TCAD model for TeraFET detectors operating in a large dynamic range
11 TCAD model for TeraFET detectors operating in alarge dynamic range
Xueqing Liu and Michael S. Shur,
Life Fellow, IEEE
Abstract —We present technology computer-aided design(TCAD) models for AlGaAs/InGaAs and AlGaN/GaN and siliconTeraFETs, plasmonic field effect transistors (FETs), for terahertz(THz) detection validated over a wide dynamic range. Themodeling results are in good agreement with the experimentaldata for the AlGaAs/InGaAs heterostructure FETs (HFETs)and, to the low end of the dynamic range, with the analyticaltheory of the TeraFET detectors. The models incorporate theresponse saturation effect at high intensities of the THz radiationobserved in experiments and reveal the physics of the responsesaturation associated with different mechanisms for differentmaterial systems. These mechanisms include the gate leakage,the velocity saturation and the avalanche effect.
Index Terms —TeraFET, terahertz detection, modeling, TCAD,HFET, MOSFET
I. I
NTRODUCTION T HE TeraFETs, plasmonic field effect transistors (FETs)applied in the terahertz (THz) frequency range, have beenproposed since the early 1990s [1], [2], [3]. Over decadesthey have found wide applications in THz mixers [4], [5],frequency multipliers [6], [7], transceivers [8], [9], imagers andsensors [10], [11], [12], etc. Recent interests include operatingTeraFET detectors at high incident power, which could beused to measure the duration and structure of the high-powerTHz pulses [13]. At large intensities of the incident THzradiation, the response has been observed to saturate in themeasurements [14]. Different propositions have been made toexplain this effect [15], [16], [17]. In this work we explore thereason for the response saturation effect by examining differentphysical mechanisms in the physics-based simulations usingthe validated TCAD models [18]. The results reveal thephysics of the response saturation associated with differentmechanisms for different material systems. These mechanismsinclude the leakage current, the velocity saturation, and theavalanche effect. This insight into the device physics allowsfor the development of the next generation compact modelsfor the TeraFET detectors that are valid over a wide dynamicoperation range. II. TH Z TCAD M
ODELS
Fig. 1 shows the structures of the TeraFETs for differentmaterial systems in Synopsys Sentaurus TCAD. The TCADmodels account for the hydrodynamic transport suitable fordeep-submicron and heterostructure devices, and the velocity (Corresponding author: Michael S. Shur.)Xueqing Liu and Michael S. Shur are with the Department of Electrical,Computer and Systems Engineering, Rensselaer Polytechnic Institute, 110 8thStreet, Troy, NY 12180, USA (e-mail: [email protected], [email protected]). (a)(b)(c)
GaAs substrateun-doped In Ga As channel un-doped Al Ga As Gate Schottky contactSi N passivationn + -GaAs cap n + -Al Ga AsSource Ohmic contact
Drain Ohmic contactGaN substrateun-doped GaN channel un-doped Al Ga N Gate Schottky contactSi N passivationn + -GaN cap n + -Al Ga NSource Ohmic contact
Drain Ohmic contact
Poly-SiSi N SiO SiSource contact Drain contactGate contactSi substrate SiO Fig. 1. Schematic of the TeraFET structures in TCAD: (a) AlGaAs/InGaAsHFET, (b) AlGaN/GaN HFET, and (c) SOI MOSFET. saturation, the generation-recombination, and the barrier tun-neling mechanisms [19].
A. AlGaAs/InGaAs HFETs
The AlGaAs/InGaAs HFET TCAD model was built tosimulate the 130 nm AlGaAs/InGaAs pHEMTs fabricated byTriQuint Inc. (now Qorvo) [20]. It has been validated by com-paring the simulated I-V characteristics and the dependence ofthe THz response on the gate bias with the measured data andthe analytical results [18].Fig. 2 shows the dependence of the simulated detectorresponse at 0.3 THz above threshold ( V gt = V a for the AlGaAs/InGaAs HFETTCAD model, compared with the measured data in [14] andthe analytical results with the open boundary condition atthe drain. To explore the device physics for the response a r X i v : . [ phy s i c s . a pp - ph ] A ug m e a s u r e d , 0 . 6 T H z m e a s u r e d , 1 . 0 7 T H z a n a l y t i c a l , 0 . 3 T H z T C A D , H d , w / o A v a , w i t h V s a t , w i t h B T T C A D , H d , w / o A v a , w i t h V s a t , w / o B T T C A D , H d , w / o A v a , w / o V s a t , w / o B T T C A D , D D , w / o A v a , w / o V s a t , w / o B T s l o p e ~ V a s l o p e ~ V Response (V)
T H z s i g n a l m a g n i t u d e V a ( V ) Fig. 2. Comparison of analytical and simulated drain response at 0.3 THzabove threshold with the measured data in [14] as a function of the THz signalmagnitude for the AlGaAs/InGaAs HFET. The analytical and measured dataare normalized to the range of the simulated results. (a)(b) E l ec t r on c u rr e n t d e n s it y - Y ( A / c m ) ungated region DD, V a =3V ungated region position (m) t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f Hd, V a =3V E l ec t r on c u rr e n t d e n s it y - Y ( A / c m ) position (m) t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f ungated regionungated region Fig. 3. Profiles of the electron current density below the gate contact withina period for the AlGaAs/InGaAs HFET TCAD model at the high intensitylevel ( V a = saturation, different physical mechanisms including the hy-drodynamic (Hd) or drift-diffusion (DD) transport, avalanche (Ava), velocity saturation (Vsat), and gate barrier tunneling(BT) were turned on and off in the simulation. It could beseen that the simulated response gives a quadratic response(proportional to V a ) at low intensities, which is consistentwith the analytical result from around 5 mV to 1 V. At highintensities ( V a > V a > V a = B. AlGaN/GaN HFETs
The TCAD model developed for the AlGaN/GaN HFETuses the same dimensions with the AlGaAs/InGaAs HFETTCAD model, as shown in Fig. 1 (b). Fig. 4 shows the m e a s u r e d , 0 . 6 T H z a n a l y t i c a l , 0 . 3 T H z T C A D , H d , w / o A v a , w i t h V s a t T C A D , H d , w / o A v a , w / o V s a t T C A D , H d , w i t h A v a , w / o V s a t T C A D , D D , w / o A v a , w / o V s a t s l o p e ~ V a s l o p e ~ V T H z s i g n a l m a g n i t u d e V a ( V ) Response (V)
Fig. 4. Comparison of analytical and simulated drain response at 0.3 THzabove threshold with the measured data in [14] as a function of the THz signalmagnitude for the AlGaN/GaN HFET. The analytical and measured data arenormalized to the range of the simulated results. (a)(b) E l ec t r on c u rr e n t d e n s it y - Y ( A / c m ) DD, V a =3V ungated region ungated region t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f position (m) E l ec t r on c u rr e n t d e n s it y - Y ( A / c m ) Hd, V a =3V ungated region ungated region t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f position (m) Fig. 5. Profiles of the electron current density below the gate contact withina period for the AlGaN/GaN HFET TCAD model at the high intensity level( V a = simulated detector response at 0.3 THz above threshold ( V gt = V a forthe AlGaN/GaN HFET TCAD model with different physicalmechanisms, compared with the measured data in [14] andthe analytical results with the open boundary condition atthe drain. Again, from the profiles of the electron currentdensity transverse to the channel at the positions closely(1 nm) below the Schottky gate contact in Fig. 5, the responsesaturation could be linked to the higher gate leakage currentobserved in the hydrodynamic transport than in the drift-diffusion transport. C. Si MOSFETs
The response saturation at high intensities of the THzradiation was also observed for Si MOSFETs [14]. Since thegate leakage could be negligible in Si MOSFETs due to thegate insulator, other mechanisms could be responsible for theresponse saturation. To investigate this, the TCAD model inFig. 1 (c) is considered. The model is set up based on an ex-emplary silicon-on-insulator (SOI) N-channel MOSFET usingthe default material parameter files for silicon in SentaurusTCAD [19]. The gate length for the model is set as 130 nm,the same as for the HFET models. The response saturation m e a s u r e d , 0 . 6 T H z a n a l y t i c a l , 0 . 3 T H z T C A D , H d , w i t h A v a , w i t h V s a t T C A D , H d , w / o A v a , w i t h V s a t T C A D , H d , w i t h A v a , w / o V s a t T C A D , H d , w / o A v a , w / o V s a t s l o p e ~ V a s l o p e ~ V Response (V)
T H z s i g n a l m a g n i t u d e V a ( V ) Fig. 6. Comparison of analytical and simulated drain response at 0.3 THzabove threshold with the measured data in [14] as a function of the THzsignal magnitude for the SOI MOSFET. The analytical and measured dataare normalized to the range of the simulated results. (a)(b) -2E-7 -1E-7 0 1E-7 2E-7-2E6-1E601E62E6 static t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f w/o Ava, V a =6V E l ec t r i c f i e l d - Y ( V / c m ) position (m) gated region -2E-7 -1E-7 0 1E-7 2E-7-2E6-1E601E62E6 E l ec t r i c f i e l d - Y ( V / c m ) with Ava, V a =6V static t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f position (m) gated region Fig. 7. Profiles of the electric field along the channel within a period for theSOI MOSFET TCAD model at the high intensity level ( V a = could also be demonstrated at large intensities with the SOIMOSFET TCAD model, as shown in Fig. 6. The comparisonof the simulated results with different mechanisms shows that (a)(b) -2E-7 -1E-7 0 1E-7 2E-7-6E6-4E6-2E602E64E66E6 D i s p l ace m e n t c u rr e n t d e n s it y - Y ( A / c m ) w/o Ava, V a =6V t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f gated region position (m) -2E-7 -1E-7 0 1E-7 2E-7-6E6-4E6-2E602E64E66E6 with Ava, V a =6V D i s p l ace m e n t c u rr e n t d e n s it y - Y ( A / c m ) position (m) t=0/20f t=1/20f t=2/20f t=3/20f t=4/20f t=5/20f t=6/20f t=7/20f t=8/20f t=9/20f t=10/20f t=11/20f t=12/20f t=13/20f t=14/20f t=15/20f t=16/20f t=17/20f t=18/20f t=19/20f t=20/20f gated region Fig. 8. Profiles of the displacement current density below the top surface ofthe gate oxide within a period for the SOI MOSFET TCAD model at the highintensity level ( V a = the response saturation could be associated with the avalancheeffect and also affected by the velocity saturation. The electricfield in the gate oxide was checked to be below the SiO breakdown field (10 MV/cm) indicating no gate breakdown.However, at high THz fields, carriers could be generated fromimpact ionization and travel into the channel and change theelectric field which could affect the response at the drain. Thiseffect of the avalanche model could be seen in Fig. 7 whichshows the profiles of the electric field transverse to the channelat the positions along the channel and closely (1 nm) belowthe SiO /Si interface within a period above threshold ( V gt = V a = ONCLUSION
In this work, the TCAD models valid in a large dynamicrange were presented. The TCAD models explain the ex- static t=0/20f t=2/20f t=4/20ft=6/20f t=8/20f t=10/20f t=12/20ft=14/20f t=16/20f t=18/20f
Fig. 9. Profiles of the impact ionization generation rate at the high intensitylevel ( V a = perimentally observed response saturation at high intensitylevels of the incident THz radiation (above 1 V). By activatingor deactivating different physical mechanisms in the TCADmodels, the reason for the response saturation effect was foundout to be associated with the gate leakage for AlGaAs/InGaAsHFETs and AlGaN/GaN HFETs and affected by the velocitysaturation and the avalanche effect for Si MOSFETs.R EFERENCES[1] M. Dyakonov and M. S. Shur, “Shallow water analogy for a ballisticfield effect transistor: New mechanism of plasma wave generation by dccurrent,”
Phys. Rev. Lett. , vol. 71, no. 15, pp. 2465–2468, Oct. 1993.[2] M. I. Dyakonov and M. S. Shur, “Detection, mixing, and frequency mul-tiplication of terahertz radiation by two-dimensional electronic fluid,”
IEEE Trans. Electron Devices , vol. 43, no. 3, pp. 380–387, Mar. 1996.[3] M. I. Dyakonov and M. S. Shur, “Plasma wave electronics: novelterahertz devices using two-dimensional electron fluid,”
IEEE Trans.Electron Devices , vol. 43, no. 10, pp. 1640–1645, Oct. 1996.[4] H. Marinchio et al. , “Room-temperature terahertz mixer based on thesimultaneous electronic and optical excitations of plasma waves in afield effect transistor,”
Appl. Phys. Lett. , vol. 96, no. 1, pp. 013502, Jan.2010.[5] S. Preu et al. , “Terahertz detection by a homodyne field effect transistormultiplicative mixer,”
IEEE Trans. Terahertz Sci. Technol. , vol. 2, no. 3,pp. 278–283, May 2012.[6] K. Wu, S. Muralidharan, and M. M. Hella, “A Wideband SiGe BiCMOSFrequency Doubler with 6.5-dBm Peak Output Power for Millimeter-Wave Signal Sources,”
IEEE Trans. Microw. Theory Tech. , vol. 86, no.3, pp. 187–200, Jan. 2018.[7] S. Muralidharan, K. Wu, and M. Hella, “A 165-230GHz SiGe amplifier-doubler chain with 5dBm peak output power,” in
IEEE RFIC Sympo-sium , San Francisco, CA, USA, 2016, pp. 302–305.[8] S. Muralidharan, K. Wu, and M. Hella, “A 216GHz 0.5 mW transmitterwith a compact power combiner in 65nm CMOS,” in
MicrowaveConference (APMC), 2015 Asia-Pacific , vol. 2, pp. 1–3, Dec. 2015.[9] K. Wu, B. Fahs, and M. Hella, “A 220 GHz OOK Outphasing Transmit-ter in 130-nm BiCMOS Technology,” , pp. 227–230, Oct. 2018.[10] W. Knap et al. , “Field effect transistors for terahertz detection: Physicsand first imaging applications,”
J. Infrared Millim. Terahertz Waves , vol.30, no. 12, pp. 1319–1337, Dec. 2009. [11] A. Lisauskas et al. , “Exploration of terahertz imaging with siliconMOSFETs,”
J. Infrared Millim. Terahertz Waves , vol. 35, no. 1, pp.63–80, Jan. 2014.[12] N. Dyakonova et al. , “Terahertz vision using field effect transistorsdetectors arrays,” in , pp. 711–714, May 2018.[13] S. Preu, M. Mittendorff, S. Winnerl, O. Cojocari, and A. Penirschke,“THz autocorrelators for ps pulse characterization based on Schottkydiodes and rectifying field-effect transistors,”
IEEE Trans. Terahertz Sci.Technol. , vol. 5, no. 6, pp. 922–929, Nov. 2015.[14] D. B. But et al. , “The dynamic range of THz broadband FET detectors,”in
Proc. SPIE, Terahertz Emitters, Receivers, and Applications IV , SanDiego, CA, USA, Sep. 2013, vol. 8846, Art. no. 884612.[15] D. B. But et al. , “Nonlinear photoresponse of field effect transistorsterahertz detectors at high irradiation intensities,”
J. Appl. Phys. , vol.115, no. 16, Apr. 2014, Art. no. 164514.[16] N. Dyakonova et al. , “Saturation of photoresponse to intense THzradiation in AlGaN/GaN HEMT detector,”
J. Appl. Phys. , vol. 120, no.16, Oct. 2016, Art. no. 164507.[17] N. Dyakonova et al. , “AlGaN/GaN HEMTs photoresponse to highintensity THz radiation,”
Opto-Electron. Rev. , vol. 23, no. 3, pp. 195–199, Sep. 2015.[18] X. Liu and M. Shur, “An efficient TCAD model for TeraFET detectors,”in , Orlando, FL, USA, Jan.2019, to be published.[19] Sentaurus Device User Guide (Version N-2017.09), Synopsys Inc.,Mountain View, CA, USA, 2017.[20] A. T. Ping, W. Liebl, G. Mahoney, S. Mahon, and O. Berger, “A High-Performance 0.13-m AlGaAs/InGaAs pHEMT Process Using SidewallSpacer Technology,” in , Apr. 2005.[21] A. Gutin, V. Kachorovskii, A. Muraviev, and M. Shur, “Plasmonicterahertz detector response at high intensities,”