Terahertz imaging with GaAs field-effect transistors A. Lisauskas, W. von Spiegel, S. Boubanga-Tombet, A. El Fatimy, D. Coquillat, F. Teppe, N. Dyakonova, W. Knap and H.G. Roskos Imaging at 0.6 THz is tested with a commercial GaAs highelectron-mobility transistor (HEMT) operated at room temperature. The results allow the assessment of the potential of future antenna-fitted HEMT arrays for real-time imaging.
Introduction: For the advancement of terahertz (THz) photonics and its applications, it is of prime importance to develop real-time active and passive camera systems that can operate at room temperature [1, 2]. Field-effect transistors (FETs) have been shown to be candidates for the realisation of suitable detector arrays, because they exhibit a rectifying response to THz radiation via electron-plasma effects [3, 4]. This perspective is intriguing because it may link the development of THz cameras to mainstream semiconductor technology. With this goal in mind, we have performed elementary sensitivity and imaging tests at a fixed frequency of 0.6 THz in order to derive an estimate for the signal-to-noise ratio (SNR) that one can expect for possible future cameras. As arrays of FETs, optimised for the detection of THz radiation, are not available yet, we employed commercial single-pixel GaAs high-electron-mobility transistor (HEMT), which have proven in the work described in [4, 5] to be good detectors for 600 GHz radiation. Measurement technique: We performed raster-scan imaging in reflection and transmission mode with measurement setups as shown in Fig. 1. The 0.6 THz radiation source is a continuous-wave emitter from Radiometer Physics consisting of a quartz-stabilised Gunn oscillator followed by a 6 multiplier cascade. The output power is 0.5 mW. The radiation is collimated and focused with ploymer plastic lenses. The object-under-test is mounted onto a mechanical stage and raster-scanned in two dimensions through the focus of the THz radiation. The Gaussian-shaped focal spot has a radius of 710 mm. The power of the reflected radiation is detected by a Fujitsu FHX05X GaAs HEMT with a gate length of 250 nm. Devices of the same type have been employed in earlier THz detection studies [4, 5]. The HEMT, positioned in the focus of a lens (spot radius: 700 mm), is operated at room temperature without antenna or lens attached to it. The radiation is guided to the HEMT gate in an uncontrolled way via the contact pads and the wire bonds.
Fig. 2 Transmission image of Croix placed in paper envelope Scaling of axes is in millimetres, and linear intensity is given in relative units
Fig. 3 Reflection image of key in envelope Inset: image of same object (without envelope) using Schottky detector unit, which has been optimised for detection of radiation at 0.6 THz
The transistor is operated at zero drain current and a gate voltage of Ug ¼ 20.45 V. The source-to-drain voltage Usd induced by the THz radiation is measured with the help of a lock-in amplifier (chopping frequency: 980 Hz) with an integration time of 10 ms. Images are generated from single amplifier readings per pixel without further averaging [6]. An image consisting of 190 160 pixels is taken in 7 min. Responsivity and NEP: We determine the HEMT’s effective responsivity RV (uncorrected for illuminated area [3]) and noise equivalent power (NEP). For an incident power P of 83 mW, the values of Usd are 13.9 mV (Ug ¼ 20.35 V) and 26.7 mV (Ug ¼ 20.45 V). As Ug approaches the threshold, the fluctuations increase. For the values of Ug as given p above, the noise voltage UN is found to be 4.8 and 12.0 nV/ Hz, leading to values of RV ¼ Usd/P of 0.167 and 0.322 V/W, p respectively, and corresponding NEP values of 28.7 and 37.3 nW/ Hz. With these values, one expects to obtain an SNR of 536 for Ug ¼ 20.45 V and for a radiation power at the HEMT detector of 83 mW.
Fig. 1 Transmission and reflection setups for THz imaging tests with FET receivers Beams propagating to top and to right side of 55/45 beamsplitter in reflectionmode geometry are stopped by absorbers a Transmission b Reflection
ELECTRONICS LETTERS 13th March 2008 Vol. 44
SNR of imaging: Fig. 2 shows a THz transmission image of a Croix (ornamental cross) hidden in a paper envelope. The interference pattern seen next to the Croix is caused by multiple reflections between the envelope sheets. The ratio between maximal signal and noise amplitude is 400:1. The SNR is 25% less than estimated, which is probably caused by a somewhat lower beam power. Fig. 3 shows a reflection image of a key in an envelope. The SNR is 40:1. The smaller SNR in reflection geometry as compared to the transmission mode is a consequence of (a) the power loss at the beamsplitter and (b) scattering at the object. A distinctive feature seen in the reflection image is the periodic intensity modulation. It originates from interference in the coupled resonators formed by the emitter, the object and the detector. A large part of the
No. 6
power incident on the HEMT is reflected from its contact pads. The inset of Fig. 3 shows another image of the key, now without an envelope and with the image taken with a Schottky detector equipped with a horn antenna (assembly from Radiometer Physics). Its voltage responsivity is 529 V/W (bias: 22 V). The SNR in the image is 1160:1 and hence about 30 larger than with the HEMT. The modulation on the image is much weaker because of the strong absorption by the Schottky detector unit. With the modulation depth M in the two images (ratio M ¼ (Imax 2 Imin)/(Imax þ Imin) of the maximal and minimal beam intensities) of M ¼ 0.7 for the HEMT and M ¼ 0.17 for the Schottky receiver, we can estimate that the absorption strength of the HEMT detector can be raised by one or two orders of magnitude by proper coupling of the radiation with an antenna. The SNR would then already have become comparable with that of the Schottky detector. Potential for realisation of real-time THz camera: In one of the few papers in the literature that quantify RpV and NEP of FET-based THz detectors, a NEP value of 0.1 nW/ Hz has been determined for a silicon MOSFET operated at 0.7 THz [3]. RV has been estimated to reach 200 V/W if the device were properly antenna-coupled [3]. These values would provide an excellent basis for the realisation of a THz camera that could compete in performance with microbolometer array cameras working pin real-time at higher frequencies (reported NEP value of 320 pW/ Hz at 4 THz [1]). A camera with 2.4 104 pixels would achieve a dynamic range of 10 when illuminated with 0.1 mW of radiation power at 0.7 THz. With the presently studied GaAs HEMTs, if unmodified, one would achieve the same dynamic range at 0.6 GHz, but with only a few tens of pixels. With a proper antenna, the number of pixels could be increased by one or two orders of magnitude. At least another order of magnitude is possible by an increase of the responsivity (RV can be raised by a factor of several tens to 100 by application of a source– drain bias voltage at the expense of increased noise [4, 5]), and by noise reduction via proper amplifier/filter integration and screening of the device. Conclusions: We have measured the NEP and responsivity of a commercial GaAs HEMT, which is used as a detector of 0.6 THz radiation, and we have demonstrated imaging with this device. The data suggests that realisation of a THz camera for real-time operation is feasible. If the radiation coupling and the electronic noise suppression are optimised, such a camera could have 103 – 104 pixels and provide a dynamic range of 10 at 0.1 mW incident power. Future work has to show whether III – V FETs or Si-based FETs with their superior noise performance [7] are better suited as THz detectors in cameras.
Acknowledgments: The work was supported by GDR-E project ‘semiconductor sources and detectors of THz frequencies’. The research at Montpellier 2 University was supported by CNRS and the Region of Languedoc-Roussillon through the ‘Terahertz Platform’ project. # The Institution of Engineering and Technology 2008 18 January 2008 Electronics Letters online no: 20080172 doi: 10.1049/el:20080172 A. Lisauskas, W. von Spiegel and H.G. Roskos (Physikalisches Institut, Max-von-Laue-Str. 1, Johann Wolfgang Goethe-Universita¨t, Frankfurt am Main D-60438, Germany) E-mail:
[email protected] S. Boubanga-Tombet, A. El Fatimy, D. Coquillat, F. Teppe, N. Dyakonova and W. Knap (GES, UMR 5650 CNRS-Universite´ Montpellier 2, Montpellier 34950, France) References 1 Lee, A.W.M., Qin, Q., Kumar, S., Williams, B.S., Hu, Q., and Reno, J.L.: ‘Real-time imaging using a 4.3-THz quantum cascade laser and a 320 240 microbolometer focal-plane array’, Appl. Phys. Lett., 2006, 89, p. 141125 2 Lo¨ffler, T., May, T., am Weg, C., Alcin, A., Hils, B., and Roskos, H.G.: ‘Continuous-wave terahertz imaging with a hybrid system’, Appl. Phys. Lett., 2007, 90, p. 091111 3 Tauk, R., Teppe, F., Boubanga, S., Coquillat, D., Knap, W., Meziani, Y.M., Gallon, C., Boeuf, F., Skotnicki, T., Fenouillet-Beranger, C., Maude, D.K., Rumyantsev, S., and Shur, M.S.: ‘Plasma wave detection of terahertz radiation by silicon field effects transistors: Responsivity and noise equivalent power’, Appl. Phys. Lett., 2006, 89, p. 253511 4 Teppe, F., Knap, W., Veksler, D., Shur, M.S., Dmitriev, A.P., Kachorovskii, V.Yu., and Rumyantsev, S.: ‘Room-temperature plasma waves resonant detection of sub-terahertz radiation by nanometer fieldeffect transistor’, Appl. Phys. Lett., 2005, 87, p. 052107 5 Teppe, F., Veksler, D., Kachorovskii, V.Yu., Dmitriev, A.P., Xie, X., Zhang, X.-C., Rumyantsev, S., Knap, W., and Shur, M.S.: ‘Plasma wave resonant detection of femtosecond pulsed terahertz radiation by a nanometer field-effect transistor’, Appl. Phys. Lett., 2005, 87, p. 022102 6 Siebert, K.J., Quast, H., Leonhardt, R., Lo¨ffler, T., Thomson, M., Bauer, T., Roskos, H.G., and Czasch, S.: ‘Continuous-wave all-optoelectronic terahertz imaging’, Appl. Phys. Lett., 2002, 80, pp. 3003– 3006 7 Levinshtein, M.E., Rumyansev, S.L., Tauk, R., Boubanga, S., Dyakonova, N., Knap, W., Shchepetov, A., Bollaert, S., Rollens, Y., and Shur, M.S.: ‘Low frequency noise in InAlAs/InGaAs modulation doped field effect transistors with 50-nm gate length’, J. Appl. Phys., 2007, 102, p. 064506
ELECTRONICS LETTERS 13th March 2008 Vol. 44 No. 6
