WO1991016476A1

WO1991016476A1 - Silicon single crystal manufacturing apparatus

Info

Publication number: WO1991016476A1
Application number: PCT/JP1991/000477
Authority: WO
Inventors: Hiroshi Kamio; Kenji Araki; Takeshi Kaneto
Original assignee: Nkk Corporation
Priority date: 1990-04-13
Filing date: 1991-04-11
Publication date: 1991-10-31
Also published as: JP2585123B2; KR920702732A; EP0484538A1; CN1056135A; JPH03295891A

Abstract

A large-diameter silicon single crystal manufacturing apparatus including a rotation-type quartz crucible, an electric resistance heater, a quartz partition member having communication holes, a heat keeping cover, etc. The total sum A of the cross-sectional areas of the communication holes formed in the partition member is between 80 and 100 mm2 when the feed rate of starting material is between 30 and 50 g/min, not less than 130 mm2 and not greater than 1200 mm2 when the starting material feed rate is in the range from 50 to 80 g/min and not less than 220 mm2 and not greater than 1600 mm2 when the starting material feed rate is in the range from 80 to 130 g/min.

Description

DESCRIPTION

■ TITLE OF THE INVENTION

SILICON SINGLE CRYSTAL MANUFACTURING APPARATUS

.TECHNICAL FIELD

The present invention relates to an apparatus for manufacturing large-diameter silicon single crystals according to the Czochralski method. More specifically, the invention relates to a silicon single crystal manufacturing apparatus including a rotation-type quartz crucible containing molten silicon, an electric resistance heater for heating the quartz crucible from the side thereof, a quartz partition member adapted for dividing the molten silicon into a single crystal growing section and a material melting section and having a plurality of communication holes for permitting unidirectional passage of the molten silicon therethrough, a heat keeping cover for covering the partition member and the material melting section, starting material feed means for continuously feeding starting material silicon to the material melting section, and dopant feed means for feeding a dopant to the material melting section. _ BACKGROUND ART^'

In the field of LSIs, the required diameter for silicon single crystals has been increased year^" after year. At present, silicon single crystals of 6 inches _c in diameter are used for the latest devices. It is now said that in future silicon single crystals of 10 inches • or more in diameter, e.g., silicon single crystals of 12 inches in diameter will be needed.

In accordance with the Czochralski method (CZ _Q method) , there are two kinds of methods concerning the process of producing silicon single crystals. They include one in which the crucible is rotated and another in which the crucible is not rotated. Today, all the silicon single crystals according to the CZ method, used ₅ for LSIs,are manufactured by methods of the type in which the crucible is rotated in a direction opposite to the rotation of a silicon single crystal and also the crucible is heated by an electric resistance heater surrounding the side of the crucible. In spite of 0 various attempts, up to date, silicon single crystals of 5 inches or more in diameter have not been manufactured by any methods in which the crucible is not rotated or any heating method other than the previously mentioned one. The reason is that without the rotation of the 5 crucible or by the use of any other heating method than the foregoing one, e.g., the magnetic induction heating or the electric resistance heating at the bottom of the crucible, it is impossible to ensure a circular temperature distribution which is completely concentric with a growing silicon single crystal. The growth of a silicon single crystal is extremely sensitive with respect .to temperature.

In the CZ method of the type which rotates the crucible (hereinafter referred to as the ordinary CZ method) , the rotation of the crucible and the electric resistance side heating produce a strong convection of molten silicon so that the molten silicon is stired vigorouly. This results in a molten silicon surface temperature distribution which is preferable for the growth of a large-diameter silicon single crystal or which is uniform and completely concentric with a silicon single crystal.

As mentioned previously, there is a great difference between the ordinary -CZ method and the other CZ method with respect to the flow of molten silicon. This difference results in wide variations of the silicon single crystal growing conditions. Thus, the two methods also differ greatly from each other with respect to the functions of the internal components of the furnace. The two methods are entirely different from each other with respect to the concept for the growth of silicon single crystals. According to the ordinary CZ method, the amount of the melt within the crucible is decreased with the growth of a silicon single crystal. Thus,-, as the silicon single crystal grows, the dopant concentration in the silicon single crystal is increased and the oxygen concentration is decreased. In other words, the properties of the silicon single crystal are varied in the direction of its growth. Since the quality required for silicon single crystals has been becoming increasingly severe year after year along with the tendency toward increasing the level of integration of LSIs, this problem must be overcome.

As a means of overcoming the problem, there are known the methods of the type in which the interior of a quartz crucible according to the ordinary CZ method is divided by a cylindrical quartz partition member having communication holes for molten silicon and a cylindrical silicon single crystal is grown on the inner side of the partition member while feeding starting material silicon "to the outer side of the partition member (hereinafter referred to as a CC-CZ method) (e.g. Patent Publication No. 40-10184, PI, L20 - L35) . As also pointed out by Laid-Open Patent No.62-241889 (P2, L12 - L16), this method has a serious disadvantage that the molten silicon tends to be solidified on the inner side of the partition member with the partition member as a starting point.. This is due to the following causes. As will be seen from the fact that quartz is used for optical fibers, the partition member made from-, ^"quartz efficiently transmits heat by radiation. In other words, the heat in the molten silicon is transmitted as light through the partition member upwardly and it is dissipated from the portion of the partition member which is exposed on the molten silicon surface. As a result, the molten silicon temperature is decreased greatly in the vicinity of the partition member. Also, in the ordinary CZ method, due to the strong agitation of the molten silicon, the surface temperature of the molten silicon is not only uniform but also just above the solidifying point. Owing to the combination of these two facts, the molten silicon surface contacting with the partition member is in a condition having an extremely high tendency toward causing the occurrence of solidification.

Laid-Open Patent No. 1-153589 is one proposing a method employing such partition member and adapted to prevent the occurrence of solification at the partition member. This laid-open patent propose to completely cover the partition member with a heat keeping member. By this method, the dissipation of heat from the partition member can be prevented and also the occurrence of solidification can be prevented. This invention further proposes to reduce the size of the molten silicon communication holes from the material melting section to the crystal growing section in such a manner that the molten silicon flows substantially in one direction from the former to the latter. Due to the synergetic effect of' .-this fact and the provision of the heat keeping member, the molten silicon temperature . in the material melting section can be maintained at a temperature sufficient for stably effecting the melting of starting material.

In accordance with inventions of the above type, it has been becoming possible to produce silicon single crystals which are uniform in composition over the whole length. The most important aim for the development of the CC-CZ method is the production of very long silicon single crystals which are uniform in characteristics over the whole length. Then, recently the kinds of semiconductor devices have been increasingly diversified. As a results, the kinds of silicon single crystal have also been increasingly diversified. The tendency of the products of silicon single crystals is toward smaller lots. There has been a demand for a CC-CZ method capable of easily changing the resistance value during the growth of a silicon single crystal. However, the CC-CZ method which have been under trial development invarious quarters give no consideration to the tendency toward smaller lots.

DISCLOSURE OF INVENTION The present invention proposes to overcome the foregoing problems in that in the CG-CZ furnace including the heat keeping member as in the case of Laid-Open Patent No. 1-153589, the cross-sectional area of the communication holes for molten silicon from the material melting section to the crystal growing section is made proper in^" consideration of the balance with the feed rate of starting material. The operation of the present invention will now be explained by taking the case where the dopant is phosphorus with reference to Figs. 4, 5 and 6. The distribution coefficient of phosphorus is about 0.33. In the steady-state pulling condition of a silicon single crystal, if the mixing of the molten silicon between a material melting section A and a crystal growing section B is effected extremely satisfactorily (mixed system) , the phosphorus concentrations of the two regions are equal. This condition is realized when the cross-sectional area of the communication holes is large. Where the concentration of phosphorus ^• taken into a silicon single crystal 5 is assumed as 1, the concentration in B is 3 and the concentration in A is also 3. A main starting material 21 and subsidiary material 22

• containing P are fed in such proportions that the concentration becomes 1. In this case, the mixing of the molten silicon in A and B is excellent so that the liquid temperatures in the two regions are substantially equal. For the growth of- the silicon

.single crystal, the liquid temperature- in B must substantially be equal to the solidifying point. As a result, the liquid temperature in A is also substantially equal to the solidifying point. In other words, the CC-CZ method cannot be materialized.

In order to avoid this problem, it is only necessary to decrease the cross-sectional area of the communication holes 10 for the molten silicon in such a manner that the molten silicon is caused to completely flow only in one direction from A to B (non-mixed system) . In this case, the concentration distribution of the dopant is such that it is 1 in the silicon single crystal, 3 in B, 1 in A and 1 in the starting material system . The transfer of heat from A to B is so small that the provision of the heat keeping cover has the effect of easily increasing the liquid temperature in A to a temperature sufficient for melting the starting material. Then, in this case, consider the case in which the concentration of the dopant in the silicon single crystal is increased two times (the resistance value is reduced to one half) during the growing of the silicon single crystal. Fig. 5 shows schematically how the dopant concentration in various part^'s are varied. Firstly, at a time T , the subsidiary material is fed in an amount required for attaining the concentration of 6 in B and the concentration of 2 in A. As a result, the concentration in A- is increased rapidly. However, the transfer to B of the molten silicon of A which contains high-concentration dopant is limited to the feed rate of the starting material (the rate of grow of the crystal) and therefore the transfer of the dopant to B is very slow. In order that the transfer may be practically completed, the starting material, (the concentration is 2) must be fed to A in an amount which is ten times the amount of molten silicon in A. In other words, in order that the concentration in B becomes 6 and hence the concentration in the crystal attains the desired value of 2, the crystal must be grown about ten times the amount of molten silicon in A. In the case of the non-mixed system, it is substantially impossible to vary the amount of the dopant during the crystal growth.

The present invention is based on the discovery that by properly adjusting the cross-sectional area of the communication holes to balance with the feed rate of starting material, it is possible to realize a condition which substantially constitutes a mixed system with respect to the dopant and which substantially constitutes a unmixed system with respect to -heat. As regards the heat, this can be realized since the flow of heat without passing through the communication holes 10 is extremely large between the regions -A- and B. The concept for realizing the above-mentioned situation will now be described.

Assume first the case in which the molten silicon flows completely unidirectionally from A to B (the unmixed system). As regards the heat, practically all the heat is introduced through the crucible bottom and the partition member and it is dissipated from the molten silicon surface. Only a small portion of the heat is introduced by being entrained on the molten silicon flowing from A into B through the communication hole. If the former is 100, then the later is less than 0.5. As regards the dopant, the whole amount is introduced through the communication holes as a matter of course.

Next, assume the case in which some mixing is added to the unmixed system. Let it be assumed that the inlet flow from A to B is increased ten times (90% of the inlet amount flow backward from B to A) . Such situation is realized by slightly increasing the cross- sectional area of the communication holes 10 as compared with the case of the unmixed system. Due to the effect of this addition of some mixing, the transfer to B of the dopant fed to A is facilitated greatly. _ In this example, it is increased by about ten times. In Fig. 5, the mixing time (T_ - ₁) or the crystal length for mixing (the transition region of the resistance value) becomes about 1/10. Such system can be - substantially considered as a mixed system with respect to the dopant.

Fig. 4 shows the manner in which the dopant concentration is varied in the various parts when the operation of increasing the dopant concentration in such a system by two times is performed. Due to the presence of the mixing, in the steady state the dopant concentration in A substantially equal to that in B. In other words, it is 3 before the time T and it is 6 after the time T». The increase in the amount of the subsidiary starting material at the time I. represents the required amount for increasing the concentration in the whole molten silicon from 3 to 6. Then, consider the manner in which the heat flows. Stating the conclusion first, the heat flow is substantially the same with that in the unmixed system. Note B first. While the amount of heat introduced by the inlet flow from A to B is increased by ten times due to the presence of the mixing, the value is still 5 at the most (the amount of heat introduced from the crucible bottom and the partition member is 100 as mentioned previously) . This represents only 5% of the whole heat amount introduced into A at the most . In other orders the heat balance does not differ much from the case of the unmixed system. As will be deduced easily, the foregoing discussion holds in the case of A. In other words, also in A, the effect of the flow rate through the communication holes being increased by ten times is not great concerning the flow of the heat due to the previously mentioned presence of mixing. Where the degree of mixing is low in this way, the flows of the heat in A and B are substantially the same as in the case of the unmixed system. In other words, as in the case of the unmixed system, the stable melting of the starting material is made possible and thus the CC-CD method is materialized.

When the cross-sectional area of the communication holes 10 is gradually increased to a large area from a very small area, the degree of mixing is increased gradually. Firstly, (a) the unmixed system is changed to the mixed system with respect to the dopant and to a condition of substantially the unmixed system with respect to the heat, and (b) the unmixed system is changed to the mixed system with respect to the two. In the range of the present invention, the lower limit to the cross-sectional area of the communication holes is the change point of (a) . In the steady state, practically there is no difference in dopant concentration between A and B at the point of this cross- sectional area (less than 1%) . The upper limit is the change point of (b) . At the point of this cross- sectional area, the temperature difference between A and B becomes less than 10 C.

The flow -from A to B is increased with increase in the feed rate of starting material. The mixing between the molten silicon in A and the molten silicon in B tends to become difficult.

The cross-sectional areas of the communication holes corresponding to the change points of (a) and (b) are increased. The suitable cross-sectional area is increased with increase in the feed rate of starting material. Fig. 3 shows the proper range of the communication holes (the range of the present invention) determined in consideration of the foregoing facts.

The lower limit of 30 g/min- for the starting material feed rate is determined from the standpoint of the prosecutive of a single crystal. While this feed rate corresponds to the pulling of a crystal of 5 inches in diameter at about lmm/min, to make the rate of crystallization less than ^' this free rate is not desirable from the productivity point of view. The upper limit of 130 g/min corresponds to the pulling of a crystal of 10 inches in diameter at a rate of 1.1 - m/min. The reason for determining this upper limit reside in that it is impossible to grow a crystal ^"at any rate of crystalization higher than that. The range of Fig. 3 can exist only in cases where there exists a metal heat keeping member for preventing -the occurrence of solidification at the partition member. While ceramics, carbons, metals, etc., can be conceived as materials for the heat keeping member, the objective of the present invention can be achieved by only metals which are high in heat keeping effect. The reason for the presence of the wide in proper range concerning the cross-sectional area of the communication holes resides in the presence of the heat keeping cover which is high in heat keeping effect. Since this heat keeping cover has the effect of maintaining the temperature in A much higher than the temperature in B, the mixing of the molten silicon between A and B is allowed to some extent. Recently, the magnetic-field applied CZ method has been attempted in many quarters. It is needless to say that the present invention holds good in cases where a magnetic field is applied. BRIEF DESCRIPTION OF DRAWINGS

Fig.l is a sectional view of an apparatus-, used in an embodiment ,

Fig. 2 is a graph showing the relation between the crystal lengthwise direction and the electric resistance value in the embodiment,

Fig. 3 is a graph showing the relation between the starting material feed rate and the total sum of the cross-sectional areas of the communication holes according to the present invention,

Fig. 4 is a schematic diagram for explaining the^' principle of the present invention,

Fig. 5 is a graph showing the relation between the dopant concentration ratio and the time in a conventional unmixed system, and

Fig. 6 is a graph showing the relation between the dopant concentration ratio and- the time in a mixed system according to the present invention.

In the drawings:

Numeral 1 designates a quartz crucible, 2 a graphite crucible, 3 an electric resistance heater, 4 a pedestal, 5 a silicon single crystal, 6 a furnace heat insulating member, 7 molten silicon, 8 a partition member, 9 a heat keeping cover, 10 communication holes, 12 the lower end of the heat keeping cover, 14 starting material feed means, 15 openings in the heat keeping cover, 16 a chamber upper cover, 20 a pull chamber, 21 a starting ^"material, 22 a subsidiary material, A a material melting section, and B a crystal growing section.

BEST MODE FOR CARRYING OUT THE INVENTION: Embodiment 1 : The present embodiment confirmed the effects of the present invention in cases where the amount of addition of the dopant (electric resistance value) was changed in the course of the pulling. Tests were made in such a manner that in the course of the pulling the resistance value was changed from 20 Ωcm to 10 Ωcm and further a change from 10 Ωcm to 5 Ωcm was effected. The kind of the dopant was phosphorus and the crystal diameter was 6 inches. The cross-sectional areas of the communication holes used for effecting the crystal growth were the following four levels.

2

(a) 100mm (8 communication holes of 4mm in diameter; within the range of the present invention) ,

2

(b) 226mm (8 communication holes of 6mm in diameter; within the range of the present invention) ,

2 (c) 942 mm (12 communication holes of 10mm in diameter, within the range of the present invention, and 2

(d) 3142 mm (10 communication holes of 20 mm in diameter; not within the range of the present invention) . The apparatus (Fig. 1) performing the present embodiment and the experimental conditions will now be described briefly. Numeral 1 designates a quartz crucible of 20 inches in diameter and it is set in a graphite crucible 2. The graphite crucible 2 is supported on a pedestal 4. The pedestal 4 is coupled to an electric motor on the outer side of the furnace and it serves to impart a rotational motion (10 rpm) to the graphite crucible 2. Numeral 7 designates molten silicon contained in the crucible 1. A silicon single crystal 5 of a cylindrical shape is pulled from the molten silicon 7 at a pull rate of 1.4 mm/min while being rotated (20 rpm) . Numeral 3 designates an electric resistance heater surrounding the graphite crucible. The pressure in the furnace (within a chamber 16) is 0.01 to 0.03 atmosphere. The foregoing are basically the same with a silicon single crystal manufacturing apparatus according to the ordinary Czochralski method.

Numeral 8 designates a partition member made from a high-purity silica glass and arranged within the crucible 1 to be concentric therewith. Its diameter is 40cm. The partition member 8 is formed with communication holes 10 that the molten starting material in a material melting ^' section flows into a single crystal growing section through the communication holes. In this embodiment, concerning the communication holes, the crystal growth -is effected by using the previously mentioned four- kinds of conditions. The lower edge portion of the partition member is preliminarily fused to the crucible 1 or fused to it by the heat produced when melting the silicon starting material.

At the start the setting conditions of molten silicon are as follows. The amount of molten silicon is 20 Kg in total, that is, 5 Kg in the material melting section and 15 Kg in the crystal growing section. The amount of addition of the dopant in each of the two regions is such that at the start the resistance value of the crystal becomes 20 Ωcm (4.5 ppba) . Even in the condition of (a) , due to the diffusion of the dopant through the communication hole dopant contents of the two regions are inevitably made equal to each other.

Numeral 14 designates starting material feed means having an opening above the material melting section and granular silicon starting material is supplied to the material melting section through the feed means. The feed rate is equal to the rate of crystallization, i.e. , 65 g/min. The starting material feed means 14 is connected to a starting material storage chamber (not shown) provided externally of the chamber upper cover 16, thereby feeding the starting material continuously. The feed rate of subsidiary material is calculated in accordance with the desired resistance value and the starting material feed rate. In accordance with the present embodiment, in the course of the pulling the feed rate of the subsidiary material is changed twice in order to change the resistance value as mentioned previously.

Numeral 9 designates a heat keeping cover made from a tantalum sheet of 0.2mm in thickness. This has the effect of reducing the dissipation of heat from the partition member 8 and the material melting section. In accordance with the present invention, the presence of the proper range of the cross-sectional areas of the communication holes is due to the presence of the partition member. Numeral 15 designates openings formed in the upper part of the heat keeping member 9.

2 The opening having an area of 100 cm is formed at each of four locations. Ar gas introduced into a pull chamber 20 is passed through the openings 15 first land then through the space between the resistance heater 3 and a furnace heat . insulating member 6 and it is discharged from the bottom of the furnace.

Due to the presence of these opening, the gas stream passing under the lower end 12 of the heat keeping cover 9 is reduced greatly. As a result, the occurrence of fine particles of SiO in the vicinity of the molten silicon surface is eliminated and thus the ■ growth of the single crystal is stabilized. However, the presence of the openings increase the dissipation of heat from the material melting section A. As a result, the use of the metal heat keeping cover 9 is made unavoidab1e.

Fig. 2 shows the changes of the resistance value in the lengthwise direction of the pulled crystal under the previously mentioned four kinds of conditions of the communication holes. (a) shows the case where the cross-sectional area of the communication holes is less than the lower limit of the present invention. Firstly, during the initial period of the pulling the resistance value is changed and also it becomes lower than the desired value. At some other intermediary points of 60 cm and 140 cm, respectively, the response characteristic with respect to the changes in the amount of the dopant is extremely bad. Practically, the desired resistance value is obtained at no point throughout the whole length. The reason is that in order that the steady state may be attained in this system, it is necessary to supply the starting material in an amount which is at least ten times the amount of the molten silicon in the .material melting section (^"5 Kg) , that is, about 108cm of the crystal in terms of the crystal length must be pulled. On the contrary, in the case of(b) and (c) which are within the.-- range of the present invention, the response characteristic with respect to the changes in the amount of the dopant is quite excellent. The transition regions of the resistance value are less than 10cm in terms of the crystal length. The CC-CZ method is well suited for the growth of long crystals involving no composition variation and also the application . of the present invention causes the CC-CZ method to become a crystal growing method suitable cope with small-lot production. In the case of (d) , the melting of the starting material is not satisfactory and the CC-CZ operation is not possible.

Embodiment 2 :

This embodiment confirmed the effect of the present invention concerning the materials of the heat keeping covers. With the tungsten sheet cover (the present invention) , the graphite cover (not of the invention) and the alumina cover (not of the invention) , crystals were pulled by the same apparatus as that of Embodiment 1. As regards the pulling conditions, they were the same as in the case of Embodiment 1 except that the crystal diameter was 8 inches, the pull rate was 0.8 mm/min (the _. starting material feed rate was 58g/min) and the cross-sectional area of the communication holes

2 was 314 mm (4 communication holes of 10 mm in diameter) .

Where the tungsten cover was used, the stable melting of the starting material and the stable crystal growth were possible and the same results as in the case of (b) and (c) in Embodiment were obtained with respect to the changes of the resistance value. In the case of the graphite cover and the alumina cover, just after the supply of the starting material was started, the occurrence of unmelted starting material was caused and thus the CC-CZ operation was no longer possible.

By working the present invention, any adjustment of the electric resistance value was simplified considerably in the CC-CZ method. As a result, to change the resistance value in the course of the pulling was also made possible while making the best use of the merit of the CC-CZ method that it was possible to manufacture very long silicon single crystals having extremely small variations of the resistance value in the growth direction. INDUSTRIAL APPLICABILITY:

As described hereinabove, the present invention is not only applicable as an apparatus for manufacturing silicon single crystals of 10 inches or more in diameter but also as an apparatus for manufacturing single crystal of any other material than silicon single crystals while maintaining the stable composition and quality.

Claims

1. A silicon single crystal manufacturing apparatus comprising a rotation-type quartz crucible containing molten silicon, an electric resistance heater for heating said quartz crucible from the side thereof, a quartz partition member for dividing said molten silicon into a single crystal growing section and a material melting section, said partition member having a plurality of communication holes for passing said molten silicon therethrough, a heat keeping cover for covering said partition member and said material melting section, starting material feed means for continuously feeding starting material silicon to said material melting section, and dopant feed means for feeding a dopant to said material melting section, wherein said heat keeping cover is made from a metallic material, and wherein a total sum of the cross-sectional areas of said communication holes formed in said partition member consists of a cross-sectional area having a value determined in a correlated relation with a feed rate of said starting material.

2. A silicon single crystal manufacturing apparatus as set forth in claim 1 , wherein a total sum A of the cross- sectional areas of said communication holes formed in

2 said partition member is not less than 80 mm and not

2 greater than 1600 mm when a feed rate of said starting material is between 30 and 130 g/min.

3. A silicon single crystal manufacturing apparatus as • set forth in claim 1, wherein a total sum A" of the cross- sectional areas of said communication holes formed in

2 said partition member is not less than 80 mm and not

2 greater than 1000 mm when a feed rate of said starting material is between 30 and 50 g/min, not less than 130 mm 2 and not greater than 1200 mm2 when said starting material feed rate is in the range from 50 to 80 g/min and not less than 220 mm 2 and not greater than 1600 mm2

when said starting material feed rate is in the range from 80 to 130 g/min.