JPH0888195A - Semiconductor process simulation method - Google Patents

Abstract

PURPOSE: To obtain simulation result of high precision in a short time, by automatically optimizing a mesh in each process. CONSTITUTION: A data retrieving means 6 retrieves each process and the execution condition of each process in order from a process data file 1. A prediction means 7 predicts concentration distribution of impurities when each process is executed, from each process and the execution condition of each process which have been retrieved. A mesh setting means 8 sets or changes automatically the interval of a mesh, according to the concentration change of the concentration distribution of impurities which has been predicted. A simulation means 9 simulates the impurity distribution in a device by using the set or changed mesh, and performs simulation by the automatically optimized mesh.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor process simulation method, and more particularly to automatic mesh setting during simulation of ion implantation and diffusion processes.

[0002]

2. Description of the Related Art Semiconductor process simulation is widely used to shorten the development period of semiconductor devices. In order to perform a semiconductor process simulation, it is necessary to perform operations such as selecting a simulation model and setting a mesh. However, the simulation may not be performed correctly in some cases because the necessary data may be omitted. Therefore, in the simulation system disclosed in Japanese Patent Application Laid-Open No. 2-272671, when the basic system data of the system and the purpose of the simulation are input, the system infers the data that is insufficient when performing the simulation and inputs the insufficient data. By making a request, it is possible to prevent the omission of necessary data. In this way, the efficiency of data input work to the system and its accuracy are improved.

[0003]

However, the semiconductor process simulation result varies greatly depending on how the mesh is set. In the case of semiconductor process simulation, if the mesh is made fine, the calculation time becomes huge, and if the mesh is made coarse, the accuracy becomes poor. For this reason,
In order to perform a highly accurate semiconductor process simulation in a short time, it is necessary not only to input mesh settings but also to engrave an appropriate mesh.

Furthermore, the mesh needs to be carved according to the selected simulation model. For this reason, in order to carve an appropriate mesh, it is necessary to have knowledge about simulation and the like, which is a very difficult work for general users.

The present invention has been made to solve such a problem, and an object thereof is to automatically perform mesh optimization for each process and obtain a highly accurate simulation result in a short time.

[0006]

A semiconductor process simulation method according to the present invention is provided with a process data file storing information relating to a semiconductor ion implantation and diffusion process, and a process data file is used to identify each process and execution conditions of each process. Searches in order, predicts the impurity concentration distribution when each process is executed from the searched process and execution conditions of each process, and automatically sets the mesh interval according to the predicted concentration change of the impurity concentration distribution. Setting or changing, and simulating the impurity distribution inside the semiconductor device using the set or changed mesh.

Further, a one-dimensional simulation only in the depth direction of the semiconductor device is performed from the retrieved processes and the execution conditions of each process, and the impurity concentration distribution when each process is performed is predicted from the result of the one-dimensional simulation. Good to do.

Further, the conditions for executing the ion implantation process are keyed.
A table for storing the projection range and the standard deviation and a table for storing the correlation data with which the concentration distribution can be predicted by using the diffusion process execution condition as a key, and the ion implantation process execution condition when the searched process is the ion implantation process The projection range and standard deviation are calculated from the table that stores the projection range, etc., and the impurity concentration distribution when the ion implantation process is executed is predicted from the calculated projection range and standard deviation, and the retrieved process When is a diffusion process, correlation data may be obtained from a table that stores correlation data based on diffusion process execution conditions, and the impurity concentration distribution when the diffusion process is performed may be predicted from the obtained correlation data.

[0009]

In the present invention, each process and the execution conditions of each process are sequentially searched from the process data file, and the impurity concentration distribution when each process is executed is predicted from the searched each process and the execution condition of each process. To do. The mesh interval is automatically set or changed according to the predicted concentration change of the impurity concentration distribution, the impurity distribution inside the semiconductor device is simulated using the set or changed mesh, and the mesh is automatically optimized. Simulate with.

Further, a one-dimensional simulation only in the depth direction of the semiconductor device is carried out from each process retrieved from the process data file and the execution condition of each process,
The impurity concentration distribution when each process is executed is predicted from the result of the one-dimensional simulation, and the mesh intervals that are simulation intervals are automatically calculated in proportion to the impurity concentration based on the predicted impurity concentration distribution. Set or change and perform accurate simulation.

Further, when the process retrieved from the process data file is the ion implantation process, the projection representing the depth from the ion injection to the ion stop with a statistical fluctuation range based on the ion implantation process execution conditions. The projection range and the standard deviation are obtained from a table that stores the range and the standard deviation of the distribution, and the concentration distribution of impurities when the ion implantation process is performed is predicted from the obtained projection range and the standard deviation. When the retrieved process is a diffusion process, the correlation data is obtained from the table that stores the correlation data based on the diffusion process execution condition, and the impurity concentration distribution when the diffusion process is executed is predicted from the obtained correlation data, and Make an accurate simulation.

[0012]

1 is a block diagram showing a semiconductor process simulator adopting the semiconductor process simulation method of the present invention. As shown in the figure, the semiconductor process simulator includes a process data file 1, a table file 2 and a processing unit 3. The shooting process data file 1 stores each process and the execution condition of each process. Here, the processes include, for example, an ion implantation process and a diffusion process, and the execution conditions of each process include, for example, the ion implantation process execution conditions such as the implant, the implantation energy and the implantation angle, and the diffusion such as the temperature and the atmosphere during the diffusion. There is a process execution condition. The table file 2 has a projection range table 4 and a correlation table 5. Projection range table 4
Stores the projection range and standard deviation with the ion implantation process execution condition as a key. The correlation table 5 stores correlation data in which the concentration distribution can be predicted based on the diffusion process execution conditions of temperature, time, and atmosphere. Here, the projective range is a distance obtained by connecting a distance from the incident point of the ion to the stop point with a straight line and projecting this on a perpendicular line from the incident point, that is, a depth from the surface. The distance from the ion injection to the ion stop is not constant for all ions, and a difference occurs for each ion depending on the way of collision with a lattice atom on the way. Further, since the number of ions to be implanted is very large, the range has a statistical fluctuation range. As a result of the range having a statistical fluctuation range, the implanted ions have a Gaussian distribution in the semiconductor substrate as shown in the distribution chart of FIG. Therefore, the projection range R becomes a value of the depth indicating the peak value of the ion distribution, and its spread can be predicted by the standard deviation.

The processing unit 3 has a data search means 6, a prediction means 7, a mesh setting means 8 and a simulation means 9. The data search means 6 is the process data file 1
To sequentially search each process and the execution condition of each process. The predicting means 7 predicts the impurity concentration distribution when each process is executed from each process searched by the data searching means 6 and the execution condition of each process. For example, each process searched by the data searching means 6 and Based on the execution conditions of each process, the projection range R and the standard deviation are read from the projection range table 4 or the correlation table 5, and the impurity concentration distribution when each process is executed from the read projection range R is read. Predict. The mesh setting unit 8 automatically sets or changes the mesh interval according to the concentration change of the impurity concentration distribution predicted by the prediction unit 7, and stores it in the process data file 1. The simulation unit 9 executes a two-dimensional process simulation of the impurity distribution inside the semiconductor device using the conditions such as the mesh interval stored in the process data file 1 by the mesh setting unit 8.

A case in which the semiconductor process simulator having the above-described structure simulates a process for producing a source and a drain of an NMOSFET, for example, will be described with reference to the flowchart of FIG.

When the start of the simulation is instructed,
The data search means 6 sequentially searches the process data file 1 for each process and the execution condition of each process, and, for example, the arsenic implantation process and the arsenic implantation process stored in the process data file 1 are configured as shown in the configuration diagram of FIG. The execution condition of is read (step S1). Here, “IMPLANT ARSENIC” in the figure indicates an arsenic implantation process, and “ENERGY = 50” specifies implantation energy as an execution condition of the arsenic implantation process. "DIFFUSE" indicates the diffusion process, and "TEMP = 950" and "TIME
= 100 "specifies temperature and time as the diffusion process execution conditions.

When the data searching means 6 reads the arsenic searching process, the predicting means 7 reads the projection range R and the standard deviation from the projection range table 4 on the basis of the arsenic as an implant and the conditions of the arsenic injection process. Based on the read projection range R and the standard deviation, the density distribution having the relationship between the depth and the density as shown by the parabola 10 in the distribution chart of FIG. 5 is obtained. The parabola 10 represents the distribution of arsenic, which is the drain, and the curves 11 and the straight lines 12 represent the distribution of the semiconductor of the gate and the semiconductor of the source, respectively. Based on the obtained concentration distribution, the predicting means 7 indicates, for example, as shown in the cross-sectional view in FIG. When is injected, it is predicted to be distributed as shown by the dotted line 17 (step S2). Based on the density distribution predicted by the prediction means 7, the mesh setting means 8 sets the meshes so that the intervals between the meshes where the density distribution changes abruptly become fine as shown in the sectional view of FIG. Store in data file 1. That is, since the density change in the area between the line A and the line B is rapid with respect to the depth direction, the mesh setting unit 8 makes the mesh interval in the area between the line A and the line B fine. Similarly, since the density change in the area between the line C and the line D is rapid, the mesh setting unit 8 makes the mesh interval in the area between the line C and the line D fine (step S3). As a result, the mesh used for simulation of the arsenic implantation process is automatically set in a short time.

When the mesh setting means 8 sets a mesh, the simulation means 9 performs a two-dimensional process simulation under the arsenic implantation process execution conditions stored in the process data file 1 (step S4). As a result, the semiconductor process simulator can perform the two-dimensional process simulation of the implantation process with high accuracy by using the optimum mesh.

Next, since the simulation of the diffusion process has not been completed, the data retrieval means 6 reads the diffusion process and the diffusion process execution condition which are the next process from the process data file 1 (step S5).
S1). When the data retrieval unit 6 reads the diffusion process and the diffusion process execution condition, the prediction unit 7 reads the correlation data from the correlation table 5 based on the read diffusion process execution condition, and the concentration distribution based on the read correlation data. Ask for. The predicting means 7 uses the calculated concentration distribution as shown in FIG.
It is predicted that the implanted arsenic diffuses as shown by the dotted line 18 as shown in the cross-sectional view (step S2). Based on the predicted concentration distribution, the mesh setting means 8 sets the mesh so that the intervals between the meshes where the concentration distribution changes abruptly become fine, and stores it in the process data file 1 (step S3). The simulation means 9 performs a two-dimensional process simulation of the diffusion process under the diffusion process execution conditions stored in the process data file 1 (step S4). If the simulation is finished and the next process is not stored in the process data file 1, the process is finished (step S5).

Although the concentration distribution of the diffusion process is predicted by reading the correlation table 5 in the above embodiment, it may be predicted by actually simulating using a one-dimensional process simulator.

Although the ion implantation process and the diffusion process have been described in the above embodiments, they may be used for the oxidation process.

[0021]

As described above, the present invention sequentially searches each process and the execution condition of each process from the process data file, and when each process is executed from the searched each process and the execution condition of each process. Predict the impurity concentration distribution, and automatically set or change the mesh interval according to the predicted impurity concentration distribution concentration change,
Impurity distribution inside the semiconductor device is simulated by using the set or changed mesh, and the simulation is automatically performed by the optimized mesh.

In addition, a one-dimensional simulation only in the depth direction of the semiconductor device is performed from each process retrieved from the process data file and the execution condition of each process,
The impurity concentration distribution when each process is executed is predicted from the result of the one-dimensional simulation, and the mesh intervals that are simulation intervals are automatically calculated in proportion to the impurity concentration based on the predicted impurity concentration distribution. Set or change and perform accurate simulation.

When the process retrieved from the process data file is the ion implantation process, the projection range and standard deviation are calculated from the table storing the projection range based on the ion implantation process execution conditions, and the calculated projection range and the standard deviation are calculated. The standard deviation is used to predict the impurity concentration distribution when the ion implantation process is performed. The projective range represents the depth from ion injection to stop with a statistical fluctuation range, and the standard deviation represents the standard deviation of the distribution.Therefore, the projective range and standard deviation indicate the impurity concentration distribution. Can be predicted, and a quick and accurate simulation can be performed. When the retrieved process is a diffusion process, correlation data is obtained from a table that stores correlation data based on the diffusion process execution conditions of temperature, time, and atmosphere, and the impurity concentration distribution when the diffusion process is performed from the obtained correlation data Predict and perform a quick and accurate simulation.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an embodiment of the present invention.

FIG. 2 is a distribution diagram showing the relationship between the distance from the substrate surface and the impurity concentration.

FIG. 3 is a flowchart showing an operation when a mesh is automatically set.

FIG. 4 is a storage configuration diagram of a process data file.

FIG. 5 is a distribution diagram showing an arsenic concentration distribution.

FIG. 6 is a cross-sectional view of an NMOSFET.

FIG. 7 is a cross-sectional view of an NMOSFET when a mesh is applied.

[Explanation of Codes] 1 Process Data File 4 Projection Range Table 5 Correlation Data Table 6 Data Retrieval Means 7 Prediction Means 8 Mesh Setting Means 9 Simulation Means

Continuation of the front page (51) Int.Cl. ⁶ Identification number Office reference number FI technical display location // G06F 17/00 9069-5L G06F 15/20 D

Claims (3)

[Claims] 1. A semiconductor process simulation method for two-dimensionally simulating the impurity distribution inside a semiconductor device in each process of implantation, oxidation, and diffusion, in which process data storing information on an ion implantation process and a diffusion process of a semiconductor are stored. With files,
Each process and the execution conditions of each process are searched in order from the process data file, the impurity concentration distribution when each process is executed is predicted from the searched each process and execution condition of each process, and the predicted impurity concentration distribution A method for simulating a semiconductor process, characterized in that a mesh interval is automatically set or changed in accordance with a change in concentration, and an impurity distribution inside a semiconductor device is simulated using the set or changed mesh. 2. A one-dimensional simulation only in the depth direction of a semiconductor device based on the retrieved processes and execution conditions of each process, and predicting the impurity concentration distribution when each process is performed from the result of the one-dimensional simulation. The semiconductor process simulation method according to claim 1. 3. A table is provided for storing projection range and standard deviation with an ion implantation process execution condition as a key, and a table for storing correlation data whose concentration distribution can be predicted with a diffusion process execution condition as a key. When the process is an ion implantation process, the projective range and standard deviation are obtained from the table that stores the projective range based on the ion implantation process execution conditions, and the ion implantation process is executed from the calculated projective range and standard deviation. When the retrieved process is a diffusion process, the correlation data is calculated from the table that stores the correlation data based on the diffusion process execution condition, and the correlation data obtained when the diffusion process is executed is predicted. The semiconductor process simulation method according to claim 1, wherein the impurity concentration distribution is predicted.