KR100742337B1 - Method for silicon single crystal of unifor quality by controlling pulling rate - Google Patents

Abstract

The present invention relates to a silicon single crystal growth method, and more particularly, to a silicon single crystal growth method by Czochralski method, comprising the step of controlling the diffusion of vertical point defects in a cooling section during pulling up of a silicon single crystal. Silicon single crystal growth method.

According to the present invention, not only the average pulling speed of the silicon single crystal, that is, the axial ΔV can be effectively controlled, but also the productivity of the high quality silicon single crystal can be increased.

Defects, average pulling speed, instantaneous pulling speed, temperature gradient

Description

Silicon single crystal growth method {METHOD FOR SILICON SINGLE CRYSTAL OF UNIFOR QUALITY BY CONTROLLING PULLING RATE}

1 is a cross-sectional view showing the inside of a typical silicon single crystal ingot growth apparatus,

2 is a graph schematically showing a change in [V] or [I] according to a change in V / G value.

3a shows the quality of the 0 to 300 mm section of the ingot grown under the conditions of Experimental Example 1 according to the instantaneous pulling speed and the average pulling speed (120 min) (the width between the reference lines is 0.1 mm / min),

Figure 3b shows the quality of the 0 ~ 300mm section of the ingot grown under the conditions of Experimental Example 1 according to the average pulling speed (240min),

Figure 3c shows the quality of the 500 ~ 700mm section of the ingot grown under the conditions of Experimental Example 1 according to the instantaneous pulling speed, average pulling speed (120min),

Figure 3d is a result showing the quality of the 500 ~ 700mm section of the ingot grown under the conditions of Experimental Example 1 according to the average pulling speed (240min),

4a shows the quality of the ingot grown under the conditions of Experimental Example 2 according to the instantaneous pulling speed and the average pulling speed (120 min),

Figure 4b is a result showing the quality of the ingot grown under the conditions of Experimental Example 2 according to the average pulling speed (240min),

5A is a result of instantaneous pulling rate profiling of a silicon single crystal grown under a comparative example condition;

5B is an average pulling speed profiling result obtained by converting the instantaneous pulling speed profiling of FIG. 5A into an average pulling speed;

6A is a result of instantaneous pulling rate profiling of a silicon single crystal grown under experimental conditions;

6B is an average pulling speed profiling result obtained by converting the instantaneous pulling speed profiling of FIG. 6A into an average pulling speed.

The present invention relates to a method for producing a silicon single crystal by the Czochralski method, and more particularly to a method for producing a silicon single crystal by controlling the pulling rate more effectively.

In general, in the method of growing silicon single crystals by the Czochralski method, polycrystalline silicon is loaded into a quartz crucible and melted polycrystalline silicon with heat radiated from a heater to form a silicon melt, followed by growing a silicon single crystal from the surface of the silicon melt. Let's do it.

When growing the silicon single crystal, the crucible is raised while rotating the shaft supporting the crucible so that the solid-liquid interface maintains the same height, and the silicon single crystal is in the opposite direction to the rotation direction of the crucible about the same axis as the rotation axis of the crucible. Pull up while rotating. The grown silicon single crystal is subjected to a wafer processing process such as slicing, lapping, polishing, cleaning, and the like to be a silicon single crystal wafer to be used as a semiconductor device substrate.

Recently, many defects are known to form in single crystal silicon in the crystal growth chamber when the crystals are cooled after solidification. These defects are caused in part due to the excess presence of intrinsic point defects (ie concentrations above the dissolution limit), known as vacancy and self-interstitials. Typically, silicon crystals grown from the melt are grown in one or another type of intrinsic point defects, i.e., excess of crystal lattice vacancy ("V") or self-interstitial ("I").

In the case of excess vacancy or self-interstitial overgrowth, if these concentrations reach critical supersaturation levels in the system, and the mobility of the caustic defects is high enough, reaction or agglomeration occurs. Aggregated intrinsic point defects in silicon can seriously affect the yield of materials in the manufacture of composite and highly integrated circuits.

Bacony-type defects include D-defects, Flow Pattern Defects (FPD), Gate Oxide Integrity (GOI) defects, Crystal Originated Particle (COP) defects, Light Point Defects (LPDs), as well as scanning infrared microscopy and laser It is known to be the cause of observable crystal defects, such as certain classes of bulk defects observed by infrared light scattering techniques such as scanning tomography. In addition, defects that act as nuclei of ring induced induced stacking faults (OISFs) are present in excess bacon.

Defects related to self-interstability are less studied, but in general they are thought to be low density self-interstitial dislocation loops or networks. While these defects are not the source of poor gate oxide integration, a major measure of wafer performance, they are widely recognized to cause other types of device failures, usually associated with current leakage issues.

Conventionally, in order to grow a high quality silicon single crystal capable of increasing semiconductor device yield, the G value is mainly improved in terms of V / G and the V value is simply controlled.

In addition, in order to increase the yield of high-quality silicon single crystal, a process window (Δ (V / G)) is sufficiently secured, and an ingot is grown within the window range (US 6,045,610 patent).

That is, numerous efforts have been made to minimize the radial ΔG because usually G values are non-uniform in the radial direction. On the contrary, it is true that efforts to effectively control ΔV in the silicon single crystal growth axis direction under limited Δ (V / G) have not been made enough.

None of the proposed ΔV control methods to date is completely satisfactory. Rather, the US 5,919,302 patent attempts to improve the yield of high quality single crystal ingot acquisition by controlling the spread of radial point defects by increasing ΔV upside down, but the effect is doubtful.

Accordingly, there is a need to develop a number of methods for growing high quality silicon single crystals that can increase semiconductor device yield.

Therefore, the inventors of the present invention solve the above problems of the prior art, and while studying a method for effectively producing a high quality silicon single crystal, the average pulling speed of the silicon single crystal for a suitable time, that is, the axial direction of the silicon single crystal It was confirmed that ΔV is in good agreement with the quality change of the single crystal ingot, and the present invention was completed.

It is therefore a main object of the present invention to provide a silicon single crystal growth method capable of effectively producing high quality silicon single crystals by controlling the average pulling speed of silicon single crystals, i.e., the axial ΔV of the silicon single crystals for a suitable time.

It is also an object of the present invention to provide a silicon single crystal growth method that can effectively produce high quality silicon single crystals by controlling the diffusion of vertical point defects in the cooling section during silicon single crystal growth.

It is also an object of the present invention to provide a silicon single crystal growth method capable of effectively producing a high quality silicon single crystal by controlling the silicon single crystal pulling rate (V) on an average for any of the previously grown sections and any sections to be grown. .

The silicon single crystal growth method by the Czochralski method according to the present invention for achieving the above object is characterized in that it comprises the step of controlling the diffusion of vertical point defects in a certain cooling section during the silicon single crystal pulling up.

In the controlling of the vertical point defect diffusion, the silicon single crystal pulling speed V is controlled on an average for a predetermined cooling period.

은In the cooling section, when the current pulling speed solidified into a single crystal is V ₀ , the pulling speed before m minutes is Vm and the pulling speed after n minutes is Vn based on the current time of solidification from the silicon melt to the single crystal.

silver

It is characterized by satisfying.

The cooling section is characterized in that the silicon single crystal is an arbitrary section that is previously grown and any section to be grown.

The arbitrary section is characterized in that it is from the solidification temperature to the temperature at which point defect coagulation begins to bacon.

Wherein said arbitrary section is from a solidification temperature to a temperature at which interstitial point defect aggregation starts.

The inventors of the present invention consider that V / G theory of Boronkov and similar effects of the pulling speed on the quality, the pulling speed control acts as a key factor in ensuring the quality uniformity of the ingot or wafer, and also the instantaneous pulling speed. While the instantaneous point defect concentration is determined during crystallization solidification, the final point defect concentration is determined after the diffusion of vertical point defects in a certain cooling section. By controlling the pulling speed, it is easy to secure the quality uniformity of the ingot or the wafer, thereby increasing the quality index.

First, in view of providing an effective method for producing high quality silicon single crystals, the behavior of point defects occurring during single crystal growth without regard to the most influential Boronkov Theory in silicon single crystals was examined.

As shown in Fig. 1, silicon single crystal is grown as the silicon solidifies from the melt. This is due to the loss of free energy in the form of latent heat held by the silicon atom or molecule in the fluid state is fixed to the liquid-liquid interface. Recently, the driving force for incorporating the silicon crystal growth unit into the crystal growth in the silicon melt is known to depend on the temperature gradient.

At this time, the driving force also affects the temperature gradient in the crystal, which is represented by the Voronkov theory on the surface. In the following, V is the single crystal pulling rate and the convection parameter of the point defect in the silicon single crystal. In addition, G is an instantaneous axial temperature gradient near the crystal melt interface and a diffusion parameter of point defects due to the temperature gradient in the crystal.

When G is dominant (relatively large), the thermodynamic point defect concentration gradient is also largely formed by a large temperature gradient, thereby increasing the point defect diffusion in the crystal direction. However, since the diffusion mobility of the interstitial is larger than the vacancy, the dominant point defect becomes the interstitial.

On the other hand, when V is large, the state of vacancy concentration higher than interstitial is maintained as it is during crystallization due to convection due to single crystal pulling (because the effect by G is small). That is, if the V / G value is greater than a certain threshold, vacancy-rich is formed, and if the V / G value is less than a certain threshold, interstitial-rich is formed. Therefore, in order to produce a defect-free single crystal throughout the ingot length, the V / G value during crystal growth must be kept within a threshold range.

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings. These examples are only intended to illustrate the invention, so the scope of the invention is not limited by these examples.

The silicon single crystal growth method according to the present invention is a silicon single crystal growth method by Czochralski method, characterized in that it comprises the step of controlling the diffusion of vertical point defects in the cooling section during the silicon single crystal pulling up.

In this case, in the controlling of the vertical point defect diffusion, it is preferable to control the silicon single crystal pulling rate V on the cooling section on average.

The single crystal pulling speed refers to the average pulling speed of the silicon single crystal, that is, the axial ΔV of the silicon single crystal. That is, the present invention is characterized by effectively controlling the axial ΔV of the silicon single crystal by adjusting the average pulling speed of the silicon single crystal in order to effectively produce a high quality silicon single crystal.

As described above, in order to manufacture a defect-free single crystal throughout the ingot length, it is necessary to keep the V / G value during the crystal growth within the threshold range, but to increase the instantaneous V / G value within the threshold range during the crystal growth. It doesn't have to be maintained. This is because a mass flow (material diffusion) occurs due to a concentration gradient, just as heat flow (thermal diffusion) occurs due to a temperature gradient. That is, even if the V / G value fluctuates at every moment, the V / G value has an average distribution through the up and down material diffusion in the grown silicon single crystal as long as the average V / G value is controlled in a certain interval. (See Fig. 2).

In addition, previously reported articles (M. Hourai et al, "GROWTH PARAMETERS DETERMINING THE TYPE OF GROWN-IN DEFECTS IN CZOCKRALSKI SILICON CRYSTALS", Materials Science Forum 196-201 (1995) p.1713. And G. Rozgonyi et al, "Crystal Originated Particles (COPs), Bulk Defects (Including the OSF-Ring), Gate Oxide Integrity (GOI), and DRAM Yield", Proceedings of Silicon Materials Science and Technology Forum (1997) p. 215) Based on the results, the outward diffusion of the point defects mainly occurs only at the outer periphery of the crystal, and the spread of the point defects in the radial direction is quite limited. Therefore, in order to effectively manufacture high quality silicon single crystal, the most important factor is to accurately determine the size of the section for calculating the average V / G value in consideration of vertical diffusion in the crystal growth axis direction.

In general, when silicon single crystal growth is affected by the doping of impurities (nitrogen, carbon, etc.) or the degree of supersaturation of point defects, the temperature at which excess point defects aggregate is as follows.

The coagulation temperature of the bacon and interstitial temperature is between about 1000 ~ 1120 ℃, and because of the presence of the caking in the temperature range above, relatively diffusion can be moved.

Therefore, the average pulling speed control section for suppressing the bacon flocculation is from the solidification temperature (or the solidification temperature, about 1410 ° C.) to an arbitrary temperature at which bacon agglomerates, and the average pulling speed control to suppress the interstitial flocculation. The section is preferably from a solidification temperature to an arbitrary temperature at which the interstitial aggregates.

The average pulling speed control section is a section in which the silicon single crystal pulling speed (V) is controlled on average with respect to a predetermined cooling section (constant cooling section, any section is used in the same sense), wherein the arbitrary section is from the solidification temperature. It is preferable that the temperature reaches the temperature at which point defect aggregation starts.

Further, the arbitrary section is preferably from the solidification temperature to the temperature at which the interstitial point defect aggregation starts.

In the present invention, the average pulling speed is controlled to the same arbitrary section on the assumption that the bacon and interstitial coagulation temperatures are the same for the convenience of process control to simultaneously control the two types of point defect coagulation.

First, it was confirmed that the average value of the pulling speeds over a predetermined period determined through Experimental Examples 1 and 2 is a meaningful index for the production of high quality silicon single crystals.

Experimental Example 1

Single crystal rotation speed (Seed Rotation, SR) 18rpm, Crucible Rotation (CR) -4rpm, silicon single crystal pulling speed (or defect free pulling speed, ΔV) under the growth conditions of 0.46 ~ 0.48mm / min The quality results of the ingot grown under the above conditions are shown in FIGS. 3A, 3B, 3C, and 3D according to the instantaneous pulling speed, the average pulling speed (120 min), and the average pulling speed (240 min), respectively.

At this time, the initial speed of the ingot was severely changed in the pulling speed, and in the middle of the body there was a slight variation in the pulling speed. Looking in more detail as follows.

(1) 0 ~ 300mm section

The pulling speed fluctuation was very large because the pulling speed was abnormally controlled by improper heater power (temperature) setting. 3A and 3B, when the simple instantaneous speed and the average speed of 120 minutes correspond to the quality results of the ingot, it can be seen that the pulling speed and the quality change do not coincide.

For example, although the interstitial abundance (I-rich) is less than before and after growth in the vicinity of 220mm, the pulling speed appears to be large, deviating from the V / G perspective. However, it can be seen that the correspondence is relatively good when corresponding to the average speed of 240 minutes.

(2) 500 ~ 700mm section

This is a section in which the rate of increase increases with slight fluctuation. Again, the instantaneous instantaneous speed and the average speed of 120 minutes do not coincide with the quality change of the ingot (see FIGS. 3C and 3D). However, it can be seen that the correspondence is relatively good when corresponding to the average speed of 240 minutes.

Experimental Example 2

Single crystal rotation speed (SR) 13rpm, crucible rotation speed (CR) -0.1rpm, silicon single crystal was grown under the growth conditions of 0.55 ~ 0.57mm / min of the growth rate of silicon single crystal, the growth of the ingot The quality results are shown in FIGS. 4A and 4B according to the instantaneous pulling speed, the average pulling speed (120 min) and the average pulling speed (240 min), respectively.

Inappropriate temperature setting caused a change in pulling speed by the Auto Diameter Control (ADC). Although the growth conditions were increased for the same reason as in Experimental Example 1, the pulling rate was increased, but as in Experimental Example 1, it can be seen that the average pulling rate of 240 minutes was relatively well consistent with the quality change of the single crystal ingot.

In addition, the pulling rate control method for high quality single crystal growth according to the present invention can be controlled as follows.

는 1분 간격으로 순간인상속도 V를 측정했을 때, -m분부터 n분까지의 순간인상속도의 합이다.here,

Is the sum of the instantaneous rate of increase from -m to n minutes when the instantaneous rate of increase V is measured at 1 minute intervals.

By the way, since it is necessary to satisfy the condition of raising the defect when growing single crystal,

조건이 주어진다.

Conditions are given.

therefore,

It is desirable to satisfy.

은 m분 전의 순간인상속도부터 1분 전까지의 순간인상속도의 합이다. 또한,

현재 순간인상속도부터 n분후까지의 순간인상속도의 합이다. At this time, V _min is the lower limit of the rate of increase of the defect, V _max is the upper limit of the rate of zero defect, the difference (V _max -V _min ) is the rate of margin for the growth of zero defect. Also,

Is the sum of the instantaneous speeds up to the minute before the minute. Also,

It is the sum of the instantaneous rate of increase from the current instantaneous rate of increase to n minutes later.

In Experimental Examples 1 and 2, it was confirmed that the average value of the pulling speeds for a specific period is a significant index for the preparation of high quality silicon single crystals. The method used to increase the defect yield is as follows.

First, Vavg. Is defined as Vmin <Vavg. <Vmax condition must be satisfied,

은

silver

을 만족시켜야 한다.

Must satisfy

That is, if the average pulling speed is higher than Vmax in any of the previously grown sections, the lower pulling speed should be maintained in the additional growing section. On the contrary, if the average pulling speed in any of the previously grown sections is smaller than Vmin, the additional pulling section should be set to the pulling speed corrected above the difference. In addition, additional temperature setting is necessary to enable the setting of the pulling speed as described above.

Hereinafter, the present invention will be described in more detail with reference to Examples. These examples are only intended to illustrate the invention, so the scope of the invention is not limited by these examples.

Comparative Example

Only the instantaneous pulling speed was controlled at the time of silicon single crystal pulling.

The silicon single crystal was grown under the growth conditions of 16 rpm single crystal speed (SR), -0.3 rpm crucible rotation speed (CR), and pulling rate (ΔV) of silicon single crystal at 0.62 to 0.65 mm / min.

Instead of controlling the average pulling speed, only the instantaneous pulling speed was controlled and the amplitude of the pulling speed was reduced as much as possible.

Referring to the instantaneous pulling speed profiling of Fig. 5A, it appears that the pulling speed control is relatively well because the amplitude of the pulling speed is significantly reduced. However, referring to the average pulling speed profiling of FIG. 5B converted to the average pulling speed, it was confirmed that the high-quality single crystal growth rate margin deviated from a considerable length.

<Example>

According to the present invention described above, the pulling speed was controlled in view of the average pulling speed.

The silicon single crystal was grown under the growth conditions of 16 rpm single crystal speed (SR), -0.3 rpm crucible rotation speed (CR), and pulling rate (ΔV) of silicon single crystal at 0.62 to 0.65 mm / min. 6A shows the instantaneous pulling speed profiling and the average pulling speed profiling of FIG. 6B, respectively.

More specifically, the pulling speed was controlled in consideration of the average pulling speed of 120 minutes before growth and 120 minutes after growth. That is, the average pulling speed for 120 minutes before growth is intact. If the pulling speed exceeds the margin margin (over / under), the average pulling speed is adjusted to be opposite (under / over), so that the average pulling speed is the total length of the ingot. It was controlled to be within the flawless range in total.

It is apparent that the present invention is not limited to the above embodiments and can be implemented by many modifications within the technical idea of the present invention by those skilled in the art.

According to the silicon single crystal growth method according to the present invention as described above, according to the silicon single crystal growth method of the present invention, it is possible to effectively control the average pulling speed of the silicon single crystal, that is, the axial direction ΔV. In addition, the productivity of a high quality silicon single crystal can be increased.

Claims (6)

delete In the silicon single crystal growth method by Czochralski method, Controlling the diffusion of vertical point defects in a predetermined cooling section when the silicon single crystal is pulled up, and the controlling of the vertical point defect diffusion on the silicon single crystal pulling averages the silicon single crystal pulling rate V in a predetermined cooling section. To control, The control may be performed such that the silicon single crystal has an average pulling speed V _avg within the predetermined cooling period greater than or equal to the defect free lower limit pulling speed V _min and smaller than or equal to the defect free upper limit pulling speed V _max . A silicon single crystal growth method characterized by maintaining a pulling rate (V). The method of claim 2, The current pulling speed solidified to single crystal is V ₀ , the pulling speed before m minutes is Vm, the pulling speed after n minutes is Vn, and the _minimum lower pulling speed is V _min . In the cooling section, when the upper limit pulling speed is V _max .

silver

Silicon single crystal growth method characterized by the above. The method of claim 2, The predetermined cooling section is a silicon single crystal growth method, characterized in that the silicon silicon crystal is grown and the section to be grown. The method of claim 2, The predetermined cooling section is a silicon single crystal growth method, characterized in that from the solidification temperature to the temperature at which point defects start aggregation when bacon. The method of claim 2, The predetermined cooling section is a silicon single crystal growth method, characterized in that from the solidification temperature to the temperature at which the interstitial point defect aggregation starts.

Images