Experimental demonstration of binary shaped pupil mask coronagraphs for telescopes with obscured pupils
Kanae Haze, Keigo Enya, Lyu Abe, Aoi Takahashi, Takayuki Kotani, Tomoyasu Yamamuro
aa r X i v : . [ a s t r o - ph . I M ] M a r Experimental demonstration of binary shaped pupilmask coronagraphs for telescopes with obscured pupils
Kanae
Haze Keigo
Enya Lyu
Abe Aoi
Takahashi
Takayuki
Kotani and Tomoyasu Yamamuro Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency (JAXA), 3-1-1Yoshinodai, Chuo-ku, Sagamihara, Kanagawa 252-5210, [email protected] Laboratoire Hippolyte Fizeau, UMR 6525 Universit´e de Nice-Sophia Antipolis, Parc Valrose,F-06108 Nice, France Department of Space and Astronautical Science, The Graduate University for Advanced Studies,3-1-1 Yoshinodai, Chuo-ku, Sagamihara, Kanagawa 252-5210, Japan Inter-University Research Institute Corporation, National Institutes of Natural Sciences, NationalAstronomical Observatory of Japan, 2-21-1 Osawa, Mitaka,Tokyo 181-8588, Japan Optcraft, 3-26-8 Aihara, Sagamihara, Kanagawa 229-1101, Japan (Received ; accepted )
Abstract
We present the fabrication and experimental demonstration of three free-standingbinary shaped pupil mask coronagraphs, which are applicable for telescopes withpartially obscured pupils. Three masks, designed to be complementary (labeled Mask-A, Mask-B, and Mask-C), were formed in 5 µ m thick nickel. The design of Mask-Ais based on a one-dimensional barcode mask. The design principle of Mask-B issimilar, but has a smaller inner working angle and a lower contrast than Mask-A.Mask-C is based on a concentric ring mask and provides the widest dark regionand a symmetric point spread function. Mask-A and Mask-C were both designedto produce a flexibly tailored dark region (i.e., non-uniform contrast). The contrastwas evaluated using a light source comprising a broadband super-luminescent light-emitting diode with a center wavelength of 650 nm, and the measurements werecarried out in a large vacuum chamber. Active wavefront control was not applied inthis work. The coronagraphic images obtained by experiment were mostly consistentwith the designs. The contrast of Mask-A within the ranges 3.3 - 8 λ/D and 8 -12 λ/D was ∼ − - 10 − and ∼ − , respectively, where λ is the wavelength and D is the pupil diameter. The contrast close to the center of Mask-B was ∼ − andthat of Mask-C over an extended field of view (5 - 25 λ/D ) was ∼ − - 10 − . Theeffect of tilting the masks was investigated, and found to be irrelevant at the ∼ − Key words: instrumentation: high angular resolutionplanetary systemstelescopes
1. Introduction
Direct observation of spatially resolved exoplanets is important in order to learn abouttheir formation, evolution, and diversity. However, the huge contrast in flux between an exo-planet and its parent star is the primary difficulty in accomplishing direct observations. Forinstance, the contrast between the sun and the earth is ∼ − in the visible wavelength region,and ∼ − in the mid-infrared (mid-IR) wavelength region (Traub & Jucks 2002). Therefore,it is necessary to develop stellar coronagraphs that can overcome the contrast between the starand the planet.Among the various coronagraphic imaging methods, binary shaped pupil maskcoronagraphs have some advantages (Jacquinot & Roizen-Dossier 1964; Spergel 2001; Vanderbeiet al. 2003a; Vanderbei et al. 2003b; Vanderbei et al. 2004; Kasdin et al. 2005a; Kasdin et al.2005b; Tanaka et al. 2006; Belikov et al. 2007; Enya et al. 2007; Enya et al. 2008; Enya et al.2011b; Haze et al. 2009; ? ; Enya & Abe 2010; Carlotti et al. 2011; Haze 2012). The function ofa binary shaped pupil mask coronagraph is to produce a high-contrast point spread function(PSF) which is much less sensitive to telescope pointing errors, and also less sensitive to wave-length (exceptwhen scaling the size of the PSF) than other coronagraphs. Simplicity is anotheradvantage of this type of optical system. In our previous experimental research (Enya et al.2007; Enya et al. 2008; Haze et al. 2009; Haze et al. 2011), we demonstrated high contrast per-formance using a checkerboard mask, which is a conventional type of binary shaped pupil mask(Vanderbei et al. 2004). After the successful demonstration of the principle of binary shapedpupil masks fabricated on substrates, free-standing metal (copper and nickel) masks were de-veloped. Since there is no substrate limiting the applicable wavelength region of the mask, suchfreestanding masks can be used for observations over a wide infrared wavelength region. Thisprovides a big advantage, since the contrast between the star and the planet is much less in theinfrared region than in the visible light region. Moreover, high contrasts ( ∼ − ) have beenconfirmed in experiments using free-standing masks at visible wavelengths (Haze 2012; Enya etal. 2012). The mask design used in those experiments, a conventional checkerboard design, isapplicable to specifically designed off-axis telescopes. However, the high contrast performanceof these masks is impaired when used with a centrally obscured telescope and in which there ispartial obscuration due to support arms. In this case, further optimization is required. Enya& Abe 2010 presented solutions using a simple one dimensional (1D) optimization of the mask2esign so that the masks could be applied to various telescope pupils including those withsegments and/or obscurations. More advanced, complicated mask designs for telescopes withobscured pupils are presented in Enya et al. 2011b; Carlotti et al. 2011. Considering this back-ground, we carried out the fabrication and experimental demonstration of free-standing maskcoronagraphs that can be used with centrally obscured onaxis telescopes. The experiments,results, and a discussion are presented in the following sections.
2. Experiments and Results
In this section, we present three new designs for free-standing masks and the results ofour first coronagraphic experiments using each mask.
The designs of the masks (Mask-A, Mask-B, and Mask-C) are shown on the left-handside of figure 1. These masks are types of binary shaped pupil masks, and their basic designis based on those reported in Enya et al. 2011b. Optimization of the basic mask shapes wasperformed with the LOQO solver presented by Vanderbei 1999. Mask-A and Mask-B are basedon an integral 1D design, and Mask-C is based on a concentric ring design (table 1). Thesemasks have the general advantages of binary pupil masks: (1) They are robust against pointingerrors, and (2) they can, in principle, make observations over a wide range of wavelengths. Thesethree masks also have a particularly important asset: (3) the design makes them applicable forthe pupil of SPICA telescope, which is partially obscured by a secondary mirror and a supportspider. The theoretical PSFs of each mask are shown on the right-hand side of figure 1. Thecentral bright region of the PSF is called the “core”, and the regions near to the core, in whichdiffracted light is reduced, are called the “Dark Regions (DRs)”. Mask-A is an example of adesign with generalized darkness constraints. The inner working angle (
IW A ) and the outerworking angle (
OW A ) are 3.3 λ/D and 12 λ/D , respectively, where λ is the wavelength and D is the pupil diameter. The required contrast for this is 10 − at 3.3 λ/D and 10 − between8 λ/D and 12 λ/D . The required contrast between 3.3 λ/D and 8 λ/D was determined bythe constraint that the contrast should be below a straight line on a log scale. Mask-B isintended to be used with a small IW A . The required contrast for this is 10 − between 1.7 λ/D and 6.2 λ/D . This mask is more useful than the other two masks for direct observationin the infrared wavelength region of young Jovian planets close to the star. With Mask-C awide-field coronagraphic image is obtained. The IW A and the
OW A are 5 λ/D , and 25 λ/D ,respectively, and the contrast for this is 10 − . at 5 λ/D and 10 − between 12 λ/D and 25 λ/D .The contrast between 5 /D and 12 /D was determined to be below a straight line on a logscale. This mask is useful for surveying unknown exoplanets far from the stars they orbit andfor observations of diffuse targets such as circumstellar disks related to planetary formation in3he infrared wavelength region. The masks were fabricated as free-standing masks, as shown infigure 2, and, consequently, can be used for infrared observations. These free-standing maskswere fabricated in nickel using nano-fabrication technology at HOWA Sangyo Co., Ltd andPhoto Precision Co., Ltd in Japan. The designs cover a 30 mm square, and comprise 10 mmmask patterns in thin nickel surrounded by thicker nickel borders designed to enable the masksto be easily handled. The target thicknesses of the patterned area and the handling area are5 µ m and ≥ µ m, respectively. First, nickel with a target thickness of 5 µ m was grown byelectrolytic plating on a temporary substrate. Then, the handling area with a target thicknessof ≥ µ m was deposited by further electrolytic nickel plating. Finally, the free-standingmask was completed by stripping it from the temporary substrate. The fabrication process forthe freestanding nickel masks is detailed in Enya et al. 2012.The pattern of the basic design of Mask-A contained some isolated and/or ultra-finefeatures, which are not viable for a free-standing mask. Therefore, for Mask-A, we addedbridge structures in a direction that had no effect on the contrast to support and link theisolated and/or ultrafine structures together. The fact that our mask design is based on a 1Dcoronagraph is an important advantage in that such bridge structures can in principle be applied(Enya et al. 2011b). Indeed, simulation showed that the contrast obtained with the amendedmasks was equivalent to the contrast obtained with masks with the basic design. On the otherhand, Mask-C has long arches and fine features not shared by Mask-A and Mask-B (the widthof the narrowest arch and the space between the arches were designed to be 33 µ m and 20 µ m,respectively). Because of this, we had the problem that the mask was irreversibly damagedduring removal from the substrate. However, the problem was finally solved by small changesto the fabrication process conditions for Mask-C (i.e., the mask design was not changed) fromthe original fabrication process described in Enya et al. 2012. Figure 3 shows the instrument used for this work. All the experimental optics werelocated in a clean-room at the Japan Aerospace Exploration Agency/the Institute of Space andAstronautical Science (JAXA/ISAS). The coronagraph optics were set up in a vacuum cham-ber, an experimental platform we call the High dynamic range Optical Coronagraph Testbed(HOCT). We used a Super luminescent Light Emitting Diode (SLED) with a center wavelengthof 650nm and wavelength width of 8 nm for the light source. Light was passed into the chamberthrough a singlemode optical fiber. The beam from the optical fiber was collimated by a 50 mmdiameter BK7 plano-convex lens (SIGMA KOKI Co., Ltd.), and the collimated beam passingthrough the pupil mask was focused by a second planoconvex lens. The pupil mask was set atan angle of θ x = 7 ◦ to the plane perpendicular to the optical axis to remove light reflected fromthe mask (see figure 3 for the definition of the coordinates and θ x ). We used 3.4 relay opticsafter the focal plane. Multi-layer anti-reflection coatings optimized for wavelengths of 400 - 7004 m were applied to both sides of the lens to reduce reflection. Though active wavefront controlhelps to improve contrast, it was not applied in this work in order to evaluate the performanceof the masks themselves. A commercially available cooled CCD camera (BJ-42L, BITRAN)with 2048 × . ± . K (1 σ ). This experimental system is capable of achieving araw contrast of 10 − (Haze et al. 2009; Haze et al. 2011; Haze 2012; Enya et al. 2012).To obtain a high-contrast image, we carried out the following procedure: we measuredthe core and the DR, each of which have different imaging times, separately.When the DR wasmeasured, we obscured the light from the core with a focal plane mask (i.e., a DR-shaped holemask) inserted at the first focal plane after the pupil mask. For measuring the core, we replacedthe DR-shaped hole mask with two neutral density (ND) filters. The transmission through theND filters is wavelength dependent, and at 650nm is 0.016%. The core image of the coronagraphic PSF was obtained with exposure times of 0.3 s and3 s. We inserted two ND filters as previously mentioned. After each imaging process, the lasersource was turned off and a “dark frame” measurement was taken with the same exposure timeand the same neutral density filters. A “raw” coronagraphic image was obtained by subtractingthe dark frame from the image with the laser light on (see left-hand side of figure 4). Theseresults are quite consistent with those expected from theory (see figure 1).
The DR of the coronagraphic image was observed with exposure times of 0.3 s, 3 s, and30 s. A “dark frame” was taken with the same exposure times, and these were then subtractedfrom the DR images with the laser light on. The observed DRs of the raw coronagraphic image,which is the area of the image through the focal plane mask, are shown on the right-hand sideof figure 4. The contrast was obtained by normalizing the observed DRs to the peak of thecore. As shown in figure 5a, contrasts of ∼ − - 10 − and ∼ − for the ranges of 3.3 - 8 λ/D and 8 - 12 λ/D , respectively, were achieved using Mask-A. These experimental contrasts almostreached the designed values. The slightly poorer contrast near the OW A is thought to be dueto contamination from light outside the
OW A . As shown in figure 5b, a contrast of ∼ − close to the center was achieved using Mask-B. The experimental contrast is almost the sameas the designed value. As shown in figure 5c, a contrast of ∼ − - 10 − over an extendedfield of view (5 - 25 λ/D ) was achieved using Mask-C. We also found speckle patterns otherthan diffraction patterns in the DRs. For example, the observed PSF of Mask-C, which has acontrast of ∼ − between 20 λ/D and 23 λ/D , has two ring structures (see figures 4f and 5c).5 . Discussion One of the critical issues for coronagraphy is preventing the pupil of the telescope frombeing obscured. So offaxis telescopes have been specifically designed for space missions spe-cializing in coronagraphye.g., TPF-C (Traub et al. 2006); SEE-COAST (Schneider et al. 2006);PECO (Guyon et al. 2009). On the other hand, the masks developed in this study are designedfor an on-axis telescope, in which the pupil is partially obscured by the secondary mirror andits supporting arms. The experiments with each of these masks described here show that thecontrast is significantly improved compared with non-coronagraphic optics. These masks alsohave the properties of a binary pupil mask coronagraph in that they are robust against point-ing errors and they can be used to make observations over a wide wavelength range. Thesemasks have no limitation to the wavelength range over which they can be used because they arefree-standing. Therefore, the application of these binary pupil mask coronagraphs to a num-ber of different normal, centrally obscured telescopes can help to open up new platforms forcoronagraphy. Many advanced ground-based telescopes (e.g., current 8 - 10m class telescopeslike Subaru, and larger future ones such as TMT, E-ELT) and space telescopes (e.g., SPICA,JWST) with an obscured pupil have the potential to be platforms for coronagraphy over a widewavelength region. The specifications of the masks developed in this study are complemen-tary, and are, therefore, useful for observing Jovian planets located at various distances fromthe central star in the mid-IR wavelength region (Fukagawa, Itoh & Enya 2009; Matsuo et al.2011; Enya et al. 2011a). Obtaining the spectra of Jovian planets is invaluable if we wish tolearn more about planetary formation processes. These complementary masks are also usefulfor studying protoplanetary disks and AGN.
Here, we consider the factors that limited the contrast obtained with the masks inthe coronagraphic experiments. The experimental PSF using Mask-B agreed very well with thedesigned PSF. The experimental PSF using Mask-A matched the designed PSF most closely. Onthe other hand, forMask-C, the discrepancy in contrast between the design and the experimentis significantly larger than for both Mask-A and Mask-B. Moreover, the experimental PSFobtained withMask-C had two ring structures not expected from the design.In the experiment described in the previous section, the pupilmaskwas set at an angle( θ x = 7 ◦ , using the coordinate system defined in figure 3) to the plane perpendicular to theoptical axis in order to eliminate light reflected by the mask. Thus, to enable further discussionof the contrast obtained from the experiment, we examined the effect of tilting the mask.Using Cartesian coordinates, the x and y variables for the 1D coronagraph masks, Mask-A and Mask-B, can be separated, while those for Mask-C, which is rotationally symmetric,6annot. Qualitatively, this suggests that the contrast can be reduced from the optimum bytilting Mask-C. However, a quantitative assessment of the influence of the tilt on the contrastobtained with Mask-C has not yet been done.Therefore, we assumed the masks to be ideally flat and, with a larger tilt angle ( θ y = 27 ◦ ,using the coordinate system defined in figure 3), made projections that took account of theinfluence of the mask thickness of 5 µ m, as shown in figure 6. The coronagraphic PSFs wereobtained from simulation using Fourier transform techniques. The results are shown in figure7. The simulation shows that a tilt angle of 27 ◦ should not affect the contrast obtained notonly with Mask-A and Mask-B but also with Mask-C.Consequently, we conducted additional experiments with the masks tilted at θ y = 27 ◦ (see figure 3). A comparison between the results from experiment and those from simulationwith the masks tilted at 27 ◦ is shown in figure 8. With Mask-A and Mask-B the experimentalresults are almost consistent with the simulations. The slightly poorer contrast near the IW A and the
OW A is thought to be due to the effect of light contamination from outside the
IW A and the
OW A .On the other hand, the experimental results using Mask-C show the contrastto be worse than that expected from the simulation. A deviation from perfect flatness of themask is a possible reason for these results. Our freestanding masks were fabricated by strippingthem from a substrate after forming them on the substrate. Indeed,many trials were neededto succeed in stripping off the long arches and fine structure of Mask-C without it breakingbecause the stripping process for Mask-C is fraught with more difficulty than for either Mask-A or Mask-B. To assess and possibly prevent the effects due to errors in flatness, comparativeexperiments with a similar mask on a substrate would be useful.
4. Conclusion
In this study, we have presented the fabrication and experimental demonstrations offree-standing coronagraphmasks with three complementary designs for telescopes, in which thepupil is partially obscured by a secondary mirror and its support structure. Coronagraphicimages obtained from the experiments weremostly consistent with the design. The masks canbe tilted with little effect on contrast, thereby giving us the freedom to choose the tilt angleto avoid ghosting and to manage stray light. The results are important for allowing general-purpose telescopes not specialized for coronagraphy to be used as platforms for high contrastcoronagraphic observations in future.
Acknowledgement
First of all we are grateful to R. J. Vanderbei and the LOQO solver presented by him(Vanderbei 1999). This work was financially supported by the Japan Science and TechnologyAgency and Grants-in- Aid for Scientific Research (Nos. 24840049 and 22244016) from the7apan Society for the Promotion of Science. We thank T. Ishii from Photo Precision Co., Ltd.,A. Suenaga from HOWA Sangyo Co., Ltd, and their colleagues in each of these companies.
References
Belikov, R., et al. 2007, Proc. SPIE, 6693, 669328Carlotti, A., Vanderbei, R., Kasdin, N. J. 2011, arXiv: 1108.4050Enya, K., et al. 2011a, Adv. Space Res., 48, 323Enya, K., Abe, L. 2010, PASJ, 62, 1407Enya, K., Abe, L., Takeuchi, S., Kotani, T., & Yamamuro, T. 2011b, Proc. of SPIE, 8146, 81460QEnya, K., Abe, L., Tanaka, S, Nakagawa, T., Haze, K., Sato, T., & Wakayama, T. 2008, A&A, 480,899Enya, K., Haze, K., Kotani, T., Abe, L. 2012, PASJ, 64, 123Enya, K., Tanaka, S., Abe, L., & Nakagawa, T. 2007, A&A, 461, 783Fukagawa, M., Itoh, Y., Enya, K. 2009, Proc. of SPICA joint European/Japanese Workshop, p.02006Guyon, O., et al. 2009, Proc. SPIE, 7440, 74400FHaze, K. 2012, Ph.D. thesis, The Graduate University for Advanced StudiesHaze, K., Enya, K., Abe, L., Kotani, T., Nakagawa, T., Sato, T., & Yamamuro, T. 2011, PASJ, 63,873Haze, K., Enya, K., Abe, L., Tanaka, S., Nakagawa, T., Sato, T., Wakayama, T., & Yamamuro, T.2009, Adv. Space Res., 43, 181Jacquinot, P., & Roizen-Dossier, B. 1964, Prog. Opt., 3, 29Kasdin, N. J., Belikov, R., Beall, J., Vanderbei, R. J., Littman, M. G., Carr, M., & Giveon, A. 2005a,Proc. SPIE, 5905, 128Kasdin, N. J., Vanderbei, R. J., Littman, M. G., & Spergel, D. N. 2005b, Appl. Opt., 44, 1117Matsuo, T., Fukagawa, M., Kotani, T., Itoh, Y., Tamura, M., Nakagawa, T., & Enya, K. 2011, Adv.Space Res., 47, 1455Schneider, J., et al. 2006, in Proc. Annual Meeting of the French Soc. Astron. and Astrophys., ed. D.Barret et al. (Montpellier: Universite Montpellier), 429Spergel, D. N. 2001, astro-ph/0101142Tanaka, S., Enya, K., Abe, L., Nakagawa, T., & Kataza, H. 2006, PASJ, 58, 627Traub, W. A., et al. 2006, Proc. SPIE, 6268, 62680TTraub, W. A., & Jucks, K. W. 2002, astro-ph/0205369Vanderbei, R. J. 1999, Optimization methods & software, 11, 485Vanderbei, R. J., Kasdin, N. J., & Spergel, D. N. 2004, ApJ, 615, 555Vanderbei, R. J., Spergel, D. N., & Kasdin, N. J. 2003a, ApJ, 590, 593Vanderbei, R. J., Spergel, D. N., & Kasdin, N. J. 2003b, ApJ, 599, 686 able 1. Required contrast in the mask design.
Mask (type) Separation angle ∗ Contrast(/D)Mask-A (integral 1D) 3.3 † − − ‡ − Mask-B (integral 1D) 1.7 † − ‡ − Mask-C (ring) 5.0 † − .
12 10 − ‡ − ∗ Separation angle for integral 1D mask is defined from the PSF core to the coronagraphic direction. † IW A of the mask design (see text). ‡ OW A of the mask design (see text). ig. 1. Left: the pupil mask designs: (a) Mask-A, (b) Mask-B, and (c) Mask-C. The transmission throughthe black and white regions is 0 and 1, respectively. The diameter of the circle encompassing the trans-missive region is 10mm. Right: the expected (theoretical) PSFs for the pupil masks. ig. 2. Panels (a), (d), and (g) are pictures of Mask-A. Panels (b), (e), and (h) are pictures of Mask-B.Panels (c), (f), and (i) are pictures of Mask-C. We adopted a 30 mm square design with a thicker handlingarea around the patterned region. The size of the patterned region is 10 mm. The thicknesses of thepatterned and handling parts in the design are 5 µ m and ≥ µ m, respectively. ig. 3. Lateral and overhead views of the configuration of the experimental optics. ig. 4. Observed coronagraphic images obtained with a SLED (650 nm). Panels (a), (b), and (c) showthe observed coronagraphic PSFs of Mask-A, Mask-B and Mask-C, respectively. Panels (d), (e), and (f)show the DR images obtained with the focal plane mask. The green arrow indicates the direction of theradial profile. The scale bars correspond to 10 λ/D . ig. 5. Radial profiles of the observed (red line) and the theoretical (black line) coronagraphic PSF. Eachprofile is normalized by the peak intensity. Panel (a): PSF profiles of Mask-A.
IW A is 3.3 λ/D . OW A is12 λ/D . Panel (b): PSF profiles of Mask-B.
IW A is 1.7 λ/D . OW A is 6.2 λ/D . Panel (c): PSF profiles ofMask-C.
IW A is 5 λ/D . OW A is 25 λ/D . ig. 6. Conceptual diagram of the mask for use in simulating a tilted pupil mask. The blue arrow indicatesthe direction of the optical axis. (a) Overhead view of the mask rotated by θ y from a plane perpendicularto the optical axis. (b) Lateral view of the mask rotated by θ y from a plane perpendicular to the opticalaxis. The dark gray region shows the mask thickness of 5 µ m. (c) Projection of the tilted mask (b) on toa perpendicular plane including the influence of the mask thickness of 5 µ m. The transmission throughthe black and white regions is 0 and 1, respectively. The PSF is simulated by using the Fourier transformof the pupil. ig. 7. Results of simulation with the pupil mask tilted (Pupil mask, PSF and Profile). Panels (a), (b),and (c) show simulated projections for the tilted masks (Mask-A, Mask-B, and Mask-C all tilted at ∼ ◦ ).Panels (d), (e) and (f) show the theoretical coronagraphic PSFs from (a), (b) and (c), respectively. Theyellow arrow indicates the direction of the radial profile. Panels (g), (h) and (i) show the PSF profileswhen using the masks tilted at 0 ◦ (black line) and at ∼ ◦ (blue line). There is no reduction in contrastin each case. ig. 8. Radial profiles of the observed (red line) and the theoretical (blue line) coronagraphic PSFs withthe mask tilted at ∼ ◦ . Panels (a), (b), and (c) show the PSF profiles of Mask-A, Mask-B, and Mask-C,respectively.. Panels (a), (b), and (c) show the PSF profiles of Mask-A, Mask-B, and Mask-C,respectively.