Insulated gate bipolar transistor physical model

Abstract

An IGBT physical model parameter extraction method includes obtaining an initial value and a transformation range of an IGBT physical model parameter; and correcting a model parameter by means of a correspondence between IGBT dynamic and static features and the IGBT physical model parameter and in combination with an experiment measurement result of the IGBT physical model parameter.

Description

BACKGROUND

The disclosure relates to the field of modeling and reliability of power electronic devices, and more particularly to an IGBT physical model parameter extraction method.

For an IGBT (Insulated Gate Bipolar Transistor) with a certain structure, the internal parameters of the components have a decisive influence on its performance. These parameters include structure size, doping concentration, lifetime of excess carrier, junction capacitance, transconductance, etc., and they will directly affect various dynamic and static performance indicators of the IGBT, such as on-state voltage drop, switching speed, turn-off tail current. Therefore, to establish an accurate IGBT semiconductor physical model and realize an accurate simulation of the electrical characteristics of the IGBT, an accurate extraction of parameters is indispensable. At the same time, the model parameters are also important for the design and manufacturing of the IGBT, the structure and performance optimization, and the safe use of the components.

A semiconductor physical model is a simulation model that characterizes the electrical characteristics of IGBT components. Since the semiconductor physical model can achieve a compromise between the simulation accuracy and the simulation efficiency, it has become a widely used IGBT model. In order to continuously improve the overall performance of the products and obtain the optimization of various performance indicators, the component manufacturers usually optimize each parameter during the manufacturing process of the IGBT, such as changing the lifetime of excess carrier of the IGBT by neutron irradiation, and improving the switching performance through buffering layer technology. IGBTs produced by different manufacturers are similar in structure, and their performance often depends on specific internal process parameters.

At present, there are two methods for extracting the IGBT physical model parameters. One is estimating the parameters by combing the directly measured electrical parameters of three ports of the IGBT with some empirical formulas. The other is indirectly extracting the internal physical parameters of the IGBT through designed circuits. The former has a large error, so that it is difficult to meet the accuracy requirements. The latter has tedious extraction steps and it is affected by the accuracy of extraction circuit, so that the extraction of the parameters is difficult to meet the requirements of the simulation accuracy, and the operability is poor.

SUMMARY

The purpose of the disclosure is to provide an IGBT physical model parameter extraction method aiming at defects in the prior art, so that the IGBT physical model parameter extraction method can be greatly simplified while the requirement of parameter extraction on model simulation precision is ensured, and the practicability of an IGBT physical model is improved.

The disclosure provides an IGBT physical model parameter extraction method, the method comprising: obtaining an initial value and a transformation range of an IGBT physical model parameter; and correcting a model parameter according to the correspondence between IGBT dynamic and static features and the IGBT physical model parameter and in combination with an experiment measurement result of the IGBT physical model parameter.

In a class of this embodiment, the dynamic and static features of the IGBT module under a typical working condition are obtained from a data handbook. A typical value of the relevant parameter of the IGBT physical model is obtained by the theoretical calculations. A reasonable variation range of the typical value of the IGBT physical model parameters is obtained based on the semiconductor physical mechanism.

In a class of this embodiment, the correspondence between the IGBT physical model parameter and the IGBT dynamic and static features is determined by analyzing the influence law and degree of the IGBT physical model parameter and the temperature on the IGBT dynamic and static features.

In a class of this embodiment, in combination with the experiment measurement result of the IGBT physical model parameter, the IGBT physical model parameter is first corrected according to the dynamic and static features that are only affected by a single parameter, and then the IGBT physical model parameter is sequentially corrected starting from a dominant parameters according to the dynamic and static features affected by a coupling of multiple IGBT physical model parameters, and finally a temperature coefficient of the IGBT physical model parameter is corrected according to the variation of the dynamic and static features under different temperatures. At last, a correction result of the IGBT physical model parameter under different temperatures is obtained and the extraction of the model parameter is finished.

In a class of this embodiment, on the basis of the initial value and reasonable range of the IGBT physical model parameters, the model parameter is corrected at a temperature of 25° C. according to the influence trend and degree of the IGBT physical model parameter on IGBT switching transient characteristics. According to an approximate linear relationship between the IGBT transient characteristics and the temperature, a temperature empirical formula inside the model are corrected at a temperature of 125° C., so that the IGBT physical model parameter can accurately represents the IGBT switching transient characteristics within an entire temperature range.

In a class of this embodiment, the method for calculating the initial values and reasonable ranges of the IGBT physical model parameter is as follows:

(1) Base Doping Concentration N_L:

$\begin{array}{cc}{N}_L=1.932\times {10}^{18}{V}_B^{-1.4};& \mathrm{symmetrical}\mathrm{structure}(\mathrm{1-a)}\end{array}$ $\begin{array}{cc}{V}_B={W_L\left[\frac{2{\mathrm{qN}}_L\left(8\times {10}^3{N}_L^{-0.75}\right)}{{\epsilon }_{\mathrm{Si}}{\epsilon }_0}\right]}^{\frac{1}{2}}-\frac{{\mathrm{qN}}_L{W}_L^2}{2{\epsilon }_{\mathrm{Si}}{\epsilon }_0}.& \mathrm{asymmetrical}\mathrm{structure}\left(\mathrm{1-b}\right)\end{array}$

In the above formulas, V_B represents a forward breakdown voltage of a collector-emitter of IGBT, W_L represents a base width, q represents an electron charge constant, ε_si represents a relative permittivity of silicon, and ε₀ represents a permittivity of air.

(2) Base Width W_L:

$\begin{array}{cc}{W}_L=2.67\times {10}^{10}·N_L^{\frac{7}{8}}.& \left(2\right)\end{array}$

(3) Lifetime of Excess Carrier in Base Region τ_L:

$\begin{array}{cc}{\tau }_L=\frac{t_{\mathrm{off}}}{\ln \left[10{\alpha }_{pnp}\right]}.& \left(3\right)\end{array}$ $\begin{array}{cc}{\alpha }_{pnp}=\sec h\left(\frac{W_L}{L_L}\right).& \left(4\right)\end{array}$

In the above formulas, α_pnp represents a common base current amplification factor of parasitic PNP transistor in the IGBT, t_off represents a current turn-off time of IGBT, L_L represents a hole diffusion coefficient of the base region.

(4) Trans-Conductance Coefficient K_p:

$\begin{array}{cc}{K}_p=\frac{2\times I_{MOS}^{sat}}{{\left(V_{GS}-V_{th}\right)}^2};& \left(5\right)\end{array}$ $\begin{array}{cc}{I}_{MOS}^{sat}=\frac{I_{CE}^{sat}}{1+{\beta }_{ss}};& \left(6\right)\end{array}$ $\begin{array}{cc}{\beta }_{ss}=\frac{\frac{1}{b}{P}_0+Q_1\left(\frac{1}{b}+\frac{1}{\cos h\left(\frac{W\left(t\right)}{L_L}\right)}\right)}{P_0+Q_1\left(1-\frac{1}{\cos h\left(\frac{W\left(t\right)}{L_L}\right)}\right)}.& \left(7\right)\end{array}$

In the above formulas, P₀ represents a concentration of the excess carrier of the base region near an edge of a collector PN junction, Q₁ represents an equivalent carrier charge of the base region, I_MOS^sat represents a saturation current of MOSFET, V_GS represents a gate-emitter voltage of IGBT, V_th represents a threshold voltage of IGBT, I_CE^sat represents an on-state current of IGBT, β_ss represents an equivalent current amplification coefficient of IGBT, W(t) represents a quasi-neutral base width, b represents a ratio of electron mobility to hole mobility, and L_L represents a hole diffusion coefficient of the base region; K_p can be obtained by putting V_GS, V_th and I_CE^sat into the formulas, I_CE^sat and the corresponding V_GS can be obtained from a device data handbook.

(5) Gate Oxide Layer Capacitance C_OXD:

According to a junction capacitance expression, an initial value of a gate-collector capacitance C_GDJ can be estimated firstly, and then an initial value of the gate oxide layer capacitance C_OXD can be obtained through a Miller capacitance C_GD, and the Miller capacitance C_GD can be obtained from the data handbook.

(6) Gate-Emitter Capacitance C_GS:

The gate-emitter capacitance C_GS can be approximately equal to the difference between an input capacitance C_ies of IGBT and a feedback capacitance C_res of IGBT. Therefore, an initial value and the reasonable order of magnitude of the gate-emitter capacitance C_GS can be calculated by the following formula (8), wherein the feedback capacitance C_res and the input capacitance C_ies can be directly obtained from the data handbook.C_GS=C_ies−C_res  (8).

(7) Other Model Parameters:

An initial value and reasonable correction range of the threshold voltage V_th can be obtained from the data handbook. A structure size parameter can be obtained by physical measurement. According to the information disclosed by an IGBT manufacturer, a reference concentration of a buffer layer N_H, the lifetime of the excess carrier of the buffer layer τ_H and a reasonable orders of magnitude of a buffer layer width W_H can be obtained. A lifetime of minority carrier in the buffer layer of IGBT is determined by a doping concentration of the carrier. When a dispersion range of a semiconductor process parameter is put into a calculation formula of an initial value of the parameter, the variation range of the initial value of the model parameter can be obtained, so that the initial value of the model parameter and its reasonable ranges can be determined.

In a class of this embodiment, the model parameters that affect a turn-on delay t_don are the gate-emitter capacitance C_GS, the threshold voltage V_th and the Miller capacitance C_GS. The model parameters that affect a current rise time t_r are the trans-conductance coefficient K_p, the threshold voltage V_th, the gate-emitter capacitance C_GS and the base doping concentration N_L. The model parameters that affect a turn-off delay taw are the gate-emitter capacitance C_GS, the threshold voltage V_th, the trans-conductance coefficient K_p and the gate oxide layer capacitance Coxa. The model parameters that affect a current fall time t_f are the lifetime of excess carrier in the buffer layer τ_H, the lifetime of excess carrier in the base region τ_L, the width of the buffer layer W_H, the base width W_L and the base doping concentration N_L.

In a class of this embodiment, the IGBT physical model parameter related to the temperature can be calculated by the following temperature empirical formulas.

$\begin{array}{cc}\tau \left(T_j\right)=\tau \left(T_0\right)\times {\left(\frac{T_j}{T_0}\right)}^a;& \left(12\right)\end{array}$ $\begin{array}{cc}{V}_{th}\left(T_j\right)=V_{th}\left(T_0\right)-b\times \left(T_j-T_0\right);& \left(13\right)\end{array}$ $\begin{array}{cc}{K}_p\left(T_j\right)=K_p\left(T_0\right)\times {\left(\frac{T_j}{T_0}\right)}^c;& \left(14\right)\end{array}$ $\begin{array}{cc}{I}_{\mathrm{sne}}\left(T_j\right)=\frac{I_{\mathrm{sne}}\left(T_0\right)\times {\left(\frac{T_j}{T_0}\right)}^d}{\exp \left[\left(1/T_j-1/T_0\right)\times e\times {10}^4\right]}.& \left(15\right)\end{array}$

In the above formulas, a, b, c, d, e are the temperature coefficients related to actual operation characteristics, and they are determined according to an actual device. τ(T₀), V_th(T₀), K_p(T₀) and I_sne(T₀) represent the lifetime of excess carrier r, the threshold voltage V_th, the trans-conductance coefficient and an electron saturation current of the emitter at a temperature of T₀, respectively. τ(T_j), V_th(T_j), K_p(T_j) and I_sne(T_j) represent the values of the aforementioned physical parameters at a temperature of T_j, respectively. Generally, To is equal to 25° C. T_j represents an actual operating junction temperature.

In a class of this embodiment, the IGBT dynamic and static features that are least affected by a coupling of the model parameter are preferentially chosen, and then the IGBT physical model parameter that have a greater impact on the IGBT dynamic and static features is preferentially corrected, so that the IGBT switching transient characteristics represented by the model having the corrected parameter are closer to the value recorded in the data handbook at a temperature of 25° C. The IGBT dynamic and static features that are least affected by a coupling of the temperature-related model parameter are preferentially chosen to correct the temperature coefficient.

In a class of this embodiment, the corrected model parameter is excluded. For the rest IGBT dynamic and static features to be adjusted, the characteristics that have the least coupling relationship with the parameter are chosen to perform the next parameter correction. When the IGBT dynamic and static features meet the requirements within a reasonable range for the first time, the correction of the IGBT physical model parameter is terminated immediately.

In this invention, the initial value of the IGBT physical model parameter and its reasonable correction range is firstly obtained based on the information provided by the data handbook and the typical calculations. Then, the parameter is corrected based on the influence law and degree of each parameter on the IGBT switching transient characteristics. Finally, simulation and experimental verification are conducted on the IGBT physical model based on the corrected parameter. The result shows that the physical model parameter obtained by this method can accurately represent the IGBT switching transient characteristics. Because only the device data handbook and the model calculation are used in this method and a complicated circuit extraction method is not required, this invention significantly reduces the difficulties of the extraction of the IGBT physical model parameter and improves the practicability of the IGBT physical model.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a structure diagram of an IGBT with a buffer layer;

FIG. 2 is a waveform diagram of the IGBT during a switching process;

FIG. 3 shows a variation rule of t_don with V_th under different temperatures;

FIG. 4 shows a variation rule of t_r with K_p under different temperatures;

FIG. 5 shows a variation rule of t_r with V_th under different temperatures;

FIG. 6 shows a variation rule of t_doff with V_th under different temperatures;

FIG. 7 shows a variation rule of t_f with τ_L under different temperatures;

FIG. 8 shows a variation rule of t_f with τ_H under different temperatures;

FIG. 9 shows a variation rule of t_r with T_j;

FIG. 10 shows a variation rule of t_f with T_j;

FIG. 11 shows a variation rule of E_on with T_j;

FIG. 12 shows a variation rule of E_off with T_j;

FIG. 13 shows an IGBT double pulse simulation test circuit; and

FIG. 14 is a flow diagram of an IGBT physical model parameter extraction method.

DETAILED DESCRIPTION

Specific embodiments of the invention are further described in details combining with the drawings hereinbelow.

This invention provides an IGBT physical model parameter extraction method, characterized by comprising the following steps of: obtaining an initial value and a transformation range of an IGBT physical model parameter; and correcting a model parameter according to the correspondence between IGBT dynamic and static features and the IGBT physical model parameter and in combination with an experiment measurement result of the IGBT physical model parameter. The method specially comprises the following steps.

1. Determination of the Initial Value and Reasonable Range of the IGBT Physical Model Parameter

FIG. 1 shows a typical structure diagram of the IGBT. The physical model parameter mainly includes the parameter directly related to the semiconductor process, such as V_th, C_GS, C_OXD, K_p, N_L, W_L, W_H, τ_L, τ_H. In the process of model parameter correction, if the value of the parameter exceeds the reasonable range, the model will lose its physical meaning. Therefore, the initial value of the parameter and its reasonable correction range should be determined firstly before the parameter correction. The specific calculation method is as follows.

(1) Base Doping Concentration N_L

According to the semiconductor theory, a depletion layer mainly extends in N-base region during forward blocking, and the base doping concentration plays a key role in the forward blocking performance of the IGBT. Therefore, the initial value of N_L and its reasonable order of magnitude can be estimated according to the formula (1), wherein V_B represents the breakdown voltage:

$\begin{array}{cc}{N}_L=1.932\times 10^{18}{V}_B^{-1.4};& \mathrm{Symmetrical}\mathrm{structure}\left(1-a\right)\end{array}$ $\begin{array}{cc}{V}_B={W_L\left[\frac{2q{N}_L\left(8\times 10^3{N}_L^{-0.75}\right)}{{\epsilon }_{\mathrm{Si}}{\epsilon }_0}\right]}^{\frac{1}{2}}-\frac{q{N}_L{W}_L^2}{2{\epsilon }_{\mathrm{Si}}{\epsilon }_0}.& \mathrm{Asymmetrical}\mathrm{structure}\left(1-b\right)\end{array}$

(2) Base Width W_L

Since the breakdown voltage is closely related to the base doping concentration and the base width, and the size of the base width depends on the base doping concentration, the initial value of W_L can be calculated by the formula (2) and the formula (1):

$\begin{array}{cc}{W}_L=2.67\times 10^{10}·N_L^{-\frac{7}{8}}.& \left(2\right)\end{array}$

(3) Lifetime of Excess Carrier in Base Region τ_L

The order of magnitude of the lifetime of the excess carrier in the base region can be estimated by t_off and semiconductor theory formulas. t_off can be obtained from the device handbook.

$\begin{array}{cc}[{\tau }_L=\frac{t_{\mathrm{off}}}{\ln \left[10{\alpha }_{\mathrm{pnp}}\right]};& \left(3\right)\end{array}$ $\begin{array}{cc}{\alpha }_{pnp}=\sec h\left(\frac{W_L}{L_{Lp}}\right).& \left(4\right)\end{array}$

In the above formulas, α_pnp represents the common base current amplification factor of parasitic PNP transistor in the IGBT, t_off represents the current turn-off time of IGBT, L_Lp represents the hole diffusion coefficient of the base region.

(4) Trans-Conductance Coefficient K_p

The initial value of K_p can be calculated by combining the following equations:

$\begin{array}{cc}{K}_p=\frac{2\times I_{MOS}^{sat}}{{\left(V_{GS}-V_{th}\right)}^2};& \left(5\right)\end{array}$ $\begin{array}{cc}{I}_{MOS}^{sat}=\frac{I_{CE}^{sat}}{1+{\beta }_{ss}};& \left(6\right)\end{array}$ $\begin{array}{cc}{\beta }_{ss}=\frac{\frac{1}{b}{P}_0+Q_1\left(\frac{1}{b}+\frac{1}{\cos h\left(\frac{W\left(t\right)}{L_L}\right)}\right)}{P_0+Q_1\left(1-\frac{1}{\cos h\left(\frac{W\left(t\right)}{L_L}\right)}\right)}.& \left(7\right)\end{array}$

In the above formulas, P₀ represents the concentration of the excess carrier of the base region near the edge of the collector PN junction, Q₁ represents the equivalent carrier charge of the base region, I_MOS^sat represents the saturation current of MOSFET. K_p can be obtained by putting V_GS, V_th and I_CE^sat into the formulas. I_CE^sat and the corresponding V_GS can be obtained from the device data handbook.

(5) C_OXD

According to the junction capacitance expression, the initial value of the gate-collector capacitance C_GDJ can be estimated firstly, and then the initial value of the gate oxide layer capacitance C_OXD can be obtained through the Miller capacitance C_GD. The Miller capacitance C_GD can be obtained from the data handbook.

(6) C_GS

The gate-emitter capacitance C_GS can be approximately equal to the difference between an input capacitance C_ies of IGBT and the feedback capacitance C_res of IGBT. Therefore, the initial value and the reasonable order of magnitude of the gate-emitter capacitance C_GS can be calculated by the formula (8), wherein C_res and C_ies can be directly obtained from the data handbook.C_GS=C_ies−C_res  (8)

(7) Other Model Parameters

The initial value and reasonable correction range of the threshold voltage V_th can be obtained from the data handbook. The structure size parameter can be obtained by physical measurement.

According to the information disclosed by the IGBT manufacturer, the reasonable order of magnitude of N_H, τ_H and W_H can be obtained. According to the semiconductor theory, the lifetime of minority carrier in the buffer layer of the IGBT is determined by the doping concentration of the carriers. When the dispersion range of the semiconductor process parameter are put into the calculation formulas of the initial value of the parameter, the variation range of the initial value of the model parameter can be obtained, so that initial value of the model parameter and its reasonable range can be determined.

2. Analysis of the Influence Law and Degree of the IGBT Physical Model Parameter on the Characteristics of the IGBT

2.1 Analysis of the Influence Law and Degree of the Model Parameter on the Turn-on Characteristics of the IGBT Under a Constant Temperature

As shown in FIG. 2, the turn-on time of the IGBT can be divided into two parts, namely a turn-on delay time t_don and a turn-on rise time t_r. On basis of the established physical model, a simulation is conducted on the influence of each parameter on t_don and t_r within the reasonable ranges of typical parameter values, wherein the model parameters are used as variables. In the phase of turn-on delay, a drive circuit charges the C_GS, and V_GE(t) gradually increases from lower than V_th to higher than V_th. When V_GE(t)=V_th, the IGBT is turned on and I_CE gradually increases from zero. Therefore, τ_don is related to C_GS, V_th and C_GS.

The variation of the rising rate of I_CE will directly affect t_r. I_CE includes MOSFET channel current I_mos, IGBT collector-emitter hole current I_P and charge-discharge current of internal junction capacitance I_c.I_CE=I_mos+I_P+I_c  (9)

In accordance with the principle of MOSFET, the variations of K_p, V_th and C_GS will have an impact on I_mos. As known from the expression (10) of the hole current of the physical model, the model parameters related to I_P are N_L, τ_L, W_H and W_L.

$\begin{array}{cc}{I}_P=\frac{\left(1+B_b·\frac{3W_H^2}{2D_{pH}{\tau }_L}\right)\frac{I_{\mathrm{CE}}}{1+b}+B_b{Q}_T\left(\frac{6D_L}{W^2\left(t\right)}+\frac{C_{BCJ}\frac{{\mathrm{dV}}_{\mathrm{AA}}}{\mathrm{dt}}}{qAW\left(t\right)N_L}-\frac{1}{{\tau }_L}\right)}{1+B_b\left(\frac{3D_L}{W^2\left(t\right)D_{pH}}+\frac{W_H^2{C}_{\mathrm{BCJ}}\frac{{\mathrm{dV}}_{\mathrm{AA}}}{\mathrm{dt}}}{2qAW\left(t\right)D_{pH}{N}_L}+\frac{W_H^2}{D_{pH}{\tau }_L}\right)}& \left(10\right)\end{array}$

In the above formula, B_b, Q_T, W_H, D_pH and C_BCJ are the semiconductor parameters related to the physical model.

Therefore, the model parameters that affect t_r are K_p, V_th, C_GS and N_L.

2.2 Analysis of the Influence Law and Degree of the Model Parameter on the Turn-Off Characteristics of the IGBT Under a Constant Temperature

As shown in FIG. 2, the turn-off time of the IGBT can be divided into two parts, namely a turn-off delay time taw and a turn-off fall time t_f. It can be known from the analysis of the waveform of the turn-off delay phase that V_GE(t) gradually decreases from V_GG to V_GP with the discharge of C_GS. V_GP is only related to V_th and K_p.

$\begin{array}{cc}{V}_{GP}={\nabla }_{th}+\sqrt{\frac{\left(I_P+I_n\right)}{K_p}}& \left(11\right)\end{array}$

In the phase of [t6-t7], V_GE(t) is in a Miller plateau. Therefore, C_GS, V_th, K_p and C_OXD will have an impact on t_doff.

The turn-off fall phase of the IGBT can be divided into two parts. a) When V_GE(t) drops slightly less than V_th, the conductive channel in the MOSFET disappears and the channel current rapidly drops to zero. b) Although the electron current disappears rapidly, a large number of excess carriers are still left inside the IGBT. The excess carriers will gradually disappear through recombination, which will result in a “tailing” of the turn-off current. As the phase a) is finished in an instant, the effect of this period on t_f is negligible. The slower recombination process of the excess carrier becomes the main factor that affects t_f. According to the semiconductor theory, the greater the carrier recombination rate, the stronger the recombination effect and the smaller t_f. The recombination rate is related to the lifetime and the concentration of the excess carriers. Therefore, the model parameters related to t_f are τ_H, τ_L, W_H, W_L, N_L.

The simulation result of the influence law and degree of the model parameter on the IGBT transient characteristics are shown in FIGS. 3-8 and Table 1.

TABLE 1
The influence law and degree of the IGBT physical model
parameter on the characteristics under switching transient
(The direction of the arrow indicates the law of influence, and
the number of arrows indicates the degree of the influence)
	tdon/μs	tr/μs	tdoff/μs	tf/μs	Eon/mJ	Eoff/mJ
Vth/V↑	↑↑	↑↑↑↑	↓↓↓	—	↑↑↑↑	—
CGS/F↑	↑	↑	↑	—	↑	—
COXD/F↑	—	—	↑↑↑↑	—	—	—
Kp/A/V2↑	—	↓↓↓	↑↑	—	↓↓↓	—
NL/cm−3↑	—	↑↑	—	↓↓↓↓	↑↑	↓↓
WL/cm↑	—	—	—	↓↓↓↓↓	—	↓↓↓↓↓
WH/cm↑	—	—	—	↓↓↓	—	↓↓↓↓
τL/s↑	—	—	—	↑	—	↑
τH/s↑	—	—	—	↑↑	—	↑↑↑

2.3 Analysis of the Influence Law and Degree of the Temperature on the Characteristics of the IGBT

The parameters of the components related to the temperature can be calculated by the following temperature empirical formulas.

$\begin{array}{cc}\tau \left(T_j\right)=\tau \left(T_0\right)\times {\left(\frac{T_j}{T_0}\right)}^a;& \left(12\right)\end{array}$ $\begin{array}{cc}{V}_{th}\left(T_j\right)=V_{th}\left(T_0\right)-b\times \left(T_j-T_0\right);& \left(13\right)\end{array}$ $\begin{array}{cc}{K}_p\left(T_j\right)=K_p\left(T_0\right)\times {\left(\frac{T_j}{T_0}\right)}^c;& \left(14\right)\end{array}$ $\begin{array}{cc}{I}_{\mathrm{sne}}\left(T_j\right)=\frac{I_{\mathrm{sne}}\left(T_0\right)\times {\left(\frac{T_j}{T_0}\right)}^d}{\exp \left[\left(1/T_j-1/T_0\right)\times e\times {10}^4\right]}.& \left(15\right)\end{array}$

In the above formula, the temperature coefficients a, b, c, d, e are positive, and they are related to the actual operation characteristics. τ(T₀), V_th(T₀), K_p(T₀) and I_sne(T₀) represent the lifetime of the excess carrier, the threshold voltage, the trans-conductance coefficient and the electron saturation current of the emitter at a temperature of T₀, respectively. τ(T_j), V_th(T_j), K_p(T_j) and I_sne(T_j) represent the values of the aforementioned physical parameters at a temperature of T_j, respectively. Generally, T₀ is equal to 25° C. T_j represents an operating junction temperature. The influence of the operating junction temperature of the IGBT on the switching transient characteristics of the IGBT can be obtained by combing the relationship between these parameters related to temperature and the variation of the temperature as well as the relationship between the parameters and the IGBT switching transient characteristics. The simulation result is shown in FIGS. 9-12.

3. The Correction Method of the IGBT Physical Model Parameter

Based on the initial value and the reasonable range of the parameter, the model parameter is corrected at a temperature of 25° C. according to the influence trend and degree of the parameter on the IGBT transient characteristics. According to an approximate linear relationship between the IGBT transient characteristics and the temperature, the temperature empirical formula inside the model will be corrected at a temperature of 125° C., so that the IGBT physical model parameters can accurately represent the IGBT switching transient characteristics within an entire temperature range.

(1) Correction of the Model Parameter at a Temperature of 25° C.

Considering the wide variety of the model parameters and their complex coupling effects on the IGBT switching transient characteristics, according to the influence trend and degree of the model parameter on the IGBT transient characteristics, the IGBT characteristics that are least affected by the coupling of the model parameter is preferentially chosen, and then the preferentially corrected parameter is chosen according to the influence degree of the parameter on the characteristics, so that the IGBT switching transient characteristics represented by the model having the corrected parameters are closer to the values recorded in the data handbook at a temperature of 25° C. The correction process is as follows:

i) t_don

As shown in Table 1, in the IGBT switching transient characteristics, the least parameters related to t_don is V_th and C_GS, and C_GS has a greater impact on t_don. Therefore, C_GS is preferred chosen and C_GS will be corrected according to the influence rule of C_GS on t_don that shown in the Table 1. When t_don meets the requirements for the first time, the correction to C_GS is immediately terminated. Otherwise, V_th can be further corrected. The method of the correction is the same as that of C_GS.

ii) t_r

The corrected model parameters are excluded. For the rest IGBT dynamic and static characteristics to be adjusted, the characteristics that have the least coupling relationship with the parameters are chosen to perform the next parameter correction. As shown in the Table 1, only K_p and N_L are left. Considering the requirements on the forward blocking of the IGBT, a large adjustment to N_L should be avoided during the correction, so that K_p is preferred for correction. According to the variation rule of t_r with K_p, when t_r meets the requirements for the first time, the correction of K_p can be terminated. Otherwise, a fine-tuning is performed on N_L to make t_r meet the requirements, and then the correction of N_L can be terminated.

iii) E_on

So far, the debugging of the transient characteristics related to the turn-on time of the IGBT has been finished. If the turn-on time of the IGBT meets the requirements, E_on will also meet the requirements at the same time.

iv) t_doff

Except for the above corrected parameters, among the remaining characteristics to be adjusted, the only parameter that affects t_doff is C_OXD. Thus, the C_OXD is corrected based on the rules shown in the table to make t_doff meet the requirements for the first time. Then, the correction can be terminated.

v) t_f and E_off

The other modifiable parameters have a coupling effect on E_off and t_f at the same time. However, since E_off is not only related to t_f but also closely related to the turn-off tail current, t_f is preferred for debugging. W_H and τ_H have a greater impact on the tail current, so that W_L and τ_L are preferred for correction when debugging t_f.

According to the variation rule of t_f with W_L, W_L is corrected. When t_f meets the requirements for the first time, the correction to W_L can be terminated. Otherwise, τ_L is corrected until t_f meets the requirements. At the same time, if E_off also meets the requirements, the parameter correction is finished. Otherwise, τ_H is corrected. When hand E_off meet the requirements at the same time, the correction is terminated and the correction of the model parameters at a temperature of 25° C. is finished.

(2) Correction of the Model Parameters at a Temperature of 125° C.

Similar to the correction process under a temperature of 25° C., the IGBT characteristics that are least affected by the coupling of the model parameters related to the temperature are preferentially chosen, and the temperature coefficients are corrected. The analysis of the correction process is as follows.

i) t_don

It is preferred to choose the temperature coefficient a that corrects V_th. When t_don meets the requirements for the first time, the correction to the temperature coefficient a will be terminated.

ii) t_r

The temperature coefficient b of the characteristic t_r-K_p that is least affected by the coupling of the temperature-related parameters is chosen to be corrected. When t_r meets the requirements for the first time, the correction to b can be terminated.

iii) t_doff

Since the hole saturation drift velocity ν_sat,p decreases with the increase of the temperature, taw increases with the increase of the temperature. Therefore, t_doff can be adjusted by correcting the temperature coefficient of ν_sat,p.

iv) t_f and E_off

According to the requirements of t_f and E_off, the temperature coefficients of τ_L and τ_H are corrected at the same time so as to complete the correction of the temperature coefficient at a temperature of 125° C.

4. Extraction Method of the IGBT Physical Model Parameter

Based on the above analysis method and conclusion, the IGBT physical model parameters extraction method is as follows. The dynamic and static characteristics of the IGBT module under a typical working condition are obtained from the data handbook. The typical values of the physical model parameters are obtained through theoretical calculations. Then, the reasonable variation ranges of the typical values of the model parameters are obtained based on the semiconductor physical mechanism. Next, the correspondence between the model parameters and the IGBT dynamic and static characteristics is determined by analyzing the influence law and degree of the model parameters and the temperature on the IGBT dynamic and static characteristics. Further, based on the actual test result, the parameter is corrected in term of the characteristic that only affected by a single parameter firstly. Then, in term of the characteristics affected by the multi-parameter coupling, the dominant parameters are corrected in turn. At last, the temperature coefficient of the parameter is corrected by combining the characteristic variations under different temperatures. Finally, the correction results of the model parameters under different temperatures are obtained, and the extraction of the model parameters is realized. The flow chart of the extraction method is shown in FIG. 14.

5. Experimental Verification

The simulation and the experimental verification are performed on Infineon 3300V/1500 A IGBT module. Firstly, based on the established physical model, an IGBT double pulse test simulation is performed on the improved model according to the initial value and the corrected value of the model parameters. The simulation and the test circuit are shown in FIG. 13, wherein L=90 uH, R_gon=0.9Ω, R_goff=2.7Ω, I_CE=1500 A, V_DC=1800V, L_S=40 nH, C_GE=330 nF.

Table 2 shows the values of the model parameters extracted by the method of this invention. Table 3 shows the comparison between the double-pulse simulation result of the IGBT module using the initial values of the model parameters and the double-pulse simulation result of the IGBT module using the extraction values of the present invention. It can be seen that the simulation result of this invention are in good agreement with the actual measurement result, which verifies the correctness of the IGBT physical model parameter extraction method of this invention.

TABLE 2
The initial values and extraction values of the
parameters of Infineon FZ1500R33HL3 IGBT model
		Initial	Extraction value of
	Parameter	value	this invention
	WL/cm	0.044	0.048
	τL/S	5 × 10−6	6.5 × 10−6
	Vth/V	5.8	6
	CGS/F	1.2 × 10−9	2.5 × 10−9
	COXD/F	5 × 10−8	7.8 × 10−8
	τH/S	5 × 10−7	3 × 10−7
	Kp/A/V2	3.0	5.0
	a(τL)	1.5	1.0
	a(τH)	1.5	1.0
	C	−1.5	−1.0

TABLE 3
Comparison of the parameter extraction effect of Infineon
FZ1500R33HL3 IGBT module physical model
	Junction Temperature
	25° C.	125° C.
		Simulation			Simulation
		result			result
	Simulation	based on		Simulation	based on
	result based	the extraction		result based	the extraction
Comparison	on the	value of this	Test	on the	value of this	Test
item	initial value	invention	value	initial value	invention	value
tdon/μs	0.21	0.52	0.5	0.55	0.52	0.5
tr/μs	0.2	0.54	0.55	0.8	0.57	0.55
tdoff/μs	2.35	3.7	4.1	3.128	4.345	4.3
tf/μs	0.21	0.36	0.4	0.48	0.36	0.4
Eon/mJ	1222.8	2389.3	2300	1617.1	3264.7	3200
Eoff/mJ	1146.6	2579.3	2400	1923.1	3100	2950

It will be obvious to those skilled in the art that changes and modifications may be made, and therefore, the aim in the appended claims is to cover all such changes and modifications.

Claims

1. An insulated gate bipolar transistor (IGBT) physical model parameter extraction method, comprising: obtaining an initial value and a transformation range of an IGBT physical model parameter; and correcting a model parameter according to a correspondence between IGBT dynamic and static features and the IGBT physical model parameter and in combination with an experiment measurement result of the IGBT physical model parameter; wherein: the IGBT dynamic and static features under a typical working condition are obtained from a data handbook; a typical value of the relevant parameter of an IGBT physical model is obtained by a theoretical calculation; and a reasonable variation range of the typical value of the IGBT physical model parameter is obtained according to a semiconductor physical mechanism;the calculation of the initial value and reasonable range of the IGBT physical model parameter comprises:(1) base doping concentration NL:

2. The method of claim 1, wherein the correspondence between the IGBT physical model parameter and the IGBT dynamic and static features is determined by analyzing an influence law and degree of the IGBT physical model parameter and a temperature on the IGBT dynamic and static features.

3. The method of claim 2, wherein the IGBT physical model parameter related to temperature is calculated by the following temperature empirical formulas:

4. The method of claim 1, wherein in combination with the experiment measurement result of the IGBT physical model parameter, the IGBT physical model parameter is first corrected according to the dynamic and static features that are only affected by a single parameter, and then the IGBT physical model parameter is sequentially corrected starting from a dominant parameter according to the dynamic and static features affected by a coupling of multiple IGBT physical model parameters, and a temperature coefficient of the IGBT physical model parameter is corrected according to the variation of the dynamic and static features under different temperatures, and at last, a correction result of the IGBT physical model parameter under different temperatures is obtained and the extraction of the model parameter is finished.

5. The method of claim 4, wherein on the basis of the initial value and reasonable range of the IGBT physical model parameter, the model parameter is corrected at a temperature of 25° C. according to the influence trend and degree of the IGBT physical model parameter on IGBT switching transient characteristics; according to an approximate linear relationship between the IGBT transient characteristics and the temperature, a temperature empirical formula inside the model is corrected at a temperature of 125° C., so that the IGBT physical model parameter accurately represents the IGBT switching transient characteristics within an entire temperature range.

6. The method of claim 5, wherein the IGBT dynamic and static features that are least affected by a coupling of the model parameter are preferentially chosen, and then the IGBT physical model parameter that have a greater impact on the IGBT dynamic and static features is preferentially corrected, so that the IGBT switching transient characteristics represented by the model having the corrected parameter are closer to the value recorded in the data handbook at a temperature of 25° C.; the IGBT dynamic and static features that are least affected by a coupling of the temperature-related model parameter are preferentially chosen to correct the temperature coefficient.

7. The method of claim 6, wherein the corrected model parameter is excluded, and for the rest IGBT dynamic and static features to be adjusted, the characteristics that have the least coupling relationship with the parameter are chosen to perform the next parameter correction; when the IGBT dynamic and static features meet the requirements within a reasonable range for the first time, the correction of the IGBT physical model parameter is terminated immediately.

8. The method of claim 1, wherein the model parameters that affect a turn-on delay tdon are the gate-emitter capacitance CGS, the threshold voltage Vth and the Miller capacitance CGD; the model parameters that affect a current rise time tr are the trans-conductance coefficient Kp, the threshold voltage Vth, the gate-emitter capacitance CGS and the base doping concentration NL; the model parameters that affect a turn-off delay tdoff are the gate-emitter capacitance CGS, the threshold voltage Vth, the trans-conductance coefficient Kp and the gate oxide layer capacitance COXD; the model parameters that affect a current fall time tf are the lifetime of excess carrier in the buffer layer τH, the lifetime of excess carrier in the base region τL, the width of the buffer layer WH, the base width WL and the base doping concentration NL.