JP2009274888A - Production method of silicon single crystal, and silicon single crystal wafer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a production method of a silicon single crystal, by which a silicon single crystal having uniform carbon concentration and BMD (bulk micro-defects) uniformly formed therein can be produced, and to provide a silicon single crystal wafer produced from a silicon single crystal produced by the method, the wafer having BMD uniformly formed within the wafer plane and having a uniform gettering ability. <P>SOLUTION: The production method of a silicon single crystal 1 by Czochralski method by doping with carbon is disclosed, wherein the crystal 1 is grown by controlling the difference between an average growing rate of the crystal center portion 15 within a plane perpendicular to the crystal growth direction and an average growing rate of the crystal peripheral portion 16 to be within ±0.1 mm/min during growing the crystal. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a method for producing a silicon single crystal by doping carbon by a Czochralski method (hereinafter abbreviated as CZ method), and a silicon single crystal wafer produced from the silicon single crystal.

A silicon single crystal from which a silicon wafer used as a substrate of a semiconductor device is cut is mainly manufactured by a CZ method.
Since semiconductor devices are formed on the surface layer, it is important that there are no defects on the surface layer. As a method of manufacturing a wafer having no such defect, there is a method of pulling up a defect-free crystal and cutting out the wafer from the defect-free crystal while controlling the growth conditions so as not to form a crystal defect in the crystal growth stage. .
Such a method is characterized in that it is difficult to control during crystal growth, and the production cost of the crystal is high.

On the other hand, after growing crystals under conditions that cause crystal defects during single crystal growth, wafers with annealed crystal defects near the surface layer or epitaxial layers without crystal defects are manufactured. There is a way.
A wafer manufactured by such a method is generally manufactured from a crystal having a high growth rate that does not include an I region, as shown in Patent Document 1, for example. Since there is an additional step for eliminating defects, the total cost is higher than that of a wafer cut out from a defect-free crystal.

In the method of detoxifying crystal defects in the vicinity of the surface layer in the subsequent process using a single crystal with crystal defects, the added value is as much as the total cost becomes higher than the wafer cut out from the defect-free crystal as described above. Is required.
As an added value, a BMD (Bulk Micro Defect) is formed inside the wafer, and an IG (Intrinsic Gettering) capability for capturing contaminant atoms such as heavy metals inside the wafer and improving device characteristics is added. .

  In the case of an annealed wafer, it is possible to form a BMD by devising the annealing process conditions. However, in general, since epitaxial layers are formed at high temperatures in epitaxial wafers, BMD nuclei formed during crystal growth disappear, and BMDs are difficult to form.

In order to improve these, carbon-doped epitaxial wafers shown in Patent Document 2 and Patent Document 3 have been proposed for a long time.
Even in a crystal grown by carbon doping, the growth rate is shown in Patent Document 4 or the FPD defect is shown in Patent Document 5 and is pulled up at a relatively high speed in the so-called V region. Crystals are used.

  In these wafers, since BMD is formed due to carbon, if there is non-uniform carbon concentration in the wafer surface, there is a problem that non-uniform BMD occurs in the wafer surface. In addition, if there is non-uniform BMD in-plane, gettering ability becomes non-uniform in-plane, and there is a problem that the device pattern cannot be stably written because the wafer warps during the device process. It was.

JP 2000-219598 A JP 59-82717 A JP 59-94809 JP 2001-102384 A JP 2002-293691 A

  The present invention has been made in view of the problems as described above, and a silicon single crystal manufacturing method capable of manufacturing a silicon single crystal in which the carbon concentration is made uniform and the BMD is formed uniformly, and the manufacturing method. An object of the present invention is to provide a recon single crystal wafer having a uniform gettering capability in which BMDs are uniformly formed in the surface of a wafer manufactured from the silicon single crystal.

In order to achieve the above object, according to the present invention, in the method for producing a silicon single crystal by doping carbon by the Czochralski method, a crystal center portion in a plane perpendicular to the crystal growth direction is obtained during crystal growth. A method for producing a silicon single crystal is provided, wherein a crystal is grown such that a difference between an average speed at which the crystal grows and an average speed at which a crystal peripheral portion grows is within ± 0.1 mm / min. 1).
Thus, during crystal growth, the difference between the average speed at which the crystal center in the plane perpendicular to the crystal growth direction grows and the average speed at which the crystal periphery grows should be within ± 0.1 mm / min. If the crystal is grown, a silicon single crystal in which the carbon concentration is made uniform in the plane and the BMD is formed uniformly can be manufactured.

At this time, the difference between the average growth rate of the crystal and the average growth rate before a predetermined time can be within ± 0.1 mm / min.
Thus, by setting the difference between the average growth rate of the crystal and the average growth rate before a predetermined time to be within ± 0.1 mm / min, the crystal center portion in the plane perpendicular to the crystal growth direction during crystal growth The difference between the average speed at which the crystal grows and the average speed at which the crystal periphery grows can be controlled within ± 0.1 mm / min.

At this time, the average growth rate is preferably an average growth rate of 40 minutes.
Thus, by setting the average growth rate to an average growth rate of 40 minutes, it becomes easy to control the crystal defect region of the growing silicon single crystal to a desired crystal defect region.

At this time, it is preferable that the predetermined time is 5 minutes or more and 80 minutes or less.
As described above, by setting the predetermined time to 5 minutes or more and 80 minutes or less, the average rate of growth of the crystal central part in the plane perpendicular to the crystal growth direction in the crystal interface height region and the crystal peripheral part grow. It is possible to control the difference from the average speed to be within ± 0.1 mm / min more reliably.

At this time, when the crystal rotation speed during the crystal growth is R [rpm] and the diameter of the crystal is Dia [mm], the crystal can be grown while satisfying R × Dia ≧ 1200. ).
As described above, when the crystal rotation speed during the crystal growth is R [rpm] and the diameter of the crystal is Dia [mm], the crystal grows by satisfying R × Dia ≧ 1200, so that the crystal growth interface It is possible to suppress the boundary diffusion layer thickness from becoming non-uniform in the surface, and to more reliably exhibit the effect of making the crystal carbon concentration uniform.

At this time, the silicon single crystal may be grown by doping carbon so that the carbon concentration in the silicon single crystal is 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc (ASTM 75). Preferred (claim 6).
Thus, the silicon single crystal is grown by doping carbon so that the carbon concentration in the silicon single crystal is 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc (ASTM 75). Thus, BMD can be more easily formed on the crystal without inhibiting the single crystallization of the crystal, and thus suppressing the manufacturing cost.

At this time, it is preferable to grow the silicon single crystal so that the oxygen concentration in the silicon single crystal is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (ASTM 79).
Thus, by growing the silicon single crystal so that the oxygen concentration in the silicon single crystal is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (ASTM 79), the effect of doping carbon is matched. Thus, an appropriate amount of BMD can be more reliably formed in the crystal.

Moreover, according to this invention, the silicon single crystal wafer manufactured from the silicon single crystal manufactured using the silicon single crystal manufacturing method in any one of the said invention is provided.
In such a silicon single crystal wafer manufactured from a silicon single crystal manufactured using any one of the above-described silicon single crystal manufacturing methods of the present invention, the BMD is uniformly formed in the wafer surface, and the uniform It is a good silicon wafer having gettering capability.

In addition, according to the present invention, the difference between the carbon concentration at the center of the crystal and the carbon concentration at the periphery of the crystal manufactured from the silicon single crystal manufactured using any one of the above-described silicon single crystal manufacturing methods of the present invention. A silicon single crystal wafer in which is within 20% is provided (claim 9).
The difference between the carbon concentration at the center of the crystal and the carbon concentration at the periphery of the crystal manufactured from the silicon single crystal manufactured using any one of the above-described silicon single crystal manufacturing methods of the present invention is within 20%. The silicon single crystal wafer is a good silicon wafer having BMD uniformly formed in the wafer surface and having more uniform gettering ability.

Further, according to the present invention, the produced silicon single crystal produced by using any of a silicon single crystal production method of the present invention, the carbon concentration of ^1.2 × 10 ¹⁶ ~20.0 × 10 16 A silicon single crystal wafer that is atoms / cc (ASTM 75) is provided.
A carbon concentration of 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc manufactured from the silicon single crystal manufactured using any one of the above-described silicon single crystal manufacturing methods of the present invention is used. A silicon single crystal wafer that is ASTM 75) is a good silicon wafer having BMD uniformly formed in the wafer surface and uniform gettering ability in the surface.

Further, according to the present invention, the oxygen concentration is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (manufactured from the silicon single crystal manufactured using any one of the silicon single crystal manufacturing methods of the present invention). A silicon single crystal wafer that is ASTM 79) is provided.
The oxygen concentration manufactured from the silicon single crystal manufactured by using any one of the above-described silicon single crystal manufacturing methods of the present invention is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (ASTM 79). If it is a silicon single crystal wafer, it is a good silicon wafer in which BMD is more reliably formed in the wafer surface and has uniform gettering ability.

  In the present invention, in the method for producing a silicon single crystal by doping carbon by the Czochralski method, the average speed at which the crystal central portion in the plane perpendicular to the crystal growth direction grows and the crystal peripheral portion during crystal growth. Since the crystal is grown such that the difference from the average growth rate is within ± 0.1 mm / min, it is possible to produce a silicon single crystal in which the carbon concentration is made uniform and the BMD is formed uniformly. A silicon single crystal wafer manufactured from the silicon single crystal is a good silicon single crystal wafer having a uniform gettering capability in which BMDs are uniformly formed in the wafer surface.

  Hereinafter, although an embodiment is described about the present invention, the present invention is not limited to this.

In a general epitaxial wafer, since the epitaxial layer is formed at a high temperature, BMD nuclei formed at the time of crystal growth disappear, which makes it difficult to form a BMD.
Therefore, in recent years, in order to form a BMD and provide sufficient gettering capability, carbon is doped to manufacture a silicon single crystal.

  Thus, in a wafer cut out from a single crystal manufactured by doping carbon, BMD is formed due to carbon, so that the gettering ability is relatively high compared to the case where carbon is not doped. However, if the carbon concentration is non-uniform in the wafer surface, there is a problem that non-uniform BMD occurs in the wafer surface. If the in-plane BMD is non-uniform, uniform gettering ability cannot be obtained within the wafer surface, and the device pattern cannot be stably written because the wafer warps during the device process. There was a problem.

Therefore, the present inventor has intensively studied to solve such problems.
By the way, the phenomenon in which carbon is taken into the silicon crystal is known as a segregation phenomenon, and it is known that a concentration obtained by multiplying the melt concentration by the segregation coefficient is taken into the crystal.
In general, this segregation coefficient is often considered as a constant value, but actually acts as an effective segregation coefficient that is affected by various conditions as shown in the following equation.

The segregation phenomenon is expressed in the following formula.
[C] _S = keff × [C] ₀ × (1-g) ^Keff−1
[C] _S : Impurity concentration in crystal, [C] ₀ : Impurity concentration in raw material melt g: Solidification rate, keff: Effective segregation coefficient

The effective segregation coefficient keff is expressed as follows.
keff = k ₀ / [k ₀ + (1−k ₀ ) exp (−vδ / D)]
k ₀ : equilibrium segregation coefficient (0.07 in the case of carbon), v: growth rate δ: boundary diffusion layer thickness, D: impurity diffusion coefficient in raw material melt

Here, the thickness of the boundary diffusion layer is a value influenced by the rotation of the crystal, the viscosity of the raw material melt, the impurity diffusion coefficient, and the like.
As can be seen from the above formula, the effective segregation coefficient is affected by various conditions and should change.

The present inventor has found that this effective segregation coefficient is influenced by various conditions, but when producing a normal product crystal, it is mainly influenced by the growth rate of a single crystal. It has been found that a stable pulling is important for bringing the crystal into a uniform carbon dope and forming a BMD.
Then, during crystal growth, the difference between the average speed at which the crystal central part in the plane perpendicular to the crystal growth direction grows and the average speed at which the crystal peripheral part grows is within ± 0.1 mm / min. As a result of the growth, it was conceived that a silicon single crystal in which BMD was formed uniformly by in-plane carbon doping to form a BMD in-plane could be manufactured, and the present invention was completed.

FIG. 1 shows an example of a single crystal pulling apparatus 21 that can be used in carrying out the method for producing a silicon single crystal of the present invention. In the main chamber-2 of the single crystal pulling apparatus 21, a quartz crucible 6 for containing the melted raw material melt 5 and a graphite crucible 7 for supporting the quartz crucible 6 are provided. And the heater 4 is arrange | positioned so that these crucibles 6 and 7 may be surrounded. In order to prevent heat from the heater 4 from being directly radiated to the main chamber-2, a heat insulating member 8 is provided on the outer periphery of the heater 4 so as to surround the periphery thereof.
The crucibles 6 and 7 can be moved up and down in the direction of the crystal growth axis, and the crucibles 6 and 7 can be raised so as to compensate for the lowering of the liquid level of the raw material decreased during the single crystal growth. .

Further, a gas rectifying cylinder 9 is provided in the main chamber-2. Further, a heat insulating material 10 is provided at the outer lower end of the gas flow straightening tube 9 so as to face the raw material melt, so that radiation from the melt surface is cut and the melt surface is kept warm.
Then, the quartz crucible 6 is filled with polycrystalline silicon, which is a silicon single crystal raw material, and a dopant for doping carbon.
Here, high purity carbon powder or carbon lump (granular carbon) can be used as the dopant.

  In this way, after filling the raw material 5 and the dopant into the quartz crucible 6, an inert gas flows in from the gas inlet 11 installed in the pulling chamber 3 while exhausting from the gas outlet 12, and the inside is inert. Replace with gas atmosphere. Next, the crucibles 6 and 7 are heated by the heater 4 disposed so as to surround the graphite crucible 7, and the raw material is melted to obtain the raw material melt 5.

  After melting the raw material filled in this manner, the seed crystal 13 is immersed in the raw material melt 5, the silicon single crystal 1 is grown while rotating the seed crystal 13, and the pulling wire 14 is wound up by a pulling mechanism (not shown). Thus, the silicon single crystal 1 is pulled up and grown below the seed crystal 13.

  In the growth interface of the silicon single crystal 1 pulled by the silicon single crystal manufacturing method of the present invention, the crystal center portion 15 is raised from the crystal periphery 16 as shown in FIG. This is because the crystal periphery 16 is easily cooled and the crystal center portion 15 is difficult to cool, and the peripheral portion 16 tends to have a higher growth rate. The height h from the lower end of the peripheral portion 16 of the swelled portion varies depending on various factors such as crystal diameter, crystal growth rate, presence / absence of magnetic field application, melt convection, etc. There is a height of about 10-40 mm under relatively high growth conditions for growth.

Therefore, the peripheral portion 16 of the crystal first becomes a crystal, and the central portion 15 of the crystal grows later, and the growth rate is different between the central portion 15 and the peripheral portion 16 of the crystal.
On the other hand, when growing a single crystal, the growth rate is changed from moment to moment in order to control the diameter of the crystal to a desired thickness. This momentary change is taken into the crystal in the form of stripes as shown in FIG.

  As shown in FIG. 2C, the wafer 31 is cut out from this crystal. However, since the momentary change is averaged in the thickness direction of the wafer 31, it is not a big problem. When considered, the average growth rate differs between the crystal peripheral portion 16 that grows first and the crystal center portion 15 that grows later. Therefore, the effective segregation coefficient of carbon greatly influenced by the growth rate is different between the wafer central portion 15 and the peripheral portion 16, and the carbon concentration to be incorporated becomes non-uniform in the plane.

At this time, in the silicon single crystal manufacturing method of the present invention, during crystal growth, an average speed of 40 minutes in which the crystal central portion 15 in the plane perpendicular to the crystal growth direction grows and 40 minutes in which the crystal peripheral portion 16 grows. Crystals are grown so that the difference from the average speed is within ± 0.1 mm / min.
Thus, during crystal growth, the difference between the average speed of 40 minutes in which the crystal central portion 15 in the plane perpendicular to the crystal growth direction grows and the average speed of 40 minutes in which the crystal peripheral portion 16 grows is ± 0. By growing the crystal so as to be within 1 mm / min, the crystal defect region can be controlled to be a desired region, and the effective segregation coefficient described above changes in-plane during single crystal growth. Can be suppressed, and the carbon concentration in the crystal central portion 15 and the peripheral portion 16 can be made uniform.
And the silicon single crystal 1 in which uniform BMD was formed can be manufactured by equalizing the carbon concentration in the central portion 15 and the peripheral portion 16 of the crystal.

At this time, the difference between the average growth rate for 40 minutes of the entire crystal growth interface and the average growth rate for 40 minutes before the predetermined time can be within ± 0.1 mm / min.
As described above, the center part 15 of the growth interface of the silicon single crystal 1 to be pulled up is raised from the crystal peripheral part 16. That is, the central portion 15 of the crystal grows after a predetermined time from the peripheral portion 16 of the crystal.

  Therefore, by pulling up the crystal so that the difference between the average growth rate for 40 minutes of the entire crystal growth interface and the average growth rate for 40 minutes before the predetermined time is within ± 0.1 mm / min, The difference in the average growth rate for 40 minutes between each of 15 and the crystal peripheral part 16 can be surely set to ± 0.1 mm / min, and the difference in carbon concentration between the crystal central part 15 and the crystal peripheral part 16 can be reduced. it can.

Here, the predetermined time is preferably 5 minutes or more and 80 minutes or less.
V-rich crystals used for epitaxial wafers and the like are pulled at an average growth rate of 0.5-2 mm / min. Therefore, considering that the height h of the crystal growth interface is 10-40 mm, the crystal center The time difference between the portion 15 and the crystal peripheral portion 16 is a minimum of 5 min (10 [mm] (height h at the crystal growth interface) / 2 [mm / min] (average growth rate)), and a maximum of 80 min (40 [mm] ( The height h) /0.5 [mm / min] (average growth rate)) at the crystal growth interface.

  That is, when the predetermined time is 5 minutes or more and 80 minutes or less, the difference between the average growth rate for 40 minutes of the entire crystal growth interface and the average growth rate for 40 minutes before the predetermined time is within ± 0.1 mm / min. By pulling up the crystal in a certain manner, the difference between the average speed at which the crystal center in the plane perpendicular to the crystal growth direction in the crystal interface height region grows and the average speed at which the crystal periphery grows can be more reliably It can be controlled within ± 0.1 mm / min.

At this time, when the crystal rotation speed during the crystal growth is R [rpm] and the diameter of the crystal is Dia [mm], the crystal can be grown while satisfying R × Dia ≧ 1200.
The segregation phenomenon by the CZ method is called a boundary diffusion layer in which impurities that have not been incorporated into the crystal are collected just below the crystal growth interface, and the concentration is higher than the average impurity concentration in the raw material melt. There is a part. The effective segregation coefficient also changes as the thickness of the boundary diffusion layer changes.

  On the other hand, convection is generated when winding is caused by the rotation of the crystal. When this convection is reduced, the thickness of the boundary diffusion layer immediately below the crystal central portion 15 is increased. The thickness of the boundary diffusion layer in the crystal peripheral portion 16 is a certain thickness due to the convection generated from the lower side of the crucible 6, 7 toward the upper side and further to the crystal side by the rotation of the crucibles 6, 7. Therefore, if the crystal rotation is extremely slow, the boundary diffusion layer thickness at the center of the crystal becomes extremely thick compared to the boundary diffusion layer thickness at the periphery of the crystal, resulting in a large nonuniform boundary diffusion layer thickness. . As a result, a difference in carbon concentration occurs between the central part and the peripheral part. Of course, the convection generated by the rotation of the crystal increases as the crystal rotation speed increases, and also increases as the crystal diameter increases. Therefore, in order to keep this convection above a certain level, it is important that R × Dia ≧ 1200 is satisfied when the crystal rotation number R [rpm] during crystal growth and the crystal diameter is Dia [mm]. is there.

  Thus, when the crystal rotation speed during crystal growth is R [rpm] and the diameter of the crystal is Dia [mm], the convection is reduced by growing the crystal satisfying R × Dia ≧ 1200. And the boundary diffusion layer thickness can be prevented from becoming non-uniform. Thereby, the effect of making the carbon concentration of the crystal uniform can be more reliably exhibited.

At this time, the silicon single crystal 1 is grown by doping carbon so that the carbon concentration in the silicon single crystal 1 is 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc (ASTM 75). preferable.
In this way, by growing the silicon single crystal 1 by doping carbon so that the carbon concentration in the silicon single crystal 1 is higher than 1.2 × 10 ¹⁶ atoms / cc (ASTM 75), more BMD is added to the crystal. It can be easily formed.
The carbon solid solubility limit in a silicon single crystal is said to be 33 × 10 ¹⁶ atoms / cc (ASTM 75). Therefore, by growing carbon single crystal 1 by doping carbon so that the carbon concentration in silicon single crystal 1 is less than 20.0 × 10 ¹⁶ atoms / cc (ASTM 75), the crystal does not become a single crystal. Can be reliably prevented, and the manufacturing cost can be suppressed by reducing the cost of the dopant used.

At this time, it is preferable that the silicon single crystal 1 is grown so that the oxygen concentration in the silicon single crystal 1 is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (ASTM 79).
As described above, by growing the silicon single crystal 1 so that the oxygen concentration in the silicon single crystal 1 is higher than 8 × 10 ¹⁷ atoms / cc (ASTM 79), it becomes easier to form a desired amount of BMD in the crystal. be able to.
In addition, by growing the silicon single crystal 1 so that the oxygen concentration in the silicon single crystal 1 is smaller than 18 × 10 ¹⁷ atoms / cc (ASTM 79), the effect of doping with carbon is combined with excessive precipitation. Can be suppressed.

The silicon single crystal 1 manufactured by the silicon single crystal manufacturing method of the present invention is sliced by a cutting device such as a wire saw in accordance with a normal method, and then subjected to steps such as chamfering, lapping, etching, polishing, etc. To 31.
Then, by annealing the silicon single crystal wafer 31, a defect-free crystal can be formed on the surface and a BMD layer can be formed in the bulk portion. An epitaxial wafer can also be obtained by forming an epitaxial layer on the surface.

Of course, these steps are merely exemplified and enumerated, and there may be other steps such as washing and heat treatment, and the order of steps and the omission of some steps may be appropriately performed according to the purpose.
The silicon single crystal wafer 31 manufactured from the silicon single crystal manufactured using the silicon single crystal manufacturing method of the present invention has a uniform silicon gettering capability in which BMD is uniformly formed in the wafer surface. It is.

In particular, the difference between the carbon concentration at the wafer center and the carbon concentration at the periphery is preferably within 20%.
Further, the carbon concentration of the wafer is 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc (ASTM 75), and / or the oxygen concentration is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (ASTM 79). ) Is preferable.
Such a silicon single crystal wafer 31 is a good silicon wafer in which a BMD is formed more uniformly in the wafer surface and has a more uniform gettering capability.

  As described above, in the present invention, in the method of manufacturing a silicon single crystal by doping carbon by the Czochralski method, the average of the crystal center in the plane perpendicular to the crystal growth direction grows during crystal growth. Since the crystal is grown so that the difference between the speed and the average speed at which the periphery of the crystal grows is within ± 0.1 mm / min, the silicon single crystal in which the carbon concentration is made uniform and the BMD is formed uniformly is obtained. Can be manufactured. A silicon single crystal wafer manufactured from the silicon single crystal is a good silicon single crystal wafer having a uniform gettering capability in which BMDs are uniformly formed in the wafer surface. In particular, the wafer is suitable for an annealed wafer and an epitaxial wafer.

  EXAMPLES Hereinafter, the present invention will be described more specifically with reference to examples and comparative examples of the present invention, but the present invention is not limited to these.

(Experiment)
A single crystal pulling apparatus as shown in FIG. 1 was equipped with a crucible having a diameter of 550 mm to grow a silicon single crystal having a diameter of 200 mm.
At this time, the carbon doping amount is adjusted so that the carbon concentration is 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc (ASTM 75), and the average growth rate is 10% at a rate of 1.05 mm / min. Single crystals of books were grown. Here, the average growth rate was defined as a value obtained by dividing the length of the straight body of the crystal by the production time of the straight body.

As a result, as shown in FIG. 3, the carbon concentration was 6 × 10 ¹⁶ atoms / cc on the top side of the single crystal and 17 × 10 ¹⁶ atoms / cc on the bottom side of the single crystal.
It was found that the effective segregation coefficient was 0.09, calculated from this result, the amount of doped carbon and the axial profile. This resulted in a large deviation from the parallel segregation coefficient of 0.07, which is often used when growing single crystals.
Thus, it was found that when the crystal growth rate is 1.05 mm / min, the effective segregation coefficient is 20% or more greater than the parallel segregation coefficient.

Further, in order to obtain the relationship between the growth rate and the effective segregation coefficient, crystal growth was performed under the same conditions as described above except for the growth rate.
As a result, it was found that the effective segregation coefficient was 0.08 when the average growth rate was 0.6 mm / min, and the effective segregation coefficient was 0.10 when the average growth rate was 1.5 mm / min. . A graph in which the effective segregation coefficient is plotted against the average growth rate is shown in FIG.
From the above experiments, it was found that the effective segregation coefficient greatly depends on the growth rate.

(Example)
A single crystal pulling apparatus as shown in FIG. 1 was equipped with a crucible with a diameter of 550 mm, and a silicon single crystal with a diameter of 200 mm was grown with a growth rate profile as shown in FIG.
Here, the actual crystal growth rate is a value that changes from moment to moment because feedback is applied from the diameter to keep the diameter of the crystal constant. Although the growth rate at this moment varies greatly, the average growth rate is more important than the instantaneous growth rate in consideration of the formation and diffusion of crystal defects. Therefore, the average growth rate for 40 minutes is used as the effective growth rate of the crystal, and this is plotted as the average growth rate. The average growth rate for 40 minutes was defined as the value obtained by dividing the length of the straight body of the crystal grown in 40 minutes by 40 minutes.

Further, the rotation speed of the crystal was set to 18 rpm so as to satisfy R × Dia> 1200.
In this crystal, the average growth rate was about 1 mm / min, and the height of the growth interface was about 20 mm. FIG. 6 shows the difference between the average growth rate for 40 minutes and the average growth rate for 40 minutes before 20 minutes.
As shown in FIG. 6, the average growth rate difference from 20 minutes ago is within ± 0.1 mm / min, most of which is stable growth rate increase within 0.05 mm / min. I understood.

In addition, a slice sample was cut out from the obtained silicon single crystal, and a portion having a carbon concentration of 6 × 10 ¹⁶ atoms / cc and an oxygen concentration of 14 × 10 ¹⁷ atoms / cc at the center of the sample was defined as a block. A slice wafer was cut out from this block to produce a polished wafer, and an epitaxial layer was formed thereon to produce an epitaxial wafer.

When the in-plane distribution of the carbon concentration was measured with this wafer, a substantially flat in-plane distribution was obtained as shown in FIG. Further, a value obtained by dividing the difference between the carbon concentration in the wafer central portion and the carbon concentration in the peripheral 10 mm portion by the carbon concentration in the central portion was 14.9%, which was 20% or less.
This wafer was heat-treated at 800 ° C. for 4 hours and at 1000 ° C. for 16 hours, and BMD inside the wafer was examined by a selective etching method.

As a result, it was confirmed that BMD of about 10 × 10 ⁴ / cm ² was formed almost uniformly in the wafer surface. The epiwafer was intentionally contaminated with Ni, subjected to a drive-in heat treatment at 800 ° C. and rapidly cooled, and then the surface shallow pits were observed. As a result, as shown in FIG. 10A, it was confirmed that there was no entire shallow pit.
Thus, it was confirmed that the silicon single crystal produced by the silicon single crystal production method of the present invention has uniform BMD or gettering ability in the plane.

(Comparative example)
A silicon single crystal was grown under the same conditions as in the example except that the growth rate profile as shown in FIG. 8 was used.
In this crystal, the average growth rate for 40 minutes was about 1 mm / min, and the height of the growth interface was about 20 mm. FIG. 9 shows the difference between the 40-minute average growth rate of the crystal and the 40-minute moving average growth rate 20 minutes before.

As shown in FIG. 9, in the growth of this crystal, the average growth rate difference for 40 minutes from 20 minutes before did not fall below ± 0.1 mm / min, and was 0.15 mm / min at the maximum. I understood.
When the in-plane distribution of the carbon concentration of this silicon single crystal was measured, the value obtained by dividing the difference between the carbon concentration at the center of the crystal and the carbon concentration at the 10 mm portion around the crystal by the carbon concentration at the center of the crystal was 29.1%. It was over 20%.

A slice sample was cut out from this single crystal, and the portion where the carbon concentration at the center of the sample was 6 × 10 ¹⁶ atoms / cc and the oxygen concentration was 13 × 10 ¹⁷ atoms / cc was taken as a block. A slice wafer was cut out from this block to produce a polished wafer, an epitaxial layer was formed thereon, and an epi wafer was produced.
This wafer was heat-treated at 800 ° C. for 4 hours and 1000 ° C. for 16 hours, and the BMD inside the wafer was examined by a selective etching method.

As a result, variation in BMD density was observed in the wafer surface, and the BMD was high at about 10 × 10 ⁴ / cm ² and low at 2.5 × 10 ⁴ / cm ² , which is the lower limit of detection.
The epiwafer was intentionally contaminated with Ni, subjected to a drive-in heat treatment at 800 ° C. and rapidly cooled, and then the surface shallow pits were observed. As a result, as shown in FIG. 10B, the portion 17 with the shallow pit and the portion without the shallow pit are mixed.

  Thus, it was found that the epiwafer obtained in the comparative example has a non-uniform BMD or gettering capability and is worried that the wafer is warped during the device process.

  The present invention is not limited to the above embodiment. The above-described embodiment is an exemplification, and the present invention has any configuration that has substantially the same configuration as the technical idea described in the claims of the present invention and that exhibits the same effects. Are included in the technical scope.

It is the schematic of the silicon single crystal pulling apparatus which can be used with the silicon single crystal manufacturing method of this invention. It is a schematic diagram of the growth stripe based on the growth rate fluctuation taken in the silicon single crystal wafer according to the present invention. It is the carbon concentration distribution in the crystal axis direction obtained from the results of the experiment. A graph showing the relationship between the crystal growth rate and the effective carbon segregation coefficient obtained from the experimental results. It is the growth rate profile of the crystal | crystallization grown in the Example. It is a graph which shows the difference of the average growth rate of the crystal | crystallization grown in the Example, and the average growth rate 20 minutes ago. It is an in-plane carbon concentration distribution of the crystal grown in the example. It is a growth rate profile of the crystal grown in the comparative example. It is a graph which shows the difference of the average growth rate of the crystal | crystallization grown by the comparative example, and the average growth rate 20 minutes ago. It is an evaluation result of in-plane uniformity of gettering ability evaluated in an example and a comparative example. (A) Evaluation results of examples. (B) Evaluation results of comparative examples.

Explanation of symbols

1 ... silicon single crystal, 2 ... main chamber, 3 ... pulling chamber,
4 ... heater, 5 ... raw material melt, 6 ... quartz crucible,
7 ... Graphite crucible, 8 ... Heat insulation member, 9 ... Rectifier tube, 10 ... Heat insulation material,
11 ... Gas inlet, 12 ... Gas outlet, 13 ... Seed crystal,
14 ... Pulling wire, 15 ... Crystal center part, 16 ... Crystal peripheral part,
17 ... Shallow pit, 21 ... Silicon single crystal pulling device,
31: A silicon single crystal wafer.

Claims (11)

  In the method of manufacturing a silicon single crystal by doping carbon by the Czochralski method, the average speed at which the crystal center in the plane perpendicular to the crystal growth direction grows and the average speed at which the crystal periphery grows during crystal growth. A method for producing a silicon single crystal, wherein the crystal is grown so that the difference between the two is within ± 0.1 mm / min.   The method for producing a silicon single crystal according to claim 1, wherein a difference between the average growth rate of the crystal and the average growth rate before a predetermined time is within ± 0.1 mm / min.   The method for producing a silicon single crystal according to claim 1, wherein the average growth rate is an average growth rate for 40 minutes.   4. The method for producing a silicon single crystal according to claim 2, wherein the predetermined time is 5 minutes or more and 80 minutes or less.   The crystal is grown by satisfying R × Dia ≧ 1200 when the crystal rotation speed during the crystal growth is R [rpm] and the diameter of the crystal is Dia [mm]. 5. The method for producing a silicon single crystal according to any one of 4 above. The silicon single crystal is grown by doping carbon so that a carbon concentration in the silicon single crystal is 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc (ASTM 75). The method for producing a silicon single crystal according to any one of claims 1 to 5. The silicon single crystal is grown so that the oxygen concentration in the silicon single crystal is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (ASTM 79). 2. A method for producing a silicon single crystal according to item 1.   A silicon single crystal wafer manufactured from a silicon single crystal manufactured by the method according to any one of claims 1 to 7.   9. The silicon single crystal wafer according to claim 8, wherein the difference between the carbon concentration at the center of the crystal and the carbon concentration at the periphery of the crystal is within 20%. 10. The silicon single crystal wafer according to claim 8, wherein the carbon concentration is 1.2 × 10 ^{16 to} 20.0 × 10 ¹⁶ atoms / cc (ASTM 75). 11. The silicon single crystal wafer according to claim 8, wherein the oxygen concentration is 8 × 10 ¹⁷ to 18 × 10 ¹⁷ atoms / cc (ASTM 79).

Images