Room-temperature detectors for millimeter waves
Scientists and engineers wanting to build and use instrumentation for the millimeter to terahertz spectral region are hampered by the lack of detectors. For these frequencies, there is no equivalent of the silicon photodiode, a photodetector that operates at room temperature and offers sensitive coverage of the entire visible spectral region with the possibility of resolving rapid fluctuations in the light intensity. A detector is required in certain applications, such as when detecting the coherent long-wavelength radiation emitted by packets of accelerated relativistic electrons in a synchrotron or other light source. In this example, the emitted light can be used for spectroscopy or analyzed to obtain information about the emitting particles.
The challenge of detecting far-infrared light is much greater than for the visible part of the spectrum. The frequency is too high for the conventional circuit-board radio techniques used for mobile telephones. In addition, photon energies are much lower than thermal energies at room temperature, and the highest and lowest frequencies differ by a factor of ten, compared with the factor of two in the visible spectrum. There are several approaches to detection, and one that offers a reasonable combination of sensitivity, speed, and spectral response is based on rectifying semiconductor diodes. These micron-sized Schottky devices convert the rapidly fluctuating signal from an antenna into a steady signal current, proportional to the signal power, which can be amplified and processed. Diode detectors have been available for decades in different configurations, and suppliers include Farran Technology Limited in Ireland, Virginia Diode Inc. in the United States, and Advanced Compound Semiconductor Technologies in Germany.
Our international and multidisciplinary team recently started producing fast, millimeter-wave detectors that work at room temperatures. Aspects we have looked into include device modeling as well as single-pixel and linear-array production. While our devices are not yet at the state of the art, we have demonstrated a wide post-detection bandwidth and a small linear array. We made our detectors from conventional forward-biased Schottky diodes and the more recently developed indium gallium arsenide zero-bias diodes. (The latter should exhibit lower noise because of the absence of a bias current.) We connected both types of diodes to planar broadband antennas and interfaced them with low-noise preamplifiers.
In our work, we employed the harmonic-balance method together with a physics-based drift-diffusion numerical device model.1 Our simulator incorporates accurate boundary and interface conditions for high-forward as well as reverse bias, including impact ionization, self-consistent incorporation of tunneling, and image-force effects. The inputs to the model include the detector device dimensions, the semiconductor layer structure, and the doping profile. This simulator allows prediction of the device properties including current–voltage characteristic, the ideality factor, responsivity, and series resistance. The devices themselves are physically similar and manufactured in a similar way to the air-bridged gallium arsenide Schottky diodes that are used for heterodyne mixers and frequency multipliers. Figure 1 shows an electron micrograph of manufactured devices on a wafer before separation into roughly 300μm×100μm area chips using a dicing saw.
After dicing, the chips were electrically connected to planar metal bow-tie antennas photolithographically defined on thin-quartz substrates. We designed and fabricated single antennas as well as a linear array of eight elements. We used a hyperhemispherical lens, made of high-resistivity silicon, to focus incident radiation onto the antenna. The ends of the antenna were connected to low-noise operational-amplifier-based preamplifiers and, in the case of Schottky devices, a constant-current bias circuit. We derived test signals for spectral response measurements from the transmit side of a vector network analyzer, which is tunable from 70GHz to over 700GHz. In order to measure the detector signal and noise level, we used a mechanical chopper and lock-in amplifier. Our best detectors demonstrated a responsivity of the order of 100V/W from 250 to 400GHz and noise equivalent powers of 0.2nW/Hz1/2.
Measuring response speeds faster than a few milliseconds, however, requires a different approach. We used photomixing2 of optical noise from a fiber amplifier at telecommunications wavelengths to generate a low-power millimeter-wave continuum. A gigahertz bandwidth electro-optic analog amplitude modulator, driven by a pulse generator, sent pulses of light of effectively arbitrary duration to the photomixer. The bottom red and blue traces in Figure 2 represent the optical signals after the modulator in response to 10 and 20ns drive pulses, respectively. The top traces show the corresponding zero-bias-detector outputs, with a measured decay time of about 10ns.
In our work, we demonstrated broadband room-temperature detectors for millimeter and sub-millimeter waves with a responsivity of about 100V/W and a response speed from DC to about 50MHz. We used a novel optical system to measure directly the speed of response to short pulses. Our future work will focus on improving the match between model and measurement, increasing responsivity, and raising the post-detection bandwidth to several hundred megahertz.
We gratefully acknowledge financial support from the UK's Science and Technology Facilities Council (STFC) Centre for Instrumentation.
Peter Huggard is a member of the Millimetre Wave Technology Group at RAL.
