JP5924090B2 - Single crystal pulling method - Google Patents

Description

  The present invention relates to a method for pulling a single crystal by the Czochralski method (hereinafter referred to as CZ method), and more particularly, to a method for measuring a liquid surface position of a melt being pulled.

  One method for growing a single crystal as a semiconductor material is a CZ method. In the CZ method, a seed crystal suspended from a wire is slowly pulled up while being brought into contact with a melt contained in a crucible, and the melt is solidified to grow a columnar single crystal.

  The melt in the crucible gradually decreases as the pulling proceeds, but in order to keep the amount of melt heated by the heater constant, the position of the melt surface is at a predetermined position with respect to the heater and the heat shield structure. It is necessary to raise the crucible to become. This is because when the melt surface position with respect to the heater or the like is not a predetermined position, the temperature history of the grown single crystal changes, and crystal defects or the like are generated, making it impossible to manufacture a high-quality single crystal.

  For this reason, a method has been proposed in which the liquid surface position of the melt is optically measured during the pulling of the single crystal, and the rising amount of the crucible is calculated from the liquid surface position. In this method, the liquid surface position of the melt is obtained directly, so that there is little measurement error and the quality of the single crystal can be improved.

  For example, Patent Document 1 proposes a method of calculating the center position of a single crystal from the position of a fusion ring generated at the solid-liquid interface between the single crystal and the melt, and measuring the liquid surface position from this center position. . In this method, the first and second predetermined distances are separated from the fusion ring image detected using the image detection means, in the vertical direction before the seed crystal landing position in the necking step. The center of the single crystal along the vertical direction in the image is set from the two intervals between the intersections on both sides of each measurement line and the fusion ring, and the first and second predetermined distances. The position is calculated, and the liquid surface position of the melt is measured based on the center position. Since this method is not affected by the inclination of the liquid level, it is effective as a method for measuring the position of the melt. In particular, even when only a part of the fusion ring can be observed in the pulling process of the single crystal, the center position of the single crystal can be calculated with a small amount of calculation. The position can be measured.

Japanese Patent No. 4089500

  However, in the above-described conventional liquid surface position measurement method, the liquid surface position is accurately determined depending on the reference value of the brightness of the fusion ring used when detecting the intersections on both sides of the first and second measurement lines and the fusion ring. May not be measured. In other words, the fusion ring appears that the meniscus is brilliant due to various synchrotron radiation, but the change in the luminance distribution of the fusion ring is only a relative change in which the entire waveform of the luminance distribution shifts up and down during crystal growth. The waveform of the luminance distribution itself was not assumed to be deformed. However, it was confirmed that the waveform of the luminance distribution of the fusion ring changes due to changes in the amount of raw material charge and the heat shield structure. If an incorrect liquid surface position is calculated from such a fusion ring and the pulling up control is performed based on the liquid surface position information, the quality of the single crystal may be deteriorated.

  In order to solve the above-described problems, the single crystal pulling method according to the present invention picks up an image of a meniscus portion that forms a boundary between a single crystal pulled by the Czochralski method and a raw material melt contained in the crucible, A fusion ring appearing in the meniscus portion is detected based on the brightness distribution of the image, and a liquid level position of the melt is calculated based on the position of the fusion ring, and the melt level is calculated based on the calculated liquid level position. A single unit that controls the height position of the crucible containing the liquid and adjusts the distance between the lower end of the cylindrical heat shield disposed above the crucible so as to surround the single crystal and the liquid surface. In the crystal pulling method, in the first period in which light from a heater provided around the crucible is reflected on the meniscus portion, the fuse is used using a first reference value. The fusion is detected using a second reference value larger than the first reference value in a second period in which light from a heater provided around the crucible is detected and is not reflected on the meniscus portion. It is characterized by detecting a ring.

  According to the present invention, since the reference value used when detecting the fusion ring is changed from the start to the end of the pulling of the single crystal, the liquid level position of the melt is accurately determined in the first and second periods. It can be measured and controlled, and high quality single crystals can be produced.

  In the present invention, the first reference value set during the first period is set within a luminance range when light from the lower surface of the thermal shield is reflected on the meniscus portion, and the second reference value is set. The second reference value set during the period is preferably set within a luminance range when light from the side wall portion of the crucible protruding above the liquid level position is reflected on the meniscus portion.

  In the present invention, during the first period and the second period, the maximum value of the detected luminance when light from the side wall part of the crucible protruding above the liquid level position is reflected on the meniscus part. Preferably, the first reference value and the second reference value are set.

  In the present invention, the coefficient of the first reference value is preferably 0.6 or more and less than 0.8, and the coefficient of the second reference value is preferably 0.8 or more and 0.95 or less.

  In the present invention, a first measurement line orthogonal to the reference line passing through the center of the single crystal image and spaced from the center by a first distance is set, and orthogonal to the reference line, the center A second measurement line separated from the first measurement line by a second distance longer than the first distance to detect two intersections of the first measurement line and the fusion ring, and the second measurement line Detecting two intersections of the line and the fusion ring, a first distance between the two intersections on the first measurement line, a second interval between the two intersections on the second measurement line. Based on the interval, the first distance, and the second distance, the center position of the single crystal located on the reference line is calculated, and the liquid surface position of the melt is calculated from the center position of the single crystal. Preferably to calculate . According to this, even when only a part of the fusion ring can be observed during the pulling of the single crystal, the liquid level position of the melt can be measured with higher accuracy than in the conventional method.

  In the present invention, the first period or the second period may be determined from the length of the single crystal monitored during the pulling of the single crystal, and the monitoring is performed during the pulling of the single crystal. You may judge whether it is the said 1st period or the said 2nd period from the residual amount of the melt in a crucible. Furthermore, it may be determined whether the first period or the second period from the height direction position of the crucible monitored during the pulling of the single crystal. In any case, the first period and the second period can be easily determined.

  As described above, according to the method for pulling a single crystal of the present invention, the single crystal is removed from the fusion ring in the photographed image without being affected by the amount of raw material charge or the heat shielding structure between the start and end of pulling of the single crystal. The center position of the melt can be calculated with a small amount of calculation, and the liquid level position of the melt can be calculated from the center position of the single crystal.

  Further, according to the single crystal pulling apparatus of the present invention, the center of the single crystal from the fusion ring in the photographed image is not affected by the charge amount of the raw material or the heat shielding structure between the start and end of pulling of the single crystal. The position can be calculated with a small amount of calculation, and the liquid level position of the melt can be calculated from the center position of the single crystal.

FIG. 1 is a side sectional view schematically showing a configuration of a single crystal pulling apparatus according to a preferred embodiment of the present invention. FIG. 2 is a perspective view schematically showing the relationship between the center position of the single crystal and the liquid surface position of the melt. FIG. 3 is a schematic diagram for explaining a method of measuring the liquid surface position of the melt using the single crystal pulling apparatus having the configuration as shown in FIG. FIG. 4 is a schematic cross-sectional view showing the state of the crucible 12 in the single crystal pulling apparatus 10, where the left half shows a relatively large amount of melt residue and the right half shows a relatively small amount of melt residue. Is. FIG. 5 is a conceptual diagram for explaining the light source of the fusion ring. FIG. 6 is a graph showing the luminance distribution in the vicinity of the fusion ring in the photographed image, where the horizontal axis indicates the pixels in the X coordinate direction, and the vertical axis indicates the luminance (relative value). FIG. 7 is a flowchart for explaining a method of pulling a single crystal by the CZ method using the single crystal pulling apparatus 10. FIG. 8 is a graph showing the measurement result of the liquid surface position of the melt in Example 1. FIG. 9 is a graph showing the measurement results of the liquid surface position of the melt in Example 2.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

  FIG. 1 is a side sectional view schematically showing a configuration of a single crystal pulling apparatus according to a preferred embodiment of the present invention.

  As shown in FIG. 1, the single crystal pulling apparatus 10 includes a chamber 11, a crucible 12 that supports the melt 1 in the chamber 11, a heater 15 provided on the outer peripheral side of the crucible 12, a heater 15, and a crucible 12. The heat shield 16 for preventing the single crystal 2 from being heated by the radiant heat from the heat and suppressing the temperature fluctuation of the melt 1, the CCD camera 18 for photographing the liquid surface of the melt 1, and each component are controlled. And a control unit 30.

  The crucible 12 includes a quartz crucible 13 and a graphite susceptor 14 that supports the quartz crucible 13, and the melt 1, which is a single crystal raw material, is accommodated in the quartz crucible 13. The diameter of the quartz crucible 13 is, for example, 800 mm, and a very large crucible is used. A cylindrical heat insulating material 20 is provided outside the heater 15. The crucible 12 passes through the center of the bottom of the chamber 11 and is fixed to the upper end of a shaft 21 provided in the vertical direction. The shaft 21 is driven up and down and rotated by a shaft drive mechanism 22. The shaft drive mechanism 22 operates according to a command from the control unit 30.

  Above the crucible 12, a seed chuck 23 that holds the seed crystal, a wire 24 that suspends the seed chuck 23, and a wire winding mechanism 25 that winds the wire 24 are provided. The wire winding mechanism 25 also has a function of rotating the wire. At the time of pulling up the single crystal, the seed crystal is immersed in the melt 1 and gradually pulled up while rotating the crucible 12 and the seed crystal in opposite directions. The wire winding mechanism 25 operates according to a command from the control unit 30.

  The heat shield 16 is a cylindrical member disposed above the crucible 12 and has an inverted frustoconical shape whose diameter is reduced from above to below. The downward inclination angle of the heat shield 16 may be gentle, for example, 10 to 45 ° with respect to the horizontal plane. The heat shield 16 is disposed inside the crucible 12, and when the crucible 12 is raised, the side wall portion of the crucible 12 is located outside the heat shield 16. There is no interference with. As the material of the heat shield 16, graphite can be used.

  The heat shield 16 also functions as a gas rectifying member that rectifies the gas flow in the vicinity of the surface of the melt 1. By appropriately adjusting the position (rising speed) of the crucible 12 in accordance with the growth of the single crystal 2, the distance (gap ΔG) from the melt surface to the lower end of the heat shield 16 can be controlled. The flow rate of the gas flowing in the vicinity of the surface (purge gas guide path) can be controlled to be constant. Therefore, the evaporation amount of the dopant from the melt can be controlled, and the stability of the resistivity distribution in the pulling direction of the single crystal can be improved.

  A viewing window 11a for observing the liquid level of the melt 1 is provided in the upper part of the chamber 11, and the CCD camera 18 is installed outside the viewing window 11a. The CCD camera 18 is an imaging device that captures an image of the fusion ring, and images the liquid level of the single crystal 2 and the melt 1 in the crucible 12 that can be seen through the heat shield 16 from the viewing window 11a. The image of the CCD camera 18 is preferably a gray scale, but may be a color image. The CCD camera 18 is connected to the control unit 30, and the captured two-dimensional image data is input to the control unit 30 and used for controlling the liquid surface position.

  The CCD camera 18 is preferably a two-dimensional CCD camera, but may be a one-dimensional CCD camera. In this case, a two-dimensional image can be obtained by a method in which the one-dimensional CCD camera is mechanically moved in the horizontal direction or a method in which the measurement angle of the one-dimensional CCD camera is changed and a fusion ring image is scanned. Further, both the measurement of the liquid surface position and the measurement of the diameter of the single crystal may be performed using the CCD camera 18.

  The image obtained by the CCD camera 18 is distorted because it is observed from obliquely above the single crystal pulling apparatus 10. This distortion can be corrected using a theoretical formula calculated from geometric optics. Moreover, it can also correct | amend by using the correction table created beforehand using the reference | standard board which wrote the scale. In this correction table, conversion coefficients indicating distances per pixel in an image are obtained in the vertical direction and the horizontal direction, respectively.

  The liquid surface position of the melt 1 has two meanings. One is the liquid surface position in the crucible 12 (particularly in the quartz crucible 13), which is mainly consumed by the pulling of the single crystal. However, it may also change when the volume of the crucible 12 changes due to the deformation of the crucible 12. The other is the liquid level position (gap ΔG) with respect to the fixed equipment such as the heater 15 and the heat shield 16, which is in addition to the change in the liquid level position in the crucible 12 described above, in the vertical direction of the crucible 12. It changes with the movement of the position (height). In the present specification, the “liquid level position” refers to the liquid level position with respect to the fixing equipment, particularly the heat shield 16 unless otherwise specified.

  FIG. 2 is a perspective view schematically showing the relationship between the center position of the single crystal and the liquid surface position of the melt.

As shown in FIG. 2, the control unit 30 calculates the position of the center C ₀ of the single crystal 2 from the fusion ring 4 generated at the solid-liquid interface between the single crystal 2 and the melt, and from the position of the center C ₀ . The position of the liquid surface 3 of the melt is calculated. The two centers C ₀ of the single crystal are intersections between the pulling axis 5 of the single crystal 2 and the liquid surface 3 of the melt. If the CCD camera 18 and the heat shield 16 are accurately installed at a predetermined position in the design at a predetermined angle, the position of the liquid surface 3 is geometrically optically determined from the position of the fusion ring 4 appearing in the image. Can be calculated.

  FIG. 3 is a schematic diagram for explaining a method of measuring the liquid surface position of the melt 1 using the single crystal pulling apparatus 10 shown in FIG.

As shown in FIG. 3, first, two measurement lines L ₁ and L ₂ are set in a two-dimensional image photographed by the CCD camera 18. The measurement lines L ₁ and L ₂ are preferably straight lines that are orthogonal to the reference line L ₀ passing through the center C _{0 of the} single crystal and are parallel to each other along the X coordinate direction of the two-dimensional image.

Here, if the measurement line L ₁ should be set near the center C ₀ of the single crystal, the fusion ring 4 when the diameter of the single crystal is reduced is in the shadow of a single crystal, can not be detected the center C _0. Therefore, the measurement line L ₁ is the outer circumferential side from the center C _0, that is, in the image set to the lower position. Further, the measurement line L _2, the surface side of the measurement line L _1, that is, in the image is set to the lower side. When the measurement lines L ₁ and L ₂ are set, since the position of the center C ₀ of the single crystal to be used as a reference is unknown, the landing position of the seed crystal in the necking step is set to the center C ₀ of the single crystal. Use as a position.

Next, the intersections on both sides of the measurement lines L ₁ and L ₂ and the fusion ring 4, that is, the intersection points C ₁ and C ₁ ′ and the intersection points C ₂ and C ₂ ′ are detected. For the detection of these intersections, a predicted value (reference value) of the brightness of the fusion ring is used. As this reference value, a value obtained by multiplying the maximum luminance in the captured image by a predetermined coefficient is used. The reference value needs to be an appropriate value that can correctly identify the fusion ring, and is preferably changed according to the pull-up situation. The setting of the reference value will be described later.

  The maximum brightness in the captured image may be targeted for one pixel having the maximum brightness alone, or in order to prevent the influence of noise, a plurality of pixels having the maximum brightness or a brightness close to it are consecutive. May be targeted. As described later, when the light from the side wall of the crucible protruding above the liquid level is reflected on the meniscus, the maximum value of the detected luminance tends to be continuous over a plurality of pixels. The influence of noise can be prevented by setting the distribution as the target of the maximum luminance.

The interval between the intersection points C ₁ and C ₁ ′ is W ₁ , the interval between the intersection points C ₂ and C ₂ ′ is W ₂ , the Y-coordinate position of the center C ₀ of the single crystal is Y ₀ , and the measurement line L ₁ If the position of the Y coordinate of Y ₁ is Y ₁ , the position of the Y coordinate of the measurement line L ₂ is Y _2, and the radius of the fusion ring is R, the relationship of Expressions (1) and (2) is obtained.

^{_{R 2 = (W 1/2}} ) 2 + (Y 0 -Y 1) 2 ··· (1)

^{_{R 2 = (W 2/2}} ) 2 + (Y 0 -Y 2) 2 ··· (2)

From the relationship of Equation 1 and Equation 2, the Y coordinate position Y ₀ of the center C ₀ of the single crystal in the Y direction in the two-dimensional image is expressed by Equation (3).

Y ₀ = {Y ₁ + Y ₂ + (W ₁ ² −W ₂ ² ) / 4 (Y ₁ −Y ₂ )} / 2 (3)

Further, the liquid surface position of the melt 1 is obtained from the Y coordinate position Y ₀ of the center C ₀ of the single crystal. The liquid surface position of the melt 1 is determined by using a linear regression line (calibration straight line) indicating the relationship between the distance from the melt surface to the lower end of the heat shield 16 (gap ΔG) and the Y coordinate in the image. It can be obtained by converting ₀ into a gap ΔG.

In the present embodiment, two or more (for example, 10) combinations of the two measurement lines L ₁ and L ₂ are set, and a value obtained by averaging the center positions of the single crystals corresponding to each combination is determined as a single crystal. It is preferable to set the center C ₀ position. When the cross-sectional shape of the single crystal 2 is a perfect circle, the measurement error is very small. However, depending on the pulling conditions, the single crystal 2 may be deformed so that it is not a perfect circle, and the measurement error may increase. In addition, when a single crystal crystal wall line appears on the measurement line, a measurement error increases in that portion. Such measurement errors can be reduced in influence by using an average value of a plurality of measurement values.

  Although the position of the crucible 12 in the height direction changes according to the growth of the single crystal 2 (consumption of the melt 1), the initial charge amount of the raw material of the single crystal is increased to increase the single crystal obtained by one pulling. Then, the amount of change in the height direction position of the crucible 12 also increases. In addition, it is necessary to consider that the heat shield 16 does not interfere with the crucible 12 or the melt 1. Due to such a change in the pulling conditions, the conditions of the photographed image as shown below are observed.

  FIG. 4 is a schematic cross-sectional view showing the state of the crucible 12 in the single crystal pulling apparatus 10, where the left half shows a relatively large amount of melt residue and the right half shows a relatively small amount of melt residue. Is.

  As shown in FIG. 4, the liquid level position is always controlled at a fixed position regardless of the melt amount. That is, the position of the liquid surface in the crucible 12 is lowered by the growth of the single crystal, so that the crucible 12 is raised and the position of the liquid surface is constant with respect to the heater 15 and the heat shield 16 (the gap ΔG is constant). It is controlled to become. In the figure, illustration of a single crystal is omitted.

  Immediately after the start of pulling with a large amount of melt, the protruding amount of the side wall portion of the crucible 12 protruding upward from the liquid surface position is small, and the height position of the upper end of the side wall portion of the crucible 12 is lower than the heater 15 position. Accordingly, a gap is formed between the upper end of the side wall portion of the crucible 12 (strictly, the graphite susceptor 14) and the heat shield 16, and light (radiant light and radiation) from the heater 15 passes through this gap as indicated by an arrow r. Radiation) enters the meniscus portion.

  On the other hand, as the growth of the single crystal progresses and the amount of melt decreases, the protruding amount of the side wall portion of the crucible 12 protruding upward from the liquid surface position increases, and the height of the upper end of the side wall portion of the crucible 12 relative to the position of the heater 15 increases. Since the position becomes higher, there is no gap between the upper end of the side wall portion of the crucible 12 and the heat shield 16, and light (radiated light and radiated light) from the heater 15 does not enter the meniscus portion.

  FIG. 5 is a conceptual diagram for explaining the light source of the fusion ring.

  As shown in FIG. 5, the fusion ring is a portion where various light incident on the meniscus 3 m is reflected and shines in a ring shape. The incident light source mainly includes light E <b> 1 from the heater 15 and from the crucible 12. The light E2 and the light E3 from the lower end of the heat shield 16 are considered. Immediately after the start of lifting, the crucible 12, the heater 15, and the heat shield 16 are in the positional relationship shown on the left side of FIG. 4, and all of the lights E1, E2, and E3 are reflected on the meniscus 3m. On the other hand, when the growth of the single crystal proceeds and the crucible 12, the heater 15, and the heat shield 16 have the positional relationship as shown on the right side of FIG. 4, only the light E2 and E3 are reflected on the meniscus 3m. Since the incident positions of these lights E1, E2, and E3 are slightly shifted in the meniscus, a luminance distribution corresponding to the light source appears in the captured image.

  FIG. 6 is a graph showing the luminance distribution in the vicinity of the fusion ring in the photographed image, where the horizontal axis represents pixels in the X coordinate direction in the image, and the vertical axis represents luminance (relative value).

  When the melt residue is as small as 210 kg (right side in FIG. 4), two small peaks appear at the top of one large peak, as shown by the waveform A in FIG. These two peaks are considered to be caused by the lights E2 and E3 in FIG.

  On the other hand, when the melt residue is as large as 290 kg (left side in FIG. 4), three small peaks appear at the top of one large peak as shown by the waveform B in FIG. One has been added. This third peak has higher brightness than the other two peaks in the initial stage of pulling up the single crystal, and gradually decreases as the growth of the single crystal proceeds. It was confirmed that the luminance of the two peaks was almost the same and disappeared at 220 kg. As shown on the left side of FIG. 4, this phenomenon can be presumed to be due to the radiation of the heater 15 being applied to the liquid surface of the melt 1 without being blocked by the crucible 12. That is, the third peak is considered to be due to the light E1 in FIG.

  Thus, the presence or absence of radiation from the heater 15 affects the imaging error of the fusion ring representing the meniscus. The luminance distribution of the fusion ring in the image affected by the radiation of the heater 15 is unstable, and the liquid level position cannot be measured accurately. If control is performed based on an erroneous measurement value of the gap, the temperature gradient of the solid-liquid interface may change, and the quality of the single crystal may deteriorate.

Therefore, in the present invention, when the liquid level is affected by the radiation from the heater 15, the level of the reference value used for detecting the intersection between the fusion ring and the two measurement lines L ₁ and L ₂ is lowered. (Reference value line TH _{1 in} FIG. 6) When the radiation from the heater 15 is not received, the level of the reference value is increased (reference value line TH _{2 in} FIG. 6) to suppress the influence of noise as much as possible. The fusion ring can be specified more accurately. And the liquid level position of a melt is accurately measured using such a fusion ring image. Hereinafter, a method for pulling a single crystal will be described in detail.

  FIG. 7 is a flowchart for explaining a method of pulling a single crystal by the CZ method using the single crystal pulling apparatus 10.

  As shown in FIG. 7, in pulling up the single crystal, first, a large amount of raw material is filled in the crucible 12, and the crucible 12 is placed in the chamber 11 (step S11). Next, after making the inside of the chamber 11 into an Ar gas atmosphere under reduced pressure, the raw material in the crucible 12 is heated and melted by the heater 15 (step S12). At this time, the seed crystal mounted on the tip of the wire 24 is at a position sufficiently higher than the crucible 12 and is separated from the raw material being melted.

  Next, after adjusting the temperature until the melt 1 is stabilized, the height of the crucible 12 is adjusted to set the initial liquid surface position of the melt (step S13). Although not particularly limited, the initial liquid surface position may be set automatically, or may be performed by the operator raising and lowering the crucible while observing the liquid surface of the melt.

Then, as the coefficient of the reference values used to detect the fusion ring as an initial setting, and it sets the first coefficient K ₁ (step S14). The first coefficient K ₁ is smaller than a second coefficient K ₂ described later, and is preferably set to 0.6 or more and less than 0.8, and particularly preferably set to 0.75. This is because if the reference value coefficient K ₁ is 0.8 or more, an incorrect liquid level position may be calculated by the fusion ring affected by the light from the heater, and the reference value coefficient K ₁ is 0. If it is less than .6, the detection accuracy of the fusion ring is lowered and the error of the liquid surface position becomes large.

  Next, the pulling of the single crystal is started (step S15). In the pulling of the single crystal by the CZ method, the single crystal is grown on the lower end of the seed crystal by slowly pulling the seed crystal while rotating the shaft 21 and the wire 24 in opposite directions.

  In the growth of a single crystal, seed drawing (necking) is first performed by a dash method in order to make the single crystal dislocation-free. Next, in order to obtain a single crystal having a required diameter, a shoulder portion is grown, and when the single crystal has a desired diameter, the body portion is grown with a constant diameter. After growing the body part to a predetermined length, tail drawing (formation of the tail part) is performed in order to separate the single crystal from the melt without dislocation.

  In necking, in order to eliminate dislocations originally contained in the seed crystal and slip dislocations generated in the seed crystal due to thermal shock at the time of landing, the minimum diameter of the seed crystal is slowly pulled upward while rotating the seed crystal relatively. Narrow it down until it becomes about 3-5mm. When the length of the neck portion becomes 10 to 20 mm and the slip dislocation is completely removed, the neck crystal diameter is enlarged by adjusting the pulling speed of the seed crystal and the temperature of the melt 1, and the shoulder portion is grown.

  When the shoulder portion reaches a predetermined diameter, the body portion is now brought up. The diameter of the body portion needs to be constant in order to increase the wafer yield. During the growth of the single crystal, the output of the heater 15, the pulling speed, and the crucible so that the body portion is grown while maintaining a substantially constant diameter. The ascending speed of 12 is controlled. In particular, as the single crystal grows, the melt 1 is reduced and the liquid level is lowered, so that the crucible 12 is raised as the liquid level is lowered.

  During the pulling of the single crystal, in order to control the position of the liquid level, an image of the liquid level is taken with the CCD camera 18, two measurement lines are set in the image, and the luminance distribution on the measurement line is used as a reference value. By comparing and determining the coincidence / non-coincidence, the intersection of the fusion ring appearing in the meniscus portion and the two measurement lines is detected (step S16). Next, the position of the center of the single crystal is calculated from the position of the intersection of the fusion ring and the two measurement lines in the image, and the liquid level position (gap value) is calculated from the position of the center (step S17). As described above, since the melt surface image is distorted, the image is corrected using a linear conversion formula (or conversion table). The control unit 30 controls the height of the crucible 12 by controlling the shaft drive mechanism 22 based on the calculated liquid level position, and controls the liquid level position so as not to deviate from the initial liquid level position.

During the pulling, the height position of the crucible 12 is monitored whether higher than a predetermined position, if it becomes higher than a predetermined position, the coefficient of the reference value from the first coefficient K ₁ It is changed to the second coefficient _{K 2} (step S19, S20). The predetermined position referred to here is a position where the crucible rises, the gap between the upper end of the side wall of the crucible and the heat shield disappears, and the radiant light of the heater does not directly enter the liquid surface.

  The timing of changing the reference value coefficient can be determined by indirectly obtaining the height position of the crucible from the length of the single crystal that is constantly monitored during the pulling of the single crystal. It is also possible to monitor the melt amount and indirectly determine the height position of the crucible from the melt amount. Of course, the height position of the crucible may be directly obtained and determined.

Coefficient _{K 2} of the second is the first value larger than the coefficient _{K 1,} preferably set to 0.8 to 0.95, particularly preferably set to 0.9. This is because the gap ΔG from the melt surface to the lower end of the heat shield 16 may be changed in the latter half of the pulling of the single crystal, and in this case, a higher reference value enables more accurate measurement. In addition, if the reference value is too high, the fusion ring cannot be extracted.

  When the body part reaches a predetermined length, the process proceeds to the formation of the tail part. In order to prevent the thermal balance between the melt 1 and the single crystal 2 existing at the crystal growth interface from being disrupted and sudden thermal shock applied to the crystal to cause quality abnormalities such as slip dislocation and abnormal oxygen precipitation, The diameter is gradually reduced to form a conical tail portion, the single crystal from the melt 1 is cut off, and the single crystal pulling is completed (step S21). Thereafter, the single crystal ingot separated from the melt 1 is cooled under predetermined conditions, and the wafers cut from the single crystal ingot are used as substrate materials for various semiconductor devices.

  As described above, according to the present embodiment, the liquid surface position of the melt can be accurately controlled in the single crystal pulling by the CZ method, and the production yield of the single crystal can be increased.

  The present invention is not limited to the above embodiments, and various modifications can be made without departing from the spirit of the present invention, and it goes without saying that these are also included in the scope of the present invention. Yes.

For example, in the above embodiment, two measurement lines L ₁ and L ₂ are set in an image showing a fusion ring photographed by the CCD camera 18, and the center of the single crystal is determined from the position of the intersection of the fusion ring and the measurement line. The position is calculated, and the liquid level position is further calculated from the position of the center C ₀ of the single crystal. However, the present invention is not limited to such a method, and the liquid level position is calculated using an image of a fusion ring. The present invention can be applied to various methods for calculating.

  In the above embodiment, the reference value is changed depending on whether it is affected by the radiation from the heater. However, the reference value can be changed in accordance with changes in various other conditions. .

  Hereinafter, although the Example of this invention is described, this invention is not limited to this Example at all.

Example 1
A single crystal pulling apparatus 10 having the configuration shown in FIG. 1 was used to pull a silicon single crystal having a diameter of 300 mmφ. A quartz crucible having a diameter of 800 mm was used for pulling, and 330 kg of polycrystalline silicon was loaded therein.

During pulling, a two-dimensional image of the fusion ring was taken with a CCD camera, and the liquid level position was calculated using this. In the calculation of the liquid surface position, first, two measurement lines L ₁ and L ₂ were set in the two-dimensional image data. Among these, the measurement line L ₁ close to the center C ₀ side of the single crystal was set at a position 150 pixels away from the surface C, ie, below the seed crystal landing position, as viewed from the center C _{0 of the} single crystal. . On the other hand, the measurement line L ₂ of the other one, the surface side of the measurement line L _1, that is, in the image is set to a position apart 150 pixels on the lower side.

Next, four intersections C ₁ , C ₁ ′, C ₂ , C ₂ ′ of the fusion ring and the measurement lines L ₁ and L ₂ are obtained using the reference values, and the center C ₀ of the single crystal is determined from these intersections. The position was determined. The reference value was a value obtained by multiplying the maximum luminance in the image by a coefficient of 0.9, and the same value was used from the start to the end of the pulling.

The position of the center C ₀ of the single crystal was calculated using Equation 3. Here, ten combinations of _two measurement lines L ₁ and L ₂ are set at a pitch of 10 pixels, the width between the first and second measurement lines L ₁ and L _{2 is} 150 pixels, and the first measurement line. the pixel coordinates L ₁ is set in the range of 833 to 743, and the average value of the center position of the corresponding single crystal each combination to the center position of the final single crystal. Further, the liquid surface position of the melt was determined from the final center position of the single crystal.

FIG. 8 is a graph showing the measurement result of the liquid surface position of the melt, where the horizontal axis indicates the number of measurements (10-second pitch), and the vertical axis indicates the amount of change (mm) in the liquid surface position. In addition, in order to make each data tendency easy to discriminate, the liquid level position when L ₁ = 833 mm was set as a reference value (change amount = 0).

  As apparent from FIG. 8, in the first period from the start of pulling up until the remaining amount of the melt reaches 220 kg, a variation of about 7 mm occurred in the liquid surface position. However, in the second period when the remaining amount was 220 kg or less, the variation in the liquid level position was about 0.5 mm.

(Example 2)
The reference value coefficient in the first period from the start of pulling up until the remaining amount of the melt reaches 220 kg is 0.75, and the reference value coefficient in the second period in which the remaining amount is 220 kg or less is 0. The silicon single crystal was pulled up under the same conditions as in Example 1 except for the point set to 9. FIG. 9 shows the measurement result of the liquid surface position of the melt.

  As is clear from FIG. 9, the variation in the liquid level position was 1 mm in the entire period from the start to the end of the pulling. That is, in the first period in which the light from the heater provided around the crucible is reflected on the meniscus portion, the coefficient of the reference value should be set to 0.75, and the light from the heater is not reflected on the meniscus portion. In the period of 2, it became clear that the reference value coefficient should be 0.9.

DESCRIPTION OF SYMBOLS 1 Melt 2 Single crystal 3 Liquid surface 3m of meniscus part 4 Fusion ring 10 Single crystal pulling apparatus 11 Chamber 11a Viewing window 12 Crucible 13 Quartz crucible 14 Graphite susceptor 15 Heater 16 Thermal shield 18 CCD camera 20 Heat insulating material 21 Shaft 22-shaft drive mechanism 23 seed chuck 24 wire 25 wire wind-up mechanism 30 control unit _{C 0} 'intersection _C 2 of the measurement line _{L 1} and the fusion ring, _{C 2'} center position _C 1, _{C 1} of a single crystal measurement line _{L 2} And fusion ring intersection K ₁ , K ₂ reference value coefficient L ₀ standard line L ₁ , L ₂ measurement line r arrow TH1, TH2 reference value line ΔG liquid level position (gap)

Claims (8)

Taking an image of the meniscus part that forms the boundary between the single crystal pulled up by the Czochralski method and the raw material melt contained in the crucible,
Detecting a fusion ring appearing in the meniscus portion based on the luminance distribution of the imaged image, and calculating a liquid surface position of the melt based on the position of the fusion ring;
A cylindrical heat shield disposed above the crucible so as to surround the periphery of the single crystal while controlling the height position of the crucible containing the melt based on the calculated liquid level position. In the single crystal pulling method for adjusting the interval between the lower end of the liquid and the liquid level,
In the first period in which light from a heater provided around the crucible is reflected on the meniscus portion, the fusion ring is detected using a first reference value;
Detecting the fusion ring by using a second reference value larger than the first reference value in a second period in which light from a heater provided around the crucible is not reflected in the meniscus portion. A single crystal pulling method characterized by the above. The first reference value set during the first period is set within a luminance range when light from the lower surface of the thermal shield is reflected on the meniscus portion,
The second reference value set during the second period is set within a luminance range when light from the side wall portion of the crucible protruding above the liquid level position is reflected on the meniscus portion. The single crystal pulling method according to claim 1. During the first period and the second period, the light from the side wall of the crucible protruding above the liquid level position reflects the maximum value of the detected luminance when reflected on the meniscus. The single crystal pulling method according to claim 2, wherein a first reference value and the second reference value are set. The coefficient of the first reference value is not less than 0.6 and less than 0.8;
The single crystal pulling method according to claim 3, wherein the coefficient of the second reference value is 0.8 or more and 0.95 or less. A first measurement line orthogonal to the reference line passing through the center of the single crystal image and spaced from the center by a first distance is set, orthogonal to the reference line, and the first measurement line from the center. Set a second measurement line separated by a second distance longer than the distance;
Detecting two intersections of the first measurement line and the fusion ring, and detecting two intersections of the second measurement line and the fusion ring;
A first distance between the two intersections on the first measurement line, a second distance between the two intersections on the second measurement line, the first distance, and the second distance. Based on the above, the center position of the single crystal located on the reference line is calculated,
The single crystal pulling method according to any one of claims 1 to 4 , wherein a liquid surface position of the melt is calculated from a center position of the single crystal.   The single crystal pulling method according to any one of claims 1 to 5, wherein it is determined whether the first period or the second period is based on a length of the single crystal monitored during the pulling of the single crystal. The single crystal pulling method according to claim 1 , wherein it is determined whether the first period or the second period from the remaining amount of the melt in the crucible monitored during the pulling of the single crystal. The single crystal pulling method according to claim 1 , wherein it is determined whether the first period or the second period from a height direction position of the crucible monitored during the pulling of the single crystal.

Images