JPH0745355B2 - Crystal growth method and apparatus - Google Patents

Description

The present invention relates to a method for growing a silicon single crystal used as a material for a semiconductor device, and an apparatus therefor.

[Prior art]

There are various methods for growing a silicon single crystal,
Among them, there is a rotary pulling method such as Czochralski method (CZ method). FIG. 6 is a schematic vertical cross-sectional view of a conventional rotary pull-up type crystal device, in which 1 is a crucible.

The crucible 1 is a bottomed cylindrical quartz inner layer holding container 1b and a bottomed cylindrical graphite outer layer holding container 1a fitted to hold the inner layer holding container 1b outside the inner layer holding container 1b. It is composed of.

A resistance heating type heater 12 is arranged outside the crucible 1, and a graphite heat insulating cylinder (not shown) is concentrically arranged outside the crucible 1. A heater with a certain weight of raw material is placed in the crucible 1.
A melt 13 melted by 12 is filled. On the central axis of the crucible 1, a pull-up rod (or wire, both of which are hereinafter referred to as a "pull-up rod") 14 which rotates at a predetermined speed in the arrow direction is arranged. Tsurubo 1 is a pulling rod 14
Crucible support shaft that rotates at the same speed in the opposite direction with the same axis as
It is supported by 14a. Then, the seed crystal 15a attached to the pulling rod 14 is brought into contact with the surface of the melt liquid 13, and the pulling rod 14 is rotated upward in accordance with crystal growth and pulled upward to solidify the melt liquid 13, A silicon single crystal 15 is grown.

Conventionally, when the silicon single crystal 4 is grown by the rotary pulling method, impurities are collectively added to the melt 13 before pulling to prepare the electric resistivity and electric conductivity type of the silicon crystal. There is a problem that impurities segregate along the pulling direction of the silicon single crystal 15 and a silicon single crystal having uniform electric characteristics in the pulling direction cannot be obtained.

This segregation is the ratio of the impurity concentration at the start of solidification at a certain point of the silicon single crystal to the impurity concentration at the end of solidification, that is, the impurity concentration C in the single crystal that actually occurs at the melt / single crystal interface during solidification. the ratio of _s and impurity concentration C _L in the melt C _s / C _L, i.e., caused by the effective segregation coefficient Ke. More specifically, when Ke <1, for example, the concentration of impurities in the molten liquid naturally increases as the silicon single crystal grows, and segregation occurs in the silicon single crystal. The effective coefficient Ke is publicly known.

There is a melt layer method as a method of suppressing the occurrence of the above-mentioned segregation and growing a silicon single crystal by the rotary pulling method.

FIG. 7 is a schematic vertical cross-sectional view of a conventional crystal growth apparatus using the melt layer method. The upper layer portion of the silicon single crystal raw material inserted in the crucible 1 having the same structure as in FIG. And melted to form a molten layer 6 on the upper layer,
The lower layer becomes the solid layer 7. The solid layer 7 is melted by the heater 12 as the pulling rod 14 is pulled up, so that the amount of the molten liquid in the crucible 1 is kept constant (the molten layer thickness constant method). In the case of this method, the impurity concentration C _L is reduced by melting a solid layer having a low impurity concentration newly as the single crystal grows regardless of the value of the effective segregation coefficient Ke (Japanese Examined Patent Publication No. 34-8242, Japanese Patent Publication No. 62-880).

Alternatively, a method of suppressing segregation by raising and lowering the crucible 1 or the heater 12 along with the growth of the silicon single crystal 15 and changing the amount of the melt in the crucible 1 (melt layer thickness changing method)
Is disclosed in Japanese Patent Laid-Open No. 61-205691.

By the way, the principle of reducing the segregation in the above-mentioned melting layer method is that the weight of the molten liquid initially filled in the crucible 1 (initial filling amount) is 1 and the impurity concentration at the position of the weight ratio x measured from the upper surface of the raw material is By expressing it as C _p (x), it can be explained by the one-dimensional model as shown in FIGS. At this time, the crystal pulling rate with respect to the initial filling amount of 1 is f _s , the weight ratio f of the melt (layer) is f _L , and the weight ratio of the raw materials (lower solid content)
Let f _p , f ₀ = f _s + f _L be defined as in the following equation (1).

f ₀ + f _p = f _s + f _L + f _p = 1 (1) In the rotary pulling method such as the CZ method, high-purity polycrystal is often used as a raw material. The case where the impurity concentration C _p ≠ 0 will be described. Further, the left side in the figure is the upper surface side of the crucible 1.

FIG. 8 shows a state immediately after the whole amount is inserted into the crucible 1, where f _p = 1. Figure 9 is a raw material of Figure 8 is melted from the raw material upper surface only f _L, it shows the state of the initial dissolution at the end of the addition of impurities.

C ₀ is the impurity concentration in the initial melt, and f _o = f _L. First
Figure 10 shows the changes during crystal pulling. The crystal is pulled up by f _s from the upper surface of the raw material, and the raw material is further melted by f _L. C _L is the impurity concentration in the melt, and C _p is the impurity concentration in the lower solid layer. C _a while further by Delta] f _s from f _s pulling the crystal
・ When impurities are added by Δf _s, f _L is f _L + Δf _L , and C _L is
C _L + ΔC _L , and f _p changes to f _p + Δf _p . C _s is the impurity concentration in the crystal. At this time, the amount of impurities in the region indicated by A in the figure is constant. As a result, the following equation (2) is established.

C _L · f _L + C _a · Δf _s + C _p · Δf ₀ = C _s · Δf _s + (C _L + ΔC _L ) · (f _L + Δf _L ) ... (2) where C _s = Ke · C _L … ( 3) However, since Ke is the effective segregation coefficient, this is applied to equation (2), and 2 in equation (2) is applied.
The following expression (4) is obtained by omitting the next minute term.

From equation (4), for example, in an ideal case, C _p = 0,
The segregation can be obtained by adding the impurity concentration C _{s in the} crystal as follows. That is, in the case of the ordinary CZ method, from f _p = 0, Δf _L ＋ Δ.f _s = 0, C _a = 0 Substituting this into Eq. (3), C _s = KeC ₀ (1-f _s ) ^Ke-1 (6).

Similarly, in the case of the melted layer method, if dC _L / df _s = 0 and C _p = 0, then according to equation (4), This is the condition for realizing the segregation-free pulling. When this is applied to the method of constant melt layer thickness, df _L / df _s
= 0, C _a = KeC _L = KeC ₀ (8) is obtained. By continuously adding the impurity amount C _a , the segregation-free condition is realized. In addition, when applied to the melt layer thickness variation method, since continuous addition of impurities is not performed, C _a = 0.
And from equation (7) In order to satisfy the above condition, the thickness of the molten layer is changed as the crystal is pulled up.

Figure 11 shows the distribution at the end of pulling. In the constant melt layer thickness method, the solid layer under the melt 13 is completely melted and f _p = 0.
After that, the non-segregation condition is not satisfied, and segregation occurs according to the equation (6). On the other hand, in the melt layer thickness variation method, if the initial melt rate is f _L0 , then f _L = f _L0 −Kef _s (10) from equation (9). Since Ke <1, by setting f _L0 = Ke, the segregation condition can be maintained until the end of pulling, and segregation is reduced.

In the molten layer thickness changing method, the molten layer thickness is controlled by appropriately selecting in advance the heat generation length of the heater 12, the depth of the crucible 1, the shape of the heat retaining cylinder, and the material. FIG. 12 is an explanatory view showing the temperature distribution on the central axis in the conventional crystal growth method and its apparatus similar to FIG. In the figure, T _m is the molten layer 6 and the solid layer 7
It is the boundary temperature with the temperature and is a constant value determined by the melting point of the raw material. T _b is the temperature of the bottom upper surface of the quartz inner layer holding container 1 b, T _c
Is the boundary temperature between the bottom lower surface of the quartz inner layer holding container 1b and the graphite outer layer holding container 1a, T _p is the bottom lower surface temperature of the graphite outer layer holding container 1a, and T _o is the temperature below the support shaft 14a.

Here the heater 12 power (heat value) is therefore substantially be set to a constant solid layer 7, the amount of heat Q _L dissipated downward by heat conduction via the support shaft 14a is constant. Therefore, the following equation (11) is approximately established from FIG.

Where λ ₇ : thermal conductivity of the solid layer 7 λ ₁ : thermal conductivity of the inner layer holding container 1b λ _c : thermal conductivity of the outer layer holding container 1a λ _p : thermal conductivity of the support shaft 14a S _c : inner layer holding container 1b Inner cross-sectional area S _p : Cross-sectional area of the support shaft 14a l ₇ : Axial length of the solid layer 7 l ₁ : Bottom direction length of the quartz inner layer holding container 1b l _c : Bottom of the graphite outer layer holding container 1a Direction length l _p : Length of the support shaft 14a in the vacuum vessel Eliminating T _b , T _c , T _p from Eq. (11), On the other hand, in normal crystal pulling, the surface position of the melt is kept constant, so l in FIG. 12 is constant, and Δl ₆ + Δl ₇ + Δl _p = 0 (13) where l ₆ : molten layer The relationship of the axial length of 6 is established. Further, using Δl ₆ / Δl ₇ = Δf _L / Δf _p (14) and Δf _s + Δf _L + Δf _p = 0 (15), Δf _s ∞Δl _p (16)

Applying these to equation (12), Becomes That is, if the heat conductivity λ ₇ S _{c of} the solid layer 7 and the heat conductivity λ _p S _{p of} the support shaft 14a are equal, Δf _L (change amount of molten layer thickness) is kept constant, and λ _p S _p > λ ₇ If it is S _c , the heat transfer condition that Δf _L decreases with the increase is satisfied. It is possible to control the melt layer thickness based on the heat transfer conditions.

[Problems to be Solved by the Invention] As described above, when pulling is performed by the molten layer thickness change method based on the non-segregation condition shown in the equation (10) and the heat transfer condition shown in the equation (17), Is a metal that can suppress segregation uniformly over the entire length of the crystal.

However, in the crystal pulling by the melt layer method, the solid density of the raw material of the crystal used as the material of the semiconductor device is smaller than the melt density, so that the crystal pulling actually progresses and the thickness of the solid layer 7 reaches the levitation limit. Within a certain value, the solid layer 7 floats in the molten layer 6 and hinders crystal pulling. Therefore, in the crystal pulling by the conventional crystal growth apparatus, the crystal pulling is finished with the raw materials (the molten layer 6 and the solid layer 7) remaining in the crucible 1, and the molten layer at the end of pulling 6 and solid layer 7
There is a problem that the manufacturing yield of the crystals is poor, which is determined by the ratio of C, that is, the crystal pulling rate at the end of pulling.

In order to solve the above problems, the inventors of the present invention conducted research and experiments to improve the production yield of crystals, and found that the ratio of the molten layer 6 and the solid layer 7 was the heat transfer condition described in the prior art. In addition, the heater 12 reflected from the heat insulation cylinder 8
It was found that it depends on the distribution of the radiant heat flow rate q. In other words, if the amount of heat diffused downward by heat conduction through the raw material and the support shaft 14a is the same, the smaller the radiant heat flow rate q radiated to the position where the solid layer 7 is to be formed, the less the radiant heat flow rate to the molten layer The ratio of the solid layer 7 becomes large.

However, in the conventional crystal growth method and apparatus, the heat insulating cylinder 8 causes the radiant heat from the heater 12 to radiate to the region extending from above the crucible 1 to the upper end of the support shaft 14a, thereby improving the heating efficiency of the heater 12. Radiant heat to be reflected from the heat insulating cylinder 8 is also radiated to the crucible 1 corresponding to the position where the solid layer 7 is to be formed. Therefore, the solid layer 7 is melted more than necessary, the ratio of the solid layer 7 becomes small, the amount of raw material at the end of pulling increases, and the production yield of crystals is poor. Therefore, in order to reduce the radiant heat in order to make the solid layer 7 thicker than the floating limit, a heat shield is provided below the heater 12 so that the solid layer 7 is formed below the crucible 1. It is sufficient to absorb the radiant heat radiated to the vicinity of the side.

The present invention has been made on the basis of such findings, the solid layer can be made thicker than the levitation limit, the raw material amount of the crucible at the end of pulling is reduced, and a crystal growth method for improving the production yield of crystals and the method thereof. The purpose is to provide a device.

[Means for Solving the Problems]

A crystal growth method according to a first aspect of the invention is a method for producing a silicon single crystal by using a melt layer method, which is a method for keeping a silicon single crystal above a crucible from a crucible filled with a raw material for a crystal placed in a chamber. In the process of pulling up toward the inside of the cylinder, the heat radiated from the heat-retaining cylinder to the lower part of the crucible is shielded by a heat shield installed on the outer circumference of the lower wall of the crucible.

A crystal growing apparatus which is a second invention is a crystal growing apparatus for producing a silicon single crystal by using a molten layer method, wherein a heater is provided only on the outer circumference of a side wall of a crucible filled with a crystal raw material, At least one heat shield is provided below the heater.

[Action]

In the crystal growth method according to the first aspect of the invention, at least one heat shield that shields the radiant heat from the heat insulation cylinder is provided below the heater provided around the crucible, and the crucible is provided by the heat shield. The radiant heat radiated to the lower side of the crucible is shielded, and the amount of radiant heat to the lower side of the crucible is reduced.

This makes it possible to increase the ratio of the solid layer formed near the lower side of the crucible to the molten layer, making the solid layer thicker than the levitation limit, efficiently pulling up the molten layer for crystal growth, and at the end of pulling. The amount of raw material in the crucible is reduced.

In the melting layer method, since a silicon single crystal is grown while a solid layer is present at the bottom of the crucible, an auxiliary heater or the like for heating the bottom of the crucible to melt the solid layer is not installed.

〔Example〕

The present invention will be specifically described below with reference to the drawings. FIG. 1 is a schematic vertical sectional view showing a crystal growth apparatus of the present invention, and 10 in the drawing shows a chamber set to a required degree of vacuum. At the center of the upper surface of the chamber 10, there is a pulling rod 4 which rotates in the arrow direction at a predetermined speed. A seed crystal 5a is attached to the pulling rod 4.

At the center of the bottom surface of the chamber 10, there is a double structure having the same axis as the pulling rod 4 as will be described later, and the supporting shaft 4a of the crucible 1 rotating at a predetermined speed in the opposite direction to the pulling rod 4 has a supporting shaft 4a. At the tip, a crucible 1 having a double structure is attached inside a cylindrical outer layer holding container 1a having a bottom and a cylindrical inner layer holding container 1b made of quartz is fitted therein. A storage box (not shown) for storing impurities is provided in the chamber 10 above the crucible 1. When the bottom lid of the chamber is opened by an opening / closing means (not shown), impurities can be added to the inner layer holding container 1b. ing.

A resistance heating type heater 2 is provided at a position slightly outside the rotation range of the crucible 1 and a length in the axial direction from above the crucible 1 to the upper end of the support shaft 4a is provided at a position between the chamber 10 and the chamber 10 further outside thereof. The heat retaining cylinders 8 are respectively arranged in concentric circles.
The heater 2 has an axial length that is appropriately shorter than that of the crucible 1, and is arranged with its lower end in the axial direction positioned slightly above the bottom of the crucible 1.

Below the heater 2 are three molybdenum heat shields.
22,22,22 are arranged coaxially with the heater 2.

An upper melting layer 6 and a lower solid layer 7 are formed in the crucible 1 by melting a predetermined weight of the upper layer portion of the solid single crystal raw material by the heater 2.

In addition, a raw material supply device 11 is provided above the chamber 10 so that the raw material can be put into the crucible 1 after the solid raw material is taken out from a hopper (not shown) for storing the solid raw material in the form of small pieces or particles and weighed. It is set up.

In the crystal growth method and the apparatus therefor configured as described above, the molten layer 6 and the solid layer 7 having a predetermined weight are formed, and the seed crystal 5a attached to the pulling rod 4 is brought into contact with the surface of the molten layer 6. Let Then, by pulling the pulling rod 4 upward while rotating the pulling rod 4 in accordance with the crystal growth, the molten liquid is solidified and the silicon single crystal 5 is grown.

As the crystal grows, the solid layer 7 is melted and pulled up by controlling the position of the crucible 1 and / or the temperature of the heater 2.

At this time, the radiant heat radiated from the heat insulating cylinder 8 to the portion where the solid layer 7 of the crucible 1 is formed, that is, near the lower side of the crucible 1, is absorbed by the heat shields 22, 22, 22.

When n heat shields 22 are provided, the amount of radiant heat radiated from the heat insulating cylinder 8 to the portion of the crucible 1 where the solid layer 7 is formed q _n
Is obtained by the following equation according to the Stefan-Boltzmann law.

Where ε ₁ is the emissivity of the heat shield 22 ε _{0 is} the emissivity of the crucible 1 and the heat insulation tube 8 θ _{n + 1 is} the absolute temperature of the heat insulation tube 8 θ _{0 is} the absolute temperature of the crucible 1 σ is the Stefan-Boltzmann constant ( From the equation (18), when the heat shield 22 is not provided (n = 0), the radiant heat quantity q ₀ and the radiant heat quantity q _n radiated from the heat retaining cylinder 8 to the portion where the solid layer 7 of the crucible 1 is formed Heat flow ratio
q _n / q ₀ is However, ε ₁ : 0.1 (emissivity of the Mo heat shield 12) ε ₀ = 0.9 (emissivity of the graphite crucible 1 and the heat insulating cylinder 8).

FIG. 2 is a graph showing the heat flow rate ratio q _n / q ₀ calculated by the equation (19) in correspondence with the number of the heat shields 22, and the vertical axis represents the heat flow rate ratio q _n / q ₀ and the horizontal direction. The axis is the number of heat shields 22. As is clear from the figure, by providing one heat shield 22, the heat flow rate ratio q _n / q ₀ is reduced by 94% as compared with the case where n = 0. When two heat shields 22 are provided, the heat flow rate ratio q _n is about 3%.
/ q ₀ can be reduced, heat flow ratio q _n / q ₀ in inverse proportion as the number of the thermal shield 22 increases as that three in the case of providing can be reduced further to about 1% is reduced.

[Experimental example]

In the crystal growth method and apparatus of the present invention, a heat shield
22,22,22 thickness 5mm, height 300mm, inner diameter 300mm, 40 respectively
Three cylinders of 0 mm and 500 mm made of molybdenum are arranged below the heater 2 concentrically with the heater 2, and polycrystalline silicon is used as a raw material, and an effective segregation coefficient Ke of the raw material is 0.5 as an impurity.
Crystals were pulled up by changing the thickness of the molten layer using phosphorus which is 35.

As a comparison, the crystal pulling was performed under the same conditions by the conventional method and the apparatus using no heat shield.

As a result, in the conventional method and its apparatus, the crystallization rate f _s
= 0.62, and it was confirmed that the solid layer 7 floated on the molten layer 6. On the other hand, in the method and apparatus of the present invention, f _s = 0.75
Up to now, the solid layer 7 could be crystallized without floating and the manufacturing yield was improved.

Further, when the thickness of the solid layer 7 was measured with a quartz pipe, it was found that the molten layer 6 and the solid layer 7 were
The interface with the solid layer 7 is not formed flat as shown in FIG.
The shape of the center part was raised. This is solid layer 7
This is because the radiant heat was radiated from the heater 2 and the end of the solid layer located near the side wall in the crucible was melted. When the melting of the end portion of the solid layer 7 proceeds as it is, the vicinity of the side wall of the crucible and the solid layer are brought into a non-contact state, and the solid layer 7
The solid layer 7 floats when the floating limit of the above is exceeded.

On the other hand, in the method and apparatus of the present invention, the interface between the molten layer 6 and the solid layer 7 is formed flat as shown in FIG. 4, and the outer periphery of the solid layer 7 is fixed along the inner peripheral surface of the crucible 1. It was As a result, it was found that the solid layer 7 of the method and the apparatus thereof according to the present invention realizes more uniform leaching and maintains a stable shape as compared with the conventional case.

This is because the radiant heat from the heater is shielded by the heat shield plate and is not radiated to the solid layer 7, so that the end portion of the solid layer 7 near the side of the crucible is less likely to melt as compared with the conventional method and apparatus. Is shown.

In this embodiment, the material of the heat shield 22 is molybdenum, but the material of the heat shield 22 has good heat resistance, is not deformed by heat, and can be any substance that does not easily generate gas. For example, tungsten or tantalum may be used.

Further, the number and size of the heat shields 22 to be provided are not limited to those in the present embodiment and the experimental examples. For example, in the method and apparatus of the present invention similar to FIG. 1 as shown in FIG. Even when one heat shield 22 is provided concentrically with the heater 2 below, the heat flow rate ratio q _n / q ₀ is greatly reduced as shown in FIG.

〔effect〕

As described in detail above, in the crystal growth method of the present invention, the solid layer in the lower part of the crucible, by using the molten layer method of pulling the silicon single crystal from the molten liquid in the state where the molten layer is present in the upper part, A silicon single crystal with extremely little segregation can be obtained, and it is possible to prevent the bottom of the crucible from being heated by the heat radiated from the heat insulation tube during the pulling process of the silicon single crystal, by blocking with a heat shield. The inconvenience that the solid layer at the bottom is melted and floated is eliminated, and efficient crystal growth can be performed.

Further, in the device of the present invention, it is of course possible to obtain a silicon single crystal with less segregation by utilizing the melt layer method,
Since it is only necessary to provide the heater on the outer periphery of the lower side wall of the crucible, the effect of reducing the equipment cost can be obtained.

[Brief description of drawings]

FIG. 1 is a schematic vertical sectional view showing the crystal growth method and apparatus of the present invention, and FIG. 2 is the radiant heat flow rate q _n of the crystal growth method and apparatus of the present invention and the radiant heat flow rate of the conventional crystal growth method and apparatus. graph showing a heat flow ratio q _n / q ₀ with q _0, FIG. 3 is a schematic longitudinal sectional view showing another embodiment of the present invention a method and apparatus, Figure 4 is a crystal growth method and the present invention FIG. 5 is a schematic vertical sectional view showing an interface between a molten layer and a solid layer in a crucible in the apparatus, FIG. 5 is a schematic vertical sectional view in the conventional crystal growth method and the apparatus, and FIG. 6 is a conventional crystal growth. FIG. 7 is a schematic vertical cross-sectional view of the method and its apparatus, FIG. 7 is a schematic vertical cross-sectional view of a conventional crystal growth method and its apparatus by the melt layer method, and FIG.
Figures 11 are explanatory diagrams showing a one-dimensional model for explaining the principle of reducing segregation of impurities, and FIG. 12 is an explanatory diagram showing a conventional crystal growth method and temperature distribution on the central axis in the apparatus. is there. 1 ... Crucible, 2 ... Heater, 6 ... Melt layer, 7 ... Solid layer, 8 ... Heat-insulating cylinder, 22 ... Heat shield

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Sumio Kobayashi 4-53-3 Kitahama, Chuo-ku, Osaka City, Osaka Prefecture Sumitomo Metal Industries, Ltd. (56) Reference JP-A-61-205691 (JP, A) Kai 55-126597 (JP, A) JP-A-2-233580 (JP, A)

Claims (2)

[Claims] 1. A method for producing a silicon single crystal by using a molten layer method, comprising pulling a silicon single crystal from a crucible filled with a raw material for crystallization arranged in a chamber toward a heat insulating cylinder above the crucible. A crystal growth method, characterized in that heat radiated from the heat-retaining cylinder to the lower part of the crucible is shielded by a heat shield installed on the outer periphery of the lower wall of the crucible in the process of raising. 2. A crystal growth apparatus for producing a silicon single crystal by using a molten layer method, wherein a heater is provided only on the outer periphery of a side wall of a crucible filled with a raw material for crystal, and at least below the heater. A crystal growth apparatus provided with one heat shield.