An Optimization Method for AlGaN / GaN Heterojunction Field-Effect Transistors Based on a Physical-Based Compact Current-Voltage Model

Abstract

The present invention relates to an optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model, belonging to the technical field of device modeling. The present invention realizes the velocity-field relationship of AlGaN / GaN HFETs devices obtained based on Monte Carlo simulation, establishes a velocity-field model dependent on the gate bias voltage, which is closer to the experimental phenomena in real AlGaN / GaN HFETs. By parameter extraction, the parameters of the velocity-field model are brought into the derived physical-based compact current-voltage model. This method cleverly retains the direct correlation between the velocity-field model parameters and the AlGaN / GaN HFET device. In our compact current-voltage model, all parameters have specific physical meanings, and the parasitic resistance factor and the channel modulation effect are also taken into account. Our solution can accurately model AlGaN / GaN HFETs, predict and optimize device performance.

Description

Technical Field

[0001] The present invention relates to an optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model, and belongs to the technical field of transistor device research. Background Art

[0002] AlGaN / GaN heterojunction field effect transistors (HFETs) are important representatives of high-frequency devices and power electronic devices. With the continuous development of GaN device processes and circuits, some compact models have been developed to characterize and optimize the device characteristics of AlGaN / GaN HFETs. Some literature reports, based on the time-delay method, the experimental phenomenon that after some gate biases of GaN-based HFETs, when the gate voltage becomes more negative, the electron drift velocity in the region under the gate decreases. However, to our knowledge, the reported literature on physical-based modeling does not accurately describe the correlation between this AlGaN / GaN HFET velocity-field relationship and the gate bias (V gs ), and these velocity-field relationship models generally contain fitting parameters. The compact current-voltage model is mainly derived from the velocity-field relationship model, so fitting parameters are introduced through the velocity-field relationship model. Although good simulation results matching the experimental output characteristic curves can be obtained by fine-tuning the fitting parameters, this method of fine-tuning fitting parameters makes many physical parameters lose their physical meanings, and it is impossible to clearly obtain the specific scattering mechanism that causes the change of the AlGaN / GaN HFET velocity-field relationship under different gate biases for different devices. Even if data consistent with the experimental values is obtained in this way, it is very difficult to optimize the device characteristics based on the scattering mechanism and mobility model. Therefore, it is very urgent and necessary to establish a velocity-field model that depends on the gate voltage and has no fitting parameters to accurately describe the physical effect of the gate voltage on velocity modulation, and to derive a compact current-voltage model with parameters having exact physical meanings, so that the GaN-based HFET can be more intuitively obtained through the mobility model and scattering mechanism and the device performance can be optimized. Summary of the Invention

[0003] In order to overcome the deficiencies of the above-mentioned AlGaN / GaN HFET modeling method, the present invention provides a velocity-field relationship model without fitting parameters and a corresponding compact current-voltage model in which all parameters have clear physical meanings, so as to improve the accuracy of the AlGaN / GaN HFET device model and the feasibility of device characteristic optimization.

[0004] The technical solution of the present invention is as follows:

[0005] An optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model, comprising the following steps:

[0006] (1) Use a semiconductor parameter analyzer to test the current-voltage (I-V) curve of the AlGaN / GaN HFET device. According to the self-consistent iteration theory of additional polarization charges, obtain the functional relationship between the additional polarization charges of the AlGaN / GaN HFET and the channel bias voltage;

[0007] (2) According to Poisson's equation, calculate the potential distribution, electric field distribution, and two-dimensional electron gas electron density distribution in the channel region under the gate when the gate-source bias (V gs ) and drain-source bias (V ds ) are equal to 6 V:

[0008] For the low-field region of the channel under the gate of the AlGaN / GaN HFET, the gradual channel approximation is satisfied. The channel potential distribution, electric field distribution, and two-dimensional electron gas electron density distribution can be expressed by the following equations respectively:

[0009]

[0010]

[0011] Ns(x) = C i (V gs - V th - V(x)) (3)

[0012] Among them,

[0013] V gt = V g - v th

[0014] I dsat is the saturation current of the A1GaN / GaN HFET, E T is the critical electric field of the A1GaN / GaN HFET, Vg is the gate voltage, V th is the threshold voltage, R s is the gate-source parasitic resistance, x is the distance from the position point in the channel to the left end of the gate, Ns is the two-dimensional electron gas density in the channel, C i is the barrier layer capacitance of the AlGaN / GaN HFET, W is the gate width, μ is the low-field mobility, ε is the dielectric constant of the barrier layer, and d is the thickness of the barrier layer;

[0015] For the high-field region of the channel under the gate, the two-dimensional electron gas electron density distribution can be calculated by formula (3), and the potential distribution and electric field distribution can be calculated by the following equations:

[0016]

[0017]

[0018] Among them, q is the electric charge, and n sat is the electron density of the two-dimensional electron gas in the channel saturation region, cosh is the hyperbolic cosine function, sinh is the hyperbolic sine function, and L 1 is the length of the first region of the channel (low-field region);

[0019] According to the present invention, preferably, in step (2), the gate-source bias is taken as 0V, -1V, -2V, -3V.

[0020] (3) According to the finite element concept, divide the region under the gate into a number of grids; according to the present invention, preferably, in step (3), the length of each grid is 30nm < x 2 -x 1 < 50nm. The electron density and electric field strength of the two-dimensional electron gas inside each grid are the same. Each grid corresponds to a coordinate (x 1 , x 2 ), as well as the polarization charge density ρ 2 . We define ρ 2 as the reference polarization charge; according to formulas (1), (2), (4), and (5), obtain the potential V1 at the leftmost end of the channel, the average potential V2 of the channel grid, and the potential V3 at the rightmost end of the channel; according to the dependence of the additional polarization charge density on the channel bias voltage, each potential corresponds to a polarization charge density, thereby determining the polarization charge density ρ 1 at the leftmost end under the gate, the polarization charge density ρ 3 at the rightmost end under the gate, and the polarization charge density ρ 2 of each grid; according to the above physical quantities, calculate the additional scattering potential of PCF scattering for each grid:

[0021]

[0022] Among them, ε s is the relative dielectric constant of the barrier layer, ε 0 is the vacuum dielectric constant, L G is the gate length, L Gs is the gate-source spacing, L GD is the gate-drain spacing, e is the electric charge, x is the direction of the channel along the gate length, y is the direction of the channel along the gate width, and z is the direction of the channel along the longitudinal broadening; ρ 0 is the polarization charge density of the gate-source region and the gate-drain region;

[0023] Based on the additional scattering potential, obtain the transition matrix element of PCF scattering for each grid:

[0024]

[0025] Among them,

[0026] m* is the electron mass, is the Planck constant, E is the electron energy, ψ k (z) is the wave function of the electron, q x , q y are the components of q in the x and y directions, A is the normalization constant, (a, b) is the coordinate interval of the two-dimensional electron gas in the channel in the x direction, θ is the scattering angle from the initial state k to the final state k′; i is the imaginary unit;

[0027] According to the transition matrix element, the PCF scattering rate of each grid is obtained:

[0028]

[0029] where S(q, T e ) is the screening factor;

[0030] The channel potential distribution and the channel electric field distribution are determined by formulas (1)-(5). Under the electric field determined for each grid, according to the Monte Carlo simulation method, the velocity distribution of different grids is calculated, so as to obtain and establish the velocity-field relationship formula of the AlGaN / GaNHFET under different gate-source biases;

[0031] (4) According to the optimized AlGaN / GaNHFET velocity-field model formula (9) and the velocity-field relationship of the AlGaN / GaNHFET under different gate-source biases, the parameters of the velocity-field model under different gate-source biases are extracted:

[0032]

[0033] where V(E) is the relationship between the electron velocity and the electric field, V sat is the saturation velocity of the AlGaN / GaN HFET under different gate voltages, ε s is the relative dielectric constant of the barrier layer, the polarization charge density at the leftmost end under the gate is ρ 1 and the polarization charge density at the rightmost end under the gate is ρ 3 , E is the horizontal electric field along the channel direction, E t is the critical electric field; V sat , ρ 1 , ρ 3 , E t will all change continuously with the gate bias voltage; among them, ρ 1 , ρ 3 are determined according to the function related to the additional polarization charge and the channel bias voltage;

[0034] (5) Compare the velocity-field relationship of the AlGaN / GaN HFET obtained by Monte Carlo with that of the AlGaN / GaN HFET obtained by the model.

[0035] Substitute the velocity parameters, device size parameters, and voltage bias V of the AlGaN / GaN HFET at different gate-source biases into the following formula to obtain the device current in the linear region of the device: ds V gs Bring in the following formula to obtain the device current in the linear region of the device:

[0036]

[0037] where a = ε s +(ρ 3 -ρ 0 ) / (ρ 1 -ρ 0 )

[0038] λ = (wε s V sat (1 + a) / E t / d)

[0039] V gt = V g - V th

[0040] R n = 1 / (λE t )

[0041] B = (2aR n V ds + 2E t LR n + 2V gt (R s + R d ))

[0042] D = (R d - R s - 2aR n )(R d + R s )(-2V gt V ds + V ds 2 )

[0043] V ds is the non-intrinsic drain-source voltage, V gs is the non-intrinsic gate-source voltage, R d is the gate-drain parasitic resistance, and L is the gate length;

[0044] The velocity parameters of the AlGaN / GaN HFETs with different gate-source biases, the device size parameters, and the voltage biases V ds , V gs are substituted into the following formula to obtain the device saturation current:

[0045]

[0046] V GT = V G - V th

[0047] where e x = E(x) / E t ; I sat is the saturation current; e 0sat is the value of the electric field E / E t at x = 0 when the device reaches the saturation state; V dsat is the saturation voltage; V G is the intrinsic gate voltage;

[0048] The velocity parameters of the AlGaN / GaN HFETs with different gate-source biases, the device size parameters, and the voltage biases V ds , V gs are substituted into the following formula to obtain the device current in the velocity saturation region:

[0049]

[0050] where V GS is the intrinsic gate-source voltage, V DS is the intrinsic drain-source voltage, and p is an eigenvalue introduced to solve the Poisson equation, which is used to describe the influence of the gate voltage on the saturation region length and output conductance in the velocity saturation region;

[0051] When the obtained current value matches well with the experimental value, preferably, the error does not exceed 5%, which is considered good, indicating that the accuracy and convergence of our current-voltage model are good and can reflect the characteristics of the device. Our current-voltage model (formulas (10)-(12)) is written into the circuit software HSPICE software or ADS software in VerilogA language, and GaN HFETs with different sizes and structures are built in the circuit software, which can be used to optimize the design of high-performance GaN-based HFET circuits; this is an important basis for the GaN-based HFET circuit tape-out.

[0052] Our velocity-field model (formula (9)) and velocity-field model parameters (V sat , ρ 1 , ρ 3 , E t)Write into TCAD software, such as Crosslight, Sentaurus, Silvaco, which can replace the previous GaN velocity field model to achieve more accurate device simulation results closer to real physical effects, facilitating the design of high-performance GaN-based HFETs.

[0053] The beneficial effects of the present invention are as follows:

[0054] Compared with the conventional velocity field model, our invention realizes the velocity field relationship of AlGaN / GaN HFETs devices obtained based on Monte Carlo simulation, establishes a velocity field model dependent on gate bias voltage, which is closer to the experimental phenomena in real AlGaN / GaN HFETs. By parameter extraction, the velocity field model parameters are brought into the derived physically-based compact current-voltage model. This method cleverly retains the direct correlation between the velocity field model parameters and the AlGaN / GaN HFET device. In our compact current-voltage model, all parameters have specific physical meanings, and the parasitic resistance factor and channel modulation effect are also taken into account. Our work can achieve accurate modeling of AlGaN / GaN HFETs, predict and optimize device performance. Description of the Drawings

[0055] Figure 1 It is the flowchart of device compact current-voltage optimization provided by the present invention;

[0056] Figure 2 It is the relationship diagram of additional polarization charge and gate bias voltage provided by the present invention;

[0057] Figure 3 It is the schematic diagram of polarization charge distribution of A1GaN / A1N layer provided by the present invention;

[0058] Figure 4 It is the comparison diagram of the velocity field relationship obtained by Monte Carlo simulation and the velocity field relationship of the velocity field model provided by the present invention;

[0059] Figure 5 It is the comparison diagram of the output characteristic curve results of the compact current-voltage model simulation and the measured output characteristic curve provided by the present invention;

[0060] Figure 6 It is the comparison diagram of the transfer characteristic curve results of the compact current-voltage model simulation and the measured transfer characteristic curve provided by the present invention. Detailed Embodiments

[0061] The present invention will be further described below through embodiments in conjunction with the drawings, but not limited thereto.

[0062] Embodiment 1:

[0063] Reference Figure 1 As described above, an optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model includes the following steps:

[0064] (1) Use a semiconductor parameter analyzer to test the current-voltage (I-V) curve of an A1GaN / GaN HFET device. Through the self-consistent iteration theory of additional polarization charges, obtain the function of the additional polarization charges of the A1GaN / GaN HFET associated with the channel bias voltage, as Figure 2 shown.

[0065] (2) According to the Poisson equation, calculate the potential distribution, electric field distribution, and two-dimensional electron gas electron density distribution under the gate of the A1GaN / GaN HFET for different gate voltages (V gs = 0V, V gs = -1V, V gs = -2V, V gs = -3V), V ds = 6V).

[0066] For the low-field region of the channel under the gate of the A1GaN / GaN HFET, the slowly varying channel approximation is satisfied, and the channel potential distribution, electric field distribution, and two-dimensional electron gas electron density distribution can be expressed by the following equations respectively:

[0067]

[0068]

[0069] Ns(x) = C i (V gs - V th - V(x)) (3)

[0070] where

[0071] V gt = V g - V th

[0072] I dsat is the saturation current of the A1GaN / GaN HFET, E T is the critical electric field of the AlGaN / GaN HFET, Vg is the gate voltage, V th is the threshold voltage, R s is the gate-source parasitic resistance, x is the distance from the position point in the channel to the left end of the gate, Ns is the two-dimensional electron gas density in the channel, C i is the barrier layer capacitance of the AlGaN / GaN HFET, W is the gate width, μ is the low-field mobility, ε is the dielectric constant of the barrier layer, and d is the thickness of the barrier layer;

[0073] For the high-field channel region under the gate, the two-dimensional electron gas electron density distribution can be calculated by formula (3), and the potential distribution and electric field distribution can be calculated by the following formulas:

[0074]

[0075]

[0076] where q is the electric charge, n sat is the two-dimensional electron gas electron density in the channel saturation region, cosh is the hyperbolic cosine function, sinh is the hyperbolic sine function, and L 1 is the length of the first channel region (low-field region).

[0077] (3) In the simulation, we divide the region under the gate into several grids, each with a grid length of 30 nm < x 2 - x 1 < 50 nm.

[0078] The electron density and electric field strength inside each grid are the same. Each grid corresponds to a coordinate (x 1 , x 2 ), as well as the polarization charge density ρ 2 . Taking ρ 2 as the reference polarization charge, according to formulas (1), (2), (4), and (5), the potential V1 at the leftmost end of the channel, the average potential V2 of the channel grid, and the potential V3 at the rightmost end of the channel are obtained; according to Figure 2 the dependence of the additional polarization charge on the channel bias voltage, each potential corresponds to a polarization charge density, thereby determining the polarization charge density ρ 1 at the left end under the gate and the polarization charge density ρ 3 at the rightmost end under the gate, as well as the polarization charge density ρ 2 of each grid. From this, the AlGaN / GaNHFET V ds = 6 V, V gs = 0 V, and the polarization charge distribution at the AlGaN / AlN interface are obtained, as shown in Figure 3 . According to the above physical quantities, the additional scattering potential of PCF scattering for each grid is calculated by the following formula:

[0079]

[0080] where ε s is the relative dielectric constant of the barrier layer, ε 0 is the vacuum dielectric constant, L G is the gate length, L GS is the gate-source spacing, and L GDis the gate-drain spacing, e is the electric charge, x is the direction of the channel along the gate length, y is the direction of the channel along the gate width, z is the direction of the channel along the longitudinal broadening; ρ 0 is the polarization charge density of the gate-source region and the gate-drain region.

[0081] Substitute the calculated additional scattering potential into the following formula to obtain the transition matrix element of the PCF scattering for each grid:

[0082]

[0083] where

[0084] m* is the electron mass, is the Planck constant, E is the electron energy, χ k (z) is the wave function of the electron, q x , q y are the components of q in the x and y directions, A is the normalization constant, (a, b) is the coordinate interval of the two-dimensional electron gas in the channel in the x direction, θ is the scattering angle from the initial state k to the final state k′; i is the imaginary unit.

[0085] Substitute the calculated transition matrix element into the following formula to obtain the PCF scattering rate for each grid:

[0086]

[0087] where S(q, T e ) is the screening factor;

[0088] Determine the channel potential distribution and the channel electric field distribution through formulas (1)-(5). Under the electric field determined for each grid, according to the Monte Carlo simulation method, calculate the velocity distribution of different grids, so as to obtain and establish the velocity-field relationship formula of A1GaN / GaNHFET under different gate-source biases.

[0089] Among them, the Monte Carlo simulation is an independently developed program that requires inputting the PCF scattering rate of different grids and the electric fields of different grids each time. The program will simulate the random motion of electrons inside the channel of AlGaN / GaNHFET; after the electrons fly freely and scatter randomly under a series of alternating electric fields, they reach a stable velocity state.

[0090] (4) According to the following optimized A1GaN / GaNHFET velocity-field model formula (9) and the velocity-field relationship of AlGaN / GaNHFET under different gate-source biases, extract the parameters of the velocity-field model under different gate biases:

[0091]

[0092] Among them, V(E) is the relationship between the electron velocity and the electric field, and V sat is the saturation velocity of the AlGaN / GaN HFET under different gate voltages, and ε s is the relative dielectric constant of the barrier layer. The polarization charge density at the leftmost end under the gate is ρ 1 and the polarization charge density at the rightmost end under the gate is ρ 3 , E is the horizontal electric field along the channel direction, and E t is the critical electric field. V sat , ρ 1 , ρ 3 , and E t will all change continuously with the gate bias voltage. Among them, ρ 1 , ρ 3 is determined according to the function related to the additional polarization charge and the channel bias voltage.

[0093] (5) Compare the velocity-field relationship of the AlGaN / GaN HFET obtained by Monte Carlo with the velocity-field relationship of the AlGaN / GaN HFET obtained by the model, as Figure 4 shown;

[0094] Input the velocity parameters of the AlGaN / GaN HFET at different gate-source bias voltages, the device size parameters, and the voltage bias V ds , V gs into the following formula to obtain the device current in the linear region of the device:

[0095]

[0096] where a = ε s +(ρ 3 -ρ 0 ) / (ρ 1 -ρ 0 )

[0097] λ = (wε s v sat (1 + a) / E t / d)

[0098] V gt = V g - V th

[0099] R n = i / (λE t )

[0100] B = (2aR n V ds + 2E t LR n + 2V gt (Rs +R d ))

[0101] D = (R d -R s -2aR n )(R d +R s )(-2V gt V ds +V ds 2 )

[0102] V ds is the non-intrinsic drain-source voltage, V gs is the non-intrinsic gate-source voltage, R d is the gate-drain parasitic resistance, and L is the gate length;

[0103] Substitute the extracted speed parameters of the AlGaN / GaN HFET at different gate-source biases, the device size parameters, and the voltage bias V ds , V gs into the following formula to obtain the device saturation current:

[0104]

[0105]

[0106] where e x = E(x) / E t ; I sat is the saturation current; e 0sat is the value of the electric field E / Et at x = 0 when the device reaches the saturation state; V dsat is the saturation voltage; V G is the intrinsic gate voltage; V dsat is the saturation voltage;

[0107] Substitute the extracted speed parameters of the AlGaN / GaN HFET at different gate-source biases, the device size parameters, and the voltage bias V ds , V gs into the following formula to obtain the device current in the velocity saturation region:

[0108]

[0109] where V GS is the intrinsic gate-source voltage, V DS is the intrinsic drain-source voltage, and p is an eigenvalue introduced to solve the Poisson equation, which is used to describe the influence of the gate voltage on the saturation region length and output conductance in the velocity saturation region;

[0110] When the error between the obtained current value and the experimental value does not exceed 5%, and the matching is good, it indicates that the accuracy and convergence of our current-voltage model are good, and it can reflect the characteristics of the device. Write our current-voltage model (Equations (10)-(12)) into circuit software such as HSPICE software or ADS software using VerilogA language, and build GaN HFETs with different sizes and structures in the circuit software, which can be used to optimize the design of high-performance GaN-based HFET circuits; this is an important basis for the GaN-based HFET circuit tape-out.

[0111] Write our velocity field model (Equation (9)) and velocity field model parameters (V sat , ρ 1 , ρ 3 , E t ) into TCAD software such as Crosslight, Sentaurus, Silvaco, which can replace the previous GaN velocity field model and achieve more accurate device simulation results closer to the real physical effects, facilitating the design of high-performance GaN-based HFETs.

[0112] In this embodiment, Figure 5 Figure 6 The current-voltage (I-V) of the compact current-voltage model simulation is compared with the current-voltage curve of the A1GaN / GaN HFET measured experimentally to verify the accuracy of the compact current-voltage model.

Claims

1. An optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model, characterized in that, it includes the following steps: (1) Use a semiconductor parameter analyzer to test the current-voltage (I-V) curve of the AlGaN / GaN HFET device to obtain the functional relationship between the additional polarization charge of the AlGaN / GaN HFET and the channel bias voltage; (2) Calculate the potential distribution, electric field distribution, and two-dimensional electron gas electron density distribution in the channel region under the gate when the gate-source bias (V gs ) and drain-source bias (V ds ) of the AlGaN / GaN HFET are equal to 6 V: (3) Divide the area under the gate into several grids; the two-dimensional electron gas electron density and electric field strength within each grid are the same, and each grid corresponds to a coordinate (x 1 , x 2 ), as well as the polarization charge density ρ 2 . We define ρ 2 as the reference polarization charge; obtain the potential V1 at the leftmost end of the channel, the average potential V2 of the channel grid, and the potential V3 at the rightmost end of the channel; according to the dependence of the additional polarization charge density on the channel bias voltage, each potential corresponds to a polarization charge density, so as to determine the polarization charge density ρ 1 at the leftmost end under the gate, the polarization charge density ρ 3 at the rightmost end under the gate, and the polarization charge density ρ 2 of each grid; according to the above physical quantities, calculate the additional scattering potential of PCF scattering for each grid; According to the additional scattering potential, obtain the transition matrix element of the PCF scattering for each grid; According to the transition matrix element, obtain the PCF scattering rate for each grid; Determine the channel potential distribution and the channel electric field distribution. Under the electric field determined for each grid, according to the Monte Carlo simulation method, calculate the velocity distribution of different grids, so as to obtain and establish the velocity-field relationship formula of the AlGaN / GaN HFET under different gate-source biases; (4) According to the optimized AlGaN / GaN HFET velocity-field model and the velocity-field relationship of the AlGaN / GaN HFET under different gate-source biases, extract the parameters of the velocity-field model under different gate-source biases; (5) Compare the velocity-field relationship of the AlGaN / GaN HFET obtained by Monte Carlo with that obtained by the model. According to the extracted velocity parameters, device size parameters of the AlGaN / GaN HFET at different gate-source biases, and the voltage bias V ds ,V gs , obtain the device current in the linear region of the device; according to the extracted velocity parameters, device size parameters of the AlGaN / GaN HFET at different gate-source biases, and the voltage bias V ds, V gs , obtain the saturation current of the device; according to the extracted velocity parameters, device size parameters of the AlGaN / GaN HFET at different gate-source biases, and the voltage bias V ds ,V gs , obtain the current in the velocity saturation region of the device; when the obtained current values match well with the experimental values, write the current-voltage model into the circuit software HSPICE software or ADS software in Verilog A language, build GaN HFETs with different sizes and different structures in the circuit software to optimize the design of high-performance GaN-based HFET circuits; write the velocity-field model and velocity-field model parameters into the TCAD software to replace the previous GaN velocity-field model and design high-performance GaN-based HFETs. The velocity-field model parameters include V sat ,ρ 1, ρ 3, E t .

2. The optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model according to claim 1, characterized in that, in step (2), the gate-source bias values are 0V, -1V, -2V, -3V; For the low-field region of the channel under the gate of the AlGaN / GaN HFET, the channel potential distribution, the electric field distribution, and the two-dimensional electron gas electron density distribution can be expressed by the following formulas respectively: Ns(x) = C i (V gs - V th - V(x)) (3) Among them, V gt = V g -V t h I dsat is the saturation current of the AlGaN / GaN HFET, E T is the critical electric field of the AlGaN / GaN HFET, Vg is the gate voltage, V t h is the threshold voltage, R s is the gate-source parasitic resistance, x is the distance from the left end of the gate to the position point in the channel, Ns is the two-dimensional electron gas density in the channel, C i is the barrier layer capacitance of the AlGaN / GaN HFET, W is the gate width, μ is the low-field mobility, ε is the dielectric constant of the barrier layer, and d is the thickness of the barrier layer; For the high-field region of the channel under the gate, the two-dimensional electron gas electron density distribution can be calculated by formula (3), and the potential distribution and the electric field distribution can be calculated by the following formulas: where q is the electric charge, and n sat is the two-dimensional electron gas electron density in the channel saturation region, cosh is the hyperbolic cosine function, sinh is the hyperbolic sine function, and L 1 is the length of the first region of the channel (low-field region); In step (3), calculate the additional scattering potential of the PCF scattering for each grid: where ε s is the relative permittivity of the barrier layer, ε 0 is the permittivity of vacuum, L G is the gate length, L GS is the gate-source spacing, L GD is the gate-drain spacing, e is the electric charge quantity, x is the direction of the channel along the gate length, y is the direction of the channel along the gate width, z is the direction of the channel along the longitudinal broadening; ρ 0 is the polarization charge density of the gate-source region and the gate-drain region; According to the additional scattering potential, obtain the transition matrix element of the PCF scattering for each grid: Among them, $m^*$ is the electron mass, $\hbar$ is the Planck constant, $E$ is the electron energy, $\psi$ k $(z)$ is the wave function of the electron, $q$ x , $q$ y $q_x,q_y$ are the components of $q$ in the $x$ and $y$ directions, $A$ is the normalization constant, $(a,b)$ is the coordinate interval of the two-dimensional electron gas in the channel in the $x$ direction, $\theta$ is the scattering angle from the initial state $k$ to the final state $k'$; $i$ is the imaginary unit; According to the transition matrix element, obtain the PCF scattering rate for each grid: where S(q,T e ) is the shielding factor; In step (4), the optimized AlGaN / GaN HFET velocity-field model is as formula (9): where V(E) is the relationship between the electron velocity and the electric field, and V sat is the saturation velocity of the AlGaN / GaN HFET under different gate voltages, ε s is the relative dielectric constant of the barrier layer, the polarization charge density at the leftmost end under the gate is ρ 1 and the polarization charge density at the rightmost end under the gate is ρ 3 , E is the horizontal electric field along the channel direction, and E t is the critical electric field; V sat , ρ 1, ρ 3, E t will all change continuously with the gate bias voltage; among them, ρ 1, ρ 3 is determined according to the function related to the additional polarization charge and the channel bias voltage; In step (5), the specific calculation process of the device current in the linear region of the device is: where a = ε s +(ρ 3- ρ 0 ) / (ρ 1- ρ 0 ) λ=(wε s v sat (1 + a) / E t / d) V gt = V g - V th R n = 1 / (λE t ) B = (2aR n V ds + 2E t LR n + 2V gt (R s + R d )) D = (R d - R s - 2aR n )(R d + R s )(- 2V gt V ds + V ds 2 ) V ds is the non-intrinsic drain-source voltage, V gs is the non-intrinsic gate-source voltage, R d is the gate-drain parasitic resistance, and L is the gate length; The specific calculation process of the device saturation current is: V GT = V G - V th where e x = E(x) / E t ; I sat is the saturation current; e 0sat is the electric field E / E at x = 0 when the device reaches the saturation state; t V dsat is the saturation voltage; V G is the intrinsic gate voltage; The specific calculation process of the device current in the velocity saturation region is: where V GS is the intrinsic gate-source voltage, V DS is the intrinsic drain-source voltage, p is an eigenvalue introduced to solve the Poisson equation, which is used to describe the influence of the gate voltage on the saturation region length and output conductance in the velocity saturation region.

3. The optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model according to claim 1, characterized in that, In step (3), the length of each grid is 30 nm < x 2 - x 1 < 50 nm.

4. The optimization method for AlGaN / GaN heterojunction field effect transistors based on a physical-based compact current-voltage model according to claim 1, characterized in that, in step (5), when the error between the obtained current value and the experimental value does not exceed 5%, it is considered a good match.

Images