3
Component Technology
This chapter reviews areas in which the committee believes there is potential for technology breakthrough devices or techniques that will result in enhanced performance of millimeter-wavelength/terahertz technologies or that may even serve as the minimum required to provide a useful capability. This information could serve as a flag for the attention of Transportation Security Laboratory (TSL) personnel in the future.
One of the best reviews of component maturity in the millimeter-wavelength/ terahertz region is by Siegel,1 describing in detail the state of the art of numerous
techniques for the generation and detection of radiation above 100 GHz. Although there are a few exceptions, little has changed in the state of the art since that review was written. The present report provides a more basic review of the technology, and the referenced paper can be consulted where additional details are required.
The three regions of the electromagnetic (EM) spectrum being examined in this chapter have different levels of component maturity but are not divided along the lines previously defined (millimeter wave, submillimeter wave, and terahertz). The component development has been led by military applications and then followed by commercial applications. Starting from the radio-frequency (RF) side of the electromagnetic spectrum, numerous commercial foundries exist to provide components below 50 GHz, and limited capability is available up to about 100 GHz. Above 100 GHz, choices are more limited. Some amplifiers have been developed to date that operate up to 200 GHz, primarily as low-noise amplifiers (LNAs) for receivers. Above 200 GHz, high-sensitivity receivers and associated low-power sources have been primarily developed for the radio astronomy and remote sensing community.
SOURCES
Sources are probably the first component requirement for active spectroscopy or imaging systems operating in the millimeter-wavelength/terahertz spectra. They serve as a means to calibrate systems, as illuminators for sensors measuring reflections from or transmissions though materials, or as the local oscillator (LO) for heterodyne receivers. Heterodyne receivers convert a received signal into a lower frequency through multiplication with a reference source. The heterodyne receiver is generally more sensitive than a direct detecting receiver, as this down-conversion process allows the use of lower frequency and thus more sensitive detection schemes. The use of stable narrowband sources with a heterodyne receiver also provides an ability to measure spectral features rapidly with high resolution.
Figure 3-1 briefly summarizes the challenge with achieving a system concept between 0.1 THz and 10 THz. Generally, the approach to achieving source power has been either to use multipliers to generate radiation from RF sources or to translate down in frequency from optical regions using laser and various forms of nonlinear mechanism. There are exceptions to this trend in that backward wave oscillators (BWOs), a vacuum electronic device, and carbon dioxide (CO2) pumped gas lasers have been available for many years and have provided power adequate for the applications of interest, namely, spectroscopy. The various techniques currently being investigated for generating power for security are discussed below with respect to the frequency range over which they function most efficiently.
Figure 3-2 shows the state of the art in RF sources. A first point is that obviously there appears to be a “cliff” that technology falls off above 100 GHz. Most source advances in the past 30 to 40 years have been due either to high-energy physics research or to military applications, with most of the investment in affordable and readily manufacturable sources below 100 GHz for the military. While commercial interest has further improved the affordability of sources, primarily below 30 GHz, it has done little to improve power levels. Little has happened since Figure 3-2 was developed to change
these performance levels. The Defense Advanced Research Projects Agency (DARPA) has recently funded a program developing new sources in the region of 0.5 THz to 2.0 THz and a second, which began in 2006, to produce useable monolithic microwave integrated circuits (MMICs) up to 340 GHz.
A second point, somewhat less obvious, is that the power levels that are currently available (with some individual vacuum electronic device [VED] exceptions) above 100 GHz are below the power levels actually represented in Figure 3-2.
Backward wave oscillators have been a staple in the region of 100 GHz to 1,200 GHz as tunable sources for instrumentation. BWOs have been built by numerous concerns over the years but are now primarily available from the former Soviet Union. These sources have power ranging from tens of milliwatts at 100 GHz to a few milliwatts above 1 THz. While they operate at high voltages (above 2 kilovolts [kV]) and suffer from some tuning problems when using wideband cavities, they continue to be a useful tool for spectroscopy and diagnostics. The DARPA Terahertz Imaging Focal Plane Array Technology (TIFT) program is currently investing in improved designs for VEDs at 650 GHz that offer higher efficiencies and power levels that could reach 50 milliwatts. These designs are making use of micromachining techniques to improve the tolerances of the resonant structures. These structures that are being micromachined also have the potential of being used for traveling wave tubes. This development would result in potential coherent operation that should enhance resolution, both spatially and spectrally. By using
these devices with multipliers, it should be possible to realize several milliwatts of power above 1 THz.
While numerous fundamental sources, such as Gunn,2 Impact Ionization Avalanche Transit Time (IMPATT) diodes, or amplified phase-locked oscillators, can be found below 200 GHz, the most practical technique for generating coherent energy above 200 GHz is through multiplication. This is the primary technique for providing LO power to heterodyne receivers for radio astronomy or high-resolution spectroscopy. While in the past it was common to use whiskered Schottky diodes as multipliers, planar diodes have become readily available and offer suitable performance. Their output power levels can be limited by the maximum amount of input power that they can handle or by even the existence of a suitable source of RF that can be multiplied efficiently. Multipliers have been used successfully above 1 THz by chaining together a series to obtain the desired frequency. Power levels from various sources available from Virginia Diodes, a leading producer of the device to the community, is shown in Figure 3-3.
It can be seen that as frequencies increase to about 500 GHz, the power output capability drops to about 5 mW, which is generally sufficient as a source for spectroscopy that operates over short ranges or as an LO for heterodyne receivers.
Figure 3-4 summarizes the various choices for sources from Siegel3 in the region from 500 GHz to 2,500 GHz. The trend for multipliers continues downward with power levels reaching a minimum at 2.5 THz, a level that is barely sufficient to provide LO power for receivers operating at cryogenic temperatures.
Time domain systems have evolved from first developing extremely short (pico-and femtosecond) pulsed lasers that can illuminate semiconductor material that illuminates a photoconductor, such as gallium arsenide, to generate carriers. The carriers then form a current in the material, which is radiated through an antenna. The radiated energy then has a frequency content equal to the Fourier content of the envelope of the
laser pulse. The spectrum would have peak power at a frequency commensurate with the inverse of the pulse-width and would consist of spectral lines separated by repetition rate. As an example, if the envelope of the laser pulse was very short and the repetition rate was great enough to form a square wave, the roll-off with increasing frequency would be 6 decibels (dB) every other harmonic of the peak frequency, reducing the efficiencies of harmonic generation. This, of course, assumes a perfect pulse shape, which is never true. Published data4 for these types of sources typically show peak powers in the 100 GHz to 700 GHz region, with the submillimeter-wave and terahertz energy produced exhibiting this roll-off in power. The average power of the sources is not great, but by using a time-gated detector, instantaneous power is sufficient for applications where ranges are short, such as transmission through packages.
This energy can also be produced by illuminating a nonlinear material with two lasers whose difference frequency is in the region of interest. The nonlinear material then acts in a manner similar to an RF mixer and is called a photomixer. A key element of this technique is being able to couple the laser energy into the photomixer while at the same time being able to efficiently couple the RF energy out. Provided that the lasers are tunable and phase-lockable, very stable sources can be created over the operational range of the photomixer. A significant challenge remaining is to improve the efficiency of both the photomixer and the coupling structure to increase power output.
Lastly, investment continues in the development of quantum cascade lasers, which have demonstrated the generation of power in the 1 THz to 10 THz band. They are created by building tailored superlattices of semiconductors that form transition regions in the device. These devices currently operate at spot frequencies in the band, are usually pulsed, and require cooling in order to operate.
RECEIVERS
For any RF system, the maximum sensitivity will be achieved when the instantaneous bandwidth of the receiver is matched to the instantaneous bandwidth of the source. For a purely passive receiver, it is important to maximize the bandwidth of operation to obtain as much received energy from the emitting or reflecting field of observation as possible, since the source is essentially infinitely broad. For an active system, it is important to reduce the receiver bandwidth as much as possible to reduce the receiver-generated noise as much as possible while preserving the entire illumination signal. Some applications will also require that the phase of the illumination function be preserved so that advanced imaging techniques such as holography or sparse aperture array synthesis can be performed.
In the region of 30 GHz to 300 GHz, numerous types of receiver components are available commercially, so there can be systems trade-offs that include issues of cost and complexity as well as strict component performance. Typical receivers now use low-noise amplifiers as the first stage prior to either down conversion, for heterodyne systems, or diode detectors, for noncoherent receivers. The major difficulty with
4 |
The Johns Hopkins University Center for Materials Science and Detection. n.d. Terahertz (THz) Imaging and Spectroscopy. Available at http://www.wse.jhu.edu/~cmsd/Thz/. Accessed August 25, 2006. |
performing imaging in this region is that of obtaining adequate spatial resolution with a simple aperture like an optical reflector or lens. On the positive side, because of the maturity of the components, more advanced techniques can be used for imaging scenes. Computed tomographic and holographic techniques can be applied to arrays of receivers or antennas used to image objects in close proximity. These techniques are dependent on having components or architectures that can measure or control the phase as well as the amplitude of the RF energy used. The Agilent and SafeView systems (described in Chapter 4) are examples of these classes of systems.
It can be shown that for passive techniques, the only suitable receivers are detectors cooled to liquid helium temperatures, to reduce noise, or heterodyne receivers. For active systems, room-temperature detectors may also be used as long as sufficient illumination power is available. An example of system performance is presented below.
The basic receiver most used by individuals trained in RF techniques is the heterodyne receiver, which is usually formed from a Schottky diode mixer and local oscillator that is followed by an intermediate-frequency amplifier and may be preceded by a low-noise amplifier (LNA). These components may all be operated at room temperature but may be temperature-stabilized for calibration or cooled to enhance performance. State-of-the-art noise figures are about 2.5 dB at 100 GHz for an LNA. At higher frequencies (above 300 GHz) where LNAs do not yet function, noise figures of 10 to 13 dB are typical.
As Siegel points out in the paper cited above, space-qualified heterodyne receivers have been fabricated for operation up to 2.5 THz. Whether the technology can be pushed to higher frequencies remains to be seen, but performance may still ultimately be limited by the availability of sources for LOs with power levels approaching 1 mW. Of course by cooling these receivers, performance can be enhanced by about 3 to 5 dB over performance at room temperature. Given the potential benefits in detection that come from cooling, further research into the cost and complexity of installing coolers is warranted. However, it is unlikely that this technology will have any near-term applications.
Looking beyond the classic Schottky diode as mixer, one finds other forms of heterodyne detectors which, while requiring cooling, have impressive performance. These include bolometers, hot electron bolometers (HEBs), Josephson effects devices, and the superconductor-insulator-superconductor device. While these devices do require cooling down to liquid helium temperatures, they also require reduced LO drive to obtain their best sensitivities. The performance of some of these devices approaches the quantum limit of performance. Figure 3-5 shows the performance of some of these devices compared with Schottky diodes.
A great deal of recent work has concentrated on developing direct detectors that can be used to replace heterodyne receivers in some applications. This has primarily been for active systems, but there have been attempts to use them for passive systems. As was pointed out above, direct detectors have yet to achieve the sensitivity of heterodyne receivers, but they are more viable for combining into arrays to perform imaging using architectures similar to what is used in the visible and infrared region. Other than sensitivity, limitations to date for uncooled detectors have included long output time constants, which is a detriment for video frame rate imaging. The TIFT program in DARPA is trying to overcome some of these problems, with different levels of success.
Ultimately, if cooling to very cold temperatures (~10 K) can be used, direct detectors may be able to achieve characteristics sufficient for security applications using spectroscopy, imaging, or both.
Detectors can take on many forms, including those of Schottky diodes and bolometers. They are usually simple enough to be produced in arrays but must be integrated with micro antennas to efficiently couple RF energy into them. Since detectors are less sensitive than heterodyne receivers, they also must operate over a wide RF bandwidth, which makes the design of the micro antenna coupling more difficult. Some recent advances in detector technologies have been seen in the University of California at Santa Barbara (UCSB) work on unbiased metal-semimetal-semiconductor diodes that have good sensitivity and reduced 1/f noise, on zero bias backward tunnel diodes from HRL Laboratories, and on HEB devices that must be cooled to very low temperatures.
Some recent programs are also looking at other transition-edge bolometers that use the rapid transition in sensitivity of high-temperature superconductors, a nonlinear mechanism, for operation at liquid nitrogen temperatures.
SOURCES AND RECEIVERS: SUMMARY
In summary, the recent investment in component technology has been toward developing room-temperature sources and receivers that deliver performance required for the particular application. Sources and receivers that are cooled to low temperatures may exhibit performance that is reasonable for demonstrating imaging but have the potential burden of reliability and maintenance of vacuum and cooling systems.
The availability of millimeter-wave and terahertz components at a reasonable cost depends on the commercialization of the devices and systems based on those devices. While most millimeter-wave and terahertz devices over 100 GHz are handmade in small businesses and laboratories, DARPA has recently started a new program to push the state of the art of MMIC technologies beyond 300 GHz. Its Sub-Millimeter Wave Imaging Focal-Plane Technology program is intended to demonstrate an active subaperture operating at 340 GHz. Two key objectives will be the development of a coherent source capable of producing at least 50 mW from a single MMIC with a power-added efficiency of 5 percent and the development of an LNA having a noise figure of 8 dB operating at 340 GHz. If successful, these two MMIC components will not only raise the frequency of operational RF systems to 340 GHz but will also provide sources that can be used with multipliers to improve performance up into the terahertz region.
SYSTEM PERFORMANCE
Table 3-1 summarizes the performance trends of component technology as a function of increasing frequency or operating range that impact how well any millimeter-wavelength/terahertz system would be expected to perform.
While Table 3-1 only indicates trends, it does show how the phenomenological effects and component performance issues interact. In order to examine all of these effects on performance, a system model was developed to predict the performance of an imaging system designed to identify metal objects concealed underneath clothing in various weather conditions. The model is designed to predict the range at which a concealed object can be identified underneath clothing. The characteristics of the system, which were held constant, are delineated in the Table 3-2. This system is composed of a single detector and source pair that is scanned over the object being screened. It is not intended to describe any particular sensor architecture but rather to show the performance characteristics.
TABLE 3-1 Summary of Trends in Phenomenology and Component Technology
The component characteristics shown in Table 3-2 reflect the component goals of the DARPA TIFT program. The clothing material characteristics were developed by the Ohio State University and UCSB, as shown in Figure 2-4 in Chapter 2. The model was carried out under three different environmental conditions. The least challenging environment represented a winter condition at 25°F, which results in very low humidity and good atmospheric transmission. The most stringent condition represented what would be expected in the state of Maryland in August with conditions of high humidity: 95°F and 90 percent relative humidity. As the most critical component in the atmosphere is water vapor, these tests provide a good baseline for potential real-world operations.
TABLE 3-2 Parameters Used for Systems Analysis of Standoff Imaging Sensor
Model Component |
Parameter |
Transmitter and receiver antenna diameter |
0.5 m |
Transmitter power |
10 mW |
Receiver noise equivalent power |
10-12 W/(Hz) |
Receiver bandwidth |
5% |
Frame rate |
30 Hz |
One-way path loss through clothing |
1 dB per 100 GHz center frequency |
Variation in clothing reflectivity |
5% |
Signal to clutter required for a good image |
6:1 |
Size of object |
75 mm |
Resolution lines required for identification |
7 |
There are three primary features to Figure 3-6. The first is that the range to achieve the required performance is first limited by the resolution of the system, which is a function of wavelength and antenna size; as previously shown for a given aperture (see Figure 2-1 in Chapter 2), higher frequency results in better resolution. If ranges of no greater than 10 m are desired to identify objects, according to the trends stated in Table 3-1, the atmosphere does not generally limit the range; resolution does.
The second feature visible in Figure 3-6 is the “dropouts” at various ranges. These are occurring at the frequencies where the atmosphere is so absorbing that negligible energy reaches the object to be imaged. Significant increases in system power or receiver sensitivity will have little impact on these features, as the attenuation is extremely high.
The third apparent feature in the figure is that the curves stop at approximately 1 THz. This is the point at which the variation in the reflection from the clothing covering the weapon is approaching the contrast of the clothing-covered object. At this point, the apparent contrast of the concealed object is competing with the variation in the reflections from the person, reducing performance below the desired 6:1 ratio. This effect is due to the increasing losses in the clothing as a function of frequency.
It has to be pointed out that there are very limited data on materials to accurately quantify the performance of systems in this region. Analysis of performance in this millimeter-wavelength/terahertz spectral region would benefit greatly from the extensive characterization of the reflectivities and losses of materials pertinent to the problem as well as the variability with aspect angle, surface roughness of materials, and environmental conditions. It is important to note that the components defined in the model have not yet been realized but are being developed, at fixed frequencies, under the DARPA TIFT program. Thus, at this point it is not prudent to develop applications based on more than one model.
Conclusion: The technology base for millimeter-wavelength/terahertz security screening is expanding rapidly internationally, yet there is insufficient technology available to develop a system capable of identifying concealed explosives.
Recommendation: To perform an accurate assessment of the applicability of millimeter-wavelength/terahertz-based technology to explosive detection, the Transportation Security Administration will need to do the following: (1) decide on the range of materials to be detected, (2) assess the state of knowledge of what chemical structures and/or features of the scope of materials lend themselves to detection by millimeter-wavelength/terahertz-based spectroscopy, (3) assess the presence of these features in other common materials (such as clothing) within the range of uncertainty for such features, and (4) assess the contribution of additives to explosives to the millimeter-wavelength/terahertz signature.