JPH02233581A - Crystal growth - Google Patents

Abstract

PURPOSE:To pull up crystal and to extremely improve yield without being influenced by flotation of solid layer causing reduction in yield of raw material for crystal by combining melt layer method with Czochralski method. CONSTITUTION:This method consists of a process wherein one crystal 8 is pulled up from a melt layer L in a crucible 2 placed in a chamber 1 in a state of existence of the melt layer L and a solid layer P having approximately the same quality as that of the melt layer positioned under the melt layer L while controlling thickness of the melt layer L and pulling of the crystal is completed before flotation of solid layer P in the melt layer L resulting from reduction in thickness of the solid layer, a process of wholly melting the solid layer P remaining in the crucible 2, a process of taking out the pulled crystal to the outside of the chamber 1 simultaneously or before or after the melting process and a process of pulling up another crystal from a melt layer in a wholly molten state of the solid layer P.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for growing crystals such as silicon single crystals, which are mainly used as semiconductor materials.

[Conventional technology]

In general, as a method for growing this type of single crystal, the Czychralski method (CZ method) has been widely used. FIG. 4c is a schematic diagram showing the manufacturing state of a single crystal by the cz method. For example, a raw material for crystal is put into a crucible 2 placed in a chamber, and after being heated and melted by a heater 3, this This is carried out by immersing a seed crystal 7 suspended by a pulling shaft 6 into the melt NL and pulling it upward while rotating it to grow a single crystal 8 at the lower end of the seed crystal 7.

By the way, when a single crystal is normally used as a semiconductor substrate, etc., in order to adjust the electrical resistivity and electrical conductivity type of the single crystal,
Impurity elements are added to the melt in the crucible, but these impurities segregate in the pulling direction of the single crystal, and it is extremely difficult to maintain a uniform concentration distribution over the entire length of the single crystal in the growth direction. Segregation of impurities is determined by the ratio Cs/CL between the impurity concentration C in the single crystal and the impurity concentration CL in the melt at the growth interface between the melt and the single crystal, that is, the effective segregation coefficient ke.
is not 1, and this is because the impurity concentration in the melt as the single crystal grows and the impurity concentration in the single crystal change during the crystal pulling process.

A fused layer method is known as a method for suppressing such segregation. FIG. 5 is a schematic diagram showing the implementation state of the general molten layer method, in which a molten liquid layer L of the raw material for crystallization is located above the crucible 2, and a molten liquid layer L is located below it. In this method, the single crystal 8 is pulled from the molten liquid layer L in a state in which the solid layer P, which is a material for crystal of substantially the same quality as the solid layer P, coexists with the molten liquid layer L.

In addition, during the process of pulling the single crystal 8, by controlling the heater 3, the decrease in the thickness of the molten liquid layer L due to the pulling of the single crystal 8 is compensated for by melting the solid layer P, and the volume of the molten liquid layer L is kept constant. A method of keeping the impurity concentration in the melt layer L constant by continuously adding impurity elements during the single crystal pulling process (constant melt layer thickness method, Japanese Patent Publication No. 34-8242)
Alternatively, there is a method in which the volume of the melt layer L is intentionally changed and the impurity concentration in the melt layer L is maintained constant without adding impurity elements during single crystal pulling (melt layer thickness (called the change method).

According to the inventor's analysis, the conditions for pulling a single crystal without deflection of impurities using this molten layer method can be roughly determined as follows.

Figure 6 (a) shows the change in layer thickness of the molten liquid layer L and the solid layer P in the crucible 2 during the single crystal pulling process shown in Figure 5.
This will be explained using one-dimensional model diagrams of ~(e).

Figure 6 (a) shows the state immediately after charging the crucible with raw materials, and Figure 6 (opening) shows the state at the end of the initial melting.
Figures (C) and (2) show the state during crystal pulling, and Figure 6 (E) shows the state at the end of crystal pulling.

As shown in FIG. 6(a), the weight of the raw material initially charged in the crucible is 1 (=fF), and the impurity concentration at the position of the weight ratio X measured from the top side of the raw material toward the bottom side is CP (
X).

As shown in Fig. 6 (a), after charging the raw material for crystal into the crucible, it is melted from the upper surface side, and as shown in Fig. 6 (opening), the melting liquid at the melt rate fL (=fo) is deposited on the upper part of the crucible. liquid layer L
and a solid layer P having a solid fraction r are made to coexist below it, and then a single crystal is pulled from 1 iL of the melt. Weight ratio of pulled single crystal to initial charge raw material weight 1 (
When the crystal pulling rate) is f3, the weight ratio of the melt layer (
f, the weight ratio of the raw material remaining as a solid (melt ratio)
Let the impurity concentration in the initial melt layer L be Co, the impurity concentration in the solid phase at the crystal-melt interface after adding the impurity be Cs, and the impurity concentration in the liquid phase be C. do.

Now, as shown in Figure 6 (c), it is a single crystal. Molten liquid layer. Assuming that each solid layer exists at a crystal pulling rate [,, melt rate f,, solid rate f, then crystal pulling rate f3 + melt rate fL+
The following equation (1) holds true between the solid fraction f.

(,-+lL+f,-f.+f,=1...(1)
However, r, +t, -r0 The melt rate is Assume that the field rate changes from fL to fL+Δ′, and from f, to f,+Δf, and the amount of impurity added during this period is c1・Δfs (where c, is the unit amount of impurity added to the crystallization rate). Then, Figure 6 (c)
In the process from the state to the state shown in Figure 6 (2), the impurity concentration C, in the molten liquid layer L is uniform, and the diffusion of impurity into the solid layer P is considered to be zero. Figure 6 (c)
, (2), the impurities in the A region are preserved, and the following (2) 2 holds between them.

c, - rL+c, ・Δf3+CP ・Δro=C
,・Δf,+ (C,+ΔCL) ・(rt+Δr
t) ...(2) Since the following relationship (3) 2 holds between the impurity concentration C in the solid phase and the impurity concentration CL in the liquid phase at the single crystal-melt interface, Cs ''ke' CL = (33 where ke: segregation coefficient Omitting the second-order minute term in equation (2) and applying equation (3) yields equation (4) below.

Therefore, in the fused layer method, the non-segregation condition is dC in equation (4).
Assuming t/d fs = 0, Cp ”0, the following (5)
It is given by Eq.

When the solid layer P is melted during the crystal pulling process and the thickness of the molten liquid layer L is kept constant, d fL/d
Since rs = O, the second condition of (5) above can be satisfied by continuously adding an impurity represented by the following formula (6).

C@ = keCL = keCo ... (6) On the other hand, in the melt thickness variation method, impurity is not continuously added (C- = 0)
, d fL/d fs =-ke may be satisfied by changing the melt layer thickness as the crystal is pulled so as to satisfy the following equation (7).

f L va f L. -ke-f 3...(
7) However, fLO: Initial melt rate In many cases, the effective segregation coefficient ke of impurities is smaller than 1, so by setting fLo-ke, segregation-free conditions can be maintained until the end.

Note that after the solid layer is completely melted, the no-segregation condition is not established, and segregation progresses according to the following equation (8).

Cs −keCo (1 f s ) ”−’
"'(8) However, C0: Initial concentration of impurities in the melt layer. FIG.
Figure (a) shows the thickness dimensions of the crucible and the molten liquid layer and solid layer in the crucible, and Figure 7 (opening) shows the temperature distribution, which can be qualitatively explained as follows.

The amount of heat supplied from the heater to the molten liquid layer includes the amount of heat Q, J such as conductive heat through the single crystal, radiant heat from the surface of the melt, etc., and the conductive heat dissipated via the raw material in the crucible 2 and the crucible axis 2c. It becomes the sum with NQL. The amount of heat Qu is set to be approximately constant throughout the pulling period in order to stabilize the crystal pulling conditions. Also, assuming that the electric power (heat amount) of the heater is approximately constant and the thermal IQL is also constant, the following (
9) 2 is established.

= h (Tb − Tp) However, T,: Boundary temperature between the molten liquid layer and the solid layer (a constant value determined by the melting point of the crystal raw material) T0: Temperature at the lower part of the crucible axis (usually a substantially constant value) λ,: Solid Thermal conductivity of the layer Sc: Internal cross-sectional area of the crucible h: Heat transfer rate from the solid layer to the crucible axis λP; Thermal conductivity of the crucible axis 1P: Length of the crucible axis inside the chamber Equation (9) can be calculated from T, When T, is eliminated, the following α0 formula is obtained.

Q, λh Sc h λ
P SF In the normal crystal pulling process, the position of the melt surface is kept constant, that is, e=constant in FIG. 7(a), so ΔJt+Δβ, +Δ1F=0.

In addition, the following αυ formula holds true, and while Δl, 8f, Δf, +Δfc+Δf, = 0, Δ(,ocΔN
, holds true.

Therefore, we obtain the following (goods) 2 from the αΦ formula, ΔfL
Δf, +Δf, Δf, Δf! 'pS. From now on, the heat conductivity of the solid layer is λ6 5C, and the crucible axis 2
If the heat conductivity of c is equal to λP and SP, the melt layer thickness can be maintained constant, and if λ, S, > λ6 sc, the melt layer thickness will decrease as it is pulled up, and the melt It becomes possible to control the liquid layer thickness.

[Problem to be solved by the invention]

Theoretically, the fused layer method should be able to make the impurity concentration distribution in a single crystal uniform as long as the above-mentioned conditions are followed. However, with this molten layer method, it is difficult to single-crystallize the entire raw material charged into the crucible 2. In particular, silicon raw materials used as semiconductor materials have a solid density lower than that of the molten liquid, so the thickness of the solid layer When the temperature falls below a certain value, the solid layer floats in the molten layer, causing vibrations and temperature fluctuations in the molten layer, which impede the pulling of the single crystal. For this reason, the pulling of the single crystal had to be stopped with a considerable amount of raw material (solid layer, melt ratio rs) left in the crucible, resulting in a problem of poor raw material yield.

The present invention was made in view of the above circumstances, and its purpose is to combine the fused layer method and the CZ method to influence the floating of the solid layer, which is a factor in reducing the yield of crystal raw materials. It is an object of the present invention to provide a crystal growth method that enables the crystal to be pulled up without being damaged, thereby significantly improving the yield.

[Means to solve the problem]

In the crystal growth method according to the present invention, the thickness of the molten layer is determined by coexisting a molten liquid layer and a solid layer of substantially the same quality as the molten liquid layer located below the molten liquid layer in a crucible placed in a chamber. pulling a crystal from the melt layer in a controlled manner and finishing the pulling of the crystal before the solid layer becomes suspended in the melt layer due to a decrease in the layer thickness; A process of melting all remaining solid layers, a process of taking out one crystal pulled out of the chamber at the same time or before or after this process, and pulling other crystals from the molten liquid layer in a state where all the solid layers have been melted. It is characterized by including the process of raising.

[Effect]

In the present invention, as a result of this, it is possible to pull crystals with no segregation of impurities in the molten layer process, and to pull crystals to a state where impurity segregation exists but with almost no residual raw material in the CZ process. It becomes possible to pull it up.

〔Example〕

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be specifically described below based on drawings showing embodiments thereof. FIG. 1 is a schematic vertical cross-sectional view of a crystal growth apparatus showing the implementation state of the method of the present invention, in which 1 is a chamber, 2 is a crucible, 3 is a heater, 4 is a shutter, 5 is a pull chamber,
1l indicates a solid raw material feeder.

A crucible 2 is disposed in the center of the chamber 1, and the crucible 2
A heater 3 is installed between the chamber 1 and the heat insulating material, and a cylinder 1 is mounted on the upper wall of the chamber 1, facing the crucible 2 and opened and closed by a shutter 4.
a stands upright, and a pull chamber 5 is removably erected on top of this cylindrical body 1a.

The crucible 2 has a double structure with an inner crucible 2a made of quartz and an outer crucible 2b made of graphite arranged around the outer periphery of the crucible 2, and the upper end of a shaft 2c penetrating the bottom wall of the chamber 1 is connected to the center of the bottom of the crucible 2. It is designed so that it can be raised and lowered while being rotated by the shaft 2c.

Above the pull chamber 5 is a rotating and lifting mechanism (not shown).
The upper end of the pulling shaft 6 connected to the is introduced, and a seed crystal 7 held by a chuck is suspended from the lower end of the pulling shaft 6. After the seed crystal 7 is blended into the molten liquid layer L in the crucible 2, By rotating and raising the silicon single crystal 8 at the lower end of the seed crystal 7. It is designed to encourage growth.

The solid raw material feeder 11 is equipped with a weigher, a hopper (none of which is shown), and a raw material input pipe 11a, and the lower end of the raw material input pipe 11a passes through the chamber 1 and faces upward toward the periphery of the crucible 2. The raw material that has fallen from the hopper is fed into the upper end of the raw material input pipe 11a via a weighing device, and is supplied into the crucible 2 from here.

Although not shown in the drawings, a supply device for impurities is also provided in the same manner.

The main steps in the method of the present invention are shown in Figures 2(a) to 2(a).
The explanation will be based on the one-dimensional model shown in ).

FIG. 2 is an explanatory diagram showing a one-dimensional model showing the changes in the proportions of the molten liquid layer, fixed layer, and single crystal in the crucible in the main process of the method of the present invention.
The process of pulling the second single crystal using the CZ method and the process of pulling the second single crystal using the CZ method will be explained separately.

[Process of pulling the first single crystal using the molten layer method] ■ First, a solid raw material for crystals, such as high-purity polycrystalline silicon (impurity concentration C
, = O) (Fig. 2 (a)).

(2) Next, after replacing the atmospheric gas in the chamber 1 with an inert gas or the like, the electric power of the heater 3 is set to the melting condition, and the charged solid raw material is melted.

The charged solid raw material has a predetermined initial melt rate r. An impurity is added to the melt from an impurity supply device (not shown) so that the impurity concentration in the melt becomes C0. This creates a melt layer L with an initial melt rate fLo and an impurity concentration C0, and a solid rate fP+ located below it.
A state is set in which a solid layer P with an impurity concentration C coexists.

■ The crystal starts to be pulled up, and the accompanying decrease in the thickness of the molten liquid layer L is replenished by melting the solid NP. Then, the layer thickness of the solid layer P decreases, and the solid layer P becomes the molten liquid layer L.
Finish pulling the single crystal before it begins to float inside. At this time, the crystal pulling rate of the single crystal is fS + the impurity concentration in the crystal is C. Then, the melt rate of the melt layer L is fL
, the impurity concentration is CG, the solid rate of the solid layer P is fP+the impurity concentration is C.

(2) After the pulled single crystal 8 is drawn into the pull chamber 5 through the shutter 4, the shutter 4 is closed.

This completes the process of pulling the first single crystal using the fused layer method.

(2) The remaining molten liquid layer L1 solid layer P in the crucible 2 is heated by heater control until the solid layer P is completely melted, and the entire solid layer P is made into a molten liquid layer L. The melt rate of this melt layer L becomes f, and the impurity concentration becomes C0' due to the melting of the solid layer P.

Step of pulling the second single crystal using the CCZ method] ■ When the impurity concentration distribution in the single crystal predicted from the impurity concentration C' in the melt layer matches the impurity concentration distribution to be set for the second single crystal. The seed crystal 7 may be immersed as it is in the molten liquid layer in the crucible 2 to start pulling the single crystal, but if you want to obtain a second single crystal with a different impurity concentration distribution, An impurity element is introduced into the liquid NL, and the impurity concentration of the molten liquid layer is set to 00.

■ After setting the impurity concentration in the melt layer to 00,
The second single crystal is pulled by the CZ method shown in the figure until the melt runs out.

In the Cz method, there is no solid layer, so fy"'O. Also, Δ
fL+Δf, -o. c. =o, so equation (4) can be expressed as the following α1 equation.

d l 3 Therefore, the impurity concentration C3 in the single crystal is segregated in a state given by the following aa formula.

C3-keCo(1-fs)"-'=Q4
) This makes the impurity concentration 03. Thus, a first single crystal whose impurity concentration is substantially constant in the crystal growth direction and a second single crystal whose impurity concentration is segregated with C expressed by the above-mentioned α-temperature formula are obtained.

[Numerical example]

A quartz crucible with an inner diameter of 150 mm was used as a crucible, polycrystalline silicon was used as a raw material, phosphorus with a segregation coefficient ke = 0.35 for silicon was added as an impurity, and a first silicon single crystal with a diameter of 5011 llm was pulled without segregation. I went there once.

The rotation speed of the crucible is 0.5 rpm. The rotation speed of the pulling shaft was 15 rpm (opposite to the crucible).

Next, after melting the remaining solid layer P, the impurity concentration in the melt layer is set to be the same as the impurity concentration in the melt layer during the first single crystal pulling process, and the second single crystal is pulled by the CZ method, Crystal pulling rate fs of pulled silicon single crystal
The distribution of resistivity at each site of 0.1 to 0.9 was measured. The results are shown in Figure 3. In Figure 3, the horizontal axis represents f3, and the vertical axis represents resistivity ρ/ρ. [-
] is shown. In the graph, the results plotted with O marks are the results for the first single crystal, and the results plotted with ● marks are the results for the second single crystal.

As is clear from this graph, it can be seen that the first single crystal can be pulled with almost no segregation, in other words, with no change in resistivity.

Note that the second single crystal also has a predetermined impurity concentration distribution.
In other words, it can be seen that a single crystal with resistivity having a gradient is obtained.

In addition, in the above-mentioned embodiment, each of the molten layer method and CZ method was used.
We have explained the case of pulling a single crystal in this book, but in the fused layer method, each time a single crystal is pulled, a solid raw material is supplied, two or more single crystals without segregation are pulled, and finally the CZ method is applied. Of course, a single crystal having a predetermined impurity concentration distribution may also be pulled.

(Effects) As described above, in the method of the present invention, by combining the fused layer method and the CZ method, it is possible to pull crystals with a high yield, improving work efficiency, and reducing equipment costs and effective use of heat sources. The present invention exhibits excellent effects, such as a significant improvement in manufacturing yield.

[Brief explanation of the drawing]

FIG. 1 is a schematic vertical cross-sectional view of a crystal growth apparatus showing the implementation state of the method of the present invention, FIG. 2 is an explanatory diagram showing the transition of phase change of the charge material in the crucible, and FIG. Graph showing the relationship between f and resistivity for a single crystal, 4th,
Figure 5 shows the Czochralski method. Explanatory diagrams of the fused layer method, No. 6
The figure is an explanatory diagram showing the transition of the phase change of the charge material in a general fused bed method, and FIG. 7 is an explanatory diagram showing an example of crucible temperature control in the fused bed method. 1... Chamber 2... Crucible 3... Shimaichita 4.
...Shutter 5...Pull chamber 6...Pulling shaft 7...Seed crystal 8...Single crystal 11...Solid raw material supply device patent Applicant: Sumitomo Metal Industries Co., Ltd. Agent
Patent attorney Noboru Kono! 'ts paper calculation! En-men dedicated Hishima I {i) (mouth) No.

Claims (1)

[Claims] 1. In a crucible placed in a chamber, a molten liquid layer and a solid layer of substantially the same quality as the molten liquid layer located below the molten liquid layer coexist, and the thickness of the molten liquid layer is determined. pulling a crystal from the melt layer in a controlled manner and finishing the pulling of the crystal before the solid layer becomes suspended in the melt layer due to a decrease in the layer thickness; A process of melting all the remaining solid layer, a process of taking one crystal pulled out of the chamber at the same time or before or after this process, and removing other crystals from the molten liquid layer in a state where all the solid layer has been melted. A crystal growth method characterized by comprising a pulling process. 2. The crystal growth method according to claim 1, which includes the step of adding impurities to the molten liquid layer in a state where the solid layer is completely melted.