3
Key Current Technologies and Evolutionary Developments

INTRODUCTION

Advanced-technology sensors, coupled with data processing and fusion and networked communications, have enabled new approaches to warfighting and have been essential for reducing combat losses, minimizing collateral damage, and allowing a smaller but more effective military force. U.S. military strategy is expected to remain dependent on maintaining technological superiority, including the development of increasingly sophisticated sensor systems and new sensor types that exploit novel threat signatures. While the possibility exists that radically new sensor technologies will be invented, there are many opportunities to customize and improve the performance of existing, and in some cases relatively mature, focal plane technologies. There is an active research community pursuing these goals on a global scale. The focus of this chapter is to describe some of the likely near-term developments to existing visible and infrared (IR) detector technologies.

The proper definition of a sensing problem drives optimization of the design features of the focal plane to lead to the best system-level performance. In many cases, new sensor systems will be enabled not just by driving closer to “physical limits” but by tailoring designs and performance to be exceptionally well matched to a specific application. Clever design architectures, adding powerful on-focal-plane processing features, lowering power dissipation, increasing detector operating temperature, and dramatically reducing costs are examples of seemingly evolutionary changes that could enable revolutionary capabilities. Innovation will likely not be driven exclusively by moving closer to the physical limits of detector performance.



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 60
3 Key Current Technologies and Evolutionary Developments INTRODUCTION Advanced-technology sensors, coupled with data processing and fusion and networked communications, have enabled new approaches to warfighting and have been essential for reducing combat losses, minimizing collateral damage, and allowing a smaller but more effective military force. U.S. military strategy is expected to remain dependent on maintaining technological superiority, including the development of increasingly sophisticated sensor systems and new sensor types that exploit novel threat signatures. While the possibility exists that radically new sensor technologies will be invented, there are many opportunities to customize and improve the performance of existing, and in some cases relatively mature, focal plane technologies. There is an active research community pursuing these goals on a global scale. The focus of this chapter is to describe some of the likely near-term developments to existing visible and infrared (IR) detector technologies. The proper definition of a sensing problem drives optimization of the design features of the focal plane to lead to the best system-level performance. In many cases, new sensor systems will be enabled not just by driving closer to “physical limits” but by tailoring designs and performance to be exceptionally well matched to a specific application. Clever design architectures, adding powerful on-focal- plane processing features, lowering power dissipation, increasing detector oper- ating temperature, and dramatically reducing costs are examples of seemingly evolutionary changes that could enable revolutionary capabilities. Innovation will likely not be driven exclusively by moving closer to the physical limits of detector performance. 0

OCR for page 60
key current technologies evolutionary develoPments  and A number of advances, such as putting more processing into a pixel or making a smaller pixel, depend on continued improvements in silicon complementary metal oxide semiconductor (CMOS) process technology, driven by the semiconductor industry’s push to stay on the Moore’s law scaling curve. It is worth noting that device scaling is becoming increasingly technically challenging and expensive and that physical limits are becoming real roadblocks to CMOS scaling. Many process technologies developed for 65 nm, 45 nm, or 32 nm, while excellent in terms of digital circuit performance, are less than ideal for the analog portions of a pixel. They may have high transistor leakage levels and very limited dynamic range due to smaller voltage swings, thus limiting performance. Accordingly, device scaling is likely to be exploited by adding digital functionality to pixels. As more advanced CMOS process technologies are used, the cost per transistor drops, but the cost per area of CMOS chips increases. This, in turn, carries negative implications for focal planes that require physically large areas. Advanced processes are also costly and have large nonrecurring expenses as- sociated with design, mask generation, and initial design and debugging, making it expensive to develop custom devices needed only in small volumes. It is also worth noting that changes in the lithography processes used at smaller feature sizes require specialty techniques, such as field stitching, to produce large-area devices that exceed the field size of modern lithography tools. In short, while CMOS scaling will dramatically shape the design options for advanced focal planes, significant learning and innovation will be required to apply these advanced technologies optimally. In many cases, visible sensor applications do not benefit from the high- volume manufacturing imperatives that both drive and allow amortization of the increasing tool cost that accompanies CMOS scaling. A number of areas in which near-term technology advances are expected include ultralarge-format arrays; mosaic tiling technologies; pixel size reduction; smarter pixels and on-focal plane processing; improved three-dimensional (3-D) integration and hybridization; higher-operating-temperature devices; multicolor pixels; improved short-wavelength infrared (SWIR) arrays; photon counting tech- nologies and lower readout noise; curved focal surfaces; lower-power operation; radiation hardening; cost reduction; and improved cooler technology. KEY TECHNOLOGIES EXPECTED TO DRIVE ADVANCEMENTS IN EXISTING DETECTOR TECHNOLOGIES OVER THE NEXT 10-15 YEARS With respect to areas in which near-term technology advances are expected, this section examines each of the advances and addresses benefits, risks or draw- backs, impact on system performance, and implications for military applications of the expected advances.

OCR for page 60
seeing Photons  Ultralarge-format Focal Plane Arrays Progress is expected to continue in developing increasingly large arrays for the visible, SWIR, mid-wavelength infrared (MWIR), long-wavelength infrared (LWIR), and very long wavelength (VLWIR) portions of the spectrum. While both visible and IR arrays depend on the development of larger CMOS readout inte- grated circuits (ROICs), IR arrays have additional challenges—for example, yield- ing large detector arrays in difficult material systems such as mercury cadmium telluride (MCT), availability of large substrates, and developing high-yielding and small pixel-pitch bump bonding. Low-cost and high-performance (by some metrics) CMOS active pixel sensor (APS) multi-megapixel arrays are commercially available for the visible spectrum. Charge-coupled devices (CCDs) for scientific applications are routinely made with pixel counts exceeding 20 megapixels, and the trend to ever-larger single chips will continue. Using four-side buttable tiling, CCD focal planes of 1.4 gigapixels have been made and are in use by astronomers. In addition to monolithic silicon active-pixel sensors and CCDs, hybrid megapixel visible arrays will also become more available, providing sensitivity advantages over the traditional commercial products. Visible 50-megapixel arrays are now available with digital outputs hav- ing ROIC noise levels of less than 10 electrons and offering a sensitivity advantage over consumer products. In the IR, highly sensitive, multi-megapixel infrared focal plane arrays (FPAs), up to 4 megapixels, are also widely available for the MWIR and SWIR spectrum. Recently, 16-megapixel MWIR FPAs have been demonstrated. As shown in Fig- ure 3-1, detector array pixel count has paralleled the exponential growth of silicon dynamic random access memory (DRAM) bit capacity.1 Visible arrays with close to 100 megapixels offer the largest formats. In the IR, 16-megapixel arrays are now available. System-level benefits of large FPAs are generally related to providing a large instantaneous field of view (FOV). As focal planes become less expensive per pixel, it increasingly makes sense to eliminate costly, power-hungry, heavy, and unreli- able mechanical scanning or optical pointing systems and replace these with a fully electronic selection of the FOV by reading out a region of interest from a larger FPA. Large FPAs allow monitoring of large areas and are important for persistent surveillance applications. These FPAs enable important applications, such as high- resolution, wide-area airborne persistent surveillance and distributed aperture systems providing full spherical coverage of platforms. The single 8-inch ROIC wafer shown in Figure 3-2 contains 2K × 2K, 2K × 4K, 1 Paul Norton. 2006. Third-generation sensors for night vision. Opto-Electronics Review 14(1):1-10. Available at http://www.springerlink.com/content/h126r11q13747524/fulltext.pdf. Accessed March 24, 2010.

OCR for page 60
key current technologies evolutionary develoPments  and FIGURE 3-1 Detector array pixel count has paralleled the exponential growth of silicon DRAM bit capacity. SOURCE: Paul Norton. 2006. Third-generation sensors for night vision. Infrared Photoelectronics, SPIE Proceed- ings Vol. 5957. and 4K × 4K (4-, 8-, and 16-megapixel) die. Scaling up to the 16-megapixel FPA provides larger sensor FOV and improved full-Earth coverage of the ballistic mis- sile theater. Individual larger arrays are advantageous over tiling multiple smaller FPAs and result in 100 percent coverage without the additional effort required to account for the gaps between tiled arrays. FINDING 3-1 The evolutionary trends are semiconductor detectors characterized by in- creased pixel pitch and count, higher readout speed, higher operating tem- perature (especially MWIR), lower power consumption, and decreased sensor thickness. The need for larger fields of regard is a significant driver for larger arrays. Even beyond the diffraction limit of the optical system, oversampling can lead to slightly enhanced resolution. Mosaic Tiling Technologies For a number of years, visible-band CCD imagers have been built in three-side buttable and four-side buttable formats, allowing tiling of large focal planes from

OCR for page 60
seeing Photons  FIGURE 3-2 A single 8-inch ROIC wafer from 2007 Raytheon industry research and development. SOURCE: An - gelo Scotty Gilmore, Stefan Baur, and James Bangs. 2008. High-definition infrared focal plane arrays enhance and simplify space surveillance sensors. Raytheon Technology Today 1:5-8. Available at http://www.raytheon.com/newsroom/rtnwcm/groups/public/documents/content/rtn08_tech_sensing _pdf2.pdf. Accessed March 26, 2010. smaller, higher-yielding chips, with only modest gaps between the chips. For ex- ample, the 1.4-gigapixel orthogonal transfer array CCD imager used in PanSTARRS is composed of 60 chips, each of 22 megapixels. An array of this size could not be made monolithically since it exceeds the size of the largest silicon wafers used by the IC (integrated circuit) industry. Currently, and for the better part of the next de- cade, silicon wafer sizes are not expected to exceed the 300 mm diameter currently in use by leading-edge semiconductor facilities, although the silicon IC industry is actively exploring a transition to 450-mm-diameter wafers. Many imaging chips are made using process technologies being run on 200-mm-diameter wafers, for both cost and technological reasons.

OCR for page 60
key current technologies evolutionary develoPments  and In many cases, the size of a silicon chip is not limited by the wafer size, but it is limited by the ability to yield working chips, which falls off rapidly with increasing chip size. Achieving high yields on large chips can be expected to be most chal- lenging for extremely dense and complex designs using the most advanced process technologies. Tiling large arrays from smaller chips addresses the practical and economic limits of making larger monolithic chips. While tiled arrays are already in common use in high-end scientific and mili- tary applications, there are a number of areas in which improvements can be ex- pected. Gaps between chips can be reduced from hundreds of micrometers to as little as a few tens of micrometers, especially for monolithic technologies. Greater use of four-side buttable designs is expected. Techniques will be developed to sim- plify interconnections to the tiles and to lower the cost of tiled arrays. Improved manufacturability and repairability are also active areas of research. Application areas for large-format tiled arrays with minimal seam loss are likely to include wide-FOV telescope systems with large optics and long focal lengths for such diverse applications as astronomy, space surveillance, and persis- tent surveillance. Pixel Size Reduction A general trend has been to reduce pixel sizes, and this trend is expected to continue. Several reasons exist for reducing pixel size, and the desirability of reducing pixel size is dependent on details of the application and the operating wavelength. In general, systems operating at shorter wavelengths (e.g., visible and ultraviolet) are more likely to benefit from small pixel sizes because of the smaller diffraction-limited spot size. Diffraction-limited optics with low F-numbers (e.g., F/1) could benefit from pixels on the order of one wavelength across, as small as about 0.5 µm in the vis- ible or about 10 µm in the LWIR. Oversampling the diffractive spot may provide some additional resolution for smaller pixels, but this saturates quickly as the pixel size is decreased. Commercial CMOS imagers and CMOS chips have been demonstrated with pixels sized in the 1 to 2 µm range, with some examples less than 1 µm. On the sili- con CMOS imager side, much of the interest in pixel size reduction has been driven by the desire to deliver a large number of megapixels to a consumer while keeping the silicon area used by the chip as small as possible to minimize cost for consumer applications, such as cell phone cameras. Interestingly, many of these cameras have low-performance optics, leading to much poorer resolution than might be expected based on the megapixel count for the imager. Visible-band small-pixel imagers are useful with small optics and can find application in unattended ground sensors as well as other systems requiring small surveillance cameras. In the IR there remains a steady emphasis on improving uncooled microbo-

OCR for page 60
seeing Photons  lometers that will continue to mature with smaller detector sizes and larger formats. Current products utilize a 17 µm pitch and are available in high-definition formats (640 and 1280) primarily in the LWIR band. MWIR arrays have been fabricated, but they are limited by detector noise due to the lower MWIR photon flux. In the near future, uncooled 10-12 µm detector pitch arrays will be available in high-definition format (1920 × 1080). This reduction in pitch will enable a reduction in optics size allowing increased range capability without an increase in weight for man-portable applications. For wide-area persistent surveillance, programs such as ARGUS-IS,2 as shown in Boxes 3-1 and 3-2, are already working on extremely large mosaics of visible FPAs to enable constant surveillance of large city areas.3 With the smaller-pitch FPAs becoming available, these types of systems should be more affordable and smaller. The technology challenge is to maintain sensitivity and reduce the thermal time constant as the pitch size is decreased. Uncooled IR technologies also will be developed that exploit piezoelectric and other temperature response mechanisms, such as bimetallic microelectromechanical systems (MEMS) structures, diode for- ward voltage changes, or capacitance changes. FINDING 3-2 The global proliferation of low-cost, commodity imagers, such as cell phone cameras and automobile thermal imagers, enables adversaries to develop sens- ing systems at relatively low cost, reducing the barrier to achieving limited operational capabilities. As an example, the rapid proliferation of low-cost “night vision technology” is eroding the overwhelming dominance of the United States in nighttime operations, even with the superior performance of advanced systems. FINDING 3-3 The availability of very low cost imagers developed for large consumer markets is providing opportunities to develop new sensor systems and architectures, even though the component-level imagers may not have the capabilities typical of high-performance sensors developed specifically for military applications. Additionally, the technology and manufacturing base used to make these low- cost imagers will extend the manufacturing base that can be used for fabricat- ing customized military parts. 2 Brian Leininger, Jonathan Edwards, John Antoniades, David Chester, Dan Haas, Eric Liu, Mark Stevens, Charlie Gershfield, Mike Braun, James D. Targove, Steve Wein, Paul Brewer, Donald G. Madden, and Khurram Hassan Shafique. 2008. Autonomous Real-time Ground Ubiquitous Surveil- lance—Imaging System (ARGUS-IS) Defense Transformation and Net-Centric Systems. Proceedings of SPIE 6981:69810H. 3 The data management challenges posed by a system such as ARGUS-IS are addressed in Chap - ter 4.

OCR for page 60
key current technologies evolutionary develoPments  and RECOMMENDATION 3-1 The intelligence community should pay careful attention to the new capa- bilities inherent in both the proliferation of commodity detector technologies and their integration into novel sensor systems. ARGUS-IS and Gnuradio are examples of how available, low-cost, mature commodity visible FPA technol- ogy (cell phone camera chips) and commercial off-the-shelf (COTS) commu- nications circuitry, through sensor integration, have enabled new, advanced, high-performance imaging capabilities. Smarter Pixels and On-focal-plane Processing A trend has been to take advantage of progress in electronics scaling to integrate as many functions as possible on a single chip. Visible-band CMOS active pixel sensors are particularly well developed examples of this trend. Usually the desire is to achieve cost reductions for high-volume applications by having a single-chip solution, but monolithic integration can also have other substantial benefits, such as power and noise reduction and enabling new interconnection-rich processing architectures that would not be feasible using off-chip inputs and outputs. In the IR and high-performance visible imaging area, fully digital focal planes are just entering the market. In some cases their electro-optical (EO) performance exceeds that of traditional analog focal planes coupled to discrete electronics. High-performance IR scanners and large-area staring visible digital focal planes pixels are currently being demonstrated and show size, weight, and power (SWaP) advantages. On-ROIC digital logic enables future digital signal processing such as nonuniformity correction (NUC), image stabilization, and compression. These focal planes will enable much smaller systems (microsystems). This technology is poised to go into large-area cryogenic infrared sensors over the next few years. Sending data off-chip requires substantial power and sending those data through communication links—for example, a radio-frequency (RF) link for an unattended ground sensor—can be even more energy intensive. There is increas- ing recognition of the value of trying to identify and transmit only small amounts of high-level and actionable information rather than large numbers of raw data bits. This is particularly important for systems with severe power or communica- tion bandwidth constraints. This is leading to specialized imaging chips that have on-chip processing tuned to a specific application. Such customization can lead to dramatic improvements in system performance, but it has the drawback that longer design and fabrication cycles and large nonrecurring costs may be required to make application-specific chips. The amount of processing that can be placed right at the pixel has generally been limited by the modest numbers of transistors that can be placed within a small pixel. Moore’s law device scaling has steadily been improving that situation, and focal planes are now being developed with reasonably small pixels and significant

OCR for page 60
seeing Photons  BOX 3-1 Case Study: DARPA ARGUS-IS The DARPA ARGUS-IS unmanned aerial system is designed for persistent surveillance and incorporates a 1.8- gigapixel composite visible sensor system composed of 184 FPA pairs, for a total of 368 FPAs. ARGUS-IS is demonstrating on-board real-time processing of wide-area video imagery. Figure 3-1-1 depicts the components of the system. Figure 3-1-2 outlines the specific on-board processing solution, details of which are discussed in Chapter 4, and Figure 3-1-3 provides a sample output image. Airborne Sensor Processing Gigapixel Sensor Power / cooling etc Airborne Stabilized 6 axis gimbal Equipment & Processor Processing Serial Data link Hardware & connections in Software A-160 Pod (modified) Many MB/sec down Some KB/sec up Ground Processing Processing Ground LAN/WAN Decom- Data Window Based pression control link Equipment/ User User User window ••• window window Processing FIGURE 3-1-1 ARGUS-IS block diagram. SOURCE: Leininger, Brian, Jonathan Edwards, John Antoniades, David Chester, Dan Haas, Eric Liu, Mark Stevens, Charlie Gershfield, Mike Braun, James D. Targove, Steve Wein, Paul Brewer, Donald G. Madden, and Khurram Hassan Shafique. 2008. Autonomous Real-time Ground Ubiq - Figure 3-1-1 uitous Surveillance-Imaging System (ARGUS-IS). Proceedings of the SPIE 6981:69810H-1. PMC with ADV212 - JPEG 2000 Serial sensor Virtex 5 FPGA control ADV212 - JPEG 2000 EMCORE Multimedia Fiber Dual Core Pentium 3 Fibers - 6 FPA Video Streams Optic Converter Single Board Computer 160 GB 2.5” with 3 1Gbit/sec SATA HD EMCORE Ethernet 3 Fibers - 6 FPA Video Streams Multimedia Fiber Optic Converter ADV212 - JPEG 2000 PMC with Virtex 5 FPGA ADV212 - JPEG 2000 1 Gbit Ethernet 6U cPCI Single Board Computer FIGURE 3-1-2 ARGUS-IS airborne processing module. SOURCE: Leininger, Brian, Jonathan Edwards, John Antoniades, David Chester, Dan Haas, Eric Liu, Mark Stevens, Charlie Gershfield, Mike Braun, James D. Targove, Steve Wein, Paul Brewer, Donald G. Madden, and Khurram Hassan Shafique. 2008. Autonomous Real-time Ground Ubiquitous Surveillance-Imaging System (ARGUS-IS). Proceedings of the SPIE 6981:69810H-1. Figure 3-1-2

OCR for page 60
key current technologies evolutionary develoPments  and FIGURE 3-1-3 Sample of ARGUS-IS imagery. Mounted under a YEH-60B helicopter at 17,500 feet over Quantico, Va., Argus-IS images an area more than 4 km wide and provides multiple 640 × 480-pixel real-time video windows. SOURCE: Image courtesy of BAE Systems.

OCR for page 60
seeing Photons 0 BOX 3-2 Impact of Commodity Components One observation to highlight is the use of readily available commodity components in the ARGUS-IS system; the ARGUS-IS designers state: Though many other processing solutions exist, the rapid deployment schedule implied use of low-risk COTS [commercial off-the-shelf], or close to COTS, processing hardware. Hence, other processing elements were considered, but rejected for various reasons including: power consumption (multicore central processor units (CPUs) and graphics processor units (GPUs)), lack of configurability and development time (custom ASICs), or lack of proven integration tools or processing margin. Multicore processors and GPUs have become COTS since 2007 and enjoy certain performance and power benefits. More importantly, the observation about rapid deployment schedules using commod - ity components has not been lost on other design teams (e.g., at the Iran University of Science and Technology,a Xidian University,b and Nanjing University of Science and Technology).c Additional examples of commercial developments relevant to the communication architectures of remote imaging systems include the Gnuradio architecture,d with its signal processing architecture partitioned between a universal software radio peripheral with selectable daughtercards and a host processor with more or less arbitrary signal processing performance. A third example is the use of multiple-input multiple-output (MIMO)e,f in radio systems, where multiple antennas and significant signal processing are used to overcome localized radio-frequency challenges such as multipath. aB. Zamanlooy, V.H. Vaghef, S. Mirzakuchaki, A.S. Bakhtiari, and R.E. Atani. 2007. A real time infrared imaging system based on DSP & FPGA. Pp. 16-23 in Lecture Notes in Computer Science. Berlin: Springer-Verlag. bHuixin Zhou, Shangqian Liu, Rui Lai, Dabao Wang, and Yubao Cheng. 2005. Solution for the nonuniformity correction of infrared focal plane arrays. Applied Optics 44(15):2928-2932. cJ. Zhang, Y. Qian, B. Chang, S. Xing, and L. Sun. 2005. Signal processing of microbolometer infrared focal-plane arrays. Pp. 137-145 in Infrared Components and Their Applications. Bellingham, Wash: SPIE. dAvailable at http://gnuradio.org/redmine/wiki/gnuradio. eGregory G. Raleigh and John M. Cioffi. 1998. Spatio-temporal coding for wireless communication. IEEE Transactions on Communications 46(3):357-366. fG.J. Foschini. 1996. Layered spacetime architecture for wireless communication in a fading envi - ronment when using multiple antennas. Bell Labs Syst Tech J 1:4159.

OCR for page 60
seeing Photons 0 Cost Reduction Visible sensors, such as interline-transfer CCDs and CMOS active-pixel sen- sors, have seen dramatic cost reductions as large commercial markets, such as consumer cameras and cell phone cameras, have driven large production volumes. These parts are high performance in the sense that they represent state-of-the-art technology and are highly optimized for their intended markets. These parts may also be very low performance compared to parts optimized for a specific DOD ap- plication. While these low-cost, consumer-driven parts are not expected to replace all high-performance custom parts, they are opening up new approaches to DOD sensors because of their low cost and high performance-price ratio. One example is the recent use of large numbers of multi-megapixel cell phone CMOS imagers in wide-FOV video surveillance applications and airborne persistent surveillance systems (see above and Chapter 4 for additional information on this system). The proliferation of low-cost, visible sensors offers opportunities for DOD if it is quick to exploit this technology; however, such technologies are proliferated globally and are also accessible to current and potential adversaries.11 As discussed above, in the IR, alternatives to MCT such as InAs-GaSb SLS detectors may provide both a large decrease in cost and an increase in key per- formance capabilities. This detector technology will exploit the established III-V technology and manufacturing base and may also avoid some of the inherent producibility challenges associated with II-VI IR detectors. These materials can be grown by more traditional III-V molecular beam epitaxy (MBE) wafer vendors that can provide these wafers to the IR industry very much as bulk InSb wafers are supplied today. If successful, these material systems would enable an evolutionary cost reduction in sensors due to simpler technology requirements once it has been established. SLS devices are predicted theoretically12 to have lower dark currents than HgCdTe; SLS is limited by Shockley-Read generation mechanisms that pro- duce current via point defects in the as-grown material.13,14,15 Future advancements in SLS technology will require a significant effort and investments to reduce the defect densities. 11 There is a fuller discussion of these trends in Chapter 4. The ready availability of capable, off-the- shelf imagers will have a dramatic impact on the availability of future imaging systems for friends and foes alike. 12 M.E. Flatte and C.H. Grein. 2009. Theory and modeling of type-II strained-layer superlattice detectors. Proc. SPIE 7222:72220Q. 13 J. Pellegrino and R. DeWames. 2009. Minority carrier lifetime characteristics in type II InAs/GaSb LWIR superlattice n+pp+ photodiodes. Proc. SPIE 7298:72981U. 14 D.R. Rhiger, A. Gerrish, and C.J. Hill. 2008. Estimation of carrier lifetimes from I-V curve fitting for InAs/GaSb and HgCdTe LWIR diodes. Proc. MSS Parallel Meeting. 15 M.E. Flatte and C.H. Grein. 2009. Theory and modeling of type-II strained-layer superlattice detectors. Proc. SPIE 7222:72220Q.

OCR for page 60
key current technologies evolutionary develoPments  and Both cost reduction and the multiple open-source supplier structure of SLS detectors could have significant impacts in the widespread proliferation of large- format IR FPA technology. FINDING 3-7 There is significant opportunity to customize image sensor architectures for specific applications that can lead to dramatic improvements in system-level performance, including size, weight, and power. Advanced architectural design, including integration of sensing and processing (in-pixel and on-chip), can have greater system-level impact than making small gains in driving detector performance incrementally closer to fundamental detectivity limits. RECOMMENDATION 3-3 The intelligence community should evaluate and track system capabilities rather than focusing solely on component technical achievements. These in- clude technologies that enable in-pixel and on-chip processing, lower-power operation, and higher operating temperatures, as well as technologies that improve manufacturability. Improved Cooler Technologies The availability of improved cooler technology can have a huge impact on the system-level performance of detector systems and may make or break whether a given focal plane technology can be considered for a given application. At least four dimensions for improvements in coolers can be identified—each dimension results in different system impacts—including the following: 1. Development of higher-coefficient-of-performance coolers: Visible and SWIR systems using thermoelectric coolers, as well as IR systems using refrig- eration cycles, would benefit from more efficient coolers, because cooling power often dominates total sensor power consumption. This is particularly important for power-constrained systems, such as unattended ground sen- sors, and systems that are used only intermittently but must continuously be in a ready-to-operate state. 2. Reliability improvements and space qualification: A number of space systems would benefit from the availability of highly reliable and fully qualified cryogenic coolers. 3. Lower cost: For systems that must be produced in large quantities—for example, for individual soldiers—and especially for “disposable” systems, low cost is critical. 4. More compact form factor: The availability of compact and lightweight

OCR for page 60
seeing Photons  coolers is a prerequisite for a number of applications. An example would include deployment of an imaging system on a micro-air vehicle. Temperature control and stability are crucial parameters in determining the ultimate resolution, signal to noise, and other performance attributes of various IR and visible EO detectors. For this reason, detectors are often categorized as being either cooled or uncooled. Cooled detectors refer to platforms that require cryogenic temperatures in order to operate. Typical temperatures of operation for cooled sensors range from <10 to 150 K or slightly higher. Uncooled detectors, despite their title, typically incorporate some degree of temperature control near or slightly below room temperature (~250-300 K) to minimize noise, optimize resolution, and maintain stable operating conditions. The two technologies cur- rently available for addressing the cooling requirements of IR and visible detectors are closed-cycle refrigerators and thermoelectric coolers. Closed-cycle refrigerators can achieve the cryogenic temperatures required for cooled IR sensors, while ther- moelectric coolers are generally the preferred approach to temperature control for uncooled visible and IR sensors. Although uncooled sensors offer significant advantages in terms of cost, life- time, and SWaP, cooled sensors offer significantly enhanced range, resolution, and sensitivity as a result of the reduced dark current and, therefore, lower-noise opera- tion. For this reason, cooled IR sensors have been the technology of choice for many military operations where performance, not cost, is the prime driver. Conversely, for many commercial applications, uncooled sensors are the preferred technology because of their lower cost and higher reliability. In this section, the current status of various cooling technologies is reviewed with respect to their present and near- term applicability to IR and visible detectors. Thermoelectric coolers and closed-cycle or mechanical refrigerators both in- volve the use of a working “fluid” to transfer heat from a thermal source to a thermal sink. The major difference between the thermoelectric and mechani- cal cryocoolers is the nature of the working fluid. A thermoelectric cooler is a solid-state device that uses charge carriers (electrons or holes) as a working fluid, whereas mechanical cryocoolers use a gas such as helium as the working fluid. While each cooling technology has its advantages and disadvantages, limitations in these cooling technologies translate directly to performance limitations in the IR and visible detectors to which they are applied. Conversely, any improvements and breakthroughs achieved in cooling technologies could provide enhancements to detector applications in SWaP metrics and in overall detector performance. Cryocoolers As mentioned previously, mechanical cryocoolers or refrigerators represent the only currently available technology that can reach the cryogenic temperatures

OCR for page 60
key current technologies evolutionary develoPments  and required for many IR and visible sensors. As shown in Figure 3-4, cooling of IR sensors represents one of a number of commercial and research applications for cryocoolers. The major commercial applications for mid- to large-scale cooling in- clude cryopumps for semiconductor fabrication facilities, magnetic resonance im- aging (MRI) magnet cooling, and gas separation and liquefaction. For low-power applications, IR sensors represent the largest single application for cryocoolers. The most common types of cryocoolers can be classified as either recuperative or regenerative. In recuperative systems, gas flows in a single direction. The gas is compressed at ambient fixed temperature and pressure and allowed to expand through an orifice to the desired cryogenic fixed temperature and pressure. The Joule-Thompson and Brayton cycle refrigerators are examples of recuperative systems. In a regenerative system, the gas flow oscillates back and forth between hot and cold regions driven by a piston, diaphragm, or compressor, with the gas being compressed at the hot end and expanded on the cold end. Stirling, Gifford- McMahon, and Pulse Tube cryocoolers are the most common types of regenerative cryocooler systems. An important parameter in measuring the performance of a cryocooler is the coefficient of performance (COP). COP is a measure of efficiency and is defined as the ratio of cooling power achieved at a particular temperature to total electrical input power to the cryocooler. Often, the COP is given as a fraction of the ideal FIGURE 3-4 Cooling of IR sensors represents one of several commercial and research applications for cryocoolers. NOTE: FCL = freon coolant line; HTS = high-temperature superconductivity; LNG = liquid natural gas; LTS = low-temperature superconductivity; MRI = magnetic resonance imaging; SMES = superconduct- ing magnetic energy storage; SQUID = superconducting quantum interference device; ZBO = zero boil-off. SOURCE: Radebaugh, Ray. 2009. Cryocoolers: the state of the art and recent developments. Journal of Physics: Condensed Matter 21:164219.

OCR for page 60
seeing Photons  or Carnot efficiency, which is defined as the ratio of Tc /(Tc – Th), where Tc and Th refer to cold and hot temperatures, respectively, of operation. In general, recupera- tive systems have advantages in terms of reduced noise and vibration, whereas regenerative systems tend to obtain higher efficiencies and greater reliability at the temperatures of interest for many IR detector applications. The relative perfor- mances of the different technologies as a fraction of the limiting Carnot efficiency are shown in Figure 3-5. Significant developments in cryocooler technologies during the past several decades have facilitated their increased use in commercial markets. One major area of improvement has been reliability. Because impurities in the working gases can freeze and either clog or damage the various internal components of a cryocooler responsible for gas flow, maintaining a high-purity gas over the lifetime of the cryo- cooler is a key challenge. The development of novel adsorber materials and designs, as well as improved sealing techniques for maintaining the high pressures required FIGURE 3-5 Relative performances of the different technologies as a fraction of the limiting Carnot efficiency. SOURCE: Radebaugh, Ray. 2009. Cryocoolers: the state of the art and recent developments. Journal of Physics: Condensed Matter 21:164219.

OCR for page 60
key current technologies evolutionary develoPments  and for cryocooler operation, has significantly contributed to the improved lifetimes now achievable in many cryocooler systems (5,000-10,000 hours). Improvements in heat exchanger and recuperator designs and materials have yielded absolute COP values for cryocoolers in the range of 0.01 to 0.1 for IR detector applications. Despite these advances, the cryocooler remains a major failure point for cooled IR detectors, as well as more than doubling the weight and power requirements of the integrated IR sensor system. While the needs of commercial and military cryocooler applications will con- tinue to drive improvements in cost, reliability, and efficiency, it is likely that these improvements will be incremental over the next 10-15 years. Areas for improve- ment include the ability to operate regenerative systems at higher frequency and the ability to operate using gases at higher pressures.16,17 Operation at higher frequency enables the use of smaller, lighter-weight compressors that are often the dominant volume in regenerative cryocoolers. Similarly, increasing the gas pressure can enable larger COP by increasing the effective thermal capacitance per cycle of operation. Designs and concepts currently being considered for operation at higher frequency and pressure, as well as ongoing material and process developments, will ultimately lead to improved seals and structural performance for cryocooler components and enclosures.18 There will likely continue to be a trade-off between cost and perfor- mance as commercial and military demands for improvements in SWaP drive the technology to more demanding operating conditions where maintaining lifetimes in excess of 10,000 hours will continue to be a challenge. Thermoelectric Coolers Thermoelectric materials can be used to convert thermal energy to electricity or to use electricity to pump heat. In a generator configuration, thermoelectric devices exploit the Seebeck effect—the voltage created between two dissimilar conductors in the presence of a temperature difference. Thermoelectric coolers work by exploiting the Peltier effect, which refers to the creation of heat flux at the junction of two dissimilar conductors in the presence of current flow. In either configuration, the optimal performance of a given set of thermoelectric materials in a device is determined by the dimensionless figure of merit ZT = S 2σ/κ, where S = Seebeck coefficient (thermopower), σ = electrical conductivity, κ = thermal conductivity, and T is the average temperature (T1 + T2)/2. Due to performance 16 S. Vanapalli, M. Lewis, Z. Gan, and R. Radebaugh. 2007. 120 Hz pulse tube cryocooler for fast cooldown at 50 K. Appl. Phys. Lett. 90:072504. 17 S.L. Zhu G.Y. Yu, W. Dai, E.C. Luo, Z.H. Wu, X.D. Zhang. 2009. Characterization of a 300 Hz thermoacoustically-driven pulse tube cooler. Cryogenics 49:51. 18 Ray Radebaugh. 2009. Cryocoolers: the state of the art and recent developments. Journal of Physics: Condensed Matter 21(16):164219-164228.

OCR for page 60
seeing Photons  limitations, thermoelectric generators and coolers have found only a few niche commercial applications. As generators, these applications include radioisotope thermoelectric generators (RTGs) for space applications and waste heat conversion to power small electric devices such as fans, lights, or battery chargers. Commercial applications of thermoelectric coolers include (1) climate-controlled seating for automobiles; (2) temperature control and stability for bolometric and ferroelectric detectors, laser diodes, and ink jet printers; (3) dark-current reduction in mid-wave IR detectors; and (4) noise reduction in CCD arrays. Thermoelectric coolers are essentially solid-state heat pumps where the flow of thermal energy is determined by the polarity of the applied current. The coef- ficient of performance for a thermoelectric cooler is a function of ZT as well as the overall temperature difference between the hot side and the cold side of the cooler. At maximum temperature difference, ∆T, the COP of a thermoelectric generator goes to zero (i.e., no heat can be removed). Conversely, at zero ∆T, a thermoelec- tric cooler achieves maximum heat pumping capacity. The typical load profile for a thermoelectric cooler is shown in Figure 3-6. Therefore, for most thermal management applications, including temperature stabilization for IR and visible FIGURE 3-6 Typical load profile for a thermoelectric cooler. SOURCE: Rama Venkatasubramanian and colleagues, Research Triangle Institute, Research Triangle Park, N.C.

OCR for page 60
key current technologies evolutionary develoPments  and “uncooled” detectors, thermoelectric coolers typically operate at the minimal ∆T that provides acceptable detector performance. Historically, the maximum achievable material ZT stood at ~1 and saw little improvement despite many decades of research starting in the 1960s through the early 2000s. However, in the early 1990s, DOD began investing significant research-and-development funding to explore the potential for achieving higher ZT values for both cooling and power generation applications. These investments paid off with the reports of breakthrough ZT values in low-dimensional thin-film thermoelectric materials. The enhancement in ZT achieved via nanostructuring in thin-film materials has prompted similar investigations in bulk materials. Recent reports have indicated that bulk thermoelectric performance in the ZT range of about 1.5 near room temperature and approaching 2.0 at higher temperatures is achievable.19 The bulk material discoveries have prompted the formation of several com- mercial new start companies—for example, GMZ and ZT Plus—and it is likely that these activities will transition directly to improved thermoelectric devices within the next 5-10 years. The potential impact of the thin-film materials is less certain since reproducibility, scalability, and fabrication costs remain significant challenges. For this reason, they are discussed in detail in Chapter 4. This section focuses on what is currently achievable with existing bulk commercial thermoelectric materi- als and what developments are likely to occur over the next 10-15 years.20 Current commercially available thermoelectric coolers are based on alloys of bismuth telluride and antimony telluride materials. These materials exhibit ZT values close to 1 unity, but in a device configuration achieve values closer to ZT = 0.7. In a single-stage device, this equates to a maximum ∆T of about 70 K as mea- sured from an ambient temperature of ~ 293 K. An approximation of maximum achievable ∆T can be obtained using the relationship ∆Tmax ~ 1/2ZTc2. However, as previously mentioned, at maximum ∆T, a thermoelectric cooler cannot dissipate any heat. Therefore, in order to achieve ∆T of 70 K or more and still have some cooling capacity, thermoelectric coolers are stacked in “stages.” The idea is to have each stage operating at less than ∆Tmax, with its “hot” side starting at the “cold” 19 Poudel, Bed, Qing Hao, Yi Ma, Yucheng Lan, Austin Minnich, Bo Yu, Xiao Yan, Dezhi Wang, An - drew Muto, Daryoosh Vashaee, Xiaoyuan Chen, Junming Liu, Mildred S. Dresselhaus, Gang Chen, and Zhifeng Ren. 2008. High-thermoelectric performance of nanostructured bismuth antimony telluride bulk alloys. Zhifeng Science 320:634. 20A good literature summary of the current status of TE technology is found in Lon E. Bell. 2008. Cooling, heating, generating power, and recovering waste heat with thermoelectric systems. Science 321:1457-1461. Available at http://www.sciencemag.org/cgi/reprint/321/5895/1457.pdf?maxtoshow= &HITS=10&hits=10&RESULTFORMAT=&author1=bell&titleabstract=an+information-maximizat ion+approach+to+blind+separation+and+blind+deconvolution&searchid=1&FIRSTINDEX=170& fdate=//&tdate=//&resourcetype=HWCIT. Accessed March 24, 2010.

OCR for page 60
seeing Photons  FIGURE 3-7 Theoretical range for maximum ∆T achievable using existing commercial materials as a function of the number of stages. SOURCE: Rowe, D.M., ed. 1995. CRC Handbook of Thermoelectrics. Boca Raton, Fla.: CRC Press, Inc. side of the prior stage. The theoretical range for ∆Tmax achievable using existing commercial materials is plotted in Figure 3-7 as a function of the number of stages. Although the figure shows values for up to a 10-stage device, commercially available coolers typically do not go beyond 6 stages. If recently reported ZT advances in bulk bismuth telluride can be transitioned to commercial products, it is reasonable to assume that device-level ZTs will see some modest improvements, perhaps as much as a 50 percent improvement in ∆Tmax for a single-stage device. In terms of efficiency, existing thermoelectric cool- ers can achieve COPs of ~1.0 (i.e., 1 W of cooling for 1 W of electrical input) at ∆T of ~30 K. If the bulk materials available for commercial coolers were to achieve ZT values closer to 1.5, then the achievable COPs would be anticipated to approach 1.5 (1.5 W of cooling for 1 W of electrical input) at similar 30 K temperature dif- ferentials. There is some evidence that modest improvements in bulk materials

OCR for page 60
key current technologies evolutionary develoPments  and are transitioning to commercial thermoelectric (TE) products.21 The reduction in material usage is attributable to anticipated improvements in material properties. There are a number of commercial factors driving improvements in TE ma- terials. In fact, the efforts to commercialize new, advanced bulk materials, coupled with industry efforts to improve cooler design and minimize parasitic losses, make it quite likely that these projected improvements will be realized within a 10-15 year time frame. For example, based on available data, the current market for TE coolers is relatively small, but it appears to be growing at an average rate of between 10 and 40 percent per year across a number of applications. One recent estimate puts the total TE cooler market at approximately $200,000,000.22 These estimates include TE coolers for a number of applications including climate control, EO cooling, personal coolers, and biomedical refrigeration. Figure 3-8 shows an aver- age growth of roughly 40 percent per year between 2001 and 2008. In contrast, a Wall Street Journal 2004 article reported that Igloo products estimated the personal cooler (small coolers for refrigerating drinks and food) market at $50 million and predicted a 10 percent annual growth. Given these estimates it is reasonable to as- sume that a significant portion of the commercial drivers for improved TE cooler technology are coming from nondetector applications. In summary, both cryocoolers and thermoelectric coolers are poised to achieve modest improvements in performance over the next 10-15 years in terms of ef- ficiency. In addition, cryocooler costs and reliability at today’s performance levels are likely to improve due to commercial market drivers associated with their use in the semiconductor (cryopumps), medical (MRI, cryosurgery), and gas liquefaction industries. Reliability at today’s SWaP requirements will likely increase to exceed 10,000 hours, while costs will likely be driven lower as these markets continue to grow. For niche applications, such as high-end detectors and large staring arrays, advances in SWaP will be critical. The trend toward larger-footprint arrays is particularly challenging because it requires greater thermal stability and minimal vibration across a larger area. Although concepts for operating at higher frequency and higher pressure are likely to result in improved SWaP for cryocoolers, these advances will probably be obtained, at least initially, at the expense of reliability and cost. In addition, the higher frequency of operation may induce additional unwanted vibrational noise and susceptibility to electromagnetic interference. FINDING 3-8 Low-cost uncooled infrared focal plane arrays are approaching the perfor- 21 Lon Bell, Amerigon, Inc., personal communication with the committee on March 3, 2010. Ameri - gon anticipates material usage in its products to decrease linearly over the next four years from 16 g to 11 g. 22Available at http://www.its.org/node/5263. Accessed March 24, 2010.

OCR for page 60
seeing Photons 0 Revenue ($ in Millions) CCS Unit Volume FIGURE 3-8 Estimates of the TE cooler market. SOURCE: Courtesy of Amerigon CCS. 3-8 w-replacement type.eps mance needed for applications that have traditionally relied on expensive cooled devices. FINDING 3-9 For both cryocooler and TE cooler technologies, there are a number of com- mercial market drivers, separate from sensor cooling applications, that will drive evolutionary improvements in SWaP. Over the next 10-15 years, it is reasonable to expect that these improvements will achieve overall reductions in SWaP on the order of 20-30 percent. CONCLUDING THOUGHTS In addition to a proliferation of sensor numbers and types, new sensors must be developed to exploit unique target and background phenomenologies and be capable of processing significantly larger volumes of data. Advanced technology sensors have the potential to be used in novel sensing situations—for example, the vetting of potential enemies or the identification of combatants and noncomba- tants in a counterinsurgency operations. IR sensors with high-performance imag- ing FPAs have that capability and have long been critical to U.S. relative military superiority. Maintaining this superiority requires continual improvements in the technologies required for advanced ROICs and detector materials growth and con- tinual awareness and incorporation of advances originating from both domestic and foreign developments.