CN104573330B - The extracting method of GaN high electron mobility transistor I V model parameters - Google Patents

Abstract

The invention discloses Current Voltage (I V) model parameter extraction method of one kind based on GaN high electron mobility transistor (GaN HEMT).The inventive method carries out piecemeal to model parameter first according to the physical significance of I V model parameters, reduces the complexity of I V models, then by being fitted the transfer characteristic curve of actual measurement pulse I V and static state I V, extract the model parameter of corresponding sub-block.Compared with traditional parameter extracting method, the inventive method can rapidly and accurately complete the parameter extraction of I V models, substantially increase the efficiency of organs weight.

Description

The extracting method of GaN high electron mobility transistor I-V model parameters

Technical field

The invention belongs to power device field, more particularly to based on GaN high electron mobility transistor (GaN HEMT) I-V model parameter extraction methods.

Background technology

GaN high electron mobility transistor (GaN HEMT) due to characteristics such as its high frequency, high powers, in microwave circuit In application it is more and more extensive.Because GaN HEMT need to be worked under high temperature, high power conditions, therefore large-signal model is to use GaN HEMT carry out the basis of microwave circuits.

Current-voltage model be I-V models be large-signal model core, in order to characterize GaN HEMT in high-power condition Self-heating effect and trap effect during lower work, the I-V models of GaN HEMT have more than the I-V models of the devices such as Si, GaAs Many model parameters, model parameter generally has tens of to more than 100.

The parameter extraction of I-V models is the basis of organs weight.The I-V models of the devices such as Si, GaAs are due to number of parameters It is few, parameter of the part with physical significance generally is calculated by surveying I-V data, then it is remaining using the optimization of the software debuggings such as ADS Parameter fitting actual measurement I-V curve is that can obtain all model parameters.I-V models for GaN HEMT, because number of parameters is non- Chang Duo, the huge and simple numerical optimization of traditional parameter extracting method workload is readily obtained the parameter for running counter to physical significance Value, cannot meet the requirement of high efficiency modeling.

For GaN HEMT device I-V model parameter extraction methods research in the scientific paper delivered both at home and abroad simultaneously Rare, such as G.Avolio of Leuven universities et al. proposed low using numerical optimization fitting in multiple operating points in 2013 The big time domain plethysmographic signal of frequency, the parameter in I-V models is obtained with this.The method that the document is proposed needs expensive test equipment (such as large signal network analysis instrument), realize difficulty it is larger (G.Avolio, A.Raffo, I.Angelov, G.Crupi, G.Vannini,and D.Schreurs,”A Novel Technique for the Extraction of Nonlinear Model for Microwave Transistors Under Dynamic-bias Operation”,IEEE MTT-S International Microwave Symposium,pp.1-3,June 2013)。

I.Angelov of Chalmers universities et al. proposed to first pass through DC I-V curve in specific region in 2013 Slope extracts part has the parameter of physical significance, then by remaining parameter of numerical optimization fitting low-frequency time-domain waveform extracting, But the I-V model forms that the document is used are relatively simple, just it is difficult to using (I.Angelov when number of parameters rolls up et al,”Hybrid Measurement-based Extraction of Consistent Large-Signal Models for Microwave FETs”,European Microwave Conference,pp.267-270,2013)。

Kelvin S.Yuk of California universities et al. were proposed in 2009 on the basis of Angelov et al. work A kind of improved I-V models, the model can reflect in the trap effect and self-heating effect, but text of GaN device not exactly Have the I-V models for providing specific and complete parameter extracting method (KS Yuk, GR Branner, DJ McQuate. " A Wideband Multiharmonic Empirical Large-Signal Model for High-Power GaN HEMTs With Self-Heating and Charge-Trapping Effects”,IEEE Trans.Microwave Theory Tech,2009,vol.57,no.12,pp.3322-3332,2009.)。

The content of the invention

In order to solve the problems, such as above-mentioned background technology, the present invention proposes a kind of I-V model parameters of GaN HEMT Extracting method, and the programming realization in Matlab, can rapidly and accurately obtain all parameter values in I-V models, carry significantly The efficiency of GaN HEMT devices modeling high.

It is an object of the invention to overcome the deficiencies in the prior art, there is provided a kind of parameter of GaN HEMT devices I-V models Extracting method, enables GaN HEMT large signal equivalent circuits models more accurately to reflect that the electrology characteristic and microwave of device are special Property.

The present invention carries out parameter extraction, the correlative detail of model using the modified Angelov I-V models of GaN HEMT It can be seen that paper (KS Yuk, GR Branner, DJ that Kelvin S.Yuk of California universities et al. were delivered in 2009 McQuate.”A Wideband Multiharmonic Empirical Large-Signal Model for High-Power GaN HEMTs With Self-Heating and Charge-Trapping Effects”,IEEE Trans.Microwave Theory Tech,2009,vol.57,no.12,pp.3322-3332,2009.)。

The present invention is specifically adopted the following technical scheme that：

A kind of extracting method of GaN high electron mobility transistor I-V model parameters, its flow as shown in figure 1, its It is middle to use fitting device in each drain voltage V_dsUnder I-V transfer characteristic curves, and non-traditional to propose the fitting output of ginseng method special Linearity curve, this causes that the independent variable in fit procedure only has grid voltage V_gs, independent variable number is reduced, reduce fitting difficulty And fitting precision is improve, following steps are specifically included：

Step 1. parameter piecemeal：The concrete form after piecemeal is carried out to I-V model parameters as follows：

I_ds=I_pkth(1+M_ipk(V_ds,V_gseff)·tanh(Ψ(V_ds,V_gseff)))·tanh(αV_ds) (1)

M_ipk(V_ds,V_gseff)=1+0.5 (M_ipkbth(V_ds)-1)·(1+tanh(Q_m(V_ds)·(V_gseff-V_gsm))) (2)

Ψ(V_ds,V_gseff)=P_1,th(V_ds)·(V_gseff-V_1,pk)+P_2,th(V_ds)·(V_gseff-V_2,pk)²+P_3,th· (V_gseff-V_3,pk)³ (3)

I_pkth=I_pk·(1+K_Ipk(V_ds)·ΔT) (4)

M_ipkbth(V_ds)=M_ipkb(V_ds)·(1+K_Mipkb(V_ds)·ΔT) (5)

P_j,th(V_ds)=P_j(V_ds)·(1+K_j,P(V_ds) Δ T), j=1,2,3 (6)

V_gseff=f (V_gs,V_ds,V_dsq,V_gsq) (7)

Δ T=P_diss·R_theq=I_ds·V_ds·R_theq (8)

Wherein, I_dsIt is dram-source voltage；I_pkthIt is electric current when device transconductance is maximum；M_ipk(V_ds,V_gseff) it is I_pkth Tanh multiplier, for characterizing the asymmetric property of GaN HEMT mutual conductances；V_gseff(V_gs,V_ds,V_dsq,V_gsq) sunken to characterize The equivalent grid voltage of trap effect, it is dram-source voltage V_ds, grid-source voltage V_gs, dram-source voltage it is static partially Put point V_dsq, grid-source voltage quiescent bias point V_gsqFunction, its expression can according to demand choose appropriate shape Formula；Ψ(V_ds,V_gseff) it is multinomial series centered on grid voltage by mutual conductance when maximum；α is saturation voltage parameter；I_pk It is I_pkthSelf-heating effect independent component；K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds) it is sign self-heating The continuous item of effect, this five are dram-source voltage V_dsFunction, its expression can according to demand choose appropriate Form；Δ T is device channel temperature increment, and it is equal to dissipated power P_dissWith equivalent thermal resistance R_theqProduct；M_ipkbth(V_ds) and Q_m(V_ds) it is M_ipk(V_ds,V_gseff) intermediate variable inside expression formula, wherein M_ipkb(V_ds) it is M_ipkbth(V_ds) self-heating effect without Close component；P_j,th(V_ds) it is Ψ (V_ds,V_gseff) polynomial coefficient, wherein, j=1,2,3；P_j(V_ds) it is P_j,th(V_ds) from Fuel factor independent component；V_gsm、V_1,pk、V_2,pk、V_3,pkIt is the grid voltage parameter of correspondence mutual conductance peak value；

Step 2. extracts the model parameter unrelated with self-heating effect and trap effect：For the crystal for needing extracting parameter Pipe, in its source ground, dram-source voltage quiescent bias point V_dsq=0V, grid-source voltage quiescent bias point V_gsq=- Under the conditions of the pulse test of 3V, obtain the current-voltage i.e. transfer characteristic curve of I-V of device, and now device from thermal effect Should be negligible with trap effect relevant parameter, pulse I-V transfer characteristic curves and step 1 as curve matching obtained by Described equation (1), (2), (3) are fitted, and obtain unrelated with self-heating effect and trap effect in equation (1), (2), (3) Model parameter：I_pk, α, V_gsm, V_1,pk, V_2,pk, V_3,pk, M_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds)；

Step 3. extracts the related model parameter of self-heating effect：Static I-V tests are done to transistor, and utilizes Function Fitting Method obtains dissipated power P_dissWith V_dsAnd V_gsIt is the expression of independent variable；Enter horizontal pulse I-V tests, gained to transistor Data can extract out the equivalent thermal resistance R of device_theq, bring the expression that formula (8) can obtain channel temperature increment Delta T into；

The static I-V data that this step is measured can obtain static I_ds-V_gsTransfer characteristic curve, using curve-fitting method handle Equation (4), (5), (6) and the static I_ds-V_gsTransfer characteristic curve is fitted, and obtains self-heating in equation (4), (5), (6) The related model parameter of effect：K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds)；

Step 4. extracts the related model parameter of trap effect：For the transistor for needing extracting parameter, connect in its source electrode Ground, dram-source voltage quiescent bias point V_dsq=0V, grid-source voltage quiescent bias point V_gsqCarried out under conditions of=- 3V Pulse I-V tests to obtain pulse I-V curve, and in its source ground, dram-source voltage quiescent bias point V_dsq=0V, grid- Source voltage quiescent bias point V_gsqEnter horizontal pulse I-V under conditions of=0V and test to obtain pulse I-V curve；With reference to step 2 and step Trap effect relevant parameter V is removed in 3 equation tried to achieve (1)~(8)_gseffOuter all parameters, using curve-fitting method handle Two group pulse I-V curves obtained by this step are fitted with equation (7), you can obtain with V_gs、V_ds、V_dsq、V_gsqThis four ginsengs It is the related model parameter V of the trap effect of variable to measure_gseffExpression, finally realize all I-V moulds of test transistor The extraction of shape parameter.

The beneficial effects of the invention are as follows：

First, parameter extracting method of the invention employs the thought of piecemeal extraction, and piecemeal is carried out by model parameter, Both the complexity of I-V models had been reduced, the Physical Mechanism of device work had been met again so that extracting the model parameter value for obtaining had had Correct physical significance, can accurately reflect the working condition of device；

Second, parameter extracting method of the invention is using fitting device in each drain voltage V_dsUnder I-V transfer characteristics Curve, and unorthodox method fitting output characteristic curve, this causes that the independent variable in fit procedure only has grid voltage V_gs, reduce Independent variable number, reduces fitting difficulty and improves fitting precision；

3rd, parameter extracting method of the invention can be used for the I-V models of multi-form, widely applicable, portable good, Disclosure satisfy that engineers and technicians change I-V model equations but do not change the requirement of parameter extracting method according to actual needs；

4th, parameter extracting method of the invention can in Matlab programming realization, program once runs and can obtain I-V All parameter values of model, compared with traditional parameters extracting method, greatly reduce workload, significantly improve organs weight effect Rate.

Brief description of the drawings

Fig. 1 is the inventive method I-V model parameter extraction flow charts.

Fig. 2 is the flow chart for extracting self-heating effect and trap effect independent parameter.

Fig. 3 is in each dram-source voltage V_dsUnder with dram-source voltage V_gsIt is abscissa, drain current I_dsIt is vertical The I of coordinate_ds-V_gsCurve.

Fig. 4 is function M_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) fitting effect.

Fig. 5 is the flow chart for extracting self-heating effect relevant parameter.

Fig. 6 is dissipated power P_dissFitting of a polynomial design sketch.

Fig. 7 is function K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds) fitting effect.

Fig. 8 is to extract to obtain the static I-V curve fitting result chart after all I-V model parameters.

Specific embodiment

Below with the parameter M of materialization_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds), K_Ipk(V_ds), K_Mipkb (V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds), V_gseffV=f (V_gs,V_ds,V_dsq,V_gsq) expression formula as a example by, and with reference to accompanying drawing The present invention is described in further detail.

The specific steps of the GaN high electron mobility transistor I-V model parameter extraction methods that the present invention is provided are such as Under：

Step 1. parameter piecemeal：

The concrete form after piecemeal is carried out to I-V model parameters as follows：

I_ds=I_pkth(1+M_ipk(V_ds,V_gseff)·tanh(Ψ(V_ds,V_gseff)))·tanh(αV_ds) (1)

M_ipk(V_ds,V_gseff)=1+0.5 (M_ipkbth(V_ds)-1)·(1+tanh(Q_m(V_ds)·(V_gseff-V_gsm))) (2)

Ψ(V_ds,V_gseff)=P_1,th(V_ds)·(V_gseff-V_1,pk)+P_2,th(V_ds)·(V_gseff-V_2,pk)²+P_3,th· (V_gseff-V_3,pk)³ (3)

I_pkth=I_pk·(1+K_Ipk(V_ds)·ΔT) (4)

M_ipkbth(V_ds)=M_ipkb(V_ds)·(1+K_Mipkb(V_ds)·ΔT) (5)

P_j,th(V_ds)=P_j(V_ds)·(1+K_j,P(V_ds) Δ T), j=1,2,3 (6)

V_gseff=f (V_gs,V_ds,V_dsq,V_gsq) (7)

Δ T=P_diss·R_theq=I_ds·V_ds·R_theq (8)

Wherein, I_dsIt is dram-source voltage；I_pkthIt is electric current when device transconductance is maximum；M_ipk(V_ds,V_gseff) it is I_pkth Tanh multiplier, for characterizing the asymmetric property of GaN HEMT mutual conductances；V_gseff(V_gs,V_ds,V_dsq,V_gsq) sunken to characterize The equivalent grid voltage of trap effect, it is dram-source voltage V_ds, grid-source voltage V_gs, dram-source voltage it is static partially Put point V_dsq, grid-source voltage quiescent bias point V_gsqFunction, its expression can according to demand choose appropriate shape Formula；Ψ(V_ds,V_gseff) it is multinomial series centered on grid voltage by mutual conductance when maximum；α is saturation voltage parameter；I_pk It is I_pkthSelf-heating effect independent component；K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds) it is sign self-heating The continuous item of effect, this five are dram-source voltage V_dsFunction, its expression can according to demand choose appropriate Form；Δ T is device channel temperature increment, and it is equal to dissipated power P_dissWith equivalent thermal resistance R_theqProduct；M_ipkbth(V_ds) and Q_m(V_ds) it is M_ipk(V_ds,V_gseff) intermediate variable inside expression formula, wherein M_ipkb(V_ds) it is M_ipkbth(V_ds) self-heating effect without Close component；P_j,th(V_ds) it is Ψ (V_ds,V_gseff) polynomial coefficient, wherein, j=1,2,3；P_j(V_ds) it is P_j,th(V_ds) from Fuel factor independent component；V_gsm、V_1,pk、V_2,pk、V_3,pkIt is the grid voltage parameter of correspondence mutual conductance peak value；

Step 2. extracts the parameter unrelated with self-heating effect and trap effect, and its flow is as shown in Figure 2：

Step 2-1：Enter horizontal pulse I-V tests to GaN HEMT devices under room temperature (22 DEG C)：The source of GaN HEMT devices Pole is grounded, the static bias voltage V of gate-to-source_gsq=-3V, the static bias voltage V of Drain-Source_dsq=0V；Grid arteries and veins Width is rushed for 1.2 μ s, drain electrode pulse width is 1 μ s, drain electrode pulse daley is 100ns；Scanning grid-source voltage scope for- 4V to 0V, is spaced 0.2V, and scanning dram-source voltage scope is 0V to 30V, is spaced 0.5V；

Step 2-2：Under the test condition of step 2-1, the self-heating effect and trap effect of device are negligible, device I-V model simplifications be following form：

I_ds=I_pk(1+M_ipk(V_ds,V_gs)·tanh(Ψ(V_ds,V_gs)))·tanh(αV_ds) (9)

M_ipk(V_ds,V_gs)=1+0.5 (M_ipkb(V_ds)-1)·(1+tanh(Q_m(V_ds)·(V_gs-V_gsm))) (10)

Ψ(V_ds,V_gs)=P₁(V_ds)·(V_gs-V_1,pk)+P₂(V_ds)·(V_gs-V_2,pk)²+P₃·(V_gs-V_3,pk)³ (11)

The pulse I-V test datas measured by step 2-1 can obtain in each dram-source voltage V_dsUnder, with drain electrode-source Pole tension V_gsIt is abscissa, drain current I_dsIt is the transfer characteristic curve of ordinate, as shown in Figure 3；For every transfer characteristic Curve, its corresponding V_dsIt is known definite value, thus the M in I-V models_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃ (V_ds) etc. with V_dsFor the function of independent variable becomes constant, whole I-V model simplifications are with V_gsIt is the function of a single variable of independent variable, And then reduce fitting difficulty；

Step 2-3：Using least square method equation (9), (10), (11) and every transfer spy described in step 2-2 Linearity curve is fitted, and obtains function M_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) in each V_dsUnder discrete take Value, and constant I_pk, α, V_gsm, V_1,pk, V_2,pkAnd V_3,pkValue；

Step 2-4：To the function M obtained in step 2-3_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) every Individual V_dsUnder discrete value, be fitted respectively using least square method and obtain function M_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) with V_dsIt is the expression formula of independent variable, described in detail below, its corresponding fitting effect is as shown in Figure 4：

M_ipkb(V_ds)=(P_M0+P_M1V_ds+P_M2V_ds ²+P_M3V_ds ³)tanh(α_MV_ds)+P_Mo (12)

Q_M(V_ds)=(P_Q0+P_Q1V_ds)tanh(α_QV_ds)+P_Qo (13)

P_j(V_ds)=j (P_{J, 0}+P_{J, 1}V_ds)tanh(α_Pj)+P_j,c, j=1,2,3 (14)

Wherein, P_M0,P_M1,P_M2,P_M3It is M_ipkb(V_ds) on V_dsThe multinomial coefficient of expansion, α_MIt is M_ipkb(V_ds) voltage Parameter of saturation, P_MoIt is M_ipkb(V_ds) modifying factor；P_Q0,P_Q1It is Q_M(V_ds) on V_dsThe multinomial coefficient of expansion, α_QIt is Q_M (V_ds) voltage saturation parameter, P_QoIt is Q_M(V_ds) modifying factor；P_j,0、P_j,1It is P_j(V_ds) on V_dsThe system of polynomials of expansion Number, α_PjIt is P_j(V_ds) saturation voltage parameter, P_j,cIt is P_j(V_ds) modifying factor, j=1,2,3；

Step 3. extracts the related parameter of self-heating effect, and its flow is as shown in Figure 5：

Step 3-1：Static I-V tests are done to GaN HEMT devices：The source ground of GaN HEMT devices, gate-to-source Scanning voltage scope be -4V to 0V, be spaced 0.2V, the scanning voltage scope of Drain-Source is 0V to 30V, is spaced 0.5V；

Step 3-2：Choose more than one test temperature higher than room temperature, such as it is right respectively at 55 DEG C, 85 DEG C and 115 DEG C GaN HEMT devices enter horizontal pulse I-V tests：The source ground of GaN HEMT devices, the static bias voltage V of gate-to-source_gsq =0V, the static bias voltage V of Drain-Source_dsq=0V；Gate pulse width is 1.2 μ s, and drain electrode pulse width is 1 μ s, leakage Pole pulse daley is 100ns；Scanning grid-source voltage scope is -4V to 0V, is spaced 0.2V, scans dram-source voltage model It is 0V to 30V to enclose, and is spaced 0.5V；

Step 3-3：By the pulse I-V test datas obtained in step 3-2, the side proposed using A.M.Darwish et al. Method extracts the equivalent thermal resistance R for obtaining GaN HEMT devices_theq(A.M.Darwish,A.Bayba,and H.A.Hung, “Thermal resistance calculation of AIGaN/GaN devices”,IEEE Trans.Microwave Theory Tech.,vol.52,no.11,pp.2611-2620,2004)；

Step 3-4：Due to containing the I for not putting forward ginseng completely in the channel temperature increment Delta T in formula (8)_ds, according to tradition Carry after Δ T substitutes into I-V models by ginseng method, can cause that I-V models turn into nested recursive form, be significantly greatly increased that to put forward ginseng difficult Degree；To avoid the occurrence of the I-V models of nested recursive form, the static I-V test datas that the present invention is obtained using step 3-1, meter Calculate in each V_dsAnd V_gsUnder dissipated power value P_diss=I_ds·V_ds, and using Polynomial curve-fit dissipated power P_diss To with V_dsAnd V_gsIt is the P of independent variable_dissExpression, then by P_dissExpression substitute into formula (8) to raceway groove temperature The expression of degree increment Delta T, P_dissFitting effect it is as shown in Figure 6；

Step 3-5：The static I-V test datas measured by step 3-1 can obtain in each dram-source voltage V_dsUnder with Dram-source voltage V_gsIt is abscissa, drain current I_dsIt is the transfer characteristic curve of ordinate；For every described in this step With bar transfer characteristic curve, its corresponding V_dsIt is known definite value, thus the K in I-V models_Ipk(V_ds), K_Mipkb(V_ds), K_1,P (V_ds), K_2,P(V_ds), K_3,P(V_ds), wait with V_dsFor the self-heating effect correlation function of independent variable becomes constant, whole I-V models It is reduced to V_gsIt is the function of a single variable of independent variable, and then reduces fitting difficulty；

Step 3-6：The model parameter unrelated with trap effect and self-heating effect that step 2 is obtained substitutes into I-V models, adopts Equation (4), (5), (6) and each transfer characteristic curve in step 3-5 are fitted with least square method, are obtained every Individual V_dsUnder, function K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds) discrete value；

Step 3-7：To the function K obtained in step 3-6_Ipk(V_ds), K_Mipkb(V_ds), K_P1(V_ds), K_P2(V_ds), K_P3(V_ds) In each V_dsUnder discrete value, be fitted respectively using least square method and obtain function K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P (V_ds), K_2,P(V_ds), K_3,P(V_ds) with V_dsIt is the expression formula of independent variable, described in detail below, its corresponding fitting effect such as Fig. 7 It is shown：

K_Ipk(V_ds)=(K_Ipk0+K_Ipk1V_ds)tanh(α_KIpkV_ds)+K_Ipko (15)

K_Mipkb(V_ds)=(K_KMipkb0+K_Mipkb1V_ds)tanh(α_KMipkbkV_ds)+K_Mipkbo (16)

K_j,P(V_ds)=(K_j,P0+K_j,P1V_ds)tanh(α_j,KPV_ds)+K_j,Pf, j=1,2,3 (17)

Wherein, K_Ipk0,K_Ipk1It is K_Ipk(V_ds) on V_dsThe multinomial coefficient of expansion, α_KIpkIt is K_Ipk(V_ds) voltage saturation Parameter, K_IpkoIt is K_Ipk(V_ds) modifying factor；K_Mipkb0,K_Mipkb1It is K_Mipkb(V_ds) on V_dsThe multinomial coefficient of expansion, α_KMipkbIt is K_Mipkb(V_ds) voltage saturation parameter, K_MipkboIt is K_Mipkb(V_ds) modifying factor；K_j,P0、K_j,P1It is K_j,P(V_ds) close In V_dsThe multinomial coefficient of expansion, α_j,KPIt is K_j,P(V_ds) voltage saturation parameter, K_j,PfIt is K_j,P(V_ds) modifying factor；

Step 4. extracts trap effect relevant parameter：

Step 4-1：Pulse I-V tests are done to GaN HEMT devices under room temperature (22 DEG C)：The source electrode of GaN HEMT devices Ground connection, the static bias voltage V of gate-to-source_gsq=0V, the static bias voltage V of Drain-Source_dsq=0V, grid impulse is wide It is 1.2 μ s to spend, and drain electrode pulse width is 1 μ s, and drain electrode pulse daley is 100ns, scanning grid-source voltage scope for -4V extremely 0V, is spaced 0.2V, and scanning dram-source voltage scope is 0V to 30V, is spaced 0.5V；

Step 4-2：Pulse I-V tests are done to GaN HEMT devices under room temperature (22 DEG C)：The source electrode of GaN HEMT devices Ground connection, the static bias voltage V of gate-to-source_gsq=-3V, the static bias voltage V of Drain-Source_dsq=20V, grid impulse Width is 1.2 μ s, and drain electrode pulse width is 1 μ s, and drain electrode pulse daley is 100ns, and scanning grid-source voltage scope is -4V To 0V, 0.2V is spaced, scanning dram-source voltage scope is 0V to 30V, is spaced 0.5V；

Step 4-3：With reference to the I-V model parameters obtained in step 2 and step 3, using least square fitting step 2- 1st, the pulse I-V curve for obtaining is measured in step 4-1 and step 4-2, you can obtain with V_gs、V_ds、V_dsq、V_gsqThis four parameters It is the model parameter V related to trap effect of variable_gseffExpression formula, it is described in detail below：

V_gseff=V_gs+γ_surf1(V_gsq-V_gsqpinch)(V_gs-V_gsqpinch)+γ_subs1(V_dsq+V_dssubs0)(V_ds-V_dsq) (18)

Wherein, four parameter γ_surf1,V_gsqpinch,γ_subs1,V_dssubs0By being known constant after curve matching, γ_surf1It is surface trap modulation parameter, V_gsqpinchIt is pinch-off voltage, γ_subs1It is buffering trap modulation parameter, V_dssubs0It is leakage Pole-source voltage quiescent bias point modifying factor.

Complete to have obtained after all of above step all parameter values of I-V models, Fig. 8 show extraction and obtains all I- Static I-V curve fitting result chart after V model parameters.

It should be noted that the form that embodies that equation (12)~(18) are used in this specific embodiment can basis Practical application is changed, and is not limited to the expression formula that this specific embodiment is used.

Claims (6)

1. a kind of extracting method of GaN high electron mobility transistor I-V model parameters, specifically includes following steps：

Step 1. parameter piecemeal：The concrete form after piecemeal is carried out to I-V model parameters as follows：

I_ds=I_pkth(1+M_ipk(V_ds,V_gseff)·tanh(Ψ(V_ds,V_gseff)))·tanh(αV_ds) (1)

M_ipk(V_ds,V_gseff)=1+0.5 (M_ipkbth(V_ds)-1)·(1+tanh(Q_m(V_ds)·(V_gseff-V_gsm))) (2)

Ψ(V_ds,V_gseff)=P_1,th(V_ds)·(V_gseff-V_1,pk)+P_2,th(V_ds)·(V_gseff-V_2,pk)²+P_3,th·(V_gseff- V_3,pk)³ (3)

I_pkth=I_pk·(1+K_Ipk(V_ds)·ΔT) (4)

M_ipkbth(V_ds)=M_ipkb(V_ds)·(1+K_Mipkb(V_ds)·ΔT) (5)

P_j,th(V_ds)=P_j(V_ds)·(1+K_j,P(V_ds) Δ T), j=1,2,3 (6)

V_gseff=f (V_gs,V_ds,V_dsq,V_gsq) (7)

Δ T=P_diss·R_theq=I_ds·V_ds·R_theq (8)

Wherein, I_dsIt is drain current；I_pkthIt is electric current when device transconductance is maximum；M_ipk(V_ds,V_gseff) it is I_pkthTanh Multiplier, for characterizing the asymmetric property of GaN HEMT mutual conductances；V_gseff(V_gs,V_ds,V_dsq,V_gsq) be characterize trap effect etc. Effect grid voltage, it is dram-source voltage V_ds, grid-source voltage V_gs, dram-source voltage quiescent bias point V_dsq, grid Pole-source voltage quiescent bias point V_gsqFunction, its expression can according to demand choose appropriate form；Ψ(V_ds, V_gseff) it is multinomial series centered on grid voltage by mutual conductance when maximum；α is saturation voltage parameter；I_pkIt is I_pkthFrom Fuel factor independent component；K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds) it is the correlation for characterizing self-heating effect , this five are dram-source voltage V_dsFunction, its expression can according to demand choose appropriate form；Δ T is Device channel temperature increment, it is equal to dissipated power P_dissWith equivalent thermal resistance R_theqProduct；M_ipkbth(V_ds) and Q_m(V_ds) be M_ipk(V_ds,V_gseff) intermediate variable inside expression formula, wherein M_ipkb(V_ds) it is M_ipkbth(V_ds) self-heating effect independent component； P_j,th(V_ds) it is Ψ (V_ds,V_gseff) polynomial coefficient, wherein, j=1,2,3；P_j(V_ds) it is P_j,th(V_ds) self-heating effect without Close component；V_gsm、V_1,pk、V_2,pk、V_3,pkIt is the grid voltage parameter of correspondence mutual conductance peak value；

Step 2. extracts the model parameter unrelated with self-heating effect and trap effect：For the transistor for needing extracting parameter, Its source ground, dram-source voltage quiescent bias point V_dsq=0V, grid-source voltage quiescent bias point V_gsqThe arteries and veins of=- 3V Rush under test condition, obtain the current-voltage i.e. transfer characteristic curve of I-V of device, now the self-heating effect and trap of device Effect relevant parameter is negligible；Pulse I-V transfer characteristic curves as curve matching obtained by and the side described in step 1 Journey (1), (2), (3) are fitted, and obtain unrelated with self-heating effect and trap effect model ginseng in equation (1), (2), (3) Number：I_pk, α, V_gsm, V_1,pk, V_2,pk, V_3,pk, M_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds)；

Step 3. extracts the related model parameter of self-heating effect：Static I-V tests are done to transistor, and utilizes Function Fitting method Obtain dissipated power P_dissWith V_dsAnd V_gsIt is the expression of independent variable；Enter horizontal pulse I-V tests, the data obtained to transistor Can extract out the equivalent thermal resistance R of device_theq, bring the expression that formula (8) can obtain channel temperature increment Delta T into；

The static I-V data that this step is measured can obtain static I_ds-V_gsTransfer characteristic curve, using curve-fitting method equation (4), (5), (6) and the static I_ds-V_gsTransfer characteristic curve is fitted, and obtains self-heating effect in equation (4), (5), (6) Related model parameter：K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds)；

Step 4. extracts the related model parameter of trap effect：For the transistor for needing extracting parameter, in its source ground, leakage Pole-source voltage quiescent bias point V_dsq=0V, grid-source voltage quiescent bias point V_gsqEnter horizontal pulse under conditions of=- 3V I-V tests to obtain pulse I-V curve, and in its source ground, dram-source voltage quiescent bias point V_dsq=0V, gate-to-source Voltage quiescent bias point V_gsqEnter horizontal pulse I-V under conditions of=0V and test to obtain pulse I-V curve；With reference to step 2 and step 3 Trap effect relevant parameter V is removed in equation (1)~(8) tried to achieve_gseffOuter all parameters, walk this using curve-fitting method Two group pulse I-V curves of rapid gained are fitted with equation (7), you can obtain with V_gs、V_ds、V_dsq、V_gsqThis four parameters are The related model parameter V of the trap effect of variable_gseffExpression, finally realize test transistor all I-V models ginseng Several extractions；

Wherein, the process of the parameter unrelated with self-heating effect and trap effect of the extraction described in step 2 specifically includes following steps：

Step 2-1：Enter horizontal pulse I-V tests to GaN HEMT devices：The source ground of device, the quiescent biasing of gate-to-source Voltage V_gsq=-3V, the static bias voltage V of Drain-Source_dsq=0V；

Step 2-2：Under the test condition of step 2-1, the self-heating effect and trap effect of device are negligible, the I- of device V model simplifications are following form：

I_ds=I_pk(1+M_ipk(V_ds,V_gs)·tanh(Ψ(V_ds,V_gs)))·tanh(αV_ds) (9)

M_ipk(V_ds,V_gs)=1+0.5 (M_ipkb(V_ds)-1)·(1+tanh(Q_m(V_ds)·(V_gs-V_gsm))) (10)

Ψ(V_ds,V_gs)=P₁(V_ds)·(V_gs-V_1,pk)+P₂(V_ds)·(V_gs-V_2,pk)²+P₃(V_ds)·(V_gs-V_3,pk)³ (11)

The pulse I-V test datas measured by step 2-1 can obtain in each dram-source voltage V_dsUnder, with gate-to-source electricity Pressure V_gsIt is abscissa, drain current I_dsIt is the transfer characteristic curve of ordinate；For every transfer characteristic curve, its is corresponding V_dsIt is known definite value, thus the M in I-V models_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) etc. with V_dsIt is certainly The function of variable becomes constant, and whole I-V model simplifications are with V_gsIt is the function of a single variable of independent variable；

Step 2-3：Using curve-fitting method equation (9), (10), (11) and each transfer spy described in step 2-2 Linearity curve is fitted, and obtains function M_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) in each V_dsUnder discrete take Value, and constant I_pk, α, V_gsm, V_1,pk, V_2,pkAnd V_3,pkValue；

Step 2-4：To the function M obtained in step 2-3_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) in each V_ds Under discrete value, be fitted respectively using Function Fitting method and obtain function M_ipkb(V_ds), Q_m(V_ds), P₁(V_ds), P₂(V_ds), P₃(V_ds) with V_dsIt is the expression formula of independent variable, it is described in detail below：

M_ipkb(V_ds)=(P_M0+P_M1V_ds+P_M2V_ds ²+P_M3V_ds ³)tanh(α_MV_ds)+P_Mo (12)

Q_M(V_ds)=(P_Q0+P_Q1V_ds)tanh(α_QV_ds)+P_Qo (13)

Wherein, P_M0,P_M1,P_M2,P_M3It is M_ipkb(V_ds) on V_dsThe multinomial coefficient of expansion, α_MIt is M_ipkb(V_ds) voltage saturation Parameter, P_MoIt is M_ipkb(V_ds) modifying factor；P_Q0,P_Q1It is Q_M(V_ds) on V_dsThe multinomial coefficient of expansion, α_QIt is Q_M(V_ds) Voltage saturation parameter, P_QoIt is Q_M(V_ds) modifying factor；P_j,0、P_j,1It is P_j(V_ds) on V_dsThe multinomial coefficient of expansion, α_PjFor P_j(V_ds) saturation voltage parameter, P_j,cIt is P_j(V_ds) modifying factor, j=1,2,3.

2. the extracting method of GaN high electron mobility transistor I-V model parameters according to claim 1, its feature It is that the process of the extraction self-heating effect relevant parameter described in step 3 specifically includes following steps：

Step 3-1：Static I-V tests are done to device；

Step 3-2：Enter horizontal pulse I-V tests to device：The source ground of device, the static bias voltage V of gate-to-source_gsq= 0V, the static bias voltage V of Drain-Source_dsq=0V；

Step 3-3：By the pulse I-V test datas obtained in step 3-2, extraction obtains the equivalent thermal resistance R of device_theq；

Step 3-4：The static I-V test datas obtained using step 3-1, are calculated in each V_dsAnd V_gsUnder dissipated power value P_diss=I_ds·V_ds, and use Function Fitting method to obtain with V_dsAnd V_gsIt is the P of independent variable_dissExpression, then will P_dissExpression substitute into formula (8) channel temperature increment Delta T expression；

Step 3-5：The static I-V test datas measured by step 3-1 can obtain in each dram-source voltage V_dsUnder with grid Pole-source voltage V_gsIt is abscissa, drain current I_dsIt is the transfer characteristic curve of ordinate；For each described in this step Bar transfer characteristic curve, its corresponding V_dsIt is known definite value, thus the K in I-V models_Ipk(V_ds), K_Mipkb(V_ds), K_1,P (V_ds), K_2,P(V_ds), K_3,P(V_ds), wait with V_dsFor the self-heating effect correlation function of independent variable becomes constant, whole I-V models It is reduced to V_gsIt is the function of a single variable of independent variable；

Step 3-6：The model parameter unrelated with trap effect and self-heating effect that step 2 is obtained substitutes into I-V models, using song Line approximating method is fitted equation (4), (5), (6) and each transfer characteristic curve in step 3-5, obtains at each V_dsUnder, function K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds) discrete value；

Step 3-7：To the function K obtained in step 3-6_Ipk(V_ds), K_Mipkb(V_ds), K_1,P(V_ds), K_2,P(V_ds), K_3,P(V_ds) Each V_dsUnder discrete value, be fitted respectively using Function Fitting method and obtain function K_Ipk(V_ds), K_Mipkb(V_ds), K_1,P (V_ds), K_2,P(V_ds), K_3,P(V_ds) with V_dsIt is the expression formula of independent variable, it is described in detail below：

K_Ipk(V_ds)=(K_Ipk0+K_Ipk1V_ds)tanh(α_KIpkV_ds)+K_Ipko (15)

K_Mipkb(V_ds)=(K_KMipkb0+K_Mipkb1V_ds)tanh(α_KMipkbV_ds)+K_Mipkbo (16)

K_j,P(V_ds)=(K_j,P0+K_j,P1V_ds)tanh(α_j,KPV_ds)+K_j,Pf, j=1,2,3 (17)

Wherein, K_Ipk0,K_Ipk1It is K_Ipk(V_ds) on V_dsThe multinomial coefficient of expansion, α_KIpkIt is K_Ipk(V_ds) voltage saturation ginseng Number, K_IpkoIt is K_Ipk(V_ds) modifying factor；K_Mipkb0,K_Mipkb1It is K_Mipkb(V_ds) on V_dsThe multinomial coefficient of expansion, α_KMipkb It is K_Mipkb(V_ds) voltage saturation parameter, K_Mipkb<It is K_Mipkb(V_ds) modifying factor；K_j,P0、K_j,P1It is K_j,P(V_ds) on V_ds The multinomial coefficient of expansion, α_j,KPIt is K_j,P(V_ds) voltage saturation parameter, K_j,PfIt is K_j,P(V_ds) modifying factor.

3. the extracting method of GaN high electron mobility transistor I-V model parameters according to claim 2, its feature It is that the process of the extraction trap effect relevant parameter described in step 4 specifically includes following steps：

Step 4-1：Pulse I-V tests are done to device：The source ground of device, the static bias voltage V of gate-to-source_gsq= 0V, the static bias voltage V of Drain-Source_dsq=0V；

Step 4-2：Pulse I-V tests are done to device：The source ground of device, the static bias voltage V of gate-to-source_gsq=- 3V, the static bias voltage V of Drain-Source_dsq=20V；

Step 4-3：With reference to being obtained in step 2 and step 3 except V_gseffOuter I-V model parameters, and use curve-fitting method The pulse I-V curve for obtaining is measured in fit procedure 2-1, step 4-1 and step 4-2, you can obtain with V_gs、V_ds、V_dsq、V_gsq This four parameters are the model parameter V related to trap effect of variable_gseffExpression formula, it is described in detail below：

Wherein, four parameter γ_surf1,V_gsqpinch,γ_subs1,V_dssubs0By being known constant, γ after curve matching_surf1For Surface trap modulation parameter, V_gsqpinchIt is pinch-off voltage, γ_subs1It is buffering trap modulation parameter, V_dssubs0It is Drain-Source electricity Pressure quiescent bias point modifying factor.

4. the extracting method of GaN high electron mobility transistor I-V model parameters according to claim 3, its feature It is that the curve-fitting method and Function Fitting method are least square method.

5. the extracting method of GaN high electron mobility transistor I-V model parameters according to claim 3, its feature It is that the partial parameters of the pulse I-V tests and static state I-V tests are by following setting：The source ground of device, grid- The scanning voltage scope of source electrode be -4V to 0V, its sweep spacing 0.2V, the scanning voltage scope of Drain-Source is 0V to 30V, Its sweep spacing 0.5V；The partial parameters of the pulse I-V tests set as follows：Gate pulse width is 1.2 μ s, and drain pulse Width is 1 μ s, and drain electrode pulse daley is 100ns.

6. the extracting method of GaN high electron mobility transistor I-V model parameters according to claim 3, its feature It is that the pulse I-V tests that step 2-1, step 4-1 and step 4-2 are carried out are carried out in identical temperature environment；Step Environment temperature of the environment temperature of pulse I-V tests described in 3-2 higher than the pulse I-V tests that other steps are carried out.