KR20050081264A - Method of extraction for model parameter - Google Patents

Abstract

The present invention relates to a method for extracting a physical MOSFET model parameter set, comprising: defining dies on one or more wafers, fabricating the basic elements of an integrated circuit on the dies, and the basic fabricated on the dies. Measuring the electrical characteristics of the device and statistically analyze the data of the threshold voltage (V _th ), saturation current (I _Dsat ) and linear current (I _Dlin ) of the measured electrical characteristic data to the type and size of the basic device Preparing a scattering curve according to the step of selecting and selecting _physical data of data of a threshold voltage (V _th ), a saturation current (I _Dsat ), and a linear current (I _Dlin ) Selecting dies according to the collected physical data, measuring current (I) -voltage (V) characteristics at the selected dies, and preparing an IV characteristic curve; A by the IV characteristic curve of the objective function to provide a model parameter extraction method comprising the step of extracting model parameters set that can reproduce the IV characteristic curve of the well. In this way, the model parameters extracted in this way accurately reflect the change in process characteristics and process dispersion due to the high integration of MOSFET devices, and can accurately predict future processes.

Description

      Model Parameter Extraction Method {METHOD OF EXTRACTION FOR MODEL PARAMETER}

The present invention relates to a method for extracting a physical MOSFET model parameter set, and more particularly, to represent a type and size of a MOSFET device based on a statistical analysis of electrical characteristic data measured on all manufactured dies. A method of selecting dies and measuring IV characteristics to extract a set of physical MOSFET model parameters.

In the process of developing a semiconductor integrated circuit, whether the designed circuit will function properly and how its performance will be performed, and further, the distribution and deviation of the performance will be statistically simulated in advance, and feedback of the result to the design process is currently It is becoming an almost essential process for the development of semiconductor integrated circuits.

When designing an integrated circuit, most designers use a circuit simulator represented by Simulation Program with Integrated Circuit Emphasis (SPICE) to determine whether the circuit they are designing is capable of operating and reaching their desired design goals. Easy to verify Moreover, because the devices that can use the SPICE have been speeded up and enlarged, it is not difficult to simulate not only the circuits to be designed, but also the entire system including the circuits. Therefore, in recent years, from the development stage of the product, in addition to the electrical specification (electrical specification) of the circuit to be designed to predict and control the dispersion characteristics (dispersion characteristics) that may appear when the integrated circuit is manufactured.

SPICE is a program that mathematically solves an equivalent equation representing the electrical properties of a single device, taking into account the number of devices used and their electrical connections. The reliability of the results of the SPICE simulation depends on whether the model parameters (or process parameters) and the values of the various coefficients included in the equivalent equation are provided correctly. Therefore, for accurate and predictable simulation, a set of model parameters for single devices to be integrated in a semiconductor wafer must be accurately extracted and provided in advance.

 The process of extracting the set of model parameters will now be described by taking a metal oxide semiconductor field effect transistor (MOSFET) device as an example.

FIG. 1 is a flow chart illustrating a method of extracting a MOSFET model parameter set using a prior art.

2 is a top view of a schematic wafer for illustrating a selected die that satisfies a predetermined target value of a design rule in the prior art.

1 and 2, the conventional method,

S100: defining dies 20 on a wafer 10, and fabricating MOSFET devices on the dies 20;

S101: selecting any one die 30 among the dies 20 manufactured in step S100;

S102: measuring electrical characteristic data at the die 30 selected in step S101;

S103: selecting data in a design rule for a threshold voltage V _th and a saturation current I _Dsat among the electrical characteristic data measured in step S102;

S104: Determine whether the data selected in step S103 satisfies a predetermined target value of the design rule, and if it is determined that the target value is not satisfied, proceed to step S101;

S105: if it is determined in step S104 that the selected data reaches a predetermined target value of the design rule, measuring an I-V characteristic for a selected die to prepare an I-V characteristic curve;

S106: extracting the MOSFET model parameter set using the I-V characteristic curve provided in step S105.

This completes the extraction of the MOSFET model parameter set.

As described above, any one die whose electrical characteristic data satisfies a predetermined target value range of the design rule is selected from a number of dies, and IV characteristics according to the type and size of the MOSFET element are selected from the selected die. A method of extracting model parameters by measurement is disclosed in US Pat. No. 6,618,837.

However, as the MOSFET devices become more integrated, the distribution of measured electrical characteristics due to the effect of process dispersion is increasing, and the electrical characteristic data among a large number of dies satisfying the predetermined target range of the design rule. The method of extracting the MOSFET model parameter set according to the type and size of the MOSFET device by selecting one die of does not satisfy the characteristics of the current semiconductor device.

And, when the designer designs the semiconductor circuit using the model parameters extracted in this way, the SPICE simulation results cannot accurately predict the performance of the circuit. In addition, since the current semiconductor device characteristics are not accurately represented, it is impossible to predict a next process change, that is, a change in doping concentration or the like as a parameter.

As a reflection of this, a MOSFET device of the same type and size can be found in the 2002 Proceedings of the Fourth IEEE international Caracas Conference under the heading "The influence and modeling of process variation and device mismatch for analog / rf circuit design." As the channel length decreases, it is suggested that even the same die and the same length device have a large difference in electrical characteristics due to minute process variations.

In addition, in the conventional case, only the threshold voltage (Vth) and saturation current (Idsat) characteristics were used when analyzing the electrical characteristics, but these characteristics alone do not accurately represent process characteristics such as resistance and mobility of the device. It has a disadvantage.

Accordingly, an object of the present invention is to provide a method for extracting a MOSFET model parameter set according to the type and size of a MOSFET device that accurately represents process characteristics and process changes.

In order to achieve the above object, the present invention defines dies on one or more wafers, fabricating the basic elements of the integrated circuit in the dies and the electrical characteristics of the basic elements manufactured on the die From the measuring step and the measured electrical characteristic data, data of threshold voltage (V _th ), saturation current (I _Dsat ), and linear current (I _Dlin ) are statistically analyzed to calculate a dispersion curve according to the type and size of the basic device. And selecting the _physical data of the data of the threshold voltage V _th , the saturation current I _Dsat , and the linear current I _Dlin in the prepared distribution curve, and selecting the selected physical data. Selecting matching dies, measuring current (I) -voltage (V) characteristics at the selected dies, preparing an IV characteristic curve, and using the prepared IV characteristic curve. And to provide a model parameter extraction method comprising the step of extracting model parameters set that can reproduce the IV characteristic curve of the well.

And after the extracting the set of physical MOSFET model parameters, correcting a predetermined target value of a design rule by using the extracted set of physical MOSFET model parameters.

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the present invention will be described in detail.

3 is a signal flow diagram illustrating a method for extracting a physical MOSFET model parameter set according to the present invention.

4 is a top view of a schematic wafer for illustrating dies selected according to the type and size of a MOSFET device in accordance with the present invention.

3 and 4, an improved method of extracting a physical MOSFET model parameter set according to the present invention,

S200: defining dies 110 on one or more wafers 100, and manufacturing basic elements of an integrated circuit on the dies 110;

S201: measuring electrical property data for all dies 110 manufactured in step S200;

S203: Statistically analyze data about threshold voltage (V _th ), saturation current (I _Dsat ), and linear current (I _Dlin ) from the electrical characteristic data measured in step S201, and scatter curve according to MOSFET type and size Providing a step;

S204: selecting dies 120a, 120b,... Corresponding to the physical data according to the type and size of the MOSFET in the dispersion curve provided in step S203;

S205: In step S204, measuring an I-V characteristic for the selected dies 120a, 120b .... to prepare an I-V characteristic curve;

S206: extracting a set of physical MOSFET model parameters using the I-V characteristic curve provided in step S205;

S207: correcting a predetermined target value of the design rule using the set of physical MOSFET model parameters extracted in step S206.

More specifically, the data on the threshold voltage (V _th ) is statistically analyzed from the measured electrical characteristic data to display a scatter curve, and the dies corresponding to the physical data according to the type and size of the MOSFET are selected through the scatter curve. This is described below.

FIG. 5 is a diagram illustrating a dispersion curve according to gate length / gate width of a MOSFET device by statistically analyzing data about a threshold voltage V _th from measured electrical characteristic data.

Referring to FIG. 5, a scatter curve showing data of measuring threshold voltages V _th of MOSFET devices having a gate length of 0.36 μm and a gate width of 0.18 μm in all dies using a statistical program. to be. Here, the x-axis represents data of the threshold voltage V _th , and the y-axis represents the number of dies. In addition, a circle displayed at the center of the scatter curve indicates the physical data among a plurality of threshold voltage data. The die number corresponding to the physical data in the displayed physical data is selected to measure the IV characteristic which is a subsequent step.

6 shows a set of physical MOSFET model parameters extracted by the method according to the invention.

Referring to FIG. 6, if a die selected in FIG. 5, for example, a die of 120a number is selected, IV characteristics are measured at the selected die 120a, and an IV characteristic curve is prepared to extract a set of physical MOSFET model parameters. It is shown. The extracted set of physical MOSFET model parameters 200 shows a part 210 having a physical meaning and a coefficient part 220 of a spice model.

As described above, the improved method for extracting a physical MOSFET model parameter set according to the present invention uses the linear current (I _Dlin ) in addition to the threshold voltage (V _th ) and the saturation current (I _Dsat ). The process characteristics can be accurately represented.

After measuring and statistically analyzing the electrical characteristic data of all dies manufactured by the basic elements of the integrated circuit, the dies representing the type and size of the MOSFET element were selected to measure IV characteristics by measuring IV characteristics. Since the set is extracted, it is not necessary to determine whether the electrical characteristic data measured at any one conventional die selected satisfies a predetermined target value of the design rule. In addition, the electrical characteristic data for all dies are measured and statistically analyzed to accurately represent process variations.

In addition, a set of physical MOSFET model parameter that satisfies a predetermined target value of the design rule can be obtained by adjusting physical parameters in consideration of future process improvement.

The method of extracting the physical MOSFET model parameter according to the present invention accurately reflects the change in process characteristics and process dispersion due to the high integration of the MOSFET device, and can accurately predict future processes.

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art will be variously modified and changed within the scope of the invention without departing from the spirit and scope of the invention described in the claims below. I can understand that you can.

FIG. 1 is a flow chart illustrating a method of extracting a MOSFET model parameter set using a prior art.

2 is a top view of a schematic wafer for illustrating a selected die that satisfies a predetermined target value of a design rule in the prior art.

3 is a signal flow diagram illustrating a method for extracting a physical MOSFET model parameter set according to the present invention.

4 is a top view of a schematic wafer for illustrating dies selected according to the type and size of a MOSFET device in accordance with the present invention.

FIG. 5 is a diagram illustrating a dispersion curve according to gate length / gate width of a MOSFET device by statistically analyzing data about a threshold voltage V _th from measured electrical characteristic data.

6 shows a set of physical MOSFET model parameters extracted by the method according to the invention.

<Explanation of symbols for the main parts of the drawings>

10, 100: wafer 20, 110: die

30, 120a, 120b ....: selected dies

Claims (2)

Defining dies on one or more wafers and fabricating the basic elements of an integrated circuit on the dies; Measuring electrical characteristics of the basic devices fabricated in the dies; Statistically analyzing data of a threshold voltage (V _th ), a saturation current (I _Dsat ), and a linear current (I _Dlin ) among the measured electrical characteristic data to prepare a scatter curve according to the type and size of a basic device; Selecting _physical data of data of a threshold voltage (V _th ), a saturation current (I _Dsat ), and a linear current (I _Dlin ) from the prepared distribution curve; Selecting dies that match the selected medical data; Preparing an I-V characteristic curve by measuring a current (I) -voltage (V) characteristic at the selected dies; And And extracting a set of model parameters that can best reproduce the I-V characteristic curve by using the prepared I-V characteristic curve as an objective function. The method of claim 1, further comprising correcting a predetermined target value of a design rule by using the extracted physical MOSFET model parameter set after extracting the physical MOSFET model parameter set. Extraction method.