CN115204089B

CN115204089B - A parameter extraction method for GaN HEMT based on ASM model

Info

Publication number: CN115204089B
Application number: CN202210814388.3A
Authority: CN
Inventors: 张茹; 尹文婷; 李翡; 朱能勇; 吾立峰; 刘强
Original assignee: Chengdu Huada Jiutian Technology Co ltd
Current assignee: Chengdu Huada Jiutian Technology Co ltd
Priority date: 2022-07-12
Filing date: 2022-07-12
Publication date: 2024-11-26
Anticipated expiration: 2042-07-12
Also published as: CN115204089A

Abstract

The application discloses a parameter extraction method and device of a GaN HEMT based on an ASM model and a computer readable storage medium. Comprising the following steps: setting the technological parameters of the GaN HEMT; dividing more than one GaN HEMT device into one or more groups based on different sizes and temperatures, wherein the GaN HEMT devices in each group respectively have the same size and different temperatures, firstly selecting a 1 st GaN HEMT device with a 1 st size S ₁ and a 1 st temperature T ₁ for each group of GaN HEMT devices, adjusting different input parameter values to obtain a plurality of measurement data output by the 1 st GaN HEMT device, and generating a measurement curve based on the measurement data, wherein the measurement curve comprises a capacitance-voltage relation curve; extracting relevant parameters of a capacitance-voltage relation curve; storing an ASM model of the 1 st GaN HEMT device; and applying the ASM model to GaN HEMT devices with the same size S ₁ and other temperatures, and continuously extracting relevant parameters of the capacitance-voltage relation curve.

Description

Parameter extraction method of GaN HEMT based on ASM model

Technical Field

The invention relates to the technical field of microelectronic integrated circuits, in particular to a parameter extraction method of a GaN HEMT.

Background

Si-based power devices have moved toward the limits of the devices, and GaN materials are used as the base materials for High Electron Mobility Transistor (HEMT) devices with their excellent device characteristics (wide forbidden band, high saturation speed, high thermal stability, and high current density). GaN HEMTs exhibit their excellent characteristics in high power microwave integrated circuit applications, including high breakdown voltage, high operating frequency, high power density, and the like. ASM-HEMT is recognized by compact model alliance (CMC) as a new standard model of GaN device, and besides large signal modeling, ASM-HEMT realizes accurate modeling on characteristics such as field plate capacitance, trap effect, kink effect, noise and the like.

In general, in parameter extraction of the GaN HEMT, the influence of temperature on a device is not considered, and the accuracy of parameter extraction is reduced. In addition, the optimization is to make the objective function converged to the minimum point, and the traditional optimization algorithm is to continuously adjust the search direction and then perform one-dimensional search in the search direction, so that very practical engineering significance is obtained, but the efficiency is low and the accuracy is limited. The prior model after parameter extraction has few multiplexing conditions, and the extracted model is only applied to the current device and has low practicability to other devices. There is a need for improvements to achieve better parameter extraction requirements for GaN HEMTs.

Disclosure of Invention

In order to improve efficiency, convenience and accuracy of parameter extraction of GaNHEMT, and universality of a parameter extraction model of a GaN HEMT, and adapt to actual working requirements, the invention provides a parameter extraction method of the GaN HEMT based on an ASM model, which comprises the following steps: setting technological parameters of the GaN HEMT, wherein the technological parameters comprise one or more of size parameters, functional parameters and working environment parameters;

more than one GaN HEMT device is divided into one or more groups based on different sizes and temperatures, wherein the GaN HEMT devices in each group have the same size and different temperatures respectively,

For each group of GaN HEMT devices, the following parameter extraction is performed:

Selecting a1 st GaN HEMT device with a1 st size S ₁ and a1 st temperature T ₁, adjusting different input parameter values to obtain a plurality of measurement data output by the 1 st GaN HEMT device, and generating a measurement curve based on the measurement data, wherein the measurement curve comprises a capacitance-voltage relation curve;

extracting relevant parameters of a capacitance-voltage relation curve;

storing an ASM model of the 1 st GaN HEMT device;

and applying the ASM model to GaN HEMT devices with the same size S ₁ and at other temperatures, and extracting relevant parameters of a capacitance-voltage relation curve.

In one implementation, the measurement curve further includes a current-voltage relationship, and the parameter extraction method further includes extracting a related parameter of the current_voltage relationship.

In one implementation, the process parameters include: the device gate length L, the device gate width W, the working temperature T, the device gate index NF, the device gate source length LSG and the device drain source length LDG.

In one implementation, extracting the capacitance-voltage relationship related parameter includes: relevant parameters are extracted through the inverse conducting capacitance-drain voltage curve, the output capacitance-drain voltage curve, the input capacitance-drain voltage curve and the grid capacitance-grid voltage curve.

In one implementation, extracting relevant parameters of the current-voltage relationship includes optimizing the parameters locally and then globally, the local optimization including:

1) Extracting a threshold voltage voff and a subthreshold slope nfactor from a low current region of the low source drain voltage Vds for the drain current-gate voltage curve id_vg;

2) And extracting a low-field mobility parameter u0, a mobility attenuation coefficient ua and a mobility second-order attenuation coefficient ub from a low-source drain voltage Vds high-current region for a drain current_gate voltage curve id_vg. ;

3) Extracting DIBL parameters and subthreshold attenuation parameters from a leakage current-leakage voltage curve id_vd in a low current region of high source-leakage voltage Vds;

4) For the leakage current-leakage voltage curve id-vd, extracting the speed saturation parameter in the high current region of the high source-drain voltage Vds,

And fine tuning parameters obtained through the local optimization in the global optimization.

In one implementation, the above local and global optimizations are performed based on an optimizer using the Trust-region algorithm Trust-region, which performs a multidimensional search directly in one region to find the optimal value.

In one implementation, the optimizer automatically amplifies the boundary when the optimized parameters reach the boundary value during the optimization process.

In one implementation, the optimization interval is customized by using a script, and the region where the parameters need to be extracted is selected on the measurement graph by calling a function for specifying the optimization interval range in the script.

In one implementation, extracting the capacitance_voltage curve parameter includes:

Extracting drain-source capacitance parameters, edge capacitance parameters, bias voltage parameters, leakage saturation voltage parameters and bias voltage parameters from a reverse conduction capacitance-drain voltage (crss _vds) curve;

Extracting drain-source capacitance parameters and drain-edge capacitance parameters from an output capacitance-drain voltage (coss _vds) curve;

Extracting a gate-source overlap capacitance parameter from an input capacitance-drain voltage (ciss _vds) curve;

Extracting AlGaN layer thickness parameters from a grid capacitance-grid voltage (cgg-vgs) curve

The invention also provides a GaN HEMT parameter extraction device based on the ASM model, which is characterized by comprising a memory and a processor, wherein a program running on the processor is stored in the memory, and the processor executes the parameter extraction method of the GaN HEMT based on the ASM model when running the program.

The present invention also provides a computer readable storage medium having stored thereon computer instructions, wherein the computer instructions, when executed by a processor, perform the aforementioned parameter extraction method for GaN HEMT based on ASM model.

Compared with the prior art, the GaN parameter extraction method has the following advantages and positive effects:

1) And considering the influence of temperature, the GaN HEMT devices are grouped according to different sizes and temperatures, and the parameter extraction is respectively carried out, so that the accuracy of the parameter extraction is improved.

2) The optimizer 'Trust region' defined by the Trust region algorithm (Trust region) greatly reduces root mean square error (RMS) compared with the existing optimizers; the boundary can be automatically amplified, and an optimal value can be found when the region cannot meet the conditions; meanwhile, the method has the default optimization iteration times with the best possible effect and the least time consumption. The optimization process is quick and efficient, and the efficiency and accuracy of parameter extraction are improved.

3) The user-defined optimization interval is called, so that different requirements of users can be met, the flexibility and convenience are high, and the practical engineering significance is achieved.

4) Firstly extracting a capacitance-voltage (cv) curve, then extracting a current-voltage (iv) curve, and simultaneously, adopting a strategy of firstly locally optimizing and then globally optimizing the current-voltage (iv) curve, so that the extraction is more targeted, and the extraction efficiency and accuracy are improved;

5) The ASM model with the extracted parameters can be stored, and the stored ASM model is applied to devices with the same size and different temperatures, so that the effect of extracting the parameters of the whole model can be achieved by only adjusting part of parameters and curves for the rest devices, the time for extracting the parameters is greatly saved, the efficiency of extracting the parameters is increased, and the ASM model has very strong engineering practical significance.

Drawings

Fig. 1 is a flow chart of extracting GaN HEMT parameters based on an ASM model according to the present invention;

FIG. 2 is a simulation and measurement image of the present invention after the extraction of the parameters of the reverse capacitance-drain voltage (crss _vd) curve;

FIG. 3 is a simulation and measurement image of the output capacitance-drain voltage (coss _vd) curve after the extraction of parameters is completed in the present invention;

FIG. 4 is a simulation and measurement image of the input capacitance-drain voltage (ciss _vd) curve after the extraction of parameters is completed in the present invention;

FIG. 5 is a simulation and measurement image of the gate capacitance-gate voltage (cgg-vg) curve after the extraction of parameters is completed in the present invention;

FIG. 6 is a simulation and measurement image of the drain current_gate voltage (id_vg) curve after the extraction of parameters is completed in the present invention;

FIG. 7 is a simulation and measurement image of the drain current_gate voltage (id_vg) curve after the extraction of parameters is completed in the present invention;

FIG. 8 is a simulation and measurement image of the drain current-drain voltage (id_vd) curve after the extraction of parameters is completed in the present invention;

FIG. 9 is a simulation and measurement image of the prior art after the extraction of the parameters of the reverse capacitance-drain voltage (crss _vd) curve;

FIG. 10 is a prior art simulation and measurement image of the output capacitance-drain voltage (coss _vd) curve after the extraction of parameters is completed;

FIG. 11 is a prior art simulation and measurement image of the input capacitance-drain voltage (ciss _vd) curve after the extraction of parameters is completed;

FIG. 12 is a simulation and measurement image of the prior art after the extraction of the gate capacitance-gate voltage (cgg-vg) curve parameters;

FIG. 13 is a simulation and measurement image of the prior art after the drain current_gate voltage (id_vg) curve parameter extraction is completed;

FIG. 14 is a simulation and measurement image of the prior art after the drain current_gate voltage (id_vg) curve parameter extraction is completed;

FIG. 15 is a simulation and measurement image of the prior art after the completion of the extraction of the drain current-drain voltage (id_vd) curve parameters;

Detailed Description

The technical solutions of the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which are obtained by a person skilled in the art based on the embodiments of the present application, fall within the scope of protection of the present application.

The terms first, second and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged, as appropriate, such that embodiments of the present application may be implemented in sequences other than those illustrated or described herein, and that the objects identified by "first," "second," etc. are generally of a type, and are not limited to the number of objects, such as the first object may be one or more. Furthermore, in the description and claims, "and/or" means at least one of the connected objects, and the character "/", generally means that the associated object is an "or" relationship. The accurate large-signal model has important significance on the circuit design of the device, and the large-signal device model is the basis of the characteristic characterization of the microwave device and the optimization of the circuit design. The nonlinear current-voltage (I-V) model is used as the core of the device large signal model, and the extraction of model parameters is the basis of device large signal modeling. The rapid and accurate parameter extraction method not only can improve the modeling efficiency, but also can shorten the circuit design period. The application provides an effective parameter extraction method aiming at GaN HEMT.

Fig. 1 is a flowchart of a GaN HEMT parameter extraction method according to the present invention, and the GaN HEMT parameter extraction method according to the present invention is described in detail below with reference to fig. 1.

In step 101, process parameters of the GaN HEMT device are set.

The technological parameters of the GaN HEMT device set in the invention comprise: a device gate length (L), a device gate width (W), an operating temperature (T), a device gate index (NF), a device gate source Length (LSG), a device drain source Length (LDG), and the like.

In step 102, more than one GaN HEMT device is divided into one or more groups based on different dimensions and temperature parameters.

Specifically, for example, the GaN HEMT devices are divided into a plurality of groups according to different sizes S ₁、S₂、S₃ … of the GaN HEMT devices, and in each group, the GaN HEMT devices with the same size have different temperatures T ₁、T₂、T₃ …, so that the temperatures in the different groups of the plurality of groups :G₁(S₁,T₁ T₂ T_3…),G₂(S₂,T₁ T₂ T_3…),… of the GaN HEMT devices may be the same, or may be partially the same or completely different.

Referring to table 1, in the 1 st group of devices G ₁, the GaN HEMT devices W1000T-40, W1000T150, W1000T25 have the same dimensions (w=100u, l=0.6u), and the temperatures of the three devices are-40 respectively. 150. And 25.. In the group 2 device G ₂, the GaN HEMT devices W100T-40, W100T150, W100T25 have the same dimensions (w=25u, l=0.6u), and the temperatures of the three are-40 respectively. 150. And 25.. In the 3 rd group of devices G ₂, the GaN HEMT devices W500T-40, W500T150, W500T25 have the same dimensions (w=50u, l=0.6u), and the temperatures of the three devices are-40 respectively. 150. And 25..

Thus, each group of parameters of the GaN HEMT device to be extracted is a device with the same size and different temperatures.

TABLE 1

In step 103, for each group of GaN HEMT devices G _n(S_i,T_j) in the multiple groups, n, i, j=1, 2,3, …, different input and output parameters are set, different parameter values are input to each group of GaN HEMT devices to obtain measurement data output by each group of GaN HEMT devices, and a measurement curve is generated based on the measurement data.

More specifically, different input and output parameters are set for the GaN HEMT devices in each group, different parameter values are input for each group of GaN HEMT devices to obtain measurement data output by each group of GaN HEMT devices, and a measurement curve is generated based on the measurement data, wherein the measurement curve comprises one or more of the following methods: for example: aiming at a1 st device G ₁(S₁,T₁ in the 1 st group of GaN HEMT devices, setting input parameters as drain voltage (vds), and output parameters as reverse conducting capacity (crss), output capacity (coss) and input capacity (ciss), sequentially changing the value of the drain voltage (vds) from 0v to 12v, and respectively obtaining actual measurement values of the reverse conducting capacity (crss), the output capacity (coss) and the input capacity (ciss) output by the GaN HEMT devices through a measuring instrument.

As shown in fig. 2, in the measurement curve image obtained based on the measurement data, the horizontal axis drain voltage (vds) is an input value, and the vertical axis reverse conductivity (crss) is an output value. The real points in fig. 2 are measured data points, the curve is a simulated graph, RMS represents the mean value of the error values fitted to be 0.57%, where the maximum value RMAX of the error present is 1.57%.

As shown in fig. 3, in the measurement curve image obtained based on the measurement data, the horizontal axis drain voltage (vds) is an input value, the vertical axis output capacitance (coss) is an output value, the real points are measurement data points in fig. 3, and the curve is a simulation diagram. RMS represents a fitted error value of 0.35% average, with a maximum value RMAX of 0.75% error present.

As shown in fig. 4, in the measurement curve image obtained based on the measurement data, the horizontal axis drain voltage (vds) is an input value, the vertical axis input capacitance (ciss) is an output value, the solid points in fig. 4 are measurement data points, and the curve is a simulation diagram. RMS represents a fitted error value of 0.69% on average, with a maximum value RMAX of 1.15% for the error present.

Similarly, for the 1 st group G ₁(S₁,T₁ of the GaN HEMT device, the input parameter is set to be the gate voltage (vgs), the given value is changed from-6 v to 6v, and the value of the gate capacitance (cgg) of the GaN HEMT device is obtained through the measuring instrument.

As shown in fig. 5, in the measurement curve image obtained based on the measurement data, the horizontal axis gate voltage (vgs) is an input value, the vertical axis gate capacitance (cgg) is an output value, the solid points in fig. 5 are measurement data points, and the curve is a simulation diagram. RMS represents a fitted error value of 5.53% average, with a maximum value RMAX of 26.5% error.

Similarly, for the 1 st device G ₁(S₁,T₁ in the 1 st group of GaN HEMT devices, the input parameter is set to be the gate voltage (vgs), the given value is changed from 0V to 5V in sequence, the value of the leakage current (ids) of the GaN HEMT devices is obtained through the measuring instrument, and meanwhile, the source-drain voltage (Vds) is set to be the scanning value, and the values are changed from 0.1V to 12.0V in sequence.

As shown in fig. 6, the graph of the measurement curve obtained based on the measurement data is that the horizontal axis gate voltage (vgs) is an input value, the vertical axis leakage current (ids) is an output value, fig. 6 is a graph of the leakage current_gate voltage (id_vg) under the conditions of the gate width w=100 um, the gate length l=0.6 um, and the temperature t=25 ℃, the real points are measurement data points, and the graph is a simulation graph. RMS represents a fitted error value of 7.65% average, with a maximum value RMAX of 9.61% error.

As shown in fig. 7, the graph of the measurement curve obtained based on the measurement data is that the horizontal axis gate voltage (vgs) is an input value, the vertical axis leakage current (ids) is an output value, fig. 7 is that all leakage currents_gate voltage (id_vg) are plotted under the conditions of the gate width w=100 um, the gate length l=0.6 um, and the temperature t=25 ℃, and the solid points in fig. 7 are measurement data points, and the graph is a simulation graph. RMS represents a fitted error value of 1% average, with a maximum value RMAX of 4.14% for error.

Similarly, for the 1 st device G ₁(S₁,T₁ in the 1 st group of GaN HEMT devices, the input parameter is set to be the drain voltage (vds), and the given value is changed in sequence from 0V to 12V, the value of the drain current (ids) of the GaN HEMT devices is obtained through the measuring instrument, and meanwhile, the gate voltage (vgs) is set to be the scanning value, and is changed in sequence from 0.1V to 5.0V.

As shown in fig. 8, the horizontal-axis leakage voltage (vds) is an input value, the vertical-axis leakage current (ids) is an output value, the graph of fig. 8 is a graph of all leakage currents_leakage voltages (id_vd) under the conditions of a gate width w=100 um, a gate length l=0.6 um, and a temperature t=25 ℃, the real points in fig. 8 are measurement data points, and the graph is a simulation graph. RMS represents a fitted error value of 20.41% average, with a maximum value RMAX of 99.97% error.

In step 104, parameters of a Capacitance_voltage (CV) curve are extracted for each of the plurality of groups G _n(S_i,T_j of GaN HEMT devices based on the measurement curve obtained in step 103.

The Advanced Spice Model (ASM) is a physical base model based on surface potential, and besides large-signal modeling, the model realizes accurate modeling on characteristics such as GaN field plate capacitance, trap effect, kink effect and noise. Is considered an industry compact standard model of power GaN and radio frequency GaN. The ASM model includes process parameters and model parameters, wherein the process parameters are determined by the device manufacturing process, such as channel length, width, and number of gate fingers.

In the automated parameter extraction process of the ASM model using the optimizer, the parameter extraction is essentially by adjusting the model parameters and fitting the test curve with the results of the simulation. As shown in the measurement curve image of fig. 2, RMS represents the error value of the fit.

In the automatic parameter extraction of the ASM model according to the present invention, for example, for one device G ₁(S₁,T₁ in the 1 st group of GaN HEMT devices), parameters of a capacitance-voltage (CV) curve are extracted, specifically including: for a GaN HEMT device with size S ₁, a capacitance-voltage (CV) curve at normal temperature T ₁ (25 ℃) is fitted.

Extracting a drain-source capacitance parameter (cgdo), a drain-edge capacitance parameter (cfgd), a bias voltage parameter (cgdl) and a leakage saturation voltage parameter (vdsatcv) from a reverse conduction capacitance-drain voltage (crss _vd) curve;

(2) Extracting a drain-source capacitance parameter (cdso) from an output capacitance-drain voltage (coss _vd) curve, a drain-edge capacitance parameter (cfd), controlling a parameter (aj) which is irrelevant to Cds under a low source-drain voltage (Vds), accessing a zero source-drain voltage (Vds) into a regional capacitance parameter (cj 0), and setting a parameter (mz) of Caccd attenuation under a high source-drain voltage (Vds), and setting a potential parameter (vbi) in a drain;

(3) Extracting a gate-source overlap capacitance parameter (cgso) for an input capacitance-drain voltage (ciss _vd) curve;

(4) And extracting quantum mechanical effect pre-factor and switching parameters (adosi) in inversion of related parameters (tbar) such as AlGaN layer thickness parameters and the like from a grid capacitance-grid voltage (cgg-vg) curve, and inverting a CV curve slope parameter (bdosi) under QME (charge centroid parameter-QME) and inverting a starting point parameter (qm 0 i) from the charge centroid parameter-QME. ;

for the GaN HEMT device with the size of S ₁, after the parameters of the CV curve of the normal temperature T ₁ are extracted, the ASM model of the GaN HEMT device is stored, the stored ASM model of the GaN HEMT device is directly applied to other devices (T ₁、T₂、T₃ …) with the same size of S ₁ in the 1 st group of GaN HEMT devices, and parameter extraction based on the ASM model is carried out based on the same steps.

In the invention, the used software XModel has the function of Apply model to other local(s), and can directly apply the adjusted ASM model to the parameter extraction of GaN HEMT devices with the same size and other temperatures. Therefore, the parameters of the GaN HEMT device at other temperatures with the same size are extracted only by fine adjustment, so that the workload of parameter extraction can be greatly reduced.

In the invention, parameters of Capacitance_voltage (CV) curves with the dimensions of S ₁ and normal temperature (T ₁) are extracted first, then an ASM model of the GaN HEMT device is stored, and then the ASM model is applied to other temperatures (T ₁、T₂ …) with the same dimensions, for example: the corresponding parameter extraction is carried out under the conditions of high temperature (150 ℃) and low temperature (-40 ℃) and the like, so that only partial parameters influenced by the temperature need to be regulated in the high temperature and the low temperature.

In step 105, parameters of a current_voltage (IV) curve are extracted for each of the plurality of packets G _n(S_i,T_j) of the GaN HEMT device based on the measurement curve obtained in step 103.

In the invention, the IV parameter extraction process is optimized locally and then optimized globally, and the specific process is as follows:

wherein the local optimization comprises:

Under the linear drain current bias condition, parameters such as a cut-off Voltage (VOFF), a subthreshold slope (NFACTOR) and the like are extracted from a drain current-gate voltage (id_vg) curve. Wherein the rough estimate of the off-Voltage (VOFF) is the gate Voltage (VG) value of the linear proportion of the drain current (ID) starting to rise in the drain current-gate voltage (ID-VG) plot. The rough value can be adjusted in the parameter extraction process to obtain the best effect. Subthreshold slope (NFACTOR) controls the subthreshold slope of the device, extracted by matching the model to a linear drain Voltage (VD) conditional drain current_gate voltage (ID-VG) feature on a logarithmic scale.

The cut-off Voltage (VOFF) and subthreshold slope (NFACTOR) carrier low-field mobility U0 and mobility vertical-field related parameters UA and UB should be extracted from the linear drain Voltage (VD) conditional drain current-gate voltage (ID-VG) profile. The transconductance GM and its derivatives GM0 and GM00 under linear VD conditions can be accurately modeled using U0, UA and UB parameters.

After the linear drain Voltage (VD) condition fitting is completed, parameters should be extracted from the high drain Voltage (VD) condition (vds=1.2v). As described above, the drain current_gate voltage (ID-VG) characteristic under the high drain Voltage (VD) condition is first focused, and the off-Voltage (VOFF) or low current region parameter is extracted. Due to the drain induced barrier lowering effect, the cut-off voltage at high drain Voltage (VD) conditions is lowered, which can be modeled by extracting DIBL parameters ETA0 and VDSCALE. The subthreshold slope also decreases under high VD conditions, which can be modeled with CDSCD parameters in the ASM-GaN-HEMT model.

Next, the self-heating effect (shmod =1) is turned on, and the parameter extraction is performed on the drain current_gate voltage (ID-VG) curve in the high drain Voltage (VD). The key parameters extracted under this condition are the velocity saturation parameter (VSAT), the channel length modulation parameter (λ) and the nonlinear series resistance parameters NS0ACCS, S, MEXPACCS, D and U0ACCS. VSAT is preferably extracted from the intermediate current level and the transconductance (GM) is still increased with the gate Voltage (VG) under high VD conditions. The nonlinear series resistance causes the transconductance (GM) to decrease with the gate Voltage (VG), and the relevant parameters should be adjusted to the appropriate range. By using the parameters extracted in this step, the trans-conductance under different drain Voltage (VD) conditions can be accurately simulated. After the above local optimization is completed, the whole between the measured data and the simulation curve current_voltage (IV) has a better fitting effect, then global optimization is performed, fine adjustment is performed on parameters, and the current_voltage (IV) curve can be accurately modeled.

The steps complete the direct current parameter extraction flow of the ASM-GaN-HEMT model at room temperature. Specifically, for the 1 st group G ₁ of GaN HEMT devices, the parameters of the extracted current-voltage (IV) curve are as follows:

After setting the bias condition (vth_half=vth/2, vgg, vth, vdlin) and the initial value, the local optimization is performed first, including: (1) Extracting a threshold voltage (voff) and a subthreshold slope (nfactor) from a drain current_gate voltage curve (id_vg) in a low current region (a current region corresponding to a gate voltage (vgs) before reaching the threshold voltage) of a low source drain voltage (Vds);

(2) Extracting a low-field mobility parameter (u 0), a mobility attenuation coefficient (ua) and a mobility second-order attenuation coefficient (ub) from a drain current-gate voltage curve (id_vg) in a low-source drain voltage (Vds) high-current region (a current region corresponding to a gate voltage (vgs) reaching a threshold voltage);

(3) Extracting parameters such as DIBL parameters, subthreshold attenuation and the like from a leakage current_leakage curve (id_vd) in a high source-drain voltage (Vds) low current region, extracting a leakage-induced barrier reduction effect parameter (eta 0) from a vthgm _vds curve, extracting a leakage voltage-induced subthreshold slope change parameter (cdscd), and extracting a leakage-induced barrier reduction effect parameter (vdscale) related to the source-drain voltage Vds;

(4) For a leakage current-leakage voltage curve (id_vd), extracting relevant parameters such as a speed saturation parameter self-heating and the like in a high-source-leakage voltage (Vds) high-current region, extracting a parameter leakage contact resistance (rdc), a source contact resistance (rsc), a source access region unit area two-dimensional electron gas density (n 0 accs), a leakage access region unit area two-dimensional electron gas density (n 0 accd), a resistance (rth 0), a mobility temperature relevant parameter (ute), an access region two-dimensional electron gas density temperature relevant parameter (kns 0), a saturation speed (vsat), an access region saturation speed value (vsataccs) and a speed saturation parameter (thesat) from a drain current-leakage voltage (id_vd) curve.

In the embodiment of the invention, the low source drain voltage Vds is greater than 0 and less than 1/2 source drain voltage Vds; the high source drain voltage Vds is greater than 1/2 of the source drain voltage Vds.

As shown in fig. 6, the horizontal axis gate voltage (vgs) is an input value, the vertical axis leakage current (ids) is an output value, fig. 6 is a graph of all leakage currents_gate voltage (id_vg) under the conditions of a gate width w=100 um, a gate length l=0.6 um, and a temperature t=25 ℃, and the solid points in fig. 6 are measurement data points and the graph is a simulation graph. RMS represents a fitted error value of 7.65% average, with a maximum value RMAX of 9.61% error.

The parameters are finely adjusted through the optimization, so that a better fitting effect is obtained.

After the local optimization is completed, global optimization is performed

Fig. 7 and 8 show the effect of global optimization of the present invention.

As shown in fig. 7, the overall optimized image is shown, the horizontal axis gate voltage (vgs) is an input value, the vertical axis leakage current (ids) is an output value, fig. 7 is a graph of all leakage currents_gate voltage (id_vg) under the conditions of a gate width w=100 um, a gate length l=0.6 um, and a temperature t=25 ℃, the real points in fig. 7 are measurement data points, and the graph is a simulation graph. RMS represents a fitted error value of 1% average, with a maximum value RMAX of 4.14% for error.

As shown in fig. 8, the overall optimized image is shown, the horizontal axis drain voltage (Vds) is an input value, the vertical axis drain current (ids) is an output value, fig. 8 is a graph of all drain current_drain voltages (id_vd) under the conditions of a gate width w=100 um, a gate length l=0.6 um, and a temperature t=25 ℃, the real points in fig. 8 are measurement data points, and the graph is a simulation graph. RMS represents a mean value of 20.41% error of the fit.

In the invention, parameters of a current_voltage (IV) curve with the size of S ₁ and normal temperature (T ₁) are extracted first, then an ASM model of the GaN HEMT device is stored, and then the ASM model is applied to the GaN HEMT device with the same size and at other temperatures (T ₁、T₂ …) through Apply model to other local (S) functions of XModel, for example: the corresponding parameter extraction is carried out under the conditions of high temperature (150 ℃) and low temperature (-40 ℃) and the like, so that only partial parameters influenced by the temperature need to be regulated in the high temperature and the low temperature.

The local optimization and global optimization of the invention are optimizers using Trust-region algorithm (Trust-region). The optimizer is characterized by searching in an area. In a typical one-dimensional search, the process of moving from the x point to the next point can be described as: x+αd where αd is the displacement in the d direction and can be denoted as s. The trust domain algorithm used by the optimizer of the present invention directly determines the displacement s, and meanwhile, unlike one-dimensional search, it does not determine the search direction d first. If the displacement can sufficiently lower the objective function value, the trust zone is enlarged; if the objective function value cannot be sufficiently reduced, the reliability area is reduced. This iterates until convergence. The optimization is to make the objective function converged to the minimum point, the traditional optimization algorithm is to continuously adjust the searching direction and then perform one-dimensional searching in the searching direction, and the newly defined optimizer is to directly perform multidimensional searching in a region, so that the speed of searching the optimal value can be greatly increased. The optimizer has the advantages that root mean square error (RMS) is obviously reduced in application, and the optimization speed is greatly improved in actual operation; in the optimization process, when the optimized parameters reach the boundary values, the optimizer can also realize the automatic expansion of the boundaries, and the principle of the automatic expansion is as follows: when the parameter value reaches the set boundary value, 0.5 times of the self boundary value is added or subtracted to be set as a new boundary value on the basis of the current boundary value. The support setting allows the soft boundary to automatically expand times (Beyond Boundary Times); meanwhile, the default maximum iteration number (Max Iterations) is set on the basis of not affecting the accuracy of the optimization result and the extraction efficiency. Therefore, the optimization speed and the precision are improved.

The optimizer can also customize an optimization interval extracted by parameters by using a script based on real points of measured data, and select a region needing to extract the parameters on a measuring curve chart by calling a function for designating the range of the optimization interval in the script. The functions called in the Script are all based on python type scripts, so that the function is convenient and accurate to call, and the practical engineering application significance is increased. The callable functions are listed as follows (only part is listed):

double [ ] getData (String exp): acquiring all data values in the selected area;

int nearestIndexOf (double value) return index of the value closest to the value in the array;

double abs (double x) return the absolute value of x;

returning the minimum value in the array;

double max (): return the maximum value in the array;

double [ ] derivative (double [ ] y, double [ ] x) y derives from x;

void addRegion (String regionName, double min, double max): adding a selection interval, and selecting an area with min < x < max;

as shown in fig. 2 and 6, the blue region selected by the frame is defined by Script. After the maximum value and the minimum value selected by the frame are specified by the function, the parameter extraction ranges of the steps 103 and 104 are performed in the region, so that the simulation curve can be quickly and directly fitted. The user can implement more efficient free-box selection as desired.

The invention, in connection with the specific embodiments, is exemplified as follows:

The method comprises the following process parameters: for example, the measurement data of the GaN HEMT with w=100um, l=0.6um, nf=10, t=25 ℃, ldg =0.7m, lsg=0.35 m, and the parameter extraction results obtained after the dc parameter extraction according to the above steps are as follows: capacitance_voltage (cv) curve parameter extraction results:

Current_voltage (iv) curve parameter extraction results:

drawing corresponding CV curves and IV curves by using the extracted parameter values, and combining the measured values to respectively obtain a curve image of a measured value and an imitated value of the reverse capacitance_drain voltage (crss _vd) shown in fig. 2, a curve image of a measured value and an imitated value of the output capacitance_drain voltage (coss _vd) shown in fig. 3, a curve image of a measured value and an imitated value of the input capacitance_drain voltage (ciss _vd) shown in fig. 4, a curve image of a measured value and an imitated value of the gate capacitance_gate voltage (cgg_vg) shown in fig. 5, a curve image of a measured value and an imitated value of the drain current_gate voltage (id_vg) shown in fig. 6, a curve image of a measured value and an imitated value of the drain current_gate voltage (id_vg) shown in fig. 7, and a curve image of a measured value and an imitated value of a leak current_drain voltage (id_vd) shown in fig. 8.

Fig. 2-8 are optimized simulation results of the present invention, and fig. 9-15 are optimized simulation results of the prior art, specifically as follows:

Fig. 2 and fig. 9 are simulation fitting results of the reverse conducting capacitor_drain voltage (crss _vd) curve under the same size, the same temperature condition, the same input parameter and the same output parameter, respectively, as those of the prior art. It can be seen intuitively that the average error RMS of the fitting in the present invention is 0.57%, and the average error RMS of the fitting in the prior art is 2.58%, and the present invention reduces the average error value of the fitting.

Fig. 3 and fig. 10 are simulation fitting results of the output capacitor_drain voltage (coss _vd) curve under the same size, the same temperature condition, the same input parameter and the same output parameter, respectively, as those of the prior art. It can be seen intuitively that the average error RMS of the fit in the present invention is 0.35%, and the average error RMS of the fit in the prior art is 1.52%, and the present invention reduces the average error value of the fit.

Fig. 4 and fig. 11 are simulation fitting results of the input capacitance-drain voltage (ciss _vd) curve under the same size, the same temperature condition, the same input parameter and the same output parameter, respectively, as the prior art. It can be seen intuitively that the average error RMS of the fit in the present invention is 0.69%, and the average error RMS of the fit in the prior art is 0.87%, and the present invention reduces the average error value of the fit.

Fig. 5 and fig. 12 are simulation fitting results of the gate capacitance_gate voltage (cgg_vg) curve under the same size, the same temperature condition, the same input parameter and the same output parameter, respectively, as compared with the prior art. It can be seen intuitively that the average error RMS of the fit in the present invention is 0.69%, and the average error RMS of the fit in the prior art is 0.87%, and the present invention reduces the average error value of the fit.

Fig. 6 and fig. 13 are simulation fitting results of the leakage current_gate voltage (id_vg) curve under the same size, the same temperature condition, the same input parameter and the same output parameter, respectively, as compared with the prior art. It can be seen intuitively that the average error RMS of the fitting in the present invention is 7.65%, and the average error RMS of the fitting in the prior art is 25.42%, and the present invention reduces the average error value of the fitting.

Fig. 7 and fig. 14 are simulation fitting results of the leakage current_gate voltage (id_vg) curve under the same size, the same temperature condition, the same input parameter and the same output parameter, respectively, as compared with the prior art. It can be seen intuitively that the average error RMS of the fitting in the present invention is 1%, and the average error RMS of the fitting in the prior art is 1.88%, and the present invention reduces the average error value of the fitting.

Fig. 8 and 15 are simulation fitting results of the leakage current-leakage voltage (id_vd) curve under the same size, the same temperature condition, the same input parameter and the same output parameter, respectively, as those of the prior art. It can be seen intuitively that the average error RMS of the fitting in the present invention is 20.41%, and the average error RMS of the fitting in the prior art is 28.07%, and the present invention reduces the average error value of the fitting.

Steps 104 and 105 in the present invention may be processed in parallel or sequentially, and the present invention is not particularly limited thereto.

Compared with the prior art, the GaN parameter extraction method has the following advantages:

1) And considering the influence of the temperature of the GaN HEMT, grouping the GaN HEMT according to different sizes and different temperatures, and respectively extracting parameters, thereby improving the accuracy of parameter extraction.

5) The model with the parameters extracted can be stored and applied to devices with the same size and different temperatures, so that the effect of extracting the parameters of the whole model can be achieved by only adjusting part of parameters and curves for the rest devices, the time for extracting the parameters is greatly saved, the efficiency of extracting the parameters is increased, and the method has strong engineering practical significance.

The invention also provides a GaN HEMT parameter extraction device based on the ASM model, which can be realized by a general computer or a computer system, wherein the computer or the computer system is provided with an input device, a display device, an external I/F, a communication I/F, a processor and a memory. These respective hardware are connected to each other in such a manner as to be able to communicate via a bus.

The processor is a processing component such as a CPU (Central processing Unit) in the computer or a special CPU, a DSP (digital signal processor) and the like. The Memory device includes a computer readable storage medium such as a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic or optical disk, and the like.

The memory stores a program running on the processor, and the processor executes the parameter extraction method of the GaN HEMT based on the ASM model when running the program.

The invention also provides a readable storage medium, on which a program or an instruction is stored, which when executed by a processor, implements each process of the above-mentioned parameter extraction method for GaN HEMT based on ASM model, and can achieve the same technical effect, and in order to avoid repetition, the description is omitted here.

The embodiments of the present application have been described above with reference to the accompanying drawings, but the present application is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those having ordinary skill in the art without departing from the spirit of the present application and the scope of the claims, which are to be protected by the present application.

Claims

1. The parameter extraction method of the GaN HEMT based on the ASM model is characterized by comprising the following steps of:

setting technological parameters of the GaN HEMT, wherein the technological parameters comprise one or more of size parameters, functional parameters and working environment parameters;

For each group of GaN HEMT devices, the following processing is performed:

Selecting a1 st GaN HEMT device with a1 st size S ₁ and a1 st temperature T1, adjusting different input parameter values to obtain a plurality of measurement data output by the 1 st GaN HEMT device, and generating a measurement curve based on the measurement data, wherein the measurement curve comprises a capacitance-voltage relation curve;

extracting relevant parameters of a capacitance-voltage relation curve;

storing an ASM model of the 1 st GaN HEMT device;

And applying the ASM model to GaN HEMT devices with the same 1 st size S ₁ and other temperatures, and extracting relevant parameters of a capacitance-voltage relation curve.

2. The method for extracting parameters of a GaN HEMT based on an ASM model according to claim 1, wherein,

The measurement profile also includes a current-voltage relationship,

The parameter extraction method further comprises the step of extracting relevant parameters of the current-voltage relation curve.

3. The method for extracting parameters of a GaN HEMT based on an ASM model according to claim 1, wherein,

The technological parameters comprise: the device gate length L, the device gate width W, the working temperature T, the device gate index NF, the device gate source length LSG and the device drain source length LDG.

4. The method for extracting parameters of a GaN HEMT based on an ASM model according to claim 1, wherein,

Extracting the relevant parameters of the capacitance-voltage relation curve comprises the following steps: relevant parameters are extracted through the inverse conducting capacitance-drain voltage curve, the output capacitance-drain voltage curve, the input capacitance-drain voltage curve and the grid capacitance-grid voltage curve.

5. The method for extracting parameters of a GaN HEMT based on an ASM model according to claim 2, wherein,

Extracting relevant parameters of the current-voltage relation curve comprises locally optimizing the parameters, and then globally optimizing the parameters, wherein the locally optimizing comprises the following steps:

2) Extracting a low-field mobility parameter u0, a mobility attenuation coefficient ua and a mobility second-order attenuation coefficient ub from a low-source drain voltage Vds high-current region for a drain current-gate voltage curve id_vg;

6. The method for extracting parameters of GaN HEMT based on ASM model according to claim 5, wherein said extracting parameters are locally optimized and globally optimized based on an optimizer using Trust-region algorithm Trust-region, said optimizer searching for optimal values in a multi-dimensional search directly in a region.

7. The method for extracting parameters of a GaN HEMT based on ASM model as recited in claim 6, wherein,

In the optimization process, when the optimized parameters reach the boundary values, the optimizer automatically amplifies the boundaries.

8. The method for extracting parameters of a GaN HEMT based on ASM model as recited in claim 6, wherein,

And (3) customizing an optimization interval extracted by parameters by using a script, and selecting a region needing to extract the parameters on a measurement graph by calling a function for designating the range of the optimization interval in the script.

9. The method for extracting parameters of a GaN HEMT based on an ASM model according to claim 2, wherein,

Extracting the capacitance-voltage curve parameter includes:

extracting drain-source capacitance parameters, edge capacitance parameters, leakage saturation voltage parameters and bias voltage parameters from the reverse conduction capacitance-drain voltage crss-vds curve;

Extracting drain-source capacitance parameters and drain-edge capacitance parameters from an output capacitance-drain voltage coss-vds curve;

Extracting a gate-source overlap capacitance parameter from an input capacitance-drain voltage ciss-vds curve;

and extracting an AlGaN layer thickness parameter from the grid capacitance-grid voltage cgg-vgs curve.

10. An ASM model-based GaN HEMT parameter extraction apparatus comprising a memory and a processor, wherein the memory stores a program running on the processor, and the processor executes the ASM model-based GaN HEMT parameter extraction method of any one of claims 1 to 9 when the processor runs the program.

11. A computer readable storage medium having stored thereon computer instructions, which when executed by a processor perform the parameter extraction method of the ASM model-based GaN HEMT of any of claims 1-9.