Research On IGBT Physical Model Optimization And Parameter Identification

With the rapid development of renewable energy technology,power semiconductor devices represented by insulated gate bipolar transistor（IGBT）are applied in electric vehicles,photovoltaic generation system,wind power generation system and other fields as core components.In power electronic devices,power semiconductor devices operate in various operating conditions,and there are complex coupling situations of various power semiconductor devices.In order to predict IGBT switching loss,evaluate electromagnetic interference and design the dead zone time,an accurate IGBT model is needed.The Fourier-series based（FSB）IGBT model,as one of the physics-based models,has clear physical significance and can predict the carrier dynamics and electric field distribution inside the device.The model is very effective in terms of simulation speed and convergence and is suitable for various simulation platforms.However,the existing FSB IGBT model still has the following problems:Firstly,the method of using Fourier series to solve the ambipolar diffusion equation will produce Gibbs phenomenon,which is caused by the truncation error of Fourier series.The influence of the Gibbs phenomenon will cause certain distortion of carrier concentration distribution of FSB IGBT model in the switching process,then the distortion of carrier concentration distribution will have an impact on the electric field in the carrier storage region.Then,the V_CE waveform of FSB IGBT model will appear errors.Therefore,how to mitigate the effects of Gibbs phenomenon is a key problem in FSB IGBT modeling.Secondly,The Miller capacitance formula in FSB field-stop（FS）IGBT model does not consider the two-dimensional distribution of depletion layer.It is generally believed that miller capacitor consists of oxide layer capacitor and depletion layer capacitor directly in series.However,in IGBT switching process,the depletion layer expands/shrinks not only in the vertical direction,but also in the horizontal direction.the unit depletion layer capacitance of different regions is different.Miller capacitance formula is one of the key factors affecting the accuracy of IGBT model switching waveform,therefore the two-dimensional distribution of depletion layer in transverse direction should be considered when describing Miller capacitance which can improve the accuracy of FS IGBT model in dynamic process.Finally,device manufacturers rarely provide detailed device parameters available for physics-based models,which seriously hinders the use of physical models.The general method is to obtain IGBT and diode parameters by fitting IGBT static characteristic curve and IGBT switching waveforms under inductive load.However,it is very difficult to obtain parameters that can be applied to different temperatures,voltages and currents at the same time.Meanwhile,it is difficult to ensure the accuracy of both dynamic and static waveforms of the model.In view of the influence of the Gibbs phenomenon,the turn on process finite element simulation of IGBT is established in Sentaurus TCAD.According to the carrier distribution in finite element simulation,the boundary conditions of FSB IGBT model are modified,and the improved FSB IGBT model is implemented in Simulink/MATLAB.Then,the improved FSB IGBT model is compared with the finite element simulation results in TCAD,and the consistency of the key parameters of the two models is guaranteed.Results show that the improved scheme proposed in this paper solves the key problem of FSB IGBT model,simulates the two-dimensional distribution of carriers in the switching process,effectively mitigates the influence of Gibbs phenomenon,improves the carrier concentration distribution,electric field intensity distribution and voltage/current waveform of FSB IGBT model in the turn-on process,which implements significant improvements to the FSB IGBT modeling.To solve the problem of Miller capacitance calculation,the simulation of field-stop（FS）IGBT under inductive load circuit in Sentaurus TCAD is firstly established in this paper.By observing the change rule of depletion layer in IGBT switching process,combined with the finite difference calculation method,a calculation formula of Miller capacitance for trench-gate FS IGBT is presented.The simulation results of different Miller capacitance formulas are compared,and the results show that the proposed Miller capacitance formula improves the accuracy of FSB FS IGBT model.The robustness and accuracy of the improved FSB FS IGBT model are verified by comparing the simulation results of FSB FS IGBT model with TCAD simulation results under different operating conditions.The improved FSB FS IGBT model can accurately predict the switching waveform of IGBT under different working conditions and can predict the dynamic changes of carriers and electric fields in carrier storage region,which promotes the application of FSB FS IGBT model in the field of power electronics.Aiming at the problem of parameter extraction,this paper proposes an automatic extraction method of IGBT parameters based on the improved FSB IGBT model.Particle swarm optimization（PSO）algorithm was used to iterate parameters to reduce the error between model output and benchmark,so as to search the optimal parameter set and realize automatic identification of IGBT parameters.In this paper,two extraction schemes are proposed.The first scheme uses IGBT turn-off waveform to extract IGBT parameters and uses IGBT turn-on waveform to extract diode parameters.The advantage of this parameter extraction scheme is that the influence of freewheel diode on IGBT model can be basically ignored when extracting IGBT parameters,which improves the accuracy of extracted IGBT parameters.On the basis of the first scheme,the second scheme uses IGBT turn-off waveform and IGBT output characteristic curve simultaneously to obtain IGBT parameters.Although the switching waveform precision of scheme 2 is not as good as scheme 1,scheme 2 can make the static characteristics of FSB IGBT model more accurate.The two parameter identification processes are verified by the finite element and experimental results.In this paper,the modeling theory and parameter identification method of FSB IGBT model are studied,the physical significance of FSB NPT model is further clarified,the accuracy of FSB NPT IGBT and FSB FS IGBT models is improved,and the parameter identification method of IGBT is explored in depth.It is of great significance to the application of FSB IGBT models in power electronics field.