Full text of "Solubility and biocompatibility of glass"

See other formats

SOLUBILITY AND BIOCOMPATIBILITY OF GLASS

ARTHUR E. CLARK, JR,

A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF

THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA
1974

UNIVERSITY nc

ACKNOWLEDGMENTS

The author extends his sincere appreciation to his
advisor, L. L. Hench , for his guidance and encouragement
throughout the course of this study. Thanks are also extended
to H. A. Paschall for his time and patience in helping the
author with interpretation of the histological results of
this study. The author will be forever indebted to his wife,
Lisa, whose patience and encouragement made this work possible
To C, G. Pantano, thanks are extended for the extensive use
of his equipment and personal time. Finally, the author
wishes to extend his appreciation to the many students, co-
workers, and friends who have afforded assistance throughout
the course of this work.

This work was supported by the U.S. Army Medical Research
and Development Command, Washington, D.C.

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS ii

LIST OF TABLES v

LIST OF FIGURES vi

ABSTRACT xi

CPiAPTER

I INTRODUCTION 1

II THE INFLUENCE OF P"*"^, B^^ AND F"^ ON THE
CORROSION BEHAVIOR OF AN INVERT SODA- LIME-
SILICA GLASS 8

Introduction 8

Experimental Procedures 11

Data Analysis 17

Results '. 18

Discussion ■ 76

Conclusions 89

III AUGER SPECTROSCOPIC ANALYSIS OF BIOGLASS

CORROSION FILMS 9 3

Introduction . - 93

Theory 9 3

Experimental Procedure 98

Results 103

Discussion 122

Conclusions 129

IV THE INFLUENCE OF SURFACE CHEMISTRY ON

IMPLANT INTERFACE HISTOLOGY 130

Introduction 130

Experimental Procedure 130

Results and Discussion 153

Conclusions 159

111

TABLE OF CONTENTS - Continued

CHAPTER

V CONCLUSIONS AND SUGGESTIONS FOR FUTURE
WORK

BIBLIOGRAPHY *

BIOGRAPHICAL SKETCH

Page

160
166

LIST OF TABLES

Table Page

1 Bioglass Compositions for Surface Chemistry
Analyses 10

2 d-Spacings Obtained from Corrosion Films on
45S-6I P2O5 and 45B5S5 Glasses Corroded for
1,500 Hrs . Corresponding d-Spacings of

Dahllite are Included 86

3 Bioglass Compositions Selected for Auger
Spectroscopic Analysis 99

4 Bioglass Compositions Implanted in Rat Tibiae. . 131

5 Energy Dispersive X-ray Analysis of the
Effect of Conditioning Treatment of Bioglass
Surface 134

LIST OF FIGURES

Figure Page

1 Schematic block diagram of the atomic
emission spectrophotometer employed for

solution analyses 14

2 Time dependent release of Si02 from bulk
bioglass surfaces into aqueous solution at

37°C 20

3 Time dependent release of Na ions from
bulk bioglass surfaces into aqueous solution

at 37°C 22

+ 2

4 Time dependent release of Ca ions from

bulk bioglass surfaces into aqueous solution

at 37°C 24

5 Time dependent release of P ions from
bulk bioglass surfaces into aqueous solution

at 37°C 26

6 Effect of P2O5 content of bioglasses on the
variation of alpha with corrosion time 29

7 Effect of P2O5 content of bioglasses on the
variation of epsilon with corrosion time .... 32

8 Infrared reflection spectra of freshly
abraded Si02 and bioglass composition

45S-6I P2O5 34

9 Changes in infrared reflection spectra of
four bioglasses with increasing phospJiorus

content as a function of corrosion time .... 37

10 Changes in infrared reflection spectrum of
bioglass composition 45S-6'6 P2O5 as a

function of corrosion time 40

11 Compositional surface changes of a 45S-6o
P2O5 bioglass exposed to a buffered aqueous
solution 43

LIST OF FIGURES - Continued

Figure Page

12 Scanning electron micrographs of corroded

surface of bioglass compositions 46

13 Effect of P2O5 content on the ratio of Si/Ca
for bioglasses corroded 1 hour in an aqueous
solution buffered at pH of 7.4 and maintained

at 37°C 48

14 Time dependent release of Si02 from bulk
bioglass surfaces into aqueous solution

at 37°C 50

15 Time dependent release of Na ions from
bulk bioglass surfaces into aqueous solution

at 37°C 52

16 Time dependent release of Ca ions from
bulk bioglass surfaces into aqueous solution

at 37°C 54

17 Time dependent release of P ions from
bulk bioglass surfaces into aqueous solution

at 37°C '^ 56

18 Effect of B"^^ and F' additions to the bio-
glass composition 455-6% .P2O5 O'"^ ^he varia-
tion of alpha with corrosion time 59

19 Effect of b"^"^ and F' additions to the
45S-6% P2O5 bioglass on the variation of

epsilon with corrosion time 61

20 Changes in infrared reflection spectrum of
the bioglass 45B5S5 as a function of

corrosion time "4

21 Changes in infrared reflection spectrum of
the bioglass 45S5F as a function of corro-
sion time 66

22 A comparison of the infrared reflection
spectra of the bioglasses 45S-6''o P2O5 >
45B5S5 and 45S5F after a corrosion treatment
of 100 hours in an aqueous solution buffered

at pH 7.4 and maintained at 37°C 69

LIST OF FIGURES - Continued

Figure P^g®

23 A comparison of the infrared reflection
spectra of the biogla^ 45B5S5 which had
been corroded for 1,500 hours in an aqueous
solution and reagent grade hydroxyapatite ... 71

24 X-ray diffraction analysis of the crystal-
lization of hydroxyapatite on the surface
of a 455-6% P2O5 bioglass as a function

of corrosion time 73

25 X-ray diffraction spectrum of the crystalline
hydroxyapatite film on the surface of a

45B5S5 bioglass corroded for 1,500 hours .... 75

26 Influence of P2O5 content on the time
required to override the pH of a buffered

aqueous solution 83

27 Influence of B"^ and F" additions to the
^SS-6-6 P2O5 bioglass on the time required
to override the pH of a buffered aqueous

solution 91

28 X-ray energy level diagram depicting a

KL,L^ Auger transition 96

29 Schematic diagram of recording profilometer
and the type of depth measurement plot

generated by the profilometer 102

30 Typical Auger spectra for three depths of
ion milling of a 45S-6'o P2O5 bioglass

corroded one hour at 37°C and pH = 7.4 105

31 Corrosion film profile produced by plotting
peak magnitudes versus ion milling time for
a 45S-6°6 P2O5 bioglass corroded one hour

at 37°C and pH = 7.4 107

32 Chemical profile expressed in atomic percent
of a 45S-6''o P2O5 bioglass corroded one hour

at 37°C and pH = 7.4 110

33 Chemical profile expressed in mole percent
for a 45S-6% ^2^5 bioglass corroded one

hour at 37°C and p?I - 7.4 112

VI 11

LIST OF FIGURES - Continued

Figure Page

34 ConTDaris on of photoelectron spectra o£ a
freshly abraded 45S-6% P2O5 bioglass with
the spectra of a 45S-6% Pz^S bioglass

corroded for one hour at 37°C and pH = 7.4 . . . 115

35 Chemical profile expressed in mole percent
of a 45S-0I P?05 bioglass corroded one

hour at 37°C and pH = 7.4 117

36 Chemical profile expressed in mole percent
of a 45S-3°o P2O5 bioglass corroded one

hour at 37°C and pH = 7.4 119

37 Chemical profile expressed in mole percent
of a 45S-12''ci P2O5 bioglass corroded one

hour at 37°C and pH = 7.4 121

38 Changes in the Auger peak heights of 0, Ca,
P and Si as a function of corrosion time

for a 45S-6?d P2O5 bioglass 124

39 Changes in infrared reflection spectrum
of 45S-0''o P2O5 glass during conditioning
treatment 137

40 Changes in infrared reflection spectrum
of 45S-6'o P2O5 glass during conditioning
treatment 139

41 Electron micrograph of junction between
45S-0% glass and bone three weeks after
implantation in rat tibia 143

42 Light microscopy three weeks after implan-
tation of a 45S-3% P2O5 glass 145

43 Photomicrograph of a 455-66 P2O5 glass-bone
interface three weeks after implantation

in rat tibia 148

44 Electron micrograph of the junction between
the corrosion film of a 45S-6% ^2*^5 glass

and mineralized bone 150

45 Light microscopy three weeks after implan-
tation of a 45S-12I glass 152

LIST OF FIGURES - Continued

Figure Page

46 Photomicrograph of a 45S-12% P2O5 glass-
bone interface eight weeks after implantation. . 154

47 Electron microscopy of capillary in Figure 8 . . 156

Abstract of Dissertation Presented to the Graduate Council

o£ the University of Florida in Partial Fulfillment of the

Requirements for the Degree of Doctor of Philosophy

SOLUBILITY AND BIOCOMPATIBILITY OF GLASS

By
Arthur E. Clark, Jr.
December, 19 74

Chairman: L. L. Hench

Major Department: Materials Science and Engineering

The influence of phosphorus, boron and fluorine addi-
tions on the surface chemical reactivity of a soda-lime-
silica glass has been investigated. Several techniques,
including infrared reflection spectroscopy, ion solution
analysis, scanning electron microscopy, energy dispersive
x-ray analysis, x-ray diffraction. Auger electron spectros-
copy and ion beam milling, have been employed to develop
insight into the morphological and chemical changes which
occur on glass surfaces corroded in a simulated physiologic
environment .

The resulting corrosion layers and the influence of phos-
phorus, boron and fluorine on their compositions and rates of
formation are defined. Surface ion concentration profiles
determined with Auger spectroscopy and ion beam milling
detail the structural alterations produced by aqueous attack.
A mechanism is postulated which explains the sequence of
events leading to the formation of the multiple - layer corro-
sion structures.

Having defined the surface chemical behavior of the
glasses in an invitro environment, an effort is made to
relate these observations to the response elicited when iden-
tical glasses are implanted ki laboratory animals. Stable
interfacial fixation results when specific surface chemistry
conditions are satisfied. Insufficient or excess surface ion
concentrations produce negative osteogenesis and fixation
results .

Based upon the invivo observations, a theory is proposed
that an ideal implant material must have a dynamic surface
chemistry that induces histological changes at the implant
surface wrhich would normally occur if the implant were not
present.

CHAPTER I
INTRODUCTION

Orthopedic prosthetic devices are employed for fixation,
stabilization, and replacement of damaged or diseased bone.
A wide variety of implant configurations are in use today.
These include plates, nails, screws and pins for fixation,
and weight-bearing devices such as hip, femoral, and total
knee prostheses.

Historically, metals have played the predominant role as
prosthetic devices. As early as 1775 AD, evidence in the
literature documents the use of iron wire to suture fractured
bone segments together [1]. Since that time numerous metals
ranging from gold, silver, aluminum, zinc, lead, copper,
nickel, high carbon steel, low carbon steel, cobalt chromium
molybdenum alloy, copper aluminum alloy, magnesium, iron,
titanium, and ti tanium- aluminum- vanadium alloy have been
investigated as candidates for prosthetic devices [2-7]. As
might be expected, a wide range of responses is elicited by
the various metals and alloys. These responses range from
gross corrosion of the metal and bone necrosis adjacent to
the implant, to situations in which the presence of the im-
plant in a physiological environment is well tolerated and
bone formation occurs in close proximity to the implant. As

the investigation o£ metallic implants has progressed, a
series of requirements for an ideal implant material has
evolved. Included in this list are: (a) high corrosion
resistance, (b) suitable mechanical properties for the appli-
cation, (c) excellent wear and abrasion resistance where
required, (d) good tissue compatibility, (e) structural homo-
geneity and soundness, (f) non- thrombogenicity , and (g) rea-
sonable cost [8] .

Metal devices predominantly in use in this country fall
into three categories: Type 316, 316L and 317 stainless
steels (wrought); cobalt- chromium based alloys (cast and
wrought); and titanium (unalloyed, wrought). These materials
all exhibit superior corrosion resistance in the physiologi-
cal environment of the body. However, it has been demonstra-
ted that there is an absence of adherence between implants
made from these materials and bone, because there is always
a fibrous capsule or sheath surrounding the implant and iso-
lating it from tissue [9,10].

The thickness of the fibrous capsule is an indication
of the degree of tissue acceptability; i.e., the thinner the
capsule the better the acceptability. The development of the
fibrous tissue is due to either corrosion of the implant or
mechanical irritation produced by movement of the implant
[11,12].

The lack of direct attachment of living tissue to metal-
lic implants can lead to loosening and motion. The resulting
pain can force surgical removal. Sufficient movement can

lead to implant failure or bone fracture. As a result of
this situation, numerous investigations have been initiated
to find a material which will firmly adhere to bone.

One approach has involved the use of porous metallic
implants. The concept involves bone ingrowth into a porous
surface providing mechanical interlocking. The mechanical
load is distributed over a wide area, reducing the chance of
bone necrosis due to stress concentrations at localized sites.

Hirschhorn e_t a_l. reported deep bone ingrowth into speci-
mens of sintered Ti and Ti-6A-4V alloy with a pore size of
200 ym [13]. Welsh ejt al^. documented bone ingrowtli into
porous Co-Cr-Mo alloy (Vitallium) coatings on solid Vitallium
rods [14].

Galante ej^ al_. [15] used titanium fibers which were com-
pacted in dies and vacuum sintered. The resulting pore size
was reported to be within an order of magnitude of the fiber
diameter. Specimens placed in rabbit and dog femurs revealed
bone ingrowth after 12 weeks. In another related study, hip
prostheses were evaluated after 3 months to a year in dogs.
Deep bone ingrowth and firm stabilization were reported [16].
Pore size was 230 ym.

A process to produce porous metal implants which involves
the use of a sacrificial metal with a low vaporization temper-
ature has been developed at Battelle Northwest Laboratories
[17]. A composite containing the sacrificial metal and the
implant material is formed and machined to the desired size
and shape. The implant is heated to vaporize the sacrificial

metal and then sintered. Cylindrical plugs made with 304
stainless steel, Ti , and Ti-6A1-4V powders have been implanted
into dog femurs for time intervals up to 12 weeks. Bone in-
growth was reported to depths of 2,500 ym [18].

A method for plasma spraying titanium hydride powder on
solid titanium specimens has been developed by Hahn and
Palich [19]. Implants with a porous surface (pore size 50-75
ym) were implanted into femurs of sheep for 14 and 26 weeks.
A significant increase in bond strength was noted when porous
specimens were compared with implants with smooth surfaces.
Although histological examination of the bone-porous surface
was not reported, bone penetration into the pores was postu-
lated on the basis of the differences in bond strength between
the porous and non-porous implants.

The use of porous metal surfaces to anchor prosthetic
devices to bone seems promising.. One of the major points
which remains to be shown is the effect of the increase in
surface area associated with a porous surface and the result-
ing corrosion which would occur over long periods of time.

Another area of interest has centered around the use of
inert porous ceramic materials. Due to their highly oxidized
state, ceramics are inert materials capable of resisting
degradation in severe environments [20]. In addition, ions
incorporated into most ceramics (Na, K, Mg , Ca) are normally
found in the body. Thus, release of these ions from a cer-
amic implant would not present as serious a problem as release
of foreign or toxic elements.

One of the first attempts involved the use of a slip
cast mixture of alumina, silica, calcium carbonate and mag-
nesium carbonate. The resulting porous material (average pore
size 17 ym) was strengthened by vacuum impregnating with an
inert epoxy [21]. Openings at the surface were obtained by
dissolving the epoxy to a depth of 50-70 mils with methylene
chloride. The composite material was called Cerosium and
exhibited mechanical properties similar to bone, Evaluation
of this material revealed little bone ingroivth into the pores.
This was attributed to a small pore size. In addition, a
reduction in the strength values of Cerosium which had been
implanted was related to epoxy degradation by body fluids [22],

The use of porous calcium aluminate has been investigated
by Klawitter and Hulbert [23], Calcium carbonate and alumina
were mixed with water, pressed into pellets, dried, and fired.
An interconnected pore structure was produced by the break-
down of the calcium carbonate and the subsequent release of
C0_ . Pore size was controlled by varying the particle size
of the calcium carbonate. Invivo studies revealed that a
minimum interconnection pore size of 100 pm was necessary for
mineralized bone growth. In addition, there was a lack of
inflammatory responses due to the calcium aluminate implants.
The one unusual response was the presence of a layer of
osteoid ('^^50 ym thick) separating mineralized bone from the
ceramic composite. The authors speculated that a local alka-
line pH change produced by hydration of the surface of the
ceramic composite inhibited mineralization within 50 ym of

the ceramic. Although there was a lack of inflammatory
response elicited, the porous ceramic cannot be considered
completely inert, because of the hydration and resulting
effect on bone mineralization.

Hulbert ej^ a_l. [24] have reviewed the invivo behavior of
numerous porous ceramic materials and found no adverse tissue
response and .mineralized bone ingrowth into several materials.

Preliminary investigations have been conducted employ-
ing dense aluminum oxide (A1_0_) as a prosthetic device [25].
The development of a fibrous sheath separating bone and cer-
amic was noted as the major drawback.

Graves et_ aJ. have recently reported on the development
of a resorbable ceramic implant [26]. The concept of a
resorbable ceramic material has several attractive features.
The initial pore size can be restricted to values less than
optimum for bone penetration. This will result in an increase
in the initial strength of the ceramic. As resorption pro-
ceeds, enlargement of the pore structure will stimulate bone
ingrowth. The drop in strength associated with the increase
in pore size will be compensated for by the presence of the
new bone. The stress concentration at the implant-bone
interface of permanent devices is not a problem as the mate-
rial is completely resorbed with time. There is the poten-
tial for influencing ossification through the release of
specific ions incorporated into the ceramic [26].

Calcium aluminate ceramics with additions of phosphorus
pentoxide were implanted into femurs of mature Rhesus monkeys.

The results pointed to an enhancement of bone formation at
the ceramic- tissue interface as well as within the ceramic
as the PtO|- concentration was increased [26].

A completely unique approach to the problem of permanent
fixation has been initiated by L. L. Hench et_ al_. [27-30].
The concept involves the use of surface reactive bioglasses
to achieve intimate bonding between an implant and bone tis-
sue. Invivo results, obtained at an early stage in the pro-
gram, in the form of transmission electron micrographs,
demonstrated glass - ceramic implants intimately bonded to bone
at 6 weeks with no indication of an inflammatory response to
the implant [31]. It was suggested that some chemical char-
acteristics of the implant may have enhanced ossification at
the glass-bone interface.

The purpose of this text is to describe a systematic
study of a series of glasses (referred to as bioglasses) with
the intent of developing an understanding of their chemical
surface behavior. New surface sensitive techniques such as
Auger Electron Spectroscopy and Infrared Reflection Spectros-
copy along with several other tools have been employed to
examine the response of bioglasses to an aqueous environment
maintained at physiologic temperature and pH. An effort is
then made to relate the observed invitro reactions to a series
of invivo responses. It is the author's opinion that such an
approach has been lacking in many previous investigations of
candidate biomaterials and, hopefully, will serve as a model
for future studies.

CHAP»ER II

THE INFLUENCE OF P^^ , B^"^ AND F"^ ON THE
CORROSION BEHAVIOR OF AN INVERT SODA-LIME-SILICA GLASS

Introduction

The corrosion o£ silicate based glasses can occur by
either selective leaching or complete dissolution, but usually
involves a combination of the two. In general, the process
leads to the formation of a thin film or gel on the exposed
glass surface with the composition of the gel being signifi-
cantly different from that of the uncorroded glass.

The composition and profile of the gel layer are usually
a direct measure of the durability of the glass. Studies on
binary soda-silica and lithia-silica glasses have established
that the corrosion resistance is maximized when the reactions
at the glass surface lead to the formation of a thin gel with
a high surface silica concentration [32].

A series of invert silica glasses are under investigation
for use as prosthetic devices [27-30], and it has been demon-
strated that it is possible to achieve bonding between glass
and living bone in the body [31]. The biological accepta-
bility of a soda-lime-silica glass is significantly affected
by the presence of small amounts of phosphorus, boron, or
fluorine [ 33- 36 ] .

The mechanism by which the bond is developed is essen-
tially a controlled corrosion of the glass which produces a
surface composition that is compatible ;vith bone. The results
of this study have shown that the corrosion behavior of the
bioglasses is directly related to the effects of additions of
phosphorus, boron, and fluorine on the composition and pro-
file of the resulting gel.

Four nondestructive techniques, infrared reflection .
spectroscopy (IRRS) , ion concentration analysis of the corro-
sion solution, scanning electron microscopy coupled with
energy dispersive x-ray analysis and x-ray diffraction are
employed to characterize the corrosion gels. IRRS provides a
direct measure of the surface silica concentration [37] , while
two parameters calculated from the solution data provide a
measure of the total amount of silica available for gel forma-
tion [38]. The parameter a is a measure of the extent of
selective dissolution and varies in magnitude from 0 to 1.
IVhen a approaches 0, selective leaching predominates. As a
approaches 1, total dissolution is the controlling process.
The second parameter, e, referred to as excess silica, is a
measure of the amount of silica available for gel formation
and is calculated from a and the concentration of SiO^ in
solution. (For a detailed discussion see Ref. 38.)

Six glasses were chosen for study. This series of compo-
sitions provides information as to the influence of phosphorus
on the corrosion behavior of the ternary soda- lime-silica
glass (see comp. 1, Table 1) as well as the influence of

Table 1

Bioglass Compositions
for Surface Chemistry Analyses

Weight

. 45S-0% P^O^

45% SiO^

2 4.5% CaO

30.5% Na20

. 45S-3% P-0^

45% SiO

2 4.5% CaO

27.5% Na20

3% P^O^

. 45S-6% P^O^

45% SiO^

2 4.5% CaO

24.5% Na^O

6% P^O^

Weight %

. 45S-12% P„0,

45% SiO„

2 4.5% CaO

18.5% Na20

12% P^O^

. 45B^S5

40% SiO^

5% B^O^

2 4.5% CaO

2 4,5% Na^O

6% P^O^

. 45S5F

43% SiO

12% CaO

16% CaF^

23% Na20

6t PjOj

boron and fluorine on the behavior of glass number 3. Glass
number 3, which contains 6-6 P^O,- is the most compatible with
bone. Boron and fluorine were added to facilitate flame
spraying onto metal substrates as they both reduce the melt-
ing temperature of the glass [39].

Experimental Procedures

The glasses were prepared from reagent grade sodium car-
bonate, reagent grade calcium carbonate, reagent grade phos-
phorus pentoxide, reagent grade boric anhydride, and 5 ym
silica. Premixed batches were melted in platinum crucibles
in a temperature range of 1250 to 1550°C for 24 hours. Sam-
ples were cast in a steel mold and annealed at 450°C for 4 to
6 hours .

Bulk samples of each composition were prepared by wet
grinding with 180, 320, and 600-grit silicon carbide paper.
After a final dry grinding with 600-grit silicon carbide paper,
samples were immersed in 200 ml of aqueous solution buffered
at a pH of 7.4. Buffering was accomplished with a physiologi-
cal buffer (trishydroxymethyl aminomethane) [40]. Stock
solutions of .2 M tris (hydroxymethyl) aminomethane and .2 M
HCl were mixed with distilled and deionized water to produce
a pH of 7.4. Temperature was maintained at 37°C and the
duration of exposure was varied from .1 to 1,500 hours. All
sample solutions were maintained in a static state. A Cole-
man Metrion IV pH meter with ±0.05 pH accuracy was used to
monitor change in pll.

Each sample was subjected to infrared reflection analy-
sis immediately upon removal from the corrosion solution and
compared with the spectrum of an uncorroded sample. The IR
radiation reflected from the glass surface is measured over
a spectrum of wavenumbers from 1,400 to 250 cm . The peaks
produced are characteristic of the vibrations of specific
ionic bonds in the glass structure [41]. By comparing the
reflectance spectra of corroded versus uncorroded glasses and
also the spectra of glasses of varying composition, informa-
tion about the type of structural change as well as the rates
of changes can be obtained [37]. All measurements were taken
on a Perkin-Elmer 467 Grating Infrared Spectrophotometer
equipped with a specular reflectance accessory.

Solution analysis was performed employing atomic emis-
sion spectroscopy and colorimetric techniques. Figure 1 is a
schematic block diagram of the atomic emission spectrophotom-
eter employed for these analyses. Samples of buffered aqueous
solutions in which glass specimens had been immersed for
specific periods of time are introduced into the flame through
the nebulizer burner system. An atom vapor which consists of
atoms in the ground state and thermally excited states is
produced in the flame. As atoms in the thermally excited
states return to the ground state, they emit radiation with a
wavelength characteristic of the type of atom involved. This
characteristic radiation, which is isolated in the monochro-
mator and intensified in the photomultiplier module, can be

t/l

I— 1

-p

■P

1— 1

<4H

J-l

1— 1

r-t

■ H

■P

-M

-p

E
o

O 3

3 TJ
^ C

Z X

related to the concentration of the atoms in the original
sample solution.

The normal procedure consisted of running undiluted sam-
ples and comparing the results with a series of premixed
standards witli concentrations ranging from 10, 25, 50, 100,
150 and 200 ppm of the ionic species being analyzed. Based
on these results, the unknown samples were diluted into a
range of 1-10 ppm. Premixed standards of 1 , 2, 4, 6, 8 and
10 ppm were analyzed and a plot of intensity versus concen-
tration (ppm) was obtained. The diluted samples were run
along with the second series of standards. Plotting the
intensities of the unknown samples on the predetermined stan-
dard curve enabled one to obtain an accurate measurement of
the unknown ionic concentration. This method was employed
to determine calcium and sodium released into solution.

The colorimetric procedure . involves the use of a Ilach
Direct Reading Colorimeter which relates the intensity of
light at a specific wavelength passing through a sample solu-
tion to the concentration of a particular ion in the solution,

The colorimetric molybdos i licate method and heteropoly
blue method were used for silica determination [42]. In both
of these procedures ammonium molybdate is added to the un-
known solution, and reacts with any silica present to form
molybdosilicate acid which has a yellow color. The intensity
of the yellow color is proportional to the concentration of
silica in solution. In the heteropoly blue method, the
yellow molybdosilicate acid is reduced with aminonaptholsul -

fonic acid to heteropoly blue. The resulting blue color is
more intense than the yellow and provides a more sensitive
measurement of the amount of silica [43]. The molybdosili-
cate method has a range of O-KO ppm, whereas the heteropoly
blue method has a range of 0-3 ppm. Normal procedure in-
volved measurement of undiluted samples with the molybdosili-
cate method, followed by dilution and a second measurement
with the heteropoly blue method. In both tests oxalic acid
was used to eliminate interference from phosphate groups.

The Phos Ver III method [42] was employed for total
phosphate determination. This method has a range of 0-3 ppm.
Dilutions were made until two successive dilutions yielded
the same results.

Several samples of each composition were examined with a
Cambridge Scanning Electron Microscope equipped with an Ortec
Energy Dispersive X-ray Analysis System. In this system a
lithium drifted silicon detector is used to separate radia-
tion according to its energy. X-rays, produced as a result
of the primary electron beam striking the sample surface,
excite electrons of the silicon atoms. Each of the excited
electrons absorbs 3.8 eV of energy. Since numerous electrons
are excited by a single x-ray, the total charge generated
produces a current which is proportional to the energy of
the x-ray. The current is then stored in a multichannel
analyzer according to its amplitude, until a sufficient number
of x-rays have been counted [44].

X-ray diffraction patterns of selected samples were
utilized to identify the corrosion films which formed on the
glass surfaces. A Phillips Vertical Dif fractometer with a
graphite diffracted beam monochromator was employed. Cu Ka
radiation was used, with tube settings of 40 kV and 15 milli-
amps. Pulse height selection was utilized to reduce back-
ground noise .

Data Analysis

Sanders and Hench have presented the following equation
for the calculation of a for binary silicate glasses:

moles of SiO^ in solution /moles SiO^ in glass
^ -' ^ " moles R2O in solution 7 moles R2O in glass

t23 - ^^2 pp^ ^. Mw SiO^ 1-Pm

where Pm = mole fraction R^O in glass, MW = molecular weight,
PPM SiO^ = concentration of SiO in solution, and PPM R =
concentration of R in solution [38].

Extension of the relation to a ternary soda- lime-silica
glass leads to the following modification of equation (2).

PPM SiO^
MW SiO^ ^ ' ^SiO^

(3) a = ^ — p

1/2 PPM Na ^ PPM Ca ^ ' SiO
MW Na MW Ca

where P^-^ = mole fraction of SiO^ in glass and all other
S1O2 2

symbols are as presented in equation (2). All alpha values
presented in this text were calculated from equation (3).

The presence of small amounts of phosphorus, boron, and fluor-
ine in the bioglasses may introduce slight inaccuracies into
the absolute magnitudes of the individual alpha values.
However, the significant in*Formation obtained from the a data'
is the extent of selective leaching from the silicate network
Avith time and its effect on the resulting corrosion layers
which are produced. In this respect, the equation employed
for the alpha calculations (3) becomes a sensitive indicator
of the influence of the phosphorus, boron, and fluorine addi-
tions on the corrosion behavior of the silicate network.

The equation utilized for the calculation of the excess
silica (e) was introduced by Sanders and Hench [38] and is
presented in equation (4) .

(4) e = PPM Si02 (— ^)

Results •

The time dependent behavior of ion release into solution
is presented in Figures 2-5 for the four glasses with increas-
ing phosphorus content. The glasses containing 0, 3 and 6
wt.''6 PoOp exhibit an orderly decrease in the amount of Na, Ca
and SiO„ in solution, whereas the glass containing 12°6 PoO,.
reverses the trend \\/ith an increase in SiO„ and Ca released
compared with the 6°b PoOj. glass.

Figure 5 shows the phosphorus solution data for the
three glasses with increasing phosphorus content. The
behavior of all three compositions is similar in that a linear

■H

•

f-H

r--

-P

•H

+->

I— 1

(/)

i-H

f-(

•H

P-

M-i

f-^

•H

t/)

CLCO

m
u

•H

M-i

03 o
CD -H

t:)

r-^

ft ;3

Cvi

+->

■rH

■H

ho
O

!h U

<+^'

r--

•M

•H

CTJ

-J

•H

■P

<-H

I— 1

t-H

■M

+-»

•H

PL,

(/)

M-^

f-<

O
•H

o u

•p

131

•H

d, 1/)

U-t

5-.

O
O

increase is followed by a drop in the phosphorus level. The
glass containing 3"6 ^yOr exhibits an increase in phosphorus
released for 100 hours, whereas the glasses containing 6 and
12% ^9^^ ^^'^'^^ ^ dron after 10 hours.

The theoretical parameters a and £ are calculated from
the solution data. Figure 6 is a plot of a, the extent of
selective leaching, versus time for the four glasses. The
glass containing 0% ^7*^^ exhibits a behavior which suggests
that selective leaching predominates throughout the entire
process. Although the curve initially increases, indicating
a tendency towards complete dissolution [38], the maximum a
value attained is only 0.37 and this is followed by a levelinj
off to an a value of 0.28. As the phosphorus content of the
glass is increased, the maximum a value achieved increases,
with the glass containing 12 0 Po^r having an a value of 0.6
at 100 hours.

In evaluating the influence of PoO^ content on the over-
all corrosion process, Figure 6 can be divided into 3 time
regimes. During the initial 20 hours of exposure the glasses
containing 0, 3 and 6 ivt.l Po^c show a fairly consistent
increase in their respective a values. The curve obtained
for the glass containing 12°^ PoO^ fluctuates above and below
the curve of the 61 PoOp glass. In region II a uniform trend
is observed, i.e., as the P^O;- content increases the a values
increase. At 100 hours this behavior reverses with the
glasses containing a larger percentage of PoO;. exhibiting a
more negative slope as the a values drop (i^egion III).

•H
4-1

nj
■H
U
as
>

4->

P!
O

1— 1

• H

-P

■H

<+H

• p

-p

■p

o 4:;

+->

• p

CNI/ ^

'4^

■P

i-H

M-l

i+H t+H

H
P
bO

•H

••■' '' '

)

\\ \ /

in in m lo
OOOO

'•i .' /

C\J CvJ C\J OJ

Q. CL D- Q.

V-.. ; /

-9 -o -9^
5~ 6~5~o ,

o n CD ^

\ ■••..»' /

1 1 1 1
W W CO CO

/ ••• \\

u^ tn in Lo

'f -^ •^ >*

I'M

( / ^\

III:

V \

1 ^\

1 •■ Av

i •' V

L* \n

.1 v>«

\ N

•'1 \ *>■

1— 1

•• ' \ >

M V

\ /\

i *• ' \

1 '. ' \

i • ' \

1 •. / ^

1 1

, \ .!

O K

O s
h- »-

Epsilon is plotted as a function of time in Figure 7.
As was stated earlier, epsilon is a measure of the amount of
silica available for film formation. An increase in epsilon
indicates that a film is forming while a decrease is a result
of film breakdown. In order for a film to be protective it
should have a high epsilon value. However, the magnitude of
epsilon alone does not completely characterize the effective-
ness of a corrosion film. The profile of the film is an
important parameter. Thin films with a high concentration of
silica at the surface (within 5 ym) are much more effective
at retarding network breakdown and release of silica into
solution than are thicker films with a more even silica dis-
tribution.

The data of Figure 7 illustrate that, as the P20^ content
increases, the amount of silica available for film formation
decreases for the glasses containing 3 and 6 wt.% P2°5" '^^^
curve for the 12"^ P^^^ glass deviates from this pattern.

Infrared reflection spectra of vitreous silica and the
glass containing 6% V^O^ are shown in Figure 8. The vitreous
silica peak at 1,115 cm" has been attributed to a bond stretch-
ing vibration of silicon-oxygen-silicon atoms [45,46], while
the peak at 475 cm"-^ is produced by bending or rocking motions
of silicon-oxygen-silicon atoms [45,46]. As alkali or
alkaline-earth oxides are added to vitreous silica several
events occur. The Si-O-Si (S) stretching peak experiences
a reduction in intensity and a shift to a lower wavenumber.
Also, the intensity of the Si-0 rocking (R) peak is suppressed,

•H
■P

■H

4->
O

1/1 (U

(/I S

.-H -P

<+-! O

O P

4-1 O

c o

-13

t/) O

M-i

vT)

i*-i

•^

+->

• H

■P

•H

■P

1—1

M-l

t/i

1—1

!-i

•H

cti

-Q

!-i

U-^ Tj

i — 1

39NVi331d3U

In addition, a new peak develops in the region of 950 cm
The addition of alkali and alkaline earth oxides (i.e., Na^O,
CaO) disrupts the continuous three-dimensional vitreous silica
network by producing s i licon-nonbridging oxygens to satisfy
the new cations (i.e.,- Na or Ca) . The intensity drops of
the S and R peaks of vitreous silica are due to the decrease
in the number of Si-O-Si bonds. The new peak at 950 cm has
been ascribed to bond stretching of the s i 1 icon-nonb ridging
oxygen atoms (NS) [37]. The shift of the S peak to a lower
wave number is a result of the change in local environment
brought about by the presence of the s ilicon-nonbridging
oxygen-cation groups. Simon and McMahon have indicated that
the Si-0 bond force constant is decreased by the presence of
the cationic field of the network modifiers [47].

Infrared reflection spectra of corroded and uncorroded
surfaces from the series of glasses containing PoOr ^^^ pre-
sented in Figure 9. Comparison of the uncorroded spectra
with the short and long corrosion times reveals several
interesting facts. The silicon-oxygen-silicon stretching peak
(S) at 1,000 cm" begins to sharpen and shift towards the
location of the Si-O-Si stretching peak for pure vitreous
SiO (1,115 cm' ) after 15 minutes for the glass containing
0% P9O . Simultaneously there is a considerable drop in the
intensity of the si licon-nonbridging oxygen peak (NSX) at
950 cm' . The silicon- oxygen rocking peak (R) located at
500 cm' also increases in intensity and sharpness after 15
minutes' corrosion. In addition, there is a shift in

Figure 9, Changes in infrared reflection spectra

of four bioglasses with increasing phosphorus
content as a function of corrosion time.
Solutions were buffered at a pH of 7.4 and
maintained at 37°C.

-, s*---

UNCORRODED

15 MIN.

120 MIN.

A. 45S-0%PoO5

"istRy

1200

800 600

C. 45 S - 6%P205

400

1200

1000

800

600

400

— UNCORRODED

— 15 MIN.
— 120 MIN.

1200

1000 800 600

WAVENUMBER(CM-I)

400

location towards the Si-O-Si rocking vibration frequency o£
pure silica (475 cm' ). These trends continue for a corro-
sion exposure of 120 minutes with one exception. The inten-
sity of the rocking peak at #475 cm is somewhat lower than
it was at 15 minutes.

For glasses with higher phosphorus contents, the 15-
minute spectra show an increasing preferential attack of the
silicon-oxygen-silicon stretching peak (S) , and a decreasing
preferential attack of the silicon-nonbridging oxygen peak
(NSX) . The increase in intensity, and location of the shift
of the silicon-oxygen rocking peak (R) are also retarded for
the higher phosphorus glasses.

At corrosion times varying from 75 to 120 minutes, there
is a complete reversal in behavior. For each of the three
glasses containing PoO,. there is an increase in the intensity
of the S peak while the intensity of the NSX peak is signifi-
cantly reduced. The longer corrosion times for each composi-
tion represent the maximum exposure before the glass surface
has roughened to the point where the intensity of the spectra
is reduced to the extent that reliable data cannot be obtained,
The time required before surface roughening dominates is
shortened as the PoO,. content of the glass increases. Even-
tually the spectra of the glasses containing PoOr become flat
curves with a very low intensity.

However, with sufficient corrosion time a new infrared
spectrum develops which is different from that of the glass.
Figure 10 contains a series of IR spectra which illustrate

Figure 10. Changes in infrared reflection spectrum
of bioglass composition 45S-6°6 PoO- as a
function of corrosion time.

1400 1200 1000 800 600 400

WAVENUMBER (CM I)

the sequence of reactions for the glass containing 6% P^O .
This new spectrum (see Figure lOd and e) develops for all
three glasses containing P^O^ , the only variable being the
length of corrosion treatment required to produce it. The
new spectrum begins to appear in as short a time as 4 hours
for the glasses containing 12°6 P^O,-, and takes 12 hours to
develop for the glass containing 3''o P^Oq-

X-ray spectra taken from the glass containing 6% PoO^
with the energy dispersive system of the SEM are shown in
Figure 11. The iron peak seen in each of the spectra is pro-
duced by x-rays originating from a pole piece in the SEM
column. The variance in the size of the iron peak indicates
that identical conditions (i.e., specimen tilt angle and
counting rate) were not achieved for each spectrum. A crude
comparison of peaks from different spectra can be obtained
by dividing the peak intensities of the various elements by
the intensity of the iron peak in the same spectrum. Another
way of achieving the same end is by comparing the ratio of
two peaks in one spectrum with the same ratio from another
spectrum.

After two hours in solution, the Si/Ca ratio for the
glass with 6% PoOp has increased from 0.9 to 2.2. In addi-
tion, the sodium and phosphorus peaks have completely dis-
appeared. The Si/Ca ratio began to drop after two hours and
at 1,500 hours was 0.23. The phosphorus peak reappears at
20 hours and continues to increase with corrosion time. The
24-hour spectrum shows that the ratio of Si/Ca has dropped to

f-l

■p

■P

<f)

t/)

PuX

CtJ

• H

, — ,

f-i

1
X

Lni~~~

•

•H

Crt

c>\°

p.f-1

^ — '

Pj t/)

t/)

•H

+->

•iH

>>2;

I— 1

M-4

If)

If) w

•P

t-H

P w

cr'

!-i

rt 0

n:3

■H

!=l

(=1

M-l

0) rc:

5-1 M-l

<+^

•H

-ID

I— 1

t:)

•H

■p

+->

4->

■ H

TIJ

cti

0 u

1 0

, f-i

AllSNaiNI

1.43 while the ratio of Ca/P is 2.4. At 1,500 hours the
phosphorus peak has reached a sufficient magnitude to make
the ratio of Si/P (.47) and Ca/P (2.04) several times smaller
than was observed in the uncorroded glass.

Micrographs of the corroded surfaces of the four glasses
with variable phosphorus content are shown in Figure 12.
Although the exposure time was only 1 hour, a thick film has
formed on the surface of each glass, indicating a significant
amount of corrosion has already occurred. Figure 13 is a
plot of the change in ratio of Si/Ca as a function of PoO,.
content for the four samples shown in the preceding figure.
The Si/Ca ratio of each glass in the uncorroded state is also
included. The ratio of Si/Ca drops significantly as the PoO-
content of the glass increases. However, the ratio of Si/Ca
is greater in the corroded glass than in the uncorroded glass
for all four compositions.

Figures 14-17 present the time dependent behavior of ion
release into solution for the glasses which contain boron and
fluorine. Since these two glasses are variations of the com-
position containing 6-6 P-Or, its solution data are included
for comparison. The release of SiO^ and Na into solution
is similar for the three compositions. However, it should be
noted that after .1 hour of exposure, the amount of silica
released into solution is slightly higher for the boron-
containing glass at every point on the curve. Comparison of
the glass compositions (see Table 1) reveals that 5 wt.°6 B^O^

Figure 12. Scanning electron micrographs o£ corroded
surface of bioglass compositions,
(A) 45S-0°s P2O5, (B) 45S-3"ti P2O5 ,
(C) 45S-6I P2O5, (D) 45S-12% P2O5.
Samples were corroded for one hour in an
aqueous solution buffered at pH of 7.4
and maintained at 37°C. The surfaces were
ground with dry 600. grit SiC prior to the
corrosion treatment.

A. 45S-0% PjOg

B. 45S-3% P2O5

^^^^.

C.45S-6%P20Sj D.45S-12O>t)P205

m U-H

+->

rt o

f-H

r— 1

O M

md:

o a

1^ -rH

•H

•M C

^ -P

•H C

:S 03

O t3 -TS (jO

"-H O

f-i

C (U

oj 0)

■ H M

U U-{

rt t3

~-^U-i

■P -H

•H :3

JD fH

W 4D

(D Oj

4^ C

o o

f- U

•pH

O -M

Is rt

•H 3

■P t-l

m ^

rt o

■P o

!h in

Q e

0) tyi

^ P

■P

■P o

• (fl

(DUX

C ^

o c/5

o cr

K^ X

■P

ct)

C C

■p f^

0) oi

Clj 1

-p

X (U

c rt

^ CX

O -H

CD (D O

C > U

■H .H (fl

LnHS

rt (fl O

O O

■P f-l fH

INl^

d 0 u

•H pu-H

i-H

rt cfl S

e-H

O "XJ

Q C

n3 o

+J Tj

0 >. f^

U O

rt bo +j

<U M

f-i u

<+-l i-^

^ <D O

m o

• C '-H

w u

f^ m p::

..I CO

ca|u

■H

•

d <->

r-^

5-1

+->

<4H

(XI

•H

■P

(/)

5-1

+->

t:)

f:J

•H

M-(

1-1

•H

CN4

ri4

3
^

Mh o

CO fo

O 4->

• H Oj

-H S

f O

2 -P

U-l ,-1

o o

5-H

■P

■p

-d

•H

<S)

•H

(/)

bO
O

^13

<4^

-P

•H

+-)

<-H

1— 1

r— 1

•M

+-)

•H

1+^

•H

O U
!-io

oi id

C o

•H

Pu to

■D

M-l

•H

t0i

was substituted for SiO Thus, the glass which contains
boron has the least amount of silica in its bulk composition.
There is a significant difference in the behavior of

calcium released into solution (Figure 16). At 10 hours

+ 2
there has been more Ca released from the glasses containing

boron and fluorine than from the glass containing 6-0 P^O^.
The level of calcium released remains fairly constant through-
out the remaining 1,490 hours for the glass containing fluor-

+ 2
ine , while the Ca release level of the glass containing

6% PoOr surpasses it at approximately 150 hours. The level

of calcium released into solution for the glass containing

boron continues to increase at a slower rate after 10 hours,

+ 2
but it remains above the Ca release level of the glass con-
taining 6% PoOr ior the entire duration of the corrosion
treatment .

Up to 10 hours, the concentration of phosphorus in solu-
tion is very similar for the glass containing fluorine and
the glass containing 6% PoOq (see Figure 17). After this
point there is a drastic drop in the P level for the glass
containing fluorine. The glass containing boron parallels
the glass containing 61 ^o^c but the P level is signifi-
cantly lower at every point.

Figures 18 and 19 show the alpha (a) and epsilon (e)
data for the glasses containing fluorine and boron as Avell
as the glass containing 6% ^^7^q- "^he alpha curve (Figure 18)
for the glass with boron rapidly attains a maximum value of
.58. After two hours there is a gradual decrease in alpha

•H

4->

•H

(/5

a o

f-l

I/)

■P

t-H

•H

M ^

•H

/ — X

4->

I— I

+-)

•H C

H 03
I >

C -p

hO O

+
o

O D-,

M-^ LTJ

/ / ^"

1 / ^'

'• A^

1 /'"

/ ^

! /^"

1 / {

i / \

i / j

^ )

(■

/
/

\ /

CO CD CO

in vn in

•^ ■* ■*
1 1

1 i

{

\
1
1
1

^ '

\\ ,

o
o
in

o
o

•

fiC

^"^

1—

1^.

- CM

• H

•H

CSl-P

o\°

•H

l/l

■=d-

•px:

■(->

■H

+->

,_^

^ — '

•H

■P

■H

I— 1

■H

t/)

Ph m

(=;

■H

■P

•H

l-H

CIj

+->

4-1

and at l,50n hours it has dropped to a value of .25. The
alpha curve for the glass containing fluorine remains con-
stant at a value of .45 for two hours, and then increases to
a maximum value of .56 at 40 hours. After 40 hours alpha
decreases linearly to a valu^ of .4 at 1,500 hours.

The amount of silica available for film formation (e)
increases uniformly for all three compositions for the initial
10 hours (see Figure 19). After 10 hours, the epsilon values
for the glass containing boron are significantly higher than
those of the glass containing 6 °s P9O5 » while the epsilon
values of the glass with fluorine are lower than those of the
glass with 6% P^'^r-

Infrared reflection spectra of the glass containing
boron (Figure 20) reveal the same sequence of steps as was
seen for the glass containing 6% PoOc- Initially there is
selective attack of the silica peak (15-minute exposure), but
by one hour a silica-rich layer has formed on the surface.
Surface roughening leads to a drop in intensity of the entire
spectrum, producing a flat curve at three hours. A new spec-
trum begins to develop within 7 hours , and is identical to
the spectrum which was described previously for the glasses
containing 3, 6, and 12"6 PtO^.

A similar series of reactions was observed for the glass
containing fluorine and the results are presented in Figure
21. One difference between the glass containing fluorine and
all other compositions was the shape of the peaks in the IR
spectrum which developed after the spectrum of the glass

Figure 20. Changes in infrared reflection spectrum
of the bioglass 45B5S5 as a function of
corrosion time.

D. 3 HRS. IN SOL.

1400 1200

1000 800

WAVENUMBER (CM'I)

600 400

Figure 21. Claanges in infrared reflection spectrum
of the bioglass 45S5F as a function of
corrosion time.

45S5F

FRESHLY ABRADED

/ #

—^

B. 15 MIN. IN SOL.

C. 1 HR. IN SOL.

1400 1200

E. 7.5 HRS. IN SOL.

1000 800

WAVENUMBER (CM'I)

600 400

disappeared. Figure 22 enables one to compare the IR spectra
of the glass containing 60 PoCi the glass containing boron,
and the glass containing fluorine, after each had been in
solution for 100 hours. There are three peaks in the wave-
number region 500-650 cm and the peak at 600 cm has the
greatest intensity for the glass containing fluorine. The
spectra of the other two compositions have only two peaks in
this region and the peak at 560 cm is dominant. In addi-
tion, the main peak at 1,035 cm is sharper and more intense
for the glass with fluorine than for either of the other two
compos itions .

Infrared reflection spectra of the glass containing
boron (which had been exposed for 1,500 hours) and reagent
grade hydroxy apatite are shown in Figure 23. The two spectra
are very similar, the main differences being the lack of defi-
nition of the shoulder at 1,085 cm and the broadness of the
peak at 1,035 cm for the spectnun of the glass surface.

Figure 24 contains x-ray diffraction curves of the glass
containing 6''6 P^O^ which was immersed for 15, 100, and 1,500
hours. This series illustrates the gradual development of an
amorphous film into a crystalline product. Figure 25 illus-
trates the diffraction curve of the glass containing boron
which had been in solution for 1,5 00 hours.

Figure 22. A comparison of the infrared reflection
spectra of the bioglasses 45S-6% P2O5 >
45B5S5 and 45S5F after a corrosion treat-
ment of 100 hours in an aqueous solution
buffered at pH 7.4 and maintained at 37°C,

45 S - 6% P2O5 - 100 HRS. IN SOL.

-X

1 1

1^ —

1400

1200

1000

800 600

400

1400 1200 1000

800

600 400

1400 1200

1000 800

WAVENUMBER (CM'1)

600 400

if) 0)

(D ^ +->

,£^ 3 -H

■P O +->

^ l#

<+-! Ph

O CD n5

o >,

rt LO X
(h - o

+-> t— I f-i

O 13)

(D !-i >^

Ph O ^
<f) '4-(

O CD Oj

•H Ti fH
-POM
O fn
CD f-i

!-■

M
<U <D

oj Id p;

?H 03 rt
•H ^

•H
P

p 5 l-l
o

<4-i Ln (/5
O CO

O LD O

t/) ^* <D

f-i t/i C7
nJ I/) rt
p. Oj

o U) rt
o o

< JZi -H

aoNviaauan

•

■H

•H

+->

■P '

CO CO

M ■=*
o

^ Mh .
■P O (D

O U +J

CO >4H f::

•rH !-< O

>^ t/) tAl

rH O

03 +-)

fn
O
U

O M^

■p n

■H O
P -H

•4H C3 P

MH P. O

•H 03 pl

i:} >^ 3

X o

03 ;-^ 03

I >N in

AilSNaiNI

■p

•H
+->

ni
X O

o m
u

>^ <D

^ t3

(U fn

•H O
rH O

t/)

>-.

1— i

f-l

• H

l+H

■p

!/)

■-H

•H

}-i

-(->

I/)

M-i

■p

• H

•xi

c ^

i-H

I-H
•H

I-H

UJ
UJ

AIISNBINI

Discussion

The behavior of the glass containing 0% P20^ is easily
interpreted since the results all point to the development
o£ a silica-rich film through a corrosion reaction dominated
by selective leaching. The evidence in support of this
statement is :

(1) The maximum value of a is .37 (see Figure 6) and
this occurs at an early stage (10 hours). In order for com-
plete dissolution to occur, a must approach a value of 1 [37],

(2) After reaching its maximum value, a rapidly drops to
.3 and remains near this value for over 1,400 hours, indicat-
ing no tendency for the film to break down.

(3) Epsilon (Figure 7) increases linearly with time for
100 hours and then levels off. The rapid increase in e which
occurs during the initial 100 hours indicates that a silica-
rich film is developing. Any tendency for film breakdown
would result in a drop in the £ curve. Clearly, no such ten-
dency is observed throughout the entire 1,500 hours of expo-
sure.

(4) The infrared reflection spectrum in Figure 9a shows
immediate selective attack of the silicon-nonbridging oxygen
peak (NSX) and the development of stretching (S) and rocking
(R) peaks associated with pure vitreous silica. After two
hours , the intensity of the entire spectrum begins to drop
uniformly. This drop is due to greater light scattering as
the surface roughens. This phenomenon is unfortunate because

it does not enable one to obtain a quantitative estimate of
the surface composition.

Sanders and Hench have shown that infrared reflectances
are proportional to the amount of species causing them [37].
This relationship assumes that the surface is sufficiently
smooth to produce predominantly specular reflection. This is
not the case with the glasses under investigation. However,
qualitative interpretation can lead to information concerning
the extent of selective leaching from the surface. It should
be pointed out that IRRS has a maximum depth penetration of
less than 1 pm for silicate glasses, and is therefore provid-
ing information about changes occurring at the surface of the
corrosion film. In this case it can be seen that a surface
film composed almost entirely of silica forms within 2 hours.

(5) The use of energy dispersive x-ray analysis shows
that after 1 hour in solution the ratio of Si/Ca on the glass
surface increased from .9 to 5.6 (see Figure 13), again demon-
strating that the glass is being selectively leached, leaving
behind a silica-rich film.

The influence of P^O content on the corrosion behavior
as seen in the data is somewhat complex. Referring to region
I of Figure 6, the initial change in alpha suggests that the
glass structure is more uniformly attacked as the P^O;- content
increases. The glasses containing 6 and 12°6 PoOj. have alpha
values slightly above 0.5, indicating that a significant part
of the corrosion mechanism is total dissolution. This is
substantiated by the IR spectra of Figure 9. Referring to

the 15-minute exposures, the decrease in intensity of the S
peak as the phosphorus content increases is a result o£
preferential attack of the silicon- oxygen-silicon bonds. The
thickness of the corrosion film at very early corrosion times
is less than 1 ym, so the IR spectra are representative of
the entire film. Within an hour the film thickness has been
observed with scanning electron microscopy to increase to
values on the order of 5-10 ym [48]. Then the IR spectra are
providing information about the surface of the corrosion film.

The Si/Ca ratios in Figure 13 of the four glasses with
increasing phosphorus content indicate that a silica-rich
film has formed on each of the glasses within one hour. How-
ever, the level of the Si/Ca ratio on the surface decreases
as the phosphorus content increases, suggesting that the sur-
face is more uniformly attacked as the phosphorus content of
the glass increases. The corrosion films in Figure 12 exhibit
less surface roughness as the phosphorus content increases ,
as would be expected if the glass structure was being uni-
formly attacked. Examination of the corroded glass surfaces
with a scanning electron microscope equipped with an energy
dispersive x-ray system leads to the same conclusion derived
from solution analysis of the ions leached from the glass
structure .

The glass containing 3% P20^ forms a silica-rich layer
almost immediately, while the 6 and 12% P^^^ glasses show
preferential silica attack within the first 15 minutes of
exposure. This behavior is reversed within two hours for the

glasses containing 6 and 12°o P^^S ^^ ^'^® intensity of the S
peak increases while the intensity of the NSX peak is reduced
(see Figure 9), As was discussed earlier, light scattering
resulting from surface roughness leads to an intensity drop
in an IR spectrum. The fact that the intensity of the S peak
increases after the initial drop indicates that a significant
amount of silica is present on the surface.

The amount of silica available for film formation (Figure
7) increases uniformly witli time in region I for all four com-
positions. It is during this period that the silica-rich
film forms on the glasses. A break occurs in each of the
curves in region II. This event corresponds to the formation
of a calcium phosphate film for the three glasses containing
P^O- and occurs earlier as the P^O^. content increases.

Direct evidence for the existence of the calcium phos-
phate film is presented in Figure 11. The series of spectra
show the clianges which occur at the surface of the glass con-
taining 6°6 PoO when it is exposed to an aqueous environment.
A silica-rich film forms within 2 hours as has already been
discussed. The phosphorus peak has reappeared in the 24-hour
spectrimiand the ratio of Si/Ca has dropped. By 1,500 hours
the phosphorus peak has continued to grow while the silicon
peak has been drastically reduced. Comparison in Figure 11 of
the respective ratios of Si/Ca, Si/P, and Ca/P clearly demon-
strates the formation of a calcium phosphate rich layer.

The calcium phosphate film is responsible for the infra-
red reflection spectra which develop after surface roughening

causes the spectra of the glasses containing phosphorus to
diminish. The new spectrum is very similar for all the
glasses containing phosphorus and it develops more rapidly
as the phosphorus content increases. Figure lOe illustrates
the spectrum for the glass with 6 °6 P^O which had been immersed
for 1,500 hours. The peaks occur in two regions, 1,045 cm
and 560 cm . Levitt et^ al_. have identified fundamental wave-
numbers for the phosphate ion of hydroxy apatite in these same

regions [49]. In addition, Nakamoto [50] has predicted that

- 3
the infrared active fundamentals of the PO. ion in aqueous

solution are at 1,080 cm and 500 cm . This evidence,
along with the simultaneous buildup of calcium and phosphorus
at the surface, identified from Figure 11, is the basis for
specifying the origin of the new spectrum as a calcium phos-
phate compound. . •

The details of the calcium phosphate compound film forma-
tion are not completely understood. It has been established
that after 10 hours, phosphorus which has been leached into
solution precipitates back onto the glass surface (see Figure

5) for the compositions containing 6 and \1% P';,0 . In addi-

+ 2
tion, Ca release is retarded during this same time period.

+ 2
Figure 4 shows a leveling off in the amount of Ca released

after 10 hours and the effect is more pronounced as the PoO-
content of the glass increases. The data points in Figure 13
emphasize this concept. The ratio of Si/Ca drops significantly
with increasing PoOp content when the four glasses are cor-
roded under identical conditions. The decrease indicates

that proportionally less Ca is removed as the phosphorus
content o£ the glass increases.

The formation of the calcium phosphate film influences
the corrosion behavior of the glasses significantly. Its
effect is seen in region III of Figure 6. As the phosphorus
content of the glass increases, the a curves descend with
increasing negative slopes, indicating selective leaching is
the controlling mechanism. The solution data (see Figures
2-4) show that both the silicon and sodium release levels off
during region III but that the Ca release actually increases

after the calcium phosphate film is formed. This could be

+ 2
due to the excessive amount of Ca present in the glass

compositions as compared to the P^O^ content. Once all the
phosphorus has been used up in the film formation, the remain-
ing Ca goes into solution. However, the film acts as a
barrier to further attack of the bulk glass structure.

The relative effectiveness of the films in isolating the
bulk glass from the aqueous environment is demonstrated in
Figure 26. It can be seen that the time required to override
the pH of a buffered solution increases as tlie PpOp content
increases. Since the pH increase results from a sodium-
proton exchange between the glass and solution [51], the
formation of the calcium phosphate film retards this reaction
and the effect is more pronounced as the film formation is
accelerated.

Now let us turn our attention to tlie influence of boron
and fluorine additions on the corrosion behavior of the glass

•H

■M

CTt-H

>-4

t/i

■M

> ^

C ^H

■M

M-i

LO O

la:

4-(

O X

+->

CTJ

U-i

1— 1

•H

d
o

o
d
o

€/i

,<M

in in m lo

o oo o

5* o o cvj
o coco T-

i I

CO cow CO

in in If) lo

• ■ 4 o

containing 6% ^o^c- There is a pronounced difference in the
protectiveness of the calcium phosphate film which forms on
these glasses. Figure 27 demonstrates the effect of adding
boron or fluorine to the bulk glass on the time required to
override the pH of a buffered SK)lution. Obviously, the glass
containing fluorine is much more effective than either of the
other two glasses in preventing an increase in pH due to a
sodium-proton exchange. In fact, the addition of boron
actually reduces the reaction time necessary to overcome the
buffering capacity of the solution.

The reasons for the drastic difference in behavior are
not intuitively obvious. Both the glass with boron and the
composition containing fluorine exhibit a behavior similar to
that of the glass with 6% ^y^S' ^^^^ i^ » initially there is
selective leaching of silica which ceases after approximately
15-30 minutes. Within the next -30 minutes a silica-rich film
is established, and finally a calcium phosphate film is pro-
duced at the silica-rich film-water interface (see Figures
20 and 21) .

The key to the variable corrosion resistance appears to
be associated with the calcium phosphate films. Initially,
they appear to be amorphous. Figure 24a contains an x-ray
diffraction pattern of the surface of the composition contain-
ing 6*^ ^2*^5 ^^hi*^^^ ^^*^ been in solution for 15 hours. Infrared
reflection spectra of this sample showed that a calcium phos-
phate film was present on the surface. The absence of any
diffraction peaks indicates that the film is completely

amorphous. However, it is possible that some crystalline
material is present but not in sufficient quantity to pro-
duce peaks. A diffraction pattern of the same composition
after 100 hours in solution shows peaks beginning to appear
(Figure 24b). Figure 24c is a diffraction pattern of the
glass containing 6 "o PtO;. which had been in solution for 1,500
hours. The d spacings obtained from the film show reasonable
agreement with the d spacings of carbonate hydroxyapatite
(dahllite). The values are compared in Table 2. There is
one discrepancy in the relative intensities and that is for
the 3.402 d value. It is the sharpest peak and has the high-
est intensity for the calcium phosphate film, whereas it has
a relative intensity of 70 for dahllite. This effect could
be accounted for if growth occurred along a preferential
direction. Figure 25 contains a diffraction pattern of the
calcium phosphate film on the surface of the glass containing
boron which has been in solution for 1,500 hours. Again
there is reasonably good agreement between its d spacings and
those of dahllite. The relative intensities are also in good
agreement .

Referring to Figure 23, the similarity between the infra-
red reflection spectrum of the reagent grade hydroxyapatite
and the spectrum of the glass containing boron which had been
in solution for 1,500 hours takes on added significance.
Considering the x-ray diffraction patterns, the infrared
reflection spectra and the energy dispersive analysis which
shows calcium and phosphorus to be the main components on the

Table 2

d-Spacings Obtained from Corrosion Films

on 45S-6% P2O5 and 45B5S5 Glasses Corroded

for 1,500 [Irs. Corresponding d-Spacings

of Dahllite Are Included.

Dahllite

. 120

.402

. 76 8

.687

,607

,232

931

834

721

4SS-6% PoOr
1,500 Ilrs in Sol

3.411
2 . 769
2.688
2.619
2.268
1.939
1. 832
1. 717

45B5S5
1,500 Hrs in Sol

.411

. 777

.697

,619

,257

,931

839

713

surface after 1,500 hours in solution (see Figure 11), it
would indicate that the crystalline calcium phosphate mate-
rial which forms contains a considerable amount of hydroxyapa-
tite. It has been stated by Korber and Tromel [52] that in
the system CaO-P^O^, hydroxy apatite will form at temperatures
up to 1050°C if water is not carefully excluded.

It should be pointed out that the most synthetic calcium
phosphate precipitates form nons toichiometric crystal com-
pounds with numerous possible substitutions existing, i.e.,
sodium for calcium, carbonate for phosphate, fluorine for
hydroxyls , water for hydroxyls. McConnell [53] has stated that
unless special precautions are taken it is practically impos-
sible to obtain apatite crystals which do not contain carbon-
ate groups. Furthermore, he suggests that carbonate substitu-
tion for phosphate groups can produce distortion in the hexa-
gonal apatite structure which can lead to line splitting in
diffraction patterns.

It thus seems likely that the calcium phosphate film
which forms at the silica-rich film-water interface of the
glasses containing phosphorus is indeed hydroxyapatite .
However, it almost surely deviates from s toichiometry due to
substitution of carbonate, sodium and possibly silicon.

One explanation for the significant difference between
the protectiveness of the calcium phosphate film of the glass
containing fluorine and all of the other compositions is that
the fluorine substitutes for the hydroxyl ions in the apatite
structure. It has been reported that if water containing

trace amounts of fluorine is brought into contact with hydroxy-
apatite, fluorapatite will form as an insoluble product [54],
Another source [55] has stated that in aqueous systems con-
taining trace amounts of fluorine, fluorapatite is the most
stable calcium phosphate compound. Referring to Figure 17, it
can be seen that there is a drastic drop in the phosphorus
level in solution between 10 and 100 hours for the glass con-
taining fluorine. The level of calcium released into solution
is also significantly lower after 100 hours for the glass con-
taining fluorine, when compared to the data for all other
glasses examined (see Figure 16).

The main influence of boron is an acceleration of the
initial attack of the glass network. Figure 14 illustrates
that even though the glass containing boron has the least
amount of silica in the bulk composition, more silica is
released into solution than is released from the glass con-
taining 6% PoOr or the glass with fluorine. This effect is
thought to be due to a weakening of the three-dimensional
silica network due to the presence of the boron atoms. Boron
can exhibit either three-fold or four-fold coordination. It
has been reported [56] that at high temperatures, boron pres-
ent in borosilicate glasses exhibits three-fold coordinati-
which changes to four-fold at lower temperatures. However,
during the cooling process there is not sufficient time for
complete reordering and some of the boron remains in three-
fold coordination. It is the presence of the boron atoms
with three-fold coordination which produce weak regions in

.on

the glass network. Aqueous solutions attack these areas,
releasing substantial amounts of boron and sodium.

A similar type of behavior could account for the observed
surface reactions of the glass containing boron. The pres-
ence of three-fold coordinated boron atoms lead to an accel-
erated release of sodium and boron atoms. This would pro-
duce a more rapid overriding of a buffered solution which
has been observed (see Figure 27). Release of silica would
also be accelerated due to the increased basicity of the solu-
tion. The data in Figure 18 substantiate this hypothesis.
The addition of boron to the glass containing 6°6 P-^O results
in an increase in the initial alpha values , which is a sign
that the extent of total dissolution is increasing. It
should be noted that this event is only temporary as a silica-
rich film is established within 1 hour. Tlie epsilon curve of
Figure 19 shows an increase in magnitude of e for the glass
containing boron which is greater than the glass containing
6% PyOr, indicating there is more silica available for film
formation .

Conclusions

In summary, the following facts have been established:
(1) The glass containing 0 -d Pt'^c forms a silica-rich
film which protects the glass throughout 1,500 hours of expo-
sure .

CMO

o\°

\r>

■p

■H

■M

■H

■P tJ

■H

■Xi

•H

•

cr o

•H

■P

•H

t/)

4- (D :=!

m ^ o

^-> cu

<-H

C CJ'

O nj

in "x!

W CD

Oj p

;=!

rH (U

I— 1

bcqn

O M-(

■H 3

1— 1

^ -£3

3.
bO
•H

o
d

0)00 CO

4 • ■

tf>

o
d

(2) The glasses containing phosphorus also form silica-
rich films. However, in the case of the glasses containing
6 and 12% phosphorus, the silica-rich film formation is pre-
ceded by a short period (15-30 minutes) of selective silica
attack. •

(3) After the silica-rich film formation, the phosphorus
containing glasses form a calcium phosphate film at the
silica film-water interface. The rate of formation of the
calcium phosphate film is accelerated as the amount of phos-
phorus in the bulk glass composition is increased.

(4) Although the calcium phosphate film appears to be
amorphous initially, it crystallizes with time into an apa-
tite structure.

(5) The calcium phosphate film is more effective than
the silica-rich film in isolating the glass from its aqueous
environment.

(6) The addition of fluorine to the glass containing
6% PoOr significantly increases the resistance of the glass
to aqueous attack.

(J) The addition of boron to the glass containing 6%
PpO^ accelerates the initial dissolution process in an aqueous
solution .

CHAPTER III

AUGER SPECTROSCOPIC ANALYSIS OF
BIOGLASS CORROSION FILMS

Introduction

Auger electron spectroscopy has been employed to further
characterize the corrosion films which form on a series of
bioglasses. An investigation by Clark and Hench [48] has
established that when exposed to an aqueous environment, a
silica-rich film forms on the glasses within two hours. A
second film composed primarily of calcium and phosphate is
produced at the silica film-water interface. This second
film is produced only when phosphorus is contained in the
glass composition and the rate of formation is related to the
amount of phosphorus in the bulk glass. IRRS, EDXA, and X-ray
diffraction confirmed that the film crystallized into an apa-
tite structure with time. Auger electron spectroscopy has
been utilized to obtain detailed chemical profiles of the
corrosion films in hopes of elucidating the mechanism of film
formation.

Theory

The technique involves bombarding the sample surface with
a beam of monoenergetic electrons. A series of interactions

leads to the release of electrons which were contained in the
electronic structure of the surface atoms. Figure 28 illus-
trates such a series of interactions. Impinging electrons
from the beam create a vacancy in the K shell. An electron
from one L shell then cascades back into the empty slot in
the K shell. In the process, sufficient energy is available
for the ejection of an electron from another L level. This
process is termed an Auger transition and the electron with
an energy characteristic of the atom from which it was
elected is called an Auger electron. The Auger electrons
produce peaks in the secondary electron energy spectrum and
thus by monitoring the energy distribution due to Auger elec-
trons, it is possible to identify the atoms producing them.
In actual practice, the derivative of the energy spectrum is
taken, which enhances the Auger peaks and suppresses the
background present in the secondary electron distribution
[57]. Due to a short mean free path, Auger electrons have a
maximum escape depth of 50 A, making this a truly surface
sensitive process. In addition to atom identification, it is
possible to relate the amplitude of the Auger peaks to the
concentration of the atoms producing them.

A complementary process of Argon ion bombardment removes
surface atoms a layer at a time. By simultaneously ion mil-
ling the surface and measuring Auger spectra it is possible
to obtain a chemical profile of the structure.

The raw data directly observed are the changes in peak
height with ion milling time. In order to obtain quantitative

Figure 28. X-ray energy level diagram depicting a
KL-.L„ Auger transition.

AUGER DE-EXCITATION

KL^L2 ^^iV Electron

lence/ Band

initial
ionization

information about the amount of atoms present at the surface,
the differences in Auger transition probabilities for differ-
ent atoms must be considered. Factors contributing to these
differences are the influence of the environment on an atom's
electronic structure as well as the distribution of atoms
within the volume of material producing the detected Auger
electrons.

To overcome this problem, sensitivity factors were deter-
mined by a recently developed process [58], These factors
normalize the Auger peaks, enabling one to make a quantitative
comparison of one component with respect to another. The
sensitivity factors were obtained by analyzing Auger spectra
of uncorroded glasses which had been ion milled for long
periods of time to expose the bulk structure, and comparing
these data to the known glass composition. Modifying the raw
data with the sensitivity factors allows one to obtain a mea-
sure of relative atomic percent versus ion milling time.

By assuming that the cations are present as specific com-
pounds with oxygen, i.e., SiO^ , CaO, P^^"; ' ^^® relative atomic
percent data can be altered to provide a measure of mole per-
cent versus ion milling time. There was usually an excess of
oxygen near the surface which was unaccounted for. The extra
oxygen atoms are probably associated with hydrogen atoms
(which cannot be detected with AES) as water molecules.
Although approximations are involved in determining the amount
of species present, the observed changes in peak height with
ion milling time correspond to an increase or decrease in the

amount of species at the surface and are unaffected by the
approximations.

Experimental Procedure

The four glass compositions selected for investigation
are listed in Table 3. The glasses were prepared from reagent
grade sodium carbonate, reagent grade calcium carbonate,
reagent grade phosphorus pentoxide, and 5 ym silica. Pre-
mixed batches were melted in covered Pt crucibles in a tem-
perature range of 1250 to 1350°C for 24 hours. Samples were
cast in a steel mold and annealed at 450°C for 4 to 6 hours.

Bulk samples of each composition were prepared by wet
grinding with 180, 320, and 600 grit silicon carbide paper.
After a final dry grinding with 600 grit silicon carbide
paper, samples were immersed in 200 ml of aqueous solution
buffered at a pH of 7.4 (trishydroxymethyl aminomethane
buffer). Temperature was maintained at 37°C, and all sample
solutions were maintained in a static state. Samples of each
of the four compositions were immersed in buffered aqueous
solution for one hour. In addition, samples of the glass
containing 6% ^2'^5 ^^^^ exposed to the buffered aqueous solu-
tion for 10, 20, 30, 40, 50, and 60 minutes.

The samples were placed in a stainless steel vacuum
chamber maintained at a background pressure of 1 x 10" Torr.
To prevent destruction of the corrosion films, the beam cur-
rent was held at a low value (5-10 ya) and was slightly

• Table 3

Bioglass Compositions Selected for
Auger Spectroscopic Analysis

45S-05„ 1

-2°5

Wt.''6

SiO,

24.

5 wt.

, % CaO

30.

5 wt.

, % Na20

45S-6"6 ]

^2^5

wt . %

SiO^

24,

5 wt

. -6 CaO

24.

5 wt

.% Na20

6 wt."o P2O5

45S-5I P-,0

45 wt

SiO

2 4.5 wt. % CaO

2 7.5 wt.% Na20

3 wt. % ?^0^

45S-12°6 \\0,

wt. %

SiO,

24.

.5 wt.

, % CaO

18,

,5 wt,

, ?6 Na20

wt. %

P2O5

100

defocused. Previous attempts to obtain spectra with a beam
current of 75-100 Ma resulted in complete degradation of the
films. The beam energy was 3 KV for the series of samples
corroded for one hour and 2 KV for the 10-60 minute exposures
of the glass containing 61 ^y^c- The angle of incidence of
the electron beam was kept at 45° to prevent unstable charg-
ing on the surface. The energies of the emitted Auger elec-
trons \Nrere measured with a cylindrical mirror electron analy-
zer.

Ion bombardment of the sample surface with 2 KV Argon
ions was employed to remove the outermost atoms. As discussed
in the previous section, the concurrent use of milling and AES
produces a chemical profile of the corrosion films.

Profiles were determined for each of the four composi-
tions corroded for one hour. Two silicon peaks can be seen
in the Auger spectra of Figure 30. It was observed that the
low energy silicon peak (78 eV) changed shape as the sample
was ion milled. The correlation between peak size and atom
concentration does not hold if the peak shape varies. As a
result, the high energy silicon peak (1,630 eV) was measured
for the silicon profiles.

A recording profilometer with a sensitivity of .02 ym was
employed to calibrate the ion milling rate. Figure 29 con-
tains the type of plot generated by the profilometer. Using
the value obtained and assuming a uniform milling rate, cal-
culations were made to convert ion milling time to depth,
yielding an estimate of the corrosion film thickness.

rt^

rt ■

O Xi

CD '

•H

i-<

P-

CxO-M

•H

1-H

-d

+->

•0

f-i

M-l

f-1

ct3

•H

■p

!-i

Ph 0

4->

O tD

•H

■P

U-i

I— t

• H

<-M

P^ o

X ^H

■M

102

■D

O
U

a
Q

■o

a
Q

103

Ion milling was not employed on the series of samples
corroded at ten-minute intervals, as only Auger spectra o£
the surface were taken. An attempt was made to measure a
layer as thin as possible. Since the electrons wliich produce
the low energy silicon peak have an escape depth (^8 A) about

one-fourth that of the high energy peak (^30 A), the magni-
tude of the low energy peak was monitored. The lower beam
energy (2 KV] was used for these samples to minimize the
thickness of the detected volume and to prevent radiation
damage which can lead to splitting of the low energy silicon
peak .

Results

Figure 30 shows Auger spectra obtained at three differ-
ent ion milling times for the glass containing 6% ^i^z which
was corroded for one hour. The location of the peaks on the
abscissa enables one to identify the atoms producing them. As
was discussed earlier, changes in peak height are caused by
an increase or decrease in the amount of element in the sur-
face layer. These changes are most pronounced for the phos-
phorus and calcium peaks in Figure 30. Plotting the peak
magnitudes versus ion milling time produces a chemical profile
as is seen in Figure 31.

Features of importance are the buildup of phosphorus and
calcium at the surface, followed by a region in which the
oxygen, calcium, and phosphorus levels fall off drastically,

Figure 30. Typical Auger spectra for three depths of
ion milling of a 45S-6% P2O5 bioglass
corroded one hour at 37°C and pH = 7.4.

105

.X4

t=3min

dN(E)
dE

Bulk

Glati

1000

2000

Electron Energy, eV

Figure 31. Corrosion film profile produced by
plotting peak magnitudes versus ion
milling time for a 45S-6I P2O5 bioglass
corroded one hour at 37°C and pH = 7.4.

107

Corrosion-Film l^rofile

Ion Milling Time,min.

108

and finally a buildup in the oxygen, calcium and phosphorus
levels to values characteristic of the uncorroded glass.
Modifying the raw data with the sensitivity factors and con-
verting ion milling time to depth of milling produces a semi-
quantitative chemical profile of the corrosion film. Figure
32 illustrates the results of this process for the glass con-
taining 6% PoO;^ which was corroded for one hour. When com-
paring Figures 31 and 32 it is important to note that,
although the magnitudes of the elements have been altered
with respect to each other, the changes observed with milling
time or depth of milling have been maintained. Ion milling
through the corrosion films into the bulk glass was achieved
only for the glass containing 6% ^o'^c. (Figure 32). The thick-
ness of the silica-rich film is on tlae order of 2.0-2.5 ym,
while the outermost film rich in calcium and phosphorus is
only 0,5 ym thick.

Figure 33 is the result of converting atomic percent of
surface species to mole percent. This final adjustment of
the data can only be applied for the corrosion films, because
the sodium has been leached out. Since the bulk glass con-
tains a significant amount of sodium which is not detected
with AES, it would be very difficult to accurately compute
mole percentages in the region of uncorroded glass.

The absence of sodium which will be seen in all of the
chemical profiles is not unexpected. It has been reported by
several investigators that leaching of alkali is one of the
initial steps in the corrosion of silicate glasses in aqueous

O 4-1

u a

u o

•H ^

O dJ

oj o

• H (D

Td o

<U 't-<

I/) u

m o

0 u

P. t/)

X (/)

CD nj

t— i

O bJj

.-H O

CNl .

t— I CU '^
CD

•H \o
(U oo

X Lo a:

110

03
CO
<

-J
3
CQ
4k

-;r

LU
CO

LL.

o
cr

-J

2'x

1 O
CO cc
O

K- C3

ct!

fH U

O !

t~-

4-1

+->

!-i

p. o

(^^

•H

cyi

r-i

t/)

!-i

P-

r— 1

• H

•H

CNl .

I— 1

Ph 'd-

o\o I^

• H

1 II

LTl "X

■* p.

112

(/}
IT)

0)
U
(0

3
C/)

113

solution [59]. In spite of these findings, one factor which
had to be considered is the difficulty in detecting the pres-
ence of sodium with AES. Previous work [60] has suggested
that electrostatic conditions produced by electron bombard-
ment cause the extremely mobile sodium atoms to migrate out
of the area of analysis. Another possibility is that the
Argon ion milling process preferentially removes the sodium.
For these reasons two samples of the glass containing 6% PoOr
were examined with Electron Spectroscopy for Chemical Analysis
(ESCA) . This technique involves bombarding the surface with
a beam of x-rays and detecting the ejected photoelectrons .
Information on composition and chemical binding can be ob-
tained from this process. By examining a sample which had
been corroded for one hour along with an uncorroded sample,
the absence of sodium in the corrosion films was shown to be
real and not an artifact of AES.' Figure 34 compares the
sodium, phosphorus, and silicon peaks for the uncorroded and
corroded samples using ESCA or photoelectron spectroscopy.
Chemical profiles of the glasses containing 0, 3, and
121 PoOq are shown in Figures 35, 36, and 37. They were
determined by the same technique previously described for the
glass containing 6% ^2*^5 " Note in Figure 36 that the F2*^5
level is intensified near the surface but the CaO level remains
relatively constant and even drops within .05 ym of the sur-
face. Immediately underlying the phosphorus -enriched region
is a silica-rich film. The profiles of the glasses contain-
ing 6 and 121 P„0^ (Figures 33 and 37) both contain areas of

r-l O ro

^1 aJ

"+^ U ^H

nJ PL, o
, (/) X

O 0 cu

-P ^ !-i
U 4-J O
0) •M <4-l
P. 12
CO 13
to <u
c! (/) TJ
O 03 o
>-i t— I 5h

+-> w; ^
U O o

CJ 'H (J

O CO

o Lo to
■M O 03

O (NJt— I

^ P-, to

D. O

o\a .H

^ VO ^ .
O I -^

C/D LO .

C Lo O t---

O ^ (NO

to iDh II

■H TJ

03 "Tj \0 p^
C-i 03 I

o jD LO fi;
U 03 >* 03

115

u
a>

c
o

■*■>

u
_aj

■«->
o

■5

Figure 35. Chemical profile expressed in mole percent
of a 45S-O1; P2O5 bioglass corroded one
hour at 37°C and pH = 7. 4.

Relative

Mole

Percent

117

45S-0% P2O5

SiO.

'2V

CaO

1.0 1.2

Depth From Surface [** mj

Figure 36. Chemical profile expressed in mole percent
of a 45S-3"o P2O5 bioglass corroded one
hour at 37°C and pH = 7.4.

45S-3% P2O5

119

SiO

2^L

Relative
Mole 40

Percent

J I L

III III I L

Depth From Surface L^^nJ

r-i •

CO •

Pi rt
0

U U

(U 4->

p; o

Tj 0)

in o
(/I

f-< 0)

X O

0) f-i

CD O

rH O

O CO

U rt

tH o

Oj .H
O^
■ H

o o

121

Q)
O
CO

E
o

-^-i

•*-' ^

03 0)

— -1 ■""

0 O

DC^

122

P 0 and CaO enrichment near the surface with silica-rich
regions below them. The calcium-phosphorus - rich film of the
glass containing 12% ^7*^1; ^^ larger than that of the glass
containing 6°s Po^S' ^

Figure 38 presents the raw data from the Auger spectra
of the sample corroded at 10-minute intervals. The silicon
peak was not detected after 20 minutes of corrosion, whereas
the Ca and P levels remained above their uncorroded values
for the entire 60 minutes.

Discussion

The profiles of Figures 33 and 35-37 clearly show the
existence of silica-rich films for all four glasses. Further-
more, as the phosphorus content of the glass increases, a
calcium phosphate film of increasing thickness overlaps the
silica-rich film.

The profile of Figure 36 indicates that there is a mini-
mum phosphorus level which must be reached near the surface
before the calcium begins to buildup. This level should
depend on the phosphorus content of the uncorroded glass as
well as the length of the corrosion treatment. In the case
of the glass containing 3% PoOr there is not a sufficient
amount of P^Op to initiate the calcium buildup within one
hour. Previous work [48] has shown that the calcium phosphate
film will form at the surface of the glass containing 31 PoOq
with time.

00 X!

p; o

U CO

>+H

+->

bog

•H

r!^

■H

(/)

ft O

bO O

<;4H

■M

■H

•P

•H

(fl

tfl

124

c^ 5 " <o ^ S

(s;!un-qje)}qB!8H >je8d -o;- )|ead JeBnv

125

The results shown in Figure 38 point to the formation
of a thin surface layer (10-15 X) rich in calcium and phos-
phorus. This layer is established within 20 minutes of cor-
rosion time during which silicon is preferentially removed.
This thin calcium phosphorus film is present on the surface
during the time when the silica-rich layer is forming beneath
it. In fact, the change from selective silica leaching to
the formation of the silica-rich film coincides with the time
when the thin calcium phosphorus layer has formed. The evi-
dence indicates that the thin calcium phosphorus film prevents
further preferential silica removal, but allows the other com-
ponents of the bulk glass composition to be continually
leached. Once a sufficient amount of calcium and phosphate
has been leached into solution the thin calcium phosphate
film serves as a nucleation site for the formation of the
calcium phosphate layer which eventually crystallizes into
an apatite structure. One point which is not clear is whether
the silica-rich film formation which is produced only after
the thin calcium phosphate layer has formed, plays a role in
the growth and crystallization of the calcium phosphate film.

These results are in complete agreement with those pre-
sented in the previous chapter, and add some additional in-
sight into the sequence of steps involved in the corrosion
process. The following series of reactions are now known to
occur when the glass containing 6% P^Or is placed in an
aqueous environment buffered at a pH of 7.4 and maintained at
37°C:

126

(1) Within the first 15-30 minutes silica is preferen-
tially leached.

(2) During this same time a thin layer rich in calcium
and phosphorus is established at the surface (10-15 A thick).

(3) Once the thin calcium-phosphorus layer has formed,
the preferential silica attack ceases and a silica-rich
layer, 2-3 ym thick, is formed within one hour.

(4) After the silica-rich layer has formed and there is
sufficient calcium and phosphate in solution the thin calcium
phosphate layer begins to grow. It was reported in the pre-
vious chapter that the calcium phosphate film formed at the
silica-rich film-water interface. The techniques which were
used to characterize the corrosion process were not suffi-
ciently sensitive to detect the presence of the thin calcium
phosphate film which forms initially. Only through the use
of Auger Electron Spectroscopy w.as the detection of this thin
film possible.

(5) The calcium phosphate film crystallizes into an apa-
tite structure with time.

This sequence of steps can be explained through the
following mechanism. Phosphorus is a network former which
exists in four-fold coordination. Due to the +5 charge of
the phosphorus atom one of the phosphorus oxygen bonds must
exist as a double bond. McMillan has stated that the exis-
tence of the double bond in the phosphorus tetrahedra leads
to conditions which promote separation of the phosphate groups
from the silica network. Furthermore, he states that it would

127

be probable for the P^O- to be associated with alkali or
alkaline earth oxides present in the glass composition [61].
Tomozawa has reported that ^n^r additions to sodium silicate
and lithium silicate glasses promote phase separation by
widening the immis cibility boundary and accelerating the
kinetics [62]. The influence on the immiscibility boundary

is related to the relative magnitude of the cationic field

+ 4 +5
strength with respect to that of Si . P , which has a

2 2

larger cationic field strength [Z/a (P) = 1.91, Z/a (Si) =

1.58] than Si, was shown to promote phase separation while

+ 4 +4
Ti and Zr , which have smaller field strengths than Si,

were both found to suppress phase separation in the soda
silica system [62]. Although this effect was only substanti-
ated for simple binary systems, Tomozawa felt that the chances
for this relation to hold in more complex silicate glasses
were quite possible.

Based on these findings, it seems likely that the PoOr
additions to the soda- lime -sili ca glass promote a tendency
towards phase separation and, in the process, disrupt the
silicate phase by tying up some of the calcium from the
ternary phase. This would have the effect of reducing the
corrosion resistance of the silicate phase as calcium addi-
tions have been shown to increase the durability of soda
silicate glasses [63]. Evidence for phase separation of the
glass containing 61 P^Or was presented by Hench et_ a_l. [27].
A scanning electron micrograph showed a second phase which
existed as droplets, and was thought to be tlie phosphorus-
rich phase.

128

The net result of this situation would be that the soda
silica phase would be preferentially attacked by the alkaline
aqueous solution. This effect would be enhanced as additional
phosphorus tied up an increasing amount of calcium. As the
silicate phase is attacked, a surface layer rich in calcium
and phosphate would be produced which would then shield the
remaining silicate phase from further network breakdown.
Diffusion of Ca and Na into solution would still be pos-
sible, thus leading to the formation of a silica-rich layer
under the calcium phosphate layer, Wien sufficient phosphate
and calcium have been released into solution, a reaction
between these two components and water would cause the calcium
phosphate layer to grow and eventually crystallize into the
apatite structure.

Reactions of this type have been cited in the literature.
Weyl has postulated that phosphate opacification in soda-lime
silica glasses is produced by the formation of apatite crys-
tals [64]. The crystal formation occurs when calcium and
phosphorus react with water in the glass melt. It was also
reported that the reaction of calcium and phosphorus with
moisture in the atmosphere can lead to apatite formation at
the glass surface, producing surface roughness and brittleness
of the phosphate opacified glass [64].

129

Conclusions

1. Chemical profiles have been measured with Auger
Electron Spectroscopy and ion beam milling which define the
silica-rich and calcium phosphate corrosion layers.

2. IVhen the bioglasses are corroded under identical
conditions, the thickness of the calcium phosphate layer
increases as the phosphorus content of the bullc glass compo-
sition increases.

3. There is a minimum phosphorus level which must be
reached near the surface before the calcium begins to build up

4. A thin surface layer ("^10-15 A) rich in calcium and
phosphate forms during the initial 15 minutes of corrosion of
the 45S-6°6 PoO,. bioglass. The data indicate that the thin
calcium phosphate layer initiates the formation of the silica-
rich layer and serves as the nucleation site for growth of
the calcium phosphate layer once sufficient calcium and phos-
phorus have been leached into solution.

CHAPTER IV

THE INFLUENCE OF SURFACE CHEMISTRY
ON IMPLANT INTERFACE HISTOLOGY

Introduction

A series of bioglasses with variable phosphorus content
have been implanted in rat femurs and their response has been
related to the previously defined invitro chemical behavior.
In previous invivo studies bioglass implants were treated in
a conditioning solution prior to implantation. The influence
of this process on the structure of the bioglass surface has
been investigated. Infrared reflection spectroscopy and
scanning electron microscopy with energy dispersive x-ray
analysis have been utilized to characterize the surface
changes produced by the conditioning solution. Light micros-
copy and transmission electron microscopy were employed to
examine histological sections of the glass-bone tissue inter-
face . . .

Experimental Procedure

Bioglass compositions 1-4 (see Table 4) were selected to
study the influence of phosphorus additions on the behavior
of bioglass implants. Samples were prepared under identical
conditions employed for the invitro studies (see page 11).

130

Table 4

Bioglass Compositions Implanted
in Rat Tibiae

131

45S-0?6 F\0^

45 wt. % SiO
2 4.5 wt
30.5 wt

2
CaO

Na20

45S-6°5 P-,0^

45 wt. % SiO
24. 5 wt
2 4.5 wt
6 wt

2
CaO

Na20

^2^5

45S-3I P,0,

wt.%

Si02

wt ,

.^0 CaO

wt,

.% Na 0

3 wt.

.i P^Oj

5S-12^o P^O^

45 wt.%
24.5 wt
18. 5 wt

SiO,

CaO
Na20

12 wt.% P2OP

132

One series containing the glasses with 0 ^o and 61 V^O^ was gas
sterilized and soaked in conditioning solution for 72 hours.
Samples of each of these two ^compositions were subjected to
IRRS and SEM analysis after gas sterilization, 24, 48 and 72
hours in the conditioning solution.

A second series was gas sterilized and soaked in condi-
tioning solution for 72 hours before implantation. The con-
ditioning solution contains Eagles MEM (Minimum Essential
Medium) and Earle's balanced salt solution, 10°o fetal calf
serum, and 10''o newborn calf serum [65].

Samples of bioglass 5 mm by 5 mm by 1 mm were placed in
defects produced in the metaphysis of the tibia just distal
to the epyphyseal plate of Sprague Dawley male rats. The
limbs were not immobilized and the animals were s acrif iced at
3 and 8 weeks.

The tibiae were dissected clean of all soft tissues and
the area of bone surrounding the bioglass was cut into 1 mm
thick sections with bone on either side of the glass. The
slices of bone and glass were immediately placed in cold
cacodylate buffered gluteraldehyde , fixed for two hours and
then washed with fresh cold buffer. The tissue sections were
then placed in 2% osmium tetraoxide collidine buffered at a
dH of 7.4 and fixed for an additional hour. After a final
wash with additional buffer, the blocks were dehydrated in
■graded alcohols and embedded in Epon 812. Sections were pre-
pared on a Porter-Blum MT-2 ultra microtome. Thick sections
(1 ym) were cut with glass knives, stained with Richardson's

133

methylene blue azure II stain and examined with a light
microscope. A diamond knife was used to cut thin sections
(600 A thick) . Prior to TEM analysis the thin sections were
stained with saturated fresh alcoholic uranyl acetate and
lead citrate [66]. All TEM sections were examined with a
Hitachi HU IIC electron microscope.

Results and Discussion

Table 5 illustrates the time dependent change in the
surface ratios of Si/Ca and Ca/P for the glasses containing
0 and 6% P7O1. during the conditioning treatment. These ratios
were obtained with a scanning electron microscope equipped
with an energy dispersive x-ray analysis system. X-rays pro-
duced as a result of the electron beam striking the sample
surface are detected and identified according to their energy.
As different atoms have their own discrete energies, the
resulting spectrum can be used to determine the atoms present
on the surface. For a more detailed discussion refer to
page 16. The gas sterilization treatment produces little or
no change for either composition. After 24 hours in the solu-
tion there is a significant increase in the ratio of Si/Ca
for both glasses. In addition, the Ca/P ratio for the glass
containing 6-6 PoO^ drops drastically. These trends continue
through 48 hours. Between 48 and 72 hours of exposure the
ratio of Si/Ca remains constant for the glass containing 0%
P^O^.. During the same period, the ratio of Si/Ca has dropped

134

Table 5

Energy Dispersive X-ray Analysis of the Effect
of Conditioning Treatment on Bioglass Surfaces

Condition of
Sample

45S-0^o P^
Si/Ca

°5

45S-6I
Si/Ca

-i^205
Ca/P

Freshly abraded

.910

.912

6.2

Gas sterilized

.912

,.912

6.1

Gas sterilized +

24 hrs in cond. sol.

2,03

1.43

2.38

Gas sterilized +

48 hrs in cond. sol.

2.41

1.75

1.97

Gas sterilized +

72 hrs in cond. sol,

2.40

0. 80

135

from 1.75 to 0.80 for the glass containing 6% PpOr , while the
ratio of Ca/P continued to drop to a value of 1.89.

Figures 39 and 40 show infrared reflection spectra of
the glasses containing 0 and 6% PoOr ^'^ selected intervals
during the conditioning treatment. The spectra of the glass
with 01 PoOp (Figure 39) reveal the formation of a silica-rich
surface layer which is present at the conclusion of the 72-
hour conditioning treatment. Little change is noted between
the freshly abraded spectrum and the spectrum of the gas
sterilized sample. After 24 hours in solution, there is
selective attack of the silicon-nonbridging oxygen peak at
840 cm . The silicon- oxygen- si licon stretching (S) and
rocking (R) peaks, located at 955 and 500 cm respectively,
begin to sharpen, increase in intensity and shift towards the
location of the S and R peaks of vitreous silica. These
changes continue to occur through 48 hours of exposure. The
curve after 72 hours exhibits no additional changes indicating
a stable condition has been achieved. The data obtained with
infrared reflection spectroscopy and the x-ray system of the
scanning electron microscope both point to the formation of
a silica-rich surface layer on the glass with 0 "o Pt'^'s* This
glass exhibited the same type of behavior in the invitro
studies presented in Chapters II and III.

The IR spectra of the glass containing 6-6 P2O2 (see
Figure 40) are similar to the spectra of the glass with 01.
P„Oj^ through 24 hours of exposure. That is, little change
can be noted between the freshly abraded and gas sterilized

Figure 39. Changes in infrared reflection spectrum
of 45S-0% ^z'-'s gl^ss during conditioning
treatment.

137

1200 1000 800 600

WAVENUMBER (CM-1)

400

Figure 40. Changes in infrared reflection spectrum
of 45S-6I P2O5 glass during conditioning
treatment.

133

1200

1000 800 600

WAVENUMBER (CM-1)

400

140

spectra. After 24 hours in solution, selective attack of the
silicon-nonbridging oxygen peak occurs, and the peaks associ-
ated with the silicon-oxygen-silicon bonds exhibit changes in
shape and location which indicate the concentration of silica
is increasing on the surface. The 48-hour spectrum of Figure
40 contains the S and R peaks of silica but their intensities
have dropped to values below their level at 24 hours. This
trend continues with the 72-hour spectrum. Behavior of this
type was also observed in the invitro studies on the glass
containing 6% ^y^c- After the silica-rich layer is formed,
the calcium phosphate layer begins to grow. Apparently the
rate of these reactions is slower in the conditioning solution
and there is not a sufficient amount of calcium phosphate on
the surface at 72 hours to produce the infrared reflection
spectrum seen invitro. However, the data obtained with the
x-ray analysis shows the ratio of Ca/P is becoming smaller
with time, while the ratio of Si/Ca drops significantly from
its 48-hour level, indicating an increase in the calcium and
phosphorus concentration on the surface.

These observations clearly show that the surface struc-
ture of a bioglass implant is drastically influenced by the
conditioning treatment and interpretation of the histological
results of conditioned samples should take these changes into
consideration.

Small pieces of glass implant were attached to bone in
almost every case, but a distinct variation ivas observed in
the tissue responses evoked by the different compositions

141

which had been conditioned prior to implantation.

Figure 41 is a transmission electron micrograph of a
45S-0% P^Or glass-bone interface at three weeks. The mate-
rial which exhibits the regular fracture pattern appears to
be the silica-rich corrosion film (CF) which forms on the
surface of the glass implant. The relative softness of the
corrosion layer compared to the glass produces the uniform
fracture pattern, with long non-branching fracture lines.
The corrosion film contains a tear which was probably produced
during the sectioning process. Close examination reveals that
a thin layer of the corrosion film (CF) remains attached to
bone (B) along the interface (I) , indicating the corrosion
film-bone interface has considerable strength. The elongated
cell (EC) in close proximity with bone has the appearance of
a normal endosteal cell on a resting bone surface and does
not appear to be actively engaged in laying down new bone.
Examination of thick sections containing the glass with 0%
P^O;- revealed a small number of viable osteocytes present in
newly formed bone and bone surfaces characterized by a lack
of active bone formation and very few active osteoblasts.

A 45S-3-6 P-pOr glass-bone interface at three weeks is
shown in Figure 42. Small pieces of implant are attached
along the surface. It should be pointed out that before sec-
tions are cut, the glass is chipped out of the block. If
this was not done it would be very difficult to cut sections
as glass knives are used and they would constantly break.
The presence of small pieces of glass attached to bone

Oj <D

t/) H

(/)

t— 1

rH •

^ — ,

M rt

•H

W o

o\° ^

"— 'o

O -H

LTi

1 -M

(D "

PJ 00

LO +-1

O V— '

^ nJ

-P

o •

0) C

+-> /— >

(U -H

-a ^

■P S

<D O

Hi (/)

XI -H

o ;s

+->

rTi o

sn rt

O 4-"

>^!h

■H fH rH rt

+J oj

-p

U rH

^ X

d P^ bO^

=5 S

•H

•^■H

4-1 t3

M-i ^

I/) +->

O O

•H OS

^'+H

,-^ bO

p.; Oj pL, .H

U t/1

Jh t/)

^— ' OJ

■1:3

O (D

U CD

I— ( (D

O 5

•H O

• H

Mh rt

6 CD

C ^^

a !-

O (D

O X

• H -P

fH 4->

tfl C

•p

O -H

O <U

0) C

f-1 0

rH O

O ^

w ^

U 4->

U ■

■;3

bO:

143

WO 0)

00 u

'^ -p
■p

Mh 0

O f-H
O ,— ^

•H U

+-> ^-^
+-> +->

I— I 1—1
Oh P-

e s

•H -H

Sh CO

(D -P
0

O
0

0 10
i^ P

^ c

P 03

Ph 0 X
O Cii o

u o

u 01

■ H Oj .

e i-i 0

P o

tXO

■ H CO O
-J &. P

145

/> /^

146

indicates that there is considerable strength associated with
the glass-bone interface because fracture occurs within the
glass implant rather than at the interface.

The mineralized bone adjacent to the implant interface
of Figure 42 contains several osteocytes and an area of
unmineralized osteoid. There is a layer of plump osteoblasts
which appear t'o be laying down new bone.

Figure 43 is a photomicrograph of a 45S-6% ^y^c. glass-
bone interface at three weeks. Large pieces of bioglass (G)
are intimately attached to bone (B) and several normal osteo-
cytes (0) are present in the mineralized area. There is a
well-defined layer of osteoblasts actively engaged in laying
down new bone (OF) and this front is separated from the
mineralized area by a transition zone of partially mineralized
osteoid. These features indicate that induction of normal
osteogenesis has been achieved. An electron micrograph of
the same section (Figure 44) shows the corrosion layer directly
attached to mineralized bone along the wavy interface I.

A 45S-12°6 P-jO^ glass-bone interface at three weeks is
shown in Figure 45. There is an absence of activity along
the ossification front with no evidence of osteoid and only
one osteoblast in the area. Figure 46 is a photomicrograph of
a 45S-12% P^Oq glass-bone interface at eight weeks. An impor-
tant feature to note is that the implant G has been separated
from the bone B by an interval containing a capillary C.
Electron microscopy of this section (Figure 47) reveals inter-
cellular crystallization (X) has been induced along the edges

u o a

'^ u o c

(U cd +-> oi

•H (U O I

QJ 03 L)

C -H nJ t/)

0 42 +-> (U
^ .H 4-> -P

1 -P 03 >.
t/1 U ,
(/) P >-, O
o3 03 1—1 0)

T-H Jh 0) P

W) P CO

C ^ O

O -H rH

CNl pi P Oj

ciH o p; 5h

•H -H 0)

o\= p >

^O 03 0) (D

I P !-i LO

00 c; 03

Ln 03 {/)

'^ t-l ^> p;

03 S^

03
P

CD Cti O

^ P rH

o3 oS o a

fn -HO

tiO in ^ ^
o ^

U (D O 0

O (D <D
p (D O .
O f-i 0 ^— ,

^ ,^ -H pq
CL, p cx^-^

CD 00

tiO

148

o m

• H " — '

o <u

>H o

o ^
o

<U 0)

^ N

(D 0)
O 03

U P-.

;3^

(Nl

P^Cn

f-<

cJP

CxOvO

S-H

•H

■^

^ ^

O Mh

f-l

w m

150

";"' ?^'*>("'' # •■■■■"■■■

■'^^;:^-\..^'''-'/^^V^

» . 'iS V ^■^'■

Figure 45. Light microscopy three weeks after implan-
tation of a 45S-12% glass. Glass (G) is
attached to bone (B) . There is an absence
of activity along the new bone surface
(OF). (1,800X)

152

Figure 46. Photomicrograph of a 45S-12I P2O5 glass-
bone interface eight weeks after implan-
tation. Glass implant (G) has been
separated from bone (B) by an interval
containing a capillary (C) . (1,800X)

154

Figure 47. Electron microscopy of capillary in

Figure 8. Note intercellular crystal-
lization (X) along edges of capillary,
(44,200X)

156

157

of the capillary. It can also be observed that part of the
corrosion film (CF) remained attached to the bone when the
interval containing the capillary separated the implant from
the bone.

Referring to Figure 46, note the unhealthy appearance
of the osteocytes (0), They have withdrawn from their lacunar
walls and the nuclei are pyknotic. There is also an absence
of new bone formation at the bone surface.

The invivo results of this study show that direct attach-
ment of glass to bone is achieved within three weeks for the
four compositions studied.

The invitro studies in Chapters I and II establish that
silica-rich corrosion films form on the surface of the bio-
glasses in a simulated physiologic environment. Furthermore,
the invitro results of this chapter show that the conditioning
treatment produces the same response.

Carlisle has reported that silicon-rich regions are
associated with active mineralization sites in young mice and
rats and, once mineralization has gone to completion, the
silicon content drops [67]. Recent invitro investigations by
Hench and Paschall [36] have shown that 45S-6'd PoOp glass
implants are bonded to bone by an amorphous cement- like layer,
probably comprised of SiO^ , CaO, and PoOi- , which serves as
the active site for collagen attachment followed by mineral-
ization.

In view of the findings of this study as well as those
in the literature, it seems likely that the silica-rich layer

158

serves as an induction site for osteoblasts to lay down the
organic intercellular substance of bone. This substance
contains collagen and mucopolysaccharides. Normally, miner-
alization would begin to occur §s soon as the organic inter-
cellular substance was secreted by the osteoblasts. The
exact mechanism of mineralization is not completely defined;
however, the concentration of Ca and PO^ ions in the area is
thought to play an important role [68].

The phosphorus content of the bioglasses may be the
important parameter which influences mineralization. The
buildup of calcium and phosphorus which occurs on the surface
of the silica-rich films could provide a source of ions for
mineralization. The results obtained indicate that, as the
phosphorus content of the glass increases from 0 through 6%
P 0 the appearance of the total ossification process becomes
increasingly healthy. In the cs^se of the glass containing 6%
P„0 the resulting situation is one of normal ossification.

The results obtained with the glass containing 121 P20^
suggest that there is an optimum phosphorus content which
should not be exceeded. The ectopic crystallization seen in
Figure 9 might well have been induced by an excessive amount
of phosphorus. Matthews et al . have reported that the addi-
tion of phosphates to a fixative, followed by incubation,
will result in apatite crystal formation [69]. Furthermore,
they reported that release of phosphate from cells which led
to the formation of an amorphous calcium phosphate was
prompted as a response to administered doses of thyrocalcitonin,

159-

In the case o£ a bioglass, a specific enzyme would not
be necessary to release large amounts of calcium and phos-
phorus as the response of the bioglass surface to body fluids
would accomplish the same end. If tlie calcium and phosphorus
released from the glass when combined with calcium and phos-
phorus present in the body fluids resulted in a critical
supersaturation , apatite crystal formation would result.

Conclus ions

Based upon the evidence obtained, the following theory
is proposed for implant materials design and selection:

An ideal implant material must have a dynamic surface
chemistry that induces histological changes at the implant
interface which would normally occur if tlie implant were not
present .

In the case of the bioglasses the optimal response is
elicited by a composition which has the ability to form a
silica-rich corrosion film and provide an adequate but not
excessive supply of ions to be incorporated in the minerali-
zation process. The glass containing 6-0 PoOr appears to be
the best candidate based upon the relatively short implanta-
tion times of this study.

CHAPTER V
CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK

The objectives of this study fall into two categories.
The first has been an effort to understand the influence of
compositional variations on the surface chemical behavior of
a series of bioglasses in a simulated physiologic environment,
and the relation of this behavior to that exhibited when
identical glasses are implanted in animals. The second objec-
tive has been an attempt by the author to bridge the gap
between the fields of materials science and the biological
sciences so that an intelligent and practical approach may
be developed for the selection of a material for potential
use as a prosthetic device. This has involved developing an
awareness of problems associated with the body's response to
prosthetic devices and some of the procedures which are em-
ployed to examine normal and abnormal responses to foreign
devices .

The results of Chapter II have shown that the glasses
investigated develop a corrosion layer or layers in response
to attack by an aqueous solution buffered at a pH of 7.4 and
maintained at 37°C. Sodium and calcium are preferentially
leached from the soda- lime-silica glass (45S-0I P^Op) , pro-
ducing a silica-rich film which serves as a buffer zone

160

• 161

protecting the remaining bulk glass from aqueous attack. As
phosphorus is added to the glass composition, a second film
is generated at the silica-rich film-water interface. The
second film is an amorphous calcium phosphate compound which
crystallizes to an apatite structure with time. Increasing
the phosphorus content of the glass reduces the time required
for the calcium phosphate film to form. Partial substitution
of B„0_ for SiO„ leads to weakening of the silicate network
and acceleration of the initial dissolution process. Fluorine
additions significantly enhance the resistance of the glass to
aqueous attack, probably by substituting for hydroxyl ions in
the apatite structure of the corrosion film.

The results of Chapter III confirm the observations of
Chapter II by providing chemical profiles of the corrosion
films which define the silica-rich layer and the calcium phos-
phate layer. The thickness of the calcium phosphate layer
was found to increase as the phosphorus content of the bulk
composition increased when glasses were corroded under iden-
tical conditions. The application of Auger spectroscopy and
ion beam milling to obtain detailed maps of compositional
changes over a depth of several micrometers has turned out
to be a valuable technique in characterizing the corrosion
behavior of the bioglasses. It should also be noted that
the results obtained with Auger spectroscopy have substan-
tiated the usefulness of the techniques employed in Chapter
II such as infrared reflection spectroscopy and ion solution
analysis, which infer rather than directly measure information

162

about the corrosion films and which are somewhat easier to
apply to a large number of samples.

Additional results in Chapter III point to the existence
of a thin surface layer (10-15 *) rich in calcium and phos-
phorus which forms during the initial 15 minutes of corrosion.
The observed sequence of events indicate that the thin cal-
cium phosphate layer initiates the formation of the silica-
rich layer and serves as the nucleation site for growth of
the calcium phosphate layer once sufficient calcium and phos-
phorus have been leached into solution.

Based upon the results of Chapters II and III a mechanism
which explains the formation of multiple corrosion layers
has been proposed (see page 126) which includes phase
separation induced by phosphorus. Future work should include
an investigation of the influence of phosphorus additions on
the micros tructure of the bioglasses. A new instrument
ideally suited for such a study is the scanning transmission
electron microscope [70] with supplemental attachments which
enable one to obtain elemental analysis and crystallographic
identification via electron diffraction on a very fine scale.

The invivo results of Chapter IV have demonstrated that
the four compositions (see Table 4) implanted all exhibited
direct attachment to bone. There was a wide variation in the
appearance of the tissue near the implant. Only the inter-
face of the glass containing 6 wt.% ^?^c, exhibited a healthy
zone of ossification characterized by numerous osteocytes in
close proximity to the glass, a layer of unmineralized

163

osteoid, and a layer of osteoblasts actively engaged in lay-
ing down new osteoid. The other three glasses exhibited a
lou density of viable osteocytes and an absence of an active
osteoid front. At 8 weeks this situation had degenerated
further for the 12 wt.% P^O glass. The osteocytes that were
present appeared to be dying and osteoblasts were not actively
producing new osteoid. In addition, the glass had been split
near the glass-bone interface. This area was filled by a
capillary containing several types of cells and electron
microscopy revealed an intercellular crystallization that
was apparently induced by the excess phosphorus.

The induction of normal bone growth was related to the
ability of a bioglass to form a silica-rich corrosion film
and provide an adequate but not excessive supply of ions to
be incorporated in the mineralization process. It has not
been established whether the undesirable results attributed
to an excess phosphorus concentration are related to the
amount of phosphorus present or an unbalance produced in the
ratio of Ca to P. This question could be answered by implant-
ing a series of bioglasses in which the phosphorus content
would be held constant while varying the Ca content. It
would be desirable to analyze the invitro corrosion behavior
of the same series employing the techniques discussed in
Chapters II and III.

The invivo results presented in this study have been
limited to some type of visual observation of the glass-bone
interface. The positive results obtained in the invitro

164

studies employing Auger spectroscopy to define the corrosion
profiles (see Chapter III) have opened up the possibility of
a similar analysis on glass-bone samples. If successful,
the results would provide a maj:^ of the change in atomic
composition from the glass through the attachment zone into
bone.

Re -examination of the EM grids containing glass-bone
sections with the scanning transmission electron microscope
described previously would allow one to achieve interfacial
compositional and crystallographic identification of the
interfacial zone of bonding.

The invitro results presented in Chapter IV describe the
effect of the conditioning treatment on the surface structure
of the bioglass implants. Corrosion layers similar to the
layers produced in the invitro studies of Chapters II and III
form on the implant surface.

It is important to know whether the conditioning treat-
ment is necessary to produce the observed invivo responses.
Possibly the body would produce the same structural changes
on the glass surface if unconditioned glasses were implanted.
In other words, how is the time sequence of events of the
interfacial reactions influenced by the conditioning treat-
ment? To answer this question it would be necessary to sub-
ject the four bioglass compositions employed in Chapter IV to
an identical implantation experiment eliminating the condi-
tioning treatment. The results might indicate that a critical
mixture of Si, Ca, and P ions on the surface is necessary for

165

the induction of bone growth. If this were the case, it
would produce new possibilities for materials for prosthetic
devices, such as ion impregnation of metals or ceramics with
the desired amounts of calcium, phosphorus, and silicon.

BIBLIOGRAPHY

1. Icart: J. de Med. ; Chir. et Phar. de Roux 44:170, 1775.

2. A. A. Zierold, Arch. Surg. , 9_, 365-412 (1924),

3. P.P. Bowden , J.B.P. Williamson and P.G. Laing, Journal
of Bone and Joint Surgery. 37-B, 676-690 (1955).

4. A.B. Ferguson, P.G. Laing and E.S. Hodge, Journal of
Bone and Joint Surgery, 42-A, 77-90 (1960).

5. J. Cohen, Journal of Materials, !_ [2], 354-365 (June
1966) .

6. H.D. Greene and D.A. Jones, Journal of Materials, 1_ [2],
345-353 (June 1966).

7. G.H. Hille, Journal of Materials . 1 [2], 373-383 (June

1966) .

8. S. Weisman, Biomechanical and Human Factors Symposium
1967, The American Society of Mechanical Engineers.

9. P.G. Laing, A.B. Ferguson and E.S. Hodge, Journal of
Biomedical Materials Research, 1_ [1], 135-150 (March

1967) .

10. D.H, Collins, Journal Pathology and Bacteriology, 65 ,
100-121 (1953).

11. N.K. Wood, E.J. Kaminski , and R.J. Oglesby, Journal of
Biomedical Materials Research. 4 [1], 1-12 (March 1970).

12. J.T. Scales, Proceedings Royal Society Medicine, 63 ,
1111 (1970).

13. J.S. Hirschhorn, A. A. McBeath, and M.R. Dustoor, Biomed-
ical Material Symposium No. 2, Bio ceramics -Engineering
in Medicine , Interscience Publishers , 19 72 , p"^^ 49 .

14. R.P, Welsh, R.M. Pilliar, and I, Macnab , Journal Bone and
Joint Surgery, 53A [5], 963 (July 1971).

15. J, Galante, W. Rostoker, R. Lueclc, and R.D. Ray, Journal
Bone and Joint Surgery, 53A [1], 101 (January 1971).

166

16 7

16. E. Lembert, J. Galante, and W. Rostoker, Clinical
Orthapaedi cs and Related Research, No. 87 (September
1972] , p. 303.

17. "Voids Help Attach Metal to Bone," Industrial Research,
December 19 71, p. 25.

18. J.L. Nilles and M. Lap it sky, Journal Biomedi cal Mats .
Res . Symp . , No. 4, John Wiley and Sons, New York, 19 73,
pp. 6 5-84.

19. H. Hahn and W. Palicli, Journal Biomedical Materials
Research, 4, 571 (1970).

20. W.W. Kriegel and H. Palmour, eds.. Ceramics in Severe
Environments , Plenum Press, New York, 19 71.

21. L. Smith, Archives of Surgery, 87 , 653 (October 1963).

22. R.P. Welsh and I. Macnab , Bioceramics -Engineering in
Medicine , Interscience Pub lishers , 19 72, p . 2 31 .

23. J.J. Klawitter and S.F. Hulbert, Biomedical Materials
Symposium No. 2, Biocerami cs -Engineering in Medicine,
Interscience Pub lishers , 19 72 , p . 161.

24. S.F. Hulbert, F.W. Cooke, J.J. Klawitter, R.B. Leonard,
B.W. Saver, D.D. Moyle, and H.B. Skinner, Journal of
Biomedical Research Symposium, No. 4, 1973, pp. 1-3 3.

25. P. Griss , G. Heimke, H. von Andrean- Werburg , B. Krempien,
S. Reipa, H.J. Lauterbacli. and H.J. Hartung, Journal
Biomedical Materials Research Symposium, No. 6 (in press)

26. G.A. Graves, F.R. Noyes , and A.R. Villanueva, Journal
Biomedical Materials Research Symuosium. No. 6 (in press)

27. L.L. Hench, T.K. Greenlee, Jr., and W.C. Allen, Annual
Report ^1, U.S. Army Med. R and D Command, Contract No.
DADA-17-70-C-0001 (1970).

28. L.L. Hench, T.K. Greenlee, Jr., and W.C. Allen, Annual
Report ''12, U.S. Army Med. R and D Command, Contract No.
DADA-17-70-C-0001 (1971).

29. L.L. Hench, H.A. Paschall, W.C. Allen and G. Piotrowski ,
Annual Report #3, U.S. Army Med. R and D Command, Con-
tract No. nADA-17-70-C-0001 (1972).

30. L.L. Hench, H.A. Paschall, W.C. Allen, and G. Piotrowski,
Annual Report H, U.S. Army Med. R and D Command, Con-
tract No. DADA-17-70-C-0001 (1973).

168

31. L.L. Hench, R.J. Splinter, W.C. Allen, and T.K. Greenlee,
Jr., J Biomed. Mats. Res. Symp., No. 2, Interscience
Publishers, New York, 19 72, pp. 117-143.

32. D.M. Sanders and L.L. Ilench , J. American Ceramic Society,
56i [7], 373-377 (July 1^73).

33. T.K. Greenlee, Jr., C.A. Beckham, A.R. Crebo and J.C.
Malmorg, J. Biomed. Mat. Res . , 6, 244 (1972).

34. C.A. Beckham, T.K. Greenlee, Jr. , and A.R. Crebo,
J. Calcified Tissue Res., 8_, 2 (1971).

35. L.L. Hench and H.A. Pas ch all, J. Biomed. Mats. Res.
Symp . , No, 4, John Wiley and Sons , New York , 1973 , pp.
25-42.

36. L.L, Hench and H.A. Paschall, "Prostheses and Tissue:
The Interface Problem," to be published in J. Biomed,
Mats. Res. Symp.

37. D.M. Sanders and L.L, Hench, Applied Spectroscopy, 2 8
[3], 247-255 (May/June 1974).

38. D.M, Sanders and L.L, Hench, J. American Ceramic Society,
54_ [7] , 373-378 (1973) .

39. L.L, Hench, Medical Instrumentation, 7 [2], 136-144
(1973) ,

40. G, Gomori , Methods in Enzymology, Vol. 1, Academic Press,
New York, 1955, pp. 138-146.

41. Scott Anderson, J. American Ceramic Society, 33 [2],
45-51 (February 1950).

42. D.R. Hach, Colorimeter Methods Manual, 7th Ed., October
1971.

43. American Public Health Association, Standard Methods of
Waste Water Analysis, American Public Health Association,
New York, 1969 , p. 258.

44. R.E. Ferrell and G.G. Paulson, Energy Dispersive Analy-
sis of X-ray Spectra Generated in the SEM , ORTEC Manual.

45. R.J. Bell and P. Dean, "The Vitreous State," in Discus-
sions of the Faraday Society, But terworths , London,

19 70, p. 50.

46. P.H. Gaskell, "The Vitreous State," in Discussion of the
Faraday Society, But terworths , London, 19 70, p . 5 0 .

• 169

47. I. Simon and M.D. McMahon, J. Chemical Physics . 21 [1],
23-30 (January 1953).

48. A.E. Clark and L.L. Hench , "Effects of p""^, B^"^, and F'
on the Surface Chemistry of Bioglass," Annual Report #4,
U.S. Army Med. R and D Command, Contract No. DADA-17-
70-C-OOOl (1973) , p. 37.

49. S.R. Levitt, K.C. Blakeslee, and R.A. Condrate , Sr. ,
Mem. Soc. Roy. Sci. Liece. 2^, 121-141 (1970).

K. Nakamoto, Infrared Spectra of Inorganic and Coordi-
nation Compounds, John Wiley and Sons, Inc., New York,
T9TT.

51. L. Holland, "Surface Chemistry and Corrosion of Glass,"
in The Properties of Glass Surfaces, John Wiley and
Sons, New York, 1964.

52. F. Korber and G. Tromel, Z. Elektrochem. , 38,578-82
(1932).

53. D. McConnell, Arch, oral Biol. , 10^, 421-431 (1965).

54. D. McConnell, Science , 136 , 241-244 (1962).

55. T.D. Farr, G. Tarbutton and H.T. Lewis, J. Phys . Chem. .
6^, 318 (1962).

56. N.V. Belov, The Structure of Glass, Acad. Sci. U.S.S.R.,
Chapman and Hall, Ltd. , London , r9"53.

57. L.A. Harris, J. Appl. Phys. . 39_ [3], 1419-1431 (1968).

58. C.G. Pantano, G.Y. Onoda, and D.B, Dove, Unpublished data,

59. C.R. Das, Trans. Ind. Coram. Soc. , 2± [1], 12 (1965).

60. G.Y. Onoda, First Annual Progress Report , "Glass Surface
Chemistry: Application of Auger Electron Spectroscopy,"
Glass Container Industry Research Corp., July 1974.

61. P.W. McMillan, Glass Ceramics , Academic Press, New
York, 1964.

62. M. Tomozawa, Advances in Nucleation and Crystallization
in Glass , Special Publication No. 5, yVmer. Cer. Soc. ,
1971, pp. 41-50.

63. J. Enss, Glasstech. Ber. , 5_, 449-474 (1928).

64. W.A. Weyl, J. American Ceramic Society, 24_ [7], 221-225
(1941) .

170

65. H. Eagle, Science, 150, 432 (1959).

66. D. Kay, Techniques for Electron Microscopy, Charles and
Thomas , Springfield , 111., 1961.

67. E.M. Carlisle, Science ,^167, 279-80 (January 16, 1970).

68. A.W. Ham, Histology, 6th Ed., J.B. Lippincott , 1969,
p. 39 4. "~

69. J.L. Matthews, J.H. Martin, B.J. Collins, J.W. Kennedy III
and E.I. Powell, Jr., Calcium Parathyroid Hormone and

the Calcitonins , Proceedings of tlie Fourth Parathyroid
Conference, Excerpta Medica, Amsterdam, 19 72.

70. Philips Operation Instructions S(T)EM-Unit 94320657001,
Provisional First Edition, Philips, 1974.

BIOGRAPHICAL SKETCH

Artliur E. Clark, Jr., was born in Savannah, Georgia, in
1947. He attended hifih school at the American School, Makati ,
Rizal, Philippines. He received a Bachelor of Science degree
in Metallurgical Engineering in June of 1969. Since obtain-
ing his bachelor's degree, the author has been pursuing his
doctorate at the University of Florida.

171

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.

L. L. Hench, Chairman
Professor of Materials Science
and Engineering

yv- /•

J-

R. T. DeHoff

Professor of Materials Science

and Engineering

E. D. Verink, Jr. Tf xV

Professor of Materials'' Science
and Engineering

H. A. Paschall
Associate Professor of
Orthopedic Surgery

0 ^ ,jy\Gj^

^x^

HI) 1 -^

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Solubility and biocompatibility of glass"

See other formats