JPS62278190A - Production of single crystal - Google Patents

Abstract

PURPOSE:To keep the liquid level in a crucible in the pulling up of a single crystal using CZ process, by continuously measuring the weight of the single crystal and lifting a lower shaft at a rate calculated from the weight data according to a specific formula or by lowering the temperature distribution according to a calculated rate. CONSTITUTION:A seed crystal 12 attached to the lower end of an upper shaft 11 is immersed in a molten raw material 10 in a crucible 9 and the shaft is pulled up under rotation to enable the pulling-up of a single crystal 2. In the above process, the first differential coefficient dW/dt and the second differential coefficient d<2>W/dt<2> of the weight W of the single crystal are calculated by a computer 5 and the constants a, b and c of the formula are determined under the condition that the height of the surface of the molten raw material 10 in the crucible 9 is invariable when the lifting speed V of the lower shaft 6 is the speed calculated by the formula. The lifting speed of the lower shaft 6 is determined from the continuously measured data of the weight W of the crystal using the constants determined above and the lower shaft 6 is lifted at the speed V. As an alternative method, the height of the lower shaft 6 is maintained at a constant level and the temperature distribution is lowered at a speed V to obtain the objective single crystal.

Description

[Detailed description of the invention] 3. Detailed description of the invention (sword technical field)
o), Si, CdTe, etc. single crystal manufacturing method.

One of the methods for producing single crystals is the pulling method. This is the Czochralski method (CZ), the liquid capsule method (LE)
C) etc.

The pulling device consists of a lower shaft that supports the crucible, an upper shaft that hangs the seed crystal, a heater that heats the crucible, a pressure-resistant container that surrounds each mechanism, and a peephole installed on the wall of the container to observe the pulling condition. There are windows etc.

A thermocouple is provided to detect the temperature of the crucible and the temperature of the heater.

The upper shaft can be rotated up and down. At the top of the upper shaft is a rotating mechanism,
A lifting mechanism is provided. There is also a device for measuring load, such as a load cell, at the upper end of the upper shaft. This allows the weight of the crystal attached to the lower end of the upper axis to be monitored.

(b) Conventional technology The temperature distribution near the crucible is provided by a heater.

The position of the heater remains unchanged.

As the crystal grows, the raw material melt in the crucible decreases.

The height of the liquid level in the crucible decreases relative to the crucible. The temperature of the liquid surface must be constant. This is because the temperature at the interface between solid and liquid must be the melting point of the substance.

Therefore, in many cases, in order to compensate for the drop in the liquid level, the lower shaft is raised by an amount equal to the drop.

Since the crucible is fixed directly above the lower shaft or via a susceptor, when the lower shaft is raised, the liquid level also rises.

Alternatively, the lower shaft may not be raised, but the output of the heater may be varied to lower the temperature distribution as the liquid level falls.

Now, let's go back to the beginning and think about how to raise the lower axis in order to keep the liquid level unchanged.

What is being measured is the weight of the entire upper shaft.

This is the sum of the crystal weight and the upper axis weight. Since the upper axis weight is constant, it can be said that what is being measured by the load cell is the crystal weight.

The amount of increase in crystal weight per unit time is defined as dW/dt. If the weight measured by the load cell is M, the weight of the upper axis of L1, and the weight of the crystal is W, then L = M +
w (1) dL dW −= −(2) dt dt holds true. That is, an increase in crystal weight can be detected by a load measuring device such as a load cell.

The inner radius of the crucible is a, the height of the melt level in the crucible is H1, and the density of the melt is ρ. The rate of decrease in liquid level is -dH
/dt. Increase in crystal weight dW/dt
is expressed by .

Let V be the rising speed of the lower axis, and U be the rising speed of the upper axis.

The condition that the rising speed V of the lower axis is equal to the falling of the liquid level is, after all, as follows from (3) and (4).

When the desired crystal growth rate is q, the upper axis speed U is as follows: U = Q (6). This is because the liquid level does not move up and down.

Furthermore, when the lower shaft is not raised (V=O), the upper shaft speed U becomes (Q, -'/). Here, V is the value of equation (5).

(i) Problem to be solved If the rising speed of the lower shaft is determined by equation (5), the liquid level will be maintained at a constant position on average. Therefore, if the crystal growth is stable and dW/dt is constant, the method for determining equation (5) is fully satisfactory.

However, when applied to actual crystal growth, there are the following problems.

dW/dt is the weight increase rate immediately before that time.

(2) is the lower axis rising speed after that time. Since they are not the same over time, when dW/dt changes quickly, the liquid level will fluctuate even if V is determined as in equation (5).

When the change in dW/di is fast, there is a time delay in the measurement system and the drive system of the lower shaft, so the rising speed (2) may not be able to follow correctly.

Another problem is what could be called an offset. There are different problems than normal servo systems. In a normal servo system, position or speed is controlled, and a target value is set for the position or speed. Apart from the target value, there is a means for measuring the current position or speed. The object is then driven in a direction that reduces the difference between the current value and the target value.

In this way, [in a normal servo system, the target value is clearly given. This is a general assumption for any servo system.

However, this does not hold true for the upward movement of the lower shaft in the pulling method. The current value of the height and rate of rise of the lower axis can be clearly measured.

However, if you think about it carefully, the target value is not obvious. In equation (5), the value on the right side is the target value. This is obvious. However, its actual value is by no means obvious.

Although it is not easy to notice, there are always errors in the density ρ, crucible inner diameter R, and load cell precision W. If velocity is measured in its raw form, there can be no offset. However, the meaning of equation (5) is that the value on the right side is
The right side is not the speed itself. There is room for offset to occur here.

The effect of offset can be almost ignored at first.

However, offcents accumulate over time.

Therefore, as the crystal growth progresses, it has a large influence. This is a problem that is often overlooked, but it can sometimes cause serious difficulties.

For ρ, R%dW/dt, offset δρ, δR1
Suppose there is δ(dW/dt). In other words, for example, for H, the true value is R1, the known value (value used for calculation) is Ro
Suppose that there is a :5 relationship called R = : R + δR.

Then, instead of equation (5), exactly should be replaced by (2). (・). is a known quantity used in the calculation.

A value other than 1 in the parentheses of equation (7) represents the offset.

However, among these, the melt density ρ and the crucible inner diameter R are easy to measure quantities, so the error can be brought close to zero.

Regarding the weight L=M+W applied to the upper axis, even if the code cell is calibrated strictly, there will be an error, and this will often not be OK.

If the offset is written as β, there is a difference between a positive weight W and a reading W of a load cell, etc., as follows: W=(1+β) w (8). Then, the inside of the parentheses in equation (7) is approximately (1-
β), and there is an offset.

Consider the offset including all of these.

The offset is an unknown amount.

Although it is an unknown quantity, it is a constant.

Therefore, we will write the offset amount as F.

This is a constant with the dimension of speed.

The correct answer should be . However, I don't know the amount of offset F, and I don't know that there is an offset. Also, since I am not aware of the possibility of the existence of an offset, (5
) expression. ■ is the correct lower axis rising speed.

(9) (...) of equation. is a calculated value. In equation (5), (
) Although 0 is not added, this is also a calculated value.

It's the same thing.

The position of the crucible is determined by integrating V in equation (5).

In order for the liquid level to remain constant even if the crucible rises, (8
) must be increased by ■ in the equation. Since there is a certain difference between V and V, the height P of the liquid level is not constant.

If we write the offset of P as ΔP, we get Δp: (V-y)t This represents a slow fluctuation in the liquid level.

The previous one is a fast variation and this one is a slow variation.

As the liquid level changes, the crystal growth rate changes.

For example, when this method is applied to the production of GaAs single crystals for IC substrates, crystals with the expected low dislocation density cannot be obtained.

It may also cause cell growth to begin.

For this reason, single crystal growth cannot be performed with good yield.

Similar problems occur even when the upper shaft speed U is changed without raising the lower shaft.

B) Structure of the Invention The first problem is due to delays in the measurement system and drive system. Assume that the delay is represented by one delay time τ.

The left side of equation (5) is the current control value and can be written as v(t).

The right side of equation (5) is not the current value, but the past value by τ is set as the current value. Ideally current CdW/dt)
should be assigned the value of But I don't know this value. Since there is a delay of τ in the system, this value is the future value dW/dt after τ seconds.

The value of (t+τ) can be predicted by Taylor expansion. In other words, by using equation (11) as (dW/dt) instead of the right side of equation (5), the time delay problem can be overcome.

Next, regarding the issue of offset, in equation (9), it is sufficient to know offset (F) in advance, but this is not known. It shall have a portion proportional to the crystal weight W. Instead of equation (9). If the difference in height P of the liquid level is written as δP, it is calculated by instead of α0.

Assume that the pulling starts from 1=0 and ends at t=T.

Assume that it is desired to determine the difference δP in the liquid level height P to be O at a time during crystal pulling where O<t3<T. t=(crystal weight at 3 is W3, t='T
Let Wt be the crystal weight at .

get.

From equations (11) and (12), it can be seen that if the lower axis rising speed is calculated using equation (5) instead of equation (5), it is possible to suppress fluctuations in the liquid level height P in the crucible.

However, τ and γ are not directly measurable values.

Rather, it is more realistic to find τ and γ through repeated experiments under the condition that the liquid level height P remains unchanged.

In the right side of equation (16), the second and third terms are smaller values than the first term. The second term is clear from the fact that the time constant τ of the measurement system and drive system is smaller than the time constant of the change in the crystal growth rate dW/dt.

Therefore, in general, the lower axis rising speed is given as a function of the lifted crystal weight W by the following equation.

This is the method of the invention. However, a is 1/πρR
2 or a value close to this.

Depending on how b and c are selected, a can be slightly shifted from this value.

In the device described above, the lower shaft is raised to keep the liquid level constant.

As mentioned above, if the temperature distribution created by the heater can be lowered, the lower shaft is not raised (V=O) and the upper shaft speed U is set as follows. Q is the crystal growth rate.

The single crystal manufacturing method of the present invention is as described above.

A single crystal manufacturing apparatus for carrying out this method will be explained with reference to FIG.

This is a known device.

Inside the pressure vessel 1, there is a lower shaft 6 that can be rotated up and down, and an upper shaft 1.
1 are provided from below and above.

A crucible 9 is installed on the lower shaft 6. A single crystal 2 is attached to the lower end of the upper shaft 11.

A weight sensor 3 is provided at the upper end of the upper shaft 11 to measure the weight W of the crystal.

The weight sensor 3 detects the weight W as an analog quantity. The A/D converter 4 converts this into a digital signal. The value of weight W, which has become a digital value, is calculated by computer 5.
sent to.

A raw material melt 10 is contained in the crucible 9 . The raw material melt 1 is heated by the heater 8 provided around the crucible 9.
0 is heated.

A seed crystal attached to the lower end of the upper shaft 11 is dipped into the raw material melt 10. When this is pulled up while rotating, the single crystal 2 is pulled up following the seed crystal 12.

As the single crystal 2 rises, the raw material melt 10 decreases, so it is slowly raised while rotating the lower shaft. The pulling speed on the upper axis is approximately equal to the crystal growth speed.

This is a Czochralski furnace. When raising a compound semiconductor such as GaAs, a liquid capsule of B2O3 is placed in a crucible 9 and high pressure is applied with N2 gas.

Now, the computer 5 calculates the first differential dW/dt and the second differential d2 from the value of W input from the A/D converter.
Calculate s/dt2.

Then, according to equation (17), the rising speed V of the lower shaft or the rising speed U of the upper shaft is calculated (formula (18)).

A lower shaft drive unit 7 is provided at the lower end of the lower shaft 6, which rotates the lower shaft and raises the lower shaft according to the calculated value V.

(4) Effect Even if there is a delay in measuring the crystal growth rate dW/dt, according to the present invention, the liquid level in the crucible can be kept constant. Therefore, the growth rate dW/dt can be kept constant.

Furthermore, even if there is a steady error in the measurement of the crystal growth rate dW/dt, this influence can be removed and the constancy of the growth rate can be maintained.

Therefore, a single crystal with a low dislocation density can be pulled. It can also prevent cell growth.

(Example A A GaAs single crystal to which 3,000 wtppm of In was added was grown by the LEC method.

Since it is difficult to measure the position of the upper surface of the melt, the advantages of the present invention will be explained by the cell growth yield of the produced single crystal.

In was added to 6 kg of polycrystalline raw material GaAs to grow a GaAs single crystal with a diameter of 3 inches.

Crystal A was pulled using conventional methods.

According to the present invention, crystal B was controlled so that the liquid level was kept at a constant height while raising the lower axis.

Both crystals A and B weighed 5.0 kg.

'Both crystals A and B were sliced along the plane containing the pulling axis, and the crystals were photographed using an X-ray topography.

In crystal A, cell growth started at a solidification rate of 0.53.

For Crystal B, Principal Cell started at a solidification rate of 0.76. It can be seen that cell growth starts later in the case of the invention.

[Brief explanation of drawings]

FIG. 1 is a schematic vertical sectional view of a single crystal pulling apparatus for carrying out the single crystal manufacturing method of the present invention.

Claims (4)

[Claims] (1) A pressure vessel 1, an upper shaft 11 provided in the pressure vessel 1 from above and capable of rotating up and down, a lower shaft 6 provided in the pressure vessel 1 from below and capable of rotating up and down, and an upper shaft. a weight sensor 3 provided at the upper end of the lower shaft 6;
A crucible 9 that is attached to the upper end and should contain the raw material melt 10
, a heater 8 provided around the crucible 9 to heat the raw material melt 10, an A/D converter 4 that converts the crystal weight value W measured by the weight sensor 3 into a digital value, and a crystal weight value W. The lower shaft drive section 7 calculates the vertical speed of the lower shaft 6 or the upper shaft 11 by inputting Seed crystal 1 attached to the lower end of upper shaft 11 in liquid 9
2, and then pull up the single crystal by pulling it up while rotating, and calculate the decrease in the liquid level of the raw material melt in the crucible based on the weight value W of the single crystal measured by the weight sensor. In a single crystal manufacturing method in which the lower shaft 6 is raised so that the liquid level becomes a constant height, or the lower shaft 6 is not raised and the temperature distribution is lowered, the computer 5 calculates the crystal weight W.
Calculate the first differential dW/dt and second differential d^2W/dt^2, and the rising speed V of the lower axis 6 is V=a(dW)/dt+bW+c(d^2W)/dt^
2, the height P of the raw material melt surface in the crucible remains unchanged, determine the constants a, b, and c, and use these constants to calculate the continuous value of the crystal weight W. The unit is characterized in that the rising speed V of the lower shaft 6 is determined from the measured value and the lower shaft is raised at a rate of V, or the height of the lower shaft 6 is kept constant and the temperature distribution is lowered at a speed of V. Crystal manufacturing method. (2) The density of the melt is ρ, the inner diameter of the crucible is R, and a=1
/πρR^2
1) Single crystal production method described in section 1). (3) The method for producing a single crystal according to claim (1), characterized in that the lower shaft 6 is raised at a speed V, and the upper shaft 11 is raised at a speed equal to the growth rate Q. (4) Lower the temperature distribution at speed V without raising or lowering the lower shaft 6,
Claim (1) characterized in that the upper axis 11 is increased at a rate of growth rate Q minus V (Q-V).
Single crystal production method described in Section 1.