JPH0437038B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for accurately measuring the diameter of a single crystal in the production of a single crystal by the CZ method (Cyochralski method).

[Conventional technology]

The CZ method is well known as a method for manufacturing single crystals of silicon, etc. used in manufacturing ICs, LSIs, etc. This method, as shown in the principle diagram in Figure 4,
A crystalline melt 4 of silicon or the like contained in a crucible 2 is pulled up into a cylindrical shape while being rotated relative to the crucible 2 using a wire 6 or the like and solidified. The pulled cylindrical single crystal 8 is cut into an ingot with a predetermined diameter. Due to the need to reduce the cutting allowance during this cutting process and increase manufacturing yield, the diameter of the single crystal 8 is increased during pulling. Accurate measurement is required. Conventionally, this diameter measurement has been mainly performed by the following method.

A difference in curvature due to the meniscus occurs at the boundary between the growing part of the single crystal 8 during pulling, that is, the solidified part, and the crystal melt 4, and by changing the apparent radiance, a so-called fusion ring 9 appears. .
This fuselage ring 9 can be seen to have a semi-elliptical shape when observed from diagonally above, so by photographing it in the long axis direction XX with a one-dimensional line sensor 10 such as a CCD camera, it can be seen as shown in Fig. 5A. Intersection with fusion ring 9 x, x
An analog signal waveform having two peaks Pa and Pb is obtained corresponding to (FIG. 4). In conventional diameter measurement, this analog signal waveform is binarized as shown in Fig. 5 A and B, and furthermore, as shown in Fig. 5 B and C, the binarized signal first becomes 0 at the peak Pa.
The actual length between Na and Nb is calculated based on the number of pixels (Na - Nb) from pixel Na, which switches from 1 to 1, to pixel Nb, where the binarized signal finally switches from 1 to 0 at peak Pb. This was done by calculation. However, such a diameter measuring method causes a large measurement error for the following reasons.

[Problem that the invention seeks to solve]

Usually, the amount of light from the fusion ring 9 changes greatly during the pulling of the single crystal 8. When the light intensity of the fusion ring 9 changes, as shown in Fig. 6, not only the maximum values of peaks Pa and Pb change, but also the width of the tail changes, and the measured diameter value changes accordingly. It will also change. Specifically, even if the diameter of the single crystal 8 is the same, if the light intensity of the fusion ring 9 increases, the binarized output for Pa and Pb will widen, and the spacing between Na and Nb will increase. If the measured value is increased and the amount of light from the fusion ring 9 is decreased, the distance between Na and Nb becomes narrower and the measured value becomes smaller.

Further, such a measurement error also occurs due to a difference in light receiving sensitivity of the one-dimensional line sensor 10.

An object of the present invention is to provide a highly accurate diameter measurement method that can completely eliminate these measurement errors.

[Means for solving problems]

By the way, when the fusion ring 9 is photographed with a one-dimensional line sensor 10 such as a CCD camera,
Strictly speaking, the peaks Pa and Pb of the output waveform are the first
As shown in Figures A and B. Each consists of two peaks Pa', Pa'' and Pb'', Pb'. That is,
The fusion ring 9 consists of two parts: an outer bright ring represented by Pa' and Pb', and an inner bright ring represented by peaks Pa'' and Pb''. The present inventors focused on the fact that the fusion ring 9 consists of two inner and outer luminous rings, and as a result of various investigations and considerations regarding the positional relationship between these luminous rings and the growth part of the single crystal 8, the inner The bright rings of crystal melt 4 and single crystal 8
It was found that the outer luminous ring corresponds to the boundary with the solidification part, that is, the outer periphery of the single crystal 8 growth part, and is a reflection image of the edge of the crucible 2 and the like reflected on the surface of the crystal melt 4.

The present invention was made based on this knowledge, and when producing a single crystal by pulling it by the CZ method,
A single crystal characterized by optically measuring the diameter of the inner bright ring among two bright rings appearing as a fusion ring around the growth part of the single crystal being pulled as the diameter of the single crystal growth part. The gist is how to measure the diameter of.

Here, optical measurement refers to detecting the actual length between peaks Pa'' and Pb''.

[Effect]

Since the inner bright ring corresponds to the outer circumference of the single crystal 8 growth area, by detecting the actual length between the peaks Pa'' and Pb'' obtained for the inner bright ring, the diameter of the single crystal 8 growth area can be determined. It depends on precision.

In addition, although the heights of peaks Pa″ and Pb″ will change when the light intensity of the fusion ring or the sensitivity of the sensor changes, the actual length between Pa″ and Pb″ will not change, so they can be measured. The diameter is not affected by changes in the light amount of the fusion ring or differences in sensor sensitivity.

〔Example〕

FIG. 3 illustrates an actual single-crystal manufacturing apparatus incorporating diameter control.

In the figure, 1 is a chamber having a transparent window, and a crucible 2 is placed on a rotating support 3 inside the chamber. A heater 5 is installed around the crucible 2 to maintain the crystal melt 4 in the crucible 2 at an appropriate temperature.
is provided. A wire 6 inserted perpendicularly into the chamber 1 from the top of the chamber 1 has a seed 7 at its tip, and a single crystal 8 is grown by gradually pulling up the wire 6 from a state of being immersed in the crystal melt 4 while rotating it. .

The optical means for measuring the diameter of the single crystal 8 is such that a one-dimensional line sensor 10 such as a CCD camera passes through the window of the chamber 1 from diagonally above the chamber 1 and detects a fusion ring appearing in the growth area of the single crystal 8. (looks like a semi-ellipse) in the major axis direction
X (see Figure 4).
11 is a calculator that calculates the diameter of the growing portion of the single crystal 8 from this photographic data, and 12 is a controller that controls the pulling speed of the single crystal 8 so that the calculated diameter matches the target value.

In order to carry out the diameter measuring method of the present invention in such a single crystal manufacturing apparatus, first, as shown in the flow sheet of FIG. Convert and expand into memory space. An analog/digital converter for converting an analog signal into a digital signal is preferably 7 bits or more. Next, the features of the expanded digital signal waveform are extracted and the positions of peaks Pa'' and Pb'' are detected. Standard pattern recognition techniques are used for this. peak
When the positions of Pa″, Pb″ are detected, finally Pa″,
The length between Pa″ and Pb″ is measured as the diameter of the single crystal 8 growth part. This length measurement is performed by measuring the number of elements between Pa″ and Pb″ (Nb″- Na″）
This is possible by multiplying by the length of one element.
The above calculations are performed by the calculator 11.

In the case of the conventional method of calculating the diameter by binarizing the signal waveform of the fuselage ring 9 obtained by photographing with the one-dimensional line sensor 10 (Fig. 5), even if there is no change in the actual crystal diameter, the fuselage ring Changes in the measurement diameter occur due to changes in the light intensity of the fuselage ring and changes in the light receiving sensitivity of the sensor, but when the length between the peaks Pa'' and Pb'' is directly measured by pattern recognition as described above, the diameter of the fuselage ring changes. There is no effect from changes in light intensity or changes in sensor sensitivity, and the inner luminous ring represented by peaks Pa″ and Pb″ inherently coincides well with the outer circumference of the single crystal 8 growth area, so these together provide accurate results. This makes accurate diameter measurement possible.

When the diameter of the single crystal 8 growing portion is determined, the controller 12
This diameter is compared with a target value, and the pulling speed of the single crystal 8 is controlled so that the difference between the two becomes zero.

At this time, it goes without saying that if the measured diameter is accurate, the accuracy of the actual diameter will increase accordingly, and the target diameter can be made correspondingly smaller and the product yield can be improved.

According to the experience of the present inventors, when diameter measurement using conventional binarized waveform processing is adopted in the production of 6-inch class silicon single crystals, the finishing error due to measurement error must be expected to be ±2 mm. Accordingly, the target diameter (final product diameter +
5mm),
In the case of direct length measurement between peaks Pa″ and Pb″ mentioned above,
Finishing errors caused by measurement errors can be expected to be around ±0.5 mm, which makes it possible to reduce the target diameter to (final product diameter + 3 mm) and improve product yield by 4%.

〔Effect of the invention〕

As is clear from the above explanation, according to the diameter measurement method of the present invention, the diameter measurement accuracy is improved, which makes it possible to perform strict diameter control, so that the control target diameter can be brought closer to the final product diameter. This reduces cutting loss, which has a significant effect on improving product yield and lowering product costs.

[Brief explanation of drawings]

Figure 1 A to C are graphs schematically showing the processing procedure of the measurement method of the present invention, Figure 2 is a flow sheet showing the same processing procedure step by step, and Figure 3 is the measurement method of the present invention. A schematic diagram showing an example of a single crystal manufacturing apparatus using the CZ method. Figure 4 is a diagram of the principle of the CZ method. Figure 5
Figures A to C are graphs showing the processing procedure of the conventional measuring method, and Fig. 6 is a graph showing measurement error factors in the conventional method. In the figure, 2: crucible, 4: crystal melt, 8: single crystal, 9: fusion ring, 10: one-dimensional line sensor.

Claims (1)

[Claims] 1 When producing a single crystal by pulling it using the CZ method,
A single crystal characterized by optically measuring the diameter of the inner bright ring among two bright rings appearing as a fusion ring around the growing part of the single crystal being pulled as the diameter of the single crystal growing part. Diameter measurement method.