JP2024055609A - Single crystal pulling method - Google Patents

Abstract

[Problem] When pulling a silicon single crystal by the Czochralski method, it is possible to control the thermal history of the latter half of the straight body of the silicon single crystal, and prevent thermal shock when the tail and the melt are separated. [Solution] In the tail forming process, when the final pulling speed of the straight body C2 is v0 and the initial pulling speed of the tail C3 continuing from the straight body is v1, v1 ≧ 0.88 · v0 is satisfied in a first stage in which the crucible 3 is raised to maintain a constant liquid level of the silicon melt M, after the first stage is completed, the pulling speed is gradually reduced to a pulling speed v2 slower than v1 and the lifting of the crucible is stopped, and after the second stage is completed, the pulling speed is adjusted to a pulling speed v3 at which the tail and the silicon melt can be separated. [Selected Figure] Figure 2

Description

The present invention relates to a method for pulling a single crystal, and in particular to a method for pulling a single crystal that can control the thermal history of the latter half of the straight body of a silicon single crystal and prevent thermal shock when the tail and the melt are separated.

In growing silicon single crystals by the Czochralski method (CZ method), as shown in FIG. 10 , a quartz crucible 51 installed in a chamber 50 is filled with polysilicon as a raw material, and the polysilicon is heated and melted by a heater 52 provided around the quartz crucible 51 to obtain a silicon molten liquid M.
Thereafter, a seed crystal P (seed) attached to a seed chuck is immersed in the silicon melt M, and a single crystal C is grown inside the radiation shield 53 by lifting up the seed chuck while rotating the seed chuck and the quartz crucible 51 in the same or opposite directions.

Specifically, the ingot diameter is gradually expanded to form a shoulder portion C1, followed by the growth of a straight body portion C2, which will become the product portion. When the growth of the straight body portion C2 is completed, a tail portion C3 is formed, which gradually reduces the ingot diameter, as shown in FIG. 11, and this tail portion C3 is separated from the silicon melt M. By forming the tail portion C3, dislocations caused by thermal shock that occur when the tail portion C3 is separated from the silicon melt M are prevented from slipping back into the straight body portion C2.

In the prior art, in order to stably separate the tail portion C3 from the silicon melt M, for example, Patent Document 1 (JP Patent Publication No. 2000-26197) attempts to automate the process by utilizing the amount of change in the ingot weight and the amount of change in the melt surface temperature.

The pulling speed of the silicon single crystal is an effective condition for controlling the thermal history. However, in pulling control of the tail section C3, the ingot diameter cannot be reduced unless the pulling speed is changed. Therefore, the melt temperature increase control or the crucible rise control for maintaining a constant liquid surface position is stopped. As a result, in pulling the tail section C3, the pulling speed is limited to a slower speed both thermally and relatively.

Since the thermal history of a certain portion of an ingot depends on the pulling speed of a portion that is pulled up later than that portion, when trying to control the thermal history of the latter half of the straight body portion, it depends on the pulling speed of the tail portion C3. Therefore, for example, when it is desired to shorten the experience time of the thermal temperature zone that promotes the generation of Grown-in defects, if the pulling speed of the tail portion C3 is low, there is a problem that the desired thermal history cannot be achieved in the latter half of the straight body portion.
To address this issue, in Patent Document 2 (JP Patent Publication No. 11-268991), the pulling speed of the tail portion C3 is increased to perform thermal history control to suppress OSF and abnormal oxygen precipitation occurring in the body portion.

JP 2000-26197 A Japanese Patent Application Laid-Open No. 11-268991

However, in the method disclosed in Patent Document 2, when the tail portion C3 is pulled up, the remaining amount of melt decreases and, if the pulling speed is high, a locally low-temperature portion is formed on the melt surface, causing freezing of the silicon melt M. In this case, it is necessary to forcibly separate the tail portion C3 and the frozen silicon melt M before they solidify, which poses a problem of causing a large thermal shock at that time.
In order to solve these problems, a method for manufacturing a silicon single crystal ingot was required that controls the thermal history during the formation of the tail portion C3 and prevents thermal shock when the tail portion C3 is separated from the melt M.

The present invention was made under the above circumstances, and aims to provide a method for pulling a single crystal that can control the thermal history of the latter half of the body of a silicon single crystal when pulling the silicon single crystal by the Czochralski method, and can prevent thermal shock when the tail and the melt are separated.

The single crystal pulling method according to the present invention, which has been made to solve the above problems, is a single crystal pulling method that pulls a single crystal from a silicon melt contained in a crucible in a chamber by the Czochralski method, and includes a straight body forming process in which pulling is controlled so that the diameter of the single crystal ingot is constant, and a tail forming process in which the ingot diameter is reduced to form a tail after the straight body forming process. In the tail forming process, when the final pulling speed of the straight body is v0 and the initial pulling speed of the tail continuing from the straight body is v1, v1 ≧ 0.88 · v0 is satisfied in the first stage in which the crucible is raised to maintain a constant liquid level of the silicon melt, after the first stage is completed, the pulling speed is gradually reduced to a pulling speed v2 slower than v1 and the lifting of the crucible is stopped, and after the second stage is completed, the pulling speed is adjusted to a pulling speed v3 at which the tail and the silicon melt can be separated.

In the tail portion forming step, when a target overall length of the tail portion is L and the length of the tail portion in the first stage is L1, it is preferable that L1≦0.5·L.
In the tail portion forming step, when the final ingot diameter in the first stage is D1 and the ingot diameter in the second stage is D2, it is preferable that D2≦1/2·D1.
In the tail portion forming step, when the length of the tail portion in the third stage is L3, it is preferable that L3≧D2.

Thus, according to the present invention, in the tail portion formation process, when the final pulling speed of the body portion is v0 and the initial pulling speed of the tail portion continuing from the body portion is v1, v1 ≧ 0.88 · v0 is satisfied, and a first stage is provided in which the crucible is raised to maintain a constant liquid level of the molten silicon.
As a result, in the latter half of the body section, particularly in the body section immediately above the tail section, the thermal experience time in the temperature range in which grown-in defects and the like are generated can be shortened, and thermal history control can be performed.
In addition, after the first stage is completed, a second stage is performed in which the pulling speed is gradually slowed down to a pulling speed v2 slower than v1 and the raising of the crucible is stopped (the pulling speed is relatively slowed down), and after the second stage is completed, a third stage is performed in which the pulling speed is adjusted to a pulling speed v3 at which the tail portion and the silicon melt M can be separated.
This makes it possible to raise the melt temperature and suppress freezing of the melt before the remaining amount of melt becomes small in the second stage of tail formation, which in turn makes it possible to mitigate the thermal shock that occurs when the tail is cut off in the third stage of tail formation and to prevent slip-back to the body.

According to the present invention, when pulling a silicon single crystal by the Czochralski method, the thermal history of the latter half of the body of the silicon single crystal can be controlled, and a single crystal pulling method can be provided that can prevent thermal shock when the tail and the melt are separated.

FIG. 1 is a cross-sectional view showing an example of a single crystal pulling apparatus for carrying out the single crystal pulling method according to the present invention. FIG. 2 is a flow chart showing an example of the method for pulling a single crystal according to the present invention. FIG. 3 is a front view showing the tail portion of a silicon single crystal pulled by the single crystal pulling method according to the present invention. FIG. 4 is a graph showing the relationship between the tail length and the pulling rate in the first to third stages of tail formation in Example 1. FIG. 5 is a front view of the tail portion, showing the results of controlling the tail length, pulling rate, and ingot diameter in the formation of the tail portion in Example 1. FIG. 6 is a graph showing the relationship between the tail length and the pulling speed in the first to third stages of tail formation in Comparative Example 1. FIG. 7 is a front view of the tail portion, showing the results of controlling the tail length, pulling rate, and ingot diameter in the formation of the tail portion in Comparative Example 1. FIG. 8 is a graph showing the relationship between the tail length and the pulling speed in the first to third stages of tail formation in Comparative Example 2. FIG. 9 is a front view of the tail portion, showing the results of controlling the tail length, pulling rate, and ingot diameter in the formation of the tail portion in Comparative Example 2. FIG. 10 is a cross-sectional view for explaining a conventional method for pulling a single crystal. FIG. 11 is a cross-sectional view showing a state subsequent to the state of FIG.

The single crystal pulling method according to the present invention will be described below with reference to the drawings. However, this embodiment will be described as an example of the present invention, and the present invention is not limited thereto.

Figure 1 is a cross-sectional view showing an example of a single crystal pulling apparatus in which the single crystal pulling method according to the present invention is carried out. This single crystal pulling apparatus 1 has a furnace body 10 formed by stacking a pull chamber 10b on a cylindrical main chamber 10a, and is equipped with a carbon crucible (or graphite crucible) 2 that is installed in this furnace body 10 so as to be rotatable around a vertical axis and movable up and down, and a quartz glass crucible 3 (hereinafter simply referred to as crucible 3) held by the carbon crucible 2. This crucible 3 is rotatable around the vertical axis together with the rotation of the carbon crucible 2.

Further, below the carbon crucible 2, there are provided a rotation drive unit 14 such as a rotation motor for rotating the carbon crucible 2 around a vertical axis, and an elevation drive unit 15 for moving the carbon crucible 2 up and down.
The rotation drive unit 14 is connected to a rotation drive control unit 14a, and the elevation drive unit 15 is connected to an elevation drive control unit 15a.

The single crystal pulling apparatus 1 also includes a heater 4 of a resistance heating type or a high-frequency induction heating type for heating and melting the semiconductor raw material (raw polysilicon) placed in the crucible 3 to form a silicon melt M.
The single crystal pulling apparatus 1 also includes a pulling mechanism 9 that winds up the wire 6 and pulls up the grown single crystal C. A seed crystal P is attached to the tip of the wire 6 of the pulling mechanism 9. A rotation drive control unit 9a that controls the rotation drive of the pulling mechanism 9 is connected to the pulling mechanism 9.

A radiation shield 7 is placed above the silicon melt M formed in the crucible 3 to surround the single crystal C. This radiation shield 7 has openings at the top and bottom, and serves to block excess radiant heat from the heater 4 and the silicon melt M to the single crystal C being grown, as well as to straighten the gas flow inside the furnace.

The single crystal pulling device 1 also includes an optical diameter measuring sensor 17, such as a CCD camera, for measuring the diameter of the single crystal being grown. A small observation window 10a1 is provided on the top surface of the main chamber 10a, and the change in position of the crystal end (position indicated by the dashed arrow) at the solid-liquid interface can be detected from outside this small window 10a1.

The single crystal pulling device 1 also includes a controller 11 having a memory device 11a and an arithmetic and control device 11b, and the rotation drive control unit 14a, the lift drive control unit 15a, the rotation drive control unit 9a, and the diameter measurement sensor 17 are each connected to the arithmetic and control device 11b.

In the single crystal pulling apparatus 1 thus constructed, when a single crystal C having a diameter of, for example, 300 mm is grown, pulling is performed as follows.
That is, first, raw polysilicon (for example, 460 kg) is loaded into the crucible 3, and a crystal growing process is started based on the program stored in the storage device 11a of the controller 11.

First, a predetermined atmosphere (mainly an inert gas such as argon gas) is created inside the furnace body 10. For example, a furnace atmosphere with an internal furnace pressure of 65 torr and an argon gas flow rate of 90 l/min is created.
Then, while the crucible 3 is rotated in a predetermined direction at a predetermined rotation speed (rpm), the raw material polysilicon loaded in the crucible 3 is melted by heating with the heater 4 to become silicon melt M (step S1 in FIG. 2).

The pulling conditions are adjusted using parameters such as the initial power supply to the heater 4 and the pulling speed, and the seed crystal P starts to rotate around its axis at a predetermined rotation speed. The rotation direction is opposite to that of the crucible 3.

Next, the wire 6 is lowered to bring the seed crystal P into contact with the silicon melt M, and after the tip of the seed crystal P is melted, necking is performed to form a neck portion P1 (step S2 in FIG. 2).
Then, the crystal diameter is gradually enlarged to form a shoulder portion C1 (step S3 in FIG. 2). When the ingot diameter measured by the diameter measuring sensor 17 reaches the desired diameter D0, the controller 11 controls the lift drive control unit 15a to drive and control the lift drive unit 15, keeps the pulling speed v0 constant, for example, at 0.55 mm/min, and proceeds to a process of forming a straight body portion C2 that will become the product portion (step S4 in FIG. 2).

When the straight body portion C2 is formed to a predetermined length, the process proceeds to the final tail portion process (step S5 in FIG. 2).
In this tail section process, if the final pulling speed of the body section C2 is v0, then in the first stage, as shown in Fig. 3, pulling is controlled so that the pulling speed v1 ≥ 0.88 v0 (step S6 in Fig. 2). In this first stage of tail section formation, the melt temperature is slightly increased and the pulling speed v1 is adjusted while the ingot diameter is gradually reduced.

Specifically, in the first stage of tail formation, the tail pulling speed v1 is increased (v1≧0.88·v0). The pulling speed is increased as the ingot diameter is gradually reduced, but the crucible elevation, which is applied to keep the liquid level constant from the time the straight body is pulled up, is continued to increase the relative pulling speed. This shortens the thermal exposure time of the temperature range where grown-in defects are generated in the latter half of the straight body, especially in the straight body directly above the tail. Pulling up the tail, which performs this thermal history control, is considered to be the first stage.
If the final ingot diameter of the body portion C2 is D0 and the target value of the tail portion is L, the first stage of tail formation is terminated when the tail length L1 in the first stage satisfies L1≦0.5·L and the ingot diameter D1 is 0.6×D0 (step S7 in FIG. 2). If the tail length L1 in the first stage is greater than 0.5×L, the tail length L1 may become too long, making it difficult to control the second and subsequent stages.

If the first stage of tail formation were to continue, the reduction in the ingot diameter would be gradual, the remaining amount of melt would be reduced, and the melt temperature would drop due to the faster pulling speed, making the melt more likely to freeze. Therefore, in the second stage of tail formation shown in FIG. 3, the melt temperature is raised before the remaining amount of melt becomes small, thereby preventing the silicon melt M from freezing.
Furthermore, the lifting of the crucible is stopped to relatively slow down the pulling speed (step S8 in FIG. 2), and the reduction is promoted at least until the ingot diameter D2 becomes 50% or less of the first stage final diameter D1 (D2≦½·D1). If the second stage control is not continued until the ingot diameter D2 becomes D2≦½·D1, there is a risk that the silicon melt M will freeze in the subsequent third stage when the pulling speed is increased to v3 at which the tail and the silicon melt M can be separated.
Specifically, the output of the heater 4 is increased to sufficiently increase the melt temperature so that the silicon melt M does not freeze, while the pulling speed v2 is gradually decreased from the pulling speed v1 to promote the reduction of the ingot diameter (step S9 in FIG. 2). Then, when the ingot diameter D2 becomes, for example, 0.3×D1, the second stage is terminated (step S10 in FIG. 2).

After the ingot diameter has been sufficiently reduced in this second stage, the third stage shown in Figure 3 begins, in which the tail, which is longer than the final diameter D2 of the second stage, is pulled up at a pulling speed v3 that allows the tail to be separated from the silicon melt M (step S11 in Figure 2).
Then, when the tail length L3 in the third stage becomes equal to or larger than the ingot diameter D2 (L3≧D2) and equal to or smaller than 10 mm (step S12 in FIG. 2), the tail portion is separated from the silicon melt M to end the third stage and produce a silicon single crystal C (step S13 in FIG. 2).
Here, when the ingot has a minimum diameter such that the tail length L3 in the third stage is finally set to be equal to or greater than D2, the pulling speed v3 for separating the tail from the silicon melt M can be set to a high or low speed without any problem, and can mitigate the thermal shock during separation and prevent slip back to the straight body portion C2 (when the ingot has a minimum diameter such that the tail length L3 in the third stage is less than D2, there is a risk that the thermal shock during separation will be large).

As described above, according to this embodiment, in the tail portion formation process, when the final pulling speed of the body portion C2 is v0 and the initial pulling speed of the tail portion C3 continuing from the body portion C2 is v1, v1 ≧ 0.88 · v0 is satisfied, and a first stage is provided in which the crucible 3 is raised to keep the liquid level of the molten silicon M constant.
As a result, in the latter half of the body section C2, particularly in the body section C2 immediately above the tail section C3, the thermal experience time in the temperature range in which grown-in defects and the like are generated can be shortened, and thermal history control can be performed.
In addition, after the first stage is completed, a second stage is performed in which the pulling speed is gradually slowed to a pulling speed v2 slower than v1 while the temperature of the silicon melt M is increased by controlling the heater output, and the lifting of the crucible is stopped (the pulling speed is relatively slowed), and after the second stage is completed, a third stage is performed in which the pulling speed is adjusted to a pulling speed v3 at which the tail portion C3 and the silicon melt M can be separated.
This makes it possible to raise the melt temperature and suppress freezing of the melt before the remaining amount of melt becomes small in the second stage of tail formation, which in turn makes it possible to mitigate the thermal shock that occurs when the tail is cut off in the third stage of tail formation and to prevent slip back to the body portion C2.

The single crystal pulling method according to the present invention will be further explained with reference to examples.

In this example, in the single crystal pulling apparatus shown in Fig. 1, a quartz crucible having a diameter of 32 inches was filled with 460 kg of silicon raw material to form a silicon melt. Boron was added to the silicon melt as a dopant, and a silicon single crystal doped with high concentration of boron was pulled.
The evaluation was carried out by inspecting the pulled silicon single crystal for the presence or absence of grown-in defects (LPDs) using a surface inspection instrument (SPI).
LPDs are likely to occur in the latter half of the formation of a silicon single crystal ingot, and in order to reduce them, thermal history control was performed during pulling to shorten the thermal exposure time in the temperature range of 1000° C. to 900° C.

Example 1
In Example 1, assuming that the final pulling speed of the body portion is v0 and the final ingot diameter of the body portion is D0, the melt temperature is raised slightly in the first stage of the tail portion to gradually reduce the ingot diameter while adjusting the pulling speed v1 in the first stage (v1≧0.88×v0). Thereafter, assuming that the target length of the entire tail portion is L, the first stage is terminated when the tail length L1 in the first stage becomes 0.3×L and the ingot diameter D1 becomes 0.6×D0.

In the next second stage, the lifting of the crucible was stopped, and the melt temperature was raised sufficiently to gradually slow down the pulling speed v2 in the second stage while being careful not to freeze the melt, and to promote the reduction of the ingot diameter. Then, the second stage was terminated when the ingot diameter D2 in the second stage reached 0.3 x D1.

In the next third stage, the temperature of the melt was further increased, but since the ingot diameter showed a tendency to expand, the pulling speed v3 in the third stage was increased to reduce the ingot diameter, and pulling of the tail continued until the tail length in the third stage L3 was equal to or greater than D2, ultimately reducing the ingot diameter to the minimum diameter of 10 mm or less, and the tail was separated from the melt.

The relationship between the tail length (mm) and the pulling rate (mm/100 min) in the first to third stages of tail formation in Example 1 is shown in the graph of Figure 4. The results of controlling the tail length, pulling rate, and ingot diameter in the tail formation are shown in Figure 5.
By forming the tail portion in this manner in Example 1, it was possible to prevent slip back to the body portion.
Furthermore, when the number of defects in the ingot portion corresponding to the rear half of the straight body was checked using a surface inspection device (SPI), it was confirmed that the number of LPDs had been reduced.

(Comparative Example 1)
In Comparative Example 1, in the first stage of tail formation, the melt temperature was actively increased to promote reduction in the ingot diameter while raising the crucible, so that v1 was slowed down to 0.56×v0.
In the second stage, the lifting of the crucible was stopped and the pulling speed v2 was gradually decreased. Then, when the ingot diameter D2 became 0.3×D1, the second stage was terminated.
In the third stage, the temperature of the melt was further increased, and when the tail length L3 became equal to or greater than D2 and the ingot diameter became 10 mm or less, v3 was adjusted to separate the tail from the melt.

The relationship between the tail length (mm) and the pulling rate (mm/100 min) in the first to third stages of tail portion formation in Comparative Example 1 is shown in the graph of Fig. 6. The results of controlling the tail length, pulling rate, and ingot diameter in the tail portion formation are shown in Fig. 7.
Although the formation of the tail portion in this manner in Comparative Example 1 prevented slip back to the body portion, the desired thermal history control was not achieved.

As described above, from the results of Example 1 and Comparative Example 1, it was confirmed that, in the first stage of tail formation, by adjusting the pulling speed v1 to v1 ≧ 0.88 × v0, it is possible to reduce the number of defects in the ingot portion corresponding to the latter half of the body portion (it is possible to perform the desired thermal history control).

(Comparative Example 2)
In Comparative Example 2, in the first stage of tail formation, the crucible was elevated while suppressing the temperature rise of the melt to satisfy the desired thermal history control, and the pulling speed v1 was increased to 1.05×v0. However, the tail length L1 was increased.
In the second stage, the lifting of the crucible was stopped, but the ingot diameter tended to increase because the melt temperature was not sufficiently high. Therefore, the speed of v2 was increased, but the ingot diameter was not reduced sufficiently and the remaining amount of melt was insufficient, so the tail and the melt were separated halfway through pulling up the tail.

The relationship between the tail length (mm) and the pulling rate (mm/100 min) from the first stage to the second stage of tail portion formation in Comparative Example 2 is shown in the graph of Fig. 8. The control results of the tail length, pulling rate, and ingot diameter in the tail portion formation are shown in Fig. 9.
Such tail formation in Comparative Example 2 caused slip back to the body portion. It was also confirmed that when the tail length L1 in the first stage was greater than 0.5×L, the tail length L1 became too long, making it difficult to control the second and subsequent stages.

As a result of this example, it was confirmed that the present invention can prevent slipback up to the body section and achieve the desired thermal history control in the latter half of the body section.

1. Silicon single crystal (single crystal)
3. Quartz glass crucible (crucible)
4 heater 6 wire 7 radiation shield C silicon single crystal M silicon melt C2 straight body C3 tail

Claims (4)

A method for pulling a single crystal by the Czochralski method from a silicon melt contained in a crucible in a chamber, comprising the steps of:
a straight body forming step of controlling pulling so that the diameter of the single crystal ingot is constant;
a tail portion forming step of reducing an ingot diameter to form a tail portion after the body portion forming step,
In the tail portion forming step,
a first stage in which the crucible is raised so that v1≧0.88·v0 is satisfied, where v0 is a final pulling speed of the body portion, and v1 is an initial pulling speed of a tail portion continuing from the body portion, and the liquid level of the molten silicon is kept constant;
After the first stage is completed, the pulling speed is gradually reduced to a pulling speed v2 which is slower than v1, and the lifting of the crucible is stopped in a second stage.
After the second stage is completed, a third stage is performed in which the pulling speed is adjusted to a pulling speed v3 at which the tail portion and the silicon melt can be separated from each other.
A method for pulling a single crystal, comprising the steps of: In the tail portion forming step,
2. The method for pulling a single crystal according to claim 1, wherein L1≦0.5·L is satisfied, where L is a target total length of the tail portion and L1 is a length of the tail portion in the first stage. In the tail portion forming step,
2. The method for pulling a single crystal according to claim 1, wherein, when the final ingot diameter in the first stage is D1 and the ingot diameter in the second stage is D2, D2≦1/2·D1. In the tail portion forming step,
4. The method for pulling a single crystal according to claim 3, wherein, when the length of the tail portion in the third stage is L3, L3≧D2.

Images