Improved simulation method for nonlinear direct current model of gallium nitride high-electron-mobility field effect transistor
By measuring and optimizing the nonlinear DC model on the GaN high electron mobility field effect transistor, and introducing self-heating parameters and softplus functions, the problem that existing models are difficult to accurately describe the characteristics of GaN transistors is solved, and the simulation accuracy and Simplicity, especially in the simulation of output transconductance.
Patent Information
- Application Number
- CN202411057881.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-02
- Publication Date
- 2024-11-12
AI Technical Summary
Existing empirical models are difficult to accurately describe the characteristics of gallium nitride high electron mobility field effect transistors, especially in the simulation of output transconductance.
By forming a high electron mobility field effect transistor of gallium nitride on a silicon carbide substrate, measuring its DC signal and transfer characteristic curve, extracting the cutoff voltage and optimizing the nonlinear DC model using test data, introducing the autothermal parameters μ and softplus functions, Calculate the output transconductance and electrical conductance to compare and optimize the simulation curve.
The simulation accuracy of the model is improved, especially in the saturation and amplification zones, the fitting process of output transconductance is simplified, the mean square error is reduced, and it is directly used for current, transconductance and simulation fitting and simulation.
Smart Images

Figure CN118940527A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of power devices, and in particular to a simulation method for an improved nonlinear direct current model of a gallium nitride high electron mobility field effect transistor. Background Art
[0002] Large signal models play an important guiding role in device characterization and nonlinear circuit design. The empirical large signal equivalent circuit model has been widely used in large signal modeling of transistor devices due to its simple structure, high accuracy and easy embedding in simulation software.
[0003] In order to maximize the efficiency of circuit design and shorten the design time, improving the accuracy of current-voltage models, output admittance, transmission and transconductance has always been the focus of research on equivalent circuit models of field effect transistors. At present, research on high electron mobility field effect transistors mainly uses empirical models to characterize their performance. The expressions of these empirical models are usually relatively simple and adjustable, so they can be widely used in various industrial scenarios. Due to the similarities between gallium nitride and gallium arsenide materials, some classic models originally developed for metal oxide semiconductor field effect transistors, such as the Statz model, Crutice model, and Angelov model, are used to characterize the performance of gallium nitride high electron mobility field effect transistors.
[0004] Compared with metal oxide semiconductor field effect transistors, the advantage of high electron mobility field effect transistors is that they have two materials with different band gaps. When a large number of electrons are transferred from the highly doped material with a higher band gap to the material with a lower band gap, a two-dimensional electron gas is generated at the interface of the two materials due to the limitation of the bound potential well in the vertical direction. Therefore, the traditional empirical voltage-current model, especially the model developed for gallium arsenide, is not suitable for describing the characteristics of the emerging gallium nitride high electron mobility field effect transistor. Summary of the invention
[0005] The technical problem to be solved by the present invention is to provide an improved simulation method of a nonlinear DC model of a gallium nitride high electron mobility field effect transistor, which can improve the simulation accuracy of the model and make the simulation of output transconductance simpler.
[0006] In order to solve the above technical problems, the present invention provides an improved simulation method of a nonlinear DC model of a GaN high electron mobility field effect transistor, comprising the following steps:
[0007] Step 1, forming a gallium nitride high electron mobility field effect transistor on a silicon carbide substrate;
[0008] Step 2, measuring the DC signal and transfer characteristic curve of the GaN high electron mobility field effect transistor;
[0009] Step 3: Extract the cut-off voltage V according to the transfer characteristic curve TH ;
[0010] Step 4: Optimize the established nonlinear DC model using test data and extract corresponding parameters;
[0011] Step 5, using the test data to calculate the output transconductance gm and output conductance gds of the GaN high electron mobility field effect transistor;
[0012] Step 6: Simulate the nonlinear DC model to obtain a simulation curve, and compare it with the test data.
[0013] Preferably, in step 4, the nonlinear DC model established is:
[0014]
[0015] Among them, a and n are fitting parameters, m is the voltage offset parameter, V ds is the drain voltage, γ is the shaping parameter, V gs is the gate voltage, V $% is the cut-off voltage, λ is the channel modulation parameter, μ is the self-heating parameter, and α is the fitting parameter.
[0016] Preferably, in step 4, the model is iteratively optimized using the Levenberg-Marquardt optimization algorithm within a 95% confidence interval, with a maximum number of iterations of 400 and a tolerance of 10 -9 . .
[0017] Preferably, in step 5,
[0018] Preferably, in step 5, the gate width of the silicon carbide substrate of the gallium nitride high electron mobility field effect transistor is 0.15 microns, the gate length is 10 microns, the number of gate fingers is 2; the gate length is 40 microns, the number of gate fingers is 4; the gate length is 60 microns, and the number of gate fingers is 8.
[0019] Preferably, for the measurement of the transfer characteristic curve, the Vgs range is -5V to 0V, the step size is 0.1V, and the Vds is set to 4V, 8V, 12V, 16V, and 20V, respectively.
[0020] Preferably, for the measurement of drain-source current, the range of Vds is 0-20V with a step size of 0.5V, and the range of Vgs is -5V to 0V with a step size of 0.25V.
[0021] Preferably, the cut-off voltage V TH is -3.25V.
[0022] The beneficial effects of the present invention are as follows: the present invention introduces the self-heating parameter μ, which improves the accuracy of the device in the saturation region. Due to the existence of the trap effect, the current of the device in the saturation region decreases. By introducing this parameter in the characterization of the source current, the current decrease in the saturation region can be effectively represented, thereby improving the accuracy in the saturation region; the softplus function is introduced to represent the gate current, and the softplus function is introduced in the representation of the gate current: This partial form is obtained by improving the softplus function, which improves the accuracy of the device in the amplification area, simplifies the fitting process of the output transconductance, and improves the fitting accuracy. For the fitting of gm, the process of obtaining the partial derivative of the proposed nonlinear model is avoided, and it can be directly used for the fitting of gm; the present invention can not only be directly used for the fitting of current, but also can be directly used for the fitting of gm; there is no need to obtain the partial derivative of the drain-source current, and it can be directly used for the fitting and simulation of the output transconductance, thereby improving the DC characteristics, transfer characteristics, and the accuracy of the output transconductance and the drain-source transconductance. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 This is a process diagram of the gallium high electron mobility field effect transistor of the present invention.
[0024] Figure 2 It is a schematic diagram for comparing the DC simulation data of the present invention, the simulation data of the traditional method and the DC measurement data.
[0025] Figure 3 It is a schematic diagram comparing the output transconductance simulation curve of the present invention, the simulation curve of the conventional method and the measurement data. DETAILED DESCRIPTION
[0026] like Figure 1 As shown, a simulation method for an improved nonlinear DC model of a GaN high electron mobility field effect transistor comprises the following steps:
[0027] Step 1, forming a gallium nitride high electron mobility field effect transistor on a silicon carbide substrate;
[0028] Step 2, measuring the DC signal and transfer characteristic curve of the GaN high electron mobility field effect transistor;
[0029] Step 3: Extract the cut-off voltage V according to the transfer characteristic curve TH ;
[0030] Step 4: Use the test data to train the established nonlinear DC model and extract corresponding parameters;
[0031] Step 5, using the test data to calculate the output transconductance gm and output conductance gds of the GaN high electron mobility field effect transistor;
[0032] Step 6: Simulate the nonlinear DC model to obtain a simulation curve, and compare it with the test data.
[0033] like Figure 2 and Figure 3 As shown in the figure, the squares are measured data, the dashed lines are the simulation results using the traditional model, and the solid lines are the simulation results using the model proposed in this paper. Compared with the traditional large signal model, the mean square error of the nonlinear DC model is as low as 2.44×10 -6 , the mean square error of gm is as low as 1.07×10 -4 .
[0034] The present invention introduces the self-heating parameter μ to improve the accuracy of the device in the saturation region. Due to the existence of the trap effect, the current of the device in the saturation region decreases. By introducing this parameter in the representation of the source current, the current decrease in the saturation region can be effectively represented, thereby improving the accuracy in the saturation region; the softplus function is introduced to represent the gate current, and the softplus function is introduced in the representation of the gate current: This partial form is obtained by improving the softplus function, which improves the accuracy of the device in the amplification area; the softplus function is introduced to simplify the fitting process of the output transconductance and improve the fitting accuracy. For the fitting of gm, the process of obtaining the partial derivative of the proposed nonlinear model is avoided, and it can be directly used for the fitting of gm; the present invention can not only be directly used for the fitting of current, but also can be directly used for the fitting of gm; there is no need to obtain the partial derivative of the drain-source current, and it can be directly used for the fitting and simulation of the output transconductance, thereby improving the accuracy of the DC characteristics, transfer characteristics, and the output transconductance and the drain-source transconductance.
Claims
1. An improved simulation method for a nonlinear DC model of a gallium nitride high electron mobility field effect transistor, characterized in that: The steps include: Step 1, forming a gallium nitride high electron mobility field effect transistor on a silicon carbide substrate; Step 2, measuring the DC signal and transfer characteristic curve of the GaN high electron mobility field effect transistor; Step 3: Extract the cut-off voltage V according to the transfer characteristic curve TH ; Step 4: Optimize the established nonlinear DC model using test data and extract corresponding parameters; Step 5, using the test data to calculate the output transconductance gm and output conductance gds of the GaN high electron mobility field effect transistor; Step 6: Simulate the nonlinear DC model to obtain a simulation curve, and compare it with the test data.
2. The simulation method of the improved GaN high electron mobility field effect transistor nonlinear DC model as claimed in claim 1, characterized in that: In step 4, the nonlinear DC model established is: Among them, a and n are fitting parameters, m is the voltage offset parameter, V ds is the drain voltage, γ is the shaping parameter, V gs is the gate voltage, V $% is the cut-off voltage, λ is the channel modulation parameter, μ is the self-heating parameter, and α is the fitting parameter.
3. The simulation method of the improved GaN high electron mobility field effect transistor nonlinear DC model as claimed in claim 1, characterized in that: In step 4, the model was iteratively optimized using the Levenberg-Marquardt optimization algorithm within a 95% confidence interval, with a maximum number of 400 iterations and a tolerance of 10 -9 . .
4. The simulation method of the improved GaN high electron mobility field effect transistor nonlinear DC model as claimed in claim 1, characterized in that: In step 5, 5. The simulation method of the improved GaN high electron mobility field effect transistor nonlinear DC model as claimed in claim 1, characterized in that: In step 5, the gate width of the silicon carbide substrate of the GaN high electron mobility field effect transistor is 0.15 microns, the gate length is 10 microns, the number of gate fingers is 2; the gate length is 40 microns, the number of gate fingers is 4; the gate length is 60 microns, and the number of gate fingers is 8.
6. The simulation method of the improved GaN high electron mobility field effect transistor nonlinear DC model as claimed in claim 4, characterized in that: For the measurement of the transfer characteristic curve, the Vgs range is -5V to 0V with a step size of 0.1V, and the Vds is set to 4V, 8V, 12V, 16V, and 20V respectively.
7. The simulation method of the improved GaN high electron mobility field effect transistor nonlinear DC model as claimed in claim 4, characterized in that: For the measurement of drain-source current, the range of Vds is 0 to 20V with a step size of 0.5V, and the range of Vgs is -5V to 0V with a step size of 0.25V.
8. The simulation method of the improved GaN high electron mobility field effect transistor nonlinear DC model as claimed in claim 1, characterized in that: In step 3, the cut-off voltage V TH is -3.25V.
