An innovative integral field unit upgrade with 3D-printed micro-lenses for the RHEA at Subaru
Theodoros Anagnos, Pascal Maier, Philipp Hottinger, Chris Betters, Tobias Feger, Sergio G. Leon-Saval, Itandehui Gris-Sánchez, Stephanos Yerolatsitis, Julien Lozi, Tim A. Birks, Sebastian Vievard, Nemanja Jovanovic, Adam D. Rains, Michael J. Ireland, Robert J. Harris, Blaise C. Kuo Tiong, Olivier Guyon, Barnaby Norris, Sebastiaan Y. Haffert, Matthias Blaicher, Yilin Xu, Moritz Straub, Jörg-Uwe Pott, Oliver Sawodny, Philip L. Neureuther, David W. Coutts, Christian Schwab, Christian Koos, Andreas Quirrenbach
AAn innovative integral field unit upgrade with 3D-printedmicro-lenses for the RHEA at Subaru
Theodoros Anagnos a,b,c , Pascal Maier d,e , Philipp Hottinger c , Christopher H. Betters i , TobiasFeger a,j , Sergio G. Leon-Saval i , Itandehui Gris-S´anchez g,k , Stephanos Yerolatsitis g , JulienLozi h , Tim A. Birks g , Sebastian Vievard h , Nemanja Jovanovic l , Adam D. Rains o , Michael J.Ireland o , Robert J. Harris c,p , Blaise C. Kuo Tiong a,b , Olivier Guyon h , Barnaby Norris i ,Sebastiaan Y. Haffert m,n , Matthias Blaicher d,e , Yilin Xu d,e , Moritz Straub q , J¨org-Uwe Pott p ,Oliver Sawodny q , Philip L. Neureuther q , David W. Coutts a,b , Christian Schwab a,b , ChristianKoos d,e,f , and Andreas Quirrenbach ca Department of Physics and Astronomy, Macquarie University, NSW 2109, Australia b MQ Photonics Research Centre, Department of Physics and Astronomy, MacquarieUniversity, NSW 2109, Australia c Landessternwarte, Zentrum f¨ur Astronomie der Universit¨at Heidelberg, K¨onigstuhl 12, 69117Heidelberg, Germany d Institute of Microstructure Technology (IMT), Karlsruhe Institute of Technology (KIT),Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany e Institute of Photonics and Quantum Electronics (IPQ), Karlsruhe Institute of Technology(KIT), Engesserstr. 5, 76131 Karlsruhe f Vanguard Photonics GmbH, Hermann-von-Helmholtz-Platz 1,76344Eggenstein-Leopoldshafen, 76227 Karlsruhe g Department of Physics, University of Bath, Claverton Down, Bath, BA2 7AY, UK h National Institutes of Natural Sciences, Subaru Telescope, National AstronomicalObservatory of Japan, Hilo, Hawaii, United States i University of Sydney, Sydney Institute for Astronomy, Institute for Photonics and OpticalScience, School of Physics, Camperdown, Australia j Redback Systems Pty Ltd, Sydney, Australia k ITEAM Research Institute, Universitat Polit`ecnica de Val`encia, Camino de Vera, 46022Valencia, Spain l California Institute of Technology, 1200 E. California Blvd., Pasadena CA, 91125, USA m Leiden Observatory, Leiden University, PO Box 9513, Niels Bohrweg 2, 2300 RA Leiden, TheNetherlands n Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, Arizona o Research School of Astronomy and Astrophysics, Australian National University, Canberra,ACT 2611, Australia p Max-Planck-Institute for Astronomy, K¨onigstuhl 17, 69117, Heidelberg, Germany q Institute for System Dynamics, University of Stuttgart, Waldburgstr. 19, 70563 Stuttgart,Germany
ABSTRACT
In the new era of Extremely Large Telescopes (ELTs) currently under construction, challenging requirementsdrive spectrograph designs towards techniques that efficiently use a facility’s light collection power. Operating in
Further author information: (Send correspondence to Th.A.)E-mail: [email protected] a r X i v : . [ a s t r o - ph . I M ] J a n he single-mode (SM) regime, close to the diffraction limit, reduces the footprint of the instrument compared to aconventional high-resolving power spectrograph. The custom built injection fiber system with 3D-printed micro-lenses on top of it for the replicable high-resolution exoplanet and asteroseismology spectrograph (RHEA) atSubaru in combination with extreme adaptive optics of SCExAO, proved its high efficiency in a lab environment,manifesting up to ∼
77% of the theoretical predicted performance.
Keywords: astrophotonics, spectroscopy, micro-lenslets, SCExAO, radial velocity, optical fibers, fiber injection,diffraction-limited spectrograph, integral field unit
1. INTRODUCTION
A wealth of crucial information can be collected through astronomical spectroscopy, such as chemical composition,motion parameters as well as the indirect discovery of celestial bodies in orbit around other stars. Conventional spectrograph designs began to make use of fibers half a century ago
2, 3 in order to enable moreefficient observations, as it became possible to locate the instrument off the telescope. Soon after, fiber basedintegral field unit (IFU) systems were developed that allowed flexibility in arranging spectra on a given detectorspace. Early on, multi-mode fibers (MMFs) were used for the IFU, which had high throughput for seeing-limitedstarlight (e.g. Ref. 5, 6). Afterwards, new designs of IFU systems emerged taking advantage of single-modefibers (SMFs) (e.g. Ref. 7, 8).Using SMFs and operating in the diffraction limit reduces the footprint of the instrument, however majorlimitations apply in coupling efficiency under seeing-limited conditions. By making use of the extreme adaptiveoptics (ExAO) systems installed in state-of-the-art telescopes, the coupling efficiency gets significantly better(e.g. Ref. 9, 10).While high-spatial resolution spectroscopy is achieved by using SM-IFUs, giving access to many new sciencecapabilities, the coupling losses are high due to the low fill fraction of SMFs and the requirement for sub- µ mprecision in alignment.In this study, we present an upgrade of the IFU system on the RHEA at Subaru.
11, 12
This custom IFU makesuse of a multi-core fiber (MCF) with 19 SM cores, with 3D-printed micro-lenses on top of the cores manufac-tured by the two-photon polymerization lithography technique.
13, 14
This custom injection system significantlyincreases the free-space coupling of starlight into the fiber cores while allowing more tolerance for misalignmenterrors in targeting. The IFU system is optimized using
Zemax optical software for instantaneous angular skyareas of 11 and 18 milli-arcseconds (mas) per lenslet. The system also offers a relatively high coupling efficiencyand fill factor due to the 3D-printed micro-lens array (MLA).In Section 2 we present the core design and parameters, complemented by the detailed experimental designdescription for the characterization of its performance. In Section 3 the laboratory results are presented. Wedraw conclusions in Section 4 and detail our future plans in Section 5.
2. METHODS
To increase the efficiency of light coupling from the 8-m Subaru telescope into the IFU feeding the RHEA, thefollowing components are necessary: the Subaru Coronagraphic Extreme Adaptive Optics (SCExAO) system,the IFU itself with the 3D printed MLAs, the MCF and the spectrograph adapted to the output of the MCF.Below, these components are presented in more detail.
In order to simulate the intensity distribution in the entrance of the IFU system, all the optical componentsof the SCExAO were taken into account. First of all, the output beam of the 8-m Subaru telescope undergoesadaptive optics (AO) correction in the AO188 unit and is then routed to the visual bench of SCExAO. Thebeam intensity distribution at this stage can be described as an Airy pattern.For an efficient coupling of the Airy disk into the SM cores of the fiber, a special injection unit is necessary.This unit is specially designed to match the specifications of the MCF. The MCF has a 6.1 µ m mode-field igure 1. MCF end face after polishing and packaging into an FC/PC connector. The cores are positioned in a hexagonalformation with a pitch of 30 µ m. Each of the cores has a 6.1 µ m MFD at 600 nm (1 / e ). A HeNe laser source was usedto back illuminate the fiber cores, which are labeled with red numbers for referencing. diameter (MFD) at 650 nm (1 / e ) where the SM cut-off limit is at ∼
600 nm (Figure 1 shows a microscope imageof the MCF end face). The cores have a pitch of 30 µ m in hexagonal formation, which leads to a ratio of 4.9:1,giving enough separation to eliminate the cross-coupling between cores. The 3D printed structure of the MLAsis applied directly on top of the fiber end face using the two-photon lithography technique described below. Thiscustom IFU will be installed into RHEA to maximize the system potential. The structure of the MLAs wasoptimized for a platescale of 11 and 18 mas on the sky per lenslet given that the diffraction limit of the Subarutelescope with the SCExAO at 650 nm is 17 mas. The SCExAO is located in the near-infrared (NIR) Nasmyth focus of the 8 m Subaru Telescope. A completeand far more detailed description of both the visual and NIR paths of SCExAO is provided in Ref. 15. A briefoverview starts with the starlight entering the Subaru Telescope and a 30-40% Strehl correction on the pointspread function (PSF) for the H-band accomplished by AO188.
The starlight then enters the SCExAO NIRbench, where the wavefront is further corrected for higher-order aberrations caused by the atmosphere. In thenext step, a dichroic filter separates the light beam in two paths: the visible ( <
900 nm) and the NIR ( >
900 nm)channel. Finally, the light beam is focused down with optical lenses onto the 3D printed MCF surface (see Figure2).
For our simulations, an Airy disk was used an an input to the IFU system, taking into account the Subarutelescope profile and the key components of SCExAO.
9, 15
The simulation of the the Airy profile was performed by using the physical-optics propagation (POP) moduleof
Zemax . This Airy disk output feeds a performance optimizer that selects for the best MLA structure tobe 3D printed on top of the MCF. POP operands were used for varying MLA geometries that affect the Airy
Periscope WFS beamsplitter
VISIBLE BENCH
To FIRST recombination bench
RHEA pickoff mirror Single-mode IFUMicro lensesAlignment camera
To RHEA spectrograph
RHEA focusing lensField stop
VAMPIRESFIRST INJECTIONRHEA
Calibration fiber
VAMPIRES/FIRST splitter
Figure 2. The visual bench of the SCExAO. The IFU injection is located in the bottom center of the illustration. Theinput beam from AO188 to the periscope to the IFU is represented with green and brown color. beam coupling into the MCF cores. Several geometrical shapes were tested for the MLA structure in orderto increase the coupling efficiency into the cores. Finally, a spherical surface was selected, as the alternativesprovided negligible improvement in performance. A spherical surface MLA of 272 µ m in height and 115 µ m inradius of curvature achieved the highest coupling efficiency for the wavelength range of 600-800 nm, resulting ina throughput of 50% for the 18 mas platescale for the central lens and a throughput of 21.9% for the platescale of11 mas for the central lens. To achieve a fill factor of ∼ λ/ λ/
21 at the working wavelength, as a surface roughness of 37 nm is achieved using the3D printing technique. The MLA was printed to the cleaved facet of the MCF as a single model block using the commercially avail-able negative-tone photoresist IP-Dip and an in-house built two-photon lithography machine. This system isequipped with a 780 nm femtosecond laser and a 40x Zeiss objective lens with numerical aperture (NA) = 1.4.A custom control software was developed in-house to guarantee optimum shape-fidelity of the printed MLA andallow for high-precision alignment with respect to the fiber cores of the MCF.As a first step, the MCF was manually glued to an FC-PC connector and subsequently polished to achieve aflat fiber end-facet accessible for the lithography machine for printing. Thereafter, the fiber is back-illuminatedby coupling in the light of a red light-emitting diode (LED) to accommodate machine vision for the detectionof the 19 cores of the MCF. After the detection procedure, the individual lenses of the MLA are aligned withrespect to the detected core positions of the MCF, thereby taking into account variations of the core positionsand pitch. All individually positioned lenses are then merged into a single 3D-model of the MLA to preventunnecessary double-illumination in the overlap regions during the printing process. The structure is furtherautomatically adapted to compensate for any tilt of the fiber end-facet. For the purpose of reducing the requiredprinting time, the MLA is divided into two parts: the first block of the model up to just below the lens surfaceswas written with a distance between subsequent layers, i.e., slicing distance, of 600 nm. For optimal printingquality, the remaining second model block comprising the 19 lens surfaces of the MLA was written with a slicingdistance of 100 nm. The writing distance between subsequent lines, i.e., hatching distance, was set to 100 nmthroughout the full model. The fabricated structure was afterwards developed in propylene-glycol-methyl-ether-acetate (PGMEA), flushed with isopropanol, and subsequently blow dried. In the next stage, scanning electronmicroscopy (SEM) and vertically-scanned white-light interferometry (VSI) images of the structure were acquiredto check the quality of the manufacturing process (see Figure 3). igure 3. The top structure of the 3D printed MLA on top of the MCF end face acquired using the SEM technique. For measuring the throughput performance of the custom 3D-printed MCF, a set of opto-mechanical parts wasconstructed as an addition to the K¨onigstuhl Observatory Opto-mechatronics Laboratory (KOOL) test-bed. Thethroughput setup is presented in Figure 4. The HeNe laser light (632 nm) passes through a 50:50 non-polarizingbeamsplitter (BS) L1 (Thorlabs CM1-BS014) and is collimated using an achromat lens L2 (Thorlabs AC127-025-B-ML). Later on, the beam is split using another 50:50 non-polarizing BS L4 (Thorlabs CM1-BS014). One beamis focused down using an 100 mm achromat (AC254-100-B-ML) to the CMOS detector (Thorlabs DCC1545M),and the other is routed through a flip mirror L6 to an achromat L7 (AC254-060-B-ML or AC254-100-B-MLdepending on the platescale) and focused down to the 3D-printed MCF that is mounted onto a 4-axis mount(Thorlabs MBT401D). After that, the fiber exit is re-imaged using a combination of achromats L8 and L9(AC127-019-B-ML and AC127-050-B-ML) to the CMOS detector (Thorlabs DCC1545M). To calculate the totalthroughput of the MCF including the the coupling losses, a power meter was used (Thorlabs S120C) in order toperform the measurements and calibrate absolute flux through the re-imaging system.
3. RESULTS3.1 Throughput efficiency results
To asses the performance of the custom IFU system before installation into RHEA, laboratory tests were per-formed as described in section 2.5. The outcome of these tests are presented here.As mentioned in section 2.5 the throughput measurements were monochromatic using a HeNe laser at 632 nm.After a series of optical elements in the KOOL test-bed, the beam was focused down to the 3D-printed MCF.Data frames with exposure times of a fraction of a second were collected with both achromats L7 (AC254-060-B-ML or AC254-100-B-ML), using the setup described above. The setup was able to sample the near-field of theMCF exit and filter the light between the adjacent cores of the fiber. Averaged dark data frames were recordedas well, and subtracted from the data for further processing.To characterize the absolute performance of the IFU, two separate experiments were conducted; in the first,the total coupling efficiency of the light into each of the 19 cores was determined after the alignment of the coreson-axis with the injected beam. In the second experiment, the misalignment tolerances of the injected beamwere measured by translating the injected beam by steps of 5 µ m in respect to the central core of the MCF.This was more representative of realistic on-sky conditions where the star would be moving due to atmosphericperturbations. igure 4. The throughput experimental setup for measuring the efficiency of the custom IFU. A power meter is usedto calibrate the throughput. A set of achromat lenses (L2-L5-L7) is used for collimation and focusing of the beam,beamsplitter (BS) and CMOS detector (D1, D2) for imaging the near-field output of the MCF (L8, L9). The results of the first experiment for the 18 mas platescale are shown in Figure 6. The average couplingefficiency was 21.45 ±
3% with a maximum of 30.89 ±
3% for the core
Zemax . For the 11 mas platescale, thecoupling efficiency of the central core was 16.9 ±
3% (77.2% of the simulated value). The total throughput fromall of the cores summed was calculated, representative of an unresolved target, and was measured to be 41.5 ± ∼ ∼ µ m, retaining ∼
40% of the maximum throughput.
4. CONCLUSIONS
In this work we presented a novel IFU system upgrade for RHEA at the Subaru telescope. This IFU is composedof a custom MCF with 3D printed micro-lenses on top of the cores to increase the coupling efficiency for off andon-axis targets from SCExAO at the Subaru 8 m telescope. The IFU system is optimized using the
Zemax
POPmodule for an on-sky angular dimension of 11 and 18 mas using an Airy profile beam produced by the visualarm of SCExAO.The custom MCF is composed of 19 cores in the same cladding with a core-to-core spacing of 30 µ m anda 6.1 µ m MFD at 650 nm (1 / e ) which leads to a negligible cross-coupling between the cores. The cores arepositioned in a hexagonal formation and their cut-off SM limit is above 600 nm.The structure of the MLA was manufactured with two-photon lithography and 3D-printed on top of the coresof the MCF, significantly enhancing the throughput of light into the fiber cores from levels of few percent to amaximum of 30.89 ±
3% for on-axis targets for a platescale of 18 mas. Furthermore, the custom MLA improvedthe off-axis light losses even for a 10 µ m lateral injection. The throughput performance across all the MLA asa representative of a single unresolved target was 41.5 ± ∼
77% of thesimulated results. The difference in throughput performance from simulated results are likely associated with theimperfectly polished MCF, Fresnel reflections ( ∼ igure 5. Left panel:
2D image of the injected beam in logarithmic color scale for better clarity, as measured in thelaboratory.
Right panel : Intensity profile of the injected point spread function normalized to its maximum, from the2D image data. T h r o u g h p u t Measured in lab
Figure 6. Coupling efficiency for all of the cores of the MCF (see Figure 1 for the numbering of the cores). This is shownfor the 18 mas platescale.igure 7. Coupling efficiency of the central core of the MCF as a function of off-axis target, compared with the simulateddata from
Zemax . Results are normalized to the maximum coupling efficiency including the errors (smaller than the datapoints).
5. FURTHER WORK
Plans for further work include a separate 3D-printed MCF, optimized for only the 18 mas platescale with
Zemax simulations. Both IFU systems will be tested on the KOOL test bed including the effect of atmospheric turbulenceusing the AO system of the KOOL infrastructure.Future work will be to integrate the fibers into the RHEA and perform on-sky tests on a variety of targets(resolved, un-resolved stars, confirmed exoplanets, spectroscopic standard stars and double star systems) in orderto probe its scientific potential.
ACKNOWLEDGMENTS
T.A. is a fellow of the International Max Planck Research School for Astronomy and Cosmic Physics at theUniversity of Heidelberg (IMPRS-HD) and is supported by the Cotutelle International Macquarie UniversityResearch Excellence Scholarship. P.M., M.B., Y.X. and C.K. are supported by Bundesministerium f¨ur Bildungund Forschung (BMBF), joint project PRIMA (13N14630), the Helmholtz International Research School forTeratronics (HIRST), Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’sExcellence Strategy via the Excellence Cluster 3D Matter Made to Order (EXC2082/1-390761711). R. J. H. andP.H. are supported by the Deutsche Forschungsgemeinschaft (DFG) through project 326946494, ’Novel Astro-nomical Instrumentation through photonic Reformatting’. T.B. & S.Y. are supported from the European Union’sHorizon 2020 grant 730890, and from the UK Science and Technology Facilities Council grant ST/N000544/1.S.Y.H. is supported by the NASA Hubble Fellowship grant
Python package forAstronomy,
23, 24
Numpy and Matplotlib. Furthermore, this publication makes use of data generated at the¨onigstuhl Observatory Opto-mechatronics Laboratory (KOOL) which is run at the Max-Planck-Institute forAstronomy (MPIA, PI J¨org-Uwe Pott, [email protected]) in Heidelberg, Germany. KOOL is a joint project of theMPIA, the Landessternwarte K¨onigstuhl (LSW, Univ. Heidelberg, Co-I Philipp Hottinger), and the Institutefor System Dynamics (ISYS, Univ. Stuttgart, Co-I Prof. Oliver Sawodny). KOOL is partly supported by theGerman Federal Ministry of Education and Research (BMBF) via individual project grants.
REFERENCES [1] P. Massey and M. M. Hanson, “Astronomical Spectroscopy,”
Planets, Stars and Stellar Systems. Volume2: Astronomical Techniques, Software and Data , 35 (2013).[2] E. N. Hubbard, J. R. P. Angel, and M. S. Gresham, “Operation of a long fused silica fiber as a link betweentelescope and spectrograph.,”
ApJ , 1074–1078 (May 1979).[3] J. R. Powell, “Application of optical fibres to astronomical instrumentation.,”
Proc. SPIE , 77–84 (1984).[4] J. Allington-Smith, “Basic principles of integral field spectroscopy,”
New A Rev , 244–251 (June 2006).[5] J. Ge, J. R. P. Angel, and J. C. Shelton, “Optical spectroscopy with a near-single-mode fiber-feed andadaptive optics,” Proc. SPIE , 253 – 263 (1998).[6] S. M. Croom, J. S. Lawrence, J. Bland-Hawthorn, et al. , “The Sydney-AAO Multi-object Integral fieldspectrograph,”
MNRAS , 872–893 (Mar. 2012).[7] S. G. Leon-Saval, C. H. Betters, and J. Bland -Hawthorn, “The Photonic TIGER: a multicore fiber-fedspectrograph,”
Proc. SPIE , 84501K (2012).[8] M. Tamura, H. Suto, J. Nishikawa, et al. , “Infrared Doppler instrument for the Subaru Telescope (IRD),”
Proc. SPIE , 84461T (2012).[9] N. Jovanovic, C. Schwab, O. Guyon, et al. , “Efficient injection from large telescopes into single-mode fibres:Enabling the era of ultra-precision astronomy,”
A&A , A122 (Aug. 2017).[10] O. Guyon, R. Belikov, E. Bendek, et al. , “Wavefront Sensing and Control R&D on the SCExAOTestbed,”
American Astronomical Society , 280.06 (Jan. 2020).[11] T. Feger, C. Bacigalupo, T. R. Bedding, et al. , “RHEA: the ultra-compact replicable high-resolution exo-planet and Asteroseismology spectrograph,” Proc. SPIE , 91477I (Aug 2014).[12] A. D. Rains, M. J. Ireland, N. Jovanovic, et al. , “Precision single mode fibre integral field spectroscopy withthe RHEA spectrograph,”
Proc. SPIE , 990876 (Aug 2016).[13] P.-I. Dietrich, R. J. Harris, M. Blaicher, et al. , “Printed freeform lens arrays on multi-core fibers for highlyefficient coupling in astrophotonic systems,”
Optics Express , 18288 (jul 2017).[14] P. Hottinger, R. J. Harris, P.-I. Dietrich, et al. , “Micro-lens array as tip-tilt sensor for single-mode fiber cou-pling,” in [ SPIE Astronomical Telescopes+ Instrumentation ], International Society for Optics and Photonics(2018).[15] N. Jovanovic, F. Martinache, O. Guyon, et al. , “The Subaru Coronagraphic Extreme Adaptive OpticsSystem: Enabling High-Contrast Imaging on Solar-System Scales,”
PASP , 890 (Sept. 2015).[16] Y. Hayano, H. Takami, O. Guyon, et al. , “Current status of the laser guide star adaptive optics system forSubaru Telescope,”
Proc. SPIE , 701510 (2008).[17] Y. Hayano, H. Takami, S. Oya, et al. , “Commissioning status of Subaru laser guide star adaptive opticssystem,”
Proc. SPIE , 77360N (2010).[18] Y. Minowa, Y. Hayano, S. Oya, et al. , “Performance of Subaru adaptive optics system AO188,”
Proc.SPIE , 77363N (2010).[19] Zemax, “Opticstudio - zemax,” (2016).[20] P.-I. Dietrich, M. Blaicher, I. Reuter, et al. , “In situ 3d nanoprinting of free-form coupling elements forhybrid photonic integration,”
Nature Photonics (4), 241–247 (2018).[21] Nanoscribe GmbH, “Ip photoresists,” (2018).[22] Menlo Systems GmbH, “C-fiber 780 femtosecond erbium laser,” (2020).[23] Astropy Collaboration, T. P. Robitaille, E. J. Tollerud, et al. , “Astropy: A community Python package forastronomy,” A&A , A33 (Oct. 2013).24] A. M. Price-Whelan, B. M. Sip˝ocz, H. M. G¨unther, et al. , “The Astropy Project: Building an Open-scienceProject and Status of the v2.0 Core Package,” AJ , 123 (Sept. 2018).[25] S. van der Walt, S. C. Colbert, and G. Varoquaux, “The NumPy Array: A Structure for Efficient NumericalComputation,” Computing in Science and Engineering , 22–30 (Mar. 2011).[26] J. D. Hunter, “Matplotlib: A 2D Graphics Environment,” Computing in Science and Engineering9