Nicholas C. Miller, Matt Grupen, Gian Piero Gibiino, Justin King
©SHUTTERSTOCK.COM/ZEF ART
Gallium nitride (GaN) HEMTs are a key technological component in current and next-generation RF and millimeter-wave (mm-wave) integrated circuits and subsystems. Applications where GaN has made considerable impact include radar, communications, satellite communications, and electronic warfare. A critical aspect of the RF design cycle is accurate modeling of the GaN HEMTs. Modeling of the transistors can take two significantly different forms: one referred to as technology computer-aided design (TCAD) and the other called physics-based compact modeling. These subjects were the basis of the “Best Presentation” winning talk at the IEEE Microwave Theory and Technology Design Automation Committee [Technical Committee 2 (TC-2)] Modeling and Optimization Workshop, where the modeling research at the Air Force Research Laboratory (AFRL) was discussed. Namely, AFRL’s custom TCAD solver called Fermi kinetics transport (FKT) [1] and physics-based compact modeling using the Advanced SPICE Model for HEMTs (ASM-HEMT) [2] were presented. This article provides an overview of the physics-based modeling efforts at AFRL.
Preface
The IEEE Microwave Theory and Technology Society (MTT-S) Young Professionals Workshop on Modeling and Optimization for Active Devices, first held on 25 October 2022, was a virtual live event free of charge for all registrants. It served as a platform for IEEE Young Professionals involved in RF active device modeling to showcase their research results and interact with the community. Cosponsored by Technical Committee 2 (TC-2) and TC-3 of the MTT-S, it included presentation sessions by young professionals and a panel discussion with senior experts.
While the second edition of this workshop took place on 17 October 2023 under the new name IEEE MTT-S Young Professionals Workshop on Modeling, Optimization, and Measurement Techniques for Active Devices, here we recall the Best Presentation from last year’s event, as voted by attendees live during the event. The article below is written by the recipients of that Best Presentation Award and is based on the research work described during their presentation last year.
—Gian Piero Gibiino and Justin King
Cochairs of the IEEE MTT-S Young Professionals Workshop on Modeling, Optimization, and Measurement Techniques for Active Devices
TCAD modeling has long been a staple of transistor research and development. In the world of Boltzmann solvers, there exist mainly two categories of methods that capture hot-electron effects in field effect transistors: stochastic methods and deterministic methods called energy transport solvers [3]. The AFRL custom TCAD solver, FKT, is a type of energy transport method which employs four moments of the Boltzmann transport equation to account for charge transport and hot-electron dynamics. A key distinguishing feature of the FKT solver is the choice of closure relation [3] for the system of nonlinear partial differential equations. FKT utilizes a thermodynamic heat flow for its closure relation [4] rather than the conventional approach of Fourier’s law [3]. The result of a thermodynamic heat flow appears to be a robust and stable charge transport framework, and the favorable convergence properties have been benchmarked against a commercial hydrodynamics solver [5]. The excellent stability and convergence properties of the solver have also enabled large-signal RF simulations of GaN HEMTs as demonstrated in [6] and [7]. Complete electronic band structure and scattering mechanisms are incorporated into the FKT device simulator through isosurface integral preprocessing routines. The calculated data are then fit with power laws in order to utilize general Fermi integral calculations in the transport solver [8], [9], [10]. The inclusion of these fundamental materials properties is critical for accurate device simulations [11]. Another key difference between FKT and conventional Boltzmann solvers is the use of an “energy-dependent mobility” for simulating electron transport [8], [9], [10]. The standard approach for deterministic solvers is choice of an electric field-dependent mobility model [12]. The FKT framework provides a self-consistent solution of the charge transport equations with Maxwell’s equations. Discretizations of Maxwell’s equations are accomplished through a method called Delaunay–Voronoi surface integration [13]. The self-consistent solution of Maxwell’s equations and charge transport enables simulations of hot-electron effects coupled to propagating-wave effects, which could be critical for simulation of next-generation mm-wave transistors [11]. Defect dynamics are an important topic for GaN HEMT dc and RF operation, and the FKT solver has been recently utilized to shed light on the fundamental physics of trapping effects. The well-known dc-IV kink effect was explored with FKT, and excellent simulation results reproduced this trapping effect in a particular GaN HEMT [14]. An interesting conclusion from this work was the role of field-assisted barrier defect ionization, a trapping mechanism less commonly considered in the literature [14]. The gate lag response of GaN HEMTs was also explored using the FKT device simulator. This response is a signature of trapping effects in GaN HEMTs where the drain current transient is monitored while the gate supply voltage is pulsed from an off state to an on state with a constant drain supply voltage. FKT simulations of gate lag were experimentally validated with drain current transient measurements of AFRL’s 140-nm GaN HEMT technology (GaN140) [15]. The main results of this work are illustrated in Figures 1 and 2. These results indicate the level of accuracy provided by the FKT device simulation framework compared with dc and drain current transient measurements of AFRL’s GaN140 technology.
Figure 1. The FKT simulated (red lines) and measured (black circles) dc drain currents for the AFRL GaN140 HEMT. (Source: [15].)
Figure 2. The FKT simulated (red line) and measured (black circles) drain current transient response to a pulsed gate supply voltage for the AFRL GaN140 HEMT. (Source: [15].)
Physics-based compact modeling of GaN HEMTs continues to be an important component in the RF integrated circuit design cycle. Two prominent physics-based nonlinear models for GaN HEMTs are the MIT virtual source GaN RF model (MVSG) [16] and the ASM-HEMT [2]. The compact nonlinear modeling research at AFRL has focused on the ASM-HEMT; however, the MVSG could be used in a similar fashion for research projects. At AFRL, the ASM-HEMT was first validated for the AFRL GaN140 HEMT across a broad range of fundamental frequencies from X- to Ka-band using a single model [17], and the nonlinear harmonic modeling of the GaN140 process was validated in [18]. These first results provided confidence in the accuracy of nonlinear models extracted from the internal GaN process at AFRL. Several research projects were initiated after the successful demonstration of accurate ASM-HEMT modeling. First, nonlinear embedding modeling of GaN HEMTs was explored in collaboration with The Ohio State University [19], [20]. This type of nonlinear modeling could enable circuit designers to instantly prescribe the harmonic impedances for specific amplifier classes without the need for tedious harmonic active load pull [21]. Statistical nonlinear modeling [22] and deep learning-based nonlinear modeling [23] of GaN HEMTs are other active research topics that have the goals of analyzing process variations and enabling fast and accurate extraction of GaN HEMTs, respectively. Finally, a recent thrust has been devoted to temperature-dependent nonlinear modeling of GaN HEMTs using the ASM-HEMT [24]. This technique could provide the framework for extrapolation beyond measured ambient temperatures and integrated circuit designs for high-temperature applications. Two results from this work are illustrated in Figures 3 and 4, which demonstrate the dc and large-signal RF validation of the ASM-HEMT versus ambient temperature.
Figure 3. The ASM-HEMT simulated (dashed lines) and measured (solid lines) dc transconductance of the AFRL GaN140 HEMT. (Source: [24].)
Figure 4. The ASM-HEMT simulated (dashed lines) and measured (solid lines) power added efficiency (PAE) of the AFRL GaN140 HEMT. (Source: [24].)
The combination of custom TCAD simulations using FKT and accurate physics-based nonlinear compact modeling provides an excellent framework for understanding the fundamental physics of GaN HEMTs. Future research with the FKT device simulator could explore various state-of-the-art III-nitride transistor technologies and their dc and RF responses in extreme environments and could predict the thermal and trapping signatures of the transistors. For ASM-HEMT modeling, future research could include model extraction from FKT device simulations and nonlinear RF modeling for extreme environment applications.
This work was supported by the Air Force Office of Scientific Research under Grant LRIR 21RYCOR073.
[1] N. C. Miller, M. Grupen, and J. D. Albrecht, “Recent advances in GaN HEMT modeling using Fermi kinetics transport,” in Proc. Device Res. Conf. (DRC), 2023, p. 1, doi: 10.1109/DRC58590.2023.10187047.
[2] S. Khandelwal, Advanced SPICE Model for GaN HEMTs (ASM-HEMT): A New Industry-Standard Compact Model for GaN-Based Power and RF Circuit Design. Cham, Switzerland: Springer Nature, 2022.
[3] T. Grasser, T.-W. Tang, H. Kosina, and S. Selberherr, “A review of hydrodynamic and energy-transport models for semiconductor device simulation,” Proc. IEEE, vol. 91, no. 2, pp. 251–274, Feb. 2003, doi: 10.1109/JPROC.2002.808150.
[4] M. Grupen, “An alternative treatment of heat flow for charge transport in semiconductor devices,” J. Appl. Phys., vol. 106, no. 12, Dec. 2009, Art. no. 123702, doi: 10.1063/1.3270404.
[5] A. Tunga et al., “A comparison of a commercial hydrodynamics TCAD solver and Fermi kinetics transport convergence for GaN HEMTs,” J. Appl. Phys., vol. 132, no. 22, Dec. 2022, Art. no. 225702, doi: 10.1063/5.0118104.
[6] N. C. Miller, J. D. Albrecht, and M. Grupen, “Large-signal RF GaN HEMT simulation using Fermi kinetics transport,” in Proc. 74th Annu. Device Res. Conf. (DRC), Jun. 2016, pp. 1–2, doi: 10.1109/DRC.2016.7548427.
[7] E. White, A. Tunga, N. C. Miller, M. Grupen, J. D. Albrecht, and S. Rakheja, “Large-signal modeling of GaN HEMTs using Fermi kinetics and commercial hydrodynamics transport,” in Proc. Device Res. Conf. (DRC), 2023, pp. 1–2, doi: 10.1109/DRC58590.2023.10186900.
[8] M. Grupen, “Energy transport model with full band structure for GaAs electronic devices,” J. Comput. Electron., vol. 10, no. 3, pp. 271–290, Sep. 2011, doi: 10.1007/s10825-011-0364-9.
[9] N. C. Miller, “Large-signal RF simulation and characterization of electronic devices using fermi kinetics transport,” Ph.D. dissertation, Michigan State Univ., East Lansing, MI, USA, 2017.
[10] N. C. Miller, M. Grupen, K. Beckwith, D. Smithe, and J. D. Albrecht, “Computational study of Fermi kinetics transport applied to large-signal RF device simulations,” J. Comput. Electron., vol. 17, no. 4, pp. 1658–1675, Dec. 2018, doi: 10.1007/s10825-018-1242-5.
[11] M. Grupen, “Three-dimensional full-wave electromagnetics and nonlinear hot electron transport with electronic band structure for high-speed semiconductor device simulation,” IEEE Trans. Microw. Theory Techn., vol. 62, no. 12, pp. 2868–2877, Dec. 2014, doi: 10.1109/TMTT.2014.2365781.
[12] Sentaurus Device User Guide, Synopsys, Inc., Mountain View, CA, USA, 2020.
[13] N. C. Miller, J. D. Albrecht, and M. Grupen, “Delaunay–Voronoi surface integration: A full-wave electromagnetics discretization for electronic device simulation,” Int. J. Numer. Model., Electron. Netw., Devices Fields, vol. 29, no. 5, pp. 817–830, Sep./Oct. 2016, doi: 10.1002/jnm.2146.
[14] M. Grupen, “Reproducing GaN HEMT kink effect by simulating field-enhanced barrier defect ionization,” IEEE Trans. Electron Devices, vol. 66, no. 9, pp. 3777–3783, Sep. 2019, doi: 10.1109/TED.2019.2928536.
[15] N. C. Miller et al., “Experimentally validated gate-lag simulations of AlGaN/GaN HEMTs using Fermi kinetics transport,” IEEE Trans. Electron Devices, vol. 70, no. 2, pp. 435–442, Feb. 2023, doi: 10.1109/TED.2022.3229291.
[16] U. Radhakrishna, T. Imada, T. Palacios, and D. Antoniadis, “MIT virtual source GaNFET-high voltage (MVSG-HV) model: A physics based compact model for HV-GaN HEMTs,” Physica Status Solidi, vol. 11, nos. 3–4, pp. 848–852, Apr. 2014, doi: 10.1002/pssc.201300392.
[17] N. C. Miller et al., “Accurate nonlinear GaN HEMT simulations from X-to Ka-band using a single ASM-HEMT model,” in Proc. IEEE 21st Annu. Wireless Microw. Technol. Conf. (WAMICON), 2021, pp. 1–4, doi: 10.1109/WAMICON47156.2021.9615166.
[18] N. C. Miller et al., “Accurate non-linear harmonic simulations at X-band using the ASM-HEMT model validated with NVNA measurements,” in Proc. IEEE Topical Conf. RF/Microw. Power Amplifiers Radio Wireless Appl. (PAWR), 2022, pp. 11–13, doi: 10.1109/PAWR53092.2022.9719743.
[19] M. Lindquist, P. Roblin, and N. C. Miller, “ASM-HEMT embedding model for accelerated design of PAs,” in Proc. 34th General Assem. Scientific Symp. Int. Union Radio Sci. (URSI GASS), 2021, pp. 1–4, doi: 10.23919/URSIGASS51995.2021.9560502.
[20] M. Lindquist, P. Roblin, N. C. Miller, D. Davis, R. Gilbert, and M. Elliott, “Experimental validation of ASM-HEMT nonlinear embedding modeling of GaN HEMTs at X-band,” in Proc. 100th ARFTG Microw. Meas. Conf. (ARFTG), 2023, pp. 1–4, doi: 10.1109/ARFTG56062.2023.10148891.
[21] H. Jang, P. Roblin, and Z. Xie, “Model-based nonlinear embedding for power-amplifier design,” IEEE Trans. Microw. Theory Techn., vol. 62, no. 9, pp. 1986–2002, Sep. 2014, doi: 10.1109/TMTT.2014.2333498.
[22] F. Chavez, N. C. Miller, D. T. Davis, and S. Khandelwal, “Statistical modeling of manufacturing variability in GaN HEMT I-V characteristics with ASM-HEMT,” in Proc. IEEE/MTT-S Int. Microw. Symp. (IMS), 2022, pp. 375–377, doi: 10.1109/IMS37962.2022.9865304.
[23] F. Chavez, D. T. Davis, N. C. Miller, and S. Khandelwal, “Deep learning-based ASM-HEMT I-V parameter extraction,” IEEE Electron Device Lett., vol. 43, no. 10, pp. 1633–1636, Oct. 2022, doi: 10.1109/LED.2022.3197800.
[24] N. C. Miller, A. Brown, M. Elliott, and R. Gilbert, “Temperature dependent large-signal modeling of GaN HEMTs at Ka-band using the ASM-HEMT,” in Proc. IEEE Wireless Microw. Technol. Conf. (WAMICON), 2023, pp. 21–24, doi: 10.1109/WAMICON57636.2023.10124913.
Digital Object Identifier 10.1109/MMM.2023.3321549