Method for heat treatment of silicon single crystal wafer

Abstract

A method for a heat treatment of a silicon single crystal wafer in an oxidizing ambient, including: performing the heat treatment based on a condition determined by a tripartite correlation between a heat treatment temperature during the heat treatment, an oxygen concentration in the silicon single crystal wafer before the heat treatment, and a growth condition of a silicon single crystal from which the silicon single crystal wafer is cut out. This provides a method for a heat treatment of a silicon single crystal wafer which can annihilate void defects or micro oxide precipitate nuclei in a silicon single crystal wafer with low cost, efficiently, and securely by a heat treatment in an oxidizing ambient.

Claims

1. A method for a heat treatment of a silicon single crystal wafer in an oxidizing ambient, comprising: performing the heat treatment based on a condition determined by a tripartite correlation between a heat treatment temperature during the heat treatment, an oxygen concentration in the silicon single crystal wafer before the heat treatment, and a growth condition of a silicon single crystal from which the silicon single crystal wafer is cut out, wherein the silicon single crystal wafer is cut out from a silicon single crystal doped with nitrogen, and the tripartite correlation is represented by the following relational expression (B-1):
T37.5[Oi]+72.7Ivoid.sup.B+860(B-1) wherein, T is the heat treatment temperature ( C.), [Oi] is the oxygen concentration (ppma-JEIDA) in the silicon single crystal wafer before the heat treatment, and Ivoid.sup.B is represented by the following formula (B-2):
Ivoid.sup.B={(V/G)(V/G)crt}.sup.1/3{L(10801040)/2V}.sup.1/2(B-2) wherein, V is a growth rate (mm/min), G is a temperature gradient near an interface ( C./mm), (V/G)crt is a value of V/G when a defect is nonexistent, L(1080-1040) is a length of a temperature zone of void-defect formation of 1,080 C. to 1,040 C. (mm) when the single crystal is doped with nitrogen.

2. The method for a heat treatment of a silicon single crystal wafer according to claim 1, wherein the silicon single crystal wafer is cut out from a silicon single crystal doped with nitrogen at a concentration of 510.sup.15 atoms/cm.sup.3 or less.

3. The method for a heat treatment of a silicon single crystal wafer according to claim 1, wherein the silicon single crystal wafer is cut out from a silicon single crystal without a defect due to Interstitial-Si.

4. The method for a heat treatment of a silicon single crystal wafer according to claim 1, wherein the heat treatment temperature is 900 C. or more and 1,200 C. or less, and a heat treatment time is 1 minute or more and 180 minutes or less.

5. The method for a heat treatment of a silicon single crystal wafer according to claim 1, wherein the oxygen concentration of the silicon single crystal wafer is 8 ppma-JEIDA or less.

6. The method for a heat treatment of a silicon single crystal wafer according to claim 1, wherein the thickness of the silicon single crystal wafer is 0.1 mm or more and 20 mm or less.

Description

BRIEF DESCRIPTION OF DRAWINGS

(1) FIG. 1 shows graphs to plot conditions of void defect annihilation by the heat treatment at each temperature ((a): 1,150 C., (b): 1,100 C., (c): 1,050 C., (d): 1,000 C.) in [Experiment], wherein the conditions are plotted on each graph with respect to the oxygen concentration and the Ivoid.

DESCRIPTION OF EMBODIMENTS

(2) As described above, it has been required to develop a method for a heat treatment which can annihilate void defects or micro oxide precipitate nuclei in a silicon single crystal wafer with low cost, efficiently, and securely by a heat treatment in an oxidizing ambient.

(3) It is known from Patent Document 7 that the heat treatment temperature and the oxygen concentration in a wafer relate to whether void defects are annihilated or not by an oxidation heat treatment. The inventors have diligently investigated to reveal that an oxidation heat treatment condition which can annihilate void defects is actually related to a growth condition of a single crystal from which the wafer is cut out in addition to the heat treatment temperature and the oxygen concentration. The inventors have accordingly found that the foregoing subject can be achieved by performing a heat treatment based on a condition determined by a tripartite correlation thereof; thereby brought the present invention to completion.

(4) Thus the present invention is a method for a heat treatment of a silicon single crystal wafer in an oxidizing ambient, comprising:

(5) performing the heat treatment based on a condition determined by a tripartite correlation between a heat treatment temperature during the heat treatment, an oxygen concentration in the silicon single crystal wafer before the heat treatment, and a growth condition of a silicon single crystal from which the silicon single crystal wafer is cut out.

(6) Hereinafter, the present invention will be described in detail, but the present invention is not limited thereto.

(7) In the present specification, the wording an oxygen concentration indicates an oxygen concentration in the silicon single crystal wafer before the heat treatment, and the wording a crystal growth condition indicates a growth condition of a silicon single crystal from which the silicon single crystal wafer is cut out.

(8) In addition, when the Ivoid.sup.A and the Ivoid.sup.B are not distinguished, they are also described as Ivoid simply.

(9) As described later, the conditions to annihilate void defects were determined by experiments to alter the heat treatment temperature, the heat treatment time in oxidation heat treatments, the oxygen concentration, and the growth condition on various samples. It has found that whether void defects are annihilated or not depends on the heat treatment temperature, the oxygen concentration, and the crystal growth condition.

(10) The foregoing experiments have also revealed that void defects are likely to be annihilated when the heat treatment temperature is high, and likely to be annihilated when the oxygen concentration is low or the void size is small. On the other hand, it was not largely influenced by the heat treatment time. This can be reasoned that an ISi diffuses into the interior of a wafer in several minutes since the diffusion constant of an ISi is relatively large. Accordingly, it is very effective to perform a heat treatment based on a condition determined by the tripartite correlation between a heat treatment temperature, an oxygen concentration, and a crystal growth condition.

(11) The tripartite correlation can be concretely represented by the following numerical formula:
T37.5[Oi]+72.7Ivoid+860(1)

(12) This formula is transformed to the following formulae with representing [Oi] and Ivoid:
[Oi]0.0267T1.94Ivoid22.9(2)
Ivoid0.0138T0.516[Oi]11.8(3)

(13) Wherein, T is the heat treatment temperature ( C.), [Oi] is the oxygen concentration (ppma-JEIDA) in the silicon single crystal wafer before the heat treatment, and Ivoid is the defect size indicator determined by the crystal growth condition to reflect the void size.

(14) It is to be noted that, herein, ppma-JEIDA is used as a unit of an oxygen concentration, which is expressed in various units. This is transformed to atoms/cm.sup.3-ASTM'79, which is relatively widely used, as [Oi] (ppma-JEIDA)=[Oi] (atoms/cm.sup.3-ASTM'79)/(810.sup.16). Accordingly, when atoms/cm.sup.3-ASTM'79 is used as a unit, the foregoing formulae (1) to (3) can be used with substituting [Oi]/(810.sup.16) for [Oi].

(15) The foregoing formula (1) represents that a heat treatment is performed by setting the heat treatment temperature not less than the temperature determined by an oxygen concentration and a crystal growth condition; the foregoing formula (2) represents that the oxygen concentration is controlled not to exceed the concentration determined by a heat treatment temperature and a crystal growth condition; and the foregoing formula (3) represents that the crystal growth condition is controlled not to exceed the values determined by a heat treatment temperature and an oxygen concentration, respectively. Concretely, in the inventive heat treatment method, a heat treatment or control is performed so as to satisfy any of these.

(16) In the following, Ivoid in the relational expression will be specifically explained.

(17) As described above, it has been found by the experiments that an annihilation of void defects also depends on a crystal growth condition. By comparing conditions in which void defects are annihilated and crystal growth conditions, it has been found that void defects tend to be annihilated in crystal growth conditions such that the void size gets small. Accordingly, the inventors have inferred that it would be reasonable to express the void size by the condition in a crystal growth, thereby introducing a defect size indicator: Ivoid determined by a crystal growth condition and reflecting a void size.

(18) On the basis of a theory of Grown-in defect in a silicon single crystal advocated first and advanced by Mr. Voronkov (see V. V. Voronkov; Journal of Crystal Growth, 59 (1982) 625 to 643), it is estimated that the amount of introduced vacancy is increased by larger V/G, which is a ratio of a crystal growth rate V (mm/min) and a temperature gradient at an interface of crystal growth G ( C./mm), and the amount of ISi is increased when V/G is smaller. Accordingly, this V/G values were utilized in order to introduce Ivoid.

(19) In a single crystal, it is known that there is a boundary between vacancy-rich region and ISi-rich region, and the vicinity is defect-free. In the present invention, the V/G value at the boundary is defined as (V/G)crt and is utilized to introduce Ivoid. It is to be noted that in a condition of the present calculation, (V/G)crt is 0.180 mm.sup.2/(min.Math. C.).

(20) It is considered that a void size is determined by vacancies introduced at high temperature and aggregated during passing through a temperature zone of defect formation. Accordingly, the amount of introduced vacancy is concisely determined as {(V/G)(V/G)crt}.sup.1/3. Herein, the reason for raising to the power is that the value in which {(V/G)(V/G)crt} was raised to the power was almost proportional to the amount of introduced vacancy separately determined by simulations of reactions occurring at high temperature such as slope diffusion.

(21) Furthermore, (Dt) is determined by assuming that the amount of aggregation in a temperature zone of void defect formation is proportional to the diffusion length (Dt) in the corresponding area. Herein, the case of a single crystal without nitrogen-doping and the case of a single crystal doped with nitrogen are explained separately since they differ in the temperature zone of void-defect formation.

(22) First, the case of a single crystal without nitrogen doing will be explained.

(23) Without nitrogen-doping, the temperature zone of void-defect formation is 1,150 C. to 1,080 C. Accordingly, the diffusion length (Dt) is determined as:
(Dt){L(11501080)/V}.sup.1/2
Wherein, D is a diffusion coefficient of a vacancy (which is a constant when the temperature is determined), t is a transit time, L(11501080) is a length of a temperature zone of void-defect formation of 1,150 C. to 1,080 C., and V has the same meaning as in the foregoing.

(24) From the product of the diffusion length and the amount of introduced vacancy concisely determined, Ivoid.sup.A is determined as:
Ivoid.sup.A={(V/G)(V/G)crt}.sup.1/3{L(11501080)/V}.sup.1/2 (A-2)
wherein, Ivoid.sup.A is an Ivoid when the single crystal is not doped with nitrogen.

(25) Furthermore, this Ivoid.sup.A is applied to the foregoing formula (1) to obtain the following tripartite relational expression (A-1) between a heat treatment temperature, an oxygen concentration, and a crystal growth condition when the single crystal is not doped with nitrogen:
T37.5[Oi]+72.7Ivoid.sup.A+860(A-1)

(26) Then, the case of a single crystal doped with nitrogen will be explained.

(27) It is said that the temperature zone of void-defect formation is lowered to 1,080 to 1,040 C. with nitrogen-doping from 1,150 to 1,080 C. when a crystal is not doped with nitrogen. This vacancy diffusion coefficient of 1,080 to 1,040 C. is approximately half of the vacancy diffusion coefficient of 1,150 to 1,080 C. Accordingly, the diffusion length (Dt) is determined as:
(Dt){L(10801040)/2V}.sup.1/2
Wherein, D, t, and V have the same meanings as in the foregoing, and L(10801040) is a length of a temperature zone of void-defect formation of 1,080 C. to 1,040 C.

(28) From the product of the diffusion length and the amount of introduced vacancy concisely determined, Ivoid.sup.B is determined as:
Ivoid.sup.B={(V/G)(V/G)crt}.sup.1/3{L(10801040)/2V}.sup.1/2 (B-2)
wherein, Ivoid.sup.B is an Ivoid when the single crystal is doped with nitrogen.

(29) Furthermore, this Ivoid.sup.B is applied to the foregoing formula (1) to obtain the following tripartite relational expression (B-1) between a heat treatment temperature, an oxygen concentration, and a crystal growth condition when the single crystal is doped with nitrogen:
T37.5[Oi]+72.7Ivoid.sup.B+860(B-1)

(30) As described above, it is possible to annihilate void defects securely by setting the heat treatment temperature, the oxygen concentration, and the crystal growth condition so as to satisfy each relational expression when using either wafer cut out from a silicon single crystal without nitrogen-doping or cut out from a silicon single crystal doped with nitrogen.

(31) It is possible to judge whether void defects are annihilated or not from the crystal growth condition by using this relational expression without determining a size of a void defect in a grown crystal by various evaluation methods every time. Moreover, utilizing this correlation, it is possible to reduce the heat treatment temperature by controlling the oxygen concentration to relatively low value and controlling the crystal growth condition so as to make the void size relatively small, for example. If the heat treatment temperature can be lowered, it is possible to reduce the cost, and it is also possible to suppress a generation of slip dislocation during a heat treatment, which is liable to generate at higher temperature.

(32) The silicon single crystal wafer used for the inventive heat treatment method is preferably cut out from a general crystal which is not intentionally doped with impurities except for a dopant to control the resistivity. This is because void defects can be annihilated as long as the oxygen concentration is so low as to satisfy the foregoing condition even when using a general crystal.

(33) On the other hand, it is known that nitrogen-doping improves the durability to slip dislocation. Moreover, being doped with nitrogen, the defect-forming temperature zone is lowered, and the void sizes tend to small as described above. Accordingly, in the inventive heat treatment method, it is also preferable to use a silicon single crystal wafer cut out from a crystal which is intentionally doped with nitrogen in addition to a dopant to control the resistivity.

(34) In this case, the doping amount of nitrogen is preferably 510.sup.15 atoms/cm.sup.3 or less. Since the solid solubility limit of nitrogen in a silicon crystal is said to be an order of the 15th power, it is possible to avoid the risk of dislocation generation in the crystal due to a nitrogen-doping with high-concentration by setting the concentration as in the foregoing. On the other hand, the nitrogen concentration does not have a lower limit. This is because the inventive heat treatment method can be used without any problem when nitrogen is not doped.

(35) In calculating the foregoing defect size indicator Ivoid, however, it is preferable to treat the nitrogen concentration of 110.sup.12 atoms/cm.sup.3 or more as being doped with nitrogen. This is because the effect of nitrogen-doping such as lowering of a void size is brought at this concentration or more.

(36) The inventive heat treatment method can annihilate void defects, but cannot annihilate defects due to ISi. Accordingly, it is preferable to use a silicon single crystal wafer which is cut out from a silicon single crystal without a defect due to ISi.

(37) In a single crystal, OSF nuclei are generated in a region (OSF region) at the side with the growth rate lower than in a region in which void defects are generated. At the side with further low rate, there is a defect-free region. The defect-free region contains a region with many vacancy (Nv region) and a region with many ISi (Ni region). The Nv region has a part which contains micro oxide precipitate nuclei. At the further low rate side, there is an I-rich region in which defects due to ISi are generated.

(38) It is considered that the inventive heat treatment method can also eliminate an OSF nucleus or a micro oxide precipitate nucleus. Accordingly, the inventive heat treatment method can be effectively applied to a silicon single crystal with any of the void defects generation region, the OSF region, the Nv region, and the Ni region, excluding the foregoing I-rich region.

(39) That is, in the heat treatment method of the present invention, an improvement of the OSF region or the Nv region can also be expected not only an improvement of the void defects generation region. Accordingly, the inventive heat treatment method is effective against all regions which does not contain a defect due to ISi.

(40) In the inventive heat treatment method, the heat treatment temperature is preferably 900 C. or more and 1,200 C. or less. When the heat treatment temperature is 900 C. or more, it is possible to annihilate void defects with sizes which affect to the electrical properties. On the other hand, the heat treatment temperature of 1,200 C. or less enables to reduce the cost and to suppress a generation of slip dislocation. More preferably, the temperature is 1,150 C. or less, since it is possible to further suppress a generation of slip dislocation when the temperature is 1,150 C. or less.

(41) The heat treatment time is preferably 1 minute or more and 180 minutes or less, although it depends on the thickness of a wafer to be used. As described above, the diffusion of ISi is relatively fast and gives a diffusion length of nearly 1 mm, which is a thickness of a conventional wafer, in a minute. Accordingly, about 1 minute is enough for the heat treatment time. On the other hand, a heat treatment time more than 180 minutes is needless since a prolonged heat treatment time causes cost increase.

(42) As a silicon single crystal wafer, it is preferable to use a wafer with an oxygen concentration of 8 ppma-JEIDA or less. This is because the solid solubility limit of oxygen at 1,200 C., which is the foregoing preferable heat treatment temperature, is approximately 8 ppma-JEIDA, and more increased oxygen concentration requires a treatment at a higher temperature. It is more preferable to use a wafer with an oxygen concentration of 6 ppma-JEIDA or less. This is because the solid solubility limit of oxygen at 1,150 C. is approximately 6 ppma-JEIDA, and it is possible to further suppress a generation of slip dislocation when the temperature is 1,150 C. or less. The oxygen concentration does not have a lower limit. As can be seen from the foregoing formulae, as the oxygen concentration is lower, it is possible to lower the heat treatment temperature required to annihilate void defects, and accordingly it is possible to reduce the cost and to suppress a generation of slip dislocation during a heat treatment, which is liable to generate at a higher temperature. The inventive heat treatment method can also be utilized to a wafer which is cut out from a crystal which hardly contains oxygen such as an FZ crystal not only a wafer cut out from a CZ crystal.

(43) In the inventive heat treatment method, it is preferable to use a silicon single crystal wafer with a thickness of 0.1 mm or more and 20 mm or less. In the inventive heat treatment method, a thinner wafer does not bring any problem, however, the thickness of 0.1 mm or more is preferable since it enables to retain the wafer shape easily.

(44) On the other hand, the heat treatment is preferably performed at 1,200 C. or less for 180 minutes or less in view of cost and so on as described above. The diffusion length of ISi during a heat treatment at 1,200 C. for 180 minutes is approximately 10 mm. Since oxide films are formed on both of the front side and the back side, and ISi is supplied therefrom, it is not possible to reform a wafer thickness beyond 20 mm or so when a heat treatment at 1,200 C. for 180 minutes is performed. Accordingly, it is preferable to use a wafer with the thickness of 20 mm or less in order to annihilate void defects in an entirety of the wafer within the foregoing heat treatment temperature and heat treatment time.

(45) The state of the surface of a silicon single crystal wafer to be used for the inventive heat treatment method may be any surface state used in a producing process of a silicon wafer such as a polished surface, an etched surface, a lapped surface, a grinded surface, and a sliced surface. In the inventive heat treatment method, the surface state is not particularly concerned as long as an oxide film forms. Washing or so on is required to be put into a heat treatment furnace, but other particular surface treatment is not required. The heat treatment can be performed in any process of producing a wafer. Accordingly, the heat treatment may be performed in any surface state to be used in a production process of a silicon wafer.

(46) The inventive heat treatment method is performed in an oxidizing ambient. The conditions such as an oxygen flow rate is not particularly limited as long as being an atmosphere which contains oxygen.

(47) As described above, in the inventive method for a heat treatment of a silicon single crystal wafer, it is possible to annihilate void defects or micro oxide precipitate nuclei in a silicon single crystal wafer with low cost, efficiently, and securely while suppressing a generation of slip dislocation by a heat treatment in an oxidizing ambient.

(48) Moreover, by using a wafer without a defect due to ISi, it is possible to obtain a defect-free silicon single crystal wafer which does not contain a defect due to either void or ISi. Such a wafer is particularly suitable for a defect-free wafer used for a substrate of a semiconductor device such as a memory, a CPU, and a power device.

EXAMPLE

(49) The present invention will be more specifically described below with reference to Example and Comparative Example, but the present invention is not limited thereto.

EXPERIMENT

(50) Each silicon single crystal with a diameter of a little over 200 mm or 300 mm was grown by using a CZ method or a magnetic field applied CZ (MCZ) method. In these cases, the crystals were grown with varying the oxygen concentration and the crystal growth condition. The oxygen concentration was controlled by altering the rotation number of a crystal, the rotation number of a crucible, the pressure in a furnace, and the Ar gas flow rate, which was flown in order to purge. On the other hand, the crystal growth condition was varied by altering the temperature gradient near an interface G and the length of a temperature zone of void-defect formation L(11501080) or L(10801040) with the structure of the parts in the furnace and/or controlling the growth rate V as well as an existence or nonexistence of nitrogen-doping. In this case, the crystal growth condition was adjusted in such a way that a defect due to ISi was not contained.

(51) This crystal was grinded cylindrically to process a cylindrical block with a desired diameter. Then, a wafer-form sample with a thickness of approximately 1.2 mm was cut out from the block. In this case, each three pieces of samples were cut out from adjacent positions per a part. One piece of these was used as a sample for measuring the oxygen concentration, and an existence or nonexistence of an FPD, an LEP, and an LSTD before a heat treatment (hereinafter referred as a measure sample). Other two pieces were actually subjected to the heat treatment as described below, and used as samples for measuring an FPD and an LSTD after the heat treatment (hereinafter referred as a sample for heat treatment). The samples were prepared with 35 of levels. In these levels, 5 levels were doped with nitrogen.

(52) The measure samples were subjected to surface grinding to be high brightness, and then each oxygen concentration was measured by an FT-IR method. In these cases, the oxygen concentrations of the measure samples were within a range of 0.4 to 12.2 ppma-JEIDA (0.3 to 9.810.sup.17 atoms/cm.sup.3-ASTM'79). Furthermore, the measure samples were subjected to mirror etching by a mixed acid composed of hydrofluoric acid, nitric acid, and acetic acid. Then the measure samples were left in a selective etching liquid composed of hydrofluoric acid, nitric acid, acetic acid, and water to be subjected to selective etching without shaking. In each of these measure samples, FPDs were observed but an LEP was not observed, which confirmed that the defect due to ISi was not contained.

(53) Further, these measure samples were cleaved, and then LSTD was observed with an infrared scattering tomograph MO441 (product of Raytec Co., Ltd.). With MO441, each observation was performed to a depth of approximately 400 m from the surface. It was confirmed that an LSTD was existent at this time.

(54) Each temperature gradient near an interface G and length of a temperature zone of void-defect formation L(11501080) or L(10801040) were determined by calculating the crystal growth condition at the position from which the foregoing measure sample was cut out by a global heat transfer analysis software FEMAG (see F. Dupret et al.; Int. J. Heat Mass Transfer, 33, 1849 (1990)). In the FEMAG, calculations were performed by fixing the interface form as 10 mm. Each temperature gradient near an interface G was calculated from the melting point (=1,412 C.) to 1,400 C. From these and the growth rate V in a crystal growth, Ivoid when a crystal is not doped with nitrogen (Ivoid.sup.A) was determined as Ivoid.sup.A {(V/G)(V/G)crt}.sup.1/3{L(11501080)/V}.sup.1/2. As described above, among 35 levels of the measure samples, 5 levels were doped with nitrogen, and the concentration were 4 to 1210.sup.13 atoms/cm.sup.3. Accordingly, Ivoid of these 5 levels of measure samples doped with nitrogen (Ivoid.sup.B) were determined as Ivoid.sup.B={(V/G)(V/G) crt}.sup.1/3 {L(10801040)/2V}.sup.1/2. It is to be noted that (V/G)crt is a value of V/G when a defect is nonexistent, which was calculated as (V/G)crt=0.180 in the foregoing calculation condition.

(55) Then, heat treatments were actually performed by using the foregoing samples for heat treatment.

(56) First, two pieces of samples for heat treatment were subjected to surface grinding to be high brightness as a pretreatment for the heat treatment, and then each were divided in four and subjected to mirror etching with the foregoing mixed acid. After the etching, the samples for heat treatment were subjected to heat treatment in an oxidizing ambient of dry oxygen of 3 L/min. In these cases, the temperatures and the times of the heat treatments were set to the following 42=8 patterns: (a) at 1,150 C. for 30 minutes or 60 minutes, (b) at 1,100 C. for 30 minutes or 60 minutes, (c) at 1,050 C. for 60 minutes or 120 minutes, and (d) at 1,000 C. for 60 minutes or 120 minutes.

(57) On the surfaces of the heat treated samples, oxide films were formed, and these were removed with hydrofluoric acid.

(58) Then, the surfaces were subjected to mirror etching with the mixed acid, followed by selective etching to observe FPD in the same way as in the measure samples. In some of the samples, FPDs were disappeared compared to measure samples which were not subjected to heat treatment. Subsequently, these samples were cleaved, and then LSTD was observed with MO441. In this time, it was also confirmed in some samples that LSTDs were disappeared compared to measure samples which were not subjected to heat treatment.

(59) Although they had surface state of etching faces, the defects were annihilated without any problem by an oxidation treatment. Basically, the same trend were observed between the cases in which FPDs were annihilated and the cases in which LSTDs were annihilated. However, FPDs tended to be annihilated easily. This can be reasoned that a wafer was observed through the interior thereof in an LSTD, while several tens m from the surface was observed in an FPD. On the other hand, the heat treatment time hardly influenced whether defects were annihilated or not as a result of varying the heat treatment times with 2 levels in each heat treatment temperature.

(60) The foregoing results are shown in FIG. 1 (a) to (d) in which the abscissa and ordinate represent the oxygen concentration and the Ivoid, respectively, by plotting as when the defects were annihilated and when the defect remained. It is to be noted that the plotted results are results of LSTD when the heat treatment were performed for 60 minutes at each temperature.

(61) It is found that the void defect is likely to annihilate as the oxygen concentration is lower or as the Ivoid is smaller from the conditions which annihilated void defects at each temperature of FIG. 1. It is also found that the defects are more difficult to be annihilated as the temperature falls from 1,150 C. to 1,000 C.

(62) The process to annihilate a void defect is assumed that an oxide film of inner wall disappears first, and then ISi fills the void defect. As described above, whether an oxide film of inner wall disappear or not is determined by whether the oxygen concentration is lower than the solid solubility limit of oxygen at the heat treatment temperature or not. That is, it is considered that the solid solubility limit of oxygen determines the temperature dependence whether the defects are annihilated or not.

(63) Several solid solubility limits of oxygen are reported, and they are represented as [Oi]=2.6510.sup.4 exp[1.035/{k(T+273)}], for example. Wherein, k is the Boltzmann constant, which is 8.6210.sup.5. By using this formula, the solid solubility limits of oxygen is about 6 ppma-JEIDA at 1,150 C. and about 2 ppma-JEIDA at 1,000 C. Accordingly, the following relation approximately holds between the temperature and the oxygen concentration in this temperature range: T37.5[Oi]. On the basis of this, the condition under which void defects were annihilated in (a) to (d) of FIG. 1 is determined as T37.5[Oi]+72.7Ivoid+860. This is transformed on the basis of the oxygen concentration to give [Oi]0.0267T1.94Ivoid22.9, and transformed on the basis of the Ivoid to give Ivoid0.0139T0.516[Oi]11.8.

(64) From the foregoing experiments, it was found that the tripartite correlation between a heat treatment temperature, an oxygen concentration, and a crystal growth condition determines whether the void defects are annihilated or not by oxidation.

(65) Among these, the crystal growth condition is easy to understand by expressing Ivoid, which is an indicator to reflect a void size and is expressed by a product of (V/G)(V/G)crt}.sup.1/3 corresponding to the amount of introduced vacancy and {L(11501080)/V}.sup.1/2 or {L(10801040)/2V}.sup.1/2 corresponding to the diffusion length in a temperature zone in which defects are formed.

Example 1

(66) In the blocks prepared in the foregoing experiment, a wafer was cut out from the part of a block with a diameter of 300 mm, oxygen concentration of 3.2 ppma-JEIDA, and Ivoid.sup.A of 1.18 (V=0.66 mm/min, G=3.49 C./mm, L(11501080)=21.0 mm) without nitrogen-doping to produce a polished wafer (PW=polished side) with the thickness of 775 m. The measured oxygen concentration [Oi] and determined Ivoid.sup.A were substituted to the relational expression to determine the required heat treatment temperature as follows:
T37.53.2+72.71.18+860=1,066
On the basis of this determined heat treatment temperature, the wafer was subjected to a heat treatment at 1,150 C. for 30 minutes in an oxidizing ambient of dry oxygen of 3 L/min.

(67) After the heat treatment, the oxide film was removed, and then the defect of this wafer was observed with MO441. As a result, no LSTD was detected.

Example 2

(68) A crystal with a diameter of about 200 mm was grown by an MCZ method using a crystal pulling apparatus having the outer diameter of a crucible of about 660 mm. This crystal was doped with nitrogen. Two pieces of wafers with each thickness of 1.2 mm were cut out from adjacent positions in this crystal. The nitrogen concentration at which the wafers were cut out was 810.sup.13 atoms/cm.sup.3. One of the cut out wafers was subjected to double-side grinding, and then the oxygen concentration was measured by FT-IR to be 2.8 ppma-JEIDA. This measure wafer was subjected to mirror etching by a mixed acid composed of hydrofluoric acid, nitric acid, and acetic acid, followed by cleaving, and then observed with MO441 to find an LSTD.

(69) The growth condition at which this measure wafer had been cut out was calculated by FEMAG as described above to reveal that the temperature gradient near an interface G ranging from the melting point to 1,400 C. was 3.17 C./mm, and the length of a temperature zone of void-defect formation L(10801040) was 13.8 mm. The growth rate V at which this wafer had been cut out was 0.90 mm/min. From the foregoing values, putting (V/G)crt=0.180, Ivoid.sup.B of this wafer was determined as follows:
Ivoid.sup.B={(0.90/3.17)0.180}.sup.1/3{13.8/(20.90)}.sup.1/2=1.30
The measured oxygen concentration [Oi] and Ivoid.sup.B determined as described above were substituted to the relational expression to determine the required heat treatment temperature as follows:
T37.52.8+72.71.30+860=1,060
Another piece of cut out wafer was subjected to surface grinding, followed by mirror etching, and then subjected to a heat treatment at 1,100 C. for 30 minutes in an oxidizing ambient of dry oxygen of 3 L/min on the basis of the heat treatment temperature determined as the foregoing.

(70) After the heat treatment, the oxide film was removed, and then the defect of this wafer was observed with MO441. As a result, no LSTD was detected.

Comparative Example 1

(71) A crystal with a diameter of about 200 mm was grown by an MCZ method using a crystal pulling apparatus having the outer diameter of a crucible of about 660 mm. This crystal was doped with nitrogen. Two pieces of wafers with each thickness of 1.2 mm were cut out from adjacent positions in this crystal. The nitrogen concentration at which the wafers were cut out was 710.sup.13 atoms/cm.sup.3. One of the cut out wafers was subjected to surface grinding, followed by mirror etching, and then subjected to a heat treatment at 1,150 C. for 30 minutes in an oxidizing ambient of dry oxygen of 3 L/min without measuring the oxygen concentration and calculating the Ivoid.

(72) After the heat treatment, the oxide film was removed, and then the defect of this wafer was observed with MO441. As a result, LSTDs were detected.

(73) For confirmation, another piece of cut out wafer was subjected to double-side grinding, and then the oxygen concentration was measured by FT-IR to be 11.2 ppma-JEIDA. This measure wafer was subjected to mirror etching by a mixed acid composed of hydrofluoric acid, nitric acid, and acetic acid, followed by cleaving, and then observed with MO441 to detect very small LSTDs.

(74) The growth condition at which this measure wafer had been cut out was calculated by FEMAG as described above to reveal that the temperature gradient near an interface G ranging from the melting point to 1,400 C. was 3.82 C./mm, and the length of a temperature zone of void-defect formation L(10801040) was 11.6 mm. The growth rate V at which this measure wafer had been cut out was 0.88 mm/min. From the foregoing values, putting (V/G)crt=0.180, Ivoid.sup.B of this wafer was determined as follows:
Ivoid.sup.B={(0.88/3.82)0.180}.sup.1/3{11.6/(20.88)}.sup.1/2=0.95
The measured oxygen concentration [Oi] and Ivoid.sup.B determined as described above were substituted to the relational expression to determine the required heat treatment temperature as follows:
T37.511.2+72.70.95+860=1,349
and the reason why the LSTD was not disappeared by a heat treatment at the heat treatment temperature of 1,150 C. was suggested that the heat treatment temperature determined by the correlation was not fulfilled.

(75) From the foregoing, it was revealed that the inventive method for a heat treatment of a silicon single crystal wafer can annihilate void defects or micro oxide precipitate nuclei in a silicon single crystal wafer with low cost, efficiently, and securely by a heat treatment in an oxidizing ambient.

(76) It is to be noted that the present invention is not limited to the foregoing embodiment. The embodiment is just an exemplification, and any examples that have substantially the same feature and demonstrate the same functions and effects as those in the technical concept described in claims of the present invention are included in the technical scope of the present invention.