JPH09255468A - Part management of single crystal pulling device and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for managing life times of parts in a single crystal pulling device, quantitatively grasping the life times of the parts and enabling the quality stabilization of the pulled up single crystal and the reduction of products below standards, and a device used to perform the method. SOLUTION: This part management device 1 is equipped with a calculating part 2 for a regression coefficient and an alarming part 3, and connected to plural single crystal pulling devices 4. The calculating part 2 for the regression coefficient prescribes a life time for each of the parts based on the inputted data and feeds to the alarming part 3. When a datum on a number of uses or a used time of each of the parts obtained from the single crystal pulling up devices 4, reaches to a value prescribed by the calculating part 2 for the regression coefficient, the alarming part 3 judges that the part reaches to its life time and gives a warning by a method such as by buzzing or displaying on a displaying device to prompt the exchange of the part.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a component management method in a single crystal pulling apparatus for growing a single crystal such as a silicon single crystal used as a semiconductor material, and a component management apparatus used for implementing the method.

[0002]

2. Description of the Related Art There are various crystal growth methods,
One of them is the Czochralski method (CZ method). Figure 7
FIG. 3 is a schematic vertical cross-sectional view showing a crystal pulling apparatus used for the CZ method. In the figure, 11 is a chamber having a cylindrical shape with a bottom, and a pull chamber having a smaller diameter than the chamber 11 is provided above it.
12 are connected at the same axis. A crucible 13 is disposed in the chamber 11, and the crucible 13 is composed of a graphite crucible 13a having a bottomed cylindrical shape and a quartz crucible 13b fitted inside the graphite crucible 13a. The crucible 13 is connected to and supported by a support shaft 18 connected to a lifting / rotating mechanism (not shown).
It can be raised and lowered and rotated. A heater 14 having a cylindrical shape concentric with the crucible 13 is disposed outside the crucible 13. In addition, a seed crystal 16a was attached to the tip of the seed chuck.
The lifting shaft (wire) 16 that can be attached and detached is a crucible
It is located above the center of 13. The lifting shaft 16 is also connected to an elevating / rotating mechanism (not shown) to elevate and lower the supporting shaft.
It has the same shaft center as 18 and can rotate in the same direction as the support shaft 18 but in the opposite direction.

When performing crystal growth, first, the crucible 13 is filled with silicon (Si) which is a raw material for crystal, and the raw material for crystal is heated by rotating the crucible 13 in a predetermined direction at a predetermined rotational speed.
Heat and melt at 14. Then, the pressure inside the furnace is kept low by a vacuum pump (not shown), and an inert gas is introduced from the pull chamber 12 into the chamber 11. Further, a seed crystal 16a is attached to the tip of the pulling shaft 16, the seed crystal 16a is lowered until it comes into contact with the melt 15, and then the seed crystal 16a is pulled upward while rotating in the opposite direction to the crucible 13. Then, the melt 15 in contact with the lower end of the seed crystal 16a is solidified and the Si single crystal 17 can be grown.

When pulling the Si single crystal 17 by the CZ method,
SiO 2 is evaporated from the melt 15 and adheres to graphite parts such as the heater 14 and the graphite crucible 13a. Then the attached Si
O and graphite react to cause deterioration of these parts.
Further, the adhered SiO is removed at the end of pulling up, but at this time, the graphite parts are worn and the parts are also deteriorated.
Due to such deterioration of the graphite part, the atmosphere in the furnace is deteriorated, for example, the graphite concentration in the atmosphere in the furnace is increased.
As a result, the impurity concentration in the Si single crystal 17 increases, or the rate at which the dislocation-free Si single crystal 17 is obtained decreases. Also, when the graphite heater 14 deteriorates, the calorific value distribution changes. Further, due to deterioration of the heater 14 and the graphite crucible 13a, the amount of heat input to the melt 15 is changed, and the initial oxygen concentration in the Si single crystal 17 is varied, resulting in non-standard products.

Even in parts other than those made of graphite, the heat reflectance may change (decrease) over time due to the adhesion of SiO 2, and the yield rate of the oxide film in the case of a wafer may decrease. In addition, if the time from the start of melting all the raw materials until the start of pulling (hereinafter referred to as the pulling time), which is an operating parameter, becomes long due to a trouble or the like, the contamination degree of the melt increases, and thus the pulling rate of dislocation-free crystals decreases .

Therefore, as a conventional device for extending the life of graphite parts, a chamber for separating from the inert gas introduction part and a pull-up chamber (chamber) is provided so that SiO 2 does not adhere to the parts in the furnace.
Japanese Unexamined Patent Publication (Kokai) No. 6-16491 discloses a silicon single crystal pulling apparatus provided with an inert gas exhaust portion at the upper position. Further, JP-A-58-172292 discloses that the above-mentioned SiO
A graphite heating element coated with SiC in order to suppress the reaction between graphite and graphite is disclosed.

[0007]

However, in these conventional devices, the purpose is to extend the service life of the parts, but there is no mention of means for determining the time to replace the parts when the service life approaches. Therefore, the life of a component is determined empirically, and the component may be replaced more than necessary. Further, the quality of parts is gradually improved, and it is necessary to review the period until the parts are replaced each time, and it is very difficult to accurately determine the life of the parts only by empirical knowledge.

The present invention has been made in view of such circumstances, and by formulating the relationship between management data such as operating conditions, the number of times a part is used, and the cumulative use time, and the quality of a single crystal, It is an object of the present invention to provide a component management method for a single crystal pulling apparatus and a device used for carrying out the method, which can grasp the lifespan quantitatively and stabilize the quality of the pulled single crystal and reduce nonstandard products.

[0009]

According to the first and third aspects of the present invention, the operating data of the single crystal pulling apparatus, the number of times the parts are used, or the time when the parts are used are managed for each part and the pulling single crystal. The relationship between the quality of the single crystal, the lower limit of the quality of the single crystal is set, the limit value of the control data when the lower limit is less than is determined based on the relationship,
The management data is managed for each part, and when the management data reaches the limit value, a warning is issued to prompt replacement of the part.

Since the life management of the parts can be quantitatively performed, it becomes easy to accurately grasp the replacement time of the parts. This makes it possible to stabilize the quality of the pulled single crystal and reduce nonstandard products, and also reduce the possibility of exchanging parts more than necessary.

The invention according to claims 2 and 4 is characterized in that the relationship between the management data and the quality of the single crystal is obtained by regression coefficient calculation.

By grasping the correlation between the control data and the quality of the single crystal, the life of the component can be set accurately.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be specifically described below with reference to the drawings showing the embodiments. FIG. 1 is a schematic block diagram showing a component management device according to the present invention. The component management device 1 includes a regression coefficient calculation unit 2 and an alarm unit 3, and is connected to a plurality of single crystal pulling devices 4. The regression coefficient calculation unit 2 sets the life of each component based on the input data and gives it to the alarm unit 3. Further, the number of times of use, the time of use, and the operating condition data of each component obtained from each single crystal pulling apparatus 4 are given to the regression coefficient calculation unit 2 and the alarm unit 3. The alarm unit 3 determines that the part has reached the end of life when the number of times of use and the time of use of each part obtained from the single crystal pulling apparatus 4 reach the values set by the regression coefficient calculation unit 2. However, in order to prompt replacement of the part, a warning is issued by sounding a buzzer, displaying on a display device, or the like.

FIG. 2 is a flow chart showing a processing procedure of a component life management method in the component management apparatus according to the present invention. The number of times each part is used, the cumulative usage time of each part, each part quality data, each device operation parameter, the single crystal quality lower limit value, and multiple regression parameters are given as inputs to the regression coefficient calculation unit 2 (step S1). After each pulling, the pulling time in the pulling batch is transmitted from each pulling single crystal device 4 (step S2).
Then, the regression coefficient calculation unit 2 adds the value to the component cumulative use time and adds 1 to the component use count (step S3). Then, after confirming that these values have not reached the end of their lives (step S4), at the time when this data has been collected for several batches, the regression coefficient calculation unit 2 calculates the coefficient matrix A _i (i = 0, 1, ...) is performed (step S5), and the regression coefficient for each factor obtained by this multiple regression is given to the regression coefficient calculation unit 2 as a multiple regression parameter.

[0015]

[Equation 1]

The life of the part is estimated from the result of the multiple regression and the preset lower limit of quality of the part, and the data of the life is output to the alarm unit 3. As a result, the life data of the alarm unit 3 is set (step S
6) Wait for the next transmission data. In step S4,
When the number-of-uses data or the usage-time data of the parts obtained from each single crystal pulling device 4 reaches this life data,
The alarm unit 3 issues an alarm (step S7).
Even when one component is used while being rotated in several furnaces, the same process can be performed by managing the component numbers.

FIG. 3 is a graph showing the results of multiple regression of the relationship between the number of times the graphite part A is used and the number ratio of dislocation-free pulling (hereinafter referred to as dislocation-free pulling rate) in the entire required area of the single crystal. . It can be seen that with this part A, the dislocation-free pulling rate has been reduced by nearly 30% by using it about 30 times. FIG. 4 is a graph showing the results of multiple regression of the relationship between the extension time and the dislocation-free pulling rate. It can be seen that the dislocation-free pulling rate deteriorates as the pulling time increases. FIG.
FIG. 4 is a graph showing a multiple regression result of the relationship between the number of times the heater has been used and the oxygen concentration in a single crystal. Thus, the oxygen concentration in the single crystal gradually decreases as the number of times the heater is used increases. FIG. 6 is a graph showing a multiple regression result of the relationship between the number of times the chamber is used and the yield rate of the oxide film withstand voltage. In this way, the rate of non-defective oxide film breakdown voltage decreases with the aging of the chamber.

[0018]

As described above, according to the present invention, the life of a component is quantitatively grasped by formulating the relation between the management data such as the operating condition, the number of times of use of the component, the cumulative use time and the quality of the single crystal. Therefore, it is possible to stabilize the quality of the pulled single crystal, reduce the nonstandard product, and reduce the possibility of exchanging parts more than necessary, and the present invention has excellent effects.

[Brief description of drawings]

FIG. 1 is a schematic block diagram showing a component management device according to the present invention.

FIG. 2 is a flowchart showing a processing procedure of a component management method according to the present invention.

FIG. 3 is a graph showing a multiple regression result of the relationship between the number of times the graphite part is used and the dislocation-free pulling rate.

FIG. 4 is a graph showing the results of multiple regression of the relationship between extension time and dislocation-free pulling rate.

FIG. 5 is a graph showing a multiple regression result of the relationship between the number of times the heater is used and the oxygen concentration in a single crystal.

FIG. 6 is a graph showing a multiple regression result of the relationship between the number of times the chamber has been used and the yield rate of oxide film non-defective products.

FIG. 7 is a schematic vertical sectional view showing a single crystal pulling apparatus used in the CZ method.

[Explanation of symbols]

 1 parts management device 2 regression coefficient calculation unit 3 alarm unit 4 single crystal pulling device

Claims (4)

[Claims] 1. A relationship between management data managed for each part and the quality of the pulled single crystal regarding the operating condition of the single crystal pulling apparatus, the number of times of use of the part, or the time of use of the part is obtained, and the quality of the single crystal is calculated. When the lower limit value is set, the limit value of the management data when it falls below the lower limit value is determined based on the relationship, the management data is managed for each component, and the management data reaches the limit value. In addition, a parts management method is characterized in that a warning is issued to prompt replacement of the part. 2. The part management method according to claim 1, wherein the relationship between the management data and the quality of the single crystal is calculated by regression coefficient calculation. 3. The relationship between the operating data of the single crystal pulling apparatus, the number of times of use of the component, or the time of use of the component, which is managed for each component, and the quality of the pulled single crystal is obtained in advance and set in advance. A means for judging the limit value of the control data when it is below the lower limit value of the quality of the single crystal based on the relation, a means for managing the control data for each part, and the control data is the limit value. And a means for issuing a warning urging replacement of the part when the part management device has reached. 4. The component management apparatus according to claim 3, further comprising a regression coefficient calculation unit that calculates a relationship between the management data and the quality of the single crystal by a regression coefficient calculation.