Development of HgCdTe large format MBE arrays and and noise free high speed MOVPE arrays for ground based NIR astronomy G. Finger 1, I. Baker 2, M. Downing1, D. Alvarez 1, D. Ives 1, L. Mehrgan 1, M. Meyer 1, J. Stegmeier 1, H. J. Weller 2 1 European Southern Observatory, GarchingGermany. 2 Selex ES Ltd, Southampton, United Kingdom.
VLT / VLTI Interferometer the spatial resolution of the four 8m telescopes of the VLTI will outperform the ELT 200 m baseline (ELT 40m) fringe tracker IR AO WFS for each telescope high speed low noise sensor required
European Extremely Large Telescope Large format detectors required (4Kx4K)
Armazones Paranal
The E-ELT 40-m class telescope: largest opticalinfrared telescope in the world. Segmented primary mirror. Active optics to maintain collimation and mirror figure. Adaptive optics assisted telescope. Diffraction limited performance. Wide field of view: 10 arcmin. Mid-latitude site (Armazones in Chile). Fast instrument changes. VLT level of efficiency in operations.
FPA arrangement for MICADO
Science Detectors Instrument Number of Det Pixel format Wavelength um MICADO 13-16* 4kx4k 0.8-2.5 HARMONI 8* 4kx4k 0.95-2.45 8 4kx4k 0.47-0.95 METIS 1 1kx1k 5-14(28) 6 2kx2k 3-5 ELT-MOS 6* 4kx4k 0.6-1.7 6-8 4kx4k or 6kx6k 0.37-1 ELT-HIRES 3* 4kx4k 0.8-2.45 5 9kx9k 0.3-0.8 ELT-PCS 4* 4kx4k 0.95-1.65 8 4kx2k 0.6-0.9
Large format LPE / MBE LPE: homojunctions, not possible to vary bandgap panchromatic response, QE lower in J band, cosmetic worse, requires cooling to <60K, historical MBE: heterojunctions, solid state engineering possible, vary bandgap for each molecular layer, solid state engineering High QE > 80% Low dark current: few electrons/hour Low readout noise: DCS < 8 erms Fowler < 3 erms Readout noise limited by IR diode, not by acquisition chain Remaining problem: persistence
Hawk-I mosaic : 2x2 2Kx2K Hwaii-2RG mosaic
Hawk-I mosaic : 2x2 Hwaii-2RG
detector package Hawaii-4RG-15: SiC package four Hawaii-4RG arrays needed for MOONS
Dark current versus temperature measured at ESO generationrecombination limited above 110K I dark exp(-t eff /T) surface leakage and tunneling below 100K T=80 K is sufficient for dark current, but cosmetics improves below 80K
QE comparison of KMOS science arrays QE of all 3 KMOS science arrays systematic error ~10 % all arrays high QE array H2RG #184 has cutoff at l c =2.45 mm
Noise of KMOS arrays for single DCS H2RG #119 (X-Shooter) 25.3 erms on IR active pixels 7.7 erms on reference pixels H2RG #184 (KMOS) 6,9 erms on IR active pixels 5.8 erms on reference pixels Pond pad contact resistance improved noise reduced from 25.3 to 6.9 erms Noise below 10 erms for DCS with all four KMOS arrays
Readout noise with Fowler sampling increase number of nondestructive readouts to reduce readout noise noise 2.8 erms with 64 Fowler pairs for more than 128 Fowler pairs readout noise increases on IR active pixels but not on reference pixels limiting noise process 1/f noise of HgCdTe pixel
Persistence of substrate removed array DIT=1.65s ThAr lamp on
Persistence of substrate removed array first dark exposure with DIT=128s after 2048s exposure with open slit
Persistence electrical /optical Generated with bias change in selected area using global reset Generated with light source
Persistence of Hawaii-2RG in SINFONI and KMOS 2010 2004 KMOS #211 and #212 from same lot but #212 has 6 times higher persistence KMOS #211 and #184 from different lots but similar persistence persistence device specific persistence has to be measured also optically need FIAT test cryostat
Persistence model of Roger Smith traps populated when exposed to mobile electrons and holes pn -junction p n charge trapped when location of trap becomes undepleted and is released in next dark exposure
Persistence model of Roger Smith traps populated when exposed to mobile electrons and holes pn -junction p n charge trapped when location of trap becomes undepleted and is released in next dark exposure
Mitigation of persistence: global reset detrapping traps populated when exposed to mobile electrons and holes pn -junction p n charge trapped when location of trap becomes undepleted and is released in next dark exposure keep global reset switch closed after science exposure allow de-trapping of charge
WFIRST H4RG-10 SiC Mosaic Plate GLS designed and had fabricated the SiC mosaic plate and lightshield for the H4RG-10 mosaic focalplane.
MOVPE (metal organic vapor phase epitaxy) MOVPE: heterojunctions solid state engineering, vary bandgap for each layer on the 0.1μm scale bandgap and doping profiles can be varied independently solid state engineering sucessfully added eapd structure for noise free detectors eapd: applications: AOWFS & fringe tracking : high speed noise free is a mature technology spectroscopy:requires development to reduce dark current and to provide panchromatic response
change from LPE to MOVPE heterostructures solid state engineering MOVPE/mesa technology on GaAs substrate Wide bandgap buffer layer : spectral response limited to 1.3-2.5 mm λ c =2.6 mm absorber at junction widen bandgap : λ c =2.0 mm for reduced gr current and TAT defects narrow bandgap gain region : for high APD gain exp(α λ c ) : λ c =3.5 mm
band diagram of MOVPE homojunction Potential energy Wide bandgap heavily doped contact region Avalanche in 2.5um bandgap material Danger area for dark current and defects n-type gain Depletion region: Jn Jn in 2.5um bandgap material hv Absorber Electrons diffuse to jn p-type absorber photons absorbed in p-type drift region and amplified in n-type gain region homojunction same bandgap for absorber and gain region danger area at junction: crystallographic defects cause excessive dark current at high field due to TAT homojunction MOVPE array was used to develop bonding techniques for SAPHIRA ROIC cosmetic quality limited already at moderate APD gain higher dislocation density
band diagram of MOVPE heterojunction Potential energy Avalanche in narrow bandgap material (3.0µm) Graded region 1.0um hv Jn Jn in wide bandgap material (2.0µm) For reduced g.r. current and TAT defects Must ensure no barrier here heterojunction widen bandgap at junction photons absorbed in p-type and amplified in n-type region electrons experience only partial APD gain compensated by higher bias voltage better breakdown voltage yields higher APD gain Wide bandgap heavily doped contact region Wide depletion (3.5um) for: reduced field across jn max gain for late absorbed photons
MOVPE solid state engineering heterojunction widen bandgap at junction photons absorbed and amplified in n-type region separate absorber and gain region and optimize them independently better breakdown voltage yields higher APD gain higher operating temperature T=85K better cosmetics
K-band flatfield cosmetics of LPE APD gain 42 T detector = 45K T blackbody = 100C 20C DIT=1ms best LPE array (GRAVITY fringe tracker)
K-band flatfield cosmetics of MOVPE APD gain 64 T detector = 85K T blackbody = 100C 20C DIT=1ms only 18 bad pixels at maximum APD gain at operating temperature of T=85K fraction of good pixels: 0.99978 CCD type quality
SAPHIRA window topology reset region lager than window region because of edge effects window reset Programmable windows and reset regions with download of bit stream Wavefront sensor: 96 x72 pixels needed Fowler-12 possible for DIT=1ms fringe tracker: 48 spectra to be read with 24 x 32x1 pixel windows Fowler-90 possible for DIT of 1 ms readout noise reduced by Fowler sampling
symmetric cryogenic preamplifiers additional cooling plane for preamplifiers 320 x256 SAPHIRA SELEX eapd in LCC package cryogenic symmetric preamplifier: well proven design with Aladdin, VIRGO and H2RG arrays OPA 354: gain bandwidth 250MHz noise: 6.5nV/ Hz Preamp gain 3 cryogenic opamps for 32 channels: OPA354
IRATEC test camera filter wheel detector Offner relay f/11 clod filter wheel with bandpass filters: J,H,K test pattern with grid of holes in image plane illuminated by extended blackbody CCC Offner relay
subelectron sensitivity filter H-band single double correlated clamp chop frequency 10 Hz blackbody temperature : on 70C off 20C optics: Offner relay f/11 fluence 1.29 photons / pixel for integration time of 1.17 msec readout mode: double correlated sampling bias voltage 12.15V
subelectron sensitivity filter H-band single double correlated clamp chop frequency 10 Hz blackbody temperature : on 70C off 20C optics: Offner relay f/11 fluence 1.29 photons / pixel for integration time of 1.17 msec readout mode: double correlated sampling bias voltage 12.15V
quantum efficiency of MOVPE heterostructures MARK5 MARK3 MARK3&5 K-band H-band MARK3 at 85K: QE 65% in K QE 56 % in H QE higher in K than in H because in K photons are absorbed closer to the junction less distance to diffuse to the depletion region MARK5 at 100K: QE 82% in K QE 67% in H reduced doping does not affect drop of QE at low temperatures
APD gain MOVPE heterostructure / LPE H MARK3 K LPE MARK5 APD gain of MARK3 is 76 in H and 64 in K: larger than that of LPE APD gain of MARK3 in H-band larger than in K-band H-band: full APD gain K-band: partial APD gain because photons absorbed in gain region Increased gain of MOVPE heterostructure because of narrow bandgap in gain region (l c =3.5 mm ) low APD gain of MARK5 due to voltage drop in p-region, this reduces available voltage drop in gain region
excess noise factor F MARK5 LPE K MARK3 H excess noise factor F F is the factor by which the APD gain increases the noise MARK3: low noise factor F H-band F=1.2 at APD gain 74 K-band F=1.35 at APD gain 64 K-band photons penetrate into APD gain region absorption depth is random process with random APD gain increases noise figure F MARK5:high noise factor F electrons have spread of kinetic energy before entering gain region because of trapping and scattering
Dark current [e/s] dark current of MOVPE heterostructure 100 Dark current of MOVPE eapd at bias = 10V Signal [e] 80 60 40 20 bias voltage across diode 10V APD gain=30 dark current depends on VDD electroluminescence of mux: smaller VDD less glow Better glow protection in next multiplexer design 0 5.0 5.2 5.4 5.6 5.8 6.0 6.2 6.4 VDD [V]
sky flux J-band Paranal advanced sky model Noll & Jones A&A, Austrian in kind contribution Nyquist sampling of Airy disk Blue: sky Black: sky and filter J Flux: 14.5 phot/s/pixel
sky flux H-band Nyquist sampling of Airy disk Blue: sky Black: sky and filter H Flux: 116 phot/s/pixel
sky flux K-band Nyquist sampling of Airy disk Blue: sky Black: sky and filter K telescope & instrument at T=15C ε= 0.15 Flux: 231 phot/s/pixel
towards panchromatic response Today: GaAs etched off CdTe 0.8um Widebandgap buffer layer 1.3um in ESO APDs SW absorber Junction Gain region
towards panchromatic: Option 1 1.0um, easy processing CdTe 0.8um Widebandgap buffer layer 1.0um cutoff SW absorber Junction Gain region Need to ensure MOVPE growth is stable giving low hillocks.
towards panchromatic response Option 2-0.8um, easy processing Thicker CdTe 0.8um SW absorber Junction Gain region Need to ensure MOVPE growth is stable giving low hillocks. Also need to ensure quality in device layers
towards panchromatic response Option 3 - visible, hard processing Buffer layers etched off with chemical etch SW absorber Junction Gain region Etch uniformity hard to control also difficult surface passivation after etch. Device is fragile
Conclusions large format 4Kx4K MBE arrays fulfill most of the requirements of 8 meter class and future ELT telescopes persistence of these arrays is still an unsolved problem aggravated by the high flux due to the large collecting area of the ELT and due to diffraction limited performance of it s adaptive optics NIR electron avalanche photodiodes grown by MOVPE adequate to deploy noise free high speed detectors in wavefront sensors and fringe trackers can already be used for noise free high time resolution imaging in H and K band and have already been used for NIR lucky imaging further development is needed for high resolution spectrographs with panchromatic response and dark current levels compliant with flux levels of ~ 10-2 photons/s/pixel
test pattern at 36 C MOVPE eapd Filter Ks l c =2.65 mm HgCdTe eapd bias voltage=7.1v APD gain=7.34 T=60K DIT=7.6 ms, BW=5MHz
linearity at high bias with APD gain of 28 28.00 plot of detector signal voltage versus fluence detector linearity at high bias APD gain of 28 non-linearity increases at high APD gain APD gain changes during integration
dark current of MOVPE heterostructure Signal [e] 2000 1500 1000 500 Dark current MOVPE eapd at bias = 10V VDD = 6.40 Idark = 94.6 [e/s] VDD = 6.20 Idark = 84.8 [e/s] VDD = 6.00 Idark = 75.0 [e/s] VDD = 5.80 Idark = 65.7 [e/s] VDD = 5.60 Idark = 56.7 [e/s] VDD = 5.40 Idark = 51.4 [e/s] VDD = 5.20 Idark = 47.9 [e/s] VDD = 5.00 Idark = 44.3 [e/s] dark current depends exponentially on bias voltage dark current amplified by APD gain at integrating node dark current back referred to absorber region before APD gain (divide by APD gain) dark current may be dominated by instrumental photon background of camera (l c =3.5 mm in gain region) dark current depends on VDD dark current 0.044e/ms/pix negligible for integration times up to 30ms at T = 85K 0 0 2 4 6 8 time [s]