Abstract: We report on the development of a multipixel hot electron bolometer (HEB) receiver fabricated using silicon membrane technology. The receiver comprises a 2 × 2 array of four HEB mixers, fabricated on a single chip. The HEB mixer chip is based on a superconducting NbN thin-film deposited on top of the silicon-on-insulator (SOI) substrate. The thicknesses of the device layer and handling layer of the SOI substrate are 20 and 300 μm, respectively. The thickness of the device layer is chosen such that it corresponds to a quarter-wave in silicon at 1.35 THz. The HEB mixer is integrated with a bow-tie antenna structure, in turn designed for coupling to a circular waveguide, fed by a monolithic drilled smooth-walled horn array.

Abstract: The superconducting energy gap is a crucial parameter of a superconductor when used in mixing applications. In the case of the SIS mixer, the mixing process is efficient for frequencies below the energy gap, whereas, in the case of the HEB mixer, the mixing process is most efficient at frequencies above the gap, where photon absorption takes place more readily. We have investigated the photon absorption phenomenon around the gap frequency of HEB mixers based on NbN films deposited on silicon membranes. Apart from studying the pumped I-V curves of HEB devices, we have also probed them with microwave radiation, as previously described [1]. At frequencies far below the gap frequency, the pumped I-V curves show abrupt switching between the superconducting and resistive states. For the NbN HEB mixers we tested, which have critical temperatures of ~9 K, this is true for frequencies below about 400 GHz. As the pump frequency is increased beyond 400 GHz, the resistive state extends towards zero bias and at some point a small region of negative differential resistance appears close to zero bias. In this region, the microwave probe reveals that the device impedance is changing randomly with time. As the pump frequency is further increased, this random impedance change develops into relaxation oscillations, which can be observed by the demodulation of the reflected microwave probe. Initially, these oscillations take the form of several frequencies grouped together under an envelope. As we approach the gap frequency, the multiple frequency relaxation oscillations coalesce into a single frequency of a few MHz. The resultant square-wave nature of the oscillation is a clear indication that the device is in a bi-stable state, switching between the superconducting and normal state. Above the gap frequency, it is possible to obtain a pumped I-V curve with no negative differential resistance above a threshold pumping level. Below this pumping level, the device demonstrates bi-stability, and regular relaxation oscillation at a few MHz is observed as a function of pump power. The threshold pumping level is clearly related to the amount of power absorbed by the device and its phonon cooling. From the above experiment, we can derive the gap frequency of the NbN film, which is 585 GHz for our 6 μm thin silicon membrane-based device. We also confirm that the HEB mixer is not an efficient photon absorber for radiation below the gap frequency. 1. A. Trifonov et al., “Probing the stability of HEB mixers with microwave injection”, IEEE Trans. Appl. Supercond., vol. 25, no. 3, June 2015.

Abstract: A new class of microwave-operated THz power detectors based on the NbN hot-electron-bolometer (HEB) mixer is proposed. The injected microwave signal ( 1 GHz) serves the dual purpose of pumping the HEB element and enabling the read-out of the internal state of the device. A cryogenic amplifier amplifies the reflected microwave signal from the device and a homodyne scheme recovers the effects of the incident THz radiation. Two modes of operation have been identified, depending on the level of incident radiation. For weak signals, we use a chopper to chop the incident radiation against a black body reference and a lock-in amplifier to perform synchronous detection of the homodyne readout. The voltage measured is proportional to the incident power, and we estimate an optical noise equivalent power of  5pW/ √Hz at 0.83 THz. At higher signal levels, the homodyne circuit recovers the stream of steady relaxation oscillation pulses from the HEB device. The frequency of these pulses is in the MHz frequency range and bears a linear relationship with the incident THz radiation over an input power range of  15 dB. A digital frequency counter is used to measure THz power. The applicable power range is between 1 nW and 1 μW.

Abstract: We report the results of our study of the stability of an 800 GHz hot electron bolometer (HEB) mixer cooled with a pulse-tube cryocooler. Pulse-tube cryocoolers introduce temperature fluctuations as well as mechanical vibrations at a frequency of ~1 Hz, both of which can cause receiver gain fluctuations at that frequency. In our system, the motor of the cryocooler was separated from the cryostat to minimize mechanical vibrations, leaving thermal effects as the dominant source of the receiver gain fluctuations. We measured root mean square temperature variations of the 4 K stage of ~7 mK. The HEB mixer was pumped by a solid state local oscillator at 810 GHz. The root mean square current fluctuations at the low noise operating point (1.50 mV, 56.5 μA) were ~0.12 μA, and were predominantly due to thermal fluctuations. To stabilize the bias current, microwave radiation was injected to the HEB mixer. The injected power level was set by a proportional-integral-derivative controller, which completely compensates for the bias current oscillations induced by the pulse-tube cryocooler. Significant improvement in the Allan variance of the receiver output power was obtained, and an Allan time of 5 s was measured.

Abstract: The intermediate frequency (IF) bandwidth of a hot electron bolometer (HEB) mixer is an important parameter of the mixer, in that it helps to determine its suitability for a given application. With the availability of wideband low noise amplifiers, it is simple to measure the performance of an HEB mixer over a wide range of IF at a fixed LO frequency using the standard Y-factor method. This in-situ method allows us to measure both the gain and noise bandwidths simultaneously. We have also measured mixer output impedance with a vector network analyser. Intrinsic time constant has been extracted from the impedance data and compared to the mixer's bandwidths determined from receiver Y-factor measurement.