Optical Characterization of 1 7 m NIR Detectors

Optical Characterization of 1. 7 m NIR Detectors for SNAP M. Brown (Michigan), J. Balleza (IDTL), N. Barron (Michigan), M. Borysow (Michigan), D. Figer (IDTL), R. Foltz (DCL), B. Hill (DCL), S. Johnson (DCL), W. Lorenzon (Michigan), E. Morse (IDTL), N. Mostek (Indiana), S. Mufson (Indiana), M. Regan (IDTL), M. Schubnell (Michigan), R. Smith(Caltech), G. Tarle (Michigan), A. Tomasch (Michigan), A. Waczynski (DCL), for the SNAP Collaboration Intra-pixel Variations Abstract University of Michigan, Caltech, IDTL, JPL and DCL have been conducting an extensive program to characterize 1. 7 m detectors produced by Rockwell Scientific Company and by Raytheon Vision Systems with R&D funding from the SNAP program. Precision photometry at the ~2% level is a critical driver for the NIR detector system specifications for SNAP. A complete understanding of the quantum efficiency as a function of wavelength, time and temperature is an important element of the calibration of the NIR detector system. Raytheon Hg. Cd. Te detectors grown by Liquid Phase Epitaxy have excellent QE, which is flat from 900 to 1700 nm without removing the Cd. Zn. Te substrate: 80% prior to anti-reflective coating, with 95% promised after applying an AR coating. The MBE Hg. Cd. Te grown by Rockwell achieves similar QE longward of 1300 nm but rolls off at shorter wavelengths. Since SNAP’s diffraction limited images will be under-sampled, intrapixel variation can have a significant impact on photometry. Therefore, we have scanned small sub-arrays with a custom made precision NIR spot projection system, and find that losses at the pixel boundaries are significantly greater for the Rockwell Hg. Cd. Te devices. SNAP continues to fund both manufacturers to improve their performance in the expectation of a large competitive procurement. Please see the companion poster on dark signals and noise. Quantum Efficiency Measurements of SNAP prototype detectors We have tested Rockwell’s (RCS) Hg. Cd. Te, grown by Molecular Beam Epitaxy and hybridized to the HAWAII-2 RG 2 kx 2 k mux, and Raytheon’s (RVS) Hg. Cd. Te grown by Liquid Phase Epitaxy, hybridized to the SB-301 1 kx 1 k (VIRGO) mux. Both are grown on 0. 8 mm thick Cd. Zn. Te substrates. In. Ga. As detectors grown on In. P substrates by Sensors Unlimited have also been tested on the same muxes. The Spot-o-Matic is a NIR pinhole projection system based on an optical LBNL design. The system projects diffraction limited spots (s = 1. 4 m @ 1050 nm, 2. 1 m @ 1550 nm) onto our NIR detectors. It is mounted on a precision XYZ stage, allowing automated focusing and pixel scans. Beam Splitter Monitoring Photodiode Y Dewar Window Light Guide Filter Tube Lens Pinhole XYZ Stage Objective Lens X Z Linear Encoder Intra-pixel Variation Results RSC The plot below shows the integrated response in a 5 x 5 pixel region as a function of spot position for a 5 m FWHM spot. The RVS response is uniform to better than 2% (left plot), while the Rockwell response varies up to 25% as the spot is moved within a given pixel. Uniform response within each pixel is essential for precision photometry without dithering for an under-sampled telescope such as SNAP. RSC, Substrate removed RVS Hg. Cd. Te 141– Integrated Response RSC Hg. Cd. Te#40 – Integrated Response All Hg. Cd. Te devices from Rockwell droop at short wavelengths. The improvement gained (in some cases) by removing the Cd. Zn. Te substrate suggests that this may be partially due to absorption in the Cd. Zn. Te substrate, however the Raytheon LPE process achieves higher QE without such substrate removal. Cutoff Wavelength Shifts in Lattice Matched In. Ga. As Arsenides, such as In. Ga. As, are known to have strongly temperature dependent bandgap Energies. The cutoff wavelength at the SNAP operating temperature of 140 K is <1. 6 m, if the room temperature cutoff is 1. 7 m. QE Variations near Hg. Cd. Te cutoff wavelength QE and flat-field variations for wavelengths <1680 nm are ~ 3 to 4% for both Hg. Cd. Te manufacturers. Variations in the bandgap, and thus the cutoff, may become significant due to inconsistencies in the stoichiometry of the material. For Hg 1 -x. Cdx. Te: Eg = -0. 302 + 1. 93 x – 0. 810 x 2 + 0. 832 x 3 – 5. 35*10 -4(1 - 2 x)*T A comparison of Raytheon and Rockwell Hg. Cd. Te @ 1720 nm shows that the RVS material grown using LPE is much more uniform than the MBE growth used by Rockwell near the cutoff wavelength. RVS Hg. Cd. Te #141 @ 1720 nm RSC H 2 RG Hg. Cd. Te #40 @ 1720 nm RSC H 2 RG # 40– Single pixel response The response for a single pixel as a function of spot position is shown on the right. The structure shown is typical of all pixels measured with the spot projection system at U. Michigan. The sensitive pixel area extends beyond 18 microns, indicating a large amount of diffusion at the pixel boundary. Future development work "This work was supported by the Director, Office of Science, of the U. S. Department of Energy under Contract No. DE AC 03 -76 SF 00098. " The detector tests described here were performed on engineering prototypes. R&D funding will continue to be supplied to both manufacturers to complete the optimization of the manufacturing process and to generate science grade arrays in the near future demonstrating all desirable characteristics simultaneously. The SNAP IR detector labs, in addition to testing these new devices as reported here, will investigate gain and sensitivity stability, limits on accuracy for conventional linearity correction, and the possible dependence of sensitivity on flux, which has been reported by R. Bohlin (STSc. I) based on HST data at ~1 m. Photo courtesy NOAO, taken with LBNL fully depleted CCD