⇦Chinese Physics Letters, 2017, Vol. 34, No. 9, Article code 090701 Terahertz Direct Detectors Based on Superconducting Hot Electron Bolometers with Microwave Biasing * Shou-Lu Jiang(姜寿禄)1, Xian-Feng Li(李先峰)1, Run-Feng Su(苏润丰)1, Xiao-Qing Jia(贾小氢)1,2, Xue-Cou Tu(涂学凑)1, Lin Kang(康琳)1,2, Biao-Bing Jin(金飚兵)1, Wei-Wei Xu(许伟伟)1,2, Jian Chen(陈健)1**, Pei-Heng Wu(吴培亨)1,2 Affiliations 1Research Institute of Superconductor Electronics (RISE), School of Electronic Science and Engineering, Nanjing University, Nanjing 210023 2Synergetic Innovation Center in Quantum Information and Quantum Physics, University of Science and Technology of China, Hefei 230026 Received 12 June 2017 *Supported by the National Basic Research Program of China under Grant No 2014CB339800, the National Natural Science Foundation of China under Grant Nos 61521001, 11173015 and 11227904, the Fundamental Research Funds for the Central Universities, and the Key Laboratory of Advanced Techniques for Manipulating Electromagnetic Waves of Jiangsu Province.
**Corresponding author. Email: chenj63@nju.edu.cn Citation Text: Jiang S L, Li X F, Su R F, Jia X Q and Tu X C et al 2017 Chin. Phys. Lett. 34 090701 Abstract Terahertz (THz) direct detectors based on superconducting niobium nitride (NbN) hot electron bolometers (HEBs) with microwave (MW) biasing are studied. The MW is used to bias the HEB to the optimum point and to readout the impedance changes caused by the incident THz signals. Compared with the thermal biasing method, this method would be more promising in large scale array with simple readout. The used NbN HEB has an excellent performance as heterodyne detector with the double sideband noise temperature ($T_{\rm N}$) of 403 K working at 4.2 K and 0.65 THz. As a result, the noise equivalent power of 1.5 pW/Hz$^{1/2 }$ and the response time of 64 ps are obtained for the direct detectors based on the NbN HEBs and working at 4.2 K and 0.65 THz. DOI:10.1088/0256-307X/34/9/090701 PACS:07.57.Kp, 85.25.Pb, 85.25.Pb, 95.85.Fm © 2017 Chinese Physics Society Article Text Superconducting niobium nitride (NbN) hot electron bolometers (HEBs) are commonly used as heterodyne detectors in astronomy[1-3] with low noise temperature[4,5] (about a few of the quantum limitations). In addition to heterodyne detectors, with large intermediate frequency (IF) gain bandwidth (GBW)[6] and high sensitivity, it would be promising as direct detectors for terahertz (THz) imaging in the security inspection field. Thermal and microwave (MW) biasing methods[7] used in HEB direct detectors have been studied before. Both methods selected bias current to monitor the HEB impedance changes corresponding to the incident THz signals. A constant voltage source is applied to give a negative electrothermal feedback to the HEB to stabilize the working state.[8] When a single HEB direct detector is used in the application, it would be fine. However, with real time imaging demands in public security inspection, a single pixel THz direct detector would not satisfy the requirements for the practical applications. Thus multi pixel direct detectors (arrays) would be necessary for the THz imaging. The frequency domain multiplexing method is an effective idea to read out all the pixels simultaneously with single LNA.[9] Every detector is designed as an MW resonator with different resonant frequencies, thus the whole array can be read out with a comb of frequencies. There is also an off-the-shelf MW circuit designed to read out thousands of low temperature detectors. Thus the MW readout method would be promising used in the array. To bridge the gap between single HEB direct detector and the HEB array, we adopt the MW biasing method to investigate the HEB direct detector's performance in this study. The HEB chip is made up of a logarithmic spiral antenna with frequency-independent impedance and a superconducting bridge made from an ultra-thin NbN film. The critical temperature $T_{\rm c}$ of the chip is about 8.4 K. The critical current $I_{\rm c}$ is typically 205 μA and the normal state resistance $R_{\rm N}$ is around 155 $\Omega$. The chip is glued to the back side of a silicon hyper-hemispherical lens with a diameter of 10 mm. The Si lens is coated with a 0.65 THz anti-reflection (AR) coating for reducing the optical loss of the incident signal. An oxygen-free copper holder used to hold the Si lens is installed on the cold plate of the dewar with the bath temperature of 4.2 K. The dewar window is made of mylar film which has a good transmittance for THz radiation with the thickness of 36 μm. A G-110 Zitex, which is a porous polytetrafluoroethylene (PTFE) film, is installed in the input hole of the dewar as the infrared filter. A band pass mesh-filter made by Virginia Diodes Inc (VDI) and with central frequency of around 0.65 THz is used to define the input bandwidth of the detector. In this work, the experimental setup as shown in Fig. 1 is similar to the MW stabilization scheme setup.[10] A 20-dB attenuator located between the MW source and the bias tee is used for avoiding the 300 K ambient background noise. We selected the constant voltage mode of the dc bias. The injected MW guided by the circulator is fed into the HEB chip without being picked up by the LNA directly. The reflected weak MW is amplified by the LNA and then demodulated by an MW square-law power detector. Finally, the demodulated signal is fed into the Agilent 35670 dynamic signal analyzer in the noise equivalent power (NEP) measurement. When used as a direct detector in the Fourier transform spectrometer (FTS) or in the THz imaging system, the dynamic signal analyzer can be replaced by the lock-in amplifier.
cpl-34-9-090701-fig1.png
Fig. 1. Schematic diagram of the experimental setup for the HEB direct detectors.
To evaluate the HEB chip properties and the readout circuit working conditions, we first measured the double sideband (DSB) noise temperature $T_{\rm N}$ of the HEB mixers at the local oscillator (LO) frequency of 0.65 THz and intermediate frequency (IF) of 1.5 GHz. An improved $Y$-factor[11] method is used to measure $T_{\rm N}$ of the HEB mixers. The IF output powers, $P_{\rm hot}$ and $P_{\rm cold}$, corresponding to the hot and cold loads, were measured at the same bias point to obtain $Y= P_{\rm hot}/P_{\rm cold}$. The uncorrected $T_{\rm N}$ can be calculated by $$ T_{\rm N} =\frac{T_{\rm hot} -YT_{\rm cold} }{Y-1}. $$ As shown in Fig. 2, at the optimum bias voltage of 1.25 mV and the bias current of 47 μA, the best $T_{\rm N}$ is 403 K (uncorrected) which is 13 times that of quantum limitation of $hv/k_{\rm B}$. The good $T_{\rm N}$ demonstrates that the HEB chip is very sensitive to the THz signal and the readout circuit working well. Because the optical path of the measurement setup for the direct detection is similar to that of the noise temperature, the good $T_{\rm N}$ also means that the optical path of the dewar works well at the desired frequency. We expected that an excellent NEP can be achieved for the HEB direct detector with MW biasing. The IF GBW was also measured at the optimum bias point with two VDI THz sources of similar frequency around 0.65 THz. The IF GBW of 2.5 GHz is obtained, meaning that the HEB chip response time is about 64 ps.
cpl-34-9-090701-fig2.png
Fig. 2. The $I$–$V$ curves with and without optimized LO power. The inset is the DSB noise temperature of the HEB as a function of bias current. The results are measured at the LO frequency of 0.65 THz.
cpl-34-9-090701-fig3.png
Fig. 3. The $I$–$V$ curves of the HEB without and with MW biasing.
As described in our previous work,[12] when the biased MW frequency is lower than the IF GBW, the HEB direct detectors work at a stable state and the temperature coefficient of resistance (TCR) can be enhanced compared with the thermal biasing method. Since the IF GBW of the HEB chip is 2.5 GHz and the IF LNA's central frequency is 1.5 GHz used in the heterodyne measurement, in this study we chose the biased MW frequency of 1.5 GHz. To guarantee the biasing MW frequency is appropriate, we also recorded the $I$–$V$ curves with different powers of 1.5 GHz MW as shown in Fig. 3. The $I$–$V$ curves show no hysteresis with proper MW power and the working state is stable, which is similar to the $I$–$V$ curves with thermal biasing. Therefore, the HEB chip with 1.5 GHz MW biasing can achieve similar stability compared with the thermal biasing. NEP is a critical parameter to characterize the direct detector's sensitivity. It is defined as the signal power that gives a signal-to-noise ratio (SNR) of one in one hertz output bandwidth. Thus to obtain the NEP of the direct detectors, we first need to obtain the incident THz signal's power. A black body with temperature of 1073 K is set in front of the dewar window. The ambient temperature is 300 K. The VDI band pass mesh-filter's central frequency is 0.623 THz and the bandwidth is 75 GHz measured by the THz time domain spectrometer (TDS) as shown in Fig. 5. By using Plank's blackbody radiation law we obtained the incident THz power of 0.8 nW. An optical chopper is located between the blackbody source and the dewar window to modulate the incident THz signal. We adjusted the chopping frequency of 1.376 kHz and then obtained the square-law power detector's output signal's spectra as shown in Fig. 4. The resolution of this spectra we set is 16 Hz. The peak value of 1.376 kHz is a constant with the spectral resolution changes and the noise voltage with spectral resolution of 16 Hz is 4 times lower than the noise voltage with spectral resolution of 1 Hz, which is determined by the spectral processing algorithm of the dynamic signal analyzer. The higher ratio between the peak value and the noise voltage beside the peak, the better NEP will be. To obtain the best NEP, we scanned all the $I$–$V$ region with different MW powers and finally found an optimum bias point. The $I$–$V$ curve that passes through the optimum bias point is also recorded as shown in Fig. 3.
cpl-34-9-090701-fig4.png
Fig. 4. Spectra of the demodulated reflected MW signals. The black and red curves represent the spectrum measured with and without the modulated signal source, respectively. The spectral resolution is 16 Hz and the modulation frequency is 1.376 kHz.
The peak signal value is 308 μV and the noise voltage beside the peak signal is 2.25 μV with the spectral resolution of 16 Hz at optimum bias point. Thus we calculated the noise voltage beside the peak signal of 0.56 μV per Hz. The response voltage caused by the THz signal is 550 times the baseline when the spectral resolution is 1 Hz. Since the input THz signal power is 0.8 nW, the NEP is 550 times lower than 0.8 nW. Thus the NEP of the HEB direct detector is 1.5 pW/Hz$^{1/2}$. We also found that when the bias voltage is zero, a signal peak is exhibited in the spectra with proper power MW biasing. This means that MW would be enough to bias the HEB and readout the impedance changes of the HEB. The NEP with zero bias is 18 pW/Hz$^{1/2}$, which is about an order higher than the NEP obtained at the optimum bias point. Although the zero-bias HEB direct detector's sensitivity is worse than the constant-voltage bias one, it would be conveniently applied in some applications with lower sensitivity requirements but high response speed without dc bias.
cpl-34-9-090701-fig5.png
Fig. 5. Transmittance of the VDI band pass mesh filter. The blue line is measured by the THz-TDS. The red line is measured by the FTS with the NbN HEB direct detector.
After characterizing the performance of the HEB direct detectors, we used the detector combined with the FTS constructed by our lab[13] to measure the transmittance of the VDI mesh filter. The chopping frequency is set at 1.4 kHz, which is limited by the optical chopper and the integral time is set at 30 ms. The sample points and sample speed were 1000 and 1 point per 30 ms. Thus the whole scanning time is about 30 s. The measured transmission spectra by the FTS coincide well with the spectra measured by the THz-TDS. The scanning speed is limited by the chopping frequency of the mechanical chopper at present. With substituting the mechanical chopper with an electrical modulator based on superconducting metamaterial,[14] the scanning speed would be increased greatly in the following experiments. In summary, we have introduced a new simple biasing method of operating the HEBs as direct detectors. The 1.5 GHz MW combined with the dc bias are used to bias the HEBs to the optimum bias point. The optical NEP of the HEB direct detector is 1.5 pW/Hz$^{1/2}$. The response time is 64 ps obtained by measuring the IF GBW. This MW readout method can be easily applied in the HEB array with frequency division multiplexing digital readouts. With simple readout, high speed response time and sensitive response, the HEB direct detectors with the MW biasing scheme would be suitable in the THz real-time imaging field. References A 1-THz Superconducting Hot-Electron-Bolometer Receiver for Astronomical ObservationsA 1.3-THz Balanced Waveguide HEB Mixer for the APEX TelescopeTerahertz hot electron bolometer waveguide mixers for GREATLow Noise Receivers Based on Superconducting Niobium Nitride Hot Electron Bolometer Mixers from 0.65 to 3.1 Terahertz4.7-THz Superconducting Hot Electron Bolometer Waveguide MixerNbN hot electron bolometric mixers for terahertz receiversA superconducting bolometer with strong electrothermal feedbackDigital readouts for large microwave low-temperature detector arraysStability of Superconducting Hot Electron Bolometer ReceiversTerahertz detectors based on superconducting hot electron bolometersAppropriate microwave frequency selection for biasing superconducting hot electron bolometers as terahertz direct detectorsA simple Fourier transform spectrometer for terahertz applicationsElectrically tunable superconducting terahertz metamaterial with low insertion loss and high switchable ratios
[1] Meledin D V, Marrone D P, Tong C Y E et al 2004 IEEE Trans. Microwave Theory Tech. 52 2338
[2] Meledin D, Pavolotsky A, Desmaris V et al 2009 IEEE Trans. Microwave Theory Tech. 57 89
[3] Putz P, Honingh C E, Jacobs K et al 2012 Astron. Astrophys. 542 L2
[4] Liang M, Chen J, Kang L et al 2010 IEICE Trans. Electron. E93-C 473
[5] Buchel D, Putz P, Jacobs K et al 2015 IEEE Trans. Terahertz Sci. Technol. 5 207
[6] Kroug M, Cherednichenko S, Merkel H et al 2001 IEEE Trans. Appl. Supercond. 11 962
[7]Lobanov Y, Tong C E, Hedden A et al 2010 Proc. 21st Int. Symp. Space THz Tech. pp 420–423
[8] Lee A T, Richards P L, Nam S W et al 1996 Appl. Phys. Lett. 69 1801
[9] Mazin B A, Day P K, Irwin K D et al 2006 Nucl. Instrum. Methods Phys. Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip. 559 799
[10] Chen J, Jiang Y, Liang M et al 2011 IEEE Trans. Appl. Supercond. 21 667
[11] Jiang Y, Jin B B, Xu W W et al 2012 Sci. Chin. Inf. Sci. 55 64
[12] Jiang S L, Li X F, Jia X Q et al 2017 Supercond. Sci. Technol. 30 044004
[13] Jiang Y, Liang M, Jin B B et al 2012 Chin. Sci. Bull. 57 573
[14] Li C, Zhang C H, Hu G L et al 2016 Appl. Phys. Lett. 109 022601