2
Fundamentals of Ultraviolet, Visible, and Infrared Detectors

INTRODUCTION

Electro-optical detectors are used to measure or sense the radiation emitted or reflected by objects within the detector’s optical field of view (FOV). Passive systems operate without any illumination of the object by the observer, relying on either self-luminosity (for example, a hot rocket exhaust) or reflection-transmission of ambient light. In active systems, the observation is associated with irradiation of the scene (as in a camera flash) in the spectral region of interest. A detector converts incident radiation to an electrical signal that is often proportional to the incoming intensity. This electrical signal is processed, usually digitally, transmitted, and/or stored. A two-dimensional array of detectors, called a focal plane array (FPA), is often placed at the focal plane of an optical system so that the spatial variation of the incident intensity is recorded as an image. There are many excellent texts at both introductory and advanced levels that deal with the fundamentals and applications of ultraviolet (UV), visible, and infrared detectors.1,2 The committee’s intent is to provide a brief introduction to facilitate reading the material that follows.

1

E.L. Dereniak and G.D. Boreman. 1996. Infrared Detectors and Systems. Hoboken, N.J.: John Wiley and Sons.

2

S. Donati. 2000. Photodetectors: Devices, Circuits and Applications. Saddle River, N.J.: Prentice-Hall.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 23
2 Fundamentals of Ultraviolet, Visible, and Infrared Detectors INTRODUCTION Electro-optical detectors are used to measure or sense the radiation emitted or reflected by objects within the detector’s optical field of view (FOV). Passive systems operate without any illumination of the object by the observer, relying on either self-luminosity (for example, a hot rocket exhaust) or reflection-transmission of ambient light. In active systems, the observation is associated with irradiation of the scene (as in a camera flash) in the spectral region of interest. A detector converts incident radiation to an electrical signal that is often proportional to the incoming intensity. This electrical signal is processed, usually digitally, transmitted, and/or stored. A two-dimensional array of detectors, called a focal plane array (FPA), is often placed at the focal plane of an optical system so that the spatial variation of the incident intensity is recorded as an image. There are many excellent texts at both introductory and advanced levels that deal with the fundamentals and applications of ultraviolet (UV), visible, and infrared detectors.1,2 The committee’s intent is to provide a brief introduction to facilitate reading the material that follows. 1 E.L. Dereniak and G.D. Boreman. 1996. Infrared Detectors and Systems. Hoboken, N.J.: John Wiley and Sons. 2 S . Donati. 2000. P hotodetectors: Devices, Circuits and Applications. S addle River, N.J.: Prentice-Hall. 

OCR for page 23
seeing Photons  SOURCES Sources include self-emission from hot objects that generally follow a black- body radiation curve depending on the temperature of the source, modified by the spectral emissivity of the object. Alternatively for passive sensors, the reflection or transmission modification of ambient sources can be detected. During daytime, the dominant source in the visible is the sun. There is a significant night glow in the spectral region around 1.5 µm that makes short-wavelength infrared (SWIR) imaging an alternative to visible image intensifier night vision goggles for some night vision applications.3,4 The semiconductor absorbance ranges that enable passive night vision and the nightglow irradiance spectrum are illustrated in Fig- ure 2-1. The peak of the room temperature blackbody curve is at about 10 µm in the infrared. TRANSMISSION Spectral Regions Over the years a number of designations for spectral regions have become somewhat standard, but there is significant overlap and it is useful to define the regions used in this report to assist the reader (see Table 2-1). The transitions between these regions are not sharply defined and the designations are to be in- terpreted loosely; the detection mechanisms, the transmission, and the dominant noise sources all vary across these bands. These definitions help in discussing those variations cohesively. Electromagnetic sensors cover the entire range from 200 nm to 20 µm and beyond; this taxonomy is intended merely to provide a nomenclature for the most frequently used bands for long-range imaging. Atmospheric Transmission Atmospheric transmission is an important aspect of any terrestrial remote sensing application. Figure 2-2 shows the atmospheric transmission across the 0.2- 20 µm region (~1 km horizontal path length at sea level, temperature = 15°C, with 46 percent relative humidity) along with the wavelength bands defined above. While Figure 2-2 is representative, the transmission curve will vary with atmo- spheric conditions, as well as the path taken through the atmosphere; for example, 3 T.R. Hoelter and B.B. Barton. 2003. Extended short wavelength spectral response from InGaAs focal plane arrays. Proceedings of SPIE 5074:481-490. 4 Available at http://www.sensorsinc.com/downloads/paper_HighSpeedSWIRImagingAndRange Gating.pdf. Last accessed March 25, 2010.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and FIGURE 2-1 Nightglow irradiance spectrum under different moonlight conditions. SOURCE: Vatsia, L.M. 1972. Atmospheric optical environment. Research and Development Technical Report ECOM-7023. Prepared for the Army Night Vision Lab, Fort Belvoir, Va. TABLE 2-1 Definition of Spectral Regions with Long-range Atmospheric Transmission Designation Wavelength Band (µm) Physical Significance and Comments Solar blind 0.2-0.28 Solar radiation in this band is blocked by the Earth’s ozone layer, so any UV radiation in this region is likely man-made. Once under the ozone layer, the atmosphere is transparent to wavelengths as short as ~200 nm where oxygen absorption limits the transmission UV 0.28-0.4 Atmosphere is transparent Visible 0.4-0.7 Peak of solar spectrum Near infrared 0.7-1.0 Long-wavelength cutoff defined by silicon detector response SWIR 1.1-2.7 Overlaps with telecommunications wavelengths; large commercial infrastructure available at 1.3 and 1.55 µm MWIR 2.7-6.2 Atmospheric transmission window, molecular vibrational absorptions LWIR 6.2-15.0 Atmospheric transmission window, molecular vibrational absorptions VLWIR 15.0-20.0 Molecular vibrational absorptions NOTE: LWIR = long-wavelength infrared; MWIR = mid-wavelength infrared; VLWIR = very long wavelength infrared.

OCR for page 23
seeing Photons  The Electromagnetic Spectrum Very Long Wave- Transmittance (%) Long Wavelength Infrared (LWIR) Ultraviolet (UV) Visible Near Infrared Short Wavelength Infrared (LWIR) Mid Wavelength Infrared (MWIR) length Infrared (VLWIR) H2O H2O H2O CO2 H2O O3 O3 Q Band CO2 CO2 CO2 Ozone CO2 Y Band J Band H Band K Band L Band N Band I Band H2O G Band R Band M Band UVC UVB UVA Band B Band V Band CO2 0.7 0.9 Wavelength (µm) Vertical Atmospheric Transmittance Horizontal Transmittance of 1 km Air Path at Sea Level Horizontal Vertical Transmission from sea Level to 100 km Conditions from U.S Standard Atmosphere FIGURE 2-2 Display of the atmospheric transmittance levels. SOURCE: Data from the Santa Barbara Research Institute, a subsidiary of Hughes, and OMEGA Engineering, Inc. Available at http://www.coseti.org/atmosphe.htm. Accessed March 29, 2010. looking through the atmosphere from a space-based platform will differ in the details. FINDING 2-1 For any sensor application, the relevant spectral range is set by the overlaps of the spectral signature of the target and the pass bands of the transmission medium between the target and the detector. DETECTION In general, detectors are divided into two classes: thermal and photon (or quan- tum).5 Thermal detectors operate by the absorption of incoming radiation causing a change in temperature of the detector and by the sensitivity of some measurable parameter—for example, resistance—to that temperature. Thermal detectors are typically sensitive across a wide range of incident wavelengths. Quantum detectors depend on the direct interaction of the incoming light with the detector materials, resulting, for example, in electron-hole pair creation in a semiconductor. Photo- generated carriers can be measured by directly measuring charge collected during an integration period, by measuring photocurrent, by a change in resistance (pho- toconductive), or by voltage generation across a junction (photovoltaic). 5 R.L. Petritz. 1959. Fundamentals of infrared detectors. Proceedings of IRE 47(9):1458-1467.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and Thermal Detection In thermal detectors, photon absorption leads to a small temperature rise of the detector, which is sensed by a temperature-dependent property of the material such as a pyroelectric effect or a temperature-dependent resistance. An advantage of using thermal detectors is that they typically are very broadband; a disadvantage is that it is a challenge to make a structure that has measurable temperature rise for low power signals.6 It is also possible to track thermoelectric effects using ther- mocouples and thermopiles or with the aid of Golay cells that can track thermal expansion in a system. In general, there is a trade-off between the response speed of a thermal detector and its sensitivity. Thermal isolation allows longer integra- tion times to detect weaker signals, but this means that the detector response time is necessarily increased. Quantum Detection Quantum or photon detectors, typically semiconductors with bandgaps matched to the photon energy, operate by the generation of electron-hole pairs by the absorption of a photon. There are two major classes of photon detectors: photoconductive and photovoltaic. Photoconductors In a photoconductor, the excited carriers are detected through the change in resistance induced by the photoexcited carriers. Often the mobilities of electrons and holes are quite different in the semiconductor, with the consequence that the faster carrier can transit the detector several times before the carriers recombine. This provides a gain mechanism. Photovoltaic Detectors In a photovoltaic device, the photoexcited electron and hole are separated by the built-in field associated with a p-n junction and collected. Particularly for indirect bandgap semiconductors, such as silicon, the absorption region has to be extended to ensure a good quantum efficiency leading to p-i-n designs. There is usually a trade-off between extending the absorption region for high probability 6 B. Cole, R. Homing, B. Johnson, K. Nguyen, P.W. Kruse, and M. C. Foote. 1994. High performance infrared detector arrays using thin film microstructures. Proceedings of the Ninth IEEE International Symposium on Applications of Ferroelectrics 653-656.

OCR for page 23
seeing Photons  of absorbing a photon and shortening it to ensure that recombination mechanisms do not impact the collection efficiency. Avalanche Photodiodes Avalanche photodiodes incorporate high-field regions that lead to carrier multiplication to increase signal levels above the characteristic noise sources down- stream in the electronics. The carrier multiplication is accomplished by imparting sufficient kinetic energy to a carrier for it to create an additional electron-hole pair by impact ionization. There is always some excess noise associated with the multiplication, but this can be minimized by designs that allow primarily one carrier to be multiplied while suppressing the multiplication of the oppositely charged carrier. INFORMATION ENCODED BY PHOTONS A photon is the quantum mechanical element of all electromagnetic radiation. Photon energy is given by hc 1.986 × 10 −19 E ph = hv = = J, λ λ where h is Planck’s constant, c is the speed of light, and λ is the wavelength of the infrared photon in micrometers. By collecting photons, measurements can be made of light's intensity, temporal variations in intensity, spectrum, polarization, electric field phase, incident angle, and photon time of arrival. These types of measure- ments will now be defined in greater detail. Intensity Intensity, the incident power per unit area, is the most commonly used optical imaging signal. The variation of signal intensity across the focal plane is recorded as a gray-scale image. Spectrum Images can be panchromatic, monochromatic, multispectral (including three- color traditional RGB [red, green, blue]), or hyperspectral (multiple spectral bands across the wavelength range of interest). Spectral information can be obtained in several ways, including dispersion into different pixels (using diffraction or refrac-

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and tion), temporal modulation of spectral filters, use of on-chip absorptive filters, measuring the size of a charge packet (for example, for X-ray energy spectroscopy), or varying the bandgaps of multiple photon absorbing regions. Polarization Imagers have been developed to measure the complete polarization states of the electromagnetic field described by the Stokes parameters or, more commonly, the linear polarization components.7 Dynamics Time scales can range from still imaging, to video rates, to fast (e.g., kilohertz) amplitude fluctuations due to target phenomenology, to high-speed imaging (e.g., megahertz), to acquiring sub-ns (nanosecond) range information from single photons for active LADAR (laser detection and ranging) imaging. Time Delay The time delay from an excitation to the reception of a photon provides a mea- sure of distance to the object in the same way as in a radar receiver. This is an active sensor application and is beyond the scope of this study. However, it is worthwhile to note that advances in both ultrafast sources and high-speed photon counting detectors will make available in the visible and near-infrared (NIR) spectral regions many of the advanced radar concepts, such as chirped pulses and synthetic imaging concepts, that have been so successful in longer-wavelength spectral regions. Imagers are available with many different designs and architectures to exploit these different characteristics of optical signals, but it is difficult to design a single imager that is optimized for simultaneously measuring all of these attributes. Phase and Incidence Angle Imagers can perform heterodyne or other types of carrier-phase detection. High-speed detectors can allow detection of temporal-phase variation by measur- ing the beat frequency between a local oscillator and a return signal. Alternatively, wavefront sensors are used to measure spatial-phase variation, allowing analysis of atmospheric wavefront distortions for adaptive optical correction. These sensors enable measurement of a small phase distortion in optical waves under significant 7 For additional detailed information on Stokes polarization parameters, please see http://spie. org/x32376.xml. Last accessed on May 6, 2010.

OCR for page 23
seeing Photons 0 background wavefront aberrations. A Shack-Hartmann sensor is a frequently used wavefront sensor that consists of a microlens array in front of a multiple element detector or focal plane array to register the local wavefront tilt in the position of the imaged spots from each microlens on the sensor array.8 FINDING 2-2 While the spatial variation of signal intensity is most often the quantity evalu- ated to produce an image, spectral distributions, polarization, phase, and tem- poral characteristics are additional information channels that can be exploited in some applications. THE LIMITS IMPOSED BY DIFFRACTION Spatial Resolution Optical imaging systems can have limitations in resolution caused by imper- fections in the lenses or by their misalignment, which results in defects of the im- age and is often referred to as an optical aberration. In addition, for transmission through the atmosphere, variations of the index of refraction due to air currents and temperature variations also cause changes in the image. Aberrations describe the amount by which a geometrically traced ray misses a specific location in the im- age.9 If all of these aberrations are dealt with, diffraction is the ultimate limit of op- tical focusing. For an aberration-free optical system with uniform illumination of a circular input aperture, the result at the focus is a bright central disk surrounded by a series of concentric rings of rapidly diminishing amplitude. This is known as an Airy pattern (shown in Figure 2-3), and the diameter of the central disk is given by ~1.22λ/NA, where λ is the wavelength and NA the numerical aperture of the optical system (the half-angle of the light acceptance cone).10 Mathematically, the intensity versus position in the Airy pattern is given by 2 2 2π NAr  NA   2 J (m)  I (r ) =    m  with m = λ , λ  8 J. Schwiegerling and D.R. Neal. 1994. Historical development of the Shack-Hartmann wavefront sensor. J Opt Soc Am A 11:1949-1957. 9 Harold Rothbart and Thomas H. Brown. 2006. Mechanical Design Handbook: Measurement, Analy- sis, and Control of Dynamic Systems, Second Edition. New York: McGraw-Hill Companies, Inc. 10 NA is related to the F/# notation commonly used to describe the light acceptance cone in pho - tography. NA = sin θ, where light incident on the lens at angles up to θ is imaged onto the focal spot. In terms of the diameter of the lens, D, and the focal length, f, tan θ = D/2f. F/# = f/D, so for small angles where sin θ ≈ tan θ ≈ θ, NA ≈ 1/[2F/#].

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and FIGURE 2-3 The Airy disk. SOURCE: Figure courtesy of Cambridge in Colour. Available at http://www.cambridgein colour.com/tutorials/diffraction-photography.htm. Accessed on March 29, 2010. where r is the radial coordinate and J1 is the first-order Bessel function, with a first zero at m = 0.61. As is very well known, the minimum focal spot diameter also sets the separa- tion distance at which two point objects can be resolved as distinct. The Rayleigh resolution criterion is obtained by setting the minimum separation of two objects equal to the radius of the Airy disk,11 Rmin = 0.6λ/NA. Detector optical systems capable of producing images with angular resolu- tions that are as good as the instrument’s theoretical limit are said to be diffraction limited. For an ideal circular aperture, the two-dimensional diffraction pattern, the Airy disk, is used to define the theoretical maximum resolution for the optical system. When the diameter of the disk’s central peak becomes large with respect to the size of the pixel in the FPA, it begins to have a visual impact on the image. OPTICAL SYSTEMS Numerical Aperture and Field of View The numerical aperture, NA = sinθ ≤ 1, describes the light collection power of the optical system. A larger NA results in higher resolution (see equation above) and, therefore, requires more pixels in the focal plane array if the same area is to 11 J.W. Strutt (III Lord Rayleigh). 1879. Investigations in optics, with special reference to the spec - troscope. Monthly Notices of the Royal Astronomical Society 40:254.

OCR for page 23
seeing Photons  be imaged at this higher resolution. The field of view is the extent of the imaged region on the focal plane array referred back to the objects being imaged. Curved Focal Planes Everyone is familiar with one optical system that uses a curved focal plane array, namely the human eye. Nature chooses this curvature because it makes the optics much simpler. In contrast, the many optical elements in, for example, a standard camera lens are required to faithfully reproduce the image on the flat focal plane of the camera. Our materials technology, which relies on epitaxial crystal growth and the accompanying device fabrication technologies and has largely derived from planar silicon integrated circuit technology, make curved focal plane arrays a difficult option. Recently there has been significant work, particularly in visible systems based on silicon materials to adapt to curved focal surfaces.12 A flat surface can conformally map to a cylinder, but it cannot map to a sphere without deform- ing. Practical curved focal plane technologies are making a significant difference in image capture and in the size and weight of optical systems. DETECTIVITY The noise-equivalent power (NEP) is the input power at which a detector ex- hibits a signal-to-noise ratio of unity. The detectivity, D, is the inverse of NEP; this of course depends on the detector area (A) and the detection bandwidth (BW). For observation of an extended object, the signal scales as the area, while the noise as- sociated with the dark current scales as A ; the noise also scales as BW . These simple extrinsic parameters can be eliminated with a simple normalization; the resulting parameter is D * = A × BW / NEP which is more characteristic of detec- tor material performance. Quantum Efficiency The signal level at the detector is directly proportional to the probability that an incident photon results in an electrical signal; this is known as the quantum efficiency (QE). The external quantum efficiency includes effects such as reflec- tion from optical surfaces that can be addressed with additional engineering (for example, antireflection coatings), while the internal quantum efficiency is more characteristic of the detector material and device geometry. The prerequisite for quantum efficiency is absorption of a photon leading 12 R. Dinyari, S-B. Rim, K. Huang, P.B. Catrysse, and P. Peumans. 2008. Curving monolithic silicon for non-planar focal plane array applications. Applied Physics Letters 92:091114.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and to some, typically electronic, change in the material such as the creation of an electron-hole pair in a semiconductor. A high absorption coefficient allows thinner material, which facilitates the second component of the quantum efficiency—sens- ing the electron-hole pair. In a photovoltaic detector this is accomplished by separating the carriers across a p-n junction resulting in a voltage proportional to the number of carriers. This process can be disrupted by recombination, either radiative or nonradiative, before the carriers diffuse into the junction region. In a photoconductive detector, the carriers are sensed as change in the conductance, which is measured as the current for a fixed voltage applied across the device. If the carriers cycle more than once through the detector before recombination, there is a gain associated with the detection that can make it easier to overwhelm noise further downstream in the electronics. Noise There are many noise sources whose relative importance varies with the de- tector material properties, the ambient temperature, the detector operating tem- perature, the device design, the readout electronics, and other variables. Some of the most important sources are catalogued here. Since these noise sources are in general uncorrelated, the total noise is proportional to the square root of the sum of the squares of the individual noise sources. Photon Statistics and Background-limited Infrared Detection There is noise associated with the signal itself. Since photodetection is a discrete process, and most natural sources exhibit Poisson statistics in the fluctuations of the signal level, this noise scales as the square root of the signal level. Photon noise is unavoidable for natural signals and sets a fundamental noise floor. For an extended source (image structure large compared to an individual pixel) the current scales as the pixel area, so the noise is n photon = i photon (BW × A) . For engineered sources, it is possible to reduce the shot noise at the expense of increased phase fluctuations, and vice versa. Collectively these are known as squeezed states and have been investigated for communications applications. Any background photons impinging on the detector also contribute to the noise. While the background is usually not an issue in the UV and visible, in the infrared there is substantial background flux associated with blackbody emission from a room-temperature scene. The peak of the 300 K blackbody emission is in the middle of the LWIR at 10 µm. For cooled infrared detectors (discussed below) this dark current associated with the background radiation and the accompany- ing noise levels often set the detection limit. This is known as background-limited infrared photodetection (BLIP). Many current infrared systems are close to BLIP;

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and achieve optimum performance for specific applications. For example, it is desir- able to have a low dark count rate because this will maximize signal-to-noise ratio and minimize statistical uncertainty. A decrease of the dark count rate occurs exponentially with decreased temperatures; however, operating at low T increases the after-pulsing probability. Thus, changing one parameter, while it will improve performance in a specific area, will affect other parameters as well.65 Near Infrared Silicon Silicon sensors are sensitive throughout the visible to wavelengths as long as the silicon bandgap of ~1.1 µm. Just the opposite of the UV situation discussed above, the very long absorption length associated with the indirect bandgap of silicon requires very different optimization of the device structure for quantum efficiency and carrier collection. Intensifiers In many fields, it is common to use image intensifiers in front of a camera tube because they permit cameras to work at the lowest light levels possible. These are electron optic systems that are made up of an input phosphor-photocathode screen that converts incoming radiation into a beam of electrons, electrodes to control the movement of electrons, and an output phosphor screen that produces the output image.66 They convert spectral radiation to a visible light image, which after additional processing can be displayed on a monitor. Most commercially available image intensifiers have axial symmetry; however, some nonaxisymmetrial intensifiers have recently been designed.67 Intensifiers work utilizing an avalanche or Geiger mode gain in back of a photocathode. Thus, extremely small photon fluxes are multiplied several thousandfold, allowing viewing under extremely low light conditions. 65 B.S. Robinson, D.O. Caplan, M.L. Stevens, R.J. Barron, E.A. Dauler, S.A.Hamilton, K.A. McIntosh, J.P. Donnelly, E.K. Duerr, and S. Verghese 2005. High-sensitivity photon-counting communications using Geiger-mode avalanche photodiodes. Proceedings of the IEEE Lasers and Electro-Optics Society 559-560. 66 K.G. Vosburgh, R.K. Swank, and J.M.J. Houston. 1997. X-ray image intensifiers. Advances in Electronics and Electron Physics 43:205-244. 67 N.W. Adamiak, J. Dabrowski, and A. Fenster. 1996. Design of nonaxisymmetrical image intensi - fiers. IEEE Transactions on Industry Applications 32(1):93-99.

OCR for page 23
seeing Photons 0 Short-wavelength Infrared InGaAs detector technology is quite well developed as a result of its dominant use in fiber-optic telecommunications at ~1.3 to 1.7 µm. By varying the composi- tion, the bandgap can be shifted to as long as 2.6 µm. Mid-, Long-, and Very Long Wavelength Infrared Brief History of Infrared Detection Thallium sulfide and lead sulfide (or galena) were among the first infrared detector materials, developed during the 1930s. Many other materials have been investigated for application to infrared detection. Lead-salt detectors are polycrys- talline and are produced using vacuum evaporation and a chemical deposition process from solution followed by post-growth sensitization.68 Reproducibility has historically been poor, but well defined, although somewhat empirical recipes were eventually found. Significant improvement in detector manufacture occurred with the discovery of the transistor, which stimulated growth and material purification techniques. This resulted in novel techniques for detector production from single crystals. High-performance detectors were initially based on the use of germanium with the introduction of controlled impurities. Development of high-performance visible and NIR detectors based on silicon began to occur in the 1970s after the invention of the CCD. This resulted in the development of sophisticated readout schemes that allowed both detection and readout to occur on one common silicon chip. In the 1950s, there was extensive investigation of III-V semiconductors. As a result of its small bandgap (5 µm at 77 K), InSb showed promise as a material for MWIR detection, and indeed vastly improved InSb FPA arrays remain a mainstay of cooled MWIR imaging. Shortly thereafter, in 1959, HgCdTe (mercury cadmium telluride, or MCT) was found to exhibit semiconducting properties over much of its composition range. The alloy’s bandgap was variable from 0.0 to 1.605 eV. Later, long-wavelength photoconductivity was demonstrated in HgCdTe, leading the way to development of infrared detectors. A shift occurred in the mid-1960s toward using the PbSnTe alloy because of production and storage problems associated with HgCdTe. However, limitations in the speed of response for PbSnTe detectors and the better suitability of HgCdTe for 68A. Rogalski, and J. Pitrowski. 1988. Intrinsic infrared detectors. Progress in Quantum Electronics 12:287-289.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and infrared imaging device production, as well as improvements in the technology of the material, once again shifted the focus back to HgCdTe at the beginning of the 1970s. In the mid- to late 1980s, HgCdTe remained the most promising narrow-gap semiconductor for infrared detector arrays. Today, after many years of intensive development, photovoltaic HgCdTe is widely used across all infrared bands for high-performance IR FPAs. Indium Antimonide InSb MWIR detectors have been developed continuously since the 1950s and are a quite mature technology. InSb has a bandgap of about 5.4 µm at 77 K, making this material a good choice for MWIR detection. InSb detectors are based on bulk materials rather than epitaxy, and processing involves impurity diffusion or ion implantation. Relatively large wafers, ~3 to 4 inches (100 mm), are available. Mercury Cadmium Telluride HgCdTe is a ternary compound whose bandgap can be adjusted by varying the relative proportions of mercury and cadmium. HgCdTe is a pseudobinary al- loy between HgTe and CdTe, written as Hg1–xCdxTe. The composition range 0.21 < x < 0.26 covers the LWIR regime. Nearly all of today’s LWIR HgCdTe is grown epitaxially in thin layers by molecular beam epitaxy (MBE) or liquid-phase epitaxy (LPE). Both p-type and n-type doping can be reproducibly accomplished. The composition can be varied during growth, allowing the formation of heterojunc- tions, barriers, and multiband devices. The most commonly used substrate is CdZnTe, but there is extensive work on using both silicon and GaAs substrates, although the large lattice mismatch has resulted in slow progress. Development of photovoltaic arrays began more than 30 years ago, and a high level of technology readiness has been achieved. Recently, Tennant presented an empirical result, known as Rule 07 (the 07 refers to the year this result was obtained and was used to stress its transient status),69 that provides a characterization of dark current as a function of bandgap and tempera- ture across a wide range of HgCdTe materials. Figure 2-6 presents the measured and semiempirical model. Tennant revisited this characterization and found that it remains a reliable guide.70 This rule is relevant to high-quantum-efficiency devices, 69W.E. Tennant, D. Lee, M. Zandian, E. Piquette, and M. Carmody. 2008. MBE HgCdTe technology: a very general solution to IR detection, described by ‘‘Rule 07,’’ a very convenient heuristic. Journal of Electronic Materials 37(9):1406-1410. 70W.E. Tennant. 2010. “Rule 07” revisited: still a good heuristic predictor of p/n HgCdTe photodiode performance? Journal of Electronic Materials DOI:10.1007/s11664-010-1084-9.

OCR for page 23
seeing Photons  FIGURE 2-6 Dark current density for HgCdTe as a function of the cutoff wavelength × temperature product. NOTE: TIS = Teledyne Imaging Sensors. SOURCE: W.E. Tennant, 2010. “Rule 07” revisited: still a good per- formance heuristic predictor of p/n HgCdTe photodiode performance. Journal of Electronic Materials. DOI: 10.1007/s11664-010-1084-9. and the analysis suggests that Auger recombination (Auger 1) in the n-type HgCdTe is the dominant limiting mechanism.71 This latest paper also compared both the theoretical and the experimental results for various strain-layer superlattice (SLS) structures against the HgCdTe results. The best SLS results are approaching the Rule 07 limits, while the theoretical work shows that significant improvement remains possible pending improvements in materials quality and processing. Tennant then went on to compare this current density with the background dark current for a 4π steradian background at the operating temperature of the device (e.g., surrounded by a cold shield). This result is shown in Figure 2-7. For 77 K operation and a MWIR cutoff, the device is close (within a factor of ~5) to this background limit, while for both SWIR and LWIR operation, the dark cur- rents are substantially above the BLIP limit for this very low background situation. 71Auger processes in semiconductors are three-body interactions in which an electron-hole pair recombines without emission of a photon but rather with excitation of a second carrier to a higher- energy state. Auger processes are essentially the inverse of impact ionization in which an energetic carrier relaxes by creating an electron-hole pair.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and Tennant ascribes the LWIR result to Auger recombination processes, while at short wavelengths the increased noise results from a deviation between the optical and electrical bandgaps. Substantial efforts have already been made to reduce the Auger recombination, and it does not appear likely that much further improvement is available in present material and device configurations. Thus, the low-temperature “external radiative limit” remains an elusive goal that does not appear approachable within the constraints of current technology. It is important to recognize that the BLIP current for LWIR tactical applica- tions, looking at a 300 K background with F/1 optics, is much higher than this “external radiative” limit. At 77 K operation temperature and with a 12 µm LWIR cutoff, the dark current is ~ 10–4 A/cm2 (see Figure 2-7). From the blackbody ir- radiance (300 K, F/1 optics), the dark current is ~ 0.18 A/cm2, orders of magnitude greater than the intrinsic device dark current. FINDING 2-3 MWIR and LWIR detectors are already close to fundamental BLIP for terres- trial operations that look at a 300 K background. Future innovations will focus on device and system optimization for specific applications. Id ratio to external radiative limit Wavelength (µm) FIGURE 2-7 HgCdTe dark current from Rule 07 relative to the external radiative limit, corresponding to a cold shield at the device temperature. At MWIR there is relatively little room for improvement at the lowest temperatures; however at other wavelengths there is substantial excess dark current, which limits the 2-7 w-replacement type.eps detector performance for low-background situations. SOURCE: W.E. Tennant, D. Lee, M. Zandian, E. Piquette, and M. Carmody. 2008. MBE HgCdTe technology: a very general solution to IR detection, de - scribed by ‘‘Rule 07,’’ a very convenient heuristic. Journal of Electronic Materials 37(9):1406-1410.

OCR for page 23
seeing Photons  Strained-layer Superlattice Strained-layer superlattice material consists of alternating thin layers of InAs and GaSb. A typical LWIR example of the superlattice structure has a period consisting of 4.4 nm of InAs and 2.1 nm of GaSb. This pair is repeated 300 times or more to form the IR-absorbing region. Because of the type II band offset be- tween the two constituent materials, in which the conduction band of InAs is below the valence band of GaSb, the structure exhibits new bands for holes and electrons, which are separated by an energy difference that is smaller than either of the bandgaps of the InAs or GaSb themselves and is adjustable by varying the thicknesses of the layers. This small effective bandgap is suitable for absorbing IR photons. The structure is grown by MBE and the commonly used substrate is GaSb. Both n- and p-type doping have been demonstrated. Heterojunctions are typically formed by growing contacting regions adjacent to the absorbing region, having a shorter superlattice period and a wider effective bandgap than the absorber. SLS materials are based on very well developed III-V materials, and the vast experience in bandgap engineering in these and related systems holds promise for continuing developments. Several variants have been and continue to be introduced including “W” and “M” structures.72,73 This remains an active research area with significant potential for dramatic advances. Additionally, the basic SLS structure can be modified by inserting a very thin (a few angstroms) layer of AlSb as a barrier for the majority carrier electrons. This opens up a wide parameter space for bandgap engineering, to enable specialized barriers as well as various heterojunction designs. Devices incorporating these structures have been grown and tested recently with the aim of reducing the dark current.74 The SLS LWIR technology development effort has received substantial funding recently because of its potential as a future, low-cost, III-V compatible replacement for HgCdTe in some applications. Although progress has been made the device performance, the dark currents remain significantly greater than those of HgCdTe as shown in Figure 2-8. These are laboratory studies, the technology readiness level of SLS material is well behind that of HgCdTe. 72 B.M. Nguyen, D. Hoffman, P.Y. Delaunay, and M. Razeghi. 2007. Dark current suppression in type II InAs/GaSb superlattice long wavelength infrared photodiodes with M-structure barrier. Applied Physics Letters 91:63511-1. 73 E.H. Aifer, J.G. Tischler, J.H. Warner, I. Vurgaftman, W.W. Bewley, J.R. Meyer, J.C. Kim, L.J. Whitman, C.L. Canedy, and E.M. Jackson. 2006. W-structured type-II superlattice long-wave infrared photodiodes with high quantum efficiency. Applied Physics Letters 89:053519-1. 74 D.Z. Ting, C.J. Hill, A. Soibel, S.A. Keo, J.M. Mumolo, J. Nguyen, and S.D. Gunapala. 2009. A high-performance long wavelength superlattice complementary barrier infrared detector. Applied Physics Letters 95:023508.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and FIGURE 2-8 Comparison of theoretical (inside circled region) dark currents of SLS devices with the HgCdTe Rule 07 metric (solid line). The experimental dark currents are above those achieved in HgCdTe, while the theory shows a potential advantage of SLS pending better materials and device processes. SOURCE: Tennant, W.E., D. Lee, M. Zandian, E. Piquette, and M. Carmody. 2008. MBE HgCdTe technology: a very general solution to IR detec - tion, described by ‘‘Rule 07,’’ a very convenient heuristic. “Rule 07” revisited: still a good performance heu - ristic predictor of p/n HgCdTe photodiode performance. Journal of Electronic Materials 37(9):1406-1410 DOI: 10.1007/s11664-010-1084-9. Quantum-well Infrared Photodetectors and Quantum-dot Infrared Photodetectors Quantum-well infrared photodetectors (QWIPs) and quantum-dot infrared photodetectors (QDIPs) are unipolar photoconductive devices based on intra- band absorption between electronic levels defined by quantum confinement in traditional III-V semiconductors, principally GaAs and InP. The promise is that the III-V growth and processing technology is quite mature, substrates are readily available, and scaling to large arrays should be simpler (and have higher yield) than for HgCdTe-based devices. The issues are related to the relatively weak absorption associated with the quantum confined structures. For the case of intrasubband transitions in III-V QWs, selection rules forbid the absorption of normally incident light requiring a grating or other optical ele- ment to scatter the incident light into the QW plane. Due to the weak absorption,

OCR for page 23
seeing Photons  associated with the small fill factor of the QWs relative to the wavelength, this can lead to cross-talk issues and constraints on decreasing the pixel size. QDIPs, as a result of the three-dimensional confinement, eliminate this selection rule allowing normal incidence detection, but the absorption is still quite weak, about 2 to 3 per- cent for a single pass through a typical active layer, which results in poor quantum efficiency. This is somewhat alleviated in the detectivity by the low dark currents, which also depend on the total volume of quantum dots. Recently, there has been quite a bit of activity in adding nanostructures such as plasmonics to QDIPs; this is discussed more fully in Chapter 4. Rogolski has recently reviewed progress in both HgCdTe and QWIPs-QDIPs for FPAs.75,76,77 Very Long Wavelength Infrared Many of the same detector technologies being developed for the LWIR also can be optimized for VLWIR operation beyond 12 µm. Historically, bulk doped semiconductors have dominated in this spectral region, which is potentially im- portant for missile detection against a cold (space) background. Mercury cadmium telluride detectors suffer from increasing noise due to thermally generated carri- ers and Auger processes at these land wavelengths. Both type II superlattice and QDIP-QWIP detectors have shown promise for this spectral region. This remains an active area of investigation. FINDING 2-4 Continued detector advancement requires improved growth and processing of low defect density compound semiconductor materials. The 30-year trend has been improvements in existing materials along with the incorporation of nanoscale structures in one, two, and three dimensions. FABRICATION OF DETECTORS AND FOCAL PLANE ARRAYS Detectors Each material system brings its own unique set of fabrication issues to main- tain high performance. Overall the dimensional scale of even visible pixels is large 75A. Rogalski. 2006. Competitive technologies of third generation infrared photon detectors. Opto- Electronics Review 14(1):87-101. 76 P. Martyniuk and A. Rogalski. 2008. Quantum-dot infrared photodetectors: status andoutlook. Progress in Quantum Electronics (3-4):89. 77 P. Martyniuk, S. Krishna, and A. Rogalski. 2008. Assessment of quantum dot infrared photodetec - tors for high temperature operation. Journal of Applied Physics 0(3):034314-1.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and compared to the minimum feature size of current lithographic tools (which are following a Moore’s law curve with current manufacturing at the 45 nm node). Focal Plane Arrays A focal plane array is created by arranging individual detector elements in a lattice-like array. Individual detectors in an array are often referred to as pixels, short for picture elements. However, the process of developing an integrated ar- ray of detectors is significantly more challenging than fabricating an individual detector element. The overall scheme of silicon-based visible detectors is discussed in the sections on CMOS and CCDs. As a result of the advanced state of silicon technology, these imaging chips integrate to some extent both the detection and the electronics. For infrared detectors, in contrast, the signals have to be moved from the detector material to silicon circuitry, called the readout integrated circuit (ROIC); this is usually accomplished by bonding each pixel to a silicon readout circuit using a myriad of indium bump bonds. The number of bonds scales as the number of pixels, and for very large arrays this is a difficult manufacturing step. Typically each pixel has one independent contact and shares the second contact with other pixels in the array. The distribution of the common contacts impacts electrical cross-talk and readout speed. A fundamental limitation in the development of arrays of detectors is that light is easily coupled to neighboring pixels in an array, which leads to the develop- ment of false counts, or cross-talk.78 There are approaches that may be exercised to mitigate this limitation, but they add additional complexity to the manufactur- ing.79 In addition, the array fabrication process becomes even more complicated by the requirement to maintain low leakage current in the individual pixels, mak- ing the fabrication process even more unwieldy.80 The progress in arrays has been steady and has paralleled the development of dense electronic structures such as DRAMs. 78 Don Phelan and Alan P. Morrison. 2008. Geiger-mode avalanche photodiodes for high time resolution astrophysics. Pp. 291-310 in High Timing Resolution Astrophysics, Don Phelan, Oliver Ryan, and Andrew Shearer, eds. New York: Springer. 79 J. Ziegler, M. Bruder, M. Finck, R. Kruger, P. Menger, T. Simon, and R. Wollrab. 2002. Ad - vanced sensor technologies for high performance infrared detectors. Infrared Physics and Technology 43(3-5):239-243. 80Alexis Rochas, Alexandre R. Pauchard, Pierre-A. Besse, Dragan Pantic, Zoran Prijic, and Rade S. Popovic. 2002. Low-noise silicon avalanche photodiodes fabricated in conventional CMOS technolo - gies. IEEE Transactions on Electron Devices 49:387-394.

OCR for page 23
seeing Photons  Manufacturing Infrastructure The manufacturing infrastructure for large array fabrication is discussed in Chapter 5. At this point it suffices to recognize that the manufacturing tools are largely those developed by the integrated circuit industry and adapted for FPA manufacturing. The FPA industry is not sufficiently large to support the develop- ment of a complete set of unique tools. As the evolution of the silicon industry is driven by a different set of goals, this can lead to divergence and to gaps in the FPA tool set. One simple example is that the silicon industry has standardized on a field size of 22 × 33 mm2 for its lithography tools. The drive to larger pixel counts for FPAs often requires much larger overall FPA sizes, which can only be accomplished by abutting multiple fields, requiring special considerations in the design of the focal plane arrays. FINDING 2-5 An advanced equipment set is required for manufacturing large-pixel-count detector arrays. Equipment availability is dependent on leveraging silicon CMOS developments. The detector market is not in itself sufficiently large to drive equipment development. CONCLUDING THOUGHTS Detection and imaging of electromagnetic radiation across the UV, visible, and infrared spectrum has a long history. As a result of its very advanced stage of technological development, silicon is now, and undoubtedly will continue as, the dominant material for visible sensors. One exception is the need for solar blind detectors that are insensitive to the solar spectrum after it is filtered bypassing through the ozone layer surrounding the Earth. Large-bandgap materials such as AlGaN are being actively developed for this application. First-generation night vi- sion systems used intensified (amplified) visible detection. Increased interest is now being placed on SWIR detection using InGaAs and related materials technology. Much of the progress at these longer wavelengths was catalyzed by the needs of the telecommunications industry for fiber-optic receivers. Infrared detectors have been under development for many years, primarily for military applications. The traditional material systems for cooled detectors are InSb for MWIR and HgCdTe for both MWIR and LWIR. Emerging material systems include SLS antimonides and intersubband transition QWIPs and QDIPs in the AlGaAs system. Both of these have the advantage of epitaxial growth on GaAs and possibly silicon substrates and of leveraging off of the mature GaAs technology developed for electronics and photonics. However, they are at a much earlier stage of development and technology readiness. Figure 2-9 shows the material systems relevant for different wavelength regimes.

OCR for page 23
fundamentals u lt r av i o l e t , v i s i b l e , infrared detectors  of and FIGURE 2-9 Material systems for UV-visible-infrared detection. Except for the bottom two entries, these material systems have been known and developed for decades. SOURCE: Presented to the committee by Dr. “Dutch” Stapelbroek, University of Arizona.