Research On Surface-potential-based Compact Model For Amorphous IGZO Thin-film Transistors

Amorphous metal oxide semiconductors,represented by amorphous indium gallium zinc oxide(a-IGZO),have been recognized as the most promising candidate for the next thin-film transistors(TFTs)technology.In the near future,a-IGZO TFTs are expected to be widely used in many fields,including the active matrix/organic light emitting diode(AMOLED)and flexible circuits.Compared to traditional silicon counterparts,a-IGZO TFTs show different operation mechanisms.Based on such mechanisms,compact models for these devices should be developed to understand the impact of materials innovation on the electrical properties of TFTs,and finally to assist the design of a-IGZO TFTs based integrated circuit.These physical models will have profound significance to the promotion and development of AMOLED and integrated circuit industries.The main contents of this paper are the analysis of the a-IGZO TFTs physical mechanisms,the development of the surfaced-potential-based direct current(DC)and capacitance compact models for a-IGZO TFTs including degenerate conduction.These models are capable of fulfilling the conditions of circuit simulator implementation.The analysis of band structure and subgap trap states provides a clear theoretical foundation of developing mathematical models.Due to the strong electronegativity,the conduction band of a-IGZO is composed of spherical overlapping s orbitals from In ionic and the valence band is composed of 2p orbitals from O ionic,leading to a smaller density of tail states near the conduction band than that of covalent semiconductors like hydrogenated amorphous silicon(a-Si:H).Meanwhile,the existence of a high density of deep trap states near the top of valence band pins the Fermi level and stops it to move toward the valence band.So at present,the a-IGZO material is targeted for n-type TFTs.Such band structure and trap states distribution illustrate that for a-IGZO TFTs,the Fermi level may exceed the band edge at a certain gate bias which is large enough.Therefore,the degenerate conduction regime must be taken into account in modeling.Based on the theoretical analysis,the limitation of Boltzmann statistics is spoken out,and the charge concentration formulas for the free and trapped charges are proposed withcontinuity from the non-degenerate to the degenerate regime.Combining with Poisson's equation and Gauss' s law,the surface potential of subthreshold and accumulation regimes is non-iteratively derived by using mathematical procedures and Lambert W function.Smooth function and Schroeder series correction are also adopted.Finally,a unified and analytical model of the surface potential with a high precision is developed.Compared to numerical results,the absolute error of the proposed model is in 10-8V range.Based on the calculated surface potential and the Pao-Sah double integral formula,the direct current characteristics of a-IGZO TFTs are predicted and modeled,including subthreshold and accumulation regimes.The mobility model is based on the actual carrier transport mechanisms,including the trap-limited conduction(TLC)and percolation conduction mechanism(PM).Hence,the field mobility follows a power law.Mobility degradation due to the phonon scattering and rough surface scattering is also taken into account.Modeling of advanced effects is also discussed,including the carrier velocity saturation and the channel length modulation effect.Besides,based on Charge-Sheet Model(CSM)and Symmetric Quadrature Method(SQM),the terminal charges are derived and characteristics of capacitance at a-IGZO TFTs including degeneration have been modeled.Such model can fulfill the requirements of charge conservation and nonreciprocal property of capacitance.It can also well describe the dynamic behavior of these devices.In conclusion,the proposed direct current model and capacitance model are physically based on the a-IGZO TFTs operation mechanisms.Through comparisons between model results and numerical iteration or experimental data,the validity of the proposed models is strongly supported.Due to the simplicity of algorithm and clear physical conception,it is easy to implement these models into circuit simulators.