JP2000322456A - Method and device for extracting model parameter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an extracting method for extracting the value of a parameter with high physical property in the case of extracting the model parameter of BSIM3 as one of MOS models. SOLUTION: Concerning this extracting method, intermediate data 2 such as the short channel effect, narrow channel effect, substrate effect, sub-threshold swing characteristics or surface punch through characteristics of a device from data 1 of measured Vds-Ids and Vgs-Ids. Parameter values corresponding to respective characteristics are extracted 5 from the value of the generated intermediate data and separately measured device structure data 4. Thus, the model parameter can be extracted even the value of high physical property and when one part of device conditions is changed, changed characteristics can be accurately predicted only by changing the parameter value corresponding to the changed conditions on the basis of the extracted parameter value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for extracting MOSFET (hereinafter referred to as MOS) model parameters used for circuit simulation.

[0002]

2. Description of the Related Art At present, as one of MOS models, the University of California, Berkeley (hereinafter referred to as UCB)
BSIM3 model has been developed and is establishing its position as an international standard model. The features of this model are the short channel effect, narrow channel effect, substrate effect, etc.
Since a model formula representing the physical phenomenon of the OS is used, the model has high physical properties.

[0003] However, even if the physical properties of the model itself are high,
Depending on the parameter extraction method, the parameter value itself does not conform to the physical phenomenon. As a conventional model parameter extraction method, there is a model parameter extraction method as shown in the outline processing flow of FIG. Vds-Ids (Drain Current vs. Drain-Source Voltage) Characteristics and Vgs-Ids of MOS Devices of Multiple Sizes
(Drain current with respect to voltage between gate and source) characteristics are measured, and values of model parameters are extracted so as to match the measured waveforms. This extraction method is "B
According to the description in Chapter 6 of “SIM3v3 manual”, three parameters are selected at a time, and the Vds-Ids characteristic, V
This is a method of determining model parameters so that an error between the gs-Ids characteristic and the measured waveform is minimized.

That is, as shown in FIG. 16, a device characteristic measurement data 1 of the measured Vds-Ids characteristic and Vgs-Ids characteristic and a device structure data 4 of the MOS used for the measurement are obtained by the measurement. The above voltage-
A parameter extraction process 5 for extracting a model parameter value so as to match a waveform of a current characteristic is performed, and a model parameter set 6 of the BSIM3 used for circuit simulation is obtained.

[0005]

In the above-described conventional model parameter extraction method, although the measured device characteristics match the characteristic results simulated using the extracted model parameter values, the values of the respective model parameters are obtained. Is not a value indicating a physical phenomenon originally meant by the model parameter.

Further, in the above simulation, waveform matching is performed using fitting parameters. Therefore, even when trying to estimate the characteristics when the device conditions such as the channel length and the threshold value are changed based on the extracted model parameters by changing the corresponding model parameters, it is difficult to estimate the characteristics with the actual device characteristics. There was a problem that would be significantly different.

Therefore, an object of the present invention is to change a model parameter value corresponding to a specific device condition when a part of the device condition is changed even though the model parameter is extracted under a specific device condition. It is an object of the present invention to provide a model parameter extracting method and apparatus that can accurately predict a device characteristic after a change by simply performing the operation.

[0008]

In order to solve the above-mentioned problems, a model parameter extracting method according to the present invention is directed to a method for extracting a model parameter value of a MOS device used when performing a circuit simulation. From the Vds-Ids characteristics and Vgs-Ids characteristics measured from the MOS device, intermediate data of device characteristics such as a short channel effect, a narrow channel effect, a substrate effect, a sub-threshold swing characteristic, and a surface punch-through characteristic of the device are generated. The method is characterized in that the values of model parameters corresponding to the respective device characteristics of the intermediate data are determined independently.

Further, the model parameter extracting apparatus according to the present invention comprises: a measurement data input section for inputting and storing measured MOS device characteristic measurement data;
A device structure data input unit for inputting and storing the device structure data of S; an intermediate data generating unit for generating predetermined device characteristic intermediate data from the device characteristic measurement data stored in the measurement data input unit; From the predetermined device characteristic intermediate data and the device structure data stored in the device structure data input unit, parameter values for each model parameter of the MOS device model assigned in advance with respect to the predetermined device characteristics are independently extracted. It is characterized by having at least a parameter extraction unit and a parameter set output unit that outputs a set of values of each model parameter extracted by the parameter extraction unit for the circuit simulation of the measured MOS.

In the above-described model parameter extraction device, a first display unit for comparing and displaying device characteristic measurement data and a simulation result of device characteristics simulated by using the parameter values extracted by the parameter extraction unit, A second display unit for comparing and displaying the measurement result / simulation result display unit, the intermediate data, and the simulation result of the device intermediate characteristic simulated using the parameter value extracted by the parameter extraction unit, that is, the intermediate data simulation result It is preferable to further provide a display unit.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the method and apparatus for extracting model parameters according to the present invention will be described below using specific examples.

FIG. 1 is a schematic processing flow chart showing a method for extracting model parameters according to the present invention. The model parameter extraction method of this embodiment is based on the BSIM developed by UCB.
This is a method of extracting a set of model parameters used as model parameters of static characteristics of three models, that is, a model parameter set 6. In FIG. 1, the same parts as those in the conventional outline processing flow shown in FIG. 16 are denoted by the same reference numerals.

The outline of this processing flow is as follows. First, from the device characteristics measurement data 1 of Vds-Ids characteristics and Vgs-Ids characteristics stored in a storage device such as a hard disk, a short channel effect, a narrow channel effect, a substrate effect, a sub-threshold swing characteristic, a surface punch-through characteristic, and the like of a device. The processing of the intermediate data generation function 2 for generating the intermediate data of the device characteristics is executed, and the obtained intermediate data 3 of the device characteristics is stored as an intermediate file in a storage device such as a hard disk. Next, the intermediate data 3 of the device characteristics stored in this intermediate file and the separately measured MOS device structure such as the channel length, oxide film thickness, and profile stored in a storage device such as a hard disk are also shown. Each model parameter of the BSIM3 model is extracted from the related device structure data 4 to obtain a model parameter set 6. The device characteristic data and the device structure data are separately measured in advance by a known device characteristic measuring device such as a curve tracer or a tester or a known analyzing device such as a scanning electron microscope (SEM).

An intermediate data generating process 2 of the device characteristics as described above, a process flow of storing the generated intermediate data 3 as an intermediate file in a storage device, and extracting parameters using the intermediate data 3 and the device structure data 4 Is different from the processing flow shown in FIG. 16 of the conventional example.

Hereinafter, a method of extracting each model parameter will be described.
This will be described in more detail with reference to FIG. 17 and FIGS. Here, FIG. 17 shows the classified model parameters and
FIG. 2 is a list showing model parameters to be extracted.
FIG. 13 to FIG. 13 respectively show the device characteristics of the intermediate data 3 for extracting the model parameters and the characteristics of the simulation result, and the symbols in the squares in each figure show the names of the model parameters to be extracted from the illustrated characteristics. I have.

In the present embodiment, the values of the model parameters are determined from the intermediate data 3 stored in the intermediate file and the device structure data 4 for the BSIM3 model parameters hierarchically grouped in advance. Hereinafter, the model parameter classification method and the extraction method will be described in detail.

First, a method of classifying the model parameters of the BSIM3 hierarchically will be described. First, the model parameters are classified into four groups: physical phenomenon parameters, process parameters, constant parameters, and fitting parameters. Then, as shown in the column of “parameter type” in FIG. 17, the group of physical phenomenon parameters is subdivided into two groups: a model parameter related to a threshold value and a model parameter related to a current.

Further, the model parameters relating to the threshold value are subdivided into groups of physically meaningful model parameters representing physical phenomena such as short channel effect, narrow channel effect, substrate effect, sub-threshold swing characteristic, and surface punch-through characteristic. I do.

The model parameters relating to the current are also divided into model parameters relating to long-channel characteristics, short-channel characteristics, and narrow-channel characteristics having physical meaning.

The process parameters of the second group are subdivided into two groups of model parameters relating to the structure and model parameters relating to the process profile. Further, regarding the model parameters relating to the structure, the “physical meaning” in the table of FIG. ”, The model parameters are divided into model parameters relating to the actual channel length and model parameters relating to the oxide film thickness.

The third group of constant parameters is divided into a model parameter relating to mobility and a model parameter relating to saturation velocity.

The fitting parameters of the fourth group include model parameters other than those described above.

As described above, the model parameters of the BSIM3 are classified hierarchically. In the table of FIG.
Also shown are the parameter names of model parameters to be extracted corresponding to the physical phenomena shown in the column of “physical meaning”, and device characteristics necessary for extracting the model parameters.

Next, a method for extracting model parameters will be described. A method of extracting model parameters of a group of physical phenomenon parameters is as follows.
The ds characteristic and the Vgs-Ids characteristic are measured. At this time, V
The ds-Ids characteristic is measured under the condition that the substrate voltage is applied, and the Vgs-Ids characteristic is measured under the condition that the drain-source voltage is low and high, respectively. Then, a series of device physical phenomena are analyzed from the measured data, and parameter values are extracted from the analysis results.

First, a method of analyzing a phenomenon regarding a threshold value from measured data and extracting a parameter value from the analysis result will be described below.

Regarding the substrate effect, the threshold value Vth when the substrate voltage Vbs of a long channel device (for example, a device having a channel length of 10 μm or more) is changed is plotted (FIG. 2), and at the same time, the substrate effect of the BSIM3 model is plotted. The expression (the first and second lines of the equation (1) shown in FIG. 18; the equation (1) corresponds to the equation (2.1.25) of the BSIM3v3 manual) is calculated, and as shown in FIG. , The values of the model parameters VTH0, K1, K2 of the BSIM3 are determined so that both the measured value and the calculated value match.

In FIG. 2, points indicate measured values and lines indicate calculated values, and the same applies to the following drawings. 18 to FIG.
The equations used in this embodiment shown in FIG.
The formulas in the three manuals are quoted.

With respect to the sub-threshold swing characteristic, the gate-source voltage Vg is obtained from the Vgs-Ids characteristics of devices having the same channel width and different channel lengths L.
The slope of the waveform where s is small is calculated, and the slope with respect to the channel length L, that is, the sub-threshold swing S is plotted (FIG. 3). In addition, the same processing is performed for a substrate whose substrate voltage Vbs is changed. In the case of FIG. 3, the substrate voltages Vbs = 0 and Vbs = −Vcc. Where V
cc is a power supply voltage. Then, a model equation representing the sub-threshold swing characteristic (equation (2) shown in FIG. 19; equation (2) corresponds to equation (2.7.3) in the BSIM3v3 manual) is calculated and shown in FIG. Thus, the model parameters CDSC, CDSCD, CDSCB, DVT1, NFA of BSIM3
Determine the value of CTOR.

As for the short channel effect, the channel width W
Vgs of a plurality of devices having the same but different channel lengths L
-Ids characteristics (for a low drain voltage, e.g.
05V), and the threshold value Vth is plotted against the channel length L (FIG. 4). In addition, the same processing is performed for a substrate whose substrate voltage Vbs is changed. In the case of FIG. 4, the substrate voltage Vbs = 0
And Vbs = −Vcc. And at the same time BSIM3
A term representing the short channel effect of the model (lines 3 and 6 in equation (1) shown in FIG. 18) was calculated, and as shown in FIG.
The model parameters N of the BSIM3 are set so that they match.
Determine the values of LX and DVT0.

Surface punch-through characteristics (DIBL:
Drain Induced Barrier Lowering characteristic), the difference DIB between the threshold value and the drain-source voltage Vds different from the Vgs-Ids characteristic of the device having the same channel width.
L is plotted against the channel length L (FIG. 5). In addition, the same processing is performed for a substrate whose substrate voltage Vbs is changed. In the case of FIG. 5, the substrate voltage Vbs = 0 and V
bs = −Vcc. Then, a term representing the surface punch-through characteristic of the threshold value of the model equation (the seventh line of the equation (1) shown in FIG. 18) is calculated, and as shown in FIG. Model parameter DSU
Determine the values of B, ETA0, ETAB.

With respect to the narrow channel effect, the value of the threshold value Vth with respect to the channel width W of devices having the same channel length and different channel widths is analyzed and plotted from the Vgs-Ids characteristic (FIG. 6). Further, the same processing is performed for a substrate whose substrate voltage is changed. Then, a term representing the narrow channel effect of the model equation (4 of equation (1) shown in FIG. 18)
6), and as shown in FIG. 6, the model parameters K3, K3B, W0
Determine the value of

Next, parameters for determining the current of the group of physical phenomena and a method of extracting the parameters will be described below. Regarding the bulk charge effect, Vds-
It can be determined from the Ids characteristics. The model parameters of BSIM3 relating to this effect include A0, AGS, K
There is ETA. The model parameters A0 and AGS are the Vds-Ids characteristics of the long channel device (substrate voltage 0 V).
Is the model formula (Formula (3) and Formula (4) shown in FIG. 20. Formula (3) is the formula (3.6.1) of the BSIM3v3 manual.
Equation (4) corresponds to B.3 in Appendix B of the BSIM3v3 manual. This corresponds to the formula described in page B-5 of 1.6 Drain Current Formula. ) Is determined to match the calculated value (FIG. 7). Also, the model parameter KETA
Is determined so that the Vds-Ids characteristic (the value obtained by multiplying the substrate voltage) matches the value calculated by the equations (3) and (4) of the model equation (FIG. 7).

Items that determine the short-channel current include:
There is an effective channel length and parasitic resistance. The effective channel length Leff is the model parameter LIN of the BSIM3.
T, and its extraction method is described in Chern (Cher
n) using the method proposed by I.I.E.E. Electron Device Letter EDL-1, No. 9, 1980, pp. 170-173 (IEEE
Electron Device Letter, VOL. EDL-1. NO.9, SEPTEMBE
R, 1980, pp. 170-173, `` A New Method To Determine MO
SFET Channel Length ”)). That is, the channel width W
, The drain-source voltage Vds of a plurality of MOS transistors having different channel lengths L is low, for example, the drain-source resistance R at Vds = 0.1 V
(= Vds / Ids), ΔL = 2 dL is obtained by plotting the resistance R against the channel length L (FIG. 8), and thereby the effective channel length Leff is obtained from Leff = L−2 dL. Then, the dL obtained by the above measurement and the dL of the model formula (the formula (1) shown in FIG.
0). Equation (10) corresponds to equation (2.8.4) in the BSIM3v3 manual. ) Are determined so that the value of the model parameter LINT matches.

As for the value of the parasitic resistance RDSW, the Vds-Ids characteristics of the short channel device are measured values and model expressions (Equations (3) and (4) in FIG. 20) as shown in FIG.
Decide to match with the value calculated in.

The item that determines the narrow channel current has an effective channel width. Effective channel width Weff
Is determined by the BSIM3 model parameter WINT. Regarding the value of WINT, as shown in FIG.
The Vds-Ids characteristics of the narrow channel device are determined so as to match the values calculated by the model equations (Equations (3) and (4) in FIG. 20).

Next, a method of extracting a process parameter which is a second group will be described. First, a method of extracting parameters relating to the structure will be described. The channel length L of the gate electrode is not particularly included in the model parameters of the BSIM3, but is important in extracting process parameters, and thus is determined at the time of extracting other structural parameters. As for the channel length L, as shown in FIG. 11, an actual dimension L is measured from an SEM photograph of a cross section of the measuring device from which model parameters are extracted, and the value is used.

As for the oxide film thickness TOX, an effective oxide film thickness is used instead of an optically measured film thickness.
As shown in FIG. 12, CV measurement is performed by applying a voltage Vgs to the gate of the device, and an oxide film thickness TOX is extracted from a capacitance value C in an inversion region (a region where a high voltage is applied to the gate). .

The model parameters relating to the process profile include channel concentration NCH, substrate concentration NSUB,
Channel ion implantation (channel implantation) depth XT
These are calculated by the equations (5) to (9) shown in FIG. Equation (5) is an equation of a general threshold value. M. This corresponds to equation (69) described on page 201 of "Semiconductor Device-Basic Theory and Process Technology-" (published by Sangyo Tosho Co., Ltd.) by G. Equations (6) and (7) are BS
Equations (2.1.5) and (2.1.5) in the IM3v3 manual
From (1.6), equation (8) is equivalent to equation (2.1.
From equation (9), equation (9) becomes equation (2.1.8) in the manual.
Respectively.

Next, a method of extracting a parameter relating to the third group constant will be described. As for the parameters related to the mobility, the relationship between the electric field Eeff and the mobility is measured using a device TEG (Test Element Group), as shown in FIG. The model parameter U0 of the BSIM3 is set so that the measured value matches the model equation (Equation (2.2.4) and Eq. (2.2.4) in the BSIM3v3 manual).
(= Μ ₀ ), the values of UA, UB, and UC are extracted.

The saturation velocity is treated as a physical constant, and its value is a general value of 1 × 10 ⁵ m / sec.
Use c.

Model parameters other than those described above are:
The parameters are classified as the fitting parameters of the fourth group shown in FIG. 17 and are not used without any particular extraction, or are used with the default values.

Next, the order of extracting the model parameters will be described with reference to the flowchart of FIG. First, in step S1, model parameter values relating to the actual channel length L, the oxide film thickness TOX, the mobility, and the saturation speed are determined. The order of extraction is arbitrary.

Next, in step S2, the effective channel length L
Extract model parameters related to eff.

The values of the model parameters for determining the threshold value Vth are determined in the order of steps S3 to S7, that is, the model parameters relating to the substrate effect, sub-threshold swing characteristic, surface punch-through characteristic, short channel effect, and narrow channel effect. Extract in order.

Further, the values of the model parameters relating to the current are changed in the order of steps S8 to S10, that is,
The model parameters are extracted in the order of the bulk charge effect, the parasitic resistance, and the effective channel width.

FIG. 15 shows a configuration of a model parameter extracting device using the above-described model parameter extracting method. That is, as the device configuration, MOs of a plurality of sizes are used.
MOS device characteristic data measured from S, ie, V
A measurement data input unit 10 for inputting and storing measurement data of ds-Ids characteristics and Vgs-Ids characteristics, a device structure data input unit 11 for inputting and storing measured device structure data of MOS, and extracting parameters. A parameter extraction unit 13, a measurement result / simulation result display unit 14 for comparing and displaying the device characteristic measurement result and a simulation result simulated using the extracted parameter, and a parameter set output for outputting the extracted parameter set to a storage device such as a hard disk. The conventional Vds-Ids characteristic and Vgs-Ids
An intermediate data generation unit 12 for generating intermediate data of device characteristics including device characteristics such as a short channel effect, a narrow channel effect, a substrate effect, a sub-threshold swing characteristic, and a surface punch-through characteristic of the device from the characteristic; and This can be realized by adding an intermediate data / simulation result display section 15 for simultaneously comparing and displaying the generated intermediate data and the simulation results simulated using the parameters extracted from the intermediate data. Such a model parameter extracting device can be configured by a computer system including a workstation, a display device such as a CRT or a liquid crystal, and an external storage device such as a hard disk.

Conventionally, it has been determined that all values of a set of model parameters are adjusted to the measured Vds-Ids characteristics and Vgs-Ids characteristics. On the other hand, in the present invention, model parameters are hierarchically grouped, and intermediate data generated by processing data of the measured Vds-Ids characteristics and Vgs-Ids characteristics, that is, the short channel effect of the device, Narrow channel effect, substrate effect,
The difference is that model parameter values corresponding to each physical phenomenon are extracted from intermediate data such as surface punch-through characteristics.

The preferred embodiment of the present invention has been described above. However, the present invention is not limited to the above-described embodiment, and various design changes can be made without departing from the spirit of the present invention. It is.

[0049]

As is clear from the above-described embodiment, according to the method for extracting model parameters according to the present invention, the short channel effect and the narrow channel effect of the device are obtained from the data of the measured Vds-Ids characteristics and Vgs-Ids characteristics. By generating intermediate data such as effects, substrate effects, sub-threshold swing characteristics, and surface punch-through characteristics, and extracting model parameters corresponding to each characteristic,
A highly physical model parameter value can be extracted. Further, based on the model parameters extracted by this method, model parameters of devices having different conditions can be generated with high accuracy.

In the conventional model parameter extraction device, V
an intermediate data generator for generating intermediate data including device characteristics such as a short channel effect and a narrow channel effect from the ds-Ids characteristic and the Vgs-Ids characteristic, and intermediate data for simultaneously displaying the generated intermediate data and a simulation result; The model parameter extraction device according to the present invention, in which a simulation result display unit is newly incorporated, can extract a model parameter value according to a physical phenomenon. For this reason, even if some device conditions are changed despite the model parameter values extracted under specific conditions,
Only by changing the model parameters corresponding to the changed conditions, it becomes possible to extract model parameters that can accurately predict device characteristics after the change.

[Brief description of the drawings]

FIG. 1 is a diagram showing an outline of a processing flow of a model parameter extracting method of the present invention.

FIG. 2 is a diagram showing device characteristics of intermediate data for extracting model parameters relating to a substrate effect and characteristics of a simulation result.

FIG. 3 is a diagram showing device characteristics of intermediate data for extracting model parameters related to sub-threshold swing characteristics and characteristics of a simulation result;

FIG. 4 is a diagram showing device characteristics of intermediate data for extracting parameters related to a short channel and characteristics of a simulation result.

FIG. 5 is a diagram showing device characteristics of intermediate data for extracting model parameters relating to surface punch-through characteristics and characteristics of simulation results.

FIG. 6 is a diagram showing device characteristics of intermediate data for extracting model parameters related to a narrow channel effect and characteristics of a simulation result.

FIG. 7 is a diagram showing device characteristics of intermediate data for extracting model parameters related to bulk charge and characteristics of a simulation result.

FIG. 8 is a characteristic diagram used for extracting a model parameter relating to an effective channel length.

FIG. 9 is a characteristic diagram used in a method of extracting model parameters related to parasitic resistance.

FIG. 10 is a characteristic diagram used in a method for extracting a model parameter relating to an effective channel width.

FIG. 11 is a schematic diagram of a cross-sectional structure used for extracting model parameters relating to an actual channel length.

FIG. 12 is a characteristic diagram used for extracting model parameters relating to an oxide film thickness.

FIG. 13 is a characteristic diagram used for extracting model parameters related to mobility.

FIG. 14 is a flowchart of an extraction order of model parameters.

FIG. 15 is a diagram showing a schematic configuration of a model parameter extraction device of the present invention.

FIG. 16 is a diagram showing an outline of a processing flow of a conventional model parameter extraction method.

FIG. 17 is a list showing a hierarchical classification of model parameters used in the model parameter extraction method of the present invention.

FIG. 18 is a diagram showing a model formula (1) used for extracting model parameters.

FIG. 19 is a diagram showing a model equation (2) used for extracting model parameters.

FIG. 20 is a diagram showing model equations (3) and (4) used for extracting model parameters.

FIG. 21 is a diagram showing model equations (5) to (9) used for extracting model parameters.

FIG. 22 shows a model equation (1) used for extracting model parameters.
FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Device characteristic measurement data, 2 ... Intermediate data generation function, 3 ... Intermediate data, 4 ... Device structure data, 5 ... Parameter extraction processing, 6 ... Model parameter set, 10
... Measurement data input unit, 11 ... Device structure data input unit, 12 ... Intermediate data generation unit, 13 ... Parameter extraction unit, 14 ... Measurement result / simulation result display unit, 1
5 ... intermediate data simulation result display section, 16 ...
Parameter set output section.

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 2G003 AA02 AB01 AF02 AH00 AH01 AH05 5B046 AA08 BA03 JA04 5F040 DA00 EA00

Claims (5)

[Claims] 1. A model parameter extracting method for extracting parameter values of a MOSFET model for circuit simulation from measured device characteristic measurement data of a MOSFET and device structure data of the MOSFET, wherein a predetermined device is extracted from the device characteristic measurement data. Generating device characteristic intermediate data relating to characteristics; and independently generating parameter values for each model parameter of the MOSFET model assigned in advance for the predetermined device characteristic from the generated device characteristic intermediate data and the device structure data. A method for extracting model parameters, characterized in that: 2. The device characteristic measurement data is Vds-Ids characteristic and Vgs-Ids characteristic static characteristic data measured from a plurality of sizes of the MOSFET, and the predetermined device characteristic intermediate data is a threshold voltage. Channel length dependence, threshold voltage channel width dependence,
2. The model parameter extraction method according to claim 1, wherein the data is data on a substrate voltage dependence of a threshold voltage, a channel length dependence of a sub-threshold swing, and a channel length dependence of a surface punch-through. 3. A measurement data input section for inputting and storing measured device characteristic measurement data of the MOSFET, and inputting measured device structure data of the MOSFET,
A device structure data input section for storing; an intermediate data generating section for generating predetermined device characteristic intermediate data from the device characteristic measurement data stored in the measurement data input section; and the generated predetermined device characteristic intermediate data. A parameter extraction unit that independently extracts parameter values for each parameter of the MOSFET model previously assigned with respect to the predetermined device characteristic from device structure data stored in a device structure data input unit; and A parameter set output unit that outputs a set of values of each model parameter extracted by the parameter extraction unit for circuit simulation of the MOSFET;
A model parameter extraction device characterized by having at least: 4. A first display unit for comparing and displaying the device characteristic measurement data and a simulation result of a device characteristic simulated by using the parameter values extracted by the parameter extraction unit; The model parameter extraction device according to claim 3, further comprising: a second display unit for comparing and displaying a simulation result of device intermediate characteristics simulated using the parameter value extracted by the parameter extraction unit. 5. The device characteristic measurement data is Vds-Ids characteristic and Vgs-Ids characteristic static characteristic data measured from a plurality of sizes of the MOSFETs, and the predetermined device characteristic intermediate data is a threshold voltage. Channel length dependence, threshold voltage channel width dependence,
The model parameter extracting device according to claim 3 or 4, wherein the data is data on a substrate voltage dependence of a threshold voltage, a channel length dependence of a sub-threshold swing, and a channel length dependence of a surface punch-through.