JP2000091178A - Production control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To appropriately control production by adjusting the production according to a predicted yield by obtaining a regression expression for predicting the yield from production control data in a wafer treatment process. SOLUTION: In a semiconductor-manufacturing device 10, a wafer treatment part 11 sends production control data in a process of a treated wafer to a control part 20. When the wafer that is treated by the wafer treatment part 11 is sent to a prober inspection part 12, the prober inspection part 12 detects the electrical characteristics of a chip in the wafer and sends them to the control part 20. An assembly part 13 assembles a chip that is inspected by the prober inspection part 12 into a semiconductor product. Then, an inspection part 14 detects electrical characteristics in a semiconductor product being assembled by the assembly part 13 to the control part 20. Also, it is judged whether the semiconductor product is conforming or not based on the electrical characteristics being detected by the inspection part 14, and the number of conforming semiconductor products is set to the control part 20. The control part 20 calculates a scheduled yield using a specific yield prediction expression and throws wafers into the wafer treatment part 11 based on the scheduled yield.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement in a production control method in, for example, a semiconductor manufacturing apparatus, and more particularly to a production control method capable of minimizing a change in the number of semiconductor products due to a change in yield and suppressing the yield. Things.

[0002]

2. Description of the Related Art The production management of semiconductor products is based on a required yield and a TAT (Turn-Ar
sound-time: the time required until the product is completely finished). The yield means a non-defective product completion rate with respect to the number of products put into a production line of a semiconductor production device. The prediction formula for calculating the yield can be represented by, for example, Y = 1 / (1 + AD). Here, Y is the yield, A is the chip area of the semiconductor product, and D is the defect density per unit area. As the yield prediction formula, a yield prediction formula most suitable for the semiconductor product manufacturing line and the type of semiconductor product to be manufactured is used.

FIG. 5 is a flowchart showing a conventional semiconductor product production management method. An example of a conventional production management method will be described with reference to FIG. First, the quantity to be manufactured is set based on the semiconductor manufacturing line and the type of semiconductor to be manufactured, and the most suitable yield prediction equation is selected from a plurality of yield prediction equations (S1). Next, the number of wafers (lots) calculated by the yield prediction formula is input to the production line (S2). Then, passive elements, active elements or integrated circuits are formed on the wafer (S
3). Specifically, according to a semiconductor product manufacturing type process flow, a cleaning process, a diffusion process,
A chip is formed by repeating an oxide film forming step, a CVD film forming step, a photolithography step, an ion implantation step, a sputtering step, and the like. Then, when the wafer processing step is completed, electrical measurement of the processed wafer is performed, and electrical characteristics of the wafer are confirmed.

Thereafter, in the prober inspection process, the electrical characteristics of the chips formed on the wafer are measured, and at this stage, non-defective and defective chips are discriminated (S).
4). Then, the quantity information of the non-defective and defective products is fed back to the production management, and the quantity of wafers to be put into the production line is controlled according to the quantity information. Then, the wafer after the prober inspection process is assembled into a semiconductor product (S
5). Then, the electrical characteristics of the manufactured semiconductor product chip are inspected, and a good product and a defective product are discriminated. Then, the quantity information of the non-defective and defective products is fed back to the production management, and the quantity of wafers to be put into the production line is controlled according to the quantity information.

[0005]

Here, since the working conditions of each step in the wafer processing step are controlled within a predetermined range, the characteristics of individual wafers to be processed are theoretically almost the same. Become. However, in practice, operating conditions fluctuate depending on temperature, humidity, and the like, so that characteristics of semiconductor products also fluctuate. Semiconductor products that do not satisfy certain characteristics are defective. Therefore, the quantity of defective products is not always constant but fluctuates, and the quantity of manufactured non-defective products also fluctuates.

Here, when the actual yield is lower than the expected yield calculated from the yield prediction formula, that is, when the actual quantity of non-defective products is smaller than the expected quantity, a new wafer is put into the production line. There must be. However, there are several tens to hundreds of steps from the wafer input into the production line (S1) to the completion of the semiconductor product (lot out), and a long processing time is required. Therefore, after inspecting in the prober inspection process or the assembly process, even if the wafer is added to the production line,
There is a problem that it takes time until a certain quantity is prepared.

[0007] In order to avoid this problem, it is conceivable to increase the quantity of wafers to be put into the production line from the beginning and to put in a large number of wafers. However, when the actual yield is almost the same as the expected yield, there is a problem that semiconductor products are excessively produced. Therefore,
There is a demand for a production management method that appropriately controls production while minimizing the influence of the yield of semiconductor products due to fluctuations in yield.

Accordingly, the present invention solves the above-mentioned problems, minimizes the influence of the yield of semiconductor products due to fluctuations in the yield, makes the production control appropriate, and suppresses the decrease in the yield. It is intended to provide.

[0009]

According to the first aspect of the present invention, a predetermined yield is calculated by predicting the occurrence of a yield in advance, and a predetermined wafer is transferred to a semiconductor manufacturing apparatus based on the predetermined yield. Injection, in the production management method for controlling the production amount of semiconductor products, by analyzing the correlation between the electrical characteristics measured to inspect the chip manufactured by the semiconductor manufacturing apparatus and the expected yield The electrical characteristics and data obtained when forming an electrical element on the wafer, the correlation between the in-process production management data having a plurality of parameters is analyzed, and the electrical characteristics and the Based on the correlation of the expected yield, based on the correlation between the electrical characteristics and the in-process production management data,
This is achieved by a production management method that analyzes the relationship between the in-process production management data and the expected yield, calculates the expected yield based on the in-process production management data, and controls the number of wafers to be input. .

According to the first aspect of the present invention, the expected yield is derived by analyzing the relationship between the in-process production management data and the yield. As a result, the expected yield can be calculated at the stage when the electric elements are formed on the wafer, so that the wafer can be loaded at an early stage, and the influence of the yield fluctuation on the production of semiconductor products is minimized. Can be suppressed.

[0011] According to a second aspect of the present invention, in the configuration of the first aspect, the analysis of the correlation between the electrical characteristics and the in-process production management data is performed based on the analysis of the in-process production management data. This is achieved by a production management method performed by extracting one parameter that affects the electrical characteristics. According to the configuration of the second aspect, the expected yield is calculated based on one parameter that most affects the yield. By performing the production management based on the expected yield, it is possible to minimize the influence of the yield fluctuation on the production amount of the semiconductor product.

[0012] According to a third aspect of the present invention, in the configuration of the first aspect, the analysis of the correlation between the electrical characteristic and the in-process production management data includes the step of analyzing the correlation between the in-process production management data. This is achieved by a production management method in which a plurality of parameters affecting the electrical characteristics are extracted and performed. According to the configuration of the third aspect, the expected yield is calculated based on the plurality of parameters that most affect the yield. By performing the production management based on the expected yield, it is possible to minimize the influence of the yield fluctuation on the production amount of the semiconductor product.

[0013] According to a fourth aspect of the present invention, in the configuration of the first aspect, the analysis of the correlation between the electrical characteristics and the in-process production management data includes the step of analyzing the correlation between the in-process production management data. This is achieved by a production management method performed by extracting a first parameter that affects the electrical characteristics and a second parameter that affects the first parameter. According to the configuration of claim 4, the expected yield is calculated by analyzing the first parameter.
The parameter is calculated by analyzing the second parameter. Therefore, in the process of processing a wafer, when an operation that affects the second parameter is performed, the first parameter also changes, and the expected yield also changes. Utilizing this, a change in the expected yield is calculated from a change in the second parameter, and a first yield that minimizes this expected yield is calculated.
Parameters in the wafer processing process so that
Modify the work conditions of the work that affects the parameters.
As a result, a decrease in yield can be minimized.

[0014]

Preferred embodiments of the present invention will be described below in detail with reference to the accompanying drawings. Note that the embodiments described below are preferred specific examples of the present invention,
Although various technically preferable limits are given, the scope of the present invention is not limited to these modes unless otherwise specified in the following description.

FIG. 1 is a block diagram showing an example of a semiconductor manufacturing apparatus to which the production management method of the present invention is applied. The semiconductor manufacturing apparatus 10 will be described with reference to FIG. The semiconductor device 10 of FIG. 1 includes a wafer processing unit 11, a prober inspection unit 12, an assembly unit 13, an inspection unit 14, a control unit 20, and the like. The wafer processing unit 11 includes a cleaning device, a CVD film forming device, a sputtering device, an ion etching device, and the like, and forms a transistor, a capacitor, and the like on a wafer. The operation of the wafer processing unit 11 is controlled by the control unit 20, and the wafer processing unit 11 performs in-process process quality control data (Inl
ine Process Quality Controller
ol, hereinafter referred to as “IPQC data”) to the control unit 20.

Here, the IPQC data includes, for example, the film thickness in the oxide film forming step, the film thickness in the CVD film forming step,
Refractive index, doping concentration such as Phos / Boron, sheet resistance, line width, misalignment in photolithography process,
In the etching process, data such as the line width and the remaining base film thickness, in the cleaning process, the remaining base film thickness and particles, and in the sputtering process, data such as film thickness, sheet resistance, and particles are incorporated.

The wafer processed by the wafer processing unit 11 is sent to the prober inspection unit 12. The prober inspection unit 12
In order to efficiently test the electrical characteristics of the electronic circuit formed on the wafer, a stylus is automatically brought into contact with the electrode of each chip, and an electrical test of each chip is performed by an external tester connected to the stylus Is what you do. Here, the prober inspection unit 12 detects the electrical characteristics EC1 of the chips on the wafer and sends them to the control unit 20. This electric characteristic EC1 is composed of a plurality of parameters EC1 (1), EC1 (2),. In addition, the prober inspection unit 12 has an electric characteristic EC.
The quality of the chip is determined based on 1 and the number of non-defective products is sent to the control unit 20.

The assembling section 13 assembles the chip inspected by the prober inspecting section 12 into a semiconductor product. The assembling unit 13 includes, for example, a dicing device for cutting the wafer into individual chips, a die bonding device for mounting the chips on a lead frame or a package, a wire bonding device for electrically connecting connection electrodes on the chips to external terminals, and a chip. And a marking device for marking characters and symbols on the package surface. The inspection unit 14 detects the electrical characteristic EC2 of the semiconductor product assembled by the assembly unit 13 and sends the same to the control unit 20. This electrical characteristic EC2 is composed of a plurality of parameters EC2 (1), EC2 (2),. The inspection unit 14 detects the detected electric characteristic E
The quality of the semiconductor product is determined based on C2, and the quantity of the non-defective product is sent to the control unit 20.

The control section 20 controls the operation of the semiconductor manufacturing apparatus 10 and calculates and analyzes the expected yield. Specifically, the control unit 20 calculates a planned yield using a predetermined yield prediction formula, and loads a wafer into the wafer processing unit 11 based on the planned yield. As will be described later, the expected yield is corrected based on the IPQC data sent from the wafer processing unit 11, the electrical characteristics EC1 and EC2 sent from the prober inspection unit 12 and the inspection unit 14, and the number of non-defective products. If necessary, control is performed so that a new wafer is loaded.

FIG. 2 is a flowchart showing an example of the semiconductor production management method of the present invention. The semiconductor production management method will be described in detail with reference to FIGS. First, the control unit 20 in FIG. 1 predicts the yield in advance (S1
0), a fixed number of wafers are loaded into the wafer processing unit 11 (S11). Then, the wafer processing unit 11 performs a predetermined process on the wafer to produce a chip (S12).
At this time, each working unit of the wafer processing unit 11 measures IPQC data for each wafer and sends it to the control unit 20.

Next, the prober inspection section 12 performs electrical measurement of the manufactured chip (S13), and detects the electrical characteristic EC1 of the chip. The prober inspection unit 12 determines the quality of the chip in the state of the wafer based on the electrical characteristic EC1, and separates the non-defective product from the non-defective product. Further, the prober inspection unit 12 also sends the number of non-defective products to the control unit 20. Chips determined to be non-defective in the prober inspection process are sent to the assembling unit 13. The assembling unit 13 assembles the chip to produce a semiconductor product (S14). Thereafter, the inspection unit 14 performs an electrical measurement on the semiconductor product and detects an electrical characteristic EC2 of the semiconductor chip (S15). The inspection unit 14 determines the acceptability of the semiconductor chip based on the electrical characteristic EC2, and classifies the semiconductor chip into non-defective products and defective products. The inspection unit 14 sends the quantity of non-defective products to the control unit 20. Then, the inspected semiconductor product is shipped (S16).

The control unit 20 calculates the expected yield based on the electric characteristics EC1 or EC2 or the electric characteristics EC1 and EC2 (hereinafter, simply referred to as “EC”). Specifically, first, the correlation between the electrical characteristics EC detected by the prober inspection unit 12 and the inspection unit 14 and the yield is analyzed. Expressing this as a function, the yield function y1 is as follows. y1 = f ₁ (electrical characteristic EC) = f ₁ (electrical characteristic EC (1), electric characteristic EC (2),...) (1) And this yield function y1 has the most influence. The parameter EC (m) of the received electrical characteristic EC is extracted. Then, the yield function y1 becomes the yield function y1 using the electric characteristic EC (m) as a variable. y1 = f ₂ (Electrical characteristics EC (m)) (2)

Next, the correlation between the electrical characteristics EC (m) and the IPQC data is analyzed. This is shown in the function:
The electrical characteristic function EC (m) is as follows. EC (m) = f ₃ (IPQC) = f ₃ (IPQC (1), IPQC (2),...) (3) And this electric characteristic function EC (m) is most affected. One parameter IPQC (m) is extracted.
For example, when a CMOS transistor is manufactured on a wafer, the threshold voltage Vth of the CMOS transistor is affected by the impurity concentration of the wafer. Therefore, one parameter of the impurity concentration of the wafer is extracted from the IPQC data. Here, the electric characteristic function EC (m) is converted into an electric characteristic function EC (m) using one parameter IPQC (m) as a variable. EC (m) = f ₄ (IPQC (m)) (4)

From the equations (2) and (4), the yield function y1 is a regression equation using one parameter IPQC (m) of the IPQC data as a variable as shown in the following equation. Yield function y1 = f ₂ (f ₄ (IPQC (m))) = f ₅ (IPQC (m)) (5) Scheduled yield calculated using this yield function y 1 and transmission from inspection unit 14. The actual yield calculated based on the obtained non-defective product quantity is compared. Then, when the actual yield is lower than the expected yield, the control unit 20 controls the wafer processing unit 11 to insert a new wafer.

By predicting the yield by using the yield function y1, the yield can be predicted from the IPQC data. Therefore, when the wafer needs to be loaded, the wafer can be loaded at an early stage. The impact on product production can be minimized.

Further, the control unit 20 can change the operation condition of the operation in the wafer processing unit 11 which has the most influence on the parameter IPQC (m) of the IPQC data based on the equation (5). Specifically, by changing the working conditions so that the yield y1 becomes the best IPQC (m) in the equation (5), it is possible to minimize the decrease in the yield.

FIG. 3 is a flowchart showing a second embodiment of the present invention, and a semiconductor production management method will be described with reference to FIG. The difference between the semiconductor production management method of FIG. 3 and the semiconductor production management method of FIG.
The point is that the yield function y2 is a two-variable function having two parameters as variables as shown in the following equation. y2 = f ₁₀ (IPQC (m), IPQC (n)) (6) The reason why the yield function y2 is a two-variable function is as follows. For example, the drain current I of a CMOS transistor
ds depends on (gate width W / gate length L). Therefore, it is understood that the drain current depends on two parameters of the IPQC data, the gate width W and the gate length L. In this case, the yield function y with two parameters as variables
By calculating the yield using 2, the yield can be more accurately predicted.

Here, equation (6) will be specifically described. First, the correlation between the electrical characteristics EC detected by the prober inspection unit 12 and the inspection unit 14 and the yield is analyzed. Expressing this as a function, the yield function y2 is as follows. y2 = f ₆ (electric characteristic EC) = f ₆ (electric characteristic EC (1), electric characteristic EC (2),...) (7) And the yield function y2 has the most influence. The parameter EC (m) of the received electric characteristic is extracted, and the yield function y2 is converted into a yield function y2 using the electric characteristic EC (m) as a variable. y2 = f ₇ (Electrical characteristics EC (m)) (8)

Next, the parameter EC (m) of the electrical characteristics
The correlation between the data and the IPQC data is analyzed. Expressing this as a function, the electrical characteristic function EC (m) is as follows. EC (m) = f ₈ (IPQC) = f ₈ (IPQC (1), IPQC (2),...) (9) Then, the electric characteristic function EC (m) is affected. There are for example two parameters IPQC (m), IPQ
C (n) is extracted and the two parameters IPQC
(M), electrical characteristic function E with IPQC (n) as a variable
C (m) is as follows. EC (m) = f ₉ (IPQC (m), IPQC (n)) (10)

From equations (8) and (10), the yield function y has two parameters, IPQC (m) and IPQC
It is a regression equation using (n) as a variable. Yield function y2 = f ₇ (f ₉ (IPQC (m), IPQC (n))) = f ₁₀ (IPQC (m), IPQC (n)) (6) The yield is calculated using this yield function y2. By predicting, in each process in the wafer processing unit 11, the IP
The yield can be predicted from the QC data and production control can be adjusted.

Further, the control unit 20 changes the operation condition of the operation in the wafer processing unit 11 which has the most influence on the parameters IPQC (m) and IPQC (n) of the IPQC data based on the equation (6). Can also. Specifically, in equation (6), 2 is used such that the yield y2 is improved.
By setting the working conditions so that the two parameters are IPQC (m) and IPQC (n), it is possible to prevent the yield from decreasing to a minimum.

FIG. 4 is a flow chart showing a third embodiment of the semiconductor production management method according to the present invention.
The third embodiment will be described with reference to FIG. The third embodiment of FIG. 4 is different from the first embodiment of FIG. 2 in that the first parameter IPQC (n) is different from the second parameter IQC (n).
Considering the case where PQC (m) is affected,
That is, the yield function y3 is calculated.

More specifically, it is assumed that a plurality of pieces of IPQC data affect the electrical characteristic EC detected.
Furthermore, the parameter IPQC (m) is different from the other parameters I
It is assumed that PQC (n) is affected. In this case, in the process of processing the wafer, if an operation that affects the second parameter is performed, the first parameter also changes, and the expected yield also changes. IPQC (n) = f ₁₁ (IPQC (m)) (11)

Therefore, first, the second parameter IPQC
(M) is changed, and the thus-changed first parameter IPQC (n) and the fluctuation of the yield are analyzed. The yield function y3 at this time is as follows. Yield function y3 = f ₁₂ (f ₁₁ (IPQC (m)) (11) Then, in order to control the parameter IPQC (m), the process operation conditions in the wafer processing unit 11 are adjusted. When a decrease in yield is predicted from the IPQC data of a certain process in the processing unit 11, the work condition can be adjusted to the next process so as to minimize the decrease in yield. By predicting the yield including the number of wafers, the number of wafers to be supplied can be minimized.

Further, the variation of the expected yield is calculated from the variation of the second parameter IPQC (m), and the first parameter in the wafer processing step is set so as to be the first parameter IPQC (n) which suppresses the expected yield. Correct the work conditions of the work affecting (n). As a result, a decrease in yield can be prevented.

According to the above embodiment, in the production control of semiconductor products, the regression equation for predicting the yield from the IPQC data of the wafer processing step is obtained, and the production is adjusted based on the predicted yield. Proper production management can be performed with a minimum.
Further, in the wafer processing step, the I
By estimating the yield from the PQC data and correcting the next operation condition based on the result and performing the operation, it is possible to minimize the decrease in the yield of the production lot.

[0037]

As described above, according to the present invention,
It is possible to appropriately perform production control while minimizing fluctuations in production volume due to fluctuations in yield.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an example of a semiconductor manufacturing apparatus managed by a production management method of the present invention.

FIG. 2 is a flowchart showing a preferred embodiment of the production management method of the present invention.

FIG. 3 is a flowchart illustrating a production management method according to a second embodiment of the present invention.

FIG. 4 is a flowchart illustrating a production management method according to a third embodiment of the present invention.

FIG. 5 is a flowchart illustrating an example of a conventional production management method.

[Explanation of symbols]

10: semiconductor manufacturing apparatus, 11: wafer processing unit,
12: Prober inspection unit, 13: Assembly unit, IPQ
C data: In-process quality control data, EC: Electrical characteristics.

Claims (4)

[Claims] 1. A production management method for predicting occurrence of a yield in advance, calculating a planned yield, loading a predetermined wafer into a semiconductor manufacturing apparatus based on the planned yield, and controlling a production amount of a semiconductor product, Analyzing a correlation between electrical characteristics measured for inspecting a chip manufactured by the semiconductor manufacturing apparatus and the expected yield, and forming the electrical characteristics and an electrical element on the wafer. And analyzing the correlation between the in-process production management data having a plurality of parameters, the correlation between the electrical characteristics and the expected yield, and the electrical characteristics and the in-process production management. Analyzing the relationship between the in-process production management data and the planned yield based on the correlation of data, and calculating the planned yield based on the in-process production management data. Out, production management method characterized by controlling the quantity of introduction of the wafer. 2. The analysis of the correlation between the electrical characteristics and the in-process production management data is performed by extracting one parameter that affects the electrical characteristics from the in-process production management data. The production management method according to claim 1. 3. The analysis of the correlation between the electrical characteristics and the in-process production management data is performed by extracting a plurality of parameters affecting the electrical characteristics from the in-process production management data. The production management method according to claim 1. 4. The analysis of the correlation between the electrical characteristic and the in-process production management data includes the step of: in the in-process production management data, a first parameter affecting the electrical characteristic; 2. The production management method according to claim 1, wherein the method is performed by extracting a second parameter that affects the parameter.