Eli E. Gordon

Direct conversion CdZnTe and CdTe detectors for digital mammography

Hybrid CdZnTe and CdTe pixel detector arrays with 50 /spl times/ 50 micron/sup 2/ pixel sizes that convert X-rays directly into charge signals are under development at NOVA for applications to digital mammography. CdZnTe and CdTe have superior X-ray quantum efficiency compared to either emulsion-based film, phosphor-based detectors or other low-Z, solid-state detectors such as silicon. In this paper, latest results from thin (0.15 to 0.2 mm) CdZnTe and CdTe detectors will be presented in terms of modulation transfer function (MTF), detective quantum efficiency (DQE), and phantom images. Single-crystal CdZnTe detectors yield better results in DQE as well as phantom images, compared to the polycrystalline CdZnTe detectors. This is due to the nonuniformities in the polycrystal that degrade the charge transport properties. Because of the charge-coupling limitation of the readout ASIC that was originally designed for Si detectors, the detector is biased to collect holes from the front side. This charge collection mode limits the CdZnTe detector performance. Their DQE measurements yield 25 % and 65 % for the poly-crystal and single-crystal CdZnTe detectors respectively. Poly-crystal CdTe test detectors were also hybridized to the same type charge readout chip. Since CdTe has much longer hole-propagation lengths compared to CdZnTe, it shows better performance in the hole-collecting mode. However, a severe polarization effect degrades performance of the present device. Excellent images were also obtained from the CdTe detectors. Future work to redesign the readout ASIC and, thus, improve the detector performance is discussed. Application to industrial imaging such as nondestructive evaluation (NDE) and nondestructive inspection (NDI) is a natural extension.

640 × 512 17 μm microbolometer FPA and sensor development

RVS has made a significant breakthrough in the development of a 640 x 512 uncooled array with a unit cell size of 17 μm x 17 μm, and performance approaching that of the 25μm arrays. The successful development of this array is the first step in achieving mega-pixel formats. This FPA is designed to ultimately achieve performance of (<50mK, f/1, 30 Hz) with an 8 msec time constant. The SB-400 is a highly productized ROIC and is designed to achieve very good sensitivity (low NETD and low spatial noise) and good dynamic range. The improved performance is through bolometer structure improvements and an innovative ROIC design. It also has a simple and flexible electrical interface which allows external electronics to be small, lightweight, low-cost, and low-power. Almost all adjustments can be made through the serial interface; hence there is no need for external adjustable (DAC) circuitry. The improved power supply rejection helps maintain highly stable detector and strip resistor bias voltages which helps reduce spatial noise and image artifacts. The combination of reduced FPA pixel size and improved effective thermal sensitivity enhances weapon sight performance by providing smaller, lighter-weight sights via reduced optics size or increased range via enhanced pixel resolution without increasing mass or increased range via improved NETD (lower f/#) without increasing mass. We will also provide an update on the enhanced performance and yield producibility of our NVESD ManTech 640 x 480 25μm arrays. We will also show the improvement in our uncooled common architecture electronics in terms of reduced power and size for helmet and rifle mounted sensors and a variety of missile applications.

Mbe growth of HgCdTe avalanche photodiode structures for low-noise 1.55 μm photodetection

Molecular-beam epitaxy (MBE) has been utilized to fabricate HgCdTe heterostructure separate absorption and multiplication avalanche photodiodes (SAM-APD) sensitive to infrared radiation in the 1.1-1.6 μm spectral range, as an alternative technology to existing III-V APD detectors. Device structures were grown on CdZnTe(211)B substrates using CdTe, Te, and Hg sources with in situ In and As doping. The composition of the HgCdTe alloy layers was adjusted to achieve both efficient absorption of IR radiation in the 1.1-1.6 μm spectral range and low excess-noise avalanche multiplication. The Hg 1-x Cd x Te alloy composition in the gain region of the device, = 0.73, was selected to achieve equality between the bandgap energy and spin-orbit splitting to resonantly enhance the impact ionization of holes in the split-off valence band. The appropriate value of this alloy composition was determined from analysis of the 300 K bandgap and spin-orbit splitting energies of a set of calibration layers, using a combination of IR transmission and spectroscopic ellipsometry measurements. MBE-grown APD epitaxial wafers were processed into passivated mesa-type discrete device structures and diode mini-arrays using conventional HgCdTe process technology. Device spectral response, dark current density, and avalanche gain measurements were performed on the processed wafers. Avalanche gains in the range of 30-40 at reverse bias of 85-90 V and array-median dark current density below 2 x 10 -4 A/cm 2 at 40 V reverse bias have been demonstrated.

A CdZnTe slot-scanned detector for digital mammography

A new high-resolution detector has been developed for use in a slot-scanned digital mammography system. The detector is a hybrid device that consists of a CCD operating in time-delay integration mode that is bonded to a 150-microm-thick CdZnTe photoconductor array. The CCD was designed with a detector element pitch of 50 microm. Two devices were evaluated with differing crystalline quality. Incomplete charge collection was a source of reduction in DQE. This occurs in both devices due to characteristically low mobility-lifetime products for CdZnTe, with the greatest losses demonstrated by the multicrystalline sample. The mobility-lifetime products for the multicrystalline device were found to be 2.4 x 10(-4) and 4.0 x 10(-7) cm2/V for electrons and holes, respectively. The device constructed with higher quality single crystal CdZnTe demonstrated mobility-lifetime products of 1.0 x 10(-4) and 4.4 x 10(-6) cm2/V for electrons and holes. The MTF and DQE for the device were measured at several exposures and results were compared to predictions from a linear systems model of signal and noise propagation. The MTF at a spatial frequency of 10 mm(-1) exceeded 0.18 and 0.56 along the scan and slot directions, respectively. Scanning motion and CCD design limited the resolution along the scan direction. For an x-ray beam from a tungsten target tube with 40 microm molybdenum filtration operated at 26 kV, the single crystal device demonstrated a DQE(0) of 0.70 +/- 0.02 at 7.1 x 10(-6) C/kg (27 mR) exposure to the detector, despite its relatively poor charge collection efficiency.

Performance improvements for VOx microbolometer FPAs

Raytheon is producing high-quality 320 x 240 microbolometer FPAs with 25 μm pitch pixels. The 320 x 240 FPAs have a sensitivity that is comparable to microbolometer FPAs with 50 μm pixels. Typical NETD values for these FPAs are <50mK with an f/1 aperture and operating at 30 Hz frame rates. Pixel operability is greater than 99.9% on most FPAs, and uncorrected responsivity nonuniformity is less than 4% (sigma/mean). These 25 μm microbolometer detectors also have a relatively fast thermal time constant of approximately 10 msec. These arrays have produced excellent image quality, and are currently fielded in a variety of demonstration systems. The pixel size reduction facilitates a significant FPA cost reduction since the number of die printed on a wafer can be increased, and also has enabled the development of a large-format 640 x 480 FPA array. Raytheon is producing these arrays with excellent sensitivity and typical NETD values of <50mK with an f/1 aperture and operating at 30 Hz frame rates. These arrays have excellent operability and image quality. Several dual FOV prototype 640 x 480 systems have been delivered under the LCMS and UAV programs. RVS has developed a flexible uncooled front end (UFE) electronics that will serve as the basis for the camera engine systems using 320 x 240 arrays. RVS has developed a 640 x 480 Common Uncooled Engine (CUE) which is intended for small pixel, high performance applications. The CUE is the ideal cornerstone for ground and airborne systems, multi-mode sensor, weapon sight or seeker architectures, and commercial surveillance.

High sensitivity 640 × 512 (20 μm pitch) microbolometer FPAs

RVS has made a significant breakthrough in the development of a 640 × 512 array with a unit cell size of 20μm × 20 μm and performance equivalent to that of the 25μm arrays. The successful development of this array is the first step in achieving mega-pixel formats. This FPA is designed to ultimately achieve performance near the temperature fluctuation limited NETD (<20mK, f/1, 30 Hz). The SB-300 is a highly productized readout and is designed to achieve very good sensitivity (low NETD and low spatial noise) and good dynamic range. The improved performance is through bolometer structure improvements and an innovative ROIC design. It also has a simple and flexible electrical interface which allows external electronics to be small, light, low-cost, and low-power. Almost all adjustments can be made through the serial interface; hence there is no need for external adjustable (DAC) circuitry. The improved power supply rejection helps maintain highly stable detector and strip resistor bias voltages which helps reduce spatial noise and image artifacts. We will show updated performance and imagery on these arrays, which is currently being measured at <30mK, f/1,555 30 Hz. Pixel operability is greater than 99.5% on most FPAs, where the uncorrected responsivity nonuniformity is less than 4% (sigma/mean), and time constant for these arrays was measured at <8msec. We will report detailed FPA performance results including responsivity, noise, uniformity and pixel operability. We also plan to present video imagery from the most recent FPAs. The reduction in pixel size offers several potential benefits for IR systems. For a given system resolution (IFOV) requirement, the 20 μm pixel will allow an optical volume that is 50 % the size of a 25 μm based system! We will also provide an update on the enhanced performance and yield producibility of our NVESD ManTech 640 × 480 25 μm arrays, and also show data on 25 μm arrays that have been designed for faster time constants (5 ms), while maintaining high performance. We will also show the improvement in our uncooled 320 × 240 and 640 × 480 sensor electronics in terms of reduced power and size for helmet and rifle mounted sensors.

Uncooled detector development at Raytheon

At the 2010 meeting of the Defense and Security Symposia Raytheon reported on the status of their efforts to establish a high rate uncooled detector manufacturing capability. At that time we had just finished the transition of the 640 × 480, 25 μm product to our 200 mm wafer fab line at Freescale semiconductor and established an automated packaging and test capability. Over the past year we have continued to build on that foundation. In this paper we will report on this years progress in completing the transition of our 25 μm product line to Freescale semiconductor. Included will be the 320 × 240 product transition and a summary of SPC and defectivity data from one years production. Looking beyond 25 μm, we are well along in our transition of the 17 μm product line to Freescale, with test results being available for the 640 × 480. Additionally, we will report on progress / status of the Tailwind program, which is developing a 2048 × 1536, 17 μm uncooled sensor. Data to be reported includes the establishment of subfield stitching at a high rate commercial fab and the development of the detector package and electronics. With 17 μm transitioned to production, Raytheon has started work on the HD LWIR program, which is laying the foundation for the next generation of uncooled detectors by further shrinking the pixel to <17 μm. With the HD LWIR program just beginning, we will review our development strategy and program plan.

Expanded applications for high performance VOx microbolometer FPAs

RVS is producing high-quality microbolometer FPAs with 25 μm pixels. We have developed formats in both 320 x 240 and 640 x 480 array formats. These FPAs have demonstrated sensitivities that are comparable to microbolometer FPAs with 50 μm pixels with the best measured NETD value for these FPAs now <25mK with an f/1 aperture and operating at 30 Hz frame rates. Pixel operability is greater than 99.9% on most FPAs, and uncorrected responsivity nonuniformity is less than 3% (sigma/mean). These 25 μm microbolometer detectors also have a relatively fast thermal time constant of approximately 10 msec. These arrays have produced excellent image quality, and are currently fielded in a variety of systems. We will report on our latest performance data and IR captive flight test imagery. We will also show data on 25μm arrays that have been designed for faster time constants (5 msec), while maintaining high performance. RVS is also developing a 320 x 240 50μm mid-wave responding FPA. We will review the MWIR sensitivity improvements with this array and provide IR imagery. RVS is developing a 640 x 480 25μm uncooled FPA for a countermine detection application using a two-band assembly designed to be sensitive in both the Restrahlen and Thermal spectral bands. We will provide IR image data on these arrays. RVS has made a significant breakthrough in the development of a 640 x 512 array with a unit cell size of 20 μm x 20 μm, and performance approaching that of the 25μm arrays. The successful development of this array is the first step in achieving mega-pixel formats. This FPA is designed to ultimately achieve performance near the temperature fluxuation limited NETD (<20mK, f/1, 30 Hz). We will show updated performance and imagery on these arrays, which is currently being measured at <45mK, f/1, 30 Hz.

Advances in linear and area HgCdTe APD arrays for eyesafe LADAR sensors

HgCdTe APDs and APD arrays offer unique advantages for high-performance eyesafe LADAR sensors. These include: operation at room temperature, low-excess noise, high gain, high-quantum efficiency at eyesafe wavelengths, GHz bandwidth, and high-packing density. The utility of these benefits for systems are being demonstrated for both linear and area array sensors. Raytheon has fabricated 32 element linear APD arrays utilizing liquid phase epitaxy (LPE), and packaged and integrating these arrays with low-noise amplifiers. Typical better APDs configured as 50-micron square pixels and fabricated utilizing RIE, have demonstrated high fill factors, low crosstalk, excellent uniformity, low dark currents, and noise equivalent power (NEP) from 1-2 nW. Two units have been delivered to NVESD, assembled with range extraction electronics, and integrated into the CELRAP laser radar system. Tests on these sensors in July and October 2000 have demonstrated excellent functionality, detection of 1-cm wires, and range imaging. Work is presently underway under DARPAs 3-D imaging Sensor Program to extend this excellent performance to area arrays. High-density arrays have been fabricated using LPE and molecular beam epitaxy (MBE). HgCdTe APD arrays have been made in 5 X 5, 10 X 10 and larger formats. Initial data shows excellent typical better APD performance with unmultiplied dark current < 10 nA; and NEP < 2.0 nW at a gain of 10.

Epitaxial growth of HgCdTe 1.55-μm avalanche photodiodes by molecular beam epitaxy

Separate absorption and multiplication avalanche photodiode (SAM-APD) device structures, operating in the 1.1 - 1.6 micrometer spectral range, have been fabricated in the HgCdTe material system by molecular-beam epitaxy. These HgCdTe device structures, which offer an alternative technology to existing III-V APD detectors, were grown on CdZnTe(211)B substrates using CdTe, Te, and Hg sources with in situ In and As doping. The alloy composition of the HgCdTe layers was adjusted to achieve both efficient absorption of IR radiation in the 1.1 - 1.6 micrometer spectral range and low excess-noise avalanche multiplication. To achieve resonant enhancement of hole impact ionization from the split-off valence band, the Hg1-xCdxTe alloy composition in the gain region of the device, x equals 0.73, was chosen to achieve equality between the bandgap energy and spin-orbit splitting. The appropriate value of this alloy composition was determined from analysis of the 300 K bandgap and spin-orbit splitting energies of a set of calibration layers, using a combination of IR transmission and spectroscopic ellipsometry measurements. MBE-grown APD epitaxial wafers were processed into passivated mesa-type discrete device structures and diode mini-arrays using conventional HgCdTe process technology. Device spectral response, dark current density, and avalanche gain measurements were performed on discrete diodes and diode mini- arrays on the processed wafers. Avalanche gains in the range of 30 - 40 at reverse bias of 85 - 90 V and array-median dark current density below 2 X 10-4 A/cm2 at 40 V reverse bias have been demonstrated.