the megabar diamond- anvil cell, in
Hydrothermal Experimental Techniques, G. C.
Ulmer and H. L. Barnes, eds., Wiley-Inter-
science, New York, Chapter 1 1 , pp. 469-506,
1987.
Knittle, E., and R. Jeanloz, Synthesis and equation
of state of (Mg,Fe)Si03 perovskite to over 100
gigapascals, Science, 235, 668-670, 1987.
GEOPHYSICAL LABORATORY
89
Knittle, E., R. Jeanloz, and G. L. Smith, The
thermal expansion of silicate perovskite and
stratification of the Earth's mantle, Nature, 319,
214-216, 1986.
Kudoh, Y., E. Ito, and H. Takeda, Effect of pres-
sure on the crystal structure of perovskite-type
MgSi03. Phys. Chem. Minerals, 14, 1987.
Mao, H. K., T. Yagi, and P. M. Bell, Mineralogy
of the Earth's deep mantle: quenching experi-
ments on mineral compositions at high pres-
sures and temperature, Carnegie Instn. Wash-
ington Year Book, 76, 502-504, 1977.
Mao, H. K., J. Xu, and P. M. Bell, Calibration of
the ruby pressure gauge to 800 kbar under qua-
sihydrostatic conditions, /. Geophys. Res., 91,
4673-4676, 1986.
Ming, L. C, M. H. Manghnani, S. B. Qadri, E. F.
Skelton, J. C. Jamieson, and J. Balogh, Gold as
a reliable internal pressure calibrant at high
temperatures, J. Appl. Phys., 54, 4390-4397,
1983.
Murnaghan, F. D., The compressibility of media
under extreme pressures, Proc. Nat. Acad. Sci.
USA, 30,244-241, 1944.
Ross, N. L., and R. M. Hazen, High-pressure
crystal chemistry of MgSi03 perovskite, Phys.
Chem. Minerals, 16, 415-420, 1989.
Wolf, G., and M. Bukowinski, Theoretical study
of the structural properties and equations of
state of MgSi03 and CaSi03 perovskites: impli-
cations for lower mantle composition, in High-
Pressure Research in Mineral Physics, M. H.
Manghnani and Y. Syono, eds., Terra Scientific
Publishing Company (TERRAPUB), Tokyo/
American Geophysical Union, Washington, D.
C, pp. 313-331, 1987.
Yagi, T., H. K. Mao, and P. M. Bell, Crystal
structure of MgSi03 perovskite, Carnegie Instn.
Washington Year Book, 76, 516-519, 1977.
Yagi, T., H. K. Mao, and P. M. Bell, Hydrostatic
compression of perovskite-type MgSi03, in Ad-
vances in Physical Geochemistry, Vol. 2, S. K..
Saxena, ed., Springer- Verlag, New York, pp.
317-325, 1982.
Yeganeh-Haeri, A., D. J. Weidner, and E. Ito,
Elasticity of MgSi03 in the perovskite structure,
Science, 243, 787- 789, 1989.
Single Crystal X-ray Diffraction
Study of A New Hydrous Silicate,
Phase E
Yasuhiro Kudoh, Larry W. Finger, Robert
M. Hazen, Charles T. Prewitt,
and Masami Kanzakv
Ringwood and Major (1967) investi-
gated the system MgO - Si02 - H20 at
pressures between 10 and 18 GPa and at
temperatures from 600 to 1100°C. They
discovered three new phases denoted A, B
and C and suggested phase B as a probable
host mineral for H20 in the deep mantle.
Subsequently, Yamamoto and Akimoto
(1974) reported the discovery of phase D
with chemical composition close to that of
chondrodite Mg(F,OH)22Mg2Si04 Among
these, phase B is known to have a tempera-
ture-pressure stability field higher than the
others (Akimoto and Akaogi, 1980) but its
Mg/Si ratio is 3.0 (Finger et ai, 1989).
More recently Kanzaki ( 1 989) investigated
phase relations in the system MgO - Si02 -
H20 up to 15 GPa and 1500°C using a
uniaxial split-sphere multi-anvil apparatus
and discovered a new phase that was syn-
thesized at 1000°C pressures higher than
13 GPa, which he denoted phase E. This
hydrous silicate phase has an Mg/Si ratio
less than 2, close to that of the mantle.
In this paper we report the crystal struc-
tural and crystal chemical aspects of phase
E, which is a potential host for H20 in the
mantle transition zone.
* University of Alberta, Edmonton, Alberta,
Canada
90
CARNEGIE INSTITUTION
Experimental Results
Phase E (Kanzaki, 1989) was synthe-
sized at 15 GPa and 1000°C using a uniax-
ial split-sphere multi-anvil apparatus. The
starting material was a stoichiometric
mixture of high-purity Si02and 2Mg(OH)2.
A single crystal with approximate dimen-
sions 160x110x40 jiim was used for x-ray
diffraction, x-ray photographs indicated
systematic absences of reflections with
-h+k+1 = 3n for hkl, which is consistent
with a rhombohedral space group, indicat-
ing that the space group of phase E is either
R3m, R3 orR5m. Intensity statistics did not
indicate a definite conclusion as to whether
the crystal has center of symmetry or not.
The cell constants determined by means of
a four-circle diffractometer, using 14 re-
flections of with 20 from 26-36°, with a
wavelength of 0.7093 A were a = 2.959(1)
A, c=\ 3.844(2) A, V = 104.98(6) A3. They
are close to those refined from powder
diffraction data [a=2.9701(l) A,
c=13.842(l) A, V = 106.05(4) A3]. Electron
microprobe analysis of the specimen at 12
sampling points showed homogeneous
chemical composition with MgO 48 .4 wt. % ,
Si02 40.1 wt.% (no other element was
detected), yielding a total of 88.6. When
closest-packing of oxygen atoms is as-
sumed, the short axis allows only one
oxygen atom per layer in the (2ra2plane and
the length of the c axis allows six oxygen
layers; therefore, there are six oxygen atoms
per unit cell. If the difference of the total
weight is ascribed to H20, the formula is
calculated to be Mg227SiL26H240O6. Crystal
structure analysis is now in progress.
Since phase E was observed co-existing
with stishovite in run products of a silica-
rich starting material (Kanzaki, 1989), phase
E is considered to have a stability field
corresponding to the mantle transition zone.
Phase E has geophysical importance be-
cause it is a hydrous phase with a chemical
composition close to that of the mantle. Its
density, 2.89 g/cm3, is rather low, however.
The composition of phase E can be
derived from a brucite starting point with a
cell content of Mg3(OH)6. If a magnesium
atom is removed, silicon atoms in tetrahe-
dral coordination can be placed over this
vacancy. The two possible reactions are:
Mg(OH)2+Si4+«=>
Si02 + Mg2+ + 2H+,
(1)
which corresponds to a single Si per Mg
removed, or
2Mg(OH)2 + 2Si4+ <=»
MgSi2042+ + Mg2+ + 4H+, (2)
which results when two Si are involved.
The measured composition corresponds to
1.73 Si atoms added for each Mg removed
from the hypothesized starting position,
which indicates that both mechanisms
apply; however, the second one is more
important. From the diffraction data, it is
obvious that these substitutions result in
short-range order. Although clusters of
defects are expected, the size of the cluster
does not result in diffuse scattering with
appreciable intensity. Long-term x-ray
photographs indicate the existence of dif-
fuse maxima with the a axis seven times
longer and the c axis doubled, as compared
to the subcell.
GEOPHYSICAL LABORATORY
91
References
Akimoto, S., and M. Akaogi, The system Mg2Si04
- MgO - H20 at high pressures and temperatures-
possible hydrous magnesian silicates in the
mantle transition zone, Phys. Earth Planet. In-
ter., 23, 268-275, 1980.
Finger, L. W., J. Ko, R. M. Hazen, T. Gasparik, R.
J. Hemley, C. T. Prewitt, and D. J. Weidner,
Water in the upper mantle: crystal chemistry of
phase B and a new anhydrous magnesium silicate,
Nature, in review.
Kanzaki, M., High pressure phase relations in the
system MgO - Si02 - H20, EOS, 70, 508, 1989.
Ringwood, A. E., and A. Major, High-pressure
reconnaissance investigations in the system
Mg2Si04 - MgO - H20, Earth Planet. Sci. Lett.,
2, 130-133, 1967.
Yamamoto, K., and S. Akimoto, The system MgO
- Si02 - HzO at high pressures and temperatures-
stability field for hydroxyl-chondrodite, hy-
droxyl-clinohumite and 10 A -phase, Am. /. Sci.,
277,288-312,1977.
Spectroscopic Evidence for a new
New High-pressure Magnesium Silicate
Phase
James D. Kubicki and Russell J. Hemley
Experimental constraints on the miner-
alogy of the lower mantle have mainly been
obtained from high-pressure phases that
are quenchable to ambient pressure and
temperature (Yagi etai, 1979). A series of
experiments have been started using the
laser-heated, diamond-anvil cell and mi-
cro-Raman spectroscopy to investigate the
possible existence of non-quenchable
phases under lower mantle conditions.
Micro-Raman spectroscopy was employed
to probe the sample after melting and
quenching at high pressure. With micro-
Raman spectroscopy, it is possible to study
any heterogeneities in the sample that may
be induced by laser-heating. The in situ
nature and spatial resolution of the tech-
nique are ideally suited for identification of
small amounts of non-quenchable high-
pressure phases.
Approximately 1 weight percent plati-
num black was mixed and ground with
MgSi03 glass to absorb Nd-YAG laser-
radiation (1.06 mm line). The sample was
compressed to 35 GPa at room temperature
and the Raman spectra measured (Fig. 60).
The platinum fluoresces under the argon
laser (514.5 nm line) thereby reducing the
signal-to-noise ratio, and obscuring the
broad band between 800 and 1100 cm"1
observed in previous spectra (Kubicki et
ai, 1987). The sample was then heated
with a Nd-YAG laser. Regions exposed to
the highest temperatures (estimated to be
above 2000K) formed rounded spots more
transparent than the surrounding unheated
MgSiO} 35 GPa Raman spectra
After heating 764
626
i Before heating
50 250 450 650 850 10501250
Frequency (cm"1)
Fig. 60. Raman spectrum of MgSi03 glass mixed
with Pt black at 35 GPa before laser-heating and
after heating with Nd/Y AG laser and m^/m quench-
ing. A linear, sloping baseline due to platinum
fluorescence has been subtracted from both spec-
tra.
92
CARNEGIE INSTITUTION
regions. Platinum was also concentrated
around the inner boundary between the
heated and unheated region. From these
observations we conclude that melting has
occurred at the most intensely heated re-
gions. Heating to lower temperatures was
also performed to drive subsolidus trans-
formations such as the crystallization of
MgSi03-perovskite. In this case, the heated
region was visibly darker than either the
starting material or melted regions and the
Raman spectra are identifiable as MgSi03-
perovskite (Hemley et ah, 1989).
The Raman spectrum of the melted and
quenched sample (Fig. 60) is significantly
different from the unheated sample. Five
major peaks (193, 375, 626, 764, and 926
cm0 appear in the spectrum. These were
not present before laser-heating. These
peaks also became sharper and more in-
tense with time under the argon-laser radia-
tion used to measure the Raman spectrum.
Spectra were also measured upon decom-
pression. Broad bands centered at 631 and
997 cm"1 from MgSi03 glass (Kubicki et
ai, 1987) and peaks of MgSi03-perovskite
were observed. No unassigned peaks exist
in the spectrum of the heated region after
decompression that could be attributed to
the high-pressure, laser-heated spectrum.
There are several possible explanations
for the spectrum induced by laser-heating.
The unidentified spectrum may be of a
glass quenched from the high-pressure melt.
In this case, the spectrum of the glass melted
at high-pressure would be very different
from the glass compressed at room tem-
perature (Kubicki etal.y 1987). The sharp-
ness of the Raman peaks, however, indicate
some degree of crystalline order. Crystal-
lites may have formed under the very rapid
heating and quenching conditions of the
experiment. Very fine-grained crystals (i.e.,
less than 1 mm) and poorly crystalline ma-
terials exhibit peak-broadening effects
associated with the breakdown of crystal-
line selection rules (Hemley et aL, 1986).
The fact that the new peaks become more
intense and sharper with time under the
argon laser suggests that annealing or re-
crystallization has taken place resulting
from a small degree of heating of the sample
by the argon laser. Under these high-
temperature and rapid quench conditions,
the possibility of crystallizing a metastable
phase also exists. Another possibility is
that the high-temperature and pressure
conditions of the experiment caused a reac-
tion between the MgSi03 glass and the Pt
black. Formation of a platinum-magne-
sium silicate under the experimental condi-
tions can not be ruled out.
Raman frequencies of silicates at high-
pressure are compared with this new spec-
trum in Table 11. Although certain peaks
may be correlated with known phases, no
combination of Raman peaks for phases
previously measured will explain the entire
spectrum. For example, the peaks at 193
and 626 cm1 may be correlated with the
high-pressure Raman spectra of stishovite
and MgSi03-perovskite, respectively (Table
11); but the other peaks of stishovite and
MgSi03-perovskite at 35 GPa are not pres-
ent in this spectrum. The intensity pattern
of the spectrum is also broadly similar to
that of stishovite. The structure of the phase
may have features in common with other
high-pressure silicates, such as octahedral
Si, which gives rise to these similar Raman
GEOPHYSICAL LABORATORY
93
Table 11. Comparison of Raman frequencies for stishovite, MgSi03-perovskite, orthoenstatite, and the
spectrum of this study all at 35 GPa (s, strong; m, moderate;w, weak; vw, very weak; b, broad).
MgSiO, Orthoenstatite* Stishoviteb MgSiO -Perovskite0
This Study
—
195 m
—
193 w
265 w
—
342 w
—
360 w
—
353 w
375 b
—
—
454 w
—
525 m
—
464 s
—
—
647 m
646 s
626 b
—
—
648 vw
—
770 s
855 s
—
764 s
1140 m
969 w
1040 vw
926 m
"Kubicki (unpublished data)
bHemley (1987)
cHemley etal (1989)
features. The overall crystal structure,
however, must be different to account for
all the peaks observed. It is not known if the
MgSi03-phases melt congruently at high
pressures so the new phase may be en-
riched in either Mg or Si relative to MgSi03.
In addition, high temperature gradients
present during laser-heating could have
altered the composition of the sample lo-
cally due to the Soret effect (Heinz and
Jeanloz, 1987). The heterogeneity of the
sample in this laser-heated region also al-
lows the possibility of a combination of
phases giving rise to this Raman spectrum.
The spectrum most probably arises from
a previously unidentified phase that ap-
pears to be non-quenchable from high pres-
sures. Also, it has been demonstrated that
micro-Raman spectroscopy, used in com-
bination with the laser-heated, diamond-
anvil cell, is a useful technique for probing
the possible existence of non-quenchable,
high-pressure phases. Future work to ana-
lyze the structure of newly identified phases
should be carried out with in situ, high-
pressure X-ray diffraction and TEM tech-
niques.
References
Hemley, R. J., H. K. Mao, and E. C. T. Chao,
Raman spectrum of natural and synthetic
stishovhe,Phys.Chem.Min., 73,285-290, 1986.
Hemley, R. J., Pressure dependence of Raman
spectra of Si02 polymorphs: Quartz, coesite,
and stishovite, in High-Pressure Research in
Mineral Physics, M. H. Manghnani and Y. Syono,
eds., Terra Scientific Publishing Co. (TER-
RAPUB), Tokyo/American Geophysical Un-
ion, Washington, D. C, 1987.
Hemley, R. J., R. E. Cohen, A. Yeganeh-Haeri, H.
94
K. Mao, D. J. Weidner, and E. Ito, Raman
spectroscopy and lattice dynamics of MgSi03-
perovskite at high pressure, in Perovskite: A
Structure of Great Interest to Geophysics and
Materials Science, A. Navrotsky and D. A.
Weidner, eds., Am. Geophys. Union, Washing-
ton, D. C, pp. 35-53, 1989.
Heinz, D. L. and R. Jeanloz, Measurement of the
melting curve of Mg0 9Fe0 ^iC^ at lower mantle
conditions and its geophysical implications, /.
Geophys. Res., 92, 11437-11444, 1987.
Kubicki, J. D., R. J. Hemley, and H. K. Mao, In
situ, high-pressure Raman spectroscopy of
MgSi03, CaSi03, and CaMgSi206 glasses,
(abstract) EOS, Trans. Am. Geophys. Union, 68,
1456, 1987.
Yagi, T., P. M. Bell, and H. K. Mao, Phase rela-
tions in the system MgO - FeO - Si02 between
150 and 700 kbar at 1000°C, Carnegie Instn.
Washington Year Book, 78, 614-618, 1979.
Compression and Polymorphism of
CaSi03 at High Pressures
and Temperatures
Liang-chen Chen, Ho-kwang Mao, and
Russell J. Hemley
CARNEGIE INSTITUTION
CaSi03 at about 16 GPa and obtained a
diffraction pattern consistent with the cu-
bic perovskite. Further, they showed that
the phase reverted to a glass on release
from high pressure; this observation was
confirmed in later diamond-anvil studies
(Mao et al., 1977). Further information on
the properties of CaSi03 at high pressure
has been obtained from theoretical calcula-
tions in which only the perovskite-type
phase has been examined (Hemley et al.,
1 987 ). The present study was undertaken to
investigate phase transitions and pressure-
volume equation of state of CaSi03 up to
conditions equivalent to those at the core-
mantle boundary. We have also studied the
phase transitions at low pressure (< 1 5 GPa)
and examined the range of stability of the
low-pressure polymorphs. A new phase
named CaSi03(III) has been identified. In
addition, we have investigated the onset of
vitrification of CaSi03-perovskite decom-
pression below its stability field.
Experimental Methods and Results
Introduction
Numerous experimental studies have
determined high pressure properties of
magnesium and iron-magnesium silicates,
but much less information is available on
those of CaSi03 at under upper and lower
mantle conditions. Ringwood and Major
(1967) found a high pressure modification
of CaSi03(I) (wollastonite) at about 3 GPa.
The structure was analyzed by Trojer (1 969)
and found to be related to walstromite.
Later, Liu and Ringwood (1975) measured
in situ x-ray diffraction from laser heated
In the first set of experiments, natural
CaSi03(I) (Wollastonite) ground and mixed
with 1% platinum black was used as a
starting material. The samples were loaded
in the diamond-anvil cell, and were heated
with the Nd- YAG laser (A = 1 .06/xm) after
each pressure increment. Powder x-ray
diffraction data were collected using a sealed
tube MoKa source and film techniques with
a camera radius of about 50 mm. Pressures
were measured with the ruby fluorescence
technique.
The diffraction data demonstrate that
following increasing pressure and heating,
GEOPHYSICAL LABORATORY
95
1 1
(a) 0 GPa, CaSi03(l) (Wollastonite)
1
i
(c) 8.5 GPa,
I
CaSi03(lll)
2.753
g
ft
1.982
1.608
1 830
CO
r
2507o.
2.063 i 1-936 R1 569
8.1361 All il 1.761 II.
CD
\ il 2241,.! e5 A 1 T.
J \jJl 1680| \k
L_
f.
I
V1! " V 1 484
Ul 1.448
i\A1 402
1
15
20 25
29 (°)
30
20 25
29 (°)
30
I I
(b) 6.8 GPa,
CaSiO
I
3(ll) (Walstromite)
2.922
3*
1.810
(O
12 819 2256 2013 197<M
1.716
1 k2158 1955 J \
I l 2.490 Alil?072JF~ J "
\?** 1.516
■♦— <
_c
' I I
1
15
20 25
29 (°)
30
CO
c
(d) 10.4 GPa, CaSi03 (III)
1.813 1601
20 25
29 (°)
30
CO
c
(D
•♦— »
c
1 1
(f) 0 GPa, CaSi03(ll) (Walstromite)
2 194 1-97« ,
2.112 I 1 893/
2308?
15
20 25
29 (°)
30
fig. 61. Examples of x-ray diffraction patterns of CaSi03 as a function of pressure. The patterns were
digitized from film with an automated densitometer.
walstromite-type CaSi03(II) is stable up to
about 7 GPa at 300K. Above 8 GPa
CaSi03(II) was observed to convert to a
new non-quenchable phase named
CaSi03(III) which is stable to 1 1 GPa. This
phase converts to walstromite-type
CaSi03(II) when quenched to ambient
conditions. A list of the x-ray diffraction
lines observed at 8.3 GPa and at ambient
conditions is shown in Table 1 2. The simple
cubic perovskite phase of CaSi03 appeared
above 11 GPa. When the pressure was
96
CARNEGIE INSTITUTION
Table 12. The d-spacings and relative intensities for CaSi03(III)
CaSi03(ni)
CaSi03(n) (Walstromite)
at
8.3 GPa
;
atO.lMPa
/
d(k)
/
d(A)
2
3.050
10
2.916
10
3.002
9
2.750
5
2.889
1
2.692
1
2.762
5
2.560
3
2.608
5
2.478
5
2.393
2
2.448
7
2.240
1
2.256
4
2.122
3
2.194
5
2.062
1
2.112
5
1.980
1
2.038
5
1.919
1
1.974
1
1.893
7
1.831
4
1.823
7
1.606
4
1.644
2
1.571
4
1.597
released at room temperature, the perovskite
structure remained at 0.8 GPa and disap-
peared at zero pressure. X-ray diffraction
patterns of CaSi03(I), CaSi03(H),
CaSi03(]H) and CaSi03-perovskite formed
at different pressures are compared in Fig.
61. Changes in the measured d-spacings
for CaSi03 with increasing pressure are
shown in Fig. 62.
In a second set of experiments, the
perovskite phase was synthesized at about
17 GPa, the sample was examined by x-ray
diffraction up to 40 GPa and on pressure
release to ambient conditions. Pressure was
again measured from ruby fluorescence.
The sample was heated with the Nd-YAG
laser after each pressure change in order to
accelerate transformations in the material
to reduce pressure inhomogeneity. Four
diffraction lines of CaSi03-perovskite (110,
111, 200, and 211) were measured for each
film. Because of the incompressibility of
the perovskite phase, additional measure-
ments at higher pressure were required to
constrain the equation of state of this phase.
These measurements also permitted us to
examine possible polymorphic transfor-
mations in CaSi03 under lower mantle
pressure and temperature.
A third set of experiments was therefore
performed; these were similar to the first
set except that 3% platinum black was
added to the sample. Two x-ray diffraction
lines from the platinum (111 and 200) were
observed in addition to the four diffraction
lines from CaSi03-perovskite. The plati-
GEOPHYSICAL LABORATORY
97
3.00
2.75
co 2.50
D)
I 2.25
Q.
CO
~6 2.00
1.75 h
1.50
CaSi03(ll)
(III)
Perov.
0.0 2.5 5.0 7.5 10.0 12.5
Pressure, GPa
fig. 62. Observed d- spacings of CaSi03 as a func-
tion pressure.
45.0
_ SL
CaSiO 3 — Perovskite -
•V
281 (+4)GPa, KqS4
o< 42.5
VQ= 45.37 (±0.08) A3
•
- B-M fit
CD
i 40.0
D LP data
_
o HP data
3
O
>
37.5
35.0
I
I
i i i i®*^.
25 50 75 100 125 150
Pressure, GPa
Fig. 63. Pressure-volume data for CaSi03-
perovskite at 300K. Squares (LP): low and moder-
ate pressure data. Circles (H): high-pressure data.
Curve: Third-order Birch-Murnghan equation fit.
num diffraction served as an internal pres-
sure standard; the new ultrahigh-pressure
equation of state of platinum, recently
developed by Lawrence Livermore Labo-
ratory was used to calculate the pressure
(Holmes, et al., 1989). Again, the sample
was laser-heated after each pressure change
in order to drive possible high-pressure
transformations. The maximum pressure
reached in this set of experiments is 134
GPa, the pressure of the core -mantle bound-
ary. The diffraction data indicate that Ca-
Si03 is stable in the cubic perovskite over
this entire pressure range.
The pressure -volume data from the low-
and high-pressure x-ray diffraction meas-
urements for the perovskite phase are plot-
ted in Fig. 63. All data points were fitted
with a third-order Birch-Murnaghan equa-
tion of state. The zero-pressure parameters
are: Vo = 45.31 (±0.08) A3, Ko= 281 (±4)
GPa, and /T = 4.3 (±0.2), density po = 4.258
(±0.008) g/cm3.
Discussion
The pressure range of stability of the
CaSi03 polymorphs identified in this study
are indicated in Fig. 62. Above 8 GPa,
walstromite-type CaSi03(II) converts to a
new non-quenchable phase CaSi03(III)
which is stable to 1 1 GPa. Tamai and Yagi
(1988) have also obtained evidence for
CaSi03(III), although they report its range
of stability at 10 to 13.8 GPa. The diffrac-
tion patterns measured at high pressure
indicate that the structure is complex and
cannot be determined from the. available
powder diffraction data. Experimental
determination of the density, structure, and
elastic properties of this phase is essential
in order to assess the possible role of this
phase in the upper mantle. It is useful to
note that recent single-crystal diffraction
studies suggest that numerous upper mantle
minerals may indeed have complex struc-
tures (Finger et ai, in preparation).
98
CARNEGIE INSTITUTION
The perovskite-type of CaSi03 remains
in the simple cubic structure up to at least
134 GPa. During pressure release, the
perovskite remains metastable at pressures
close to 0.1 MPa. It is also of interest to
compare the densities of the mixed oxide
assemblage of CaO + Si02 (stishovite) with
that of CaSi03 (Richet etal, 1988; Bass et
ai, 1981). At -80 GPa the density of the
oxide assemblage exceeds that of CaSi03-
perovskite. This result may indicate that
the extrapolated equation of state of
stishovite overestimates the densities at
high pressure. Alternatively, the compari-
son may indicate that CaSi03-perovskite
may disproportionate at pressures above
134 GPa (outside the range of the lower
mantle).
Our results strongly suggest that cubic
CaSi03-perovskite is a stable phase through-
out the entire lower mantle. Under stable
conditions, the CaSi03 -perovskite exists in
the lower mantle as a major separate phase
with abundance only next to ferromagne-
sian silicate perovskite and probably to
magnesiowustite, depending on the chemi-
cal composition model (Anderson, 1989).
Further, the cubic CaSi03-perovskite phase
may also be significant as a reservoir for
rare earth elements in the lower mantle
(Mao et al., this Report). The 300K equa-
tion of state of CaSi03 -perovskite is close
to that of (Mg090Fe010) Si03-perovskite (Mao
et al.y this Report). We also note that the po
of CaSi03-perovskite, 4.26 g/cm3, is in
excellent agreement with the inferred 300K,
zero-pressure density of the lower mantle.
Hence, CaSi03-perovskite must be consid-
ered an "invisible" component, in terms of
density and bulk modulus constraints, in
the lower mantle.
References
Anderson, D. L., Composition of the Earth, Sci-
ence, 245,367-370, 1989.
Bass, J. D., R. C. Liebermann, D. J. Weidner, and
S. J. Finch, Elastic properties from acoustic and
volume compression experiments, Phys. Earth
Planet. Inter., 25, 140-158, 1981.
Hemley, R. J., M. D. Jackson, and R. G. Gordon,
Theoretical study of the structure, lattice dy-
namics, and equations of state of perovskite-
type MgSi03 and CaSi03, Phys. Chem. Miner-
als, 14,2-12, mi.
Holmes, N. C, J. A. Moriarty, G. R. Gathers, and
W. J. Nellis, The equation of state of platinum to
660 GPa (6.6 Mbar), /. Appl. Phys., in press,
1989.
Liu, L. and A. E. Ringwood, Synthesis of a
perovskite-type polymorph of CaSi03, Earth
Planet. Sci. Lett., 28, 209-211, 1975.
Mao, H. K., T. Yagi, and P. M. Bell, Mineralogy
of the Earth's deep mantle: quenching experi-
ments on mineral compositions at high pres-
sures and temperature, Carnegie Instn. Wash-
ington Year Book, 76, 502-504, 1977.
Mao, H. K., P. M. Bell, J. W. Shaner, and D. J.
Steinberg, Specific volume measurements of
Cu, Mo, Pd, and Ag and calibration of the ruby
R, fluorescence pressure gauge from 0.06 to 1
Mbar, /. Appl. Phys., 49, 3276-3283, 1978.
Richet, P., H. K. Mao, and P. M. Bell, Static
compression and equation of state of CaO to
1.35 Mbar,/. Geophys.Res., 75,279-288, 1988.
Ringwood, A. E., and A. Major, Some high-
pressure transformations of geophysical
significance, Earth Planet. Sci. Lett., 2, 106-
110,1967.
Tamai, H., and T. Yagi, High-pressure and high-
temperature phase relations in CaSi03 and
CaMgSi206 and elasticity of perovskite-type
CaSi03, Phys. Earth Planet. Inter., 54, 370-
377, 1989.
Trojer, F. J., The crystal structure of a high-
pressure polymorph of CaSi03, Z. Kristallogr.,
130, 185-206, 1969.
GEOPHYSICAL LABORATORY
99
The Polarized Raman Spectra of
Tourmaline
Mingsheng Peng, Ho-kwang Mao,
Liang-chen Chen,
and Edward C. T. Chao
Polarized Raman spectroscopy (PRS)
has been used extensively for structural
and compositional characterization of
minerals, (White, 1975, Mao etal, 1987,
Hemley, 1988). The Raman spectra of
minerals are generally analyzed in terms of
factor group analysis (McMillan, 1985).
For tourmaline with several structural
groupings, a general assumption is that its
Raman spectra are made of internal modes
of each of the individual structural units in
the crystal (Si6018, B03 OH), plus lattice
modes characteristic of the entire unit cell.
Each structural unit, [Si6018]12\ [B03]3- and
(OH)1", has its distinctive vibrational spec-
trum.
In this paper we present results of PRS
of samples of tourmaline from three differ-
ent geological occurrences in China, namely
granite pegmatites, hydrothermal veins, and
metamorphic skarns. Our interest is fo-
cussed on the correlation of the PRS to the
compositional and structural differences
among tourmalines, and the nature of or-
der-disorder of the OH ions in the tourma-
line structure in regards to the specific
geological occurrences.
Crystal Structure of Tourmaline and
Description of Samples
Tourmaline is a complex borosilicate of
aluminum varying considerably in compo-
sition with a general formula:
WX3Y6(B03)3Si6018(OH,F,Cl)4.
where W= Na and Ca; X= Mg, Fe2+, Mn,
Al, andFe3+; and
Y=Al,Fe3+,Cr,andV.
As shown by Buerger et al. (1962),
tourmaline has rhombohedral symmetry,
and is in the space group R3m - C3V. The
crystal structure is characterized by a layer
of six nearly regular Si04 tetrahedra in
hexagonal arrangement similar to that of a
phyllosilicate sheet. The octahedral layer
consists of three larger central octahedra
containing X cations, six smaller periph-
eral octahedra containing Y cations, and
three boron atoms. The three octahedra of
X cations (mainly Mg) shares edges and
forms a trigonal unit similar to a brucite
[Mg(OH)2]-like layer. The trigonal X octa-
hedra unit also share edges with the six Y
cations. Each of the boron atoms is in a 3-
fold coordination of oxygens at the vertices
of octahedra of this layer. The W cation and
OH are located along the 3 -fold axis of
symmetry in the middle of the unit cell. The
(OH) hydroxyl groups are confined to three
Mg(OH)204 octahedra lying in the same
layer as the three pairs of Al(OH)05 octa-
hedra.
*U.S. Geological Survey, Mail Stop 929, Reston,
VA 22920
100
CARNEGIE INSTITUTION
Table 13. Microprobe analysis of tourmaline of three different types
Types
Pegmatitic
Hydrothermal
Metamorphic
Sample No
T05
T04
T09
T06
T08
Color
Red
Lt. Green
Green
Blue
Deep Blue
Si02
37.60
37.05
35.33
34.37
35.1
fi02
0.22
0.50
0.61
0.32
0.21
A1203
33.30
31.63
33.09
29.83
32.89
FeO
2.19
4.89
4.79
5.43
6.43
MnO
3.30
0.23
0.37
0.18
0.4i
MgO
3.65
3.95
8.76
10.05
7.03
CaO
2.37
2.58
0.29
3.36
4.11
K20
0.11
0.20
0.30
0.64
0.29
Na20
2.86
2.18
1.66
1.48
1.68
Total
85.6
83.21
85.20
85.66
88.17
The pegmatitic tourmaline samples are
from Xinjiang Province in northwest China
(samples no. T05 and T04). The associ-
ated minerals are beryl, columbite, and
pollucite. The tourmaline crystals are al-
most of gem quality. Their color changes
from rose-red to green and blue along the c-
axis. Normal to the c-axis, color rings of the
same color occurs. The tourmaline from
the hydrothermal vein (T09) came from
Hunan province of China. It is associated
with quartz and beryl. The tourmaline crys-
tals exhibit prismatic habit.The color ranges
from light green to dark green. The tourma-
lines from metamorphic skarn (T06 and
T08) came from the tin deposit of Yunnan
province of China. The tourmalines have
the highest iron content among the three
types of samples. The associated minerals
are cassiterite, calcite, scapolite and diop-
side. Chemical compositions of tourmaline
samples are listed in Table 13.
Characteristics ofPRS of Tourmalines
and Assignment of Spectral Peaks
Polarized Raman spectra of tourmaline
samples are presented in Table 2. Raman
peaks are observed in the regions of 0-1200
cm'1 and 3400-3600 cm1. Representative
spectra are shown in Figs. 64 and 65.
The major peaks of the PRS in the 0-
1200 cm1 region are related to the [Si6018]12"
hexagonal rings (Table 14). Peak assign-
ments are based on the analysis of Raman
spectra of a powdered tourmaline sample
by Griffith (1969). In the present study,
intense Si-0 stretching vibration peaks are
observed at 1000-1200 cm1. Two symmet-
rical ring stretching peaks lie between 400
and 570 cm1. Two asymmetrical ring
stretching peaks lie at 962-999 cm1 and at
600-700 cm1. Two ring deformation stretch-
ing peaks are located between 220 - 380
cm1. These two ring deformation stretch-
GEOPHYSICAL LABORATORY
101
if)
c
Q)
C/)
c
CD
200 400 600 800 1000 1200
Wavenumbers, cm"1
200 400 600 800 1000 1200
Wavenumbers, cm"1
Fig. 64. Polarized Raman spectra of different types of tourmaline in N direction and in No direction,. The
ring deformation stretching peaks of [Si6Oj J are very strong at 220-3 §0 cm1. The PRS peak correspond-
ing to the stretching of the B-O bond in BO lies between 700-800 cm1.
ing peaks are very intense. At different po-
larization directions, the number of Raman
peaks are the same, but the positions of the
peaks shift, and the intensities of the peaks
vary. The PRS in the Ne direction is much
more intense than that of the N direction.
o
In addition, as the iron content in the tour-
maline increases, the spectral peak splits or
distorts, so that the PRS from the granite
pegmatite (Fig. 64; T05) are clearly differ-
ent from the PRS from the skarn metamor-
phic tourmaline (Fig. 64; T08) along the No
direction.
The PRS peak corresponding to the B-
O bond in [B03]3- lies between 700 and 800
cm1. Brethous etal. (1981) studied Raman
spectra of the synthetic system of B203 -
Si02 - Li20. By holding the Li20 content
constant but varying B203 and Si02 con-
tent, they found that the intensity of the 760
cm"1 peak increased with increasing B203
content, and that the intensity of the peaks
at 1040, 950, and 600 cm1 increased with
increasing Si02 content. Our finding re-
garding the ring vibrational peaks assigned
to Si-0 and B-0 vibrational peaks is gener-
ally consistent with that of Brethous et al.
(1981). However, the B-0 vibration peaks
102
CARNEGIE INSTITUTION
Table 14. Frequencies (cm1) of polarized Raman spectra of tourmaline in the Ne direction
Types
Pegmatitic
Hydrothermal
Metamorphic
Powdered
Samples
(Griffith, 1969)
Sample No.
T05
T04
T09
T06
T08
[Si6o18]12"
1049
1048
Vs (Si-O)
1114
1115
1082(s)
1016(s)
1016(s)
1040(6)
Ring Stretches*
543
540
563
510
527
526
569(5)
437
436
403
484
487
464(10)
404(s)
404(s)
Ring Stretches**
999
998
971
965
962
929(1)
671
685
638
672
633
669
634
682
Ring deformation
304
342(s)
372(s)
363(s)
363(s)
353(5)
314
306
306
340(l/2)
[B03]3-
253(s)
255(s)
220(s)
228(s)
228(s)
744
775
746
764
764
734(s)
737(s)
703(s)
693(s)
692(s)
[OH]1"
V2
3635
3636
3648
3629
3630
Vl
3573(s)
355 l(s)
3577
3589(s)
3562(s)
3555(s)
V3
3460
3472
3482
no
no
* = symmetrical stretching vibration
** = asymmetrical stretching vibration
GEOPHYSICAL LABORATORY
103
in tourmaline often split into two peaks.
The main splitting is probably due to the
variation of bond lengths between boron
and adjacent oxygen (B-02 bond length is
1.375 , B-08 bond length is 1.358 A). The
C3 symmetry of the boron atom is reduced
to C2v, and thus the peak splits into two.
Raman peaks in the region between
3400 and 3600 cm1 are assigned to OH
stretching. The peak positions and multi-
plicities in this region are complicated due
to the combined influence of octahedral
site occupancies, in- and out-of-phase ef-
fects, Al/Si ordering, OH/OH2 ordering,
alkali cations (K, Na, Li), and the extent of
Al substitution in octahedral and/or tetra-
hedral sites. The PRS in tourmaline struc-
ture peaks assigned to (OH) show clear
differences parallel to No as compared to
Ne. The intensities of the main (OH) band
(v2) in tourmalines are particularly sensi-
tive to the orientation of the sample.
The blue (Fig. 65; T06) and deep blue
(Fig. 65; T08) tourmalines from metamor-
phic skarn have two (OH)1" stretching vi-
brational peaks. The most prominent fea-
ture in the spectrum of tourmaline is a
sharp, intense peak vx at 3550-3590 cm1
and a weak peak v2 at 3630 cm1. For the
blue tourmaline, the two (OH) peaks are
located at 3562 (Vj), and 3629 (v2) cm1).
For the deep blue tourmaline, the two (OH)
peaks are located at 3555 (v^, and 3630
(v2) cm1.
Based on the crystal structure of tour-
maline, we know that the (OH) site is at the
center of the hexagonal silicon tetrahedra
and below the Na ion. The Na-OH bond is
3.285 A long. The (OH) ion is surrounded
by 3 Mg ions which form three octahedra of
C/)
c
CD
TOc
*-'V/\/^-,/,v
Na T°4
a T06
^AjunAsyW-\^rvVV\A«
TOs
3200 3333 3466 3600 3733 3866 4000
Wavenumbers, cm-1
Fig. 65. PRS of [OH]1* groups in different types of
tourmaline in Ne direction at 3400-3650 cm1.
Spectra of tourmalines from metamorphic skarn
(T06 and T08) and from pegmatite and hydroth-
ermal vein (T04 and T05) are plotted.
Mg (OH)204. The Mg-OH bond length is
2.063A. Thus the tourmaline structure
contains brucite type (OH) groups.
The vibration peak assigned to (OH) in
brucite is shown in Fig. 66 (brucite from
U.S. National Museum No. 14390, cour-
tesy of the Division of Mineralogy, U. S.
National Museum). A very sharp single
peak is observed at 3648 cm1. The singular
peak is due to its high degree of symmetry
of (OH) in the brucite structure. The OH
group is surrounded by one type of cation
(Mg) only. In tourmaline, although the (OH)
ion has symmetry similar to that of brucite,
the surrounding atoms varies. Thus its
symmetry is reduced and the Raman peak
splits into two. The deep blue and blue
tourmaline of metamorphic skarn origin is
high in Fe2+ content (Table 13). The Fe2+
104
CARNEGIE INSTITUTION
C
, , J
L
3200 3333 3466 3600 3733 3866 4000
Wavenumbers, cm*1
Fig. 66. Raman spectra of brucite [Mg (OH)J.
replacement of Mg2+ in the (Mg,Fe) (OH)204
octahedra causes distortion of octahedra
and splitting of the OH peak.
The PRS of tourmalines from the granitic
pegmatite and from the hydrothermal vein
have an additional OH peaks. There are
three peaks attributed to (OH) stretching.
The rose red Mn-bearing tourmaline has
three (OH) peaks at 3460 (v3), 3570 (vx)
and 3635 (v2) (Fig. 65; T05). The three
peaks may be accounted for if (OH) occupy
two different sites. In addition to the one at
the center of the unit all in the middle of the
ring as mentioned earlier, (OH) may also
substitute an oxygen which surrounds bo-
ron atoms, and form an (OHB02) ion group.
The location of this peak is similar to that of
B-0 stretching vibrational peak of HOB022"
ion (Grice etal., 1986). The light green Fe-
bearing tourmaline (Fig. 65; T04) from the
granite pegmatite has 4 peaks where the vx
peak splits into two. These four peaks are
located at 3472 (v3), 3551 and 3577 (v,),
and 3636 (v2).
Conclusion
Based on the PRS of single crystals of
tourmaline, we are able to assign the vibra-
tional spectra to [Si6018]12\ [B03]3\ and
[OH] 1". The general feature of the polarized
Raman spectra (PRS) of tourmaline in the
ranges of 200-1200 and 3400-3600 cm1
are presented. Strong peaks of tourmaline
were observed at 1000-1200 and 200-400
cm1. They belong to Si-0 stretching vibra-
tion and ring deformation vibration of
[Si6018]12\ Strong peaks of [B03]3- vibra-
tion were measured at 700-800 cm*1. PRS
peaks of [B03]3" shift to higher frequencies
in the N direction.
o
Strong peak of (OH)1" vibration occurs
at 3550-3565 (v^ in the Nc direction. The
(OH) vibration is strongly polarized. PRS
of (OH) can only be detected in the Ne
direction. The (OH) group in the metamor-
phic skarn tourmaline occupies a single
site. The site occupancy is ordered. In
hydrothermal and granitic pegmatite tour-
malines, (OH) occupies two sites with dis-
ordered distribution. The (OH) vibrational
can be used to characterize site occupancy,
and are potentially indicative of the mode
of geological occurrences of tourmaline.
References
Brethous, J. C, A. Levasseur, G. Villeneuve, P.
Echegut, and P. Hagenmueller, S tudies by spec-
troscopic Raman and by RMN of the glasses of
the system B203 - Si02 - Li20, J. Solid State
Chem.,39, 199-208, 1981.
Buerger, M. H., C. W. Burnham., and D. R.
Peacor, Assessment of the several structures
proposed for tourmaline, Acta Crystallogr., 15,
583-590, 1962.
Grice, D. J., and J. V. Velthuigen, Moydite (Y.
REE) [B(OH)3 (C03)], a new mineral species
from the evans-lou pegmatite, Quebec Can.
GEOPHYSICAL LABORATORY
105
Min.y 24, 665-673, 1986.
Griffith, W. P., Raman studies on rock-forming
minerals, Part I orthosilicates and cyclosilicates,
/. Chem. Soc. (A), 1372-1377, 1969.
Hemley, R. J., H. K. Mao, and E. C. T. Chao,
Raman spectrum of natural and synthetic
stishovite, Phys. Chem. Minerals, 13, 285-290,
1986.
Mao, H. K., R. J. Hemley, and E. C. T. Chao, The
application of micro-Raman spectroscopy to
analysis and identification of minerals in thin
section, Scanning Microscopy, 7,495-501, 1987.
McMillan, P., Vibration spectroscopy in the min-
eral sciences, Rev. Mineral., 14, Miner. Soc.
Am., 9-63, 1985.
White, B. W., Structural interpretation of lunar
and terrestrial minerals by Raman spectros-
copy, in Infrared and Raman Spectroscopy of
Lunar and Terrestrial Minerals, C. Karr, Jr.,
ed., Academic Press, New York, Chap. 13, pp.
325-356, 1975.
New Optical Transitions in Type Ia
Diamonds
at Very High Stresses
Russell J. Hemley and Ho-kwang Mao
The generation of ultrahigh pressures in
the megabar range is now routine with the
diamond-anvil high-pressure cell (Mao,
1988). One of the important features of the
diamond-cell arises from the transparency
of the diamond anvils to large regions of the
electromagnetic spectrum, permitting spec-
troscopic characterization of materials at
high pressures using ultraviolet, visible,
and infrared radiation (Hemley etal, 1987).
Type Ia diamonds are used in ultrahigh
pressure studies owing to the presence of
nitrogen platelets which may enhance their
strength (Mao et aL, 1979). The nitrogen
impurities in these diamonds give rise to a
variety of absorption and luminescence
systems in the visible and ultraviolet at
ambient pressures (Walker, 1979). In opti-
cal studies using the diamond-anvil cell,
the absorption system at 3 eV in type Ia
diamonds serves as an effective absorption
edge, precluding most optical measure-
ments at higher energies. Laser excitation
in this region gives rise to a broad back-
ground fluorescence that can complicate
optical measurements of samples within
the cell. Further, the enhancement of this
luminescence at very high pressure (above
200 GPa) can interfere with measurements
of ruby fluorescence used for pressure de-
termination (Mao et al.9 1978).
Recently, we performed a series of op-
tical studies of hydrogen and a variety of
materials compressed at pressures in the
200 GPa range (Mao and Hemley, 1989;
Hemley and Mao, this Report). During the
course of this work we discovered dramatic
changes in the optical characteristics of the
diamonds in the high stress regions (tips) of
the anvils. Documenting these effects is
essential for further optical studies of ma-
terials at pressures above 200 GPa. In par-
ticular, this work is a prerequisite for opti-
cal characterization of the pressure-induced
insulator-metal transition in hydrogen and
other systems.
In the present work optical measure-
ments were performed on anvils with 25-
50 mm diameter tips, 300-500 mm culets,
and bevel angles of 8-10° (see Mao, 1988).
As a result of their small tips, at a given load
these diamonds generate higher stresses
within the anvils than those used in previ-
ous spectroscopic studies. The spatial de-
106
CARNEGIE INSTITUTION
0)
2.41 eV
Excitation
?\
J \ Diamond
\ Fluorescence
V.
X
2.0 1.9 1.8
Energy, eV
1.7
Fig. 67. Fluorescence spectrum of a type la
diamond anvil at ultra-high pressure excited with
514.5 nm (2.41eV) Ar+ laser line. The sample
consisted of hydrogen and ruby at a peak pres-
sure of 250 to 300 GPa.
pendencies of luminescence and Raman
spectra within the anvils were measured
using a -135° scattering configuration
with a 2 x 2 mm aperture and an argon-ion
laser beam focused to 2-4 mm at the im-
aged spot within the diamond (Hemley et
al., 1987). An example of this fluorescence
is shown in Fig. 67. With increasing stress
(corresponding to sample pressures above
200 GPa) a fluorescence peak appears at 2
eV. At very high pressures the signal dra-
matically increases. The peak tends to shift
toward lower energies with higher energy
excitation. Absorption extending through-
out the visible region of the spectrum, with
a broad peak at 2.4 eV, is also observed.
These changes in the fluorescence and
absorption spectra of the diamonds are
accompanied by new Raman bands (Fig.
68). Changes in the Raman spectra have
been documented with samples consisting
of hydrogen, neon, ruby, NaCl, and Si02.
Peak pressures were estimated from both
the pressure profile determined from ruby
fluorescence and from x-ray diffraction
measurements on metal pressure standards.
Fig. 68 shows the dependence of the Raman
0 400 800 1200 1600 2000
Wavenumbers, cm'1
Fig. 68. Raman spectrum of a type la diamond
anvil as a function of distance from a sample at
ultrahigh pressure: (a) 30mm. (b) 20mm. (c)
10mm. (d) sample-diamond interface. The sample
consisted of NaCl and ruby at a peak pressure of
-250 GPa.
spectrum on distance from the sample-
anvil interface for a sample containing NaCl
with 10-20% ruby. At the top of the dia-
mond anvil the zone-center, Raman-active
T. mode of the diamond at 1333 cm1 is
2g
clearly visible, with no bands apparent at
lower frequencies. The stress dependence
of this band has been measured previously
in diamond anvils under load (Sharma et
al, 1985; Hanfland and Syassen, 1985). As
the tip of the diamond is approached a new
feature at 590 cm*1 with a broad shoulder at
-350 cm1 appears. In some runs, a sharper
band at -900 cm1 was also observed. At the
sample-anvil interface the intensity of the
band overwhelms the diamond T0 band.
GEOPHYSICAL LABORATORY
107
Although the relative wavenumbers of the
bands are independent of laser excitation
wavelength (indicative of Raman transi-
tions), they are superimposed on a struc-
tured fluorescent background, which is
especially strong with 488.0 and 514.5 nm
excitation. In addition, the intensities of the
Raman bands showed a large degree of
resonance enhancement with decreasing
wavelength (e.g., 647.1 to 476.5 nm). The
bands were found to be reversible on re-
leasing the stress, although the 590 cm1
peak can remain at low sample pressures (~
30 GPa) before disappearing. The tips of
the diamonds have been found to exhibit
higher luminescence intensity upon un-
loading.
The present experiments demonstrate
that significant changes in the electronic
properties of type la diamonds occur at
stresses in the 200 GPa range. The lumines-
cence may be due to pressure-induced elec-
tronic changes in deep level impurity cen-
ters (Walker, 1980). If so, the Raman bands
may be associated with localized vibra-
tional modes at these centers. In this regard,
we note that the N-V (nitrogen-vacancy)
defects have an absorption band in this
region (zero-phonon line at 1 .95 eV at zero
stress) with a fundamental vibrational in-
terval in this range (n = 525 cm1, also at zero
stress) (Davies andHamer, 1978). Alterna-
tively, the new bands may be associated
with actual changes in the diamond struc-
ture. There is a close similarity between the
new Raman features and the one-phonon
density of states of diamonds which has a
broad peak centered at 600 cm1 (Dolling
and Cowley, 1966). Such a correlation
would imply a breakdown in crystalline
selection rules, resulting perhaps from
growth of defects at the anvil tips or macro-
scopic flow of the diamond (Mao et al.,
1979). Although a structural transforma-
tion in the diamond itself induced by non-
hydrostatic stress also cannot be ruled out
(Nielsen, 1986), our results indicate that
such a transition must be reversible. Simi-
lar measurements carried out on different
diamond types (type II, lb, including syn-
thetics) can be used to determine the extent
to which the optical effects are associated
with impurities or are intrinsic to diamond.
References
Davies, G., and M. F. Hamer, Optical studies of
the 1.945 eV vibronic band in diamond, Proc.
Roy. Soc. London A, 348, 285-298, 1978.
Dolling, G., and R. A. Cowley, The thermody-
namic and optical properties of germanium,
silicon, diamond, and gallium arsenide, Proc.
Phys. Soc. (London), 88, 463-494, 1966.
Hanfland, M., and K. Syassen, A Raman study of
diamond anvils under stress, /. Appl. Phys., 57,
2752-2756, 1985.
Hemley, R. J., P. M. Bell, and H. K. Mao, Laser
techniques in high-pressure geophysics, Sci-
ence, 237, 605-612, 1987.
Mao, H. K., Static compression of simple molecu-
lar systems in the megabar range, in Simple
Molecular Systems at Very High Densities, P.
Loubeyre, A. Polian, and N. Boccara, eds., Ple-
num, New York, pp. 221-236, 1988.
Mao, H. K., and R. J. Hemley, Optical studies of
hydrogen above 200 gigapascals: evidence for
metallization by band overlap, Science, 244,
1462-1465, 1989.
Mao, H. K., P. M. Bell, J. W. Shaner, and D. J.
Steinberg, Specific volume measurements of
Cu, Mo, Pd, and Ag and calibration of the ruby
R, fluorescence pressure gauge from 0.06 to 1
Mbar, J. Appl. Phys., 49, 3276-3283, 1978.
Mao, H. K., P. M. Bell, K. J. Dunn, R. M. Chrenko,
and R. C. Devries, Absolute pressure measure-
108
CARNEGIE INSTITUTION
ments and analysis of diamonds subjected to
maximum pressures of 1.3-1.7 Mbar, /. Appl.
Phys., 50, 1002-1009, 1979.
Nielsen, 0. H., Optical phonons and elasticity of
diamond at megabar stresses, Phys. Rev. B, 34,
5808-5819, 1986.
Sharma, S. K., H. K. Mao, P. M. Bell, and J. A Xu,
Measurement of stress in diamond anvils with
micro-Raman spectroscopy, /. Raman Spectros.,
16, 350-352, 1985.
Walker, J., Optical absorption and luminescence
in diamond, Rep. Prog Phys, 42, 1605-1659,
1979.
Premonitory Twinning in the High-
Pressure Phase
Transition of Zr02
Yasuhiro Kudoh, Charles T. Prewitt, and
Haruo Arashi*
At room temperature, a single crystal of
the monoclinic phase of Zr02 with space
group P2Jc transforms to a single crystal of
the orthorhombic phase at 35 kbar (Fig.
69). At pressures higher than 130 kbar, a
further phase transition to a different
orthorhombic phase with the cotunnite-
type structure is known. In this paper we
report two observations of stress-induced
twinning: the formation of 90° twin do-
mains about the c axis in single crystal Zr02
of the orthorhombic phase, and formation
of ( 1 1 1 ) twinning in a single crystal Zr02 of
the tetragonal phase under the application
of stress in a diamond-anvil cell at 298K.
These observations provide direct evi-
dence of ferroelastic behavior in Zr02 and
further corroborate predictions of the oc-
*Tohoku University, Research Institute for
Scientific Measurements, Sendai, Japan
Tetro.
2 3 4 5 6 7
PRESSURE (GPo)
Fig. 69. In situ phase diagram for Zr02 deduced
from diffraction measurements at high tempera-
tures and pressures (after Arashi et al., 1988).
currence of a displacive phase transition to
a higher symmetry phase in this material at
high temperatures and pressures.
Twinning in the orthorhombic phase
with a tetragonal symmetry operation
A single crystal of Zr02 was pressurized
up to 45 kbar with cedar oil as the fluid-
pressure medium using a modified Merrill-
Bassett type diamond-anvil pressure cell.
After loading in the diamond-anvil cell, the
pressure on the crystal was gradually in-
creased up to 45 kbar at room temperature,
exceeding the hydrostatic limit of cedar oil.
The pressure was then reduced slowly back
to ambient pressure and the sample re-
moved from the cell. X-ray precession pho-
tographs were made on the crystal, both
before and after pressure loading and at 45
kbar, using Mo#a radiation. Fig. 70 shows
an x-ray precession photograph taken of
GEOPHYSICAL LABORATORY
109
a£ b* b*
t
h*
bL'
Fig. 70. X-ray precession photograph of Zr02 after pressure loading to 45 kbar. Twinning by a mirror
plane parallel to the b-c plane [(100) twinning] is indicated by L. Twinning resulting from fourfold
rotation about the c axis of the orthorhombic phase is indicated by L'.
the pressure-released specimen at ambient
conditions. An analysis of the precession
photograph indicates the presence of two
types of twinning:
(1 ) Twinning by mirror plane parallel to
the b-c plane [(100) twinning]. Reciprocal
lattice axes resulting from these twin op-
erations are indicated by L in Fig. 70. This
type of twinning is known to occur in the
monoclinic phase prior to transition to the
orthorhombic phase (Kudoh et aL, 1986).
The mechanism of this twinning can read-
ily be interpreted by the slip system
(100)<001>. In monoclinic Zr02 this slip
system results in a 1 80° rotation of the axes
about the c axis, such that the b axis in some
domains lies parallel to the -b axis in others.
(2) Twinning by fourfold rotation about
the c axis of the orthorhombic phase (U in
Fig. 70). This twinning is consistent with a
unit cell rotation of exactly 90° about the c
axis, such that the a axis in some domains
lies parallel to the b axis in others. Because
the pressure of 45 kbar exceeds the hydro-
static limit of the cedar oil, this twinning is
thought to have occurred under a non-
hydrostatic condition. Since the fourfold
symmetry operation is included in the
tetragonal or cubic class, this twinning is
considered to be a premonitory phenome-
non, suggesting the possible existence of a
phase transition of the orthorhombic phase
to tetragonal symmetry, which was con-
firmed by the experiment described below.
110
CARNEGIE INSTITUTION
Twinning in the tetragonal phase with a
cubic symmetry operation
Another single crystal of Zr02 was pres-
surized up to 550kbarat room temperature.
Pressure was generated using a lever-type
diamond-anvil cell and measured by using
the ruby fluorescence method. Distilled
water was employed as the pressure trans-
mitting medium. Details of the experimen-
tal procedure have been reported previ-
ously (Arashi et al., 1989). The high pres-
sure phase is quenchable to atmospheric
pressure when the diamond anvil pressure
cell is unloaded rapidly (Arashi etal., 1 989).
After reducing the pressure to room pres-
sure, the recovered crystal was examined
by x-ray precession photography.
An analysis of the precession photo-
graph indicates the presence of two crystal-
lographically distinct orientations for the
tetragonal Zr02 crystal, indicating the pres-
ence of twinning on (111). Because the
pressure of 550 kbar exceeded the hydro-
static limit of distilled water, this twinning
also occurred under a non-hydrostatic
condition. Because the (111) mirror plane
is not included in the tetragonal class, but is
included in the cubic class, this twinning is
probably a premonitory phenomenon, sug-
gesting the possible existence of a transi-
tion of the tetragonal phase to cubic sym-
metry.
References
Arashi, H., O. Shimomura, T. Yagi, S. Akimoto,
and Y. Kudoh, P-T Phase diagram of Zr02
determined by in-situ x-ray diffraction measure-
ments at high pressures and high temperatures,
in Advances in Ceramics, Vol. 24, Science and
Technology of Zirconia III, The American
Ceramic Society, Inc., Westerville, Ohio, 493-
500, 1988.
Kudoh, Y. and H. Takeda, In situ determination of
the crystal structure for high pressure phase of
Zr02 using a diamond anvil and single crystal x-
ray diffraction method, Phys. Chem. Minerals,
13, 233-237, 1986.
GEOPHYSICAL LABORATORY
111
BlOGEOCHEMISTRY
Nitrogen Isotope Tracers of Human
Lactation in Modern and Archeologi-
cal Populations
Marilyn L. Fogel, Noreen Tuross/and
Douglas W. Owsley'
Variations in the stable isotope ratios of
carbon (&3C)** and nitrogen (#5N)** are
useful for paleodietary analysis of archae-
ologically-derived skeletal material (e.g.,
van der Merve, 1982; Schoeninger and
DeNiro, 1984; DeNiro, 1986). Because
plants and animals have distinctive iso-
topic signatures, the isotopic composition
of humans can therefore be correlated with
diets. For example, the C isotopic compo-
sition of com is distinct from other crop
plants, such as beans or squash. The differ-
ence in the #3C value is due to the operation
of a different photosynthetic pathway (C-4
photosynthesis) that occurs in corn relative
to that which is operational in most other
higher terrestrial plants (C-3 photosynthe-
sis). Accordingly, the introduction of com
(maize) into the diet of prehistoric North
American Indians has been traced with
stable C isotope ratios of the protein colla-
' * = <fi-J*M - D103. where X refers to »C or
,5N, and R refers the ratio of the heavy to light
isotope of either C (*3C/12C) or N (,SN/,4N) in the
sample or the standard.
* Conservation Analytical Laboratory, Smithsonian
Institution, Washington, D.C., 20550
# Department of Anthropology, Smithsonian Insti-
tution, Washington, D.C., 20550
gen preserved infossil bones (van der Merve
and Vogel 1977; Boutton et aU 1983).
Nitrogen isotopes are useful tracers of
an animal's diet primarily because isotopic
fractionation occurs during the metabo-
lism of dietary nitrogen and its incorpora-
tion into animal biomass. The protein in the
tissue of an animal is enriched in 15N rela-
tive to the diet of the animal by about +3 %o
(Minigawa and Wada, 1983). The enrich-
ment in the 15N in the animal relative to the
diet has been used to determine a variety of
important features concerning prehistoric
human diets such as the importance of
marine-derived food sources, legumes, and
meat (e.g., Schoeninger and DeNiro, 1984).
One of the major questions in anthro-
pology is what effect did the introduction
of horticulture have on weaning and birth
intervals in prehistoric peoples? Some have
hypothesized that, before agriculture,
humans nursed their infants longer and
concomitantly, birth intervals were longer
(e.g., Buikstra et al.y 1986). They assume
that with the introduction of agriculture,
mothers weaned their babies onto alterna-
tive food sources at a younger age, and
were thus able to give birth again in a
shorter time interval. These hypotheses are
difficult to test in modem populations, and
seemingly would be impossible to test in
prehistoric ones. In this paper, we investi-
gate whether breast milk has a unique iso-
topic signature that can be used to trace
lactation in humans. Infant nutrition in both
modem and fossil populations was studied
112
CARNEGIE INSTITUTION
Longitudinal Study
o
in
to
14
13 -
12 -
11 -
10 -
9 -
8
I
I I I
□
I I
■
□ □ n
I I
I
I
—
□
□
□
□
□
D
~~ •
K
□
m
K
•
D []
M
M
*
K
K
__U
I
I I I
I I
I I
I
I
0 1
2 3 4 5 6 7 8 9 10 11 12
Age (months)
Fig. 71. Longitudinal study of the variation in #5N in the fingernail cuttings of a single mother (*) and
infant pair (□). The time (months after birth) indicates when the fingernails were sampled. The infant's
hair (■) and the father's fingernails (•) were also measured. The infant was exclusively breast fed until
5 months of age, when a bovine milk-based formula was introduced (100 ml/ day). Formula amounts
increased with time to 500 ml/ day at 1 1 months. Dairy products were introduced at 7 months (100 g/
day). Fingernails (1-3 mg) were combusted at900°C, as in Tuross etai, (1988). The error of the analysis
for^NwasztO^oo.
with carbon and nitrogen isotopic tracers.
The hypothesis that nursing infants exist
one trophic level up on the food chain from
their lactating mothers, and thus protein
from infant tissue should be enriched in 15N
relative to the mother's protein, was tested.
In our study of contemporary mothers
and infants, fingernails were sampled and
analyzed. Fingernails are a rapidly synthe-
sized tissue easily obtainable from both
infants and their mothers. Numerous stud-
ies on nail growth have documented that in
GEOPHYSICAL LABORATORY
113
Cross-sectional Nursing Study
o
14
13
12 -
11
10 -
9 -
8
0
6 8 10
Age (months)
12 14
Fig. 72. Cross-sectional study of isotope variation in fingernail samples collected from 16 mother (*) and
infant pairs (D). Several children were sampled at different times afterbirth. All infants were fully breast
fed for at least three months. At approximately 3 months after birth a variety of substitute foods were
introduced to their diets. Children who were totally weaned to milk or milk-formula are indicated with
(•).
healthy, growing infants, fingernails re-
quire 2-3 months time to grow from cuticle
to finger tip. We sampled one infant and her
mother from birth to 15 months in a longi-
tudinal study (Fig. 71), in addition to 16
separate mothers and their infants in a
cross-sectional study (Fig. 72). In all cases,
the isotopic composition of the nursing
infants' fingernails was enriched in 15N as
compared to that of the mothers' over the
age range from three months until several
months after alternate food sources were
introduced. A decrease in the infant &5N
values toward those of their mothers corre-
lates with the introduction of alternative
nitrogen sources: infant formula, milk, dairy
114
CARNEGIE INSTITUTION
products, and meat. Carbon isotopic com-
positions of infant fingernails (-17.5 %o)
were nearly identical to those of their mother
and were not useful for tracing a human
milk source.
Fingernails cut in the first three months
of life were synthesized in utero. The in-
crease in the 15N content in the nursing
infants' fingernails after 3 months corre-
sponds to the introduction of breast milk at
birth (#5N = +8.0; n=4). After three months
of age, each infant was enriched in 15N by an
average +2.4 %o, when compared to the
mother. Four babies who were totally
weaned at 4-8 months of age to bovine
milk-based formula (#5N = +4) (n=3) or
whole bovine milk (n=l) showed a de-
crease in the #5N of their fingernails 3-5
months after the dietary change. The other
infants who were not given a milk substi-
tute, or provided formula in limited
amounts, maintained the enrichment of 15N
in their fingernails for the duration of the
study. Thus, the natural abundance of stable
nitrogen isotopes provides a measure of the
nitrogen sources, especially breast milk,
being utilized by a growing infant.
No attempt was made to control for the
diet of the mothers, yet 14 females had an
average #5N of +10 ±0.6 (la). In the lon-
gitudinal study, the woman had an average
#5N of +10.2 ±0.3. The isotopic composi-
tion of three individual nails from different
fingers sampled at one time from this mother
and her baby had a standard deviation of
±0.4 %o, which is larger than the mean of
the adult isotopic signal. In a study of nine
subjects from Chicago, Schoeller et al.
(1986) reported an average #5N of +9.4.
Given the diversity of nitrogen sources in
current diets, and the range of #5N in these
sources, the lack of variation is surprising.
The difference in #5N between infants and
their mothers (+2.4 %6) is thus eight stan-
dard deviations away from the adult iso-
topic mean and provides a distinct tracer of
lactation.
Whereas fingernails have a more clearly
defined and straightforward turnover time
in infants, collagen synthesis in bone and
its relation to diet are more complex. Gen-
erally, the stable isotopic values obtained
from fossil bone collagen are thought to
reflect the dietary input over a long period
of time, because the turnover time of colla-
gen in adult bone is on the order of 10-20
years. In the modern American society,
infants triple their birth weight by one year
(Ryan and Martinez, 1 987). Therefore, even
without any resorption of the bone collagen
present at birth, the one year old infant
would be expected to have synthesized a
minimum of two-thirds of its bone mass
after birth.
To determine whether a nursing signal
could be detected in skeletal remains, bone
samples from infants, small children and
adults were analyzed for age differences in
#5N values of bone collagen. The samples
are from archaeological contexts and rep-
resent pre- and post-horticultural popula-
tions. This contrast in subsistence patterns
provides a test of the hypothesis that the
time of weaning changed after agriculture
became established. The pre-horticultural
population sample (13 adults, 34 children)
was comprised of skeletal remains from
three Tennessee Valley Middle and Late
Archaic period sites located in Benton
County, TN: Cherry (40Bn74), Eva
GEOPHYSICAL LABORATORY
115
Tennesee Valley - Pre-Agricultural
i — i — i — r*
Sully — Agricultural
1 1 I--T-*
3 4 5
Age (years)
'Adult
Fig. 73. Nitrogen Isotopic composition of bone
collagen from the Tennessee Valley as a function
of the age at the time of death of the individual.
Horticulture was not practiced at this site.
(40Bnl2), and Ledbetter (40Bn25) (Hig-
gins, 1982; Magennis, 1977; Lewis and
Kneberg, 1959; Lewis and Lewis, 1961).
The subsistence pattern was based on
hunting and gathering. There is no evi-
dence for maize horticulture during this
period, which dates from about 5500 B.C.
until 2000 B.C. Permission to use the
samples was granted by J. Chapman and
M.O. Smith of the Frank H. McClung
Museum of the University of Tennessee.
Rib bones from the Sully site (39S14),
Sully Co., South Dakota, were obtained
primarily from the Smithsonian collection
and represent a population dependent on
horticulture. This protohistoric North Plains
Coalescent Tradition site dates to A.D. 1 650-
1700 (Owsley and Jantz, 1978). This popu-
lation relied on a mixed subsistence econ-
omy, involving the hunting and collecting
of wild foods, as well as horticulture with
principal crops being com, squash, beans,
CO
0 1 2 3 4 5 6 Adult
Age (years)
Fig. 74. Nitrogen isotopic composition of bone
collagen from the Sully site, South Dakota, as a
function of the age of the individual at the time of
death. The Sully site is representative of a Plains
Indian population that practiced maize agricul-
ture.
and sunflowers (e.g., Holder, 1970). A total
of 12 adult and 27 children specimens were
analyzed.
Age determination of infants and chil-
dren in both populations were based on the
dental calcification standards of Moorrees
<?fa/.(1963),asmodifiedbyHyman(1987),
and characteristic lengths of the long bones
(Merchant and Ubelaker, 1977). Bone
samples, usually rib fragments (1-2 g),
were decalcified in either EDTA solution
or 1 N HC1 (Tuross et ai, 1988), and the
isotopic ratio was determined as described
therein.
An enrichment in 15N of total collagen
was measured in almost all of the bones
tested from one year old infants (Figs. 73
and 74). Age determination of the humans
at the time of death is a critical component
of the study. If the infant age were off by 6
months to a year, the isotopic ratios would
116
CARNEGIE INSTITUTION
be random. In the Tennessee samples, the
difference between the babies and the adults
at 1 year was +4 %o, whereas at Sully it was
+2.5 %o. Both differences are similar to that
measured between the isotopic ratio of
modem infant and maternal fingernails.
The #5N of bones from both archeological
sites declined sharply at 1 8-20 months. The
initial enrichment and subsequent deple-
tion of the isotopic ratio of collagen is
consistent with the establishment and pres-
ervation of a nursing signature in the
younger group and a weaning pattern in the
older group. The 515N of the newborns was
variable but, on average, was almost iden-
tical with that of the adults. These children
were probably too young to have expressed
the extra utero nursing pattern, as was the
case with the isotopic ratios in modem
infants under 3 months of age.
From the modem data presented here,
we conclude that a clear tracer of lactation
is established in the protein of fingernails.
In every sample tested, the isotopic ratio of
a nursing infant was always more enriched
in 15N than that of its mother. The results
from the analysis of prehistoric human
populations demonstrate that when a suite
of individuals with known ages are ana-
lyzed, then the #5N of collagen preserved
in bone can be used as a tracer of infant
nutrition: breast feeding, weaning, and the
introduction of alternate food sources.
Nursing and weaning practices in the pre-
and post-horticultural Indian populations
studied were not significantly different from
one another. In both populations alternate
food sources were introduced at 18-20
months, and breast milk became less im-
portant in the diet. Full realization of the
application of this technique will require
the analysis of well-characterized collagen
from large skeletal populations.
References
Boutton T. W. , P. D. Klein, M. J. Lynott, J. E. Price,
and L. L. Tieszen, Stable carbon isotope ratios as
indicators of prehistoric human diet, in Stable
Isotopes in Nutrition, ACS Symposium Series
258, J. R. Turnland, P. E. Johnson, eds., Ameri-
can Chemical Society, Washington, D. C. pp.
191-204. , 1984
Buikstra, J. E., L. W. Konigsburg, and J. Bulling-
ton, Fertility and the development of agriculture
in the prehistoric midwest, Amer. Antiquity, 51 ,
528-546, 1986.
DeNiro, M. J., Stable isotopy and archaeology,
Am.ScL, 75, 182-191, 1986.
Higgins, Katherine F., The Ledbetter Site: A Study
of Late Archaic Mortuary Patterning. Unpub-
lished Master's thesis. Dept. of Anthropology,
The University of Tennessee, Knoxville, Ten-
nessee, 1982.
Holder, P., The Hoe and the Horse on the Plains,
University of Nebraska Press, Lincoln, 1970.
Hyman, Suzanne A., The Relationship Between
Dental Age and Long Bone Growth in Arikara
Infants, Unpublished Master's thesis, The Uni-
versity of Tennessee, Knoxville, 1987.
Lewis, Thomas M. N. and Madeline Kneberg, The
archaic culture in the middle South, Am. Antiq.,
25(2), 161-183, 1959.
Lewis, Thomas M. N., and M. K. Lewis, Eva, An
Archaic Site, University of Tennessee Study in
Anthropology, Knoxville, Tennessee, 1961.
Magennis, Ann L., Middle and Late Archaic
Mortuary Patterning: An Example from the
Western Tennessee Valley, Unpublished Mas-
ter's Thesis, Dept. of Anthropology, The Uni-
versity of Tennessee, Knoxville, 1977.
Merchant, Virginia L. and Douglas H. Ubelaker,
Skeletal growth of the protohistoric Arikara,
Am. J. Phys. Anthropoid 46(1), 61-72.
Minagawa, M. and E. Wada, Step-wise enrich-
ment of 15N along food chains, Further evidence
GEOPHYSICAL LABORATORY
117
and the relationship between #SN and animal
age, Geochim. Cosmochim.Acta,48, 1 135-1 140,
1984.
Moorrees, Coenraad F. A., Elizabeth A. Fanning,
and Edward E. Hunt, Jr., Formation and resorp-
tion of three deciduous teeth in children, Am. J.
Phys.Anthropol.,21, 205-213, 1963a.
Owsley, D. W. and R. L. Jantz, Intracemetary
morphological variation in Ankara crania from
the Sully site (39SL4), Sully County, South
Dakota, Plains AnthropoL, 23, 139-147, 1978.
Schoeller, D. A., M. Minagawa, R. Slater, and I. R.
Kaplan, Stable isotopes of carbon, nitrogen and
hydrogen in the contemporary North American
food web, Ecol. Food Nutrition , 18, 159-170,
1986.
Schoeninger, M. J., and M. J. DeNiro, Nitrogen
and carbon isotopic composition of bone colla-
gen from marine and terrestrial animals, Geo-
chim. Cosmochim. Acta, 48, 625-639, 1984.
Ryan, A.S. and G.A. Martinez, Physical growth of
infants 7 to 13 months of age, results from a
national survey, Amer. J. Physical AnthropoL,
73, 449-457, 1987.
Tuross, N., M. L. Fogel, and P. E. Hare, Variability
in the preservation of the isotopic composition
of collagen from fossil bones, Geochim. Cosmo-
chim. Acta, 52, 929-935, 1988.
van der Merve, N. J., Carbon isotopes, photosyn-
thesis and archeology, Amer. Sci.,70, 596-606,
1982.
van der Merve, N.J., Isotopic evidence for early
maize cultivation in New York State, Amer.
Antiquity, 42, 238-242, 1977.
Nitrogen Isotope Fractionation in the
Uptake of Ammonium by a Marine Bac-
terium
Matthew P. Hoch, David L. Kirchman*
and Marilyn L. Fogel
Heterotrophic bacteria can contribute
substantially to total biomass and biomass
* College of Marine Studies, University of
Delaware, Lewes, DE 19958
production in marine environments. On the
order of 50% of primary production, in the
form of dissolved organic matter (DOM),
can be processed by heterotrophic bacteria
in marine and freshwater environments
(Ducklow, 1983). In marine nitrogen cy-
cling, bacteria have traditionally been
viewed as consumers of dissolved organic
nitrogen (Fuhrman, 1987) and regenera-
tors of ammonium (HoUibaugh, 1980).
Contrary to this, recent studies have shown
that ammonium is a significant source of
nitrogen required for bacterial biomass
production, and a large fraction of total
ammonium uptake can be attributed to
bacteria (Wheeler and Kirchman, 1986).
For example, in the Delaware Estuary 4 -
30% of the total ammonium uptake can be
attributed to bacteria (Hoch and Kirchman,
in preparation).
The influence of heterotrophic bacteria
in changing the isotopic composition of
organic and inorganic pools of nitrogen in
the water column has not been considered
extensively (Altabet, 1988; Sigleo and
Macko, 1985). In situ microbial processes
can greatly influence the isotopic signature
of nitrogen that enters the sedimentary
records. Large isotope fractionation asso-
ciated with microbial processes such as
denitrif ication and nitrification (Mariotti et
al., 1981) can alter the isotopic composi-
tion of N03 , N02 , and NH4+, and therefore,
particulate organic matter when this nitro-
gen is fixed. A large isotope fractionation
associated with the assimilation of ammo-
nium by heterotrophic bacteria would af-
fect the 15N abundance of the ammonium
pool in addition to bacteria and phyto-
plankton.
118
CARNEGIE INSTITUTION
40
F
30
^
X
20
CO
0
O
10
0
8 12 16
Time (hrs)
20
0
i
-4
b
o
IS
815N
° "ammonium
ID
-8
-12
_e =
-15.0 ±0.76%
" ^bacteria
-16
-i i
* A A »
1 1 1* 1 1 1 1 1 1 1 1 1 1
4 8 12 16 20
Time (hrs)
Fig. 75. Growth of Vibrio harveyi on 20 mM ammonium and 10 mM glucose;
a) bacterial cell numbers over time, b) #5N of bacterial nitrogen and residual ammonium in the medium
for replicate cultures (a, +, ▲ ;b, □,♦).
In this study, the isotope fractionation
between NH4+ and bacterial nitrogen has
been determined for a common marine
isolate, Vibrio harveyi, grown on ammo-
nium as its sole nitrogen source. Bacteria
have been shown to have two enzymatic
pathways for NH4+ assimilation, which are
regulated by the concentrations of this N
source in the growth media. The hypothesis
that the isotopic fractionation would corre-
late with a switch in the enzymatic pathway
for the assimilation, from glutamate dehy-
drogenase to glutamine synthetase as
ammonium concentration decreased, was
tested.
Batch cultures of Vibrio harveyi were
grown on a minimum nutrient media with
ammonium and glucose as the sole source
of nitrogen and carbon, respectively. Addi-
tions of glucose, to 10 mM, and ammo-
nium, to either 20, 10, 5, 2, or 0.5 mM, were
first filter sterilized through 0.22 fm\ Mil-
lipore™ membrane. Cell growth was fol-
lowed by monitoring the absorbance at 660
nm and by cell abundance epifluorescent
microscopy. Cultures were incubated at
25 °C until stationary phase of the growth
curve was reached.
Cultures were sampled for the measure-
ment of their ammonium concentration and
bacterial nitrogen content, and for isotopic
analysis of ammonium and bacterial nitro-
gen. Bacterial biomass was collected on
pre-combusted 25mm Whatmann™ GF/F
glass fiber filters for analysis of nitrogen
content with a Hewlett Packard™ 185B
CHN Analyzer. Ammonium concentration
of the medium was determined by the indo-
phenol blue method (Solorzano, 1969).
Bacteria for isotopic analysis were concen-
trated by centrifugation, washed with dis-
tilled water, freeze dried, and stored in
vacuo. Filtrate collected for ammonium
isotope analysis was frozen at -80°C until
distillation. A Labconco™ Rapid Kjeldahl
System, Rapid Still HI was used for alka-
line distillation of ammonium (Velinsky et
al., 1989). Bacterial biomass and zeolite
with ammonium were converted to N2 for
mass spectral analysis by combustion
GEOPHYSICAL LABORATORY
119
(Macko,eftf/., 1987).
Specific activities for the bacterial en-
zymes glutamate dehydrogenase (GDH)
and glutamine synthetase (GS) were meas-
ured in cultures of V. harveyi grown on 20,
10, 5, 2, or 0.5 mM ammonium and har-
vested at mid-exponential growth. Total
GS activity was assayed using the g-glu-
tamyltransferase assay of Bender et al.
(1977). Activity of GDH was assayed by
following the oxidation of NADH (Sanwal
and Lata, 1961).
At high concentrations (20 and 5 mM)
only a small fraction of ammonium
(<0.05%) was assimilated by cultures that
had grown to stationary phase (Fig. 75a).
These conditions are characteristic of an
open system, so the isotope enrichment (e)
approximates the difference between the
&5N of the bacteria and that of ammonium:
e ~ &5N bacteria - #5N ammonium. (1)
For the 20mM NH4+ cultures, e equaled
-15 %o (± 0.78 %o) (Fig. 75b).
-3.2
-2.8 -2.4
-2.0 -1.6
log [NHJ ] (M)
Fig. 76. Composite plots of the nitrogen isotope
fractionations for Vibrio harveyi grown on 20, 5,
2, and 0.5 mM ammonium, plotted as the log of the
molar ammonium concentration. One error bar
equals one standard deviation.
All the ammonium was assimilated by
the stationary phase in cultures with 2 and
0.5 mM NH4+ at the start of growth. The
isotope ratio of NH4+ was determined at
intervals before it was totally assimilated.
In this case e was calculated with equations
described by Mariotti et al. (1981):
£=1000 log (R/RJ/ log f,
(2)
where / is the fraction of ammonium
remaining, R is the ratio of 15N/14N in the
initial NH4+ and Ro is the ratio in the sample
at time (to). There is an inverse relationship
between the ammonium concentration of
the culture medium and the isotope frac-
tionation (Fig. 76). V. harveyi grown on 5
and 20 mM NH/ fractionated ammonium
nitrogen by ca. -15 %o. At the lowest con-
centration (0.5 mM) the isotope fractiona-
tion was the greatest, -22 %o. In order to
explain the change, the activity of the pri-
mary ammonium assimilatory enzymes
100
:g 80
o
<
.9 60
o
8.40
CO
I I I I I I I 1 I I I I I I I I
Glutamine Synthetase
:\(gs)
Glutamate
Dehydrogenase
(GDH).
20 -
» » ' »
-3.2
-2.8
-2.4
-2.0
-1.6
log [NHJ ] (M)
Fig. 77. Composite plot of percent specific activ-
ity of glutamine synthetase and glutamate dehy-
drogenase for V. harveyi grown on 20, 10, 5, 2,
and 0.5 mM ammonium, plotted as the log of the
molar ammonium concentration.
120
CARNEGIE INSTITUTION
were assayed.
The highest specific activities for V.
harveyi GDH and GS were in the cultures
with 20 mM and 0.5 mM ammonium, re-
spectively (Fig. 77). This relationship sug-
gests that isotope fractionation is depend-
ent on the enzymatic pathway for ammo-
nium assimilation. At high ammonium
concentrations, GDH is the dominant as-
similatory enzyme for bacteria, and is re-
sponsible for catalyzing the reaction:
2-oxoglutarate + NHj + NAD(P)H
<=» L-glutamate + NAD(P)+ + Hfi. (3)
At lower concentrations of ammonium, GS
becomes dominant, and catalyzes the reac-
tion:
L-glutamate + NH3 + ATP
<=> L-glutamine + ADP + P..
(4)
Nitrogen isotope fractionation associ-
ated with ammonium assimilation by other
cultured organisms is about the same as
that for V. harveyi (Table 15). When both
algal and bacterial species were grown on
millimolar concentrations of ammonium
(3.5 to 70 mM), the nitrogen isotope frac-
tionation ranges from -13.5 to -15 %o.
Pennock et al. (in preparation) measured
fractionations within the range of -19 to
-27 %o for a marine diatom, Skeletonema
costatum, grown on 50 /xM NH4+. These
values for this diatom are similar to those
for V. harveyi grown at 0.5 mM NH4+.
Apparently, nitrogen isotope fractionation
Table 15. Isotope fractionation (c) between NH4* and organic matter for various organisms studied.
Ammonium concentration of growth media are given in parentheses.
Organism
e(%o) £(%©)
High [NH4+] Low [NH4+]
(mM) (//M)
Vibrio harveyi
(marine bacterium)
-15.0
(5,20)
-22.0
(500)
(present study)
Azotobacter vinelandii
(soil bacterium)
-14.8
(70)
Delwiche and Steyn (1970)
Anabaena sp
(cyanobacterium)
-13.6
(18)
Mzcko etal. (1987)
Skeletonema costatum
(marine diatom)
-19 to -27
(100)
Pennock et al. (unpublished)
Phaeodactylum tricornutum
(marine diatom)
-13.6-
(3.5)
Wada and Hattori (1978)
* Recalculation of data from Wada and Hattori ( 1978) using equation 1 1 from Mariotti et al. ( 198 1 ), and
assuming f=0.6 and initial substrate &*N = -1.5 %o when phytoplankton biomass #5N is approximately
-12 %o.
GEOPHYSICAL LABORATORY
121
associated with ammonium assimilation is
not species specific. We conclude that the
rate-limiting step in ammonium assimila-
tion for high and for low ammonium con-
centrations is similar among different or-
ganisms.
Potential sites of an isotope effect (i.e.,
the rate-limiting step) for ammonium as-
similation are depicted in Fig. 78. The
equilibrium isotope effect between NH4+
andNI^ is -19.2 %o at 25°C (Hermes etai,
1985). The 15N is concentrated in NH4+,
whereas 14N is enriched in the NH3. In cal-
culating the kinetic isotope effect of
alanine and glutamate dehydrogenase from
bovine liver. Weiss et al. (1988) corrected
the observed values to account for equilib-
rium isotope effects during the deprotona-
tion of NH4\ because the actual substrate
for the enzyme is NH3. Therefore, the iso-
tope effects for alanine and glutamate
dehydrogenase are inverse, +8%o and
+2 %o(±l %6), respectively. Assimilation of
ammonia with a #5N of -19 by these en-
zymes would yield #5N values for bacterial
cells of - 1 7 .2 %o and - 1 1 .2 %c, respectively.
Isotope fractionation for V. harveyi grown
on 20 mM NH4+ (ca. -15 %o) is between
these values. Alanine dehydrogenase may
be involved in V. harveyi ammonium as-
similation, however, its activity was not
assayed in our cultures.
We are presently measuring the isotope
fractionation by glutamine synthetase. Also
inherent in determining the isotope frac-
tionation of GS is the pre-equilibrium iso-
tope effect between NH3 and NH4\ because
NH3 is the species taken up by the enzyme.
The rate controlling steps of many chemi-
cal reactions are preceded by rapid and
high
[NH+J
DIFFUSION
■3^ ... >4
potential isotope effect
low
[NHJ]
ACTIVE TRANSPORT
Fig. 78. Schematic diagram of the pathways for
ammonium assimilation in a bacterial cell show-
ing the potential sites of isotope effects. Values for
isotope fractionation are explained in the text.
reversible pre-equilibria (Bigeleison and
Wolfsberg, 1958). Glutamine synthetase
was assayed at two pH values to determine
the effect of this pre-equilibrium on the
measured fractionation of the reaction. At
pH 7.0, where less than 1 % of the total N is
NH3, the isotope fractionation (e) for both
the pre-equilibrium step and the enzyme
reaction itself is -8.0 ± 0.3 %o (r2 = 0.95; n
= 1 3). At pH 8.6, where a greater proportion
of the total N is NH3, the total fractionation
is -123 ± 0.5 %o (r2 = 0.94; n = 9). Accord-
ingly, the fractionation by GS itself was
determined by calculations modified from
Bigeleison and Wolfsberg (1958):
total equilibrium GS*
(5)
At pH 7.0 and 8.6, respectively, inverse
122
CARNEGIE INSTITUTION
isotope effects of +10.8 and +3.0 were cal-
culated. Both values have a similar direc-
tion and magnitude as those measured by
Weiss et al (1988) for ADH and GDH.
Therefore, at lower ammonium concen-
trations the observed isotope fractionation
for V. harveyi is a result of some other
process involved in the ammonium assimi-
latory pathway. Bacteria obtain ammonium
across cell membranes by physical diffu-
sion of NI^ or by active (i.e., energy de-
pendent) transport of NH4+ (Kleiner, 1985).
Nothing is known about isotope effects
associated with active NH4+ transport, al-
though Marotti et al. (1982) found no iso-
tope fractionation during the active uptake
of NO3- into plant cells. Conversely, the
isotope effect during the diffusion NH3
may be as large as -29 %o (See Hermes et
al., 1985). At low NH4+ concentrations,
diffusion may be the rate-limiting step in
assimilation.
The relatively large isotope fractiona-
tion associated with ammonium assimila-
tion by heterotrophic marine bacteria can
have a major influence on the nitrogen
isotope ratio of suspended and sedimentary
organic matter in estuarine and coastal
environments. With our results, nitrogen
isotopes may be useful in addressing eco-
logical questions concerning the role of
heterotrophic bacteria in marine nitrogen
cycling.
References
Altabet, M. A., Variations in nitrogen isotopic
composition among particle classes: Implica-
tion for particle transformation and flux in the
open ocean, Deep Sea Res., 35, 535-544, 1988.
Bender, R. A., K. A. Janssen, A. D. Resnick, M.
Blumenberg, F. Foor, and B. Magasanik, Bio-
chemical parameters of glutamine synthetase
from Klebsiella aerogenes, J. Bacterid., 129
(2), 1001-1009, 1977.
Biegeleison, J., and M. Wolfsberg, Theoretical
and experimental aspects of isotope effects in
chemical kinetics, in Advances in Chemical
Physics, I. Prigogine, ed., Interscience Publish-
ers, Inc., New York, 1958.
Delwiche, C. C, and P. L. Steyn-, Nitrogen isotope
fractionation in soils and microbial reactions,
Environ. Sci. Technoi, 4, 929-939, 1970.
Ducklow, H. W., Production and fate of bacteria in
the oceans, BioScience, 33(8), 494-501, 1983.
Eppley, R. W., R. W. Holmes, and J. D. H. Strick-
land, Sinking rates of marine phytoplankton
measured with fluorometer, /. Exp. Mar. Biol.
Ecol, 1, 191-208, 1967.
Fuhrman, J. A., Close coupling between release
and uptake of dissolved free amino acids in
seawater studied by an isotope dilution approach,
Mar. Ecol. Prog. Ser., 37, 45-52, 1987.
Hermes, J. D., P. M. Weiss, and W. W. Cleland,
Use of nitrogen- 15 isotope effects to determine
the chemical mechanism of phenylalanine
ammonium-lyase, Biochem., 24, 2959-2967,
1985.
Hollibaugh, J. T., A. B. Carruthers, J. A. Fuhrman,
and F. Azam, Cycling of organic nitrogen in
marine plankton communities studied in en-
closed water columns, Mar. Biol., 59, 15-21,
1980.
Kleiner, D. , Bacterial ammonium transport, FEMS
Microbiol. Rev., 32, 87-100, 1985.
Macko, S. A., M. L. Fogel (Estep), P. E. Hare, and
T. C. Hoering, Isotope fractionation of nitrogen
and carbon in the synthesis of amino acids by
microorganisms, Chem. Geol. (IGS), 65, 79-92,
1987.
Mariotti, A., J. C. Germon, P. Hubert, P. Kaiser, R.
Letolle, A. Tardieux, and P. Tardieux, Experi-
mental determination of nitrogen kinetic isotope
fractionation, some principles; illustration for
the denitrification and nitrification processes,
Plant and Soil., 62, 413-430, 1981.
Mariotti, A., F. Mariotti, M.-L. Champigny, N.
Amarger, and A. Moyse, Nitrogen isotope frac-
GEOPHYSICAL LABORATORY
123
donation associated with nitrate reduction activ-
ity and uptake of N03* by pearl millet, Plant
Physiol, 69, 880-884, 1982
Sanwal, B. D., and M. Lata, The occurrence of two
different glutamic dehydrogenases in neurospora,
Can. J. Microbiol, 7, 319-328, 1961.
Sigleo, A. C, and S. A. Macko, Stable isotope and
amino acid composition of estuarine dissolved
colloidal material, in Mar. Estuar. Geochem. A.
C. Siegleo and A. Hattori, eds., Lewis Publish-
ing Inc., Chelsea, Michigan, pp. 29-46, 1985.
Solorzano, L., Determination of ammonium in
natural water by the phenol hypochlorite method,
Limnol Oceanogr., 14, 799-800, 1969.
Velinsky, D. J., L. A. Cifuentes, J. R. Pennock, J.
H. Sharp, and M. F. Fogel., Determination of the
isotope composition of ammonium-nitrogen at
the natural abundance level from estuarine
waters, Mar. Chem., in press, 1989.
Wada, E., and A. Hattori, Nitrogen isotope effects
in the assimilation of inorganic nitrogenous
compounds by marine diatoms, Geomicrobiol.
J., 1(1), 85-101, 1978.
Wheeler, P. A., and D. L. Kirchman., Utilization
of inorganic and organic nitrogen by bacteria in
marine systems, Limnol. Oceanogr., 31(5), 998-
1009, 1986.
Weiss, P. M., C. Y. Chen, W. W. Cleland, and P.
F. Cook, Use of primary deuterium and 15N
isotope effects to deduce the relative rates of
steps in the mechanisms of alanine and gluta-
mate dehydrogenases, Biochemistry, 27, 4814-
4822, 1988.
Dissolved Nitrogen Isotopic Distribu-
tion in the Black Sea
David J. Velinsky, Marilyn L. Fogel, and
Bradley M. Tebo*
The Black Sea is the world's largest,
present-day, anoxic marine basin (Fig. 79).
As a result of intense water column strati-
fication, the flux of oxygen to the bottom
waters is not sufficient for the complete
oxidation of surface-derived organic mat-
ter. Below approximately 200 m, bacteria
use alternate electron acceptors, including
iron and manganese oxides, nitrate and
most importantly sulfate, to oxidize or-
ganic matter during respiration. Because
sulfate is the most abundant oxidant after
oxygen, microbial respiration is dominated
by sulfate reduction in the deep waters and
sediments of the Black Sea.
During the oxidation of organic matter
coupled to sulfate reduction, organically
bound nitrogen is converted to ammonium.
Fig. 79. Black Sea station location map.
* Scripps Institution of Oceanography, La Jolla,
CA 92093.
124
CARNEGIE INSTITUTION
Richards (1965) described a steady state
stoichiometric model for the production of
ammonium during anaerobic oxidation of
organic matter (Table 16). These reactions,
coupled with the sluggish mixing of Black
Sea deep waters, result in the buildup of
ammonium in the waters below approxi-
mately 100 m. As the ammonium diffuses
upward in the water column across the
oxygen-sulfide boundary, it undergoes
important transformations that affect bio-
logical production and possibly also trace
metal distributions. For example, ammo-
nium can serve as an energy source for
aerobic chemosynthetic production of or-
ganic matter (Brewer and Murray, 1973).
In the process of nitrification, chemoau-
totrophic bacteria convert ammonium to
nitrite and nitrate, which can subsequently
be reduced to nitrogen gas via denitrifica-
tion in tightly coupled reactions (Hattori,
1982). In addition, because of the broad
zone of low oxygen concentration observed
just above the oxygen/sulfide interface
during the 1988 expedition to the Black
Sea, it has been postulated that nitrate can
serve as an electron acceptor for both
ammonium and dissolved manganese oxi-
dation (Murray etaL, 1989). Thus there are
potentially both aerobic and anaerobic
processes taking place near the interface
that can consume ammonium and limit its
availability as a nutrient in the overlying
waters.
Isotope tracers at the natural abundance
level have been used to determine the flow
Table 16. Steady-State Stoichiometric Model for the Oxidation of Organic Matter
1) Oxygen Respiration:
(CH2O)106(NH3)16H3PO4+ 106 02 <=> 106 C02 + 16 NH, + HLPO, + 106 H.O (1)
Ammonia is oxidized to nitrate (i.e., nitrification):
16NH3 + 32 020 16HN03+16Hp
2) Nitrate Reduction and Denitrification:
(CH2O)106(NH3)1<H3PO4 + 84.8 HN03
<=> 106 CO, + 42.4 N2 + 16 NH, + H3P04 + 148.4 H20
Also, the NFL, released can be oxidized by HN03:
5NH3+3HN03<=>4N2 + 9H20
3) Sulfate Reduction:
(CH2O)106(NH3)16H3PO4+ 53 S04>
<=> 106 C02 + 53 S2 +16 NH3 + H3P04 + 106 H.O
(la)
(2)
(2a)
(3)
GEOPHYSICAL LABORATORY
125
and reactions for both carbon and nitrogen
in estuarine and open ocean environments
(Cifuentes et aL, 1988a; Mariotti et ai,
1984; Altabet, 1988). Processes such as
algal assimilation of nitrogen (e.g., NH4+,
N03\ and N02), nitrification, denitrifica-
tion and possibly organic decomposition
can be followed with stable isotope tracers.
Overall, the isotopic composition of both
the dissolved and particulate pool of nitro-
gen is determined by the isotopic composi-
tion of the source material and any related
isotopic fractionation in its formation or
decomposition. To understand better the
reaction pathways of the nitrogen, particu-
larly near the oxygen-sulfide interface in
stratified waters, the stable N isotope
composition (#5N) of dissolved ammo-
nium and nitrate were determined from
water samples taken from the Black Sea.
Along with particulate nitrogen isotopic
data, the transformations of nitrogen can be
further elucidated. This information is ex-
tremely important in the understanding of
the isotope biogeochemistry of material
formed in the oxic surface waters and
deposited in euxinic environments like the
Black Sea. Once the variability and proc-
esses related to the formation of the nitro-
gen isotopic composition of particulate
matter are understood, a more thorough
interpretation of the #5N distribution in the
sedimentary record can be made (Rau et
aU 1987).
Samples for nitrogen isotopic composi-
tion of dissolved ammonium and nitrate
were obtained during the 1988 Black Sea
Oceanographic Expedition on leg 3 (June
3-16, 1988). Two stations (Fig. 79) were
occupied and water samples were taken
from the R/V Knorr for full water column
chemistry, Station 2 (BS3-2; 42°50'N
32WE) and Station 6 (BS3-6; 43°04'N
34WE).
The method for the preparation and
nitrogen isotopic determination of dissolved
ammonium is described in Velinsky et al.
(1989). Briefly, an aliquot of the sample is
adjusted to pH>9 using 6 M NaOH and
distilled with a rapid steam distillation
apparatus. The distilled ammonia is col-
lected in a dilute acid trap and removed
from solution by ion-exchange onto a zeo-
lite. The nitrogen isotopic composition of
dissolved nitrate is accomplished by the
method described by Mariotti and Letolle
(1978) and Horrigan et al. (unpublished).
Dissolved nitrate is reduced to ammonia
using Devarda's Alloy (Cu-Al-Zn) in a
basic solution. The ammonia generated is
distilled with a conventional distillation
apparatus, into a dilute acid and zeolite trap
(Velinsky et aL, 1989). The zeolite and
particulate N is analyzed for the nitrogen
isotopic composition (#5N) of both ammo-
nium and nitrate by a modified Dumas
combustion technique (Macko, 1981). The
data are reported in the standard <5 notation
{i.e., #5N = [(K^A^J-lllO* where R
= 15N/14N} and the ratios are reported against
air (#5N = 0). Precision of replicate samples
for ammonium and nitrate isotopic analy-
sis is approximately ±0.5 %o and + 1 .0 %o ,
respectively.
The concentration and isotopic compo-
sition of dissolved ammonium and nitrate
varied with depth in the water column
(Figs. 80 and 81). Ammonium concentra-
tions (Fig. 80) in the surface waters (0-80
m) were close to the detection limit and
126
CARNEGIE INSTITUTION
NH* (uM) a
20 40 60 80 100
500
1000
sz
a.
a>
Q
1500
NH5 (uM) -
20 40 60 80 100
2000
J I L
Station 2
J I i i i
500 -
1000 -
SI
Q
1500 -
2000 -
01 23456789 10
515N Q
Fig. 80. Concentration and nitrogen isotopic composition of dissolved ammonium with depth in the
Black Sea at Stations 2 and 6.
reflected biological uptake. Below 80 m,
with the onset of sulfate reduction, ammo-
nium concentrations increased from less
than 0.2 to about 20 fiM at 200 m and up to
90 jiM by 2000 m. Nitrate concentrations
exhibited a broad maximum of 6.5 ^M near
the middle of the sub-oxic zone (Fig. 8 1 ). A
slight overlap between nitrate and ammo-
nium distributions near 85 m was indica-
tive of the oxidation of ammonium to ni-
trate. The presence of nitrite which is an
intermediate in the oxidation of ammo-
nium to nitrate during nitrification is fur-
ther evidence for this process (Murray et
aU 1989a,b).
The distribution of isotopic ratios of
ammonium and nitrate are presented in
Figs. 80 and 81 . The #5N of ammonium at
depths greater than 500 m was uniform for
both stations (1.71 ±0.16%c,n=9). Above
500 m dramatic shifts occurred in the nitro-
gen isotopic composition of ammonium.
As the concentration of ammonium de-
creased toward the interface, the #5N of
ammonium increased. At station 6 for
example, the #5N increased from ca. 1 .9 %o
at 500 m to 9.0 %o at 95 m. A similar
distribution was observed at station 2. The
GEOPHYSICAL LABORATORY
127
DIN (jiM)
0 4 8 12 16 20 24
0 r— i — i — i — |— i — I — I — i — I — I — i — r~
50
■B
a.
o
Q
100
150
200
Station 6
J I I I I I I I L
— flj- NH>M)
— ♦- 515 NO
3
(>iM)
6 NH4
01 23456789 10
615N
Fig. 81. Concentrations and nitrogen isotopic
composition of dissolved ammonium and nitrate
in the upper water column of the Black Sea at
Station 6. (DIN = dissolved inorganic nitrogen)
more positive isotope ratios for ammonium
are indicative of isotopic fractionation
during the consumption or oxidation of
ammonium (see below).
The #5N of nitrate also changed consid-
erably with depth. At station 6, the #5N of
nitrate increased with depth from 4.2 %o at
50 m to 9.8 %o just below the nitrate con-
centration maximum. Bazylinski et al.
( 1 988) measured maximum denitrification
rates below this maximum at station 2
during leg 2 (May 1988). The transforma-
tion of nitrate to N2 by denitrification
(Table 16) induces a large isotopic frac-
tionation (r, where e = (a-l)103) of ap-
proximately -30 %o (Cline and Kaplan,
1975). Therefore, the residual nitrate should
be enriched in 15N. The more positive #5N
values of nitrate below the nitrate concen-
tration maximum could be explained by
high denitrification rates at depth. Similar
observations of #5N of nitrate and denitri-
fication zones occurred at station 2 (data
not shown).
An advection-dif fusion model was used
to calculate the consumption or production
of ammonium in the water column of the
Black Sea with a final goal of understand-
ing isotope fractionation and biogeochemi-
cal processes in nitrogen cycling. This
model is similar to that developed by Craig
(1969) and Craig and Weiss (1970). Be-
cause the data set for station 6 is more
complete, only this station was used for the
advection-diffusion model.
The general equation for the advection,
diffusion and reaction of a chemical spe-
cies within a water system is (Craig, 1969):
Kd2C/dz2-codC/dz + J = 0.
(1)
This particular form of the equation as-
sumes steady state conditions where K is
the vertical eddy diffusion coefficient, co is
the vertical advection velocity, J is the
production or consumption of ammonium
(C). It assumes only vertical water move-
ment (i.e., where z is positive upward) and
is valid only in the linear portion of the
potential temperature-salinity profile. In
this case, J [equation (1)] was fitted by an
exponential term (J = Jo e /iz), where fi is the
decay constant for ammonium consump-
tion or production with depth (z). The
equation was first solved in terms of K/co
(Craig, 1969) for stable conservative ele-
ments (i.e., potential temperature and
128
CARNEGIE INSTITUTION
salinity). Kl co is a mixing parameter for
species that do not undergo any reaction
during two end-member mixing. Once K/co
was determined, the model was solved for
stable nonconservative species (i.e., am-
monium) in terms of JJco and ji. The use of
this type of model and its limitations is
discussed by Craig ( 1 969), Craig and Weiss
(1970) and Spencer and Brewer (1971).
The solution of equation (1), fitted to
the salinity data, yielded a K/co of 127 m,
which is in excellent agreement with the
results of Spencer and Brewer (1971). The
model was run for a constant / and expo-
nentially fitted J (see above) and both solu-
tions gave similar results. The median /
within our mixing interval (90 to 800 m) is
-3.8 • 10 2 /zM NH4+/kg yr, where \i = -2.80
km1. The negative sign means that ammo-
nium was being consumed. This median
consumption rate occurs at a depth of 390
m below the surface. Fig. 82 shows the
fitted data with an exponentially decreas-
ing J and with a model run with 7 = 0. The
consumption rate of -3.8 • 10 2 juM NH47kg
yr was approximately three times lower
than that derived by Brewer and Murray
(1973). Station to station variability and
different mixing intervals could be the
reason for this difference (Spencer and
Brewer, 1971; Murray etal., 1989). In any
case, the ammonium taken up in this inter-
val, could be oxidized via nitrification or
consumed by other chemoautotrophic or
heterotrophic bacteria around the interface.
Both of these processes are enzymatically
mediated, thus a normal isotope effect (i.e.,
14N is taken up at a faster rate then 15N) was
expected.
Qualitatively, the ammonium in the
mixing interval became more enriched in
NH>M)
20
40 60 80
~1 1 1 1 1 T
300
^ 600
E
Q.
CD
Q
900
1200
1500
Fig. 82. Advection-diffusion model results for
both predicted ammonium concentrations
assuming J decreases exponentially with depth
(predicted) and with no production or consump-
tion (/ = 0).
15N compared to that in bottom waters (Fig.
80). In other words, as ammonium was
consumed, the residual ammonium became
enriched in 15N. Isotopic fractionation by
bacteria has been determined by Hoch et
ai, (this Report) for the bacterium Vibrio
harveyi. Fractionation values, £, varied
between -15 to -22 %o during growth on
ammonium as the sole nitrogen source.
Chemosynthetic uptake by bacteria of
ammonium near the interface of the Black
Sea (Brewer and Murray, 1973) is therefore
GEOPHYSICAL LABORATORY
129
associated with a normal isotope effect
because the residual ammonium becomes
increasingly heavier as it was consumed.
Our calculations, based on a closed system,
yielded a fractionation, e, of -3.7 %o be-
tween the mixing interval of 90 to 800 m.
Closed system calculations will yield
smaller isotope fractionations than those
calculated from open system models,
because 14N is constantly diffusing upward
and reacting faster (i.e., an open system)
than 15N. We are presently developing an
advection diffusion model taking into ac-
count isotopic fractionation during the
uptake of ammonium. This type of equa-
tion has been used to model the isotopic
composition of nitrate in the low oxygen
zone of the North East Pacific (Cline and
Kaplan, 1975).
In summary, two distinct processes
appear to be occurring near the interface (<
200 m) of the Black Sea; 1) denitrification
and 2) chemosynthetic production of or-
ganic matter. Both processes result in dis-
tinct isotopic ratios for both ammonium
and nitrate. Below the nitrate concentration
maximum, the concentration of nitrate
decreases due to denitrification. This is
further evidenced by the increase in 15N
below the nitrate concentration maximum
(Fig. 81). Ammonium concentrations start
to increase because sulfate reduction and
related ammonium production (Table 16)
were faster then both transport and con-
sumption. However, our model results
show that there is a net consumption of am-
monium most likely associated with bacte-
rial chemosynthetic production near the
interface. As the bacteria consume ammo-
nium, the residual ammonium becomes in-
creasingly enriched in 15N.
The nitrogen isotope fractionation dur-
ing the chemosynthetic uptake of dissolved
ammonium was calculated using closed
system assumptions. While the fractiona-
tion for closed system calculations yielded
an e of -3.7, calculations based on more
realistic open system assumptions will yield
a slightly greater fractionation value. It will
be of interest to compare these results with
model calculations from other anoxic envi-
ronments such as the Framvaren Fjord and
the Saanich Inlet. These data would help
eventually determine the controls on the
isotopic composition of material formed in
the water column and eventually buried in
anoxic marine environments.
References
Altabet, M. A., Variations in nitrogen isotopic
composition between sinking and suspended
particles: implications for nitrogen cycling and
particle transformation in the open ocean, Deep
Sea Res., 35, 535-554, 1988.
Bazylinski, D. A., B. L. Howes, and H. W. Jan-
nasch, Denitrification, nitrogen fixation and
nitrous oxide concentrations through the Black
Sea oxic-anoxic interface, EOS, 69, 1241, 1988.
Brewer, P. G., and J. W. Murray, Carbon, nitrogen
and phosphorus in the Black Sea, Deep Sea Res.,
20, 803-818, 1973.
Cifuentes, L. A., J. H. Sharp, and M. L. Fogel,
Stable carbon and nitrogen isotope biogeochem-
istry in the Delaware estuary, Limnol. Oceanogr.,
33, 1102-1115, 1988a.
Cifuentes, L. A., M. L. Fogel, J. R. Pennock, and
J. H. Sharp, Seasonal variations in the stable
nitrogen isotope ratio of ammonium in the Dela-
ware Estuary, Annual Report of the Director of
the Geophysical Laboratory, Carnegie Instn.
Washington, 1987-1988, Geophysical Labora-
tory, Washington, D. C, 114-121, 1988.
130
CARNEGIE INSTITUTION
Cline, J. D., and I. R. Kaplan, Isotopic fractiona-
tion of dissolved nitrate during denitrification in
the eastern tropical North Pacific Ocean, Mar.
Chem., 5,271-299, 1975.
Craig, H., Abyssal carbon and radiocarbon in the
Pacific,/. Geophys Res., 74, 5491-5506, 1969.
Craig, H., and R. F. Weiss, The Geosecs 1969
intercalibration station: Introduction, hydro-
graphic features, and total C02-02 relationships,
J. Geophys. Res., 75, 1641-1647, 1970.
Macko, S. A., Stable nitrogen isotope ratios as
tracers of organic geochemical processes, Ph. D.
dissertation, Univ. of Texas at Austin, 1981.
Mariotti, A., C. Lancelot, and G. Billens, Natural
isotopic composition of nitrogen as a tracer of
origin for suspended organic matter in the Scheldt
Estuary, Geochim. Cosmochim. Acta, 48, 549-
555, 1984.
Mariotti, A. and R. Letolle, Analyse isotopique de
l'asote au niveau des abondances naturelles,
Analusis, 6, 421, 1978.
Murray, J. W., H. W. Jannasch, S. Honjo, R. F.
Anderson, W. S. Reeburgh, Z. Top, G. E. Frie-
derich, L. A. Codispoti, and E. Izdar, Unex-
pected changes in the oxic/anoxic interface in
the Black Sea, Nature, 338, 411-413, 1989a.
Murray, J. W., and E. Izdar, The 1988 Black Sea
oceanographic expedition: Overview and new
discoveries, Oceanography, 1, 15-21, 1989b.
Rau, G. H., M. A. Arthur, and W. E. Dean, 15N/14N
variations in Cretaceous Atlantic sedimentary
sequences: implications for past changes in
marine nitrogen biogeochemistry, Earth Planet.
Sci. Lett., 82, 269-279, 1987.
Richards, F. A., Anoxic basins and fjords, in
Chemical Oceanography, J. P. Riley and G.
Skirrow, eds., pp. 1-41. Academic Press, New
York, 1965.
Spencer, D.W., and P. G. Brewer, Vertical advec-
tion diffusion and redox potentials as controls on
the distribution of manganese and other trace
metals dissolved in waters of the Black Sea, J.
Geophys. Res., 76, 5877-5892, 1971.
Velinsky, D. J., J. R. Pennock, J. H. Sharp, L. A.
Cifuentes, and M. L. Fogel, Determination of
the isotopic abundance of dissolved ammonium-
nitrogen from estuarine waters at the natural
abundance level, Mar. Chem, 26, 351-362,
1989.
mlneralogical and oxygen isotope
Analyses of Manganese Oxides Precipi-
tated by Spores of a Marine Bacterium
Kevin W. Mandernack, Marilyn L. Fogel,
Bradley M. Tebo*
and Jeffrey Post"
Manganese(II) [Mn(II)] oxidation in the
environment is generally catalyzed by
bacteria (Nealson etal., 1988). The mecha-
nism of this process, however, is poorly
understood, as are the products of micro-
bial Mn oxidation. In recent experiments,
Hastings and Emerson (1986) utilized the
catalytic properties of spores of a marine
bacillus (SG-1) to perform laboratory Mn
(II) oxidation experiments under environ-
mentally significant conditions of pH and
Mn (II) concentration. They concluded that
the concentration of Mn (II) in seawater
and the oxidation state of marine Mn ox-
ides are controlled by the rapid precipita-
tion of hausmannite (Mn304), which can be
microbially mediated. The Mn^ rapidly
undergoes abiotic disproportionation to
Mn02 because of low Mn(II) concentra-
tions typically found in the marine environ-
ment.
Hastings and Emerson's (1986) work
supports the model of Hem and Lind ( 1 983)
in which manganese oxidation is believed
to occur in two steps (Fig. 83). In the
laboratory at 25 °C, the initial product is
hausmannite, which can spontaneously pro-
tonate to manganite, y-MnOOH (Murray et
Scripps Institution of Oceanography, La Jolla,
CA 92093
Smithsonian Institution, Washington, D. C.
20550
GEOPHYSICAL LABORATORY
131
Fig. 83. [Taken from Hastings and Emerson, 1986].
Dissolved Mn2+ activity - pH equilibrium relation-
ships in aerated solutions (based on the work of
Hem and Lind, 1983). Each line represents the
equilibrium for the reaction written. Based on
these lines and with Mn and pH values, it is
possible to predict which Mn phase is thermody-
namically stable. A solution with Mn and pH
values to the right of line 2 would be expected to
precipitate Mn304. Data points from Hem and
Lind (1983), Murray etai, 1985 and Hastings and
Emerson (1986; referred to this figure as "this
study") are shown as well as the field which
represents typical seawater Mn concentration and
pH.
ai, 1985; Piper ef al., 1984). Theoretically,
Mn02, which has an oxidation state of 4+,
results from a disproportionation of Mn304.
In laboratory experiments that lasted up to
9 months, however, the highest oxidation
state observed was 3+ except under more
extreme conditions of high temperature or
pH (Murray etai, 1985). This fact is rather
enigmatic considering that manganese in
the oceans is usually highly oxidized
(Kalhorn and Emerson, 1984; Murray et
al.y 1984; Piper et al„ 1984). Hastings and
Emerson (1986) suspected that more oxi-
dized phases did not form in their experi-
ments because of high concentrations of
dissolved Mn(II).
Representative strains of bacteria that
could be responsible for Mn oxidation in
the ocean have been observed to precipitate
high-oxidation state manganese oxides
(>3+), even at high dissolved Mn(II) con-
centrations in solution (Tebo> unpublished
results). Because of the uncertainties that
persist in the mechanisms of Mn oxide
formation by these bacteria, we examined
these processes with oxygen isotope tracer
and x-ray powder diffraction studies on an
array of oxides precipitated by SG- 1 spores .
The oxygen isotopic composition of
dissolved atmospheric oxygen is very dif-
ferent from that of water. Thus, the relative
proportions of these two oxygen sources to
Mn oxides can be traced. Molecular oxy-
gen in air has a #80 of +23.5, whereas
seawater has a 5180 value of around 0 %o
(Kroopnick and Craig, 1976). The percent-
age of oxygen in Mn oxides derived from
both H20 and 02 can be predicted depend-
ing on the mechanism that is proposed for
manganese oxide formation. The model of
Hem and Lind (1983), for example, pre-
dicts that 25% of the oxygen in hausman-
nite should come from dissolved 02. In
contrast, Mn02 minerals formed by direct
precipitation from seawater without an
intermediate would be expected to have
50% of the oxygen from dissolved 02, as in
the following equation:
Mn2+ + 1/2 02 + H20 <=> Mn02 + 2H+. (1)
In this paper preliminary results of
mineralogical and stable oxygen isotope
investigations of Mn oxides produced by
132
CARNEGIE INSTITUTION
SG-1 spores in buffered seawater or deion-
ized distilled water with varying Mn con-
centration and temperature are presented.
The dual approach is useful for (1) estab-
lishing the identity of the mineral phase
formed under given conditions, and (2)
determining the portion of the oxygen at-
oms in the Mn oxide product that is derived
from either dissolved oxygen or water
The marine bacillus bacterium SG-1 is
unique since only spores,. a dormant non-
metabolic resting stage in the life cycle of
the bacterium, are capable of oxidation
while the growing vegetative cells are not
(Nealson and Ford, 1980; Rosson and Ne-
alson, 1982). SG-1 is able to oxidize Mn(II)
Table 17. Mineralogy of Mn Oxides Produced by SG-1 Spores
[Mn (11)1
3°C
RT
50°C
70°C
Seawater 10 uM
todorkite(?)
ND
ND
ND
100 uM
buserite— >
birnessite
buserite — >
birnessite
todokorite(?)
todorokite(?)
ImM
todorokite(?)
todorokite(?)
todorokite(?)
hausmannite
10 mM
U
manganite,
manganite &
feitknechtite(?),
hausmannite,
trace
manganite &
MnC03(?)
hausmannite
trace MnC03
Distilled HpiOuM
birnessite(?)
8.9A phase
ND
ND
ND
100 uM
buserite — »
birnessite
buserite — >
birnessite
U (10A phase)
hausmannite
ImM
ND
feitknechtite(?),
manganite
hausmannite
hausmannite
10 mM
feitknechtite -»
manganite >
hausmannite
hausmannite
manganite,
hausmannite
trace manganite
groutite (?)
RT = Room temperature
ND = Not determined
U = Unknown phase
? = Tentative identification
-> = Process that apparently occurred upon drying
GEOPHYSICAL LABORATORY
133
over a wide range of temperatures (2-80°C)
and Mn concentrations (10 nM to 10 mM),
in both seawater and distilled water. The
use of SG-1 spores as a catalyst permitted
sufficient amounts of manganese oxides to
form within a short time frame so that the
objectives of this study could be met. SG-
1 was grown to a fully sporulated state in a
20 mM HEPES buffered (pH 7.5) seawater
medium containing 0.5 g yeast extract and
2 g peptone per liter and 100 fjM MnC^.
The spores were harvested by centrifuga-
tion and purified to remove any remaining
vegetative cells and cell debris.
The SG-1 produced Mn oxides were
prepared in 0.22 jjm filtered seawater (SW)
or in deionized-distilled water (DW).
Additions of 1 M MnSOd and 1 M HEPES
buffer (20 mM final concentration, pH 8.0)
were followed by inoculation with SG-1
spores (Table 17). The oxides were pre-
pared during 2 week incubations. The ox-
ides that were formed were collected by
centrifugation and washed with DW. They
were stored wet and frozen until analysis.
Powder x-ray diffraction patterns were
collected for the oxides at the Smithsonian
Institution on a Scintag™ automated dif-
fractometer with copper Ka radiation. The
oxides were analyzed wet because drying
can change the crystal structure of certain
manganese oxides (Paterson et al., 1986).
Fig. 84. Transmission electron micrographs of spores before (A) and after (B) organic matter extraction.
Spores in Figure 84a were grown at 3°C in lOmM Mn. Spores in Figure 84b were grown at 25°C in
100^m Mn. Note the amorphous Mn minerals coating the spore wall. In Figure 84a, two different types
of Mn minerals are present: wavey sheets and granular material.
134
CARNEGIE INSTITUTION
Some of the oxides were finely crystalline
or poorly ordered, and consequently could
not be positively identified. Table 17 shows
the inferred mineralogy based on the XRPD
results and the conditions under which the
oxides formed. Transmission electron
microscopy was performed on several of
the samples at the Department of Earth
Sciences, Johns Hopkins University, by Jill
Banfield (Fig. 84).
The microbially -produced oxides were
extracted to remove organic material prior
to #80 determination. The method that
proved the most efficient without altering
the isotopic composition of the oxides was
a modification of a DNA purification pro-
cedure, followed by hypochlorite treatment
at 4°C. Frozen cells with oxide coatings
were sequentially extracted with phenol,
chloroform, and methanol with a final wash
with DW. This final residue was treated
with dilute (3%) hypochlorite overnight at
4°C. The oxide was washed extensively
with DW and dried in vacuo at 50°C.
The fluorination method with BrF5 was
used for determining the #80 of silicates
14.72 5.53
200.0
<0
CL160.0
O
f0.06
CO
— 40.0 \-
0.0
d-spacing, A
3.42 2.49
- 1 1 —
1.97
H 80 AS
CO
^120.0 J\ii>ttMh»^^ 60 3
' ' ' '
100
16 26 36
26, degree
Fig. 85. X-ray powder diffraction pattern of man-
ganese mineral before (10A peak) and after (7 A
peak) drying. Spores were cultured in distilled
water at 3°C in 0. 1 mM Mn
(Clayton and Mayeda, 1963). For most
analyses duplicate samples of 3 to 7.5 mg
of Mn oxide were reacted at 600°C for > 1 8
hr for the highest yields. The error of the
analysis for the c5180 of technical grade
Mn02 was ±0.5 %o.
Broad interpretations can be made from
Table 17 regarding Mn oxide mineral for-
mation. In general, hausmannite formed at
high temperatures and higher Mn concen-
trations in both DW and SW. Subsamples
collected from the 10 mM DW and SW
preparations after 4 days incubation at 50
and 70 °C were essentially completely com-
posed of hausmannite (data not shown).
Conversely, those that incubated for 2 weeks
usually contained a significant amount of
manganite (Mn(ITI)). The shift in mineral
structure with time may indicate that the
hausmannite was protonating to manganite
as indicated in reaction 2. Protonation may
occur faster at lower temperatures, as the
oxides from 10 mM solutions at room
temperatures were composed mostly of
manganite.
In general, higher oxidation state ox-
ides were observed to form under condi-
tions of lower Mn concentration and lower
temperature. Qualitatively, this is what
would be predicted from thermodynamics
(Fig. 83). Buserite was evidently precipi-
tated at 100 jUM Mn at 3°C and room
temperature in both waters. The XRPD
patterns for these samples showed the
characteristic collapse of a 10A peak to 7 A
upon drying (Fig. 85). The 7 A phase is
presumed to have a birnessite-type struc-
ture.
Manganese minerals from the seawater
incubations appeared as buserite or re-
GEOPHYSICAL LABORATORY
135
Table 18. Isotope Ratios of Manganese Minerals Precipitated from SG-1 Spores
[Mn] Medium 7
°c
fl^Hp
5180 02
fl'OMin
%np
%o2
Hausmannite
ImM
DW
50
-9.5
23.5
-10.3
100
0
10 mM
DW
70
-9.5
23.5
-11.9
100
0
ImM
SW
70
0.0
23.5
Birnessite
2.5
89
11
ImM
DW
25
-9.5
23.5
Todorokite
-0.2
72
28
100 ^iM
SW
25
0.0
23.5
7.7
67
33
ImM
SW
50
0.0
23.5
8.9
62
38
sembled a disordered todorokite with a
fixed d-spacing near 10 A. Todorokite was
tentatively identified based on the charac-
teristic 10A peak both before and after
drying. Todorokite-like phases were only
observed in the sea water incubations, and
therefore the interaction with the other
cations in seawater may be significant in
the mineral formation.
Spores were cultured in distilled water
or seawater with air as the source of oxy-
gen. The source of the oxygen to the man-
ganese minerals and therefore, the mecha-
nism of formation, was elucidated using
the following mass balance equation:
(&*0 np)(% O from H20)
+ (^O 02)(% O from Cg
-9%0 Mn oxide
(2)
The determination is approximate as there
is a ±0.5 %o uncertainty in the isotopic
analysis of the Mn oxide. The calculation
also assumes negligible isotopic fractiona-
tion in the incorporation of both water and
dissolved 02. Studies with slightly l ^-en-
riched water and oxygen in comparable
minerals gave similar results with the same
equation (Tebo et al.y 1987).
The percentage of H20-derived oxygen
in the minerals decreased with increasing
oxidation state (Table 18). The #80 of
hausmannite was always slightly more
negative than that of the water in which the
precipitation occurred. Based on equation
(1), all of the oxygen in the mineral must
have originated from water, unless the
mineral freely exchanged its oxygens with
water. The possibility of rapid and total
136
CARNEGIE INSTITUTION
exchange of the isotopic signature in the
manganese minerals was tested previously
(Tebo etal, 1987). Only 10% of the oxy-
gens in manganite exchanged after 12
months. One hausmannite sample contain-
ing manganite yielded a ^80 that was +2.5
more positive than that of the water. Tebo et
al. (1987) had previously measured the
isotopic composition of synthetic mangan-
ite, and concluded that approximately 75%
of its oxygen came from water and the
remaining portion from dissolved 02.
The higher oxidation state oxides con-
tain between 62 to 72% oxygen from water
and the remainder from dissolved oxygen
(28-38%). Because no dissolved oxygen
signal was measured in the Mn304, the
reaction mechanism for Mn oxidation must
be different from that proposed by Hem
and Lind (1983) and Hastings and Emerson
(1986). Equation (1) and the following
equation are closer representations of the
reaction mechanism indicated by isotopic
analysis:
Mn2+ + 1/4 02 + 3/2 H20
<=> Mn(III)OOH + 2H+.
(3)
The minerals formed by the spores were
commonly mixtures of mineral phases
(Table 17). Accordingly, the variability in
the isotopic results for a particular set of
growth conditions is most likely the bypro-
duct of the analysis of these mixtures.
In summary, both x-ray powder diffrac-
tion and isotopic tracer studies confirm that
manganese oxides are precipitated by a
different mechanism from that proposed
previously for chemical precipitates and
for spore catalyzed oxides. No traces of
hausmannite or manganite were found in
minerals precipitated at low Mn concentra-
tion. If hausmannite were the initial prod-
uct in the formation of the higher oxidation
state oxides, oxygen bonds in the crystal
lattice must be broken and reformed during
the rearrangement of structure to buserite
or todorokite. The involvement of molecu-
lar oxygen is indicated by isotopic ratios of
these two oxides, rather than in the initial
formation of hausmannite itself. With bet-
ter quantitative mineral identification we
should be able to determine whether Mn
oxidation mechanisms for chemically
driven systems are the same as those cata-
lyzed by living organisms.
References
Clayton, R. N., and T. K. Mayeda, The use of
bromine pentafluoride in the extraction of oxy-
gen from oxides and silicates for isotopic
analysis, Geochim. Cosmochim. Acta, 27, 43-
52, 1963.
Hastings, D., and S. Emerson, Oxidation of
maganese by spores of a marine Bacillus: ki-
netic and thermodynamic considerations, Geo-
chim. Cosmochim. Acta, 50, 1819-1824, 1986.
Hem, J. D., Redox coprecipitation mechanisms of
manganese oxides: particulates in water, in Ad-
vances in Chemistry, Series No. 189, M. C.
Kavanaugh and J. O. Lecke, eds., American
Chemical Society, Washington, D. C, pp. 45-
72, 1980.
Hem, J. D., and C. J. Lind, Nonequilibrium models
for predicting forms of precipitated manganese
oxides, Geochim. Cosmochim. Acta, 47, 2037-
2046, 1983.
Kalhorn, S., and S. Emerson, The oxidation state
of manganese in surface sediments of the Pa-
cific Ocean, Geochim. Cosmochim. Acta, 48,
897-902, 1984.
Kroopnick, P. and H. Craig, Oxygen isotope frac-
GEOPHYSICAL LABORATORY
137
tionation in dissolved oxygen in the deep sea,
Earth Planet. Sci. Lett., 32, 375-388, 1976.
Murray, J. W, L. S. Balistrieri, and B. Paul, The
oxidation state of manganese in marine sedi-
ments and ferromanganese nodules, Geochim.
Cosmochim.Acta,48, 1237-1247, 1984.
Nealson, K. H., and J. Ford, Surface enhancement
of bacterial manganese oxidation: Implications
for aquatic environments, Geomicrobiol. J., 2,
21-37, 1980.
Nealson, K. H, B. M. Tebo, and R. A. Rosson,
Occurrence and mechanisms of microbial oxi-
dation of manganese, Adv. Appl. Microbiol., 33,
279-318, 1988.
Paterson, E, J. L. Bunch, and D. R. Clark, Cation
exchange in synthetic manganates: I. alkylam-
monium exchange in a synthetic
phyllomanganate, Clay Minerals, 21, 949-955,
1986.
Piper, D. Z, J. R. Basler, and J. L. Bischoff,
Oxidation state of marine manganese nodules,
Geochim. Cosmochim. Acta, 48, 2347-2355,
1984.
Rosson, R. A., and K. H. Nealson, Manganese
binding and oxidation by spores of a marine
bacillus, J.Bacteriol., 151, 1027-1034, 1982.
Tebo, B. M, M. L. Fogel, A. T. Stone, and K. W.
Mandernack, Oxygen isotope tracers of manga-
nese oxide precipitation, EOS, 68, 1702, 1987.
Separation and Purification of Phos-
phates for Oxygen Isotope Analysis
Ellen K. Wright and Thomas C. Hoering
Oxygen isotope analysis of biogenic,
marine phosphates is a potentially power-
ful tool for determining the temperature of
ancient oceans because phosphates are more
resistant towards isotope exchange during
diagenesis than the conventionally-used
carbonates. However, published tempera-
tures deduced from analyses of phosphates
indicate several problems. Some of these
are clearly extrinsic such as unknown iso-
topic composition of the ocean water
through time. However, others may be
inherent in the analytical methods and these
problems will be discussed here.
There are stringent requirements for an
analytical method designed to measure
stable oxygen isotope ratios in geological
materials. First, all steps must be either
quantitative, in order to avoid isotope frac-
tionation due to incomplete yields, or else
a reproducible fraction of the sample must
be converted. Second, there must be no
addition of extraneous oxygen through
contamination, memory effects, or isotope
exchange induced during the analysis.
Three, the final product to be introduced
into the mass spectrometer must be free of
spectral contaminants. These criteria were
used in the design of an analytical method
for the isotopic analysis of phosphates. As
will be seen below, requirements one and
three are probably met, but two presents
difficulties.
A published method (Tudge, 1960)
involves bringing the phosphate into solu-
tion with nitric acid and then oxidizing any
organic matter with potassium permanga-
nate. This is followed by three successive
purification steps wherein the sample is
first precipitated as ammonium phospho-
molybdate which is filtered, dissolved and
precipitated as magnesium ammonium
phosphate. This compound is dissolved
and finally precipitated as bismuth phos-
phate, which is heated to remove water of
crystallization. Oxygen is extracted from
anhydrous bismuth phosphate by reaction
with bromine pentafluoride at 250°C. The
138
CARNEGIE INSTITUTION
oxygen is then converted to carbon dioxide
for analysis in the mass spectrometer.
This method can be considerably sim-
plified by substituting the lengthy precipi-
tation and dissolution steps with an ion
exchange separation method and then pre-
cipitating non-hydrating silver phosphate
instead of bismuth phosphate. In the first
step, 30 to 60 mg of a calcium phosphate
sample (-200 jjM P04) is dissolved in 2 ml
of 2 molar HF (digestion time -24 hours).
Calcium ions are precipitated as calcium
fluoride and are removed by centrifuga-
tion. The resulting solution is pipetted off,
and the residue is rinsed in 2 ml distilled
water, centrifuged again, and the rinse water
pipetted off and added to the solution.
The solution is neutralized with potas-
sium hydroxide ~ 2.2 ml of 2 molar solu-
tion) and added to a disposable ion ex-
change column (6 ml volume) packed with
a strong base anion exchange resin in the
hydroxide form [Amberlite™ IRA-
400(OH)]. The column is rinsed with dis-
tilled water until neutral and free of fluo-
ride ions. Then the phosphate is eluted with
20-30 ml 0.5 Molar ammonium nitrate.
The eluate is added to an ammoniacal solu-
tion of silver nitrate and precipitated as
silver phosphate according to the method
of Firsching (1961).
The phosphate concentration in the
eluate was monitored by the method of
Parsons etal. (1984a). Collecting fractions
of eluate from a 200 U.M P04 sample showed
peak concentrations of phosphate ions
between 1 and 12 ml eluate and only traces
after 14 ml eluate. Care should be taken that
all fluoride ions are removed from the
column in the rinse step as fluoride inter-
feres with formation of the blue color, which
is used to monitor the phosphate concentra-
tion.
A large sample of potassium dihydro-
gen phosphate, KH2P04 (> 99.97% pure)
has been reserved as a oxygen isotope
reference material. This compound is not
hygroscopic and does not form a hydrate.
Oxygen isotope analyses were done first on
the potassium dihydrogen phosphate, then
on silver phosphate precipitated directly
from potassium dihydrogen phosphate, then
on silver phosphate precipitated from po-
tassium hydrogen phosphate processed
through the anion exchange column, and
finally on bismuth phosphate precipitated
directly from potassium hydrogen phos-
phate. Each of these compounds was re-
acted with an excess of bromine pen-
tafluoride at 300°C according to the proce-
dure of Clayton and Mayeda (1963) to
yield oxygen. The oxygen was converted to
carbon dioxide by reaction with hot graph-
ite. Also several samples of National Bu-
reau of Standards phosphate standards (NBS
120b and NBS 120c) were processed
through the columns and the isotopic val-
ues compared with the published values of
Shemesheftf/. (1983).
Results
The results are shown in Table 19. A
sample of quartz, that had been analyzed
previously in several laboratories, was used
as a control. The published value for &sO of
this material is +7.18 %o relative to Stan-
dard Mean Ocean Water. The value shown
in Table 19 agrees within the precision of
GEOPHYSICAL LABORATORY
139
measurement, proving that the extraction
system and the mass spectrometry are
functioning properly. The mean #80 of the
potassium dihydro gen phosphate reference
material (11.91 %o) was higher than the
mean #80 values of several batches of
silver phosphate (11.76 to 10.05 %6) and
bismuth phosphate (10.55 %6) precipitated
directly from this reagent. The standard
deviation of the potassium dihydrogen
phosphate (0.2 %c) was comparable with a
standard deviation of 0.2 %o for the quartz
run on the same extraction system. How-
ever, the standard deviation increased for
the secondary precipitates (0.3 and 0.4 %c
for the silver phosphates and 0.5 %o for the
bismuth phosphate) and was highest for the
column processed silver phosphates (0.6 to
1.0 %6). The mean #80 of the column
processed NBS 120b standard was 19.81
%o with a standard deviation of 0.8 %c. The
published values of Shemesh et al. (1983)
are 20.1 %o and a standard deviation of 0.3
%o.
The decrease of #80 values and the
increase in the variance of the the silver
phosphate precipitated from the potassium
dihydrogen phosphate reference material
may be caused by contaminants. Silver
phosphate precipitated from phosphate ions
purified by the ion exchange column has a
different color (greenish or brownish yel-
low) and smaller crystal size than silver
phosphate precipitated directly in an
Table 19. Summary of means and standard deviations of #80 of silver phosphate, bismuth phos-
phate, and quartz.
Compound
Yield(%)
mean
Yield(% )
s.d.(%o)
N
precipitate
#8(0%o)
co2
KH2P04
11.91
102
0.2
7
Ag3P04*
100
11.76
104
0.4
8
Ag3P04*
100
11.21
104
0.3
7
Ag3P04* (silica free)
100
10.75
103
0.4
7
Ag3P04* (column)
98
11.55
104
0.7
6
Ag3P04* (column)
98
10.05
103
1.0
8
BiPO*
4
99
10.55
105
0.5
8
Ag3P04 (from NBS 120b)
98
19.81
100
0.8
12
Ag3P04 (from NBS 120c)
98
19.94
101
0.6
7
Si02»
6.99
101
0.2
21
&sO = {^QptOJ^O/1*®^ - 1 } x 1000 where the subscript x refers to the unknown sample and
std refers to standard Mean Ocean water.
* denotes that the compound is derived from the KH2POA isotope reference material.
• the published value for #80 of quartz is +7.18 %o
140
CARNEGIE INSTITUTION
ammoniacal solution from pure potassium
dihydrogen phosphate. Possible contami-
nants are small inclusions of silver chloride
and coatings of organic matter and silica on
the silver phosphate crystals. The chloride
ions stem from the ion exchange resin. The
Amberlite™ IRA-400(OH) resin is mainly
in. the hydroxide form. However, a small
amount of chloride ions is present, and
attempts to eliminate it by flushing the
resin with potassium hydroxide solution
caused decomposition of the resin before
complete elimination of the chloride.
The organic coatings stem from un-
avoidable decomposition of the anion ex-
change resin. Combustion of a silver phos-
phate sample that was derived from a col-
umn processed NBS 120c sample showed
the presence of 0.3 weight percent carbon.
The &2C of the carbon was -31 %o. This low
value is indicative of petrochemicals that
were used to manufacture the exchange
resin, and not of sedimentary carbonates
contained in the NBS 1 20c sample, or from
dissolved carbon dioxide co-precipitated
as silver carbonate. Traces of silica were
detected by microprobe analysis in some
batches of silver phosphate. This contami-
nant would yield oxygen upon reaction
with bromine pentafluoride. Silica was
detected in the ammonium hydroxide used
in the final precipitation step by the colo-
rometric method of Parsons etal. (1984b).
Silica-free ammonium hydroxide was pro-
duced by dissolving ammonia gas from a
cylinder into distilled water and storing the
solution in a Teflon™ bottle. Microprobe
analysis of column processed and directly
precipitated silver phosphate samples
showed no detectable silica levels when
silica-free ammonium hydroxide was used
in the precipitation step.
Discussion
The silver phosphate procedure for
analyzing oxygen is promising, although
several problems remain to be solved. It is
definitely easier and more rapid than previ-
ous methods and more amenable to proc-
essing large numbers of samples. The pre-
cision of measurement has steadily im-
proved until it now rivals that of the bis-
muth phosphate method, although there
seems to be a constant bias between the two
methods. The procedure fulfills two of the
three criteria described in the introduction.
The wet chemistry steps are monitored by
a sensitive and specific colorometric
method. No losses of phosphate were de-
tected in any of them. Firshing's (1961)
method for precipitating silver phosphate
is quantitative and accurate. The bromine
pentafluoride method of Clayton and
Mayeda (1963) has been used on silicates
for many years and found to be completely
reliable. Quantitative yields of carbon di-
oxide are obtained from pure silver phos-
phate by this method. The reproducible
results on the control sample of quartz are
encouraging.
The source of of the variance is most
likely due to small and variable amounts of
contaminants that are precipitated with the
silver phosphate and yield oxygen on reac-
tion with bromine pentafluoride. Future
work will focus on finding the source of
this contamination and methods for elimi-
nating it. Reasons for the systematic differ-
GEOPHYSICAL LABORATORY
141
ences between the silver phosphate and the
bismuth phosphate will be sought.
It is unlikely that the problem is due to
isotopic exchange of phosphate with water
during the course of the analysis. Keisch et
al. (1958), Bunton etal. (1961), andTudge
(1960) have shown that the phosphate ion
is inert to exchange under all of the condi-
tions used in this study.
References
Bunton, C. A., D. R. Llewellyn, C. A. Vernon, and
V. A. Welch, The reactions of organic phos-
phates. Part IV. Oxygen exchange between and
water and orthophosphatic acid, /. Chem. Soc.
London, 1636-1640, 1961.
Clayton, R. N., and T. K. Mayeda, The use of
bromine pentafluoride in the extraction of oxy-
gen from oxides and silicates for isotopic analysis,
Geochim. Cosmochim. Acta., 27, 43-54, 1963.
Firsching, F. H., Precipitation of silver phosphate
from homogeneous solution, Anal. Chem., 33,
873-87, 1961.
Keisch, B., J. W. Kennedy., and A. C. Wahl, The
exchange of oxygen between phosphoric acid
and water, /. Amer. Chem. Soc, 80, 4778-4782,
1958.
Parsons, T. R., Y. Maita, and C. M. Lalli, Determi-
nation of phosphate, in A Manual for Chemical
and Biological Methods for Seawater Analysis,
Pergamon Press, New York, 22-25, 1984a.
Parsons, T. R., Y. Maita, and C. M. Lalli, Determi-
nation of silica, in A Manual for Chemical and
Biological Methods for Seawater Analysis, Per-
gamon Press, New York, 25-28, 1984b.
Shemesh, A., Y. Kolodny, and B. Luz, Oxygen
isotope variations in phosphate of biogenic
apatites, II. Phosphorite rock, Earth Plan. Sci.
Let., 64, 405-416, 1983.
Tudge, A. P., A method of analysis of oxygen
isotopes in orthophosphates - its use in the
measurement of paleotemperatures, Geochim.
Cosmochim. Acta., 18, 81-83, 1960.
GEOPHYSICAL LABORATORY 143
Scientific Highlights of the
Geophysical Laboratory
1905-1989
H. S. Yoder, Jr.
Table of Contents
1. Introduction 144
2. Experimental Petrology 147
3. Hydrothermal Techniques 155
4. High-pressure Apparatus 157
5. Ore Petrology 159
6. X-ray Crystallography 161
7. Spectral Mineralogy 163
8. Field Petrology 165
9. Statistical Petrology 167
10. Extraterrestrial Petrology 168
11. Volcanology 172
12. Geophysics 174
13. Geochemistry 177
14. Thermodynamics and Calorimetry 181
15. Heat and Mass Transfer and Kinetics 183
16. Geochronology 187
17. Stable Isotopes 189
18. Biogeochemistry 192
19. War-time Studies 194
20. Closing Remarks 196
144
CARNEGIE INSTITUTION
1. Introduction
The concept of a geophysical labora-
tory was initiated by Clarence King at the
U. S. Geological Survey in 1882. The
laboratory was placed under the direction
of Carl Barus, and continued until govern-
ment funds were cut off in 1 892. Work was
resumed in 1900 under the direction of
George F. Becker, a field geologist with a
background in physics and mathematics,
and some support for its staff was obtained
in 1904 from the newly formed Carnegie
Institution of Washington (CIW). An
Advisory Committee in Geophysics had
been set up in the Institution in 1 902, mainly
at the instigation of Charles D. Walcott,
then Director of the U. S. Geological Sur-
vey, who also served as Secretary of the
Executive Committee of the Board of Trus-
tees of CIW. Charles R. Van Hise, who had
drawn up the plan for the work of the
U.S.G.S. geophysics group, served with T.
C. Chamberlin, Carl Barus, A. A. Michel-
son, C. D. Walcott and R. S. Woodward
(Chairman) on the Advisory Committee.
A proposal for an independent, pri-
vately endowed geophysical laboratory was
hastily prepared for the Committee by
Becker. A subcommittee was formed in
July, 1902, consisting of Chamberlin, Van
Hise, and Woodward to prepare a more
detailed report on the problems to be inves-
tigated from both physical and chemical
viewpoints. Van Hise and Becker were
sent to Europe in 1903 to consult their
colleagues on the continent about forming
a new laboratory. In addition to the reports
of Van Hise and Becker, the Trustees re-
ceived a detailed outline of suggested
geophysical investigations on 1 October
1903 from Frank D. Adams, Whitman
Cross, Joseph P. Iddings, James F. Kemp,
Alfred C. Lane, Louis V. Pirsson, H. S.
Washington, and John E. Wolff. (That
outline eventually served as the charge to
the new laboratory). Because Becker's
plan required a large proportion of the CIW
budget, the more modest plan of Van Hise,
focusing on geochemical and petrological
research, was deemed more acceptable. As
a result, Becker's deputy, Arthur L. Day,
was asked to submit a proposal whereby
small grants could foster programs from
which a larger scale endeavor might evolve.
Day received a grant in 1904 to enlarge
his studies onpetrogenesis at the U.S.G.S.,
and a grant was made to Becker for the
analysis of the strain relations in the flow
and rupture of rocks. These two tests of the
practicability of experimental solutions to
geological problems proved successful.
Later that year, R. S. Woodward, a member
of the geophysics subcommittee, succeeded
to the presidency of CIW, and he urged the
Board of Trustees to approve the construc-
tion of a geophysical laboratory. On 12
December 1905, Woodward, with the help
of Walcott, succeeded in getting the
Trustee 's approval, and within a few months,
Day was named as its first Director.
The news of the authorization did not
please Mr. Andrew Carnegie, who believed
the exceptional investigator should be
supported in his own environment. Others
were equally unhappy with the proposed
specific program of work. The physico-
chemical study of mineral solutions at high
temperatures was considered by Becker a
mere "detail" in the general need to under-
GEOPHYSICAL LABORATORY
145
stand the behavior of matter under the
extreme conditions in the earth. Neverthe-
less, the following broad program outlined
in 1902, reviewed and rededicated by Day
in 1927, has served the Geophysical Labo-
ratory well for over 80 years,
"The crust of the lithosphere has thus far
been the chief field of geology in the
narrower sense, since it contains the rock
record of the earth's past; and geological
studies have been directed chiefly to read-
ing and mapping this record, but the record
needs to be interpreted on broader and
deeper lines based on a profounder knowl-
edge of physical laws. To this end the data
of geology need to be correlated and uni-
fied under these laws on an experimental
basis ....
"Some of the salient problems of the outer
lithosphere are the origin and maintenance
of the continental platforms . . . and a
whole group of intricate questions of a
chemical and chemico-physical nature,
including the flow of rocks, the destruction
and genesis of minerals, the functions of
included water and gases, the internal trans-
fer of material, the origin of ore deposits,
the evolution and absorption of heat, and
other phenomena that involved the effects
of temperature, pressure, tension and re-
sultant distortion upon chemical changes
and mineralogical aggregations.
"These questions of the earth's outer part
are inseparably bound up with those of the
interior, and here the problems involve the
most extreme and the least known condi-
tions and make their strongest demand for
experimental light. The themes here are
the kinds and distribution of the lithic and
metallic materials in the deep interior, the
states of matter; the distribution of mass
and of density, and the consequent distri-
bution of pressure; the origin and distribu-
tion of heat; ... the secular redistribution
of heat within the earth and its loss from
the surface; the possible relations of redis-
tribution of internal heat to vulcanism and
to deformation and similar profound prob-
lems.
"A series of specific laboratory questions
arise from these, e.g., the effect of pressure
on the melting point of rocks carried to as
high temperatures and pressures and
through as wide range of materials, as
possible to develop the laws of constancy
or of variation; the effect of temperature
and pressure on thermal conductivity as
indicated above, and on elasticity, espe-
cially as involved in the transmission of
seismic tremors."
In subsequent years all of the recommended
areas of geophysical research have been
investigated by the staff.
The land for the new building in the
Azadia area of the District of Columbia
was obtained in March of 1906, construc-
tion began in June of that year from plans
prepared in 1904, and the building was
occupied on 7 June 1 907. The design of the
massive structural walls resulted from
Day's experience at the Physikalisch-Tech-
nische Reichanstalt in Berlin, where pass-
ing streetcars caused the galvanometers to
vibrate, Another innovative feature was
the erection of the machine shop on a
floating slab independent of the building.
The staff was recruited primarily from
the U.S.G.S. At the end of the first year of
operation in the new building, the staff
consisted of 3 physicists (A. L. Day, J. K..
Clement, W. P. White), a chemist (E. T.
Allen), 2 physical chemists (E. S. Shep-
herd, G. A. Rankin), and 2 petrologists (F.
E. Wright, E. S. Larsen, Jr.). Grants were
alsomadebyCIWtoG.F.Becker(U.S.G.S.)
and F. D. Adams (McGill Univ.) in geo-
physics and to T. C. Chamberlin (Univ. of
Chicago) and H. S. Washington (Locust, N.
J.) in geology.
146
CARNEGIE INSTITUTION
Year
■■/■■/■•A Biogeochemistry
10
20
30
40
50 60
Percent
70
80
90
100
Fig. 1. Change in proportional effort of regular Geophysical Laboratory staff (not including Fellows or
Visiting Investigators) in various fields of research with time and directorship. (George W. Morey served
as Acting Director, 1952- 1953; Robert B. Sosman served as Acting Director, 1918-1920, for Arthur L.
Day).
GEOPHYSICAL LABORATORY
147
Over the years, more than 2100 papers
have been issued from the Geophysical
Laboratory, but this represents only one
measure of the contribution of the staff.
The ebb and tide of the focus of the work
has indeed been great. Figure 1 gives a
crude picture of the change in effort of the
regular staff throughout the years. The
effort assigned to various fields is some-
what arbitrary in view of the overlap and
integration of the fields. The designated
fields, however, serve as focal points on
which to summarize the following high-
lights of the work. It is evident the Geo-
physical Laboratory has responded dynami-
cally to the needs of the science, develop-
ing the most rewarding directions as they
evolved. It is also evident that it is the
individual staff members who have made
the concept of interdisciplinary research so
successful. In accord with the original
wishes of Mr. Andrew Carnegie, the sup-
port of exceptional individuals has resulted
in a record of discovery and invention that
is extraordinary.
2. Experimental Petrology
Igneous Petrology
The charter for the Geophysical Labo-
ratory, as recorded in the report of the
Advisory Committee on Geophysics,
clearly stated the need to unify geological
field observations under physical and
chemical laws on an experimental basis.
They believed geologists wanted to know:
"...the melting points of rocks, the tem-
peratures at which rocks crystallize from
magma, the relative specific gravities of
melted and crystallized rocks, the effects
of slow cooling upon the crystallization of
rocks with and without pressure, the solu-
tion of one kind of rock in another, and, in
short, all the phenomena which concern
the transformation of magma to crystal-
lized rock and of crystallized rock to
magma."
That statement appeared in Year Book
No. 2 of CIW for 1903, which included a
detailed plan of investigation (pp. 195-
201). It has served as the principal guide-
line for the core program of the Geophysi-
cal Laboratory for 84 years.
The experimental approach to those
goals was immediately beset with prob-
lems of the most fundamental nature. There
was no generally acceptable temperature
scale above 200°C and standard calibration
points had not been established even though
several boiling and melting points were
commonly used up to about 1100°C
(Sosman, 1952). Primary pressure calibra-
tion was available only to 2 kbar. There
was, however, considerable knowledge
about the composition of rocks and the ten
most important oxides had been identified
by chemical analyses. The main advantage
lay with the incredible intuition and per-
ception of the geological advisors who had
acquired a remarkable qualitative sense
about how rocks were formed.
The most abundant mineral in the crust
of the earth is plagioclase, and the Ab-An
system had been selected for study at the
U.S.G.S. in the formative years of a geo-
physical laboratory. That first step in a
148
CARNEGIE INSTITUTION
much broader scheme of investigation of
the common rock-forming minerals was
undertaken by Day (1906-1935)1, Allen
(1907-1932) and Iddings (Univ. of Chi-
cago) with the financial support of CIW.
The results were published in 1 905 as paper
No. 1 of the new Geophysical Laboratory.
The liquidus was determined by the heat-
ing curve method from An100 to An26 v the
remainder of the now classical solid solu-
tion loop was deduced as Roozeboom's
Type I. The temperature calibration was
based on the Reichsanstalt scale for the
melting of Cd, Zn, Hg, and Cu. Platinum-
Rhodium thermocouples were employed
with a Pt-wound resistance furnace. Be-
cause of the great difficulty in determining
the exact temperature of complete melting,
Shepherd (1904-1946) and Rankin (1907-
1916) devised a new method in 1 909 for the
CaO-Si02, MgO-Si02, and Al203-Si02
systems in which the liquid was quenched
from a known temperature to a glass and
examined optically for crystals. The new
technique was applied in a reexamination
of the plagioclase system in 1913 by N. L.
Bowen (1910-1919; 1920-1937; 1947-
1952), who proved that the solid solution
loop was indeed as deduced. He also showed
that the depression and rise of the melting
temperatures of the endmembers An and
Ab, respectively, were in close agreement
withRaoult's Law of vapor pressure. Thus,
plagioclase may be considered an ideal
solution, however, the conditions under
1 Numbers in bold-face type are the years during which the
Staff Member or Fellow officially served the Geophysical
Laboratory. The single years in text type is for a reference
or an Annual Report citation, most of which are listed in the
"Indices of the Annual Reports of the Director of the Geo-
physical Laboratory" and the "Publication List of the Geo-
physical Laboratory."
which it deviates from ideality remains a
principal focus of today. The new Geo-
physical Laboratory temperature scale
(discussed at the end of this section), cali-
brated with lithium metasilicate, diopside
and anorthite, was applied in those experi-
ments.
In the short time of seven years, Bowen
had at his disposal the data for Ne-An,2 Ab-
An, Di-Fo-Qz, An-Fo-Qz (Anderson, 1912-
1918, 1915), Di-Ab-An, and CaO-AL/V
Si02 (Rankin and Wright, 1906-1944,
1915). From these few data and a large
measure of genius Bowen produced "The
later stages of the evolution of the igneous
rocks." In 1922, he published "The reac-
tion principal in pedogenesis," which Pentti
Eskola (1921) of Finland later called "the
most important contribution to petrology
of the present century." With only the
additional information in CaO-MgO-Al203
(Rankin and Merwin, 1906-1946, 1916),
MgO-AL/^-SiC^ (Rankin and Merwin,
1918), CaO-MgO-Si02 (Ferguson, 1912-
1919, and Merwin, 1919); Ak-Geh
(Ferguson and Buddington, 1919-1920,
1 920) and the immiscibility studies of Greig
(1922-1960, 1927) in FeO-Fe^-ALft-
Si02, Bowen assembled all his previous
petrological discussions in a set of Prince-
ton lectures published in 1928 as "The
Evolution of the Igneous Rocks." Although
Bowen expressed his great prejudice for
the well established theory of crystal frac-
tionation as the guiding principle in ac-
counting for the diversity of rocks, he also
2 Bowen's Ph.D was granted in 1912 by the Massachusetts
Institute of Technology in part for his study of Ne-An at the
Geophysical Laboratory between 1910 and 1912. He ap-
pears to have been the Laboratory's first "Predoctoral Fel-
low."
GEOPHYSICAL LABORATORY
149
provided the theory for testing alternative
views. No other book has had a greater
influence on the course of petrology. On
the fiftieth anniversary of Bowen's book, a
review of the same questions raised by
Bowen was made (Yoder, 1948 — , 1979),
and it was evident that he had indeed dis-
cussed the critical issues still relevant to-
day.
Most of the other staff members not as
geologically inclined as Bowen and his
associates chose to pursue a plan whereby
oxide systems, rather than mineral systems
were investigated. By taking each of the
principal oxides alone, then two at a time,
successively adding other components, the
melting behavior of the multicomponent
rock can be ascertained. Because there are
about 10 oxides essential to rocks, there
would be 45 binary, 120 ternary, 210 quar-
ternary, and 252 quinary systems. Clearly
judicious choices were necessary, because
not all of these systems are pertinent to
rock-forming processes. Some systems
were of exceptional industrial importance,
however, and a few staff members deviated
from the principal goals initially set. It was
fortunate indeed they did, because some of
the studies resulted in the establishment of
the optical glass industry in the U. S. (see
section on Wartime Studies), solution of
the clinker problem in Portland cement,
increased metal yields in the steel industry
by adjusting the slag compositions, and
others have been important to the high-
temperature refractories industry.
During the first thirty years, there was a
special effort to deal with the oxidation
states of iron, starting with Sosman (1908-
1928) and his associates. In experiments
on the Na20-Fe203-Si02 system involving
acmite and hematite, Bowen et al. (1930)
pointed out that the liquids actually con-
tained some FeO. From this experience,
Bowen and Schairer ( 1927-1970) set out to
ensure that equilibrium was obtained on
FeO-Si02 (1932). The successful break-
through in dealing with iron came as a
result of using an iron crucible held in a
purified stream of nitrogen. The equilibria
were well defined as long as native Fe was
present, even though the Fe203 content of
the liquids varied in a systematic way. With
this new technique, reproducible results
were obtained by them, with the help of
Posnjak (1913-1947), on Ln-Fa, CaO-FeO-
Si02, MgO-FeO-Si02, Ab-Fa, and Ne-FeO-
Qz. Later, Schairer (1942) completed a
major portion of the very complex system
CaO-FeO-Al203-Si02, for which he in-
vented the summary "flow sheet" to de-
scribe the major courses of fractionation of
liquids. As will be seen in the section on
Hydrothermal Studies, the next break-
through at the Laboratory in controlling the
oxidation state of iron was through the
oxygen buffer technique devised by Eug-
ster (1952-1958, 1957). A major tool for
petrologists was provided by Lindsley
(1960-1970) who calibrated the coexisting
pairs of Fe-Ti oxides for use as thermome-
ters and oxygen barometers. Because the
oxidation state of iron has a profound influ-
ence on the differentiation trend of a magma,
the quantitative measure of the partial pres-
sure of oxygen with minerals of wide spread
occurrence has been of exceptional value.
For systems requiring only Fe203 Hucken-
holz (1966-1973) and Yoder (1971) mixed
Pt02 in the starting materials held in Pt
150
CARNEGIE INSTITUTION
tubes to ensure an excess of oxygen for the
study of andradite and fenidiopside at high
pressures. Another method for maintain-
ing the partial pressure of oxygen, devel-
oped elsewhere (Darken and Gurry, 1945)
was the use of mixtures of gases, such as
C02-H2. Eventually it was possible to
define the Mg-Fe fractionation trends for
the major rock-forming phases as a func-
tion of /(02) and T.
Characterization of the principal rock-
forming minerals and investigation of their
stability relations has been an ongoing
program of the Geophysical Laboratory.
After the determination of Ne-Ks-Qz,
"petrogeny's residua system," by Schairer
and Bowen in 1935, the course of phase
equilibria research was set for years to
come. Systematically, Schairer and col-
leagues added the endmembers of each of
the phases formed early in magma (e.g.,
forsterite, anorthite, diopside, and enstatite)
to the relevant joins in the residua system.
The ternary feldspars required consider-
able effort because of their complex stabil-
ity relations at various pressures and even
more complex structural changes. The
sluggish growth problems were overcome
when suitable hydrothermal pressure ves-
sels were developed.3 The Ab-Or-H20
system was studied by Bowen and Tuttle
(1947-1953, 1950) in preparation for their
classic work on the granite system. The
ternary feldspars were investigated by
Yoder, Stewart (1955-1957) and Smith
1954-1957) (1957) at P(U20) = 5 kbar to
avoid the incongruent melting of sanidine
and to achieve a suitable rate of reaction. In
3 The development of techniques and apparatus is given in
the following sections on Hydrothermal Techniques and
High-pressure Apparatus.
the course of that work a direct method was
established for obtaining the water content
of the liquid as defined by the phase equili-
bria. The melting relations of Ab-An were
worked out by Lindsley (1968) at lOand 20
kbar after Bell (1964-1989) and Roseboom
(1956-1959) (1965) provided the P-T dia-
gram for Ab up to 50 kbar while Hays
(1965-1966, 1967) was investigating the
P-T diagram for An and related phases up
to 35 kbar. It became evident that plagio-
clase was not a stable phase at pressures
above about 32 kbar.
In a similar way, the stability regions of
the olivines, pyroxenes, spinels, melilites,
and an array of feldspathoids were mapped
out by many staff members and their asso-
ciates at atmospheric pressure as well as at
mantle pressures. Those phases found to
be stable only at high pressures, e.g., jade-
ite, pyrope, sodium melilite, provided new
constraints on the origin of rocks in which
they occur, particularly in solid solution. It
was essential that the stability relations of
each of the major mineral groups be well
defined before their interrelationships as
assemblages in rocks could be tackled.
(Space does not permit a detailed account
of the investigations of each of these major
mineral groups even though staff members
played leading roles in defining their sta-
bility relations.) Impatience, however,
usually led to somewhat premature ven-
tures into the study of more complex as-
semblages. One successful venture is given
in the following example.
After extensive experimental experi-
ence with feldspars and pyroxenes, the
time was considered appropriate to put
these two major mineral groups together as
GEOPHYSICAL LABORATORY
151
they are found in basalt. Study of the
multicomponent system one component at
a time would require an unreasonable
amount of time. For this reason a less
rigorous approach was taken in which the
natural igneous rock, presumed to have
been at one time all liquid, could be treated
as a single bulk composition in the multi-
component system. On this basis Yoder
and Tilley (1931, 1955-1967) examined
various natural basalt types and their high
pressure analogues. The results were pub-
lished in 1962 under the title "Origin of
basalt magmas: an experimental study of
natural and synthetic rock systems." Ac-
cording to the Institute for Scientific Infor-
mation, it was the most quoted paper in 43
core earth science journals in the period
1961-1980. Their model, the generalized
basalt tetrahedron, still serves as a guide for
testing other theories of basalt magma
generation.
Attention turned to the petrology of the
mantle in the 40-200 km depth range when
the Boyd-England (1926-1971) high-pres-
sure apparatus was developed. During the
very active period 1960-1970, Boyd
(1953—) and colleagues determined the
melting curves for Di, Ab, En, Fo, and Jd.
In addition, the stability fields of pyrope,
coesite, and jadeite were defined in early
studies. The transformation of basalt to
eclogite was outlined by Yoder and Tilley
in 1961. O'Hara (1962-1963) worked out
the melting relations of natural garnet peri-
dotite at 30 kbar and B. T. C. Davis (1962-
1965) provided the liquidus for Di-Py-Fo
at 40 kbar. The critical observation that
eclogite (equivalent to basalt) was at the
minimum melting composition of ompha-
cite-garnet was based on O'Hara's study of
that join at 30 kbar with purified natural
minerals and B. T. C. Davis' work on Di-Py
at 40 kbar. The solvus on the Di-En join
was measured by B. T. C. Davis and Boyd
(1966) at 30 kbar, and they applied it as a
geothermometer to the pyroxenes in nod-
ules from kimberlites. A particularly im-
portant phase diagram was published by
Boyd (1970) for the system Wo-En-Cor in
which the alumina content of orthopyrox-
ene was later (1973) calibrated as a geobar-
ometer. During this time, the limits of
alumina content of diopside were defined
on the solvus of the join Di-Jd by Bell and
B. T. C. Davis (1969) who had earlier
determined the melting relations of that
system. All of these studies and related
investigations had indeed generated a new
outlook on the origin of magmas, the depths
of metamorphism, and placed constraints
on the plethora of earth models that erupted
with the rise of plate tectonics.
The lower mantle and core captured the
attention of some of the staff and a large
number of Postdoctoral Fellows and Guest
Investigators when the diamond-anvil cell
became a practical tool. The phase stabili-
ties and equations of state of the metals and
major oxide components of the earth were
studied first, some to pressures near the
center of the inner core. Next, combina-
tions such as CaSi03- and (Mg,Fe)Si03-
perovskite were investigated, and then Aip3
was added to the system. The samples were
heated with a YAG4 laser at each pressure
increment to accelerate the phase transi-
tions. Phase diagrams were generated for
4 The yttrium aluminum garnet (YAG) was first synthesized
by Yoder and M. L. Keith (1947-1950) in 1949 at the
Geophysical Laboratory in a study of spessartite-yttrogar-
net.
152
CARNEGIE INSTITUTION
MgO-FeO-Si02 up to 700 kbar by T. Yagi
(1976-1978), Bell, and Mao (1972—,
1979). The partitioning of Mg and Fe
between the various solid solutions placed
great constraints on the geochemical mod-
els of the earth. When the equations of state
of the various structures were fitted to
similar equations inferred from seismic
data (Preferred Reference Earth Model), it
became evident that the lower mantle was
dominated by two perovskite structures,
mostly an orthorhombic ferromagnesian
silicate perovskite and a cubic calcium
silicate perovskite (CaSi03) as a minor
phase. After the sharp phase transition
between spinel and perovskite + magne-
siowiistite was discovered, Mao (1988)
proposed a model wherein the 670-km
seismic discontinuity is a phase-transition-
driven, chemical-composition boundary.
Because magnesiowustite strongly parti-
tions iron relative to silicate perovskite,
Mao, Bell and T. Yagi constructed a new
fractionation model of the earth whereby
the magnesiowustite would sink toward
the center and lose some of its iron to the
core by chemical reduction or dispropor-
tionation. The excitement generated by
these difficult and laborious experiments
on the interior of the earth rivals that of the
returned lunar samples.
Although much of the emphasis was
placed on understanding the interrelation-
ships of the crystalline phases, their melt-
ing behavior was, from a thermodynamic
point of view, highly dependent on the
character of the liquid. For this reason,
characterization of the liquid became a
major focus. The techniques for the meas-
urement of density, viscosity, thermal ex-
pansion compressibility, refractive index,
and heat contents were developed, but
understanding why they varied with com-
position required knowledge of the liquid
structure. With the advent of spectroscopic
tools, the local short-range structure of the
"amorphous" liquid could be ascertained.
The structural units were identified and
simple coordination rules developed. A
model emerged that could be used to ra-
tionalize the chemical, physical and thermo-
dynamic properties of melts. The model
was achieved by systematically examining
the melt structure of simple systems. The
work of B. O. Mysen (1972—) and D.
Virgo (1971 — ) and associates on systems
with the common rock forming oxides,
with and without volatiles, resulted in a
book by Mysen (1988) on the "Structure
and Properties of Silicate Melts."
Temperature Scale. The very first prac-
tical problem to be faced by the staff of the
Geophysical Laboratory was the calibra-
tion of a temperature scale above about
1100°C, there being no internationally
accepted scale in 1905. Because of Day's
experience at the Physikalisch-Technische
Reichanstalt in Berlin, the first apparatus
installed was the nitrogen-gas thermome-
ter. With a new design for improved accu-
racy of measurement, Day and Sosman
gave accurate data for a scale from 300° to
630°C with a direct determination of the
boiling point of sulfur. The scale was then
extended by them in 1910 to the palladium
melting point of 1 549.2°C. The fixed points
were corrected from the constant-volume
nitrogen scale to an absolute thermody-
namic scale and expressed by Johnston
GEOPHYSICAL LABORATORY
153
(1908-1916) and Adams (1910-1952)
(1914) as an e.m.f. for copper-constantan
and platinum-rhodium thermocouples from
0°to 1755°C. The principal fixed points for
gold, copper, diopside, and palladium be-
came known as the Geophysical Labora-
tory scale,5 and it is still used today. The
melting point of platinum, 1755°C, was
based on the incremental difference of
optical pyrometer measurements of the
Bureau of Standards between palladium
and platinum.
After WWI, an international confer-
ence met and the International Tempera-
ture Scale of 1927 was adopted. Silver was
set 0.3° higher; gold, 0.4° higher, palla-
dium, 5.5° higher, and the major change
was that platinum became 1773°C. In spite
of these recommendations, the scientists at
the Geophysical Laboratory retained their
own scale because of their belief in its
thermodynamic foundation and for consis-
tency with their previously published phase
diagrams. In the meantime, physicists noted
certain discrepancies in the physical con-
stants related to the International Tempera-
ture Scale. A new international committee
met in 1939, but WWII intervened and
their results were not published until 1948.
The changes nearly restored the fixed val-
ues below 1550°C to those of the Geo-
physical Laboratory scale. One significant
change related to the extrapolation to the
melting point of cristobalite which would
require a change from 1713°C to 1723°C.
Nevertheless, the values for diopside
(1391. 5°C) and pseudo-wollastonite
(1544±2°C), anorthite (1553±2°C), and
5 The fixed boiling and melting points below gold are listed
by Sosman (p. 522, 1952).
cristobalite (1713±5°C) have all been re-
tained by the Geophysical Laboratory as
calibration points not only for consistency
but also because they are within the errors
of experimental determination of the Inter-
national Temperature Scale, which is still
subject to change. It would be useful to
ascertain in the near future the melting
point for the important endmember min-
eral forsterite that is still known only as
1890±20°C. The corrections to the e.m.f.
of thermocouples used at high pressures,
recently studied but subject to debate, also
remain an important subject for reinvesti-
gation. In principle, the triple points of
mineral phase changes may serve as an
adequate P-T scale for interlaboratory
comparisons.
Metamorphic Petrology
Prior to WWII there were only a few
studies concerned with metamorphism at
the Geophysical Laboratory, even though
an extensive program had been laid out by
the Advisory Committee. Van Hise had
already completed his enormous (1286
pages) "Treatise on Metamorphism" for
the U.S.G.S. in 1904 and was well prepared
to advise the CIW. Some of the thermody-
namic principles were stated by Johnston
andNiggli (1912-1913) (1913), and Eskola
(1922) described the sequence of meta-
morphic rocks at a limestone-granite con-
tact. A major mapping study of the meta-
morphic zones of Dutchess Co., N. Y,. was
made by Barth (1929-1936, 1936). He
demonstrated that the rocks were appar-
ently at equilibrium because the assem-
154
CARNEGIE INSTITUTION
blages accounted for the bulk composition
at each stage of the metamorphism of the
argillaceous sediments.
The principal delay in tackling experi-
mentally the metamorphic facies of Eskola
as well as Bo wen's petrogenetic grid was
the lack of apparatus to contain the hydrous
minerals. That problem was subsequently
solved (see next section) by Bowen and
Tuttle (1949) who published the first sys-
tematic study on the P-T stability of some
of the critical metamorphic minerals in
MgO-Si02-H20. Shortly thereafter a se-
ries of papers appeared on systems defin-
ing the stability fields of chlorite, cordier-
ite, muscovite, biotite, chloritoid, stauro-
lite, garnets, and especially the AL^SiC^
polymorphs.
Serious objections to the facies and
isograd concepts were raised by Yoder
(1952) in a study of MgO-Al203-Si02-H20
in which representatives of all the then
defined facies were found to be stable at the
same P and T. He also raised the issue of
the role of water in metamorphism (Yoder,
1955) in regard to its presence in excess as
a free phase or where it was "deficient"
relative to the most hydrous assemblage.
The open vs. closed system with volatiles
became the major issue of the day. A
significant contribution to the problem was
made by H. J. Greenwood (1959-1963)
from experiments on analcite-H20-Argon,
Ct-Qz-H20, and others. The influence of
P(H20) less than P was clearly demon-
strated for those simple but highly informa-
tive reactions.
The apparent loss of large volumes of
water during the metamorphism of some
sediments and the gain of water by other
rocks, implied an active flow of fluids.
Oxygen isotopic evidence and phase analy-
sis led D. Rumble (1969-1971; 1973—)
and associates to estimate that the volume
of fluid that flowed through a calc- silicate
bed at Beaver Brook, N. FL, was from 1 .5 to
4.0 times the rock volume. Their paper
(1982) emphasized the process of rock-
fluid interaction, but the proportions of
fluid to rock are still being debated vigor-
ously. The concept of reaction progress
variables espoused by J. M. Ferry (1975-
1977) have been helpful in understanding
the process. More recently, the flow of
fluids has been related to structure by C. P.
Chamberlain (1985-1987). The transport
of heat with fluid was documented by
comparing the isotherms deduced from the
metamorphic reactions with the oxygen
isotopic data. The mapping of channeled,
hot, fluid-flow regions is potentially a valu-
able tool for ore prospecting.
The extensive studies on the amphi-
boles at the Geophysical Laboratory per-
tain to both metamorphic and igneous rocks.
In general, the role of amphibole in meta-
morphism is confined to low P(H20) and
relatively low temperatures, whereas most
amphiboles appear on the liquidus above a
few kbar P(H20) Although Allen and
Clement (1904-1907) (1908) found water
to be a component in tremolite, its struc-
tural role deduced by Schaller (1916) was
confirmed by Posnjak and Bowen (1931),
and a previous report of its formation from
a dry melt was withdrawn. A few years
later, fluoroamphiboles were formed inad-
vertently when NaF was used as a flux to
grow orthopyroxenes. (That study laid the
groundwork for a later extensive program
GEOPHYSICAL LABORATORY
155
at the U. S. Bureau of Mines for the pos-
sible production of synthetic asbestos.) The
first systematic P-T studies of tremolite
and pargasite were made by Boyd (1954).
A complex hornblende was produced from
a tholeiite basalt by Yoder and Tilley (1956),
and its stability field outlined up to 10 kbar.
The study by staff and associates of most of
the critical endmember amphiboles fol-
lowed: magnesioriebeckite, arfvedsonite,
glaucophane, ferropargasite, richterite,
ferrorichterite, K-richerite, anthophyllite,
cummingtonite, and aluminous anthophyl-
lite. The breakdown of amphibole at high
pressures (20-30 kbar) was first suggested
from thermodynamic calculations by
Greenwood (1 963), but it was several years
before the general concept was demon-
strated experimentally by others. The
possible storage of water in the mantle then
shifted from the amphiboles to the micas
and hydrous forms of commonly anhy-
drous minerals (e.g., hydrogarnets).
3. Hydrothermal Techniques
Before 1900 some 80 mineral species
had been synthesized, but not under con-
trolled and reproducible conditions. The
devices used for synthesis were generally
gun barrels closed by brazing, welding, or
a screw on cap. A pressure vessel having a
screw-on-cap fitted with a flat metal washer,
made in about 1900, was brought to the
Geophysical Laboratory by E. T. Allen
from the U. S. Geological Survey. Few
experiments were tried before WWI be-
cause the devices were unreliable even at
low pressures. The review of hydrothermal
mineral formation by G. W. Morey (1912-
1957) and Paul Niggli in 1913, however,
registered the strong interest in pursuing
those experiments. Further stimulus was
provided by the analyses of gases at Kilauea
volcano, Hawaii, by Day and Shepherd
who found H20 a principal component. In
1917 Morey and Earl Ingerson (1935-1947),
no doubt inspired by the demonstration by
Day and Shepherd of H20 in the gases of
Kilauea, designed a pressure vessel with a
contained copper or silver washer that would
retain reliably pressures somewhat over 1
kbar. With that apparatus Morey and C. N.
Fenner (1910-1937) investigated the
K2Si03-Si02-H20 system up to 1000°C and
about 340 bar by the isothermal polybaric
saturation method. Their detailed exposi-
tion of theory and experimental methods
recorded a major advance in hydrothermal
studies. There followed a series of experi-
ments in the "Morey bomb" on the simple
oxide systems with H20.
The next breakthrough came with the F.
H. Smyth (1919-1925) and L. H. Adams
(1923) device in which HOOT and 1 kbar
could be sustained with a C02-gas me-
dium. With the same device, R. W.
Goranson (1926-1951) achieved the first
systematic results on the melt curves of the
rock-forming silicates, albite-H20 and
sanidine-H20, as well as the melting curve
for granite-H20. These dramatic experi-
ments would appear to be the last major de-
velopments in experimental apparatus prior
to WWII. Fortunately, Morey had also de-
scribed a new design forhis pressure vessel
that was to lay fallow throughout the war.
He adapted P. W. Bridgeman's unsupported-
area principle to the closure and connected
156
CARNEGIE INSTITUTION
a H20 pump directly to the Morey bomb,
thereby controlling the water pressure in-
dependently of the temperature.
A review of the Geophysical
Laboratory's programs after WWII by the
staff and a large host of external colleagues
greatly emphasized the need to pursue
hydrothermal systems. An inventive and
imaginative investigator employed at the
Laboratory during the war under an Office
of Scientific Research and Development
contract, O. F. Tuttle, was hired to help
Morey and N. L. Bowen carry out this
mission. The hot-seal apparatus, eventu-
ally called the Tuttle apparatus, and the
cold-seal pressure vessels soon were de-
veloped by Tuttle, using alloys found supe-
rior for machine gun barrel liners during
WWII research at the Geophysical Labora-
tory (see section on War-time Studies). The
cold-seal, pressure vessels, also called "rod
bombs" or "test-tube bombs" depending
on their position in, or orientation of, the
furnace, remain the most widely used ves-
sels for hydrothermal investigation today.
A simple, cold-seal, pressure-vessel bench
employing an intensifier was built by Yoder
for operation to 850°C, and 5 kbar for runs
of one month or more duration. The sealed-
platinum-tube technique, first introduced
at the Laboratory by P. Eskola in 1922, was
used for a wide range of gases in addition to
Hfi.
Because of the inherent limitations of
the externally-healed pressure vessels,
Yoder constructed in 1949 an internally -
heated, gas-media pressure vessel that could
sustain 1650°C and 10 kbar for months.
These conditions were adequate for the
entire range of temperatures and pressures
in the continental crust within which 99%
of the rocks observed crystallized. With
the experience gained from investigations
on pure mineral systems, Yoder and C. E.
Tilley initiated in 1956 a study of natural
igneous rocks with and without water. In
the same apparatus, sulfur-bearing systems
were studied in gold tubes by Kullerud and
Yoder.
Other new techniques were developed
rapidly thereafter. For examples, the con-
trol of oxygen fugacity by a series of min-
eral buffers was demonstrated by Eugster
in 1956. The buffers initially consisted of
iron oxides whose stability had been deter-
mined by Darken and Gurry (1945). For
pairs of oxides in equilibrium, the P(02) is
fixed for a given temperature and pressure.
By surrounding a sample contained in Pt, a
metal that is permeable to hydrogen, with
water and a pair of oxides held in an outer
gold tube that is relatively impermeable to
hydrogen, the oxygen pressure can be
maintained. Eugster 's first demonstration
of the technique was in determining the
stability of the iron-rich endmember mica,
annite, between 0.5 and 2 kbar. The tech-
nique was adopted rapidly worldwide and
continues to be a principal method for
investigating redox reactions with a large
range of buffers. The equilibrium estab-
lished through the Pt membrane is of a
restricted type because the gas phase over
the buffer has a different composition that
than over the sample. Nevertheless, the
simplicity of the technique and wide range
of application have had a major impact on
igneous and metamorphic petrology. In
1976, J. D. Frantz (1972- ) redesigned the
Shaw apparatus for controlling the H2 fu-
GEOPHYSICAL LABORATORY
157
gacity. The influence of gas mixtures, such
as C02 and H20, on metamorphic reactions
was outlined by H. J. Greenwood, after he
resolved the gas-mixing problem that frus-
trated previous attempts to measure these
effects.
With the above-described devices it was
thenpossible to establish the stability ranges
of most of the common rock-forming
minerals that contained a volatile compo-
nent. The Postdoctoral Fellows carried
these techniques to other parts of the world,
and hydrothermal studies evolved expo-
nentially. The attainment of those condi-
tions to be found in the mantle and core
required new concepts and these are out-
lined in the next section.
4. High-Pressure Apparatus
The effect of pressure on the melting
"point" of rocks has been a high priority
question since the inception of CIW. The
compressibility studies of T. W. Richards
and W. N. Stull (1903) at Harvard Univer-
sity were supported by CIW, probably on
the recommendation of C. Barus, a mem-
ber of the Advisory Committee on Geo-
physics. [Barus (1893) was the first to
obtain an estimate of the change of melting
"point" of diabase with pressure from
measurements of the volume change, melt-
ing temperature, and heat of melting.] In
1906 P. W. Bridgman began his experi-
ments at Harvard on a range of materials to
ascertain the change of physical properties
with pressure. About the same time, John
Johnston began to develop apparatus for
sustaining both high temperature and high
pressure for geological applications at the
Geophysical Laboratory. The limits ob-
tained were 400° C at 2 kbar, and Johnston
and L. H. Adams (1911) were able to report
on the melting point changes with pressure
of Sn, Bi, Cd, and Pb. The study of the
compressibility of major rock-forming
minerals and rocks, however, remained of
central interest. The measurements on
dunite and basaltic glass were especially
pertinent to theories of the constitution of
the mantle. By 1923 Smyth and Adams had
achieved 1400°C at 1 kbar, and the range
was later extended in the same apparatus to
940°C and 3.7 kbar by Goranson. In this
time period, the manganin pressure gauge
and its calibration was perfected by Adams
and his colleagues as well as valves and
other high-pressure fittings critical to the
success of high-pressure experiments.
Nevertheless, the trial-and-error stage of
testing metals and packings continued. The
availability of an on-site machine shop and
direct accessibility to imaginative instru-
ment makers contributed immeasurably to
the development of new high-pressure
techniques.
After WWII, Yoder sustained 1650°C
and 1 0 kbar by 1 949 in an internally-heated,
gas-media apparatus. In 1956 substantial
increases in pressure and temperature were
achieved by F. R. Boyd, Jr. and J. L. Eng-
land who had adapted the Griggs and
Kennedy design of Bridgman's (1935)
carbide-piston "squeezer". They attained
800° at 10 kbar and 600° at 50 kbar, illus-
trating the breakdown of nepheline to jade-
ite and an unidentified phase.
The exciting announcement of Loring
Coes (Norton Co., Worcester, MA) that he
/
158
CARNEGIE INSTITUTION
had synthesized a large number of "high-
pressure" minerals others had struggled
unsuccessfully to obtain, came in 1953. He
had devised a solid-media, high-pressure
apparatus in which a hot-pressed Al^
core supported by steel bands and tungsten
carbide pistons were used. His experi-
ments were made up to 1000°C and 45 kbar
in that device. Because of its obvious
applicability to many geological questions,
the Coes apparatus was the basis for an
improved chamber assembly designed by
Boyd and England (1958, 1960) that could
sustain 1750°C and 50 kbar. The main
breakthroughs were in fitting a thermo-
couple up to the sample and in providing
adequate support for a reuseable tungsten
carbide core. Immediately after the an-
nouncement of the details of diamond
synthesis by the General Electric Co. in
1 959, Boyd and England successfully made
diamonds in their apparatus. That highly
successful design has been used by hun-
dreds of investigators around the world on
mineralogical and petrological problems
in the mantle. Extension of the range to 1 00
kbar was accomplished by them through a
two-stage device in which the piston was
supported by a substance such as KBr, that
undergoes a phase change, thereby main-
taining the support pressure at the phase
change.
On the basis of Bridgman's anvil con-
cept, Weir et al. (1959) at the National
Bureau of Standards patented a miniature
device in which the sample is squeezed
between two opposing diamond anvils, for
pressures up to 160 kbar at relatively low
temperatures (< 175°C). Substantial im-
provements by H-k. Mao and P. M. Bell in
1978 led to the production of 1.72 Mbar,
verified by three different methods of cali-
bration. In subsequent years, pressures
beyond those in the center of the earth (3.5
Mbar) were achieved and the current rec-
ord is 5.5 Mbar. The observation of plastic
flow in diamond during one super pressure
experiment raised an array of theoretical
questions. The sample in the high-pres-
sure, diamond-anvil cell can be heated to
3500°C with an appropriate laser beam to
which diamond is transparent. The trans-
parency of the diamond was a great advan-
tage for observing absorption, scattering,
and diffraction by the sample at pressure
with Mbssbauer, x-ray diffraction, Raman,
optical, infrared and other spectral devices.
The versatility of the Mao-Bell diamond
cell has resulted in an explosion of studies
on specific materials in the earth at ambient
conditions and generated a host of opportu-
nities for the chemist and physicist.
In spite of the availability of equipment
that reproduces the entire range of P-T
conditions in the earth, their exploitation in
general systematic studies of phase equili-
bria have been slow in the geological
community. The current trend toward
development of large-volume devices for
sustaining high pressures and high tem-
peratures appears to be driven by the desire
to understand those properties of the earth's
mantle and core that require measurement
of larger volume specimens. For examples,
measurement of sound velocity, rheology,
partitioning in multiphase assemblages,
single crystals for structural study, and
particularly those measurements requiring
a constant and uniform temperature cur-
rently require larger volumes than can be
GEOPHYSICAL LABORATORY
159
accommodated in the diamond-anvil cell
without special facilities.
Pressure calibration has been a difficult
problem in itself. The primary calibration
of force per unit area is dependent on an
evaluation of the friction on the piston. A
rotating, free piston loaded with weights
was used at the Laboratory for many years
up to 10 kbar. Secondary calibrations were
usually dependent on abrupt volume
changes in a simple material. The melting
of ice VI to water at room temperature was
given as 9630 bar at 25 °C by Adams (1931).
The freezing pressure of Hg at 0°C was
taken as 7492 bar, and the transformation
pressure of CC14 1 — » II was taken as 3326
bar (Bridgman, 1911, 1914). As higher
pressures were developed the transforma-
tion of Bi I -> II (25.2 kbar at 25°C) was
employed by Boyd and England. The tran-
sition of Till -» m (37.15 kbar at 29°C)
was also used by Boyd and England ( 1 958)
indirectly by measuring its abrupt change
in electrical resistence in a silver chloride
cell.
When the pressure range exceeded 50
kbar, the limit of the tungsten carbide core,
it was necessary to turn to still other meth-
ods. The shift in the lattice constants of
certain metals (Au, Ag, Cu, Mo, Pd, Pt) was
used as the primary pressure calibration.
The shift of the ruby Rj fluorescence line
became the new secondary calibration in
the diamond-cell apparatus. Because the
ruby fluorescence weakened above 1 Mbar,
it was necessary to return to the primary
method of using force per unit area, with a
substantial loss in accuracy, or to the mul-
tiple use of calibrated metals. It is likely
that new calibrations will be required as the
investigation of certain ranges of pressure
involved with major seismic discontinui-
ties in the earth progresses.
The leadership of the Geophysical
Laboratory in developing high-pressure
techniques has been maintained through-
out its existence. Although geology is
often described as an applied science, the
geologists at the Laboratory have provided
the basic physics and chemistry in the ex-
treme regions of pressure and temperature
for use in other sciences. Challenges to
existing theory have been proposed by
experimentation under the core and mantle
conditions in the earth, and the feedback to
the basic sciences has been rewarding. The
analysis of a geological problem by isolat-
ing and evaluating the effect of each sig-
nificant variable appears to be an effective
way to understand multivariate natural
phenomena.
5. Ore Petrology
The plans submitted in 1903 by Dr. C.
R. Van Hise for a geophysical laboratory
included the need for
"experimental studies on underground
solutions and the artificial reproduction of
natural minerals [that] will lead to correct
theories of ore deposition and also give
results of practical value, the magnitude of
which cannot now be estimated."
Although Van Hise championed but one of
the several schools of ore deposition, the
recommendation was clear. Within a few
years of the opening of the laboratory build-
ing, papers appeared on the characteriza-
160
CARNEGIE INSTITUTION
tion of sulfides, the analytical procedures
for determining sulfur, and on the physical
chemistry of sulfide systems. For example,
the chemistry of the secondary enrichment
of copper sulfides was unraveled by E. G.
Zies (1913-1949), E. T. Allen, and H. E.
Merwin in 1916. The studies of Allen and
Zies (1919) on the chemistry of hot springs
was applied by Fenner (1933) to ores de-
rived from igneous origins. In the same
year, Bowen ( 1 933) proposed that the heavy
metals would be concentrated in the resid-
ual fractions of a differentiating magma. It
became evident that hot springs were an
end product of ore deposition as well as
indicators of volcanic activity.
A benchmark paper on the Cu-Fe-S
system by Merwin and Lombard (1915-
1927) appeared in 1937. They laid out the
technique for holding synthetic and natural
samples at a defined vapor pressure of
sulfur (455 mm) and temperature in silica-
glass (vitreosil) tubes. The phase diagram
was of great importance to economic ge-
ologists because it helped to constrain the
temperatures and pressures of ore forma-
tion. With continued investigation, the
system has been found to be exceptionally
complex, and it remains one of the most
intensively studied systems even today.
The concept of buffers was clearly defined
even though the "equilibrium" obtained
was of a restrictive type, the buffer and
sample being at different temperatures.
Merwin and Lombard had recognized
many of the problems arising from the
inability to quench the run products. The
unmixing of solid solutions in the FeS-ZnS
system had been determined quantitatively
as a function of temperature by Kullerud
(1954-1970). After he joined the Geo-
physical Laboratory he applied this impor-
tant principle of geothermometry to other
systems and to the estimation of the tem-
perature of formation of natural ores. During
a highly inventive period, techniques were
evolved by Kullerud (1971) for dealing
with sulfur and selenium systems, beyond
the limitations of silica-glass tubes, up to
1400°C. Kullerud and his colleagues
showed that essentially all sulfide and se-
lenide systems exhibited liquid immisci-
bility. The first high-pressure-temperature
diagram for a sulfide, pyrite, was achieved
by Kullerud and Yoder (1959). Their ob-
servation of incongruent melting in pyrite
clearly showed that it could not be a mag-
matic phase in either basalts or rhyolites. A
similar conclusion was reached forpentlan-
dite in a study by Kullerud (1963). The
phase diagram for the economically impor-
tant system Cu-Fe-Ni-S was established by
J. R. Craig (1965-1967) and Kullerud
(1969).
The systematic study of the synthetic
sulfide systems have yielded information
on the stability of mineral assemblages in a
wide range of ore bodies. Not infrequently,
the discovery of synthetic phases preceded
their discovery as minerals in the ore body!
The temperature gradients in the ore bod-
ies, as determined by sulfide geothermom-
etry, have been especially useful in explo-
ration. The rapid response of some of the
sulfides to changes in temperature in the
laboratory have brought special insight into
the metamorphism of ore bodies, a concern
ofVanHiseasearlyas 1900. Re-equilibra-
tion to temperatures as low as 200°C for
some sulfides has been useful in generating
GEOPHYSICAL LABORATORY
161
a scale of closure for determining the kinet-
ics of the cooling of the ore body (Kullerud,
1967).
The interrelationships of sulfides and
silicates, the essence of ore petrology, did
not get underway until the early 1960's.
Kullerud and Yoder, after considering the
zones around the Bodenmais, Bavaria, ore
body, reacted sulfur with fayalite and ob-
tained pyrrhotite, magnetite, and quartz.
From these and other experiments with
various iron-bearing silicates, the concept
of sulfurization emerged that helped ex-
plain the apparent high-grade metamor-
phic aureoles around low-temperature ore
bodies. This concept was also demon-
strated and applied to the interrelationship
of sulfides and carbonates as well as sul-
fides and oxides. In the latter case, Kullerud
and colleagues produced a new type of
omission solid solution in magnetite when
reacted with sulfur.
The magmatic ores, especially those
occurring in the layered igneous intrusions,
have been of more recent concern. With an
exceptionally wide range of experience in
the layered intrusions of the world, Irvine
(1972 — ) and his colleagues characterized
the Pt-Pd ores of the Stillwater Complex of
Montana. The detailed analyses of the J-M
reef illustrated how magma mixing and
double diffusive convection play a major
role in the deposition of ores high in the
layered sections of silicate rocks.
6. X-ray Crystallography
The program in crystal structure deter-
mination was initiated at the Geophysical
Laboratory in 1919, just seven years after
Laue's discovery of x-ray diffraction, by
Ralph W. G. Wyckoff (1919-1927), who
had been invited to the Laboratory by Day
during a visit to Cornell. Atomic arrange-
ments were initially deduced intuitively
and then tested by the few measured reflec-
tions. Wyckoff derived a complete analyti-
cal expression of Schoenflies space group
theory to define all the possible arrange-
ments and used the x-ray information to
select the correct structure. In this way he
worked out the relatively simple structures
of the calcite group, dolomite, aragonite,
periclase, quartz, wiistite, and zircon. The
first structure to be determined at high
temperature from powder x-ray-diffraction
data was high cristobalite (Wyckoff, 1 925).
With the help of Wyckoff, C. J. Ksanda
(1914-1940), a Swiss-trained instrument
maker, designed the twin-gas tubes for x-
ray generation. Wyckoff joined with Eu-
gene Posnjak (1913-1947) to prove the
validity of Werner's coordination theory.
When Wyckoff departed to work at the
Rockefeller Institute on organic crystals, T.
F. W. Barth joined the staff in 1929. He
worked with Posnjak on the spinel prob-
lem, which resulted in their recognition in
1931 of "variate atom equipoints" - that is,
crystallographically equivalent sites that
could be occupied by chemically different
atoms. It was the key idea in understanding
many crystal structures, especially the
aluminosilicates.
According to Donnay et al. (Yearbook
68, pp. 278-283), only the Geophysical
Laboratory and Caltech had x-ray crystal-
lography programs carried on continuously
from 1919 to 1969. This activity continues
162
CARNEGIE INSTITUTION
at the Geophysical Laboratory and Caltech
in 1989.
The sulfide minerals were the focus of
attention of George Tunell (1925-1947).
He determined the structures of tenorite,
calaverite, sylvanite and krennerite. He
also derived the Lorenz correction factor
for equi -inclination Weissenberg films,
essential for use of the intensities of dif-
fracted x-rays. The Patterson-Tunell sten-
cils were a very popular aid for the compu-
tation of Fourier synthesis prior to the
computer era.
With the arrival of the powder x-ray
diffractometer in 1949, the identification
and characterization of synthetic mineral
phases, primarily carried out under the
microscope, was supplemented by the use
of comparative powder patterns. The cell
dimensions and volumes of the alkali feld-
spar solid-solution series were determined
and the existence of a high-order transition
established by Gabrielle Donnay (1950-
1952;1955-1960;1963-1970) and L. H.
Adams. The first applications of general-
ized symmetry to magnetic-structure de-
terminations were made by J. D. H. Don-
nay (Visiting Investigator) with colleagues
at the Brookhaven National Laboratory.
The Donnays were especially successful in
relating morphology to structure, which
led to additional generalizations of the Law
of Bravais. These studies were followed
with the discovery of the relationship be-
tween crystallographic axes and morpho-
logical features of the calcite skeleton,
termed biocrystals, in Echinodermata.
During his tenure as a Postdoctoral
Fellow, J. V. Smith (1951-1954) pursued
crystallographic studies of paracelsian,me-
lilite, and alkali feldspars, much of the
latter in collaboration with W. S. MacKenzie
(1951-1952; 1953-1957). Smith also pro-
vided the theoretical and structural basis
for polymorphism in the micas with Yoder.
Although a staff member for only a brief
period, Charles W. Burnham (1963-1966)
refined the structures of sillimanite, kyanite,
and mullite. He compared in detail the
crystal structures of orthoferrosilite, clinof-
errosilite, and ferrosilite HI, and pointed
out differences in these polymorphs that
have not yet been fully explained. In addi-
tion, he provided several computer pro-
grams whose successors are still being used
in crystallographic studies. A wide range
of crystal structures among the common
rock -forming minerals were refined by L.
W. Finger (1967 — ). He also developed
computer reduction programs for the elec-
tron microprobe, automated the x-ray dif-
fractometer, and contributed many tech-
niques for resolving the more complex
mineral structures. Finger and R. M. Hazen
(1976 — ) initiated single-crystal, diamond-
cell techniques that resulted in the accurate
measurement of both compressibility (to
200 kbar) and thermal expansion (to
1000°C). From these techniques the con-
cepts of cation polyhedral analysis as a
function of pressure and temperature
evolved.
Current studies include the adaptation,
initiallybyFinger,A.Jephcoat(1982-1989),
and H-k. Mao, of synchrotron radiation to
structure determination of very small single
crystals held at pressure in the diamond
cell and the characterization, primarily by
Finger and Hazen, of the structures of phases
with high-temperature superconducting
GEOPHYSICAL LABORATORY
163
properties. C. T. Prewitt (1986 — ) joined
the Laboratory as Director and collabo-
rated with Finger and Hazen on a variety of
projects including the first structure deter-
minations of high-temperature supercon-
ductors and high-pressure silicate struc-
tures containing octahedrally-coordinated
silicon that were synthesized in the cubic-
anvil apparatus at SUNY Stony Brook. As
part of his involvement in the superconduc-
tor research, Hazen wrote a book, "The
Breakthrough", which gives an account of
the discovery of the 1-2-3 high-Tc super-
conductor and subsequent attempts by dif-
ferent research groups to develop the dis-
covery.
In addition to their individual pioneer-
ing efforts, the hallmark of the crystallog-
raphers at the Geophysical Laboratory has
been their cooperative response to the needs
of other staff members in characterizing
the common rock-forming minerals and
especially the minute synthetic phases
produced. The determination of crystal
structure at a range of conditions and its
relationship to physical and chemical prop-
erties constitutes a major and essential
contribution to the success of the Geo-
physical Laboratory's mineralogical inves-
tigations.
7. Spectral Mineralogy
The field of mineral physics at the
Geophysical Laboratory evolved mainly
from initial efforts in mineral optics, arc-
Raman, and x-ray crystallography, followed
in the 1970's by Mossbauer, laser-Raman
and infrared spectroscopy, and in the 1 980's
by the entire array of spectral tools of
modem day physics and chemistry.
An early pioneer in the U. S. in the
application of microscopy to mineralogi-
cal and penological problems was F. E.
Wright. His design of apetrographic micro-
scope was adopted by a leading U. S.
manufacturer and his improvements sub-
stantially advanced its use. His book on
"The methods of petro graphic-microscope
research" in 1911 had great influence in
promoting the quantitative measurement
of the optical properties of crystals. An-
other major contributor to crystal optics
was H. E. Merwin. He developed with E.
S. Larsen (1907-1909) special immersion
media of unusually high refraction using
mixtures of amorphous sulfur and sele-
nium. His dispersion method for measur-
ing refractive indices of grains in immer-
sion liquids is still widely used. Because of
his demonstration of the relationship of
index of refraction, density, and composi-
tion of glasses, he contributed to many of
the phase equilibria studies of the Labora-
tory.
Some of the earliest studies of the Raman
effect, discovered in 1928, were carried out
in the U. S. by J. H. Hibben (1928-1939).
He provided detailed treatises on inorganic
compounds in 1933 and on organic com-
pounds in 1939. His principal successes
were in the speciation of organic com-
pounds, and he made a special effort to
apply the technique to the petroleum indus-
try. The low intensity of the Raman effect
using an arc lamp and photographic plates
was eventually enhanced with the advent
of laser light sources and photoelectric cell
recording of the spectra. Applications to
164
CARNEGIE INSTITUTION
the common rock-forming minerals and
their melts did not begin until the arrival of
Postdoctoral Fellow S. K. Sharma (1977-
1980) who installed a modern Raman spec-
trometer and a variety of laser sources with
the help of his colleagues J. D. Frantz and
D. Virgo. Another Raman system fitted
with a microscope was added, and a multi-
channel detector was introduced by Mao,
Bell and Hemley (1984 — ). These refine-
ments led to the techniques of single-crys-
tal, micro-Raman spectroscopy as well as
ultra high-pressure optical spectroscopy.
An outpouring of highly innovative studies
at pressures in the megabar range yielded
an array of studies of crystallized gases,
new mantle-phase structures, as well as
coordination changes in common miner-
als. The crystallized gases became the
"hydrostatic" media for subsequent dia-
mond-anvil experiments and initiated the
race to make metallic hydrogen. With
these new instruments for sustaining rec-
ord pressures, the winning of that race by
Hemley and Mao appears to have been
achieved in 1988 and documented above
2.5 Mbar.
The application of Mossbauer tech-
niques to atom site preference problems
was carried out by D. Virgo. Landmark
studies of the Fe-Mg ordering in olivines
(with L. Finger) and anthophyllite (with F.
A. Seifert, 1973-1974) included an impor-
tant estimate of the kinetics of the process
over a range of pressure and temperature.
The resulting time-temperature-transfor-
mation plots are useful in defining the later
stages of the cooling path of a metamorphic
rock.
One of the ultimate goals of vibrational
spectroscopy is in deriving the thermody-
namic properties of crystals that can be
used in calculating their phase relations. A
major step forward was taken in 1985 by
Hofmeister (1983-1987) and colleagues in
modifying a Fourier-transform, far-infra-
red spectrometer with a beam condenser,
an He-cooled bolometer, and a diamond
cell in order to obtain high-pressure spec-
tra. From the data collected in the modified
apparatus, combined with the data from the
Raman active bands, the heat capacity and
the Griineisen parameters could be ob-
tained. The equation of state for the Mg-Fe
olivines in particular will be at hand when
the temperature effects are investigated.
Brillouin/scattering from a five-pass,
interferometer was used by Mao, Bell, and
other colleagues to obtain the pressure
dependence of longitudinal- and transverse-
acoustic velocities and refractive index in
solid hydrogen and deuterium for the pur-
pose of obtaining its equation of state.
These techniques can now be applied to the
principal mantle phases to obtain their
elastic parameters that are so important to
the seismologist.
The important problem of heat transfer
by radiation in the earth was examined by
S. P. Clark, Jr. (1957-1962). The strong
onset of photon absorption in natural oli-
vine appeared to be an important process
that influenced the thermal regime in the
earth. With dual-beam, crystal-field spec-
tra H-k. Mao and P. M. Bell showed that the
radiation window broadened with pres-
sure, but the shift of the absorption edge
with pressure blocked the radiative trans-
fer.
Because of the enormous advantage of
the intense, parallel x-rays from a synchro-
tron source, Mao, Hemley, and colleagues
GEOPHYSICAL LABORATORY
165
have developed the appropriate devices for
carrying out diffraction at high pressures at
the National Synchrotron Light Source at
the Brookhaven National Laboratory, Up-
ton, NY. The intense beam, properly colli-
mated, is ideal for high-resolution, low-
atomic-weight elements, and small sample
diffraction. The determination of the crys-
tal structures of solid hydrogen and helium,
for example, has been a major contribution
to fundamental physics. The structures of
oxides and silicate-perovskites, critical in
the mantle, and those of iron and nickel-
iron alloys at core pressures have also been
ascertained in the diamond cell. These
impressive experiments have been ac-
claimed by the entire scientific community.
The assembly in the 1970's and 1980's
under one roof of the wide range of tools
described, supplemented by the use of elec-
tron spin resonance, nuclear magnetic reso-
nance devices, and transmission electron
microscopes at other institutions, have
resulted in a major revolution at the Geo-
physical Laboratory and in the field of
mineralogy. The quantitative approach to
mineralogy in which mineral properties are
characterized in order to obtain thermody-
namic parameters is of critical importance
in understanding materials under the ex-
treme range of conditions in the earth,
particularly where the direct measurement
of thermodynamic parameters is not yet
possible. Theoretical computational meth-
ods for predicting structural, thermody-
namic, and elastic properties has been
developed by Hemley for minerals under
these extreme conditions. In combination
with the on-going experimental mineral
physics research, a sound theoretical basis
for interpreting condensed phase behavior
is developing.
8. Field Petrology
Throughout the years, field measure-
ments have been an integral part of the
experimental program. Most staff mem-
bers have collected their own samples to
test new concepts, but the help of col-
leagues, especially in the U. S. Geological
Survey, has also been vital. The generosity
of museum curators has also been essential
in obtaining materials with the appropriate
properties for meaningful experiments.
Some staff members have relied heav-
ily on detailed mapping for the develop-
ment of ideas. The analytical work of H. S.
Washington (1912-1934) and E. G. Zies,
for examples, was guided by first hand
experience in the field. The classic studies
of Fenner at Gardiner River, Yellowstone
National Park, Wyoming, and Katmai,
Alaska, are still highlighted. No doubt his
work was well remembered by some of the
officials of Yellowstone National Park, who
watched with "considerable apprehension"
as Fenner began drilling operations in 1929
several hundred feet west of Old Faithful
geyser. The study of that core as well as one
from Norris Basin recorded the rock altera-
tion, which correlated well with the results
of Allen's analyses of the discharge waters.
Fenner 's papers gave rise to an extended
debate on the roles of assimilation and
magma mixing. He was a strong advocate
of superheat, and he believed that an ade-
quate heat content was necessary in the
magma for assimilation to be a major proc-
166
CARNEGIE INSTITUTION
ess. Bowen, on the other hand, assigned a
minor role to assimilation because he did
not believe superheat prevailed. Perhaps
the more widely known Fenner-Bowen
debate was in regard to the fractionation
trends of magmas. Whereas Fenner's field
studies led him to believe iron was concen-
trated in intermediate liquids, Bowen was
persuaded on experimental grounds that
iron was removed continuously through
the stages of fractionation. The iron-en-
richment trend through ferrogabbro be-
came labelled the Fenner trend, and the
basalt — » andesite — > dacite — > rhyolite
series was designated as the Bowen trend.
Later work by E. F. Osborn (1938-1945;
1973-1977) in which iron oxidation was
taken into account, has shown that both
trends can result depending on the condi-
tions.
A major contribution to the granite
problem resulted from the extensive col-
lections of granites by F. Chayes (1947-
1986). His techniques of modal analysis
provided critical evidence that related the
mineralogical composition of these rocks
to the minimum melting diagram of Tuttle
and Bowen (1958). The persuasive argu-
ment that liquid was involved in the gen-
eration of rhyolites and granites was made,
but whether they arose by partial melting or
fractionation depended on other criteria.
Various granites have been distinguished
by both modal and chemical analyses, yet
attempts to relate the pyroxene-, pyroxene-
amphibole-, amphibole, mica-amphibole-,
mica-, two-mica-bearing granites have not
been rewarding. The search for chemically
equivalent granites having different miner-
alogy has not as yet been successful, so the
question of heteromorphism in granites
remains open.
Layered intrusions have provided criti-
cal tests for many theories purporting to
relate magmas. Careful field observations
by T. N. Irvine have yielded many new
ideas. Before arriving at the Geophysical
Laboratory, he had made detailed studies of
layered intrusions on Duke Island (Alaska)
and Muskox (N.W.T.), Canada. Subse-
quent studies included Skaergaard, Still-
water, Bushveld, and other classic areas.
As a result of these experiences he charac-
terized the processes resulting from new
magma influxes, side-wall accumulation,
mass slumping, density currents, troughs,
double-diffusive convection, and accounted
for compaction effects, metasomatic ex-
change, replacement reactions, and the
products of magma mixing. Debate has
been vigorous, but the thoroughness of his
documentation, supported by laboratory
model experiments, have stimulated ef-
forts in field studies by others.
The collection and identification of ul-
trabasic xenoliths from alkaline rocks has
been a major interest of F. R. Boyd. On the
basis of many thousands of electron micro-
probe analyses of coexisting minerals in
nodules collected from kimberlite pipes in
southern Africa, Boyd and associates were
able to define the depth and temperature of
origin of the nodules. From these data and
the constraints of high-pressure experimen-
tal studies, Boyd was able to construct a
section of the mantle under the Kaapvaal
Craton and adjoining younger rocks. The
model illustrates the lithosphere-astheno-
GEOPHYSICAL LABORATORY
167
sphere boundary at about 150 km with a
root zone extending to 200 km under the
craton. The root zone extended into the
stability field of diamond and is also marked
by the presence of low-Ca garnets. From
these field and mineralogical studies ex-
tending over twenty-years, a unique struc-
tural and chemical model has emerged that
will no doubt be further supported as the
great variety of nodules are characterized.
The metasomatic effects, recrystallization,
deformation, and kinetic responses to the
dynamics of eruption of the nodules are
currently under investigation.
The classical studies of T. F. W. Barth in
1936 on the "metamorphism" of paleozoic
sediments in Dutchess Co., N. Y., resulted
in the definition of a new class of rocks
called "syntectic". His detailed minera-
logical and petrological studies outlined
the passage of sediment -> slate — > schist
— » gneiss — » augen gneiss -> intrusive
granite. During these events the rocks were
"heated and stewed in liquids of magmatic
and anatectic origin". The role of fluids
became an important aspect of the mapping
by Rumble and associates of metamorphic
rocks of New Hampshire. He developed a
dynamic model for the flow of fluids through
rocks during metamorphism from stable
isotope data. His recognition of hydrother-
mal graphite as the core marker of fluid
transport generated a wide variety of inves-
tigations.
The staff members of the Geophysical
Laboratory are convinced that experiments
derived from, guided by and tested with
field relations result in principles of lasting
value. The field-laboratory-field process is
reiterated until an acceptable interpretation
of the geological field observations is ob-
tained.
9. Statistical Petrology
Four years were required for H. S.
Washington to compile the "Chemical
Analyses of Igneous Rocks" that had been
published during the period 1884-1913.
He personally recalculated all the CIPW
(Cross, /ddings, Pirsson, Washington)
norms of the rocks. The monumental work
appeared in 1917 with the statement that
rock analyses were "indispensable", and
"the study of igneous rocks is in large part
the study of silicate solutions and their
equilibria, often complicated by the pres-
ence of volatile components, and is thus
regarded as essentially a special branch of
physical chemistry". The CIPW norms
contributed greatly to the classification of
igneous rocks, but more importantly re-
duced the chemical analysis of a rock to
simple endmember components that could
be experimentally investigated. The nor-
mative system incorporates an enormous
amount of petrologic intuition and percep-
tiveness derived from field experience. The
assignment of the analyzed constituents
was made in an order that reflected field
knowledge of the physicochemical behav-
ior of rocks not yet demonstrated experi-
mentally in 1902!
A "Manual of the Chemical Analyses of
Rocks" was also published by Washington
in 1904, revised through three subsequent
168
CARNEGIE INSTITUTION
editions, with due acknowledgment of the
advice of his friend W. F. Hillebrand, a
chemist at the U. S. Geological Survey at
that time. That manual served a generation
of analysts in the production of high quality
rock analyses.
The explosion of analytical data after
WWII presented a challenge of the first
order that was met primarily by the remark-
able developments in electronic data stor-
age and retrieval at first confined to nu-
merical data. An early compilation of over
1 6,000 analyses of Cenozoic volcanic rocks
by Felix Chayes, a petrologist, proved
extremely useful both in research and as a
guide to further work. As soon as it became
apparent that large quantities of non-nu-
merical data could be effectively stored,
sorted, and selectively retrieved in digital
form, Chayes turned his efforts to system-
atic development of a world data base for
igneous petrology. He developed a suitable
international organization for the construc-
tion and maintenance of such a base. Cur-
rently, the International Geological Corre-
lation Project 263 and a Subcommission of
the International Union of Geology share
responsibility for the base, version 2 of
which is distributed by World Data Center
A.
Current estimates of the number of
published rock analyses range from 65,000
to over 100,000. It is evident that a more
intensive effort will be required to use the
information now available. The eventual
correlation of phase assemblage (mineral-
ogy) with chemical composition, not yet
done for a single rock type, is essential for
the future progress of petrology. The least-
squares approximation technique pioneered
by W. B. Bryan (1967-1970), Chayes and
Finger (1969) was a major contribution to
the estimation of these important relation-
ships.
Another important problem, still largely
unresolved, results from ratio formation, in
particular, the percentage form of state-
ment used in reporting rock analyses. For
example, the closure resulting from divid-
ing the amount of each constituent by the
sum of all imparts special properties to the
interrelationships of the constituents.
Chayes summarized these properties in a
manual for students called "Ratio Correla-
tion" in 1971. Chayes' extensive experi-
ence in the modal analysis of rocks, men-
tioned in the section of Field Petrology,
was recorded in "Petrographic Modal
Analysis" (1956) in which he showed that
the point-counting of minerals in thin sec-
tions gave an accurate estimate of their
volumes in rocks, and he provided experi-
mental evaluation of the number of points
and sections required to achieve appropri-
ate precision.
10. Extraterrestrial Petrology
Meteorites
The nickel-iron core of the Earth was at
one time believed to be surrounded by a
zone of mixed iron and silicate, pallasite.
Because of the supposed similarity be-
tween some meteorites and the interior of
the earth, the pallasites were studied both
for their structure and range of composition
by Adams and Washington (1924). The
GEOPHYSICAL LABORATORY
169
meteorites also contribute to an understand-
ing of the origin and history of the solar
system. For this reason S. P. Clark, Jr. and
Kullerud undertook a study of Fe-Ni-S and
Fe-Ni-P to establish a buffered system of
taenite andkamacite with troilite or schreib-
ersite in order to derive the temperature of
formation of the meteorites. While in resi-
dence, P. Ramdohr (1960-1962, 1964)
examined over 340 polished sections of
stony meteorites to characterize in a sys-
tematic way the mineralogy of the opaque
phases. In typical fashion, he discovered
many new minerals as well as previously
described minerals hitherto not previously
observed in meteorites. The Fe-Cr-S sys-
tem was studied by El Goresy (1967-1968)
and Kullerud to account for the Cr-bearing
compounds in meteorites. They found the
sulfides responded more readily to shock
impact than silicates thereby explaining
their disequilibrium relations.
Over the years the Allende carbona-
ceous chondrite received the special atten-
tion of the staff because of its potential
applications to the early evolution of plan-
ets of alleged chondritic composition. The
P-T diagram for the Allende meteorite up to
30 kbar with and without H20 was investi-
gated by Kushiro (1962-1965, 1968-1969,
1971-1976, 1978-1987) and M. G. Seitz
(1971-1974). They demonstrated that
separation of the phases would yield the
layered structure presumed for the earth.
Other experiments dealt with the partition-
ing of elements between the metal, oxide
and silicate portions. The amino acids
were identified by Cronin (1974-1975,
1975) and some notion of the process
whereby abiotic organic compounds could
be formed from the precursor compounds
in the meteorite. The partitioning of boron
was achieved through an etching technique
that revealed the spallation recoil tracks in
whitlockite. The high concentration of
boron and other volatile elements in a
meteorite alleged to have formed at high
temperatures remains a mystery. The fas-
saite of the Allende meteorite was interest-
ing because it was found to be iron free and
contain trivalent titanium, confirmed by
high-resolution optical spectra. The chemi-
cal incompatibility of its oxidation state
with other minerals in the meteorite, such
as metallic iron and andradite, was evident.
Meteorites frequently contain minerals
formed at high pressures through shock.
For this reason they are important in under-
standing the phase changes that take place
at depth in the earth's mantle. The discov-
ery of ringwoodite, the spinel form of oli-
vine, indicated to N. Boctor (1977-1980),
Mao and Bell that the pressure was be-
tween 100 and 225 kbar during impact
metamorphism. The presence of majorite
in association with ringwoodite suggested
to them that a large pressure gradient ex-
isted in the order of 100 kbar to > 300 kbar.
Other features such as veins of inhomo-
geneous glass from incipient melting, frac-
turing, undulatory extinction, and mosaic-
ity are also indicative of high shock pres-
sures.
Lunar Samples
The return of the successful Apollo
missions with samples of the moon begin-
ning in 1969 provided one of the most
170
CARNEGIE INSTITUTION
exciting scientific opportunities for the
following decade. The incredibly fresh and
unaltered character of the rocks greatly
facilitated their study, but the fine grain and
shock metamorphism of the minerals were
the principal challenges. The characteriza-
tion of the fine-grained material generated
a new array of techniques and the unique
conditions of rock formation on the moon
led to the new field of comparative petrol-
ogy. Needless to say, the daily excitement
of discovery has not been equaled by the
arrival of any other set of specimens. The
initial stages of inquiry were predominantly
detailed mineralogical studies, followed
by experimental studies of both the natural
samples and synthetic analogues, and then
the testing of various models of the moon's
composition and structure. The entire ar-
ray of sample types (rocks, breccias, glass
fragments, and soils), was investigated by
the staff members.
Because of the special skills of the staff,
the opaque minerals received detailed at-
tention. Ilmenite was the major opaque
phase, but members of the chromite-
ulvospinel series, the newly discovered
armalcolite series, as well as troilite and
metallic iron alloy were studied. (Armal-
colite was also discovered independently
by several laboratories, and subsequently
named by combining the initial letters of
the names of the astronauts Armstrong,
A/drin, and Collins.) Haggerty (1968-1971)
found that the spinels were bimodal at
some sites, but other samples exhibited a
complete series of solid solutions. A new
pyroxenoid was discovered by Lindsley,
which he then prepared synthetically at
high pressures. The new phase, pyroxfer-
roite, is apparently metastable and had
persisted in that state for at least 3 b.y. ! The
olivines also appeared to be bimodal in
some samples. The Cr content of olivine
was more than twice that of earth olivines,
and more importantly, was in a reduced
state (Haggerty et al., 1970). The first
demonstration of Fe-Mg ordering in any
olivine was made by Finger on the lunar
material.
The pyroxenes were studied in excep-
tional detail by Boyd, whose electron micro-
probe data clearly reflected their chaotic
crystallization behavior. The zoning of the
pigeonites and oscillatory augite rims, for
example, suggested cooling and mixing
with new magma batches. Two distinct
rate-determining steps were found by Virgo
in the cooling of two lunar pigeonites. A
steady state of Fe-Mg ordering was achieved
at about 810°C involving a few hours of
time, then an exceptionally slow rate to
about 480°C below which no further an-
nealing was possible. (The 57Fe resonance
spectroscopy technique for determining
valence state and geometrical configura-
tion of iron in a crystal structure had been
previously proved useful in kinetic studies
of terrestrial pyroxenes). The plagioclases
contained Fe2+ and its site preference was
determined by Finger. The range of shock
features in the plagioclase were particu-
larly impressive. The lunar glasses were
studied with the new high-resolution, opti-
cal-spectra apparatus of Bell and Mao.
Some of the glasses were igneous in origin
and others formed from the meteorite
impacts. Various colored glasses were found
to result from the reduced states of Fe and
Ti. A particularly exciting event, although
GEOPHYSICAL LABORATORY
171
short lived, was the discovery of "rust"
(goethite and akaganeite) on some of the
lunar specimens, eventually attributed to
accidental contamination in the earth's
atmosphere. The event was not only an
exhibit of the great care taken by the ob-
servers, but also of the advanced state of the
art of characterizing fine materials.
The experimentalists also contributed
to the understanding of the formation on
the moon of the principal rock types, called
basalts and anorthosites even though they
were quite different in composition and
mineralogy from their earth-bound name-
sakes. Because of the extreme rarity of the
lunar samples, Muan and Schairer (1969)
made a synthetic analogue and studied its
behavior at a series of temperatures in iron
crucibles. This "basalt" composition
yielded pyroxene on the liquidus at 1 1 85° C
and had a solidus between 1075°-1090°C
with plagioclase and ilmenite. These val-
ues are not greatly different from those of
some earth basalts. Later, as more material
became available, lunar samples themselves
were studied at a range of P and T by
Kushiro and Hodges (1973-1974). Of the
three models of lunar composition they
tested, the model composition of Ganapa-
thy and Anders (1974) appeared to fit the
observations best. A most interesting ob-
servation of Bell and Mao in support of the
high-pressure experiments was recogni-
tion that a spinel + two-pyroxene symplec-
tite in olivine had the bulk composition of
garnet. The reaction of garnet and olivine
to the symplectite assemblage had been
previously demonstrated to be a high-pres-
sure reaction. Another set of pioneering
experiments was performed by M. G. Seitz.
He was concerned about the chemical frac-
tionation that resulted from the volatiza-
tion of materials during the impact events.
He showed in a vacuum furnace that all of
the alkalies and some of the iron was lost by
volatilization in short heating events. These
experiments were precursory to the major
program of study, described below, most
pertinent to the origin of the solar system.
Condensation Petrology
An understanding of the evolutionary
processes of the solar system require data
on the P, T, and composition at various
times and places as the solar nebula col-
lapses. The fundamental issue is whether
the minerals formed by direct condensa-
tion from a gas or by crystallization from an
intermediate liquid phase of the proto-solar
system. Experiments conducted at the
Geophysical Laboratory by Mysen, Virgo,
and Kushiro bear on these processes. The
materials found in meteorites are believed
to be representative of the oldest solar
system material. For this reason, they
worked on the very low-pressure region of
stability for minerals such as akermanite,
diopside, corundum, spinel, and hibonite,
found in the carbonaceous chondrites. By
means of a Knudsen-cell technique, in a
high-vacuum, high-temperature furnace,
they established the P-T curves separating
the crystal, vapor, and liquid regions. From
the experimental results, they suggest that
for the solar-gas composition, the pressure
would have to exceed 10 2 bar for liquid to
form, increasing with decreasing oxygen
fugacity. In general, it appears that the
172
CARNEGIE INSTITUTION
early solar nebula resulted from gas-crystal
reactions in the absence of melting at pres-
sures below 10"4 bar and at an/(02) at least
3 orders of magnitude below that of the
iron-wiistite buffer. In short, these dra-
matic experiments place severe constraints
on the collapse of the solar nebula and
emphasize the systematic chemical differ-
ences between the terrestrial planets as a
function of their distance from the sun.
Other constraints are placed on the
gaseous planets by experiments in a totally
different realm. Materials that are nor-
mally gaseous condense to form liquids
and crystals at high pressures . Studies up to
5.5 Mbar are particularly pertinent to the
early evolution of the solar system as well
as the interior of Jupiter, for example. The
P-V curves for crystalline hydrogen, deu-
terium, argon, neon, xenon, and oxygen
have been determined, and, in some cases,
their crystalline structure determined with
synchrotron-generated radiation. The first
single-crystal structure determination of n-
H2 at 54 kbar by Hazen et al. (1987) with
conventional x-ray diffraction was of fun-
damental interest to condensed-matter and
planetary physicists. In addition, methane
and water, important in the Giant Planets
were also studied. In this way Mao, Hem-
ley, and a large number of colleagues were
able to set limits on the conditions required
to collapse these gases in the nebula.
11. VOLCANOLOGY
In 1902, when the CIW was only five
months old, the city of St. Pierre, Martin-
ique, was destroyed by the eruptions of
Mount Pelee. The event no doubt helped
persuade the Board of Trustees of the need
for a geophysical laboratory that would
investigate the phenomena of volcanic
eruption. Field and analytical investiga-
tions of the rocks from an active volcanic
region by H. S. Washington (1906) were
promptly supported by CIW. His mono-
graph on "The Roman Comagmatic Re-
gion" (-Italian petrographic province)
presented a detailed description of the many
rare lavas characterized by the presence of
leucite. In 1911 the Hawaiian Volcano
Observatory was founded by T. A. Jaggar
(MIT) in collaboration with R. A. Daly
(Harvard), the Volcano Research Associa-
tion of Hawaii, and the Geophysical Labo-
ratory. The measurement of the tempera-
ture of the lava lake in Kilauea and the role
of gas in the flowing lavas were undertaken
by E. S. Shepherd and F. A. Perret that year.
These epoch-making studies involved the
collection, via a cable across the lava lake,
of an iron bucket dip sample of the lava and
immersion of a thermocouple pipe in the
bubbling lake itself. In the summers of
1911 and 1912, Day and Shepherd col-
lected and analyzed the gases in the active
part of the Halemaumau crater of Kiluaea.
They clearly demonstrated that water was
an original component of the lava, contrary
to the prevailing view of the nonaqueous
quality of magmatic gases. Furthermore,
they attributed the loss of gas as the reason
for the structural change from Pahoehoe
lava to Aa lava.
Day and Allen next turned to Lassen
Peak, CA., after its catastrophic outbreak
in May of 1915, the first eruption from a
volcano within the continental boundaries
GEOPHYSICAL LABORATORY
173
of the U. S. in the memory of then living
men. In addition to the description of the
eruptive activity, they focused on the types
of hot springs and fumeroles with field and
laboratory measurements. The change from
springs of acid character transporting py-
rite, to those of alkaline character was at-
tributed to the interaction of the hot waters
with the silicate rocks.
In the meantime, Mount Katmai, Alaska,
had erupted during June, 1912, but it was
1916 before an expedition, organized by R.
F. Griggs and supported by the National
Geographic Society, reached the area. On
that and subsequent expeditions to the Valley
of Ten Thousand Smokes were C. N. Fen-
ner, E. G. Zies, and E. T. Allen, who col-
lected rocks, fumerole encrustations, meas-
ured the temperature of the hot springs,
aspirated exhalations for the "insoluble"
gases, and helped in the geologic mapping.
They concluded that the vast sheet of sili-
ceous rocks were not lavas but were of
pyroclastic origin, ejected as rhyolitic
pumice through the fractured valley floor.
The fumeroles (100°-650°C) were, there-
fore, of deep seated origin and decreased in
temperature with time. Through succes-
sive observations, the mineralogy of the
encrustations changed as the temperature
dropped, and because many economic
minerals were formed, a relationship of ore
deposits to volcanic exhalations was estab-
lished. The analyses of the gases collected
showed the highest contents of HC1 and HF
that had ever been detected. The hybrid
nature of the rocks (also found at Lassen
Peak) led Fenner to believe that a super-
heated rhyolite magma melted fragments
of old andesitic lavas and incorporated
them into the Erupted pumice and ash. The
detailed analytical work, tied closely to the
mineralogy and geology, established the
value of a multidisciplinary approach to
geologic problems.
In the course of these studies, Morey
(1922) provided a new theory for the in-
crease in pressure of a cooling hydrous
magma, based on the continuity and uni-
variancy of the crystal + liquid + gas curve
in the KN03-H20 system. Some forty
years later Yoder (1965) pointed out that
most magmas were not saturated or uni-
variant and that explosive volcanism re-
sulted from an incremental drop in pressure
when gas was liberated from an initially
undersaturated magma. This concept arose
out of an experimental study of the syn-
thetic basalt system diopside-anorthite-HjO
at 5 and 10 kbar.
Four major studies in volcanology by F.
A. Perret, an associate of the Geophysical
Laboratory, were supported by the CIW.
Detailed descriptive monographs were
published on the Vesuvius eruption of 1 906,
the eruptions of Mt. Pelee, the volcano-
seismic crises at Montserrat, and finally,
because Perret was obliged to stay in the U.
S. during WWII, a compendium of his
studies of volcanoes around the world.
These unique contributions record the
observations of one of the world's most
perceptive students of volcanic activity.
The remarkable hydrothermal activity
of Yellowstone National Park was described
in another classic study by Allen and Day
(1935), in which is recorded the physical
and chemical changes of the fumeroles,
geysers and thermal springs over a period
of seven years. The relationship of hot
174
CARNEGIE INSTITUTION
springs to fumeroles was defined and the
differences in rock alteration from the acid,
mixed, and alkaline types contrasted. They
identified superheated water up to 138°C,
which gave rise to violent effervescence,
and they contributed to the problem of
discharge and its relation to rainfall.
A further contribution to the study of
hot springs and geysers was made by T. F.
W. Barth who carried out the laboratory
study of samples collected in Iceland. The
work was done during the summers of 1 934
and 1937, but publication was held up for
eight years while Barth was detained in
occupied Norway.
Note should be made of the extensive
field and chemical investigation by E. G.
Zies of the domes of the active volcano
Santiaguito and its ancient edifice Santa
Maria in Guatemala. Unfortunately, fail-
ing health prevented him from bringing
those studies, mentioned in a series of nine
abstracts, to the publication stage. The
combination of Zies' analytical skills and
H. E. Merwin's keen microscopy had gen-
erated a detailed picture of the mixing of
magmas and the digestion of individual
crystals.
The last major work on volcanic activ-
ity supported in part by the Laboratory was
F. R. Boyd's study of the welded tuffs and
flows in the rhyolite plateau of the Yellow-
stone National Park, Wyoming. Boyd
mapped the plateau, determined its strati-
graphy, and, most importantly, discovered
the Yellowstone caldera. His thermody-
namic analysis combined with experimen-
tal evidence showed that tuffs can have
temperatures of emplacement sufficiently
high for them to weld.
More pages have been written by the
staff on volcanology than any other field of
endeavor. Other projects on active volca-
noes have been undertaken by the staff in
recent years, but the field studies have been
primarily for the purpose of keeping touch
with the principal problems uncovered by
other workers. Field work has always been
encouraged at the Geophysical Laboratory
in order to investigate experimentally the
significant problems whose solution may
be applied broadly.
12. Geophysics
Although the name of the Geophysical
Laboratory implies that a large component
of the work would involve geophysics, the
classical fields of endeavor now included
under the term have played a small role
over the years. For brief periods, however,
gravity, heat flow, electrical conductivity,
thermal conductivity, density, magnetism,
tectonophysics, oceanography and seismol-
ogy, have all been investigated. Some of
these studies were cooperative with the
Department of Terrestrial Magnetism,
which was established in 1903. It initially
dominated the field of magnetism, comple-
mentary to the national bureaus, and was a
pioneer in explosion seismology after
WWII.
Gravity
One of the many interests of F. E. Wright
was the difference in gravity between the
earth and moon and the resulting differ-
GEOPHYSICAL LABORATORY
175
ences in geomorphology and isostatic
compensation. Pursuing this interest, he
persuaded F. A. Vening Meinesz in 1928 to
install his pendulum for making gravity
determinations at sea on a U. S. submarine.
From subsequent measurements, they
concluded that some oceanic deeps were
uncompensated whereas the Mississippi
delta was practically compensated in spite
of the enormous load of sediment laid down
each year. Inspired by the facility of occu-
pying a large number of stations at sea,
Wright and J. L. England developed an
improved torsion gravity meter mounted
on a truck so that twenty or more stations
could be occupied in a day. Although most
of the stations occupied were in the eastern
U. S., the apparatus was set up in 1940 on
an active volcano in Guatemala to assess
the changes in the magma chamber in
conjunction with other geophysical meas-
urements.
Heat Flow
An early attempt (1912) to measure the
thermal gradient in the crust was made by
J. Johnston and L. H. Adams. They used
both mercury thermometers and an electri-
cal resistance thermometer in wells as deep
as 5230 feet. It was believed that such
measurements might also have economic
importance in identifying layers rich in
coal or oil, indicated by a higher tempera-
ture gradient. An opportunity for measur-
ing heat flow in long tunnels occurred in
the construction of the Arlberg and Taverin
tunnels in Austria. With the underground
temperature observations and the labora-
tory measurement of thermal conductivity
of the various rocks, S. P. Clark, Jr. found
that relatively high geothermal fluxes ex-
tend into the eastern Alps. More recently
the relatively high heat flow (2.2 meal/
cm2sec) in Arizona was investigated by
Bell and R. F. Roy (Harvard), who related
the results to the gravity and seismology
data of the region.
Geotherm
The problem of the cooling of a primi-
tive earth was undertaken by L. H. Adams
(1924) after the revealing calculations of
Holmes (1916) and Jeffreys (1924). He
generated a "most probable" geotherm
down to 300 Km, assuming the age of the
earth was 1.6 b.y., the generation of radio-
active heat was constant, and the earth was
covered with a substantial molten layer.
Urry (1949) also calculated the geotherms
for the earth at various times taking into
account the exponential decay of radioac-
tive elements and the variation of surface
heat flow with time. Later, Clark (1961)
derived geotherms as a function of time for
various models calculated with the aid of a
digital computer. He concluded that the
distribution of radioactivity cannot be in-
ferred from the near surface heat flow.
Furthermore, the variability of heat flow
cannot be attributed to different degrees of
concentration of radioactivity in the outer
few hundred kilometers of an initially
homogeneous earth. An innovative ap-
proach to the geotherms was found by F. R.
Boyd, Jr. by applying the pyroxene geobar-
ometer and geothermometer. On the basis
176
CARNEGIE INSTITUTION
of the composition of coexisting pyroxenes
in nodules from kimberlites, he obtained a
quantitative measure of the geotherm. In
general, the geotherm derived from the
nodules substantiates the geophysical esti-
mates based on surface heat flow. His
results also showed that cratons have been
cool relative to oceanic plates for at least 3
b.y. One surprising result was an inflection
in the geotherm under northern Lesotho
that may be attributed to a region of partial
melting.
Tectonophysics
Prior to the opening of the Geophysical
Laboratory and during its formative pe-
riod, grants were made to F. D. Adams
(1902-1912) by CIW to study the flow of
rocks. A 1 20-ton press was set up at McGill
University to investigate the deformation
of marble, granite, diabase, and other rock
types at temperatures up to 1000°C. One of
the goals was to understand the origins of
crystalline schists, a subject of special
concern to Van Hise who served on the
CIW Advisory Committee in Geophysics.
A theoretical interpretation of the flow in
stressed solids by R. W. Goranson, using
thermodynamic potential relations for dif-
ferent physical conditions, was corrobo-
rated by experimental studies. Those ex-
periments were not published, presumably
because of Goranson's assignments during
WW II, but support for his theory was
provided by the more detailed laboratory
studies ofD. Griggs (Harvard). For a five-
year period, two Postdoctoral Fellows from
Yale undertook a field and experimental
program relating the conditions of flow to
the resultant plastic strain. On the basis of
these and other studies, E. Hansen (1964-
1968) developed the concept of "strain
fades" recorded in a book with that name.
Oceanography
The discovery of the high radium con-
tent of ocean bottom samples collected by
the auxiliary brigantine Carnegie led C. S.
Piggot (1925-1947) to develop a device for
coring ocean bottom sediments in 1935.
Using a gun-fired sample tube and hoisting
gear built at the Geophysical Laboratory,
he obtained cores with the cooperation of
the crews on the ship Atlantis (Woods Hole
Oceanographic Institution). Cores up to
ten feet in length were recovered from
depths as great as 2700 fathoms. Changes
in the orientation of the earth's magnetic
field with depth in those cores were studied
at DTM.
Magnetism
The pressure effect on the critical tem-
perature of magnetization of iron was found
to be negligible up to 3.6 kbar by L. H.
Adams and J. W. Green (1931). They
concluded that the nickel-iron core had
little influence on the earth's magnetic field
because the core temperature was well
above the Curie point! The Curie point was
investigated for a large number of materi-
als by Posnjak during 1936-1937, and the
effects of solid solution on the magnetic
properties of spinels were the subject of
investigation for many years to follow.
GEOPHYSICAL LABORATORY
177
Seismology
Of special significance was the theo-
retical contribution of L. H. Adams and E.
D. Williamson (1914-1923) in 1923 when
they deduced a formula that related the
compressibility and density of rocks to the
seismic velocities of the longitudinal and
shear waves. In this way the laboratory
measurement of the density and compressi-
bility of rocks and minerals constrained the
kinds and proportion of phases in earth
where the seismic velocities were known.
Observational seismology had been rec-
ommended by the Van Hise Committee as
early as 1903 and proposed on numerous
occasions thereafter. The Director of the
Geophysical Laboratory, A. L. Day was
appointed chairman of the CIW Advisory
Committee in Seismology in 1921 . On his
Committee's recommendation, a program
of study was outlined and a Seismology
Laboratory built in Pasadena, California,
in 1926 in cooperation with the California
Institute of Technology. The studies were
administered by the Committee until 1 Jan
1937 when the Seismological Laboratory
was turned over to Caltech. That Labora-
tory was primarily concerned with natural
earthquakes; however, following WWII,
the DTM initiated a cooperative program
in explosion seismology. Three members
of the Geophysical Laboratory staff, J. W.
Greig, J. L. England, and G. L. Davis
(1941-1978), helped select the seismome-
ter sites for their geological advantages and
on occasion occupied those sites to receive
signals from quarry blasts and the destruc-
tion of old military explosives, for ex-
amples. In another more recent coopera-
tive project with DTM, the velocity of
transmission of both longitudinal and shear
waves in partially molten peridotite was
measured directly up to 10 kbar by T.
Murase (1976-1980) and colleagues.
13. Geochemistry
In the formative stages of the Geophysi-
cal Laboratory, the major debate about its
program of work concerned physical meas-
urements, promoted by G. F. Becker, and
chemical measurements, advocated by C.
R. Van Hise. A compromise resulted in
physical chemistry, with the emphasis on
chemistry, a position strongly supported by
C. D. Walcott who was Director of the
USGS and Secretary of the CIW Board of
Trustees. There was no disagreement on
the great need for application of the quan-
titative principles of physics and chemistry
to the science of geology. Although, geo-
chemistry, loosely defined, pervades most
aspects of geology, the following high-
lights of the work of the Geophysical Labo-
ratory are restricted to only a few investiga-
tions highly dependent on chemistry.
Element Partitioning
The accuracy and precision of the analy-
sis of the minutest amounts of an element
has reached an exceptionally high state
with the wide variety of techniques avail-
able. Nevertheless, the analysis of a min-
eral, a unique combination of elements,
178
CARNEGIE INSTITUTION
still requires research — it is not a routine
matter. The data produced are of a most
fundamental character in determining P, T,
time, and reaction path of the mineral and
its host rock. Analytical chemistry, how-
ever, has greatly outstripped the calibra-
tion of those data in defining the conditions
endured by the rock.
Phase equilibria studies have defined
limits of solid solution for a large number
of the common rock-forming minerals.
Some of those studies have already been
mentioned in the section on Experimental
Petrology; however, the minor and trace
elements would appear to be a more accu-
rate measure of conditions because they
tend to obey the thermodynamic laws of
dilute solution. The emphasis, therefore,
has been on the partitioning of elements
such as Ni, Cr, Ti, alkali metals, alkaline
earths, and various rare earths for crystal-
liquid, crystal-vapor, as well as crystal-
crystal equilibria. The major productive
period was 1970-1980, and perhaps the
principal reason for the diminution of inter-
est after 1980 can be attributed to the reali-
zation that the partitioning coefficients were
much more sensitive to bulk composition
than previously envisaged. The enormity
of the task of calibrating trace elements
then acquired dimensions beyond the scope
of a small laboratory dedicated to pioneer-
ing ventures.
In 1953, Eugster and colleagues initi-
ated some experiments on the partitioning
of Cs, Tl, and K between sanidine and a
fluid phase at a series of temperatures from
500° to 800°C and 1 and 2 kbar. Cesium,
for example, more readily entered the sani-
dine structure at high temperatures than at
low temperatures when Cs/K was 0.0002
to 0.01, but the enrichment was much less
than observed in nature. Eugster attributed
the difference to the great enhancement
that results from fractional crystallization.
These experiments appear to be the first
direct determination of the distribution
factors of minor elements in silicates.
The technique that facilitated the meas-
urement of crystal-liquid partition coeffi-
cients was fission-track mapping on mus-
covite or on emulsions, which when devel-
oped revealed the concentration of the
radioactive element. Seitz described the
method in 1973 for the partitioning of 235U
and 230Th between diopside and a liquid in
Di- Ab- An, a simple basalt system. He used
less than 30 ppm of the spikes and con-
cluded that the activity coefficients were
independent of concentration. In another
series of experiments he examined the
partitioning of 151Sm between diopside and
the simple basalt liquid to compare with the
previous results of Kushiro and Masuda
( 1 970) on clinopyroxene in a natural basal-
tic liquid. He found that the coefficients in
synthetic systems were considerably lower
than those in the natural system. Although
he indicated that chemical and physical
conditions affect the measurements, he also
attributed the higher values in the natural
system to non-equilibrium conditions.
Subsequent detailed studies by Mysen
(1976) and his associates clearly demon-
strated that the concentration range of
Henry's law behavior was limited, and the
partition coefficients became composition-
ally dependent. In the course of a few
years, it was shown that the partition coef-
ficients were dependent on pressure, tern-
GEOPHYSICAL LABORATORY
179
perature, bulk composition, presence of
other phases, and on the available sites in a
crystal for specific trace elements. Even
the structure of the melt was found by
My sen and Virgo (1980) to influence the
partition coefficients. To all those factors
can be added the influence of/(02) mdf(S2)
on the partitioning of Ni between olivine
and iron sulfide melt. The calibration of
any partition coefficient indeed required
careful control of all the variables.
Throughout the productive period, a
wide range of natural rocks were melted at
various conditions to determine specific
partition coefficients in an empirical ap-
proach to defining their conditions of for-
mation. Some of the rock types investi-
gated included basalt, kimberlite, perido-
tite, komatiite, and other ultramafic rocks,
primarily for the purpose of relating pre-
sumed source rocks to their partial melts. It
became evident that the partial melts, as
represented by the alkali basalts, were much
too enriched in the light rare earths to have
been generated from such sources by even
small degrees of melting (Harrison, 1977-
1979, 1979). That conclusion gave strong
support to the concept that metasomatism
plays an important role in the mantle.
Aside from the many experiments on
crystal-crystal partitioning, two ingenious
investigations on liquid-liquid and crystal-
vapor partitioning must be mentioned. In
1975, Watson (1975-1977) measured the
partitioning of spiked elements in coexist-
ing immiscible liquids in the K^O-Al^-
FeO-Si02 system. The compositions of the
quenched glasses were determined by elec-
tron microprobe. From these data, he
concluded that deviations from Henry's
law were confirmed for several of the ele-
ments; there are cation interaction effects;
and that there are distinct differences be-
tween the ways various cations are accom-
modated in the acid melt relative to the
iron-rich melt. The results provided a useful
test of speculations on the origin of mafic-
felsic associations such as basalt and rhy-
olite.
The partitioning of elements between a
water-rich vapor and the constituent miner-
als of a garnet peridotite was investigated
by Mysen (1978). With the beta-track
technique he measured the partition of Ce,
Sm, and Tm in its minerals and obtained the
trace element content of the coexisting
vapor by mass balance. He demonstrated
that the REE patterns were highly pressure
dependent. In summary, it was clear that
the REE content of the crystals are similar
to the depleted nodules from the mantle and
that the metasomatizing fluid in the mantle
was, therefore, probably similar to that
observed in the experiments.
Mineral Solubility
The initial studies on the solubility of
minerals were carried out in 1915 on KN03
and KC1 by L. H. Adams, who was inter-
ested in freezing point depression in dilute
solution. He used an interferometer, rather
than the customary refractometer, to achieve
higher precision in the analysis of the solu-
tions. About the same time, Johnston and
Williamson, paying particular attention to
the species dissolved in the solution, inves-
tigated the solubility of calcite. During the
depression years and up to WWII, consid-
180
CARNEGIE INSTITUTION
erable efforts were made in studying the
simple systems NaCl-H20, K2S04-H20,
B203-H20, and KCl-Hp, with emphasis
on obtaining thermodynamic properties.
With the development of high-pressure
equipment, P-V-T data were developed for
common salt solutions. One significant
breakthrough was the study by Goranson
(1936) of Ab-H20 in which he determined
the solubility of water in silicate melt, that
is, the high-temperature end of the three-
phase solubility curve. Other staff mem-
bers focussed on the CaS04-H20 system
with various salts in order to understand the
deposition of gypsum and anhydrite in salt
water. Posnjak (1941) came to the conclu-
sion that gypsum was most likely to be
deposited from ocean water and might be
converted to anhydrite after deposition.
During those years attempts were made
to resolve the experimental problems of
determining solubility at high P and T by
developing a stirring and filtering device
(Morey and Burlew 1936-1952, 1938) so
that the fluid could be quenched independ-
ently of the crystals. The increase in solu-
bility with water pressure was of special
interest, and in one study the weight of
solids dissolved in the fluid reached almost
half. Much of the theory for the solubility
curves had already been worked out, how-
ever, a paper by Morey and Niggli (1913)
was instrumental in guiding the laboratory
research. It was not until 1940 that Morey
and Fleischer (1936-1938) provided the
background for two-volatile systems, such
as K20-Si02-C02-H20, that are so impor-
tant to the entire range of geological prob-
lems. Some of the stimulus for the study of
the solubility of minerals in aqueous solu-
tions no doubt came from the detailed stud-
ies of other staff members on the study of
sublimates around fumeroles and volca-
noes.
After WWII and with the development
of many new tools for generating P and T
and characterizing products, attention again
turned to the P-V-T of simple aqueous
solutions. Morey was then able to study
Na20-Si02-H20, which exhibited retro-
grade solubility: The solubility of sodium
disilicate fell to almost zero at the first
critical end point, but became important
again at the upper part of the solubility
curve. The determination of the solubility
of quartz in steam up to 600°C and 2 kbar
was of practical importance in relation to
the fouling of turbine blades as well as in
the understanding of the formation of quartz
veins. Morey and Hesselgesser (1949-
1953) demonstrated that the solubility of
some minerals was incongruent in the vapor
phase - a property that became of special
significance in metasomatism. In this post
war period, Yoder (1958) studied the melt-
ing curves of Ab-H20, Sa-H20, Di-H20,
An-H20, Ne-H20 and Qz-H20 up to 10
kbar, but none of the systems showed a
second critical end point on the solubility
curve.
The next major thrust was on the solu-
bility of ore minerals in aqueous solution.
A new apparatus was developed by H. L.
Barnes (1956-1960) in which he could
sample and analyze the gas phase from a
reaction vessel at P and T. In this way, he
obtained quantitative solubility measure-
ments of sphalerite (ZnS) in H2S- saturated
water. In addition, he found evidence that
a bisulfide complex was the most probable
GEOPHYSICAL LABORATORY
181
transport mechanism for sphalerite. Bar-
nes also observed that adding NaOH to the
ZnS-H2S-H20 system did not change the
phase relations significantly; however, in a
study with Ernst, it was demonstrated that
NaOH lowered the stability of brucite, for
example, in a major way.
After an almost ten-year hiatus, solubil-
ity work again resumed in the new dia-
mond-anvil, high-pressure cell. By direct,
visual observation, the solubility of gyp-
sum was measured up to 8 kbar by A. Van
Valkenburg (1975-1980) and his colleagues
Mao and Bell. They also observed the
phase CaC036H20, ikaite, which had been
found in a carbonatite deposit submerged
in arctic waters.
With the development of infiltration
models for mass transport in hydrothermal
rock systems, the need for mineral solubil-
ity data became acute. Because fluid inclu-
sions in many ore deposits contained chlo-
rides, geothermal well waters are often
brines, and evaporate minerals are associ-
ated with copper porphyry deposits, Frantz
and his colleagues investigated simple
systems in the presence of HC1. Study of
the reaction of HC1 with talc, hausmannite,
albite and hematite resulted in the conclu-
sion that the metals were dominantly asso-
ciated, e.g., MgCl2°, above 400° to 600°C
at 1 and 2 kbar. With knowledge of the
associated species, accurate solubility in-
formation could be calculated for those
minerals investigated and a host of others.
In view of the above importance of
ionization behavior of electrolytes, Frantz
joined with W. L. Marshall (Oak Ridge
National Laboratory) to measure the elec-
trical conductance of salt solutions as a
function of pressure and temperature. A
large array of simple systems were investi-
gated, including carbonates, hydroxides,
and fluorides. The results were immedi-
ately applicable to steam-generated corro-
sion in nuclear power plants in addition to
their fundamental importance to the prin-
ciples of element concentration in ore
deposits. Current research is focusing on
the determination of the PVT properties of
mixed volatiles in which synthetic, fluid-
inclusion techniques are used. Methods
are being developed for the accurate analy-
sis of individual fluid inclusions in the
silicates coexisting with ore minerals with
new sophisticated microanalytical tech-
niques. Concurrently, a high-pressure cell
has been designed to examine the Raman
spectra of solutions at 600°C and 4 kbar to
ascertain the species transporting the vari-
ous metals important in ore deposits.
14. Thermodynamics and Calorimetry
The first director, A. L. Day, while at-
tending Yale occupied an office in the same
building as J. Willard Gibbs, and presuma-
bly Gibbs influenced Day's scientific fo-
cus. Later, on returning from Germany,
Day served as the personal emissary of the
Berlin Physical Society, advising Gibbs of
his election as president of the Society,
which Gibbs declined for reasons of age. It
is no wonder, therefore, that thermody-
namics became a major factor in the work
of the Geophysical Laboratory. Intensive
study sessions were conducted by mem-
bers of the staff, and several members
achieved international recognition for their
182
CARNEGIE INSTITUTION
interpretation and application of thermo-
dynamic principles. After a series of pa-
pers on laws of chemical equilibria, hetero-
geneous equilibria, and phase rule prob-
lems, G. W. Morey (1912-1957) was asked
to contribute an article to the "Commentary
on the Scientific Writings of J. Willard
Gibbs." Another major contributor to
thermodynamics was George Tunell, re-
nown for his careful exposition and explicit
derivations, particularly in regard to open
systems.
The definition of activity, chemical
potential, and related thermodynamic quan-
tities, especially their variation with tem-
perature and pressure was the principal
area of expertise of L. H. Adams. A treatise
on the "Thermodynamic relations in multi-
component systems" by R. W. Goranson
was published by CIW. He, too, was in-
volved in the thermodynamic treatment of
activity as it applied to solutions. But for
the practical PVT relations in solutions,
whether they were organic or inorganic,
one turned to R. E. Gibson (1924-1946).
Experimentalists consulted Morey for
application of Schreinemaker's principles,
especially the Morey-Schreinemaker's
coincidence theorem, in the solution of
their phase diagrams.
It was N. L. Bowen who led the way in
using the phase diagram to derive thermo-
dynamic properties. The diopside-anorthite
system is often used to illustrate those
principles, and his diopside-albite-anorthite
system is the current model for describing
thermodynamic functions. The theoretical
guidance of thermodynamic principles was
evident throughout all the work of the
Geophysical Laboratory.
The experimental determination of th-
ermodynamic properties was at various
times a significant part of the work of the
Laboratory. The construction of a calo-
rimeter was in the hands of W. P. White
(1904-1935), who was best known for the
White double potentiometer adopted by
the leading manufacturer of precision elec-
trical measuring devices. White was greatly
concerned with the factors that resulted in
high precision and accuracy, and his mono-
graph in the American Chemical Society
Series on "The Modem Calorimeter" rec-
ords detailed analysis of each facet of the
experiment. He provided specific heats for
the various forms of silica and some sili-
cates, including the feldspars, in which a
drop calorimeter was used.
Calorimetry was considered a high pri-
ority subject of investigation after WWII.
A hydrofluoric-acid-solution calorimeter
was built by F. C. Kracek (1923-1956) and
his colleagues T. G. Sahama (1947-1949)
and K. J. Neuvonen (1948-1950), with
improvements on the successful Bureau of
Mines (Berkeley, CA) design. At that time
jadeite was under intense study from both
a theoretical and experimental viewpoint,
so the thermodynamic data for the critical
reaction nepheline + albite — > jadeite were
obtained. With the newly determined heats
of formation as well as the determination of
the other relevant parameters by other
workers at the Laboratory, L. H. Adams
was able to calculate the pressure-tempera-
ture curve of stability for jadeite, as yet not
synthesized in the laboratory. In short
order, the heats of formation of the plagio-
clases, alkali feldspars, olivines, and the
hypersthenes followed. One of the divi-
GEOPHYSICAL LABORATORY
183
dends of the calorimetry program not re-
corded, was the large number of purified
mineral separates prepared that were valu-
able in related studies.
With the material and financial help of
the Geophysical Laboratory, T. G. Sahama
was able to build his own calorimeter at the
University of Helsinki and continue min-
eralogical studies with K. J. Neuvonen. In
more recent years the determination of
thermodynamic properties has been car-
ried out by calculation from more funda-
mental parameters. The elegant work of A.
H. Hofmeister (1983-1987), who used the
spectral methods of mineralogy, is outlined
in the section on Mineral Physics. The
thermochemical properties of silicate
glasses and liquids was of special interest
to Pascal Richet (1983-1984) in collabora-
tion with his colleague in France, Y. Bot-
tinga.
15. Heat and Mass Transport
and Kinetics
In all of the phase equilibrium studies
carried out at the Geophysical Laboratory,
much effort was expended in achieving
equilibrium conditions. The products of
such reproducible experiments were to be
applied to rocks, which were assumed to
approach closely equilibrium in the earth.
In addition, if the end products of an equi-
librium process were known, the non-equi-
librium paths could be deduced. It is fortu-
nate that rocks retain some of their non-
equilibrium features, thereby revealing the
path to their present state of closure. Some
of the factors that provide evidence of path,
such as heat and mass transport properties
as well as process rates, have been investi-
gated by the staff. For convenience, the
studies are grouped under diffusion, reac-
tion kinetics, crystal growth and dissolu-
tion, metasomatism, and heat transfer.
Diffusion
One of the earliest measurements of
diffusion in silicate liquids was made by
Bowen (1921). He studied the interdiffu-
sion against gravity (to avoid convection)
between a buoyant layer of plagioclase
liquid over a denser diopside liquid at about
1500°C. After holding the liquids for a
period of time, they were quenched, and the
compositions of glass were determined at
various levels by measuring their refrac-
tive indices, which had been previously
calibrated with mixtures of known compo-
sition. From the diffusion profiles an
"average diffusivity" was calculated.
Bowen concluded that the formation of
border phases of large bodies of igneous
rocks by diffusion could not be considered
possible in the time available for a cooling
magma. On the other hand, the formation
of reaction rims about inclusions could be
attributed to diffusion.
A similar experiment was carried out by
Yoder (1973) between liquids of basalt and
rhyolite compositions at 1200°C and
P^O) = 1 kbar. The gradients of the major
elements after quenching were determined
by electron microprobe. He noted the strong
coupling of the fluxes of major compo-
nents and suggested that the structural units
in the liquid were related to subspecies of
184
CARNEGIE INSTITUTION
the minerals.
The self diffusion of 45Ca in diopside at
one atmosphere (McCallister 1973-1975,
1978-1979, Brady 1978-1979, and Mysen,
1979) and up to 30 kbar (Watson, 1979)
were especially informative experiments
in regard to exsolution and homogeniza-
tion processes in pyroxenes. The study of
the coupled diffusion of Mg and Fe in
olivine by Misener (1971-1972, 1972) was
carried out at a series of temperatures for
the purpose of relating diffusion to the
creep rate. The interdiffusion coefficient
was found to be sensitive to composition
and crystallographic orientation. With the
available creep data, Misener was able to
demonstrate that the rate-controlling step
was not the volume diffusion of cations.
Similar coupled diffusion experiments were
done by Boctor and Brady (1979) involv-
ing S in HgSe (tiemannite).
The diffusion of those elements used in
geochronology received considerable at-
tention. For example, the discordant ages
that fit a chord across the concordia curve
in a plot of 207Pb/235U vs 206 Pb/23^ were
attributed by Tilton (1956-1965) to the
continuous volume diffusion of Pb from
the minerals measured. The paradox of
lead loss at the same time on all continents
from rocks of the same age was thereby
resolved. Other studies included the gain
and loss of argon in albite (Laughlin and
Yoder, 1971 ) and the disequilibrium distri-
bution of Th in diopside (Seitz, 1974). The
technique of preparing fission-track maps
developed by Seitz was successful in ob-
taining data on the diffusion of Th, U, and
O in diopside and fluorapatite. In coopera-
tive studies with DTM, the diffusion of Sr
in basalt liquid, studied by Hofmann( 1974),
was especially pertinent to the small scale
heterogeniety of the mantle as well as to
applications to Rb-Sr geochronology of
magmatic rocks.
Reaction Kinetics
The compositions of exsolved pyrox-
enes place limits on their conditions of
formation and give valuable information
on the cooling history of the host rock. In
a laboratory study of a synthetic solid solu-
tion of Di-En, McCallister measured the
rate of exsolution between 1225°C and
1300°C. The observed changes in textures
suggested to him that nucleation and growth
occur above 1300°C, whereas spinodal
decomposition, dependent solely on inter-
diffusion kinetics, is the appropriate mecha-
nism at lower temperatures. Additional
annealing experiments yielded the useful
observations, substantiated by bright-field
micrographs, that rapid exsolution produces
irregular lamellae (spinodal decomposi-
tion), whereas slow exsolution has regular
lamellae that lie within the coherent solvus.
Another major contribution to the cool-
ing rate problem was made by Seifert and
Virgo (1975). By determining the Mg-Fe
distribution with Mossbauer spectra in
anthophyllite at P(H20)=2 kbar, 400°-
720°C, and various times, they were able to
calibrate the order-disorder parameters. In
this way the cooling rates of natural an-
thophyllites could be ascertained. These
experiments were seminal to a large range
of studies on the rates of metamorphic
events.
Other studies included the influence of
Zn and Pb on the transformation rates of
GEOPHYSICAL LABORATORY
185
metacinnabar to cinnabar (Boctor and
McCallister, 1979), the influence of poros-
ity on the reaction rate of periclase and
quartz to form forsterite (Brady, 1979), and
the rate of homogenization of zoned gar-
nets from metamorphic zones (Muncill,
1983-1989, and Chamberlain, 1986). The
latter study is being applied in New Eng-
land in order to obtain the cooling rates
after peak metamorphism.
Crystal Growth and Dissolution
In 1916 Becker and Day demonstrated
by experiment the linear force of growing
crystals. They refuted earlier failures by
showing that the load increased crystal
solubility, so it was necessary to maintain
supersaturation for growth. A crystal-
growth apparatus was designed by Hostetter
(1912-1919) in 1919 in which separate
thermostats held the source fluid at a higher
temperature than the growing crystals.
Whereas the growing of crystals under
equilibrium conditions was the essence of
obtaining phase equilibria diagrams, the
growing of large crystals was only under-
taken in regard to special measurements.
The persistence of staff members in
obtaining phases that are difficult to grow
is the source of many legends. One ex-
ample, is the incredible patience of J. F.
Schairer in determining the melting point
of albite. The liquid of appropriate compo-
sition has to be prepared at a temperature
high enough to dissolve the A1203, but not
too high to volatilize the Na. The liquid is
quenched to a glass, crushed to homoge-
nize, and remelted at successively lower
temperatures. The process of "acclimat-
ing" the liquid structure to that close to
crystalline albite results in a glass that will
produce the appropriate crystalline struc-
ture just below the melting point. Only
after a five-year effort, was Schairer satis-
fied that he knew the exact melting point of
albite ! In the difficult synthesis of pyrope,
Boyd and England (1959) resorted to seed-
ing, and its field of stability at high pres-
sures was accurately delineated. Similarly,
iron cordierite was grown in the system
FeO-Al203-Si02, where previous attempts
failed, merely by crushing a natural iron
cordierite in the same room, thereby pro-
viding inadvertently sufficient dust for
seeds!
A highly sophisticated apparatus has
recently been assembled by Muncill (1988)
for the exact measurement of growth rates
of crystals in melt. He measured the iso-
thermal growth kinetics of plagioclase in a
haplogranodiorite melt at P(H20)=2 kbar.
The growth rate curves were well modeled
by a modification of the theory of Muncill
and Lasaga (1987) fox a simple system, and
the endmember mineral growth curves can
now be used to calculate growth rates in
multicomponent systems. One of the divi-
dends of the study was a videotape that
shows the in situ growth and melting of
plagioclase and other minerals that led to a
visual appreciation of the generation of
textures in igneous rocks.
The dissolution of crystals in melt is
important to the process of assimilation
and provides a test of theories of disequili-
brium melting. Minerals were separated
from a spinel lherzolite and ground into
spheres by Scarfe (1978-1980), E.
Takahashi (1979-1981) and Yoder (1980)
and held in an alkali basalt melt at high
186
CARNEGIE INSTITUTION
pressures and temperatures. By measuring
the change in diameter of the spheres it was
found that the rate of dissolution of Gr >
Cpx > OPx > Sp > 01 at the range of
conditions examined. In general, the rate
of dissolution was inversely proportional
to the enthalpy of melting. The results are
pertinent to the digestion of mantle xeno-
liths by basaltic magmas during their rapid
ascent to the surface. In another series of
experiments by Muncill and Dingwell
(1984-1986) the minerals stable in granitic
melts dissolved much more slowly under
anhydrous conditions than when volatiles
were present.
Metasomatism
The rocks that result from the transport
of various volatile and nonvolatile compo-
nents are usually explained by the diffusion
of components through a static solvent.
For example, mineral textures at metamor-
phic isograds are often accounted for in this
way. A different model, first proposed by
Korzhinskii (1936) involves the transport
of material by the flow of a solvent in
response to gradients of fluid pressure. To
obtain quantitative information on such
infiltration mass transport, J. D. Frantz and
A. Weisbrod (1972-1973) studied the K20-
AljOj-SiOj-HjO-HCl system in 1973 to
identify the sequence of zones of assem-
blages, the nature of the boundaries or
"fronts", and calculate the relative rates of
progression of those fronts. They showed
that infiltration metasomatism was ade-
quate for long distance transport if there
was an adequate fluid pressure gradient,
which may remain unchanged regardless
of the thickness of the reaction zones. These
elegant computations were particularly
instructive in predicting reaction paths and
the relative velocities of the fronts and
hence the relative thickness of the zones.
Frantz and Weisbrod took into account the
important volume term, introducing a po-
rosity factor, and concluded that unless the
rock expands, the infiltration process stops
when the pores are filled. Although the
infiltration theory is usually applied to
metamorphic rocks, Irvine ( 1 980) described
the same process for postcumulus mag-
matic metasomatism to account for the
observed compositional variations in the
cyclical units of the Muskox intrusion.
In contrast to the graphical solution of
Frantz and Weisbrod, a general mathemati-
cal model for mass transfer by both infiltra-
tion and diffusion was considered by Frantz
and Mao (1974) for a multicomponent
system. As a demonstration of their theory,
the system MgO-Si02-H20-HCl was ex-
amined analytically by them using new and
published values of the solubility of the
phases and taking porosity, tortuosity, and
diffusion coefficients into account. The
calculations yielded zone sequences and
thicknesses for various times. Further work
included the more complex system CaO-
MgO-Si02-H20-C02 in which they were
able to calculate the sequence of zones and
the modal abundance of each of the phases
in each zone with the diffusion-infiltration
model.
In an attempt to deal with the multicom-
ponent natural systems, R. C. Fletcher and
R. J. Vidale (now Buden) (1974-1975)
proposed in 1975 a finite-difference model
GEOPHYSICAL LABORATORY
187
for combined diffusion-infiltration metaso-
matism. The results were applicable to a
wide variety of reactions between fluid-
filled cracks and country rock and between
incompatible rock assemblages.
Heat transfer
One of the important modes of heat
transfer, resulting in metamorphism or
magma generation, is from an external
planar source. To obtain quantitative infor-
mation on the rates of heat transfer that
determine the width of metamorphic zones
and the rate of magma production, Yoder
devised an apparatus for measuring, during
a constant heat flux, the heat transfer prop-
erties in advance of melting and during the
partial-melting process for a binary sys-
tem. Detailed temperature profiles were
obtained as a function of time. It was found
that the melting process was decoupled
from the rapid establishment of the thermal
gradient. No evidence was observed for
convection even though the properties of
the system closely approached those of
natural magmas except for linear scale.
The results were well represented by a
theoretical model, deduced by Finger and
Muncill, from which the effective thermal
diffusivities could be calculated. The chal-
lenge of future experiments is to evaluate
quantitatively the combined heat and mass
transfer, but that project awaits a suitable
theory for guidance.
16. Geochronology
The determination of the age of miner-
als by the ratio of lead to uranium and
thorium was a well established principle by
1 928 , but a number of difficulties in analy-
sis remained. Methods for the chemical
analysis of these elements was given in
detail by C. N. Fenner. In the following
year, he and C. S. Piggot enlisted the help
of F. W. Aston (Cambridge), in making the
first calculation of a mineral (thorian uranin-
ite) age on the basis of the specific isotopes
of lead determined by mass spectroscopy.
The discrepancy between the Pb-U age and
Pb-Th age, however, led them to believe
that the assumed U-Th equivalence factor
may be in error. Although the radioactivity
of ocean sediments became a major interest
of C. S. Piggot and of W. D. Urry (1938-
1949), the age determination of minerals
was dominated by others until after WWII.
A program for the determination of the
age of minerals was initiated as a collabo-
rative effort with DTM in 1950. The goal
was to develop techniques and equipment
for determining the age of several common
minerals in the same rock for which differ-
ent methods could yield completely inde-
pendent ages. Granites of Precambrian age
were chosen from which almost a dozen
major and accessory minerals were ex-
tracted. The methods focused on the accu-
rate measure of the naturally occurring
radioactive elements having long half-lives.
The parent and daughter elements were
concentrated from each separated mineral
by ion-exchange resins after spiking the
sample with a known amount of a tracer
isotope and digesting it in acid. In this way
the absolute concentration of the isotopes
in the mineral could be ascertained by
analysis in a mass spectrometer. The min-
eral separation and solution chemistry was
performed by Davis at the Geophysical
188
CARNEGIE INSTITUTION
Laboratory and the mass spectrometry was
carried out mainly by others at DTM
(Aldrich,Wetherill,Tilton). In 1956Tilton
transferred from DTM to the Geophysical
Laboratory, and the close cooperation of
both groups, with a growing tide of Post-
doctoral Fellows and Guest Investigators,
was even more effective.
During the 1950-1955 period efforts
were concentrated on methods. With the
increased sensitivity of the isotope dilution
technique, the decay scheme of 87Rb to 87Sr
was applied successfully to the Li -bearing
micas. The 40K/40Ca and 40K/40Ar clocks
were slowly being developed. The discrep-
ancies between the various clocks were
attributed to radiation damage in the zir-
cons, differential leaching in acids, and the
transfer of Pb between minerals. These
problems were investigated in great detail.
The highest ages appeared to be given by
the Rb-Sr clock, whereas the lowest age
was found by the Th-Pb method. It was
deduced that the Rb half-life used in the
calculation was apparently too low, so a
new half-life was found by assuming the U-
Pb age was correct in a mineral giving
concordant ages from six locations having
a range of ages. Values of the half-life were
calculated from the "Rb/^Sr from Rb-
bearing minerals at the same localities.
The new Rb half-life so determined geol-
ogically, eliminated some of the previous
discordance, and it remains the present-day
value.
With improved techniques, it became
possible to begin applying the results to the
solution of geological problems. In 1957,
after discovering large groups of ancient
rocks, the concept developed that the old-
est rocks were the "nucleus" of a continent
and younger belts of rocks were subse-
quently added on. Because of the different
responses of specific minerals to metamor-
phism, it became possible to identify the
age of the critical events in the geologic
history of a region. In 1958 a major prob-
lem among discordant age values was re-
solved. Wetherill (1956) had shown in a
plot of ^Pb/^U vs. ^Pb/^Uthat the vari-
ous measurements could be related by a
line of "concordia". The line was inter-
preted by Tilton as the locus of apparent
ages resulting from the continuous loss by
diffusion of Pb. It was this discovery that
cleared the way for geological application
on a grand scale. Subsequently, the zones
of various ages mapped for the U. S. and
Canada were confirmed, and the concept of
the slow accretion of a continent began to
take shape.
Discordant ages continued to plague the
analysts, but each time the cause of the
discordancy was resolved greater insight
into geologic processes emerged. Studies
of contact aureoles and eventually regional
metamorphic grades illustrated that tem-
perature affected the diffusion of radioac-
tive daughter elements. The apparent ages
could then be used for mapping thermal
zones in a metamorphosed region.
When the amphiboles and pyroxenes
became useful indicators of age, a new
range of petrological problems could be
tackled. For example, the exceptionally
low concentration of U in dunites and
websterites suggested that the heat flow in
the oceans had to arise from another source
rock; eclogite appeared to be adequate. On
the other hand, the hornblende-bearing
peridotites in some of the islands had enough
K to yield the appropriate heat flow. In
GEOPHYSICAL LABORATORY
189
addition, the isotopic analysis of Pb indi-
cated that the basalts of the oceanic islands
were from heterogeneous sources.
With the arrival of T. Krogh (1966-
1975) the major thrust of the work turned to
the Grenville controversy. The value of
analyzing the whole rock as a closed sys-
tem instead of individual minerals became
appreciated. As belts of ages were identi-
fied, the region appeared to have analogues
with the modern-day, island-arc volcanic
zones. The Grenville front was interpreted
as an ancient plate boundary where a major
metamorphic event took place 1500-1800
m.y. ago with a second major dislocation
about 1000 m.y. ago.
In the next ten years (1968-1978) at
least three major improvements in tech-
nique took place. Krogh invented a new
dissolution method for zircons in which a
teflon-lined pressure vessel was used at
220° C. X-ray fluorescence became a stan-
dard tool for ascertaining the suitability of
samples for Rb-Sr analysis. The produc-
tion and purification (Krogh and Davis,
1975) of the ^Pb spike (with the help of the
Holifield National Laboratory) greatly
improved the precision of the clocks based
on lead. [The use of the 205Pb spike was
apparently developed independently at the
same time by Tera and Wasserburg ( 1 975)] .
With these improvements the ages of zir-
cons inkimberlites were measured by Davis.
The African diamond pipes were found to
be around 90 m.y. old and two groups of
pipes at Yakutsk, U.S.S.R., were 402-443
m.y. and 360-344 m.y. Other studies in-
cluded the dating of many other geologi-
cally significant formations .
With the resignation of Krogh in 1975
and the retirement of Davis in 1978, the
geochronology program was reevaluated.
It was evident that the Geophysical Labo-
ratory had served its role in pioneering new
methods and contributing new concepts to
the solution of geologic problems depend-
ent on knowledge of accurate isotopic
compositions. The existence of more than
50 laboratories in the U. S. devoted to the
dating of rocks and minerals, many headed
up by past associates, indicated the field
was well established, and it was appropri-
ate for the Laboratory to invest its limited
resources in new opportunities.
17. Stable Isotopes
The stable isotopes of the five elements
sulfur, carbon, hydrogen, oxygen, and nitro-
gen (the SCHON, or "beautiful" system)
provide a special set of tools to investigate
both organic and inorganic processes in the
earth. The stable isotope program at the
Geophysical Laboratory evolved after the
arrival of T. C. Hoering (1959 — ). In a very
short time, he built a mass spectrometer,
with the help of colleagues at DTM, and
began applying the C and O isotopes to the
solution of organic problems. (The appara-
tus, with several stages of improvement,
served the staff well for exactly 30 years!)
Carbon
The fractionation of C by algae was
studied by Abelson (1953-1971) and Hoer-
190
CARNEGIE INSTITUTION
ing (1959) and found to be consistent with
the general observation that 13C/12C has a
lower ratio than carbonate or C02 in the
environment. They examined a large array
of separated, individual amino acids with
marked isotopic differences, which they
attributed to the different biosynthetic path-
ways by which its constituents are incorpo-
rated into the algae. Fatty acids are the
precursors of petroleum, which is depleted
in 13C relative to whole modern organisms.
That depletion was correlated with the low
,3C in the fatty acids of living organisms
(both plants and animals) examined by
Parker (1961-1963). In a study of the
organisms of a Texas bay, Parker produced
one of the first investigations of an ecosys-
tem in which isotopes were used as tracers
of complex food webs. Even the reduced
carbon in Precambrian sediments had rela-
tively low ratios of 13C/12C, according to
Hoering 's studies. From then on, isotopes
became a common tool in studying the
processes in living and fossil organic mate-
rial.
Oxygen
Hoering set out to evaluate the effects of
T, P, and X on the isotopic fractionation of
oxygen. The solubility of C02 in H20, for
example, resulted in fractionation. Pres-
sure, on the other hand, was found not to be
an important variable up to 4 kbar. This
systematic study was soon set aside be-
cause of the demand to resolve igneous
penological problems. The lavas of Ice-
land, Snake River Plain, and rocks from the
Island Arcs were examined for evidence of
sediment contamination and their reaction
with meteoric waters, which result in 180
enrichment and depletion, respectively.
These results led to a study of the exchange
of oxygen between silicates and C02 and
02 by Muehlenbachs (1971-1974) and
Kushiro. The value of oxygen isotope
analysis was rapidly recognized by the
metamorphic petrologists, and experiments
were designed to test the control of fluid
composition by the buffering effects of
local mineral assemblages in metamorphic
rocks. Even the exchange of oxygen be-
tween fossils and minerals in metamorphic
rocks were measured (Rumble, Hoering,
and Boucot, 1978). The oxygen isotopes
were then used to test the permeability of
rocks during metamorphism, and eventu-
ally in the mapping of the principal hy-
drothermal pathways in a region. With an
ever growing demand for data on a host of
problems another mass spectrometer de-
voted to both carbon and oxygen was ac-
quired, and the oldest machine was modi-
fied to investigate nitrogen isotopes. A
dramatic new development by Z. Sharp
(1987-1989) in technique for liberating the
oxygen in a mineral on a microscale by
laser heating will bring a new dimension to
the application of oxygen isotopes to geo-
logical problems.
Hydrogen
The second mass spectrometer to be
acquired was a special type dedicated to
measuring hydrogen and deuterium. Be-
ginning in 1977, Estep (now Fogel,
1977 — ) and Hoering studied the fractiona-
tion of hydrogen isotopes in cultures of
microalgae. They discovered several bio-
GEOPHYSICAL LABORATORY
191
logical processes governing the isotope
effect when a cell converts water in the
medium to organic matter. Such informa-
tion was necessary for interpreting hydro-
gen isotopes in the organic matter of sedi-
mentary rocks. For example, marine or-
ganisms produce a larger effect than fresh-
water forms, therefore the source of or-
ganic matter in a sediment may be identi-
fied. Hydrogen isotopes in lipids were
shown to be particularly promising in trac-
ing the sources of plant matter contributed
to sediments and petroleum.
Nitrogen
In many parts of the ocean, nitrogen is a
limiting nutrient and determines the amount
of growth. Variations of 15N/14N can be
used as tracers in the biogeochemical cycle.
Hoering and Ford (1960) studied the iso-
tope fractionation during the fixation of N2
for four species of bacterium cultured in the
laboratory, and they could account for the
depletions in the heavy isotope in natural
populations. After a hiatus of almost twenty
years, the processes of nitrate assimilation
and reduction by blue-green algae attracted
the attention of Macko (1981-1983) and
colleagues. Whereas the fractionation of
isotopes was small in the fixation of mo-
lecular nitrogen, the effect was large by
contrast when the source of nitrogen avail-
able was in the form of a nitrate. In a
collaborative effort of all the members of
the Biogeochemical Group, individual
amino acids were separated from cultured
microorganisms in order to determine the
isotopic effects during biochemical syn-
thesis. Hare and Estep (now Fogel) (1983)
later explored these differences in biosyn-
thesis to trace metabolism of diets and
subsequent diagenesis of modern and fos-
sil animal bones.
In a broad survey of coastal and estuar-
ian sediments, nitrogen isotopes were used
to track the mixing and recycling of organic
matter in the nearshore environments.
Although the processes are indeed com-
plex, Cifuentes (1984-1988) and colleagues
were able to correlate some of the changes
with the processes that tended to consume
nitrogen. These results spurred the search
for more details on the course of nitrogen in
the diets of living organisms. In addition,
the value of using several isotope systems
was recognized as each system contributed
to the definition of the environmental con-
ditions. The multiple use of isotopes was
applied on a grand scale in a study of the
Delaware estuary. The continuing study
has already documented the dramatic sea-
sonal changes that take place in the waters
and sediments of the estuary.
Sulfur
New technique usually opens the door
to new opportunities, and the use of sulfur
isotopes is exemplary. Sabels (1962) and
Hoering in 1963 found that S could be
liberated as SF6 with the halogen fluorides,
and its isotopes measured in the mass spec-
trometer. The precision in the available
mass spectrometer was not sufficient to
warrant a major study on the ore minerals.
In 1983, however, in a new mass spec-
trometer fitted with four detectors, Hoering
was able to measure the four stable isotopes
of S (32S, 33S, 34S, and 36S) simultaneously.
192
CARNEGIE INSTITUTION
The first applications were to metamorphic
rocks in which the alteration of pyrite to
pyrrhotite could be studied The flow of
fluids from a nearby igneous body were
believed to be responsible for the desulfuri-
zation process.
Hoering determined the isotopic com-
position of the sulfur in sedimentary barites
(BaS04) and pyrite from the Archean of
southern India. The values for the barite
contrast sharply with that of contemporary
seawater and appeared to fall in the range
for igneous rocks. The values of the coex-
isting pyrite were even more anomalous.
One of the factors Hoering thought might
be important in explaining the results, prior
to the emergence of the sulfur-reducing
bacteria about 2.8 b.y. ago, was the role of
atmospheric oxygen in the nonbiological
oxidation of reduced sulfur molecules.
Other isotopic systems are now being
applied to resolve this fascinating paradox
in ancient environments.
It is anticipated that any future study
will require the use of all the pertinent
stable isotope systems, whether the prob-
lem be in sedimentary, igneous, or meta-
morphic rocks. The advantage in having
several dedicated mass spectrometers under
one roof at the Geophysical Laboratory
means there is no impediment to examin-
ing the same sample with all the appropri-
ate isotopic systems, and to integrating all
the observations by one or several investi-
gators.
18. BlOGEOCHEMISTRY
The application of organic chemistry to
geological problems at the Geophysical
Laboratory arose out of the research inter-
ests of the newly arrived Director, Philip H.
Abelson (1953-1971). He used paper
chromatography to demonstrate that fos-
sils as old as 300 million years retained
amino acids from some of their original
proteins. Abelson also determined that the
breakdown of amino acids in fossils could
be simulated in the laboratory by substitut-
ing elevated temperatures for geological
time, thereby demonstrating their potential
as stratigraphic markers and geochronom-
etric tools. These "chemical fossils" com-
plemented the classical methods of paleon-
tology. The book "Biochemistry of Amino
Acids," edited by P. E. Hare (1963— ), T. C.
Hoering, and F. King, Jr. (1970-1974), has
been the definitive work on the subject.
All the amino acids in proteins exist in
two configuration that are mirror images,
or optical isomers, designated D and L.
Hare was successful in separating these
isomers with gas chromatography andhigh-
pressure liquid chromatography. He learned
that biologically produced amino acids,
which are dominantly L, transformed spon-
taneously to D abiologically as a function
of time. Hare developed these observa-
tions into a method for dating fossils as old
as 20 million years.
From the early paper chromatography,
the techniques have evolved to very-high-
resolution capillary gas chromatography,
high-pressure liquid chromatography and
eventually to a combination of gas-chro-
matographic and mass spectrometric meth-
ods. A field-portable, liquid chromato-
graph was made by Hare to measure amino
stratigraphic sections on site. The fluores-
cent derivatives of NH2 groups were used
GEOPHYSICAL LABORATORY
193
by Hare to analyze for amino acids, pep-
tides, and proteins at an unprecedented low
concentration. With this technique he set a
limit of less than one part per trillion of
amino acids in the returned lunar soil. That
method was not only useful in the analysis
of fluid inclusions in igneous and meta-
morphic rocks, but also in the characteriza-
tion of blood.
The study of amino acids in fossils was
followed by investigations of the fatty ac-
ids, fatty alcohols, humic acids, prophy-
rins, kerogen and steranes by Hoering. Of
special interest was the demonstration in
the laboratory of the reaction of glucose
and amino acids to produce melanoidin, a
product closely related to humic acid, a
significant fraction of the organic material
in Holocene sediments. The work outlined
the pathway whereby organic matter is
effectively removed from the biological
carbon cycle and preserved without further
metabolization by micro-organisms.
The organic compounds in mildly meta-
morphosed Precambrian rocks were stud-
ied by Hoering. He discovered with the use
of 13C/12C that the extractable organic
compounds could be attributed to biologi-
cal origin. The insoluble fraction, kerogen,
yielded similar carbon isotope ratios, but
the relationship of soluble to the insoluble
fractions remained obscure because of
potential contamination by modem organ-
isms. Subsequent laboratory experiments
on the thermal breakdown of kerogen pro-
vided a mechanism for the production of
the high-molecular weight components
found in petroleum (Hoering, 1984). The
hydropyrolysis of shales and other rocks
resulted in a method, now standard in the
petroleum industry, for evaluating their
potential for petroleum generation.
As a result of Hoering 's skills in mass
spectrometry, the Laboratory has devel-
oped dedicated facilities for measuring the
stable isotopes of C, H, O, N, and S. For
example M. Fogel traced the food chains
with hydrogen isotopes. The specific algae
on which a snail had been feeding could be
identified in confirmation of the adage "you
are what you eat." In another important
study, Fogel showed with stable isotopes
that modem blue-green algae and bacteria
growing in C02-rich hot springs exhibited
the same depletion in 13C as in Precambrian
stromatolites formed by the same types of
organisms. She demonstrated that the
atmosphere in Precambrian times was,
therefore, probably enriched in C02 by
several percent relative to the present day
atmosphere.
In an unprecedented collaboration be-
tween the Geophysical Laboratory and the
Carnegie Institution of Washington's De-
partment of Plant Biology at Stanford, Fogel
and Joseph Berry solved one of stable-
isotope geochemistry's oldest problems,
the "Dole Effect", that was identified in
1936. Atmospheric oxygen is anomalously
enriched in the heavy isotope, 180, and
previous studies could not account for this
effect. Berry and Fogel discovered that a
large isotope fractionation occurred during
the uptake of 02 in photorespiration, a
process that accompanies the photosynthe-
sis reaction, and thus they could account
for the isotope enrichment.
The presence of a group in biogeochem-
istry in a geophysical laboratory has yielded
many unanticipated dividends. The stable
194
CARNEGIE INSTITUTION
isotope facilities initially developed by them
have been applied to a wide range of prob-
lems in sedimentary, igneous, and meta-
morphic petrology. Their techniques and
cooperation have contributed to the under-
standing of ore deposits, meteorites, and an
unusual array of mineralogical problems.
They are indeed exemplary of the outstand-
ing results that can be achieved through the
Carnegie concept of supporting scholars in
fields of their own choice.
19. War-time Studies
World War I
Before hostilities began in 1914, it
became evident that the U.S. would be cut
off from the European sources of optical
glass. Five American companies under-
took to make optical glass but the quality
was not satisfactory by the time the U.S.
entered the war. Because of the critical
need for high quality optical glass for mili-
tary fire-control instruments, methods for
its manufacture on a large scale had to be
developed. The Council of National De-
fense appealed to the Geophysical Labora-
tory for help because it had been engaged
for many years in the study of silicate
liquids, similar to optical glass, at very high
temperatures. It was the only organization
in the country with a staff trained in the
fundamentals necessary for the manufac-
ture of optical glass.
In April 1917 groups of staff members
were sent to the various plants and assigned
the responsibility for their operations,
whereas others remained at the Laboratory
to deal with specific problems. The coop-
erative attitude of the companies and the
direct liaison with the Army through the
commissioning of F. E. Wright greatly
facilitated the task. The expenses incurred
were covered by CIW and no compensa-
tion was ever received for their work. The
Director, A. L. Day, was eventually desig-
nated as "in charge of optical glass produc-
tion, War Industries Board." After the
armistice, the records show that 95% of all
optical glass manufactured in the U.S.
during the war had been made under the
supervision of the staff of the Laboratory.
The manufacturing problems were
eventually resolved by putting the secre-
tive cook-book glass making methods on a
scientific base. Formulae were devised so
that glasses of the appropriate index of
refraction or other optical constants could
be prepared from the necessary constitu-
ents with a minimum of trial and error.
Even the barium-rich glasses for aerial
camera lenses were made on short notice.
Most important contributions were made
by Adams and Williamson (1919), who
deduced the laws for relieving stress in
glass by annealing, and Roberts (1917-
1947, 1919), who by direct experiment was
able to formulate cooling schedules for the
glass pots. Sosman (1925) had investi-
gated some of the principles governing the
corrosion of the fire clays used in the pots
by the molten glass. Other problems such
as high dispersion due to successive iron
content, stones from the digestion of un-
suitable clay pots, cords and striations aris-
ing from poor stirring, and strain from
rolling were all investigated. Over 20 papers
were published by the staff on glass making
for the benefit of the future U.S. glass
GEOPHYSICAL LABORATORY
195
industry. Of these, the monograph on the
properties of glass by G. W. Morey (1938)
revised in 1954, remains a standard refer-
ence work. Although none of the 20 scien-
tifically-trained staff had previous experi-
ence with the manufacture of glass, all used
their basic knowledge of silicates to put this
new U.S. industry on a sound basis.
The skills of the chemists at the Labora-
tory were also put to use on the fixation of
nitrogen for the manufacture of explosives.
Experimental work on the Bucher-cyanide
process and the Haber-process were begun
in the summer of 1918, and, therefore, had
not proceeded far before the end of the war.
As repugnant as the task may have been,
the Laboratory also investigated some of
the physical constants of mustard "gas" in
response to a military request.
World War II
The president of CIW, Vannevar Bush,
helped establish the National Defense
Research Committee in 1940 and served as
its chairman. On the 28 June 1941 Bush
became Director of the Office of Scientific
Research and Development (OSRD) that
organized and directed most of the research
efforts during the war. In that summer a
comprehensive program of defense research
was organized to be centered at the Geo-
physical Laboratory. Some of the staff
began to collect information from military
and other sources for delineating the lines
of research. After the declaration of war in
December, the entire staff, supplemented
by thirty temporary employees, and all of
the resources of the Laboratory were de-
voted to the tasks ahead.
The Director, L. H. Adams, was ap-
pointed chairman of the committee for
investigating the erosion of gun barrels due
to high-pressure, hot, propellant gases re-
leased on firing. As the research pro-
ceeded, studies were concentrated on the
caliber-50, rapid-fire, aircraft gun where
means to counteract severe swaging of the
lands and thermal expansion of the barrel
became the principal focus. The system-
atic studies included analysis of the corro-
sion products of the steel barrels and the
propellant gases. By means of isotopically
labeled nitrogen in the explosive charge
and use of the National Bureau of Stan-
dards mass spectrometer, tracer studies
revealed the depth of penetration of the
gases. Experiments were carried out in
high-pressure vessels on controlled explo-
sions to ascertain the internal ballistics and
chemical products. Metal with high, hot-
hardness, as well as resistance to gas ero-
sion were inserted as short liners in barrels
at the origin of rifling. The liners were then
tested on a firing range on the Potomac
River or in firing ranges installed under the
tennis courts (now volley ball court) be-
hind the Laboratory. The superior metal
was found to be the cobalt-based alloy,
stellite, and it became a most useful mate-
rial for making hydrothermal pressure
vessels after the war.
Another group was concerned with the
electroplating of chromium inside the bar-
rel after the liner, in cooperation with the
electroplating group at the National Bu-
reau of Standards. Because it was not
practical to machine a taper in the large
number of barrels required, a method was
designed to taper the plating, with increas-
ing thickness of the plate toward the muzzle.
196
CARNEGIE INSTITUTION
In this way, constriction of the bore com-
pensated for the thermal expansion of the
barrel during firing. The increased life and
accuracy was documented by test firing on
the Geophysical Laboratory ranges. The
barrel adapted for military use, still being
manufactured today, contains a short stel-
lite liner and a chrome-plated, tapered
bore.
Several of the staff members were also
helpful at DTM in the development of the
proximity fuze for artillery shells. That
device was considered to be "the most
important technical improvement in weap-
onry to come out of World War H."
Almost five years of the life of the
Geophysical Laboratory were devoted to
the war work. The regular staff was paid by
CIW; however, the costs of the temporary
employees and extra expenses were pro-
vided by the government. In 1946 the war
work was phased out, the reports written,
and a comprehensive review undertaken of
the scientific programs in the light of the
irreversible changes brought about by World
Warn.
20. Closing Remarks
The most important factor in the gen-
eration of new ideas at the Geophysical
Laboratory has been the freedom of choice
to follow whatever the staff member be-
lieves to be important in the solution of a
geological problem. The scientist's over-
riding goal was to achieve an understand-
ing of the problem so that the critical vari-
ables could be recognized, evaluated, and
formulated into general concepts useful in
solving other problems. The intent, there-
fore, was to seek knowledge that has broad
application to the major problems of the
earth. Because no researcher can predict
how a fundamental discovery might be
applied to future societal needs or prob-
lems, there is no test for relevance or appli-
cability applied to the work at the Geo-
physical Laboratory as is made in indus-
trial organizations. That freedom to follow
whatever is critical to the solution of prob-
lems is why the Geophysical Laboratory
has remained unique among research or-
ganizations. The price of such a generous
measure of scientific freedom is greater
personal responsibility to produce and
greater accountability. Although peer
review provides for continual testing, the
responsibility to produce was self gener-
ated and was expressed by the high motiva-
tion and involvement of the staff.
Addendum
Most histories record the biased views
of a single observer moderated by the writ-
ten records and evaluations of others. It is
difficult to subdue the enthusiasm, admira-
tion and pride the author has for the Geo-
physical Laboratory and its past and pres-
ent staff members. Having known almost
all of the early staff members and experi-
enced directly slightly more than half of the
life of the Laboratory, the author might be
forgiven for any excessive claims of dis-
covery attributed to the staff. All science is
built on the discoveries of others, and it is
not always evident who arrives at the pin-
nacle of an idea first, demonstrates its proof,
applies the solution to a geological prob-
lem, or capitalizes on its promotion. The
GEOPHYSICAL LABORATORY
197
personal satisfaction of contributing to the
growth of science is adequate reward in
itself.
The scientific history presented above
was prepared on short notice, with a mini-
mum of time for reflection, preparatory to
the departure of the staff from the Geo-
physical Laboratory building on 280 1 Upton
St., N. W., Washington, D. C. to a new
building on the DTM campus. It is antici-
pated that a more detailed, documented
history will be prepared in the future.
GEOPHYSICAL LABORATORY
199
Publications
Reprints of the numbered publications listed below are available, except where noted, at no
charge from the Librarian, Geophysical Laboratory, 2801 Upton St., N.W, Washington, D.C.
20008-3898, U.S.A. Please give reprint number(s) when ordering.
Angel, R. J., High-pressure structure of anorthite,
Am. Mineral., 73, 1 1 14-1 1 19, 1988 (G.L. Paper
2089).
Angel, R. J., and L. W. Finger, Polymorphism of
nickel sulfate hexahydrate, Acta Crystallogr .,
Sect.C,44, 1869-1873, 1988 (G.L. Paper 2094).
Angel, R. J., T. Gasparik, and L. W. Finger,
Crystal structure of a Cr^-bearing pyroxene,
Am. Mineral, 74, 599-603, 1989 (G.L. Paper
2123).
Angel, R. J., S. A. T. Redfern, and N. L. Ross,
Spontaneous strain below the /I -PA transition in
anorthite at pressure, Phys. Chem. Minerals, 16,
539-544, 1989 (G. L. Paper 2122).
Angel, R. J., L. W. Finger, R. M. Hazen, M.
Kanzaki, D. J. Weidner, R. C. Liebermann, and
D. R. Veblen, Structure and twinning of single-
crystal MgSi03 garnet synthesized at 17 GPa
and 1800°C, Am. Mineral., 74, 509-512, 1989
(G.L. Paper 2120).
Arashi, H., O. Shimomura, T.Yagi, S. Akimoto,
and Y. Kudoh, P-T phase diagram of Zr02 deter-
mined by in situ X-ray diffraction measurements
at high pressures and high temperatures, in
Advances in Ceramics, Vol. 24: Science and
Technology ofZirconia HI, The American Ce-
ramic Society, Inc., Westerville, Ohio, pp. 493-
500, 1988 (No reprints available from Geo-
physical Laboratory).
Boctor, N. Z., and G. Kullerud, Phase relations in
the mercury-selenium sulfur system at 200° to
700°C, J. Solid State Chem., in press.
Boyd, F. R., Where do we go from here?, in
Kimberlite and Related Rocks, Proceedings of
the Fourth International Kimberlite Conference,
Perth, Australia, August, 1986, J. Ross, ed.,
Geological Society of Australia, Special Publi-
cation No. 14, Vol. 2, Blackwell Scientific Pubns,
Carleton, Victoria, Australia, pp. 1239-1251,
1989 (G.L. Paper 2142; no reprints available for
distribution).
Boyd, F. R., Compositional distinction between
oceanic and cratonic lithosphere, Earth Planet.
Sci. Lett., in press.
Chamberlain, C. P., P. H. Zeitler, and M. Q. Jan,
The dynamics of the suture between the Kohis-
tan Island arc and the Indian plate in the Hima-
laya of Pakistan, /. Metamorphic Geol., 7, 135-
149, 1989.
Chamberlain, C. P., and M. Q. Jan, Petrologic
constraints on the tectonic development of the
Nanga Parbut - Haramosh Massif, Himalayas,
Spec. Pap. - Geol. Soc. Amer., in press.
Chamberlain, C. P., and D. Rumble, Thermal
anomalies in a regional metamorphic terrane:
An isotopic study of the role of fluids, /. Petrol.,
29, 1215-1232, 1988 (G.L. Paper 2129).
Chamberlain, C. P., and D. Rumble, III, The
influence of fluids on the thermal history of a
metamorphic terrane, New Hampshire, USA. /.
Geol. Soc. London (Spec. Issue), in press.
Chayes, F., The Delesse relation in a concentri-
cally zoned sphere. I. The section-number bias,
Math. Geol, 21, 319-329, 1989 (G.L. Paper
2134).
Cifuentes, L. A., J. H. Sharp, and M. L. Fogel,
Stable carbon and nitrogen isotope
biogeochemistry in the Delaware Estuary,
Limnol. and Oceanogr., 33, 1102-1115, 1988
(G.L. Paper 2095).
Day, H. W., and C. P. Chamberlain, Implications
of thermal and baric structure for controls on
metamorphism in northern New England, J.
Geol. Soc. London, in press.
Dingwell, D. B., The structures and properties of
fluorine-rich magmas: a review of experimental
studies, in Recent Advances in the Geology of
Granite-Related Mineral Deposits, Proceedings
of the CIM Conference on Granite-Related
Mineral Deposits, Halifax, Canada, September,
1985, R. P. Taylor and D. F. Strong, eds., CIM
Bull Spec Vol. 39, pp. 1-12, The Canadian
200
CARNEGIE INSTITUTION
Institute of Mining and Metallurgy, Montreal
1988 (G.L. Paper 2098).
Dymek, R. R, S. C. Brothers, and C. M. Schiffries,
Pedogenesis of ultramafic metamorphic rocks
from the 3800 Ma Isua Supracrustal Belt, West
Greenland, /. Petrol, 29, 1353-1397, 1988.
Finger, L. W., R. M. Hazen, and R. J. Hemley,
BaCuSi206. A new cyclosilicate with four-
membered tetrahedral rings, Am. Mineral., 74,
952-955, 1989 (G.L. Paper 2131).
Fogel, M. L., E. K. Sprague, A. P. Gize, and R. W.
Frey, Diagenesis of organic matter in Georgia
salt marshes, Estuarine, Coastal Shelf Science,
28, 211-230, 1989 (G.L. Paper 2115).
Frantz, J. D., Y. G. Zhang, D. D. Hickmott, and T.
C. Hoering, Hydrothermal reactions involving
equilibrium between minerals and mixed vola-
tiles. 1. Techniques for experimentally loading
and analyzing gases and their application to
synthetic fluid inclusions, Chem. GeoL, in press.
Guy, R. D., J. A. Berry, M. L. Fogel, and T. C.
Hoering, Differential fractionation of oxygen
isotopes by cyanide-resistant and cyanide-sen-
sitive respiration in plants, Planta, 177, 483-
491, 1989 (G.L. Paper 2138).
Hare, P. E., Chiral mobile phases for the enanti-
omeric resolution of amino acids, in Chromato-
graphic Chiral Separations, L. J. Crane and M.
Zief, eds., Marcel Dekker, New York, pp. 165-
177, 1988 (G.L. Paper 2105).
Hare, P. E., and P. A. St. John, Detection limits for
amino acids in environmental samples, in De-
tection in Analytical Chemistry: Importance,
Theory, and Practice, L. A. Curie, ed., ACS
Symposium Series 361, American Chemical
Society, Washington, D.C., Chapt. 15., pp. 275-
285, 1988 (G.L. Paper 2106; no reprints avail-
able for distribution).
Hare, P. E., Detection limits in amino acid analy-
sis: An overview, in Methods in Protein Se-
quence Analysis, (Proceedings of the 7th Inter-
national Conference, Berlin, July 3-8, 1988), B.
Wittman-Liebold, ed., Springer- Verlag, New
York, Chapt. 1.1, pp. 2-9, 1989 (G.L. Paper
2141; no reprints available for distribution).
Hazen, R. M., A useful fiction: polyhedral model-
ing of mineral properties, Am. J. Sci., Special
("Wones") Volume, 288-A, 242-269, 1988 (G.L.
Paper 2059).
Hazen, R. M., Understanding perovskites of bene-
fit to science and industry - an interdisciplinary
approach, Earth in Space, 1, No. 3,. 8-10, 1988
(G.L. Paper 21 12; no reprints available for dis-
tribution).
Hazen, R. M., and Z. D. Sharp, Compressibility of
sodalite and scapolite, Am. Mineral., 73, 1120-
1 122, 1988 (G.L. Paper 2088).
Hazen, R. M., and L. W. Finger, High-pressure
crystal chemistry of andradite and pyrope:
Revised procedures for high-pressure diffrac-
tion experiments, Am. Mineral., 74, 352-359,
1989 (G.L. Paper 2114).
Hazen, R. M., L. W. Finger, and D. E. Morris,
Crystal structure of DyBa2Cu4Og: A new 77 K
bulk superconductor, Appl. Phys. Lett., 4, 1057-
1059, 1989 (G.L. Paper 2113).
Hazen, R. M., The Breakthrough: The Race for
the Superconductor, Summit Books, New York,
1988; Ballantine/Science, New York, 1989.
Foreign editions: Superconductors: The Break-
through, Unwin Hyman Ltd., London, 1988; La
Course Aux Supraconducteurs, Librairie Plon,
Paris, 1989; De Dag Dat de Wetenschap Wild
Werd, Uitgeverij Lannoo, Tielt, The Nether-
lands, 1989 (G.L.Paper 2073; obtainable by
purchase only from the publishers).
Hemley, R. J., R. E. Cohen, A. Yeganeh-Haeri, H.
K. Mao, D. J. Weidner, and E. Ito, Raman
spectroscopy and lattice dynamics of MgSi03-
perovskite at high pressure, in Perovskite: A
Structure of Great Interest to Geophysics and
Materials Science, A. Navrotsky and D. J.
Weidner, eds., American Geophysical Union,
Washington, D. C, pp. 35-53, 1989 (G.L. Paper
2111).
Hemley, R. J., A. P. Jephcoat, C. S. Zha, H. K.
Mao, L. W. Finger, and D. E. Cox, Equation of
state of solid neon from X-ray diffraction meas-
urements to 1 10 GPa, in International AIR APT
Conference,XIth,Kiev, USSR, July 12-17, 1987,
Vol. 3. High Pressure Science and Technology:
Proceedings, N. V. Novikov and Ye M.
Chistyakov, eds., Naukova Dumka, Kiev, pp.
211-217, 1989 (G.L. Paper 2135; no reprints
available for distribution).
Hemley, R. J., L. C. Chen, and H. K. Mao, New
transformations between crystalline and amor-
phous ice, Nature, 338, 638-640, 1989 (G.L.
Paper 2124).
GEOPHYSICAL LABORATORY
201
Hemley, R. J., C. S. Zha, A. P. Jephcoat, H. K.
Mao, L. W. Finger, and D. E. Cox, X-ray diffrac-
tion and equation of state of solid neon to 1 10
GPa,Phys.Rev.B,39, 11820-11827, 1989(G.L.
Paper 21 18).
Hickmott, D. D., and N. Shimizu, Trace element
zoning in garnets from the Kwoiek area, British
Columbia: Possible influence of interface kinet-
ics in metamorphism, Contrib. Mineral. Petrol.,
in press (No reprints will be available from
Geophysical Laboratory).
Hoering, T. C, Isomers of the monomethyl, acy-
clic hydrocarbons in the Messel shale and in
petroleums, Cour. Forsch.Senkenberg, 107 ', 79-
87, 1988 (G.L. Paper 2116).
Hofmeister, A. M, J. Xu, H. K. Mao, P. M. Bell,
and T. C. Hoering, Thermodynamics of Fe-Mg
olivines at mantle pressures: Mid- and far-
infrared spectroscopy at high pressure, Am.
Mineral., 74, 281-306, 1989 (G.L. Paper 2097).
Irvine, T. N., A global convection framework:
Evidence for symmetry and stratification in the
Earth's convection system, Econ. Geol., in press.
Kubicki, J. D., and A. C. Lasaga, Molecular dy-
namics of SiOz melt and glass: Ionic and cova-
lent models, Am. Mineral., 73, 941-955, 1988
(No reprints available from Geophysical Labo-
ratory).
Kudoh, Y., E. Ito, and H. Takeda, High-pressure
structural study on perovskite-type MgSi03 - A
summary, in Perovskite: A Structure of Great
Interest to Geophysics and Materials Science, A
Navrotsky andD. J. Weidner, eds., Geophysical
Monograph 45, American Geophysicial Union,
Washington, D. C.,pp. 33-34, 1989 (No reprints
available from Geophysical Laboratory).
Kushiro, I., Density of basalt magmas at high
pressures and its petrological application, in
Advances in Physical Geochemistry, "Physical
Chemistry of Magma", L. L. Perchuk and I.
Kushiro, eds., Springer- Verlag, New York, in
press.
Kushiro, I., and B. O. My sen, Experimental stud-
ies of the system Mg^SiO^F^at pressures 10"2-
10"10 bar and temperatures to 1650'C: Applica-
tion to condensation and vaporization processes
in the primitive solar nebula, in Advances in
Physical Geochemistry, L. L. Perchuk, ed.,
Springer- Verlag, New York, in press.
Luth, R. W., Natural versus experimental control
of oxidation state: Effects on the composition
and speciation of C-O-H fluids, Am. Mineral.,
74, 50-57, 1989 (G.L. Paper 2110).
Luth, R. W., and G. E. Muncill, Fluorine in alumi-
nosilicate systems: Phase relations in the system
NaAlSi308-CaAl2Si2Og-F20 A, Geochim. Cosmo-
chim. Acta, in press (G.L. Paper 2136).
Luth, R. W., D. Virgo, F. R. Boyd, and B. J. Wood,
Ferric iron in mantle-derived garnets: Implica-
tions for thermobarometry and for the oxidation
state of the mantle, Contrib. Mineral. Petrol., in
press.
Mao, H. K., Static compression of simple molecu-
lar system in the megabar range, in Simple
Molecular Systems at Very High Density, Vol.
186, Proceedings of a NATO Advance Research
Workshop/European Society Workshop, March
28-April 6, 1988, Les Houches, France, A. Po-
lian, P. Loubeyre, and N. Boccara, eds., Plenum
Publ. Corp., New York, pp. 221-236, 1989 (G.L.
Paper 2100).
Mao, H. K., R. J. Hemley, Y. Wu, A. P. Jephcoat,
L. W. Finger, C. S. Zha, and W. A. Bassett,
High-pressure phase diagram and equation of
state of solid helium from single crystal X-ray
diffraction to 23.3 GPa, Phys. Rev. Lett., 60,
2649-2652, 1988 (G.L. Paper 2083).
Mao, H. K., and R. J. Hemley, Optical studies of
hydrogen above 200 gigapascals: Evidence for
metallization by band overlap, Science, 244,
1462-1465, 1989 (G.L. Paper 2130).
Mao, H. K., L. C. Chen, R. J. Hemley, A. P.
Jephcoat, Y. Wu, and W. A. Bassett, Stability
and equation of state of CaSi03-perovskite to
134 GPa, J. Geophys. Res., in press.
McMillan, P., and N. Ross, The Raman spectra of
several orthorhombic calcium oxide perovskites,
Phys. Chem. Minerals, 16, 21-28, 1988 (No
reprints available from Geophysical Labora-
tory).
Morris, D. E., J. H. Nickel, J. Y. T. Wei, N. G.
Asmar, J. S. Scott, U. M. Scheven, C. T. Hultgren,
A. G. Markelz, J. E. Post, P. J. Heaney, D. R.
Veblen, and R. M. Hazen, Eight new high-
temperature superconductors with the 1:2:4
structure, Phys. Rev. B, 39, 7347-7350, 1989
(G.L. Paper 2127).
Muncill, G. E., and A. C. Lasaga, Crystal growth
kinetics of plagioclase in igneous systems:
Isothermal H20-saturated experiments and
202
CARNEGIE INSTITUTION
extension of a growth model to complex sili-
cate melts, Am. Mineral., 73, 982-992, 1988
(G.L. Paper 2087).
Mysen, B. O., and D. Virgo, Redox equilibria,
structure, and properties of Fe-bearing alumi-
nosilicate melts: Relationships among tempera-
ture, composition, and oxygen fugacity in the
system Na20-Al203-Si02-Fe-0, Am. Mineral,
74, 58-76, 1989 (G.L. Paper 2108).
Mysen, B. O., Relations between structure, redox
equilibria of iron, and properties of magmatic
liquids, in Advances in Physical Geochemistry,
L. L. Perchuk and I. Kushiro, eds., Springer-
Verlag, New York, in press.
Mysen, B. O., Volatiles in magmatic liquids, in
Progress in Physico-Chemical Petrology (D. S.
Korzhinskii Memorial Volume), L. L. Perchuk,
ed., Cambridge University Press, New York, in
press.
Mysen, B. O., Distribution of aluminum between
structural units in peralkaline aluminosilicate
melts in the systems Li20-Al203-Si02, Na20-
Al203-Si02 and K20-Al203-Si02, Am.
Mineral., In press.
Powell, E. N., A. Logan, R. J. Stanton, Jr., D. J.
Davies, and P. E. Hare, Estimating time-since-
death from the free amino acid content of the
mollusc shell: A measure of time averaging in
modern death assemblages? Description of the
technique, Palaios, 4, 16-31, 1989 (G.L. Paper
2140; no reprints available for distribution).
Prewitt, C. T., Annual Report of the Director of the
Geophysical Laboratory, Carnegie Instn. Wash-
ington, 1987-1988, Geophysical Laboratory,
Washington, D.C., 1988 (G.L. Paper 2102).
Richet, P., J. A. Xu, and H. K. Mao, Quasi-
hydrostatic compression of ruby to 500 Kbar,
Phys. Chem. Minerals , 16, 207-21 1, 1988 (G.L.
Paper 2071).
Richet, P., H. K. Mao, and P. M. Bell, Static
compression and equation of state of CaO to
1.35 Mbar, J. Geophys. Res., 93, B12, 15279-
15288, 1988 (G.L. Paper 2099).
Richet, P., H. K. Mao, and P. M. Bell, Bulk moduli
of magnesiowustites from static compression
measurements, /. Geophys. Res., 94, B3, 3037-
3045, 1989 (G.L. Paper 2109).
Ross, N. L., and A. Navrotsky, Study of the
MgGe03 polymorphs (orthopyroxene, clinopy-
roxene, and ilmenite structures) by calorimetry,
spectroscopy and phase equilibria, Am. Min-
eral., 73, 1355-1365, 1988 (No reprints avail-
able from Geophysical Laboratory).
Ross, N. L., and R. M. Hazen, Single crystal X-ray
diffraction study of MgSi03 perovskite from 77
to 400 K, Phys. Chem. Minerals, 16, 415-420,
1989 (G.L. Paper 2119).
Ross, N. L., J. Ko, and C. T. Prewitt, A new phase
transition in MnTi03: LiNb03 perovskite
structure, Phys. Chem. Minerals, in press (G.L.
Paper 2137).
Rumble, D., HI, and C. P. Chamberlain, Graphite
vein deposits of New Hampshire, in New Eng-
land Intercollegiate Geological Conference
Guidebook, 80th Annual Meeting, W. A. Both-
ner, ed., University of New Hampshire, Dur-
ham, NH, pp. 241-255, 1988 (G.L. Paper 2104;
no reprints available for distribution).
Rumble, D., in, C. P. Chamberlain, D. K. Zeitler,
and B. Barriero, Hydrothermal graphite veins
and Acadian granulite facies metamorphism,
New Hampshire, USA, in Fluid Movements,
Element Transport, and the Composition of the
Crust, D. Bridgwater, ed., Kluwer Academic
Publ., Dordrecht, in press (G.L. Paper 2128).
Schiffries, C. M., Liquid- absent fluid inclusions
and phase equilibria in the system CaCl2 - NaCl
- H20, Geochim. Cosmochim. Acta, in press,
1989.
Schiffries, C. M., and D. M. Rye, Stable isotope
systematics of the Bushveld Complex: I. Con-
straints in magmatic processes in layered intru-
sions, Am. /. Sci., in press, 1989.
Schiffries, C. M., amd D. M. Rye, Stable isotope
systematics of the Bushveld Complex: II. Con-
straints on hydrothermal processes in layered
intrusions, Am. J. Sci., in press, 1989.
Sharp, Z. D., G. R. Helffrich, S. R. Bohlen, andE.
J. Essene, The stability of sodalite in the system
NaAlSi04-NaCl, Geochim. Cosmochim. Acta,
in press (G.L. Paper 2133).
Sheng, Z. Z., A. M. Hermann, D. C. Vier, S.
Schultz, S. B. Oseroff, D. J. George, and R. M.
Hazen, Superconductivity in the Tl-Sr-Ca-Cu-
O system, Phys. Rev. B, 38, 7074-7076, 1988
(G.L. Paper 2103)
Spear, F. S., D. D. Hickmott, and J. Selverstone,
The metamorphic consequences of thrust em-
placement, Fall Mountain, New Hampshire,
Geol.Soc. Am. Bull., in press (No reprints will be
GEOPHYSICAL LABORATORY
203
available from Geophysical Laboratory).
Stafford, T. W., Jr., and R. A. Tyson, Accelerator
radiocarbon dates on charcoal, shell, and human
bone from the Del Mar site, California, Am.
Antiq., 54, 389-395, 1989 (G.L. Paper 2125).
Stafford, T. W., Jr., Extraction of organic fractions
from fossil bones for radiocarbon dating and
stable isotope analysis, J. Archaeol. Sci. , in
press.
Stathoplos, Linda, and P. E. Hare, Amino acids in
planktonic foraminifera: Are they phylogeneti-
cally useful? in Origin, Evolution, and Modern
Aspects of Biomineralization in Plants and
Animals, Proceedings of the Fifth International
Symposium on Biomineralization, R. E. Crick,
ed., Plenum Publ. Co., New York, in press.
Ulmer, P., The dependence of the Fe2+-Mg cation-
partitioning between olivine and basaltic liquid
on pressure, temperature, and composition: An
experimental study to 30 kbars, Contrib. Min-
eral. Petrol, 101, 261-273, 1989.(G.L. Paper
2139).
Velinsky, D. J., J. R. Pennock, J. H. Sharp, L. A.
Cifuentes, and M. L. Fogel, Determination of
the isotopic composition of ammonium-nitro-
gen at the natural abundance level from estuar-
ine waters, Marine Chemistry, in press (G.L.
Paper 2126).
Wood, B. J., and D. Virgo, Upper mantle oxida-
tion state: Ferric iron contents of lherzolite spi-
nels by 57Fe Mossbauer spectroscopy and resul-
tant oxygen fugacities, Geochim. Cosmochim.
Acta, 53, 1277-1291, 1989 (G.L. Paper 2121).
Yoder, H. S., Jr., The great basaltic "floods,"
South African J. Geol., 91 (Alex. L. du Toit
Memorial Lectures, No. 20), 139-156, 1988
(G.L. Paper 2085).
Zhang, Y. G., and J. D. Frantz, Experimental
determination of the compositional limits of
immiscibility in the system CaClj-Hj-O-COj at
high temperatures and pressures using synthetic
fluid inclusions, Chem. Geol, 74, 289-308, 1989
(G.L. Paper 2107).
204
CARNEGIE INSTITUTION
Personnel
July 7, 1988 to June 30, 1989
Research Staff
Research Associates
Charles T. Prewitt, Director
Peter M. Bell1
Francis R. Boyd, Jr.
Larry W. Finger
Marilyn L. Fogel
John D. Frantz
P. Edgar Hare
Robert M. Hazen
Russell J. Hemley
Thomas C. Hoering
T. Neil Irvine
Ho-Kwang Mao
Bjorn O. My sen
Douglas Rumble III
David Virgo
Hatten S. Yoder, Jr.
David Velinsky
Nick Oliver5
Postdoctoral Fellows
Ross Angel6
Luis Cifuentes7
Donald Hickmott
Andrew P. Jephcoat8
Yasuhiro Kudoh9
Robert W.Luth10
Nancy Ross11
Craig Schiffries
Zachary Sharp12
Peter Ulmer13
Yi-gang Zhang14
Keck Earth Sciences Research Scholar
i'112
Gregory E. Muncill
Postdoctoral Associates
Liang-chen Chen3
Ming Sheng Peng4
Jinfu Shu
Ellen K. Wright
Predoctoral Fellows
Constance Bertka
Yingwei Fei
Matthew Hoch16
Kevin Mandernack17
Linda Stathoplos18
Research Interns
Brad Herman19
GEOPHYSICAL LABORATORY
205
Virginia Mattingly20
William Merrill21
Supporting Staff
Andrew J. Antoszyk, Shop Foreman
Bobbie Brown, Instrument Maker22
Stephen D. Coley, Sr., Instrument Maker
Roy R. Dingus, Instrument Maker23
David J. George, Electronics Technician
Christos Hadidiacos, Electronics Engineer
Marjorie E. Imlay, Assistant to the Director
Lavonne Lela, Librarian
Harvey J. Lutz, Technician
Mabel B. Mattingly, Department Secretary
Mary Moore, Word Processor Operator —
Receptionist
Lawrence B. Patrick, Maintenance Super-
visor
David Ratliff, Jr., Maintenance Technician
Pedro J. Roa, Maintenance Technician
Susan Schmidt, Coordinating Secretary
John M. Straub, Business Manager
Mark Vergnetti, Instrument Maker24
Visiting Investigators
Ronald E. Cohen, Naval Research Labora-
tory
David H. Freeman, University of Mary-
land
Jaidong Ko, SUNY, Stony Brook
James Kubicki, Yale University
Julie Kokis, George Washington Univer-
sity
Yali Su, University of Maryland
Bradley Tebo, Scripps Institution of Ocean-
ography
Noreen Tuross, Smithsonian Institution
Donald J.Weidner, SUNY, Stony Brook
Emeritus
Hatten S. Yoder, Jr., Director Emeritus
Felix Chayes, Petrologist Emeritus
Retired June 30, 1989.
2Expiration of Keck Fellowship Arpil 30, 1989.
3To June 30, 1989.
4FromJuly 1, 1988.
5ToJune 1,1989.
6To September 30, 1988.
7To September 1, 1988.
8To February 28, 1989.
9From September 1, 1988.
10To September 30, 1988.
nTo October 30, 1988.
12To June 30, 1989.
13To September 30, 1988.
14From July 1, 1988 to June 30, 1989.
15From July 1, 1988 to June 30, 1989.
16FromJuly 1, 1988.
17From July 1,1988.
18To June 30, 1989.
19FromJunel, 1989.
20FromJune 1, 1989.
21FromJune 1,1989.
22FromJuly 1, 1988.
23Transferred to D.T.M. February 1989.
24From April 1, 1989.