A modeling method for radio frequency MOS devices

Abstract

The present invention discloses a modeling method for radio frequency MOS devices, and relates to the technical field of MOS device modeling. The present invention combines electromagnetic simulation with thermal field simulation to establish a dynamic parasitic parameter library to describe parasitic effects under complex working conditions. With the dynamic parasitic parameter library as the core foundation, the dynamic response capability of the model is significantly improved, and the dynamic parasitic parameters are introduced into a small signal equivalent circuit model to explicitly reflect the dynamic changes of the parasitic parameters. A partitioning modeling method is introduced to refine the dynamic behaviors of the gate, source, and drain into independent contributions, which can more accurately describe the local dynamic characteristics of the radio frequency MOS device and enhance the modeling accuracy. Large signal excitation and nonlinear polynomial description are introduced to capture the nonlinear behavior under high power conditions, and a unified behavioral model from linear to nonlinear is established. The Volterra series method is used to accurately describe high-order nonlinearity. Combined with the dynamic parasitic parameters, the key problems of signal frequency aliasing and dynamic distortion in nonlinear analysis are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of MOS device modeling, and in particular to a modeling method for a radio frequency MOS device. Background Art

[0002] As a key component in high-frequency communication systems, RF MOS devices require modeling methods that accurately describe the dynamic behavior of the devices in complex electromagnetic environments and nonlinear operating conditions.

[0003] At present, mainstream modeling methods are mainly based on equivalent circuit models or analytical expressions; these methods have significantly enhanced parasitic effects at high frequencies. In nanoscale MOS devices, the parasitic capacitance and parasitic inductance between the source, gate and drain have a significant impact on circuit performance as the frequency changes. Traditional modeling methods mostly use static parameters or linear approximations for processing, resulting in insufficient model accuracy; secondly, as the frequency increases, the electromagnetic coupling effect becomes more complex, and existing models find it difficult to accurately capture these characteristics.

[0004] In terms of large-signal nonlinear modeling, most traditional methods assume that the nonlinear behavior of RF MOS devices can be described by a fixed nonlinear charge model. Obviously, such static nonlinear assumptions are not effective under dynamic excitation. In broadband communication systems, nonlinear behavior will lead to harmonics and intermodulation distortion. These high-order dynamic characteristics need to be accurately described by introducing time-dependent charge equations or dynamic frequency responses. However, existing analytical models or methods based on experimental parameter fitting are difficult to fully capture the dynamic nonlinear characteristics under large-signal conditions. Some traditional solutions adopt the method of establishing independent models for parasitic effects, thermal effects and nonlinear characteristics respectively, and couple them in simulation. This method is more efficient in the early stages of model development, but its computing resource requirements are high, and there is a lack of a unified dynamic coupling mechanism between different models, which makes it difficult to meet the requirements of modern RF systems for high-precision modeling. Therefore, there is an urgent need for a modeling method for RF MOS devices to solve such problems. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a modeling method for radio frequency MOS devices to solve the problem that a fixed nonlinear charge model of a traditional method has a poor description effect on radio frequency MOS devices under dynamic excitation.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] The present invention provides a modeling method for a radio frequency MOS device, which comprises:

[0009] Step S1, physical field simulation to extract dynamic parasitic parameters,

[0010] Obtain the dynamic parasitic parameters and their associated characteristics of RF MOS devices under actual device operating conditions;

[0011] Step S2: combining physical characteristic parameters with equivalent circuit modeling.

[0012] Integrating the dynamic parasitic parameters extracted in step S1 into the equivalent circuit model of the radio frequency MOS device;

[0013] Step S3, nonlinear characteristics analysis,

[0014] Based on the equivalent circuit model established in step S2 and combined with the dynamic parasitic parameters extracted in step S1, the nonlinear characteristics of the RF MOS device under high power and strong signal excitation are analyzed, and a unified linear and nonlinear behavioral model, namely the RF MOS device model, is constructed.

[0015] Furthermore, in step S1, the electromagnetic characteristics of the radio frequency MOS device are simulated using the finite element method (FEM), the parasitic inductance and parasitic capacitance parameters under high frequency conditions are extracted, and their dynamic characteristics varying with frequency are analyzed;

[0016] Introducing thermal field simulation analysis, combined with the self-heating effect of the device, to evaluate the dynamic impact of power consumption and temperature rise on performance;

[0017] By combining the results of electromagnetic simulation and thermal field simulation, a dynamic parasitic parameter library of RF MOS devices is established, which includes parasitic parameters and their correlation characteristics with frequency, temperature, voltage and frequency.

[0018] Furthermore, in step S1, the electromagnetic characteristics simulation method is:

[0019] Construct a three-dimensional geometric model of the RF MOS device, including the electromagnetic structure of the gate, source, drain and its surrounding medium. Assume that the relative dielectric constant of the MOS device material is ε r , the conductivity is σ c , a high-frequency signal is applied in the boundary conditions, and the AC voltage V in (t), the input signal frequency is f;

[0020] Extract the parasitic capacitance and use the finite element method to solve the Poisson form of the static or quasi-static Maxwell equations: Where φ is the potential distribution, ε=ε0ε r is the dielectric constant of the medium, ε0 is the vacuum dielectric constant, ε r is the relative dielectric constant, ρ is the space charge density, and is defined as the excitation source;

[0021] Define the total charge Q between the gate and source gs , and calculate the parasitic capacitance C gs: Among them, Q gs =∫ S εE·ndS is the total charge passing through the surface between the gate and the source, E is the electric field strength, dS represents the small area on the surface S, n is the surface normal vector, V in is the input voltage applied to the gate; for frequency dependence, the simulation frequency is gradually swept to [f1,f2,…,f n ], and record C gs (f) changes, generating dynamic response curves;

[0022] In step S1, the parasitic inductance is extracted, and the finite element method is used to solve the magnetic field distribution under high-frequency conditions based on the vector Poisson equation of Maxwell's equations: Where A is the vector magnetic potential, defined as B is the magnetic induction intensity, is the vector differential operator, μ=μ0μ r is the magnetic permeability of the medium, μ0 is the magnetic permeability of vacuum, μ r is the relative magnetic permeability, J is the current density;

[0023] Calculate the gate parasitic inductance L from the magnetic field energy gs : Where H = μ -1 B, where is the magnetic field strength, I gs is the AC current flowing between the gate and the source, defined as the peak value of the high-frequency current, and V is the magnetic field distribution volume near the gate; the simulation frequency is gradually scanned as [f1, f2, ..., f n ], record L gs (f) changes, and obtain the frequency dynamic response curve.

[0024] Furthermore, in step S1, the step of establishing a dynamic parasitic parameter library includes:

[0025] According to the thermal field distribution and based on the heat conduction equation, the temperature field inside the device is solved:

[0026] Where κ is the thermal conductivity of the material, P gen =I 2 r j is the power density, where I is the current flowing through the device, R j is the junction resistance, ρ c is the material density, C p is the specific heat capacity, Θ is the temperature field distribution;

[0027] In steady state Obtain the steady-state temperature distribution of the MOS device. Under dynamic conditions, analyze the impact of the input signal frequency f on the device temperature fluctuation and obtain the temperature response Θ(f).

[0028] Set the bottom of the device as the heat dissipation boundary, and the boundary condition Θ boundary is the ambient temperature Θ ambient ;

[0029] Assume that the temperature field obtained by simulation is Θ(x,y,z), and use the temperature field Θ(x,y,z) and power density P gen Analyze the nonlinear relationship between power consumption and temperature rise: Among them, ΔΘ is the temperature rise of the device. The formula is divided into two parts to describe the influence of heat dissipation and heat conduction on temperature rise. h is the heat dissipation coefficient, A s is the heat dissipation surface area;

[0030] Define the effect of temperature rise on capacitance and inductance C gs (Θ) and L gs (Θ):

[0031] C gs (Θ)=C gs0 [1+α C (Θ-Θ ambient )], where C gs0 is the initial gate-source capacitance, α C is the temperature coefficient;

[0032] L gs (Θ)=L gs0 [1-α L (Θ-Θ ambient )], where L gs0 is the initial gate-source inductance, α L is the temperature coefficient;

[0033] The parasitic parameters C extracted from electromagnetic simulation gs (f),L gs (f) and the temperature response Θ(f) are mapped into the associated characteristics:

[0034] C gs (f,Θ)=C gs0 [1+α C (Θ(f)-Θ ambient )],

[0035] L gs (f,Θ)=L gs0 [1-α L (Θ(f)-Θ ambient )], the input signal frequency f is swept from low frequency to the upper limit of the working frequency band, and the parameter dynamic response curve is obtained;

[0036] Combine the correlation characteristics of frequency, temperature and voltage to build a dynamic parasitic parameter library:

[0037] Dynamic parasitic capacitance: C gs (f,V,Θ)=C gs0 [1+α C (Θ-Θ ambient )+β C V], where β C is the voltage coefficient, which describes the effect of bias voltage on parasitic capacitance;

[0038] Dynamic parasitic inductance: L gs (f,V,Θ)=L gs0 [1-α L (Θ-Θ ambient )-β L V], where β L is the voltage coefficient, which describes the effect of bias voltage on parasitic inductance;

[0039] The C obtained by simulation gs (f, V, Θ) and L gs (f, V, Θ) is mapped into a dynamic parameter library.

[0040] Furthermore, in step S2, a small signal equivalent circuit model is established based on the dynamic parasitic parameters, and correlation features are added to the equivalent circuit model. The equivalent circuit model serves as a basic framework for modeling RF MOS devices.

[0041] Based on the physical characteristics and circuit behavior, i.e., the parasitic parameters under different conditions in step S1, the currents of the gate, source, and drain are partitioned and modeled. The partitioned modeling refines the dynamic behavior in the basic framework.

[0042] Based on the equivalent circuit model and partition modeling, the frequency response function of the MOS device is derived to predict the performance within a wide frequency band.

[0043] Furthermore, in step S2, the physical characteristic parameters are combined with the equivalent circuit modeling in the following manner:

[0044] Construct a small signal equivalent circuit model and transform the dynamic parasitic parameters extracted in step S1, dynamic parasitic capacitance C gs (f, V, Θ) and gate-drain capacitance C gd (f, V, Θ) introduces a small signal equivalent circuit:

[0045] Among them, Z gs (f, V, Θ) is the dynamic impedance between the gate and the source, f is the input signal frequency, and j is the imaginary unit;

[0046] Among them, Z gd(f, V, Θ) is the dynamic impedance between gate and drain;

[0047] In the equivalent circuit, the dynamic parasitic parameters are characterized as impedance elements and are related to the transconductance current g. m v gs Together with other non-parasitic characteristics, it forms the small signal framework, where: Among them, i d is the drain current, g m is the transconductance, defined as V gs is the gate-source voltage, V gd is the gate-drain voltage.

[0048] Furthermore, in step S2, gate, source, and drain partition modeling is performed:

[0049] Gate dynamic behavior includes signal drive and parasitic effects, equivalent current: Among them, i gs is the gate-source current, is the time derivative of the input signal;

[0050] The dynamic characteristics of the source part are determined by the dynamic parasitic inductance L s Description, the source current is:

[0051] Among them, V s is the source voltage, Z s is the source impedance, R s is the source equivalent resistance;

[0052] The dynamic behavior of the drain section includes transconductance and parasitic effects:

[0053] Among them, g m v gs is the transconductance current, is the dynamic current introduced by parasitic capacitance;

[0054] According to the partition modeling results, calculate the input impedance Z in :

[0055] Among them, Z in is the input impedance of the RF MOS device;

[0056] Drain current i d The gate-source voltage v gs The relationship describing the gain characteristic is:

[0057] Among them, H(f) is the gain function, the first term g mis the contribution of transconductance to gain, and the second term 1 / Z gd is the effect of parasitic impedance on gain;

[0058] According to the frequency characteristics of the total impedance, the phase response function is derived:

[0059] Where φ(f) is the phase difference between the input signal and the drain output signal, Im(Z in ) is the imaginary part of the input impedance, Re(Z in ) is the real part of the input impedance.

[0060] Furthermore, in step S3, based on the equivalent circuit model in step S2, nonlinear characteristics are added and large signal excitation conditions are introduced;

[0061] Extract intermodulation distortion (IMD), harmonic distortion (HD), and power compression point (P1dB) nonlinear indicators. Based on dynamic parameters and nonlinear indicators, the Volterra series method is used to construct a nonlinear behavior model.

[0062] Combining the nonlinear behavior model with the linear equivalent circuit model of step S2 to form a radio frequency MOS device model;

[0063] The MOS device model can simultaneously predict the linear characteristics in a wide frequency band and the nonlinear characteristics under strong signals.

[0064] Furthermore, in step S3, the step of constructing a nonlinear model of the RF MOS device includes:

[0065] Add large signal excitation and nonlinear characteristic expansion, apply large signal input v in (t):

[0066] v in (t) = V0cos(2πf0t) + V1cos(2πf1t), where V0 and V1 are the amplitudes of the fundamental frequency f0 and the interference frequency f1, and f0 and f1 are the fundamental frequency and the interference frequency,

[0067] Introducing nonlinear characteristics, using nonlinear polynomials to represent the drain current i d The nonlinear response of:

[0068] Among them, g1, g2, g3 are the first-order, second-order, and third-order derivative coefficients, defined as: Among them, v gs is the gate-source voltage;

[0069] Extract intermodulation distortion IMD. The mixing of f0 and f1 in the input signal will produce intermodulation frequency components:

[0070] f IMD=2f0-f1,2f1-f0, where f IMD is the intermodulation frequency component;

[0071] The second-order and third-order nonlinear terms contribute the following intermodulation currents respectively:

[0072] i IMD2 =g2V0V1cos(2πf IMD t), where i IMD2 is the intermodulation component caused by the second-order nonlinearity,

[0073] Among them, i IMD3 is the intermodulation component caused by the third-order nonlinearity;

[0074] Extract harmonic distortion HD, the harmonic frequency of the large signal input is f HD =nf0,n=2,3,4,f HD is the harmonic frequency,

[0075] Calculate the harmonic component amplitude using the nonlinear term contribution: i HD2 is the second harmonic component; i HD3 is the third harmonic component;

[0076] Extract the power compression point P1dB, the power gain G changes with the input power P in It manifests as:

[0077] G(P in )=G0-kP in , where G0 is the small signal gain and k is the nonlinear compression factor;

[0078] When the gain drops by 1dB, the input power compression point P in,1dB satisfy: P in,1dB is the input power compression point.

[0079] Furthermore, in step S3, a nonlinear behavior model is constructed based on the Volterra series:

[0080] The nonlinear behavior is determined by the Volterra kernel H n Definition, the input and output relationship is:

[0081] Among them, H n is the n-order Volterra kernel, v gs is the gate-source voltage, dτ1…dτ n Represents the time variables τ1,…,τ n Differential elements for multiple integration, τ1,…,τn The gate-source voltage v gs Delay time at different moments;

[0082] For second-order and third-order nonlinear behaviors, H2(τ1,τ2)=g2δ(τ1-τ2), where H2 is the second-order Volterra kernel, representing the second-order nonlinear response of the input signal;

[0083] H3(τ1,τ2,τ3)=g3δ(τ1-τ2)δ(τ2-τ3), where H3 is the third-order Volterra kernel, representing the third-order nonlinear response of the input signal; δ is the Dirac delta function;

[0084] Combining the linear equivalent circuit model with the nonlinear Volterra model, the complete response is:

[0085] i d (t) = g1v gs (t)+∫∫H2(τ1,τ2)v gs (t-τ1)v gs (t-τ2)dτ1dτ2+…, where the first term is the linear contribution and the higher-order terms are the nonlinear contributions.

[0086] The beneficial effects of the present invention are:

[0087] The present invention combines electromagnetic simulation with thermal field simulation to extract dynamic correlation parameters of frequency, temperature, and bias voltage, establishes a dynamic parasitic parameter library, and describes the parasitic effects under complex working conditions. With the dynamic parasitic parameter library as the core foundation, the dynamic response capability of the model is significantly improved, and the dynamic parasitic parameters are introduced into the small-signal equivalent circuit model to explicitly reflect the dynamic changes of the parasitic parameters. A partitioned modeling method is introduced to refine the dynamic behavior of the gate, source, and drain into independent contributions, which can more accurately describe the local dynamic characteristics of the RF MOS device and enhance the modeling accuracy.

[0088] The present invention introduces large signal excitation and nonlinear polynomial description to capture nonlinear behavior under high power conditions, establishes a unified behavioral model from linear to nonlinear, uses the Volterra series method to accurately describe high-order nonlinearity, and combines dynamic parasitic parameters to solve the key problems of signal frequency aliasing and dynamic distortion in nonlinear analysis.

[0089] The present invention achieves a comprehensive prediction of the wide-band performance of RF MOS devices by deriving frequency response functions and phase response functions. By incorporating dynamic parasitic parameters and thermal effects, the model can respond more accurately to high-frequency non-ideal effects, and extract key nonlinear indicators such as IMD, HD, and P1dB to quantitatively evaluate the impact of high-frequency distortion on signal integrity. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0091] Figure 1 The figure is a schematic diagram of the modeling process of the radio frequency MOS device of the present invention. DETAILED DESCRIPTION

[0092] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0093] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0094] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0095] Example 1, reference Figure 1 This embodiment provides a modeling method for a radio frequency MOS device, comprising the following steps:

[0096] Step S1, physical field simulation to extract dynamic parasitic parameters,

[0097] Obtain the dynamic parasitic parameters and their associated characteristics of RF MOS devices under actual device operating conditions;

[0098] In step S1, the electromagnetic characteristics of the radio frequency MOS device are simulated using the finite element method (FEM), the parasitic inductance and parasitic capacitance parameters under high frequency conditions are extracted, and their dynamic characteristics varying with frequency are analyzed;

[0099] Introducing thermal field simulation analysis, combined with the self-heating effect of the device, to evaluate the dynamic impact of power consumption and temperature rise on performance;

[0100] Combine electromagnetic simulation with thermal field simulation results to establish a dynamic parasitic parameter library for RF MOS devices, including parasitic parameters and their correlation characteristics with frequency, temperature, voltage, and frequency;

[0101] In step S1, the electromagnetic characteristics simulation method is:

[0102] Construct a three-dimensional geometric model of the RF MOS device, including the electromagnetic structure of the gate, source, drain and its surrounding medium. Assume that the relative dielectric constant of the MOS device material is ε r , the conductivity is σ c , a high-frequency signal is applied in the boundary conditions, and the AC voltage V in (t), the input signal frequency is f;

[0103] Extract the parasitic capacitance and use the finite element method to solve the Poisson form of the static or quasi-static Maxwell equations: Where φ is the potential distribution, ε=ε0ε r is the dielectric constant of the medium, ε0 is the vacuum dielectric constant, ε r is the relative dielectric constant, ρ is the space charge density, and is defined as the excitation source;

[0104] Define the total charge Q between the gate and source gs , and calculate the parasitic capacitance C gs : Among them, Q gs =∫ S εE·ndS is the total charge passing through the surface between the gate and the source, E is the electric field strength, dS represents the small area on the surface S, n is the surface normal vector, V in is the input voltage applied to the gate; for frequency dependence, the simulation frequency is gradually swept to [f1,f2,…,f n ], and record C gs (f) changes, generating dynamic response curves;

[0105] In step S1, the parasitic inductance is extracted, and the finite element method is used to solve the magnetic field distribution under high-frequency conditions based on the vector Poisson equation of Maxwell's equations: Where A is the vector magnetic potential, defined as B is the magnetic induction intensity, is the vector differential operator, μ=μ0μ r is the magnetic permeability of the medium, μ0 is the magnetic permeability of vacuum, μ r is the relative magnetic permeability, J is the current density;

[0106] Calculate the gate parasitic inductance L from the magnetic field energy gs : Where H = μ -1 B, where H is the magnetic field strength, I gs is the AC current flowing between the gate and the source, defined as the peak value of the high-frequency current, and V is the magnetic field distribution volume near the gate; the simulation frequency is gradually scanned as [f1, f2, ..., f n], record L gs (f) changes, and obtain the frequency dynamic response curve;

[0107] Specifically, the parasitic inductance calculation is expanded to the integral form of the magnetic field energy distribution, combined with the dynamic frequency response to achieve accurate modeling of high-frequency nonlinear parasitic effects. At the same time, the coupling relationship between the input signal voltage and the total charge is combined when extracting the capacitance, thereby improving the simulation accuracy.

[0108] In step S1, the steps of establishing a dynamic parasitic parameter library include:

[0109] According to the thermal field distribution and based on the heat conduction equation, the temperature field inside the device is solved:

[0110] Where κ is the thermal conductivity of the material, P gen =I 2 R j is the power density, where I is the current flowing through the device, R j is the junction resistance, ρ c is the material density, C p is the specific heat capacity, Θ is the temperature field distribution;

[0111] In steady state Obtain the steady-state temperature distribution of the MOS device. Under dynamic conditions, analyze the impact of the input signal frequency f on the device temperature fluctuation and obtain the temperature response Θ(f).

[0112] Set the bottom of the device as the heat dissipation boundary, and the boundary condition Θ boundary is the ambient temperature Θ ambient ;

[0113] Assume that the temperature field obtained by simulation is Θ(x,y,z), and use the temperature field Θ(x,y,z) and power density P gen Analyze the nonlinear relationship between power consumption and temperature rise: Among them, ΔΘ is the temperature rise of the device. The formula is divided into two parts to describe the influence of heat dissipation and heat conduction on temperature rise. h is the heat dissipation coefficient, A s is the heat dissipation surface area;

[0114] Define the effect of temperature rise on capacitance and inductance C gs (Θ) and L gs (Θ):

[0115] C gs (Θ)=C gs0 [1+α C (Θ-Θ ambient )], where C gs0 is the initial gate-source capacitance, α C is the temperature coefficient;

[0116] L gs (Θ)=L gs0 [1-α L (Θ-Θ ambient )], where L gs0 is the initial gate-source inductance, α L is the temperature coefficient;

[0117] The parasitic parameters C extracted from electromagnetic simulation gs (f),L gs (f) and the temperature response Θ(f) are mapped into the associated characteristics:

[0118] C gs (f,Θ)=C gs0 [1+α C (Θ(f)-Θ ambient )],

[0119] L gs (f,Θ)=L gs0 [1-α L (Θ(f)-Θ ambient )], the input signal frequency f is swept from low frequency to the upper limit of the working frequency band, and the parameter dynamic response curve is obtained;

[0120] Combine the correlation characteristics of frequency, temperature and voltage to build a dynamic parasitic parameter library:

[0121] Dynamic parasitic capacitance: C gs (f,V,Θ)=G gs0 [1+α C (Θ-Θ ambient )+β C V], where β C is the voltage coefficient, which describes the effect of bias voltage on parasitic capacitance;

[0122] Dynamic parasitic inductance: L gs (f,V,Θ)=L gs0 [1-α L (Θ-Θ ambient )-β L V], where β L is the voltage coefficient, which describes the effect of bias voltage on parasitic inductance;

[0123] The C obtained by simulation gs (f,V,Θ) and L gs (f, V, Θ) is mapped to a dynamic parameter library;

[0124] Specifically, through the dynamic characteristic chain of power consumption → temperature rise → parasitic parameters, the dynamic correlation under high-frequency complex conditions is quantified, and a multi-dimensional modeling formula of dynamic parameters based on temperature rise and bias voltage is proposed to comprehensively characterize the changing pattern of parasitic effects under actual working conditions.

[0125] Step S2: combining physical characteristic parameters with equivalent circuit modeling.

[0126] Integrating the dynamic parasitic parameters extracted in step S1 into the equivalent circuit model of the radio frequency MOS device;

[0127] In step S2, a small-signal equivalent circuit model is established based on dynamic parasitic parameters, and correlation features are added to the equivalent circuit model. The equivalent circuit model serves as a basic framework for modeling RF MOS devices.

[0128] Based on the physical characteristics and circuit behavior, i.e., the parasitic parameters under different conditions in step S1, the currents of the gate, source, and drain are partitioned and modeled. The partitioned modeling refines the dynamic behavior in the basic framework.

[0129] Based on the equivalent circuit model and partition modeling, the frequency response function of the MOS device is derived to predict the performance in a wide frequency band.

[0130] In step S2, the physical characteristic parameters are combined with the equivalent circuit modeling as follows:

[0131] Construct a small signal equivalent circuit model and transform the dynamic parasitic parameters extracted in step S1, dynamic parasitic capacitance C gs (f, V, Θ) and gate-drain capacitance C gd (f, V, Θ) introduces a small signal equivalent circuit:

[0132] Among them, Z gs (f, V, Θ) is the dynamic impedance between the gate and the source, f is the input signal frequency, and j is the imaginary unit;

[0133] Among them, Z gd (f, V, Θ) is the dynamic impedance between gate and drain;

[0134] In the equivalent circuit, the dynamic parasitic parameters are characterized as impedance elements and are related to the transconductance current g. m v gs Together with other non-parasitic characteristics, it forms the small signal framework, where: Among them, i d is the drain current, g m is the transconductance, defined as v gs is the gate-source voltage, V gd is the gate-drain voltage;

[0135] In step S2, gate, source, and drain partition modeling is performed:

[0136] Gate dynamic behavior includes signal drive and parasitic effects, equivalent current: Among them, i gs is the gate-source current, is the time derivative of the input signal;

[0137] The dynamic characteristics of the source part are determined by the dynamic parasitic inductance L s Description, the source current is:

[0138] Among them, V s is the source voltage, Z s is the source impedance, R s is the source equivalent resistance;

[0139] The dynamic behavior of the drain section includes transconductance and parasitic effects:

[0140] Among them, g m v gs is the transconductance current, is the dynamic current introduced by parasitic capacitance;

[0141] According to the partition modeling results, calculate the input impedance Z in :

[0142] Among them, Z in is the input impedance of the RF MOS device;

[0143] Drain current i d The gate-source voltage v gs The relationship describing the gain characteristic is:

[0144] Among them, H(f) is the gain function, the first term g m is the contribution of transconductance to gain, and the second term 1 / Z gd is the effect of parasitic impedance on gain;

[0145] According to the frequency characteristics of the total impedance, the phase response function is derived:

[0146] Where φ(f) is the phase difference between the input signal and the drain output signal, Im(Z in ) is the imaginary part of the input impedance, Re(Z in ) is the real part of the input impedance;

[0147] Specifically, in step S2, dynamic parasitic parameters are integrated into the equivalent circuit to reflect the coupled effects of frequency, temperature rise and bias voltage on circuit performance, and partition modeling is used to refine the dynamic behavior of the gate, source and drain, thereby enhancing the accuracy of the small signal model. At the same time, the frequency response function comprehensively describes the changes in gain and phase over a wide frequency band, which can be directly used to predict the high-frequency performance of RF MOS devices.

[0148] Step S3, nonlinear characteristics analysis,

[0149] Based on the equivalent circuit model established in step S2 and combined with the dynamic parasitic parameters extracted in step S1, the nonlinear characteristics of the RF MOS device under high power and strong signal excitation are analyzed, and a unified linear and nonlinear behavioral model, namely the RF MOS device model, is constructed;

[0150] In step S3, based on the equivalent circuit model in step S2, nonlinear characteristics are added and large signal excitation conditions are introduced;

[0151] Extract intermodulation distortion (IMD), harmonic distortion (HD), and power compression point (P1dB) nonlinear indicators. Based on dynamic parameters and nonlinear indicators, the Volterra series method is used to construct a nonlinear behavior model.

[0152] Combining the nonlinear behavior model with the linear equivalent circuit model of step S2 to form a radio frequency MOS device model;

[0153] The MOS device model can simultaneously predict the linear characteristics in a wide frequency band and the nonlinear characteristics under strong signals;

[0154] In step S3, the steps of constructing a nonlinear model of a radio frequency MOS device include:

[0155] Add large signal excitation and nonlinear characteristic expansion, apply large signal input v in (t):

[0156] v in (t) = V0cos(2πf0t) + V1cos(2πf1t), where V0 and V1 are the amplitudes of the fundamental frequency f0 and the interference frequency f1, and f0 and f1 are the fundamental frequency and the interference frequency,

[0157] Introducing nonlinear characteristics, using nonlinear polynomials to represent the drain current i d The nonlinear response of:

[0158] Among them, g1, g2, g3 are the first-order, second-order, and third-order derivative coefficients, defined as: Among them, v gs is the gate-source voltage;

[0159] Extract intermodulation distortion IMD. The mixing of f0 and f1 in the input signal will produce intermodulation frequency components:

[0160] f IMD =2f0-f1,2f1-f0, where f IMD is the intermodulation frequency component;

[0161] The second-order and third-order nonlinear terms contribute the following intermodulation currents respectively:

[0162] i IMD2 =g2V0V1cos(2πf IMD t), where i IMD2 is the intermodulation component caused by the second-order nonlinearity,

[0163] Among them, i IMD3 is the intermodulation component caused by the third-order nonlinearity;

[0164] Extract harmonic distortion HD, the harmonic frequency of the large signal input is f HD =nf0,n=2,3,4,f HD is the harmonic frequency,

[0165] Calculate the harmonic component amplitude using the nonlinear term contribution: i HD2 is the second harmonic component; i HD3 is the third harmonic component;

[0166] Extract the power compression point P1dB, the power gain G changes with the input power P in It manifests as:

[0167] G(P in )=G0-kP in , where G0 is the small signal gain and k is the nonlinear compression factor;

[0168] When the gain drops by 1dB, the input power compression point P in,1dB satisfy: P in,1dB is the input power compression point;

[0169] In step S3, a nonlinear behavior model is constructed based on the Volterra series:

[0170] The nonlinear behavior is determined by the Volterra kernel H n Definition, the input and output relationship is:

[0171] Among them, H n is the n-order Volterra kernel, v gsis the gate-source voltage, dτ1…dτ n Represents the time variables τ1,…,τ n Differential elements for multiple integration, τ1,…,τ n The gate-source voltage v gs Delay time at different moments;

[0172] For second-order and third-order nonlinear behaviors, H2(τ1,τ2)=g2δ(τ1-τ2), where H2 is the second-order Volterra kernel, representing the second-order nonlinear response of the input signal;

[0173] H3(τ1,τ2,τ3)=g3δ(τ1-τ2)δ(τ2-τ3), where H3 is the third-order Volterra kernel, representing the third-order nonlinear response of the input signal; δ is the Dirac delta function;

[0174] Combining the linear equivalent circuit model with the nonlinear Volterra model, the complete response is:

[0175] i d (t) = g1v gs (t)+∫∫H2(τ1,τ2)v gs (t-τ1)v gs (t-τ2)dτ1dτ2+…, where the first term is the linear contribution and the higher-order terms are the nonlinear contributions;

[0176] Specifically, in step S2, the nonlinear characteristics of the RF MOS device under large signal conditions are analyzed, and a high-order nonlinear behavior model is established using the Volterra series. This is combined with the linear equivalent circuit to form a complete RF MOS device model. At the same time, key performance indicators such as IMD, HD, and P1dB are extracted to provide a basis for nonlinear analysis within a wide frequency band.

[0177] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A modeling method for a radio frequency MOS device, characterized by: include, Step S1, physical field simulation to extract dynamic parasitic parameters, Under the actual working conditions of the device, the dynamic parasitic parameters of the RF MOS device and their correlation characteristics with frequency, temperature and voltage are obtained through electromagnetic-thermal field coupling simulation; Step S2: combining physical characteristic parameters with equivalent circuit modeling. Integrate the dynamic parasitic parameters extracted in step S1 into the equivalent circuit model of the RF MOS device, and refine the dynamic behavior based on partition modeling; Step S3, nonlinear characteristics analysis, Based on the equivalent circuit model established in step S2 and combined with the dynamic parasitic parameters extracted in step S1, the nonlinear characteristics of the RF MOS device under high power and strong signal excitation are analyzed, and a unified linear and nonlinear behavioral model is constructed using the Volterra series method to form an RF MOS device model; In step S1, the electromagnetic characteristics of the radio frequency MOS device are simulated using the finite element method (FEM), the parasitic inductance and parasitic capacitance parameters under high frequency conditions are extracted, and their dynamic characteristics varying with frequency are analyzed; Introducing thermal field simulation analysis, combined with the self-heating effect of the device, to evaluate the dynamic impact of power consumption and temperature rise on performance; Combine electromagnetic simulation with thermal field simulation results to establish a dynamic parasitic parameter library for RF MOS devices, including parasitic parameters and their correlation characteristics with frequency, temperature, and voltage; In step S2, a small-signal equivalent circuit model is established based on dynamic parasitic parameters, and correlation characteristics are added to the equivalent circuit model. The equivalent circuit model serves as a basic framework for modeling RF MOS devices. Based on the physical characteristics and circuit behavior, i.e., the parasitic parameters under different conditions in step S1, the currents of the gate, source, and drain are partitioned and modeled. The partitioned modeling refines the dynamic behavior in the basic framework. Based on the equivalent circuit model and partition modeling, the frequency response function of the MOS device is derived to predict the performance in a wide frequency band. In step S3, based on the equivalent circuit model in step S2, nonlinear characteristics are added and large signal excitation conditions are introduced; Extract intermodulation distortion (IMD), harmonic distortion (HD), and power compression point (P1dB) nonlinear indicators. Based on dynamic parameters and nonlinear indicators, the Volterra series method is used to construct a nonlinear behavior model. The nonlinear behavior model is combined with the linear equivalent circuit model of step S2 to form a radio frequency MOS device model.

2. The modeling method of a radio frequency MOS device according to claim 1, characterized in that: In step S1, the electromagnetic characteristics simulation method is: Construct a three-dimensional geometric model of the RF MOS device, including the electromagnetic structure of the gate, source, drain and its surrounding medium. Assume that the relative dielectric constant of the MOS device material is ε r , the conductivity is σ c , a high-frequency signal is applied in the boundary conditions, and the AC voltage V in (t), the input signal frequency is f; Extract the parasitic capacitance and use the finite element method to solve the Poisson form of the static or quasi-static Maxwell equations: Where φ is the potential distribution, ε=ε0ε r is the dielectric constant of the medium, ε0 is the vacuum dielectric constant, ε r is the relative dielectric constant, ρ is the space charge density, and is defined as the excitation source; Define the total charge Q between the gate and source gs , and calculate the parasitic capacitance C gs : Among them, Q gs =∫ S εE·ndS is the total charge passing through the surface between the gate and the source, E is the electric field strength, dS represents the small area on the surface S, n is the surface normal vector, v in is the input voltage applied to the gate; for frequency dependence, the simulation frequency is gradually swept to [f1,f2,…,f n ], and record C gs (f) changes, generating dynamic response curves; In step S1, the parasitic inductance is extracted, and the finite element method is used to solve the magnetic field distribution under high-frequency conditions based on the vector Poisson equation of Maxwell's equations: Where A is the vector magnetic potential, defined as B is the magnetic induction intensity, is the vector differential operator, μ=μ0μ r is the magnetic permeability of the medium, μ0 is the magnetic permeability of vacuum, μ r is the relative magnetic permeability, J is the current density; Calculate the gate parasitic inductance L from the magnetic field energy gs : Where H = μ -1 B, where H is the magnetic field strength, I gs is the AC current flowing between the gate and the source, defined as the peak value of the high-frequency current, and V is the magnetic field distribution volume near the gate; the simulation frequency is gradually scanned as [f1, f2, ..., f n ], record L gs (f) changes, and obtain the frequency dynamic response curve.

3. The modeling method of a radio frequency MOS device according to claim 2, characterized in that: In step S1, the steps of establishing a dynamic parasitic parameter library include: According to the thermal field distribution and based on the heat conduction equation, the temperature field inside the device is solved: Where κ is the thermal conductivity of the material, P gen =I 2 R j is the power density, where I is the current flowing through the device, R j is the junction resistance, ρ c is the material density, C p is the specific heat capacity, Θ is the temperature field distribution; In steady state Obtain the steady-state temperature distribution of the MOS device. Under dynamic conditions, analyze the impact of the input signal frequency f on the device temperature fluctuation and obtain the temperature response Θ(f). Set the bottom of the device as the heat dissipation boundary, and the boundary condition Θ boundary is the ambient temperature Θ ambient ; Assume that the temperature field obtained by simulation is Θ(x,y,z), and use the temperature field Θ(x,y,z) and power density P gen Analyze the nonlinear relationship between power consumption and temperature rise: Among them, ΔΘ is the temperature rise of the device. The formula is divided into two parts to describe the influence of heat dissipation and heat conduction on temperature rise. h is the heat dissipation coefficient, A s is the heat dissipation surface area; Define the effect of temperature rise on capacitance and inductance C gs (Θ) and L gs (Θ): C gs (Θ)=C gs0 [1+α C (Θ-Θ ambient )], where C gs0 is the initial gate-source capacitance, α C is the temperature coefficient; L gs (Θ)=L gs0 [1-α L (Θ-Θ ambient )], where L gs0 is the initial gate-source inductance, α L is the temperature coefficient; The parasitic parameters C extracted from electromagnetic simulation gs (f),L gs (f) and the temperature response Θ(f) are mapped into the associated characteristics: C gs (f,Θ)=C gs0 [1+a C (Θ(f)-Θ ambient )], L gs (f,Θ)=L gs0 [1-α L (Θ(f)-Θ ambient )], the input signal frequency f is swept from low frequency to the upper limit of the working frequency band, and the parameter dynamic response curve is obtained; Combine the correlation characteristics of frequency, temperature and voltage to build a dynamic parasitic parameter library: Dynamic parasitic capacitance: C gs (f,V,Θ)=C gs0 [1+α C (Θ-Θ ambient )+β C V], where β C is the voltage coefficient, which describes the effect of bias voltage on parasitic capacitance; Dynamic parasitic inductance: L gs (f,V,Θ)=L gs0 [1-α L (Θ-Θ ambient )-β L V], where β L is the voltage coefficient, which describes the effect of bias voltage on parasitic inductance; The C obtained by simulation gs (f,V,Θ) and L gs (f, V, Θ) is mapped into a dynamic parameter library.

4. The modeling method of a radio frequency MOS device according to claim 3, characterized in that: In step S2, the physical characteristic parameters are combined with the equivalent circuit modeling as follows: Construct a small signal equivalent circuit model and transform the dynamic parasitic parameters extracted in step S1, dynamic parasitic capacitance C gs (f, V, Θ) and gate-drain capacitance C gd (f, V, Θ) introduces a small signal equivalent circuit: Among them, Z gs (f, V, Θ) is the dynamic impedance between the gate and the source, f is the input signal frequency, and j is the imaginary unit; Among them, Z gd (f, V, Θ) is the dynamic impedance between gate and drain; In the equivalent circuit, the dynamic parasitic parameters are characterized as impedance elements and are related to the transconductance current g. m v gs Together with other non-parasitic characteristics, it forms the small signal framework, where: Among them, i d is the drain current, g m is the transconductance, defined as v gs is the gate-source voltage, V gd is the gate-drain voltage.

5. The modeling method of a radio frequency MOS device according to claim 4, characterized in that: In step S2, gate, source, and drain partition modeling is performed: Gate dynamic behavior includes signal drive and parasitic effects, equivalent current: Among them, i gs is the gate-source current, is the time derivative of the input signal; The dynamic characteristics of the source part are determined by the dynamic parasitic inductance L s Description, the source current is: Among them, V s is the source voltage, Z s is the source impedance, R s is the source equivalent resistance; The dynamic behavior of the drain section includes transconductance and parasitic effects: Among them, g m v gs is the transconductance current, is the dynamic current introduced by parasitic capacitance; According to the partition modeling results, calculate the input impedance Z in : Among them, Z in is the input impedance of the RF MOS device; Drain current i d The gate-source voltage v gs The relationship describing the gain characteristic is: Among them, H(f) is the gain function, the first term g m is the contribution of transconductance to gain, and the second term 1 / Z gd is the effect of parasitic impedance on gain; According to the frequency characteristics of the total impedance, the phase response function is derived: Where φ(f) is the phase difference between the input signal and the drain output signal, Im(Z in ) is the imaginary part of the input impedance, Re(Z in ) is the real part of the input impedance.

6. A modeling method for a radio frequency MOS device according to claim 5, characterized in that: In step S3, the steps of constructing a nonlinear model of a radio frequency MOS device include: Add large signal excitation and nonlinear characteristic expansion, apply large signal input v in (t): v in (t) = V0cos(2πf0t) + V1cos(2πf1t), where V0 and V1 are the amplitudes of the fundamental frequency f0 and the interference frequency f1, and f0 and f1 are the fundamental frequency and the interference frequency, Introducing nonlinear characteristics, using nonlinear polynomials to represent the drain current i d The nonlinear response of: Among them, g1, g2, g3 are the first-order, second-order, and third-order derivative coefficients, defined as: Among them, v gs is the gate-source voltage; Extract intermodulation distortion IMD. The mixing of f0 and f1 in the input signal will produce intermodulation frequency components: f IMD =2f0-f1,2f1-f0, where f IMD is the intermodulation frequency component; The second-order and third-order nonlinear terms contribute the following intermodulation currents respectively: i IMD2 =g2V0V1cos(2πf IMD t), where i IMD2 is the intermodulation component caused by the second-order nonlinearity, Among them, i IMD3 is the intermodulation component caused by the third-order nonlinearity; Extract harmonic distortion HD, the harmonic frequency of the large signal input is f HD =nf0,n=2,3,4,f HD is the harmonic frequency, Calculate the harmonic component amplitude using the nonlinear term contribution: i HD2 is the second harmonic component; i HD3 is the third harmonic component; Extract the power compression point P1dB, the power gain G changes with the input power P in It manifests as: G(P in )=G0-kP in , where G0 is the small signal gain and k is the nonlinear compression factor; When the gain drops by 1dB, the input power compression point P in,1dB satisfy: P in,1dB is the input power compression point.

7. A modeling method for a radio frequency MOS device according to claim 6, characterized in that: In step S3, a nonlinear behavior model is constructed based on the Volterra series: The nonlinear behavior is determined by the Volterra kernel H n Definition, the input and output relationship is: Among them, H n is the n-order Volterra kernel, v gs is the gate-source voltage, dτ1…dτ n Represents the time variables τ1,…,τ n The differential elements for multiple integration, τ1,…,τn are the gate-source voltage v gs Delay time at different moments; For second-order and third-order nonlinear behaviors, H2(τ1,τ2)=g2δ(τ1-τ2), where H2 is the second-order Volterra kernel, representing the second-order nonlinear response of the input signal; H3(τ1,τ2,τ3)=g3δ(τ1-τ2)δ(τ2-τ3), where H3 is the third-order Volterra kernel, representing the third-order nonlinear response of the input signal; δ is the Dirac delta function; Combining the linear equivalent circuit model with the nonlinear Volterra model, the complete response is: i d (t) = g1v gs (t)+∫∫H2(τ1,τ2)v gs (t-τ1)v gs (t-τ2)dτ1dτ2+…, where the first term is the linear contribution and the higher-order terms are the nonlinear contributions.

Images