L'-band AGPM vector vortex coronagraph's first light on VLT/NACO: Discovery of a late-type companion at two beamwidths from an F0V star

Astronomy 8l Astrophysics manuscript no. naco_agpm_onsky_dm_letter_astroph
April 4, 2013

©ESO 2013

Letter to the Editor

L'-band AGPM vector vortex coronagraph's first light on VLT/NACO
Discovery of a late-type companion at two beamwidths from an FOV star

D. Mawet^ O. Absil^, C. Delacroix^, J. H. Girard^ J. Millii, J. O'Neal^ P. Baudoz^, A. Boccaletti^, P. Bourget^ V.

Christiaens^, P. Forsberg^, F. Gonte^ S. Habraken^, C. Hanot^, M. Karlsson^, M. Kasper^, J.-L. Lizon^, K. Muzic^, R.

Olivier^, E. Pena^ N. Slusarenko^ L. E. Tacconi-Garman-', and J. Surdej^

o

<

o

European Southern Observatory Vitacura, Alonso de Cordova 3107, Casilla 19001, Vitacura, Santiago 19, Chile

Departement d'Astrophysique, Geophysique et Oceanographie, Universite de Liege, 17 Allee du Six Aout, B-4000 Liege, Belgium

European Southern Observatory Headquarters, Karl-Schwarzschild-Str 2, 85748 Garching bei Munchen, Germany

Department of Engineering Sciences, Angstrom Laboratory, Uppsala University, Box 534, SE-751 21 Uppsala, Sweden

LESIA, Observatoire de Paris, 5 pi. J. Janssen, F-92195 Meudon, France

GDTech s.a., LIEGE Science Park, rue des Chasseurs Ardennais, B-4031 Liege, Belgium

Accepted April 1,2013

ABSTRACT

Context. High contrast imaging has thoroughly combed through the limited search space accessible with first-generation ground-
based adaptive optics instruments and the Hubble Space Telescope. Only a few objects were discovered, and many non-detections
reported and statistically interpreted. The field is now in need of a technological breakthrough.

Aims. Our aim is to open a new search space with first-generation systems such as NACO at the Very Large Telescope, by providing
ground-breaking inner working angle (IWA) capabilities in the L' band. The L' band is a sweet spot for high contrast coronagraphy
since the planet- to- star brightness ratio is favorable, while the Strehl ratio is naturally higher.

Methods. An annular groove phase mask (AGPM) vector vortex coronagraph optimized for the L' band, made from diamond sub-
wavelength gratings was manufactured and qualified in the lab. The AGPM enables high contrast imaging at very small IWA,
potentially being the key to unexplored discovery space.

Results. Here we present the installation and successful on-sky tests of an L'-band AGPM coronagraph on NACO. Using angular
diff'erential imaging, which is well suited to the rotational symmetry of the AGPM, we demonstrated a AU > 7.5 mag contrast from
an IWA ^ 0V09 onwards, during average seeing conditions, and for total integration times of a few hundred seconds.

Key words. Instrumentation: high angular resolution - Stars: planetary systems - Stars: binaries: close

>

o

CO

o

o
m

'i

1. Introduction

The goal of high contrast imaging is primarily to discover and
characterize extra-solar planetary systems. For technical motives
rather than scientific ones, most surveys have targeted young and
nearby stars. This search space is already limited, and imaging
surveys have only explored its surface, strongly limited by con-
trast and inner working angle (IWA) capabilities ( Absil & Mawet
[2010) . Despite the few spectacular objects discovered and in-
sightful lessons learned from the majority of non detections,
hunting for long-period planets has mostly been a hard and un-
fruitful ta sk, with a very low yield ( [Lafreniere et al.|2007t|Chau-

vin et al.||2010t[vrgan et al . 2012). It is expected that opening

Martinache et al.|2012| ). Coronagraphy promises to be high con-
trast imaging 's sharpest tool, but requires exquisite image qual-
ity and stability to perform efl&ciently. These new instruments
have thus been designed accordingly. However, first-generation
instruments still possess untapped potential that only ten years
of operations and understanding allow us to exploit fully (^Girardj
|et al.||2012| ), especially in the mid-infrared (L' band, from 3.5
to 4.2 yum). This wavelength range offers significant ad vantages
compared to shorter wavelengths ( [Kasper et al.||2007) : (i) The
L'-band contrast of planetary-mass companions with respect to
their host stars is predicted to be more favorable than in the H
band JBaraflTe et al.|[2QQ3l [Fortney et ari[20Q8l [Spiegel & Bur-|

the parameter space to fainter/smaller planets closer to their par-
ent stars will bring many new objects ( [Crepp & Johnson|2011| ).
The new sample is critical to exoplanet science because it will
shed some light on planet formation mechanisms at or within
the snow line, and help bridge the gap between the population of
close planets discovered by radial velocity or transit techniques
and the free-floating planets discovered by microlensing obser-
vations (Qua nz et al.|2012| ). This intermediate parameter space
should be opened by the second-generation coronagraphic in-
struments that have started to arrive at major observatories (|Mac-

fintosh etaT.|2012| [Kasper et al.|2012l|Qppenheimer et al.|2012

rows|2012 ) so that lower-mass, older objects can be addressed;
and (ii) the L'-band provides better and more stable image qual-
ity, with Strehl ratios well above 70% and sometimes as high as
90%, thus reducing speckle noise. These advantages certainly
compensate for the increased sky background in the thermal in-
frared and the loss in resolution, especially if small IWA phase-
mask coronagraphs are available.

Here we describe the successful implementation of an L'-
band annular groove phase mask (A GPM; Mawet et al.[|2005|)
vector vortex coronagraph on NACO ( [Lenzen et al.|2 003 ; Rous-[
set et al.[|2003| ), the adaptive optics instrument of ESO' s Very

Article number, page 1 of H

Fig. 1. Scanning electron microscope (SEM) images of the NACO
AGPM. a. Structure profile schematic, with h2 = 5+0.1 yum, hi = 1 ±0.1
fim, and w = 0.65 + 0.03 yum (the grating pitch is 1.42 yum). b. Zoom on
the center of the diamond AGPM. c. Overview of the structure showing
the uniformity and original cleanliness of this particular device.

Large Telescope (VLT). To our knowledge it is the first time that
an image-plane phase-mask coronagraph has been used in the
mid-infrared.

2. An AGPM vector vortex coronagraph on NACO

The AGPM is an optical vortex made from diamond subwave-
length gratings (Fig. [T]). When centered on the diflTraction pat-
tern of a star seen by a telescope, optical vortices afifect the
subsequent propagation to the downstream Lyot stop by redi-
recting the on-axis starlight outside the pupil (e.g., |Mawet et al.
|2005j . The advantages of the AGPM coronagraph over classi-
cal Lyot coronagraphs or phase/amplitude apodizers are small
IWA, down to 0.9 A/ D (e.g., 0V09 in the L' band at the VLT,
slightly smaller than the dififraction limit); clear 360° oflT-axis
field of view/discovery space; outer working angle set only
by the instrument and/or mechanical/optical constraints; achro-
matic over the entire working waveband (here L' band); high
throughput (here ^ 88%); and optical/operational simplicity. Af-
ter eight years of intense technological development, the AGPM
has reac hed a suflScient readiness level for telescope imple men-
tation ( [Delacroix et al.|2QT3}|Forsberg & Karlsson|2Q13a| ). The

AGPM selected for NACO was the third one in a series of four
realizations (AGPM-L3). Its theoretical raw null depth limited
by its intrinsic chromatism was estimated (assuming a trape-
zoidal profilejjsee Fig.[T]) and measured to be around 5 x 10"^
(corresponding to a raw contrast of 2.5 x 10"^ at 2A/D), which
is more than needed for on- sky operations where the limit is set
by the residual wavefront aberrations.

The AGPM was installed inside NACO as part of a planned
overhaul in November 2012. The AGPM was mounted on the
entrance slit wheel by means of a dedicated aluminum mount,
designed by GDTech s.a. The assembly of the mount and AGPM
was done on site at Paranal observatory in a clean room environ-
ment. Prior to on- sky tests and operations, a CONIC A internal
image of the mask was done (see Fig. |2] left), revealing signif-
icant dust contamination, marginally aflTecting the background
noise. The slit wheel was set so that the center of the AGPM
falls close to but slightly away from CONIC A' s detector quad-
rant intersection. The AGPM field of view is ^ 15'', correspond-
ing to an outer working angle (OWA) of 7V5. The OWA is only
limited by the size of the device (10 mm in diameter) and its

This AGPM diff'ers from the one tested in Delacroix et al. (2013 1

The tops of the grating walls are triangular to improve the transmittance
(Fig.[T]). This profile was etched with a p rocess similar to the fabrica tion
of broadband antireflective structures ( [Forsberg & Karlsson|2013b| ).

Article number, page 2 ofH

Fig. 2. Left: view from inside CONICA, showing the 9 mm clear
aperture corresponding to a 15''field of view (diameter), fully contained
within the 2Vfield of view of the L27 objective. Middle: full oversized
stop of CONICA, showing the VLT pupil (including the central obscu-
ration and struts). Right: AP0165 pupil mask (diameter^ 0.87 x Dpup)
available inside CONICA, aligned to cover the diffraction and thermal
background from the central obscuration and struts.

Table 1. Observing log.

Date

09/12/2012

11/02/2013

Star

HD4691

HD123888

Spectral type

FOV

Kim

Vmag

6.79

6.62

L app. mag

5.86

4.01

DIT/NDIT/# fr

0.2S/10/100

0.25S/80/40

Seeing

r - V!5

0V8 - 0V9

To

2 -4ms

4 -5ms

Strehl ratio

65 - 80%

75 - 80%

PA range

^30°

^30°

mount. The mask transmittance at L' band was measured on the
sky to be 85% ±5%, which is consistent with the theoretical value
and laboratory measurements, both ^88%, limited by imperfect
antireflective treatments and mild absorption features around 4
yum ( Delacro ix et al.||2013|). To stabilize speckles, we used the
pupil tracki ng mode enabling angular diflferential imaging (ADI,
Marois et al. 1 ,2006), which is perfectly adapted to the circular
symmetry and 360° field of view of the AGPM. The CONICA
camera is equipped with a pupil mask which blocks the tele-
scope central obscuration and spiders. Once correctly aligned
with the pupil (in x, y, and 0), this mask is optimal for use with
the AGPM in pupil tracking mode (see Fig. |2]). The measured
throughput of the AP0165 mask used here is ^ 60%. In terms
of sensitivity, it is worth noting that the throughput loss is almost
entirely compensated by the improved thermal background. The
pupil obscuration is responsible for more than 25% of the ther-
mal emissivity of the telescope, even though its area only covers
^5%. Therefore, instead of loosing 1 - Vo!6 ^ 0.225 in sensi-
tivity, only 1 - 0.6/ VO.6 ^ 0.75 ^ 0.1 is actually lost. To main-
tain its high contrast capabilities, the centering of the star on the
AGPM must be within (A/D)/10, or ^ 10 mas (a third of a pixel
at L27). This level of centering capability is now routinely ob-
tained with NACO, both in pupil and field tracking modes; we
typically measure ^ 1 mas/minute drifts across the meridian.

3. First light

On December 9, 2012, a representative observing sequence
was performed on the 1.9-Gyr old main sequence standard star
HD4691 (see Table [T]), under ^ 1V2 visual seeing conditions.
This star was chosen to maximize brightness and field rotation

D. Mawet et al.: L'-band AGPM vector vortex coronagraph's first light on VLT/NACO

b. U CORO RAW (AGPM, AP0165, DIT=0.2s) c. U CORO RCA (AGPM, AP0165, DIT=0.2s)

HD4691 B, M2V (~1 : 1 00 (g 0".2)

Fig. 3. a: L'-band NACO PSF with the AP0165 pupil mask in the beam, b: L'-band NACO coronagraphic image with the star centered on the
AGPM (the dynamic range and corresponding colorbar/scale are a factor of 10 smaller than in image a), c: Result of our PCA-ADI data reduction
pipeline, revealing a putative faint off"- axis M2V companion at only ^ 0V19 (the dynamic range and corresponding colorbar/scale are a factor of
100 smaller than in image a). The scale is linear on all images (and scaled down by a factor of 10 and 100 for b and c, respectively), illustrating in
sequence the benefits of coronagraphy and optimized data reduction technique. The bottom color bar refers to Figure a.

during the short time allocated for this technical test. A ^ 30-
minute ADI sequence was obtained with a parallactic angle (PA)
range of 30° and for a total exposure time of 200s on source; the
efficiency was mediocre for technical reasons. After acquiring
an oflT-axis PSF for photometric reference, we measured an in-
stantaneous contrast of ^ 50 peak-to-peak (despite the average-
to-bad conditions, see Table [T]). The attenuation is about 5 times
higher than measur ed with NACQ's four-q uadrant phase-mask
coronagraph at Ks ( [Boccaletti et al.|[2004| ). The coronagraph
diflTraction control yields two instantaneous benefits compared
to classical imaging: (i) the peak saturation limit is decreased by
a factor ^ 50, and (ii) the level of quasi-static speckles pinned
to the PSF and the stellar photon noise limit are potentially de-
creased by a factor ^ V50 ^ 7, both within the AO control ra-
dius of lAID. All in all, the L'-band AGPM coronagraph allows
the background limit to be reached much closer in.

3.1. Detection of a candidate companion

After applying basic cosmetic treatment to our sequence of
100 frames (background subtraction, flat fielding, and bad
pixel/cosmic ray correction), we decided to use the quality and
stability of the L'-band PSF provided by NACO to perform a so-
phisticated speckle subtraction. We used the very efficient prin-
cipal component analysis (PGA) algorithm presented in So um-,
mer et al.] ( |2012| ). The result, using the whole image and re-
taining three main components, is presented in Fig. |5] By pure
chance, the object has a ^ 1 : 100 (or AU ^ 5) ofiT-axis compan-
ion located at ^ 0V19 (< 2A/D), making this our first unexpected
scientific result. The companion flux and astrometry were ob-
tained by u sing the fake negative companion technique ( |Marois|
et al.||201Q| ). The method proceeds as follows: (i) estimate the
(biased) position and flux of the companion from the first re-
duced image; (ii) use the measured oflf-axis PSF as a template
to remove this first estimate from the cleaned data cube before
applying PC A; and (iii) iterate on the position x,y and flux un-
til a well-chosen figure of merit is minimized (x^ in a pie chart
aperture centered on the first estimate of the companion position,
2A4A/D in radius and 6 x 1.22A/D in azimuth). The minimiza-
tion was performed with the Simplex- Amoeba optimization.

Close to the center where the speckle field is intense, the
companion flux can be overestimated because the minimization
tries to subtract underlying speckles. To estimate our error bars,
we decided to proceed with an alternative method called smart-
ADI PCA: the frames used to construct the component basis are
selected according to a minimum azimuthal separation criterion
(here A^^ = 0.1 A/ D). With this technique, flux is much better
preserved. We measured up to 200% additional flux compared
to normal ADI PCA. However, fake planet tests still indicate
a 25% flux loss, confirming that this method underestimates the
flux and thus provides our lower bound. Finally, the coronagraph
oflT-axis attenuation profile, measured in the lab (Delacroix et al.|
|2013 ) was also taken into account. Using the BCAH98 model
(Baraffeet al. 1998), we derived the properties of the newly dis-
covered candidate companion, assuming association. Note that
the TRILEGAL starcount model ( [Girardi et aL][2005] ) yields a
probability of 4 x 10"^ that it is an unrelated background object.
At an absolute L'-band magnitude of 6.65, and 1.9 Gyr for the
system, the putative companion would most likely be an M2V
star at projected separation of 11.8 ± 0.4 AU, and 354?5 ± 0?6
position angle (see Table [2]).

3.2. Representative NACO AGPM detection limits

Since the presence of the companion aflfects the contrast, we took
another similar representative ADI sequence on a diflTerent stan-
dard star (HD 123 88 8, see Table [T]). This technical test was per-
formed under better conditions, and benefited from our improved
mastering of the new mode (efficiency was four times better than
during the first light). A similar instantaneous attenuation was
confirmed. To calibrate our detection limits against flux losses
induced by PCA, we injected fake companions (at I5cr) prior to
PCA and measured their throughput after PCA. We used the de-
rived throughput map to renormalize the initial contrast curve a
posteriori (see Fig. |4]). Figure |4] shows excellent detection ca-
pabilities down to the IWA of the AGPM. The final calibrated
contrast presented here (green dash-dot curve), is limited by the
small PA range, especially at small angles. The floor reached
beyond V is due to the background at L', and will be lower for
brighter targets and/or longer integrations.

Article number, page 3 of H

Table 2. Properties of HD4691B.

4. Conclusions

Distance

Age ( Holmberg et al. 2009 1

Companion abs. U mag

Mass

Temperature

Sp Type

Angular sep.

Proj. angular sep.

Position angle

62 pc
1.9 Gyr
6.65 + 0.1

^ 0.3 Mo

:^3450K

M2V

0V190 ± 0V007

11.8 + 0.4 AU

279° + 0?6

10^

^ 10

"co
c

O

O _2

D 10

B10-^

-^ -4

10'

'

— NACO PSF int. prof.

— AGPM int. prof, (raw contrast)

— - AGPM PCA-ADI (5a)

-,

2.5

.7.5

10

12.5

0.1

1 7

Angular separation (")

Fig. 4. Normalized azimuthally averaged relative intensity profiles
and contrast curve. The plain red curve shows the intensity profile of a
typical saturated NACO L' PSF (similar brightness and exposure time).
The blue dashed curve shows the AGPM intensity profile before PCA,
demonstrating the instantaneous contrast gain provided by the coron-
agraph at all spatial frequencies within the AO control radius (^ 0V7).
The green dash-dot curve presents the reduced PCA-ADI 5cr detectabil-
ity limits (40 frames, 800s, APA ^ 30°), taking both the coronagraph
off'-axis transmission and the PCA-ADI flux losses into account.

We would like to raise several flags that we will thoroughly
address in subsequent papers: (i) Classical tools assuming Gaus-
sian statistics, perfectly valid at large separation, lose signif-
icance close to the center simply because the sample size de-
creases dramatically. At a given angular separation r (in AjD),
there are Inr resolution elements, i.e., only 6 at r = 1/l/Z), 12
at r = 2X1 D, etc. (ii) The probability density function (PDF) of
speckle noise and associated confidence level for detection de-
pend on radius. ADI was shown to transform speckles' modified
Rician PDF into quasi-Gaussian PDF at large separations, but it
is expected that this property of ADI does not hold true at small

angles ( [Marois et a l. 2008 ). (iii) The flux attenuation induced by
ADI, potentially significant at small angles, does not scale lin-
early with the companion brightness, which makes its calibration
more difificult. These points should be kept in mind when inter-
preting contrast curves such as those presented in Fig.|4] but also
all contrast/detectivity plots that have been presented so far in the
literature for very small angles.

Article number, page 4 ofH

The AGPM was designed to provide exquisite IWA (and OWA)
capabilities, down to 0.9 AjD (0V09 at L'), as demonstrated in
Fig. |4] The downside of the AGPM's small IWA is its sensi-
tivity to the Strehl ratio (as all coronagraphs) and to pointing
errors. The Apodizing Ph ase Plate (APP) is another advanced
coronagraph oflTered at L' ( |Quanz et al.||2010[ |Kenworthy et al.
|2013| ). The only, but significant, benefit of this pupil plane phase
apodizer over the AGPM is its intrinsic immunity to tip-tilt er-
rors. This advantage, which has to be traded ofl' with the sig-
nificantly limited field of view provided by the APP, is decisive
when tip-tilt is an issue as was the case with NACO prior to
November 2011 ('Girard et al."2012). However, it is less obvious
when the instrument provides nominal PSF stability.

In a single technical run, the L'-band AGPM has proved to
be a reliable coronagraphic solution, and one of the best high-
contrast imaging modes of NACO (and most likely worldwide).
Combined with ADI, we demonstrated that high contrast of the
order of AL' > 7.5 mag can be reached from the IWA of 0V09
onwards, even with very modest on-source integration time, PA
variation, and average conditions. The field of view is a clear
360° discovery space 15'' in diameter. The coronagraph is opti-
mized for pupil tracking and is easy to use, thanks to the stability
of the NACO L'-band PSF.

Acknowledgements. This work was carried out at the European Southern Obser-
vatory (ESO) site of Vitacura (Santiago, Chile). OA and JS acknowledge support
from the Communaute frangaise de Belgique - Actions de recherche concertees
- Academie universitaire Wallonie-Europe. We would like to thank the referee,
Dr Christian Marois, for his constructive comments.

References

Absil, O. & Mawet, D. 2010, A&A Rev., 18, 317

Baraffe, L, Chabrier, G., Allard, E, & Hauschildt, P. H. 1998, A&A, 337, 403

Baraffe, L, Chabrier, G., Barman, T. S., Allard, E, & Hauschildt, P. H. 2003,

A&A, 402, 701
Boccaletti, A., Riaud, P., Baudoz, P., et al. 2004, PASP, 1 16, 1061
Chauvin, G., Lagrange, A.-M., Bonavita, M., et al. 2010, A&A, 509, A52
Crepp, J. R. & Johnson, J. A. 2011, ApJ, 733, 126
Delacroix, C, Absil, O., Eorsberg, P., et al. 2013, accepted to A&A
Eorsberg, P. & Karlsson, M. 2013a, Diamond & Related Materials, 34, 19
Eorsberg, P. & Karlsson, M. 2013b, Optics Express, 21, 2693
Eortney, J. J., Marley, M. S., Saumon, D., & Lodders, K. 2008, ApJ, 683, 1104
Girard, J. H. V., O'Neal, J., Mawet, D., et al. 2012, in Proc. SPIE, Vol. 8447
Girardi, L., Groenewegen, M. A. T., Hatziminaoglou, E., & da Costa, L. 2005,

A&A, 436, 895
Holmberg, J., Nordstrom, B., & Andersen, J. 2009, A&A, 501, 941
Kasper, M., Apai, D., Janson, M., & Brandner, W. 2007, Astronomy and Astro-
physics, 472, 321
Kasper, M., Beuzit, J.-L., Eeldt, M., et al. 2012, The Messenger, 149, 17
Kenworthy, M. A., Meshkat, T., Quanz, S. R, et al. 2013, ApJ, 764, 7
Lafreniere, D., Doyon, R., Marois, C, et al. 2007, ApJ, 670, 1367
Lenzen, R., Hartung, M., Brandner, W., et al. 2003, in Proc. SPIE, ed. M. lye &

A. E M. Moorwood, Vol. 4841, 944-952
Macintosh, B. A., Anthony, A., Atwood, J., et al. 2012, in Proc. SPIE, Vol. 8446
Marois, C, Lafreniere, D., Doyon, R., Macintosh, B., & Nadeau, D. 2006, ApJ,

641,556
Marois, C, Lafreniere, D., Macintosh, B., & Doyon, R. 2008, ApJ, 673, 647
Marois, C, Macintosh, B., & Veran, J.-P. 2010, Adaptive Optics Systems II.

Edited by EUerbroek, 7736, 52
Martinache, E, Guyon, O., Clergeon, C, Garrel, V, & Blain, C. 2012, in Proc.

SPIE, Vol. 8447
Mawet, D., Riaud, R, Absil, O., & Surdej, J. 2005, ApJ, 633, 1191
Oppenheimer, B. R., Beichman, C, Brenner, D., et al. 2012, in Proc. SPIE, Vol.

8447
Quanz, S. P., Lafreniere, D., Meyer, M. R., Reggiani, M. M., & BuenzH, E. 2012,

Astronomy and Astrophysics, 541, 133
Quanz, S. R, Meyer, M. R., Kenworthy, M. A., et al. 2010, ApJ, 722, L49
Rousset, G., Lacombe, E, Puget, R, et al. 2003, in Proc. SPIE, ed. P. L. Wiz-

inowich & D. Bonaccini, Vol. 4839, 140-149
Soummer, R., Pueyo, L., & Larkin, J. 2012, ApJ, 755, L28
Spiegel, D. S. & Burrows, A. 2012, ApJ, 745, 174
Vigan, A., Patience, J., Marois, C, et al. 2012, A&A, 544, A9