JPH02293394A - Production of single-crystalline compound semiconductor - Google Patents

Abstract

PURPOSE: To improve quality by rotating a bulk crystal and crucible in the same direction and executing single crystal growth in the state that the rotating speed of the crucible is higher than the rotating speed of the bulk crystal by using a Czochralski method.

CONSTITUTION: The crucible 3 which rotates in the same direction as the direction of a pulling shaft 6 and at the speed higher than the speed of the shaft is disposed within a high-pressure vessel 2 of an inert gaseous atmosphere, such as Ar, of several tens atm. A two-dimensional compd., such as GaP, is melted by heating with a heater 4 in the crucible 3. A melt 5 added with a high concn. of In is sealed by a liquid sealant 8, such as B₂O₃, of a suitable thickness and a second crystal 7 fixed to the bottom end of the pulling up shaft 6 is contacted with the melt and thereafter, the pulling shaft 6 is slowly pulled in an arrow A direction while the shaft is kept rotated, by which the bulk crystal 1 is drown to the single crystal. The semiconductor single crystal of the compd. is obtainable even from three- and higher dimensional compds., such as Ga_1-xAl_xAs and In_xGa_1-xAs_1-yPy, by this process.

COPYRIGHT: (C)1990,JPO

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a method for manufacturing compound semiconductor single crystals. To be more precise, L adds a high concentration of impurities to the melt.
The present invention relates to a method for manufacturing a compound semiconductor medium crystal using an EC method.

Compound semiconductor single crystals, which are produced as raw materials for insulating substrate crystals used in conventional technology integrated circuits (ICs), have significantly higher dislocation mobility than silicon (Si> single crystals). Because of its high price, it has been attracting attention and being put into practical use as a high-speed, a!i product circuit (IC).Moreover, some compound semiconductor single crystals have a light-emitting function that S1 single crystals do not have. It holds promise as an important device in the fields of microcommunications, optical communication, and information processing.

For f'l compound semiconductor single crystals, the horizontal Bridgman method (horizontal Bridgman method)
It is difficult to apply the HB method), and single crystal growth is generally performed using the LEC method in a high-pressure container.

FIG. 4 is a diagram schematically showing one method for forming a single crystal for a compound semiconductor by the LEC method.

In FIG. 4, a crucible 10 made of quartz or nitrided nitride is placed in a high-pressure container 9, and a multiviscosity crystal material of a compound semiconductor is heated in the crucible 10 by a pot 11. Heating and melting it to form a melt 12,
Similarly, heated boron oxide (B203) also becomes a melt and serves as a liquid sealant 13 for the melt 12.

Subsequently, after pressurizing the inside of the high-pressure container 9 to a high B with an inert gas 14 such as argon (Ar), a seed crystal 1 is applied to the melt 12.
5 in contact with each other from above in the figure, and pull up the bulk crystal 17 in the direction of arrow D while rotating the upper shaft 16. However, by rotating the crucible in the opposite direction to the bulk crystal 17, heating to the crucible 10 is uniform. It is possible to form a convex solid-liquid interface 18 that is easy to form and is convenient for single crystal growth. Note that if the solid-liquid interface 18 is concave, twins and polycrystals will occur inside the bulk crystal 17, which is disadvantageous.

At this time, in the crystal growth of a compound semiconductor crystal by the LEC method, the number of dislocations as crystal defects (Fujitsu, dislocation density E
In order to reduce as much as possible (referred to as PD), as an impurity addition effect, compound semiconductor single crystals, such as semi-insulating G
Indium (In) or the like is added to the melt 12 when growing an aAs crystal, and Ga@ is added when growing a semi-insulating indium phosphide (InP) crystal.
It is known that it is possible to suppress the generation and diffusion of dislocations during crystal growth.

When creating a compound semiconductor single crystal using the LEC method in which impurities are added to the melt, it is necessary to reduce the lagoonality gradient in the direction of the crystal pulling axis. It is well known that this type of manufacturing method improves the efficiency of LEC.
In the method, abnormal cell growth due to compositional supercooling is likely to occur, so if the temperature gradient is lowered, cell growth due to compositional supercooling is more likely to occur.

Japanese Patent Application No. 61-105910 (Japanese Unexamined Patent Publication No. 62-260794) discloses a technique for eliminating abnormal cell growth caused by compositional supercooling.

Problems to be Solved by the Invention However, the method of pulling the single crystal while rotating the crucible 10 and the bulk crystal 17 in opposite directions in FIG. 10 and the bulk crystal 17 rotate in opposite directions, convection currents in opposite directions such as upper convection 12A2 and lower convection 12B occur in the melt 12 in the crucible 10. These upper convection 12A and lower convection 12B are the intermediate stagnation 1 shown by the dotted line.
Since there is no mixing with U with 2C as a field, the following 2
There are still several issues that need to be solved.

(1) When growing a compound semiconductor single crystal by adding impurities to the melt 12 using the LEC method, the impurity concentration Cs in the solid phase, that is, the bulk crystal 17, and the average impurity concentration in the liquid phase, that is, the melt 12, are CL. Then, k = Cs / C
L. This relationship is established.

This k is a quantity called the effective segregation coefficient, and from this relational expression, when k is smaller than 1, the impurity concentration is higher in the parts that grow later, and when k is larger than 1, the impurity concentration is higher in the parts that solidify first. It can be seen that the value becomes higher. Therefore, in FIG. 5, at the rotational speed used for conventional single crystal growth, the upper convection flow 12A and the lower convection flow 128A
Since these do not mix with each other using the intermediate retention zone 12G indicated by the dotted line as a boundary, the impurity concentration in the upper convection flow 12A is promoted to be much larger than the impurity concentration in the lower convection flow 12B.

(2) Due to the existence of the upper convection 12A, the temperature distribution in the upper convection 12A becomes uniform, and the temperature gradient becomes extremely small.

By the way, as is well known, in the crystal pulling process of a compound semiconductor by the LEC method in which impurities are added, the condition for compositional overcooling to occur is when the following inequality is satisfied.

The symbols in inequality a are as follows: △T: Temperature gradient near the solid-liquid boundary in the melt D: Diffusion coefficient of impurities in the melt m: Temperature gradient of the liquidus line CL: Impurities in the melt Concentration V: Crystal growth rate k: Represents each equation of the effective segregation coefficient. Therefore, as can be seen from the above, when inequality 0 is applied to the known technique of rotating the crucible 10 and the bulk crystal 17 in opposite directions, When the impurity density CL of convection 12A increases due to the above reason (1), the CL value on the right side of inequality a increases, and the value of Δ' on the left side of inequality a increases due to the reason of Shiki (2). decreases, the inequality (3) is very likely to hold true.In other words, the known technique of rotating the crucible 10 and the bulk crystal 17 in the opposite direction has the disadvantage of promoting compositional supercooling. I understand that.

This compositional supercooling W refers to a phenomenon in which abnormal cell growth occurs in crystal growth due to the temperature gradient Δ decreasing from when the assimilation rate 9 is still small. If there is abnormal cell growth, the bulk crystal 1
The availability rate of 7 is significantly reduced, which has a large impact on the yield when commercialized.

The above patent application No. 61-105910 (Japanese Patent Application No. 62-2607
94) deals with the occurrence of compositional supercooling by changing the temperature gradient ΔT〈g〉 in the melt when the solidification rate is Q to the impurity concentration OL in the melt, which changes with the progress of crystal growth. This paper discloses a technique that improves abnormal cell growth due to compositional supercooling by changing the temperature according to the change range of . However, in order to suppress the temperature gradient △' according to the amount of change in the impurity concentration CL, it takes two to three times as long to pull the crystal, and a measuring means that can perform precise continuous measurements is required. Since automatically controlled heating means are also required, the entire device becomes very complex.

Moreover, the upper convection 12A.
This is not a self-effective solution because it does not fundamentally eliminate the occurrence of the lower convection zone 12B and the intermediate convection zone 12C.

Furthermore, if the temperature gradient △T (q) is increased, the already formed single crystal part will be affected by the relatively large temperature gradient △T (o), and the dislocation density [
This results in an unfavorable result of increasing fEPD.

The present invention was made based on the above-mentioned circumstances.
To provide a method for manufacturing a compound semiconductor single crystal that can effectively prevent abnormal cell growth due to compositional supercooling in manufacturing the compound semiconductor single crystal by the IEC method in which a high concentration of impurities is added to the melt. With the goal.

Means for Solving the Problems In order to achieve the above-mentioned objects, the present invention has developed a compound semiconductor monomer using the liquid-enclosed chemical method (EC method), in which impurities are added at a high concentration into the melt. Crystal! lI
A method for growing single crystals, in which the bulk crystal and the crucible are rotated in the same direction, and the rotation speed of the crucible is set to be thicker (,) than the rotation speed of the bulk crystal. are doing.

In a preferred embodiment according to the invention, the compound semiconductor is then advantageously formed from a binary compound, such as gallium phosphide (GaP).

Also, compound semiconductors such as gallium aluminum? Indium/Arsenic (Ga1, A IxAs) and Indium/Gallium/Hiki/Phosphorus (In, Ga1, AS1■P
, ) is conveniently formed from a ternary or more multi-component compound.

Effect: In the method for manufacturing a compound semiconductor single crystal in which impurities are added to the melt configured as described above, rotating the bulk crystal and the crucible in the same direction may cause compositional supercooling. Prevents abnormal cell formation by suppressing convection and the formation of intermediate retention zones.

Further, by performing crystal growth while maintaining the crucible rotation speed higher than the bulk crystal rotation speed, a convex solid-liquid interface favorable for crystal growth is obtained.

Embodiments Hereinafter, preferred embodiments of the present invention will be described based on the drawings.

In FIG. 1, a bulk crystal 1 is pulled up from a melt 5 heated by a heater 4 in a crucible 3 rotatably disposed inside a high-pressure container 2 that can withstand pressures of 100 atmospheres or more. A crystal is grown by bringing the seed crystal 7 fixed to the lower end of the shaft 6 into contact and slowly pulling it up in the direction of arrow A while rotating the pulling shaft 6 (for example, 10 m+/h). It is.

In this example, the melt 5 is a gallium phosphide (GaP) melt, and a high concentration of indium (In) is added as an impurity (dopant) to suppress the generation and propagation of dislocations as crystal defects. are doing. In this example, boron oxide (B2 03
The upper surface of the melt 5 is sealed by forming a suitable thickness of the melt. 2A.

The crucible 3 is made of quartz, solid boron nitride, or the like. And this crucible 3 is
Inside the high-pressure container 2, the bulk crystal 1, that is, the pulling shaft 6, is rotated by the lower shaft 3A in the same rotational direction as the pulling shaft 6, and the rotational speed WC of the lower shaft 3A is set relative to the rotational speed WS of the pulling shaft 6. Ws<Wc (1)
The relationship is as follows. That is, the bulk crystal 1 is grown as a single crystal by rotating the bulk crystal 1 and the crucible 3 in the same direction h and maintaining the rotation speed Wc of the crucible 3 larger than the rotation speed WS of the bulk crystal 1. It will be done.

Regarding FIG. 2, L to which impurities related to the present invention are added
A method for producing a compound semiconductor crystal using the EC method includes rotating a bulk crystal 1 and a crucible 3 in the same direction, and rotating a bulk crystal 1 and a crucible 3 in the same direction.
The effect on the bulk crystal 1 and the melt 5 when the single crystal is grown while maintaining the rotational speed rfWc of the bulk crystal 1 to be higher than the rotational speed Ws of the bulk crystal 1 will be explained.

First, when the bulk crystal 1 and the crucible 3 are rotated in the same direction while maintaining the relationship ws<wc...(1), there is a part of the melt 5 that follows the crucible 3 at a large rotational speed Wc. As a result, a single convection current 5A is generated, which is indicated by a curved solid line arrow in the figure. This convection 5A flows from the outer edge of the bulk crystal 1 to the center as shown by the dotted arrow B, forming a convex liquid boundary 1A that is convenient for single crystal growth.

By the manufacturing method of the present invention in which the bulk crystal 1 and the crucible 3 are rotated in the same direction, the upper convection 12A and the intermediate retention zone 12C shown in FIG. 5 are not generated, so the convection 5A is a single convection. Therefore, the occurrence of compositional supercooling as described above can be effectively eliminated.

However, as shown in FIG. 3, when the rotation speed WS of the bulk crystal 1 is larger than the rotation speed aWc of the crucible 3, that is, W s > W c
In case (2), since the flow of the melt 5 is driven primarily by the bulk crystal 1, a strong flow occurs below the center of the bulk crystal 1, although it is also a single convection flow 5B.

Therefore, the liquid flows in the direction of the dotted arrow C from the center of the same liquid interface B to the outer edge. In the LEC method, since the bulk crystal 1 is the lowest temperature part, the melt 5 flowing toward the outer edge along the solid-liquid interface 1B flows into the single crystal first from the outer edge of the bottom surface of the bulk crystal 1. This results in a concave solid-liquid interface 1B that is inconvenient for single crystal growth.

When such a concave identical liquid interface 1B is formed, crystal growth progresses from the outer edge of the bulk crystal 1 to the center, so twins and polycrystals may be precipitated inside the bulk crystal 1. As a result, the availability of the bulk crystal 1 is likely to be significantly reduced.

Therefore, in the present invention, a convex solid-liquid boundary 1Δ (see Fig. 1 and Fig. 2), which is convenient for single crystal growth, is formed by maintaining the respective rotational speeds in the relationship expressed by equation (1). It is configured to do so.

Next, I will give an example of an experiment conducted to prove the effectiveness of the present invention.

Example gallium li/v (GaP)! When 1.78 t% of In was added as an impurity to the liquid and single crystal was produced by the production method of the present invention, the crystals shown in the table below were obtained. In the table, a-e are sample numbers.

Table 1 Solidification rate Q at the time when cell growth occurs (cell)
was measured and shown.

This experimental example shows that when the bulk crystal and the crucible rotate in opposite directions, as in samples ce, a convex solid field is formed regardless of the rotation speed, but these limbs c It can be seen that cell growth occurs at -e when the solidification rate q is very small.

Conversely, it can be seen that when the bulk crystal and the crucible in samples a and b rotate in the same direction h, cell growth occurs only after the solidification rate q becomes sufficiently high. However, if the rotation speed of the bulk crystal is higher, a concave solid-liquid interface will be formed and there is a risk that twins or polycrystals will precipitate inside the medium crystal, which is impractical, but the rotation speed of the crucible If is larger, a convex solid-liquid boundary will be formed, which is advantageous.

Effects of the Invention Since the present invention is configured as described above, it exhibits at least the following effects.

In the method for manufacturing a compound semiconductor single crystal according to the present invention,
By rotating the bulk crystal and the crucible in the same h direction, abnormal cell growth due to compositional supercooling is effectively suppressed, which improves the availability of medium crystals and the yield rate when processed into semiconductors. can be done. In addition, by growing a single crystal while keeping the rotational speed of the lupus higher than the rotational speed of the bulk crystal, a convex solid-liquid interface that is favorable for single crystal growth is formed, which improves the quality of the single crystal. can be improved.

Further, in the method for manufacturing a compound semiconductor single crystal according to the present invention, fl[J
It can also be applied to a wide range of applications.

[Brief explanation of drawings]

FIG. 1 is a vertical sectional view for explaining the manufacturing method of the present invention, FIG. 2 is a sectional view of essential parts for explaining the process of forming a convex solid-liquid interface according to the present invention, The figure is a cross-sectional view of the main part to explain the process of forming a concave solid-liquid interface, Figure 4 is a vertical cross-sectional view showing a known technique, and Figure 5 is a known technique in which compositional supercooling is performed. FIG. 2 is a cross-sectional view of a main part with half of the figure omitted to explain the process that occurs. 1...Bulk crystal, 1A...Convex solid-liquid boundary, 1
B... Concave solid-liquid interface, 2... High pressure container, 2A.
... Atmosphere, 3... Crucible, 3A... Lower shaft, 4... Heat, 5... Melt, 5A. 5B... Single convection, 6... Pulling shaft, 7... Seed crystal, 8... Liquid Moon Seed agent, Ws... Bulk crystal rotation speed, Wc... Luppo rotation speed .

Claims (3)

[Claims] (1) A method for manufacturing a compound semiconductor single crystal using the liquid-enclosed Czochralski method (LEC method) in which a high concentration of impurities is added to the melt, and the bulk crystal and the crucible are rotated in the same direction. Also, a method for producing a compound semiconductor single crystal, characterized in that single crystal growth is performed while maintaining a crucible rotation speed higher than a bulk crystal rotation speed. (2) The compound semiconductor is, for example, gallium phosphide (GaP)
2. The method for producing a compound semiconductor single crystal according to claim 1, wherein the compound semiconductor single crystal is formed from a binary compound such as. (3) Compound semiconductors include, for example, gallium-aluminum-arsenic (Ga_1_-_xAl_xAs) and indium-gallium-arsenic-phosphorus (In_xGa_1_-_
2. The method for manufacturing a compound semiconductor single crystal according to claim 1, wherein the compound semiconductor single crystal is formed from a ternary or more multi-component compound such as xAS_1_-_yP_y).