Skip to content

Accurate, Efficient and Reliable Small-Signal Modeling Approaches for GaN HEMTs

MetadataDetails
Publication Date2023-01-01
JournalIEEE Access
AuthorsSaddam Husain, Anwar Jarndal, Mohammad Hashmi, Fadhel M. Ghannouchi
InstitutionsUniversity of Calgary, Nazarbayev University
Citations12
AnalysisFull AI Review Included

This analysis details the development and validation of accurate Small-Signal Equivalent Circuit Models (SSECMs) for Gallium Nitride (GaN) High Electron Mobility Transistors (HEMTs) grown on diamond substrates.

  • Core Achievement: Successfully implemented and compared a Scanning-Based Systematic (SBS) extraction method against three novel Optimization Algorithm (OA)-based hybrid methods: Marine Predators Algorithm (MPA), Pelican Optimization Algorithm (POA), and Tunicate Swarm Algorithm (TSA).
  • Performance Validation: All four modeling approaches achieved excellent agreement between measured and simulated S-parameters, with minimum errors consistently around 0.06% at the Cold-FET pinch-off condition.
  • Frequency Range: The models are validated and reliable across an extensive frequency range, from 0.1 GHz up to 40 GHz, suitable for mmWave applications.
  • Methodological Trade-offs: The SBS method is significantly faster (7.03 s execution time) but requires assumptions (e.g., Cpga = Cpda), limiting its physical relevance and generalization.
  • OA Superiority: OA-based hybrid methods (MPA, POA, TSA) produce physically relevant and reliable SSECMs without requiring simplifying assumptions, and accurately capture complex device physics (kink effect, RF-DC dispersion).
  • Efficiency Winner: The TSA-based hybrid approach demonstrated the fastest execution time among the OAs (50.48 s for 50 iterations), making it the most efficient hybrid technique tested.
ParameterValueUnitContext
Device TechnologyGaN HEMTN/AGrown on 500 µm Diamond Substrate
Frequency Range0.1 to 40GHzRF Characterization
Gate Length (L)0.25µmDevice Specification
Gate Finger Width (Wg)125µmDevice Specification (4-finger device)
Barrier Layer Thickness20nmAlGaN
VGS Operating Range-3 to 0VGate-Source Voltage
VDS Operating Range0 to 30VDrain-Source Voltage
Minimum S-Parameter Error (TSA Hybrid)0.060343%Cold-FET Pinch-off (VGS = -3 V, VDS = 0 V)
Transconductance (Gm)251.86mSVGS = -1 V, VDS = 10 V (TSA Model)
Output Conductance (Gds)6.74mSVGS = -1 V, VDS = 10 V (TSA Model)
Execution Time (SBS)7.03sTotal systematic scanning process
Execution Time (TSA Hybrid)50.48s50 iterations (Fastest OA)
Execution Time (MPA Hybrid)100.35s50 iterations
Execution Time (POA Hybrid)99.33s50 iterations

The study utilized two primary methodologies—Scanning-Based Systematic (SBS) and Optimization Algorithm (OA)-Based Hybrid—to extract the 16 parameters of the SSECM.

  1. Cold-FET and Unbiased Measurements: S-parameters were measured across 0.1 GHz to 40 GHz at two key conditions: Cold-FET pinch-off (VGS = -3 V, VDS = 0 V) for capacitance extraction, and unbiased (VGS = 0 V, VDS = 0 V) for parasitic inductance and resistance extraction.
  2. Total Capacitance Extraction: Total capacitances (Cgst, Cgdt, Cdst) were calculated from the imaginary parts of the Y-parameters (Im[Yij]) using simple linear regression analysis at pinch-off.
  3. Scanning-Based Systematic (SBS) Extraction:
    • Parasitic pad capacitances (Cpga, Cpgi) were determined via a systematic search (scanning) across incremental values.
    • The method assumes Cpga = Cpda and Cds = 0 at pinch-off to simplify the calculation of Cgs and Cpdi.
    • Parasitic inductances (Lg, Ls, Ld) and resistances (Rg, Rs, Rd) were extracted from the slopes of the Z-parameters derived from unbiased measurements.
  4. OA-Based Hybrid Extraction (MPA, POA, TSA):
    • The initial values and bounds for all 16 SSECM parameters were set using results from the direct extraction method.
    • Optimization algorithms (MPA, POA, TSA) were initialized with hyperparameters (50 iterations, 2500 search agents) and reliability conditions to filter non-physical values.
    • The OA iteratively minimized the objective function (Eab), which quantifies the weighted error between measured and simulated S-parameters across the full frequency range.
    • The intrinsic elements (Gm, Gds, τ, Cgs, Cgd, Cds, Ri, Rgd) were calculated following the de-embedding of the extrinsic elements determined by the OA.
  5. Model Validation: The final SSECMs were validated across multiple active bias points (e.g., VGS = -1 V, VDS = 10 V) to confirm the frequency-independent behavior of the intrinsic elements and the model’s ability to replicate complex device physics.

The accurate and reliable small-signal modeling of GaN-on-Diamond HEMTs is foundational for designing high-performance microwave and millimeter-wave circuits, supporting the following industries and applications:

  • 5G and 6G Wireless Communication: Enabling the design of high-efficiency, wide-bandwidth Radio Frequency Power Amplifiers (RFPAs) required for next-generation base stations and massive MIMO systems.
  • High Power RF Systems: GaN-on-Diamond technology offers superior thermal management, making these devices ideal for high-power density applications where self-heating is critical.
  • Radar and Defense Systems: Used in high-frequency, high-power transmitters and receivers for advanced radar, electronic warfare, and satellite communication systems.
  • Millimeter Wave (mmWave) Technology: The validated 40 GHz frequency range directly supports the development of high-speed data links and short-range communication systems operating in the mmWave spectrum.
  • Internet of Things (IoT) Infrastructure: Providing reliable, high-frequency components for complex IoT network backbones and high-throughput gateways.
View Original Abstract

This article presents accurate, efficient and reliable small-signal model parameter extraction approaches applied to Gallium Nitride (GaN) High Electron Mobility Transistor (HEMT). Firstly, a scanning-based systematic model parameter extraction methodology is developed. Then, newly reported Optimization Algorithms (OAs) namely Marine Predators Algorithm (MPA), Pelican Optimization Algorithm (POA) and Tunicate Swarm Algorithm (TSA) in combination with direct extraction method are utilized to develop hybrid model parameter extraction methodologies. Lastly, both the scanning-based systematic and OA-based hybrid modelling procedures are thoroughly validated and demonstrated on a GaN HEMT grown on diamond substrate to identify their pros and cons in distinct application settings. Moreover, reliability, accuracy, convergence behavior, complexity and execution time of MPA-, POA- and TSA-based hybrid extraction procedures are also discussed. We found that both classes of the approaches are able to produce an excellent agreement between the measured and modelled S-parameters for a wide frequency range up to 40 GHz. However, OA-based hybrid modelling procedures are more physically relevant.