TW202106935A - Method for pulling a single silicon crystal by the czochralski method from a melt - Google Patents

Abstract

A method for pulling a single silicon crystal by the Czochralski method from a melt, comprising the pulling of at least one end portion of a seed cone and of a cylindrical portion of the single crystal at a pulling velocity which follows a mandated profile, and the repeated implementation of steps a) to c) below as part of controlling the diameter of the end portion of the seed cone and of the cylindrical portion of the single crystal: a) the measurement of a current diameter of the single crystal; b) the establishment of the deviation of the current diameter from a target diameter of the single crystal; and c) the use of the heating power of an annular resistance heater which has a diameter which is greater than the current diameter of the single crystal, and which is disposed concentrically to the seed cone over the melt, as controlled variable for controlling the diameter of the single crystal toward the target diameter of the single crystal.

Description

Method for pulling silicon single crystal from melt through Czochralski crystal pulling method

The present invention relates to a method for pulling a silicon single crystal from a melt through the Czochralski method, which includes pulling a seed cone and a cylindrical part of the single crystal.

According to the Czochralski method, the single crystal is usually pulled from the melt by the following methods: melting the polycrystalline material in a crucible to form a melt, immersing the seed crystal of the single crystal in the melt, and removing the crystal from the melt. Pull the immersed crystal in the middle. The single crystal pulled in this process can be subdivided into seed cone, cylindrical part and end cone. Of particular interest is the cylindrical part of the single crystal, because this part undergoes further processing into semiconductor wafers, which in turn are raw materials for the production of electronic components.

According to US 2003/0033972 A1, it is desirable to pull the seed cone of a single crystal of semiconductor material in such a way that the seed cone obtains a desired shape, and the single crystal happens to have inherent point defects (intrinsic Point defect) and its agglomerates have the desired characteristics in terms of defect distribution.

The method described in WO 00/56956 A1 envisages changing the pulling speed during the first part of pulling the single crystal in order to control the diameter of the single crystal, and pulling the single crystal at a predetermined pulling speed during the second part of pulling.

US 2011/0126757 A1 proposes to pull a cylindrical portion of a single crystal of semiconductor material at a prescribed pulling speed, and to control the diameter of the cylindrical portion by means of the heating power of a heating source arranged above the melt. Knowing the thermal conditions during the pulling of the cylindrical part of the single crystal, the pulling speed curve is selected in such a way that the pulling speed v _{p is} proportional to the axis at the phase boundary between the melt and the grown single crystal _{The ratio v p} /G to the temperature gradient G is kept within the expected value range, and this ratio strictly simulates the defect distribution of intrinsic point defects. Disturbances that deviate from the target diameter can be offset by the penetration control of the heating source.

However, in the ratio mentioned above, the pulling speed v _p is only an approximate expression of the crystallization speed v of single crystal growth. Moreover, this approximation is relatively inaccurate especially in the transition stage from the seed cone to the cylindrical part. In this stage, the crystallization rate v is especially affected by temperature fluctuations in the melt. A common result of this is that when the single crystal cylindrical portion is just started to be pulled, the dynamic balance with the target ratio v/G _{cannot be achieved through the specified ratio v p /G.} Such a result is a yield loss because the first part of the cylindrical portion of the single crystal does not have the desired physical characteristics in terms of the defect distribution of the internal point defects.

According to WO 2018/069051 A1 a method is contemplated, the seed crystal during the pulling cone, repeatedly a predetermined pulling velocity v _p and arrangement of the heating power of the heating source above the melt. However, this method is more susceptible to interference.

The purpose of the present invention is to effectively limit the aforementioned yield loss by providing a relatively robust method.

The object of the present invention is achieved by means of a method for pulling a silicon single crystal from a melt through the Czochralski crystal pulling method, the method comprising: Pull at least one end portion and cylindrical portion of the seed cone of the single crystal at a pulling speed that follows the prescribed curve, and repeat the following steps (a) to (c), and steps (a) to (c) are used as control units Part of the diameter of the end of the seed cone and the cylindrical part of the crystal: (a) Measure the current diameter of the single crystal; (b) Establish the deviation of the current diameter from the target diameter of the single crystal; and (c) Use the heating power of the ring resistance heater as a control variable for controlling the diameter of the single crystal toward the target diameter of the single crystal. The ring resistance heater has a larger diameter than the current diameter of the single crystal and is compatible with the seed crystal. Concentrically arranged above the melt.

The inventors have discovered that still during the pulling of the seed cone, the diameter must be controlled in a way that affects the ratio v/G as little as possible. Therefore, it can be ensured that _{the deviation between the prescribed pulling speed v p} and the crystallization speed v is kept sufficiently small to avoid the occurrence of the above-mentioned yield loss. Obviously, it is not enough to start the diameter control only by pulling the cylindrical part of the single crystal. Instead, it must be ensured that the control is started before the cylindrical portion of the single crystal is pulled. It is particularly preferable that when the seed cone with a length of at least 15 mm needs to be pulled up, the diameter of the single crystal is controlled with the aid of a ring resistance heater. By carrying out the present invention, it is possible to achieve the desired defect distribution of the intrinsic point defects in the cylindrical portion without any loss or only a small loss in yield.

The curve of the pulling speed is specified as a function of the expected defect distribution of the intrinsic point defects, and the known fact is that the pulling speed is between the pulling speed and the axial temperature gradient G at the phase boundary between the melt and the single crystal The ratio seriously affects the defect distribution. In addition, it is preferable to consider the influence of the stress field in the single crystal on the point defect and its movement, and the influence of the temperature field in the single crystal on the radial diffusion of the point defect. For example, the way this happens is described in US 2008/0187736 A1. The radial curve of the axial temperature gradient G at the phase boundary between the melt and the single crystal during the pulling of the cylindrical portion of the single crystal is preferably determined in advance through simulation calculation.

In order to control the diameter of the single crystal during the pulling of at least one end of the seed cone of the single crystal and during the pulling of the cylindrical part of the single crystal, a feedback control loop is set up, in which the specified target diameter is used as the reference variable, and the current diameter is used as the control The variable, the heating power of the ring resistance heater is used as the manipulated variable. The heating power of the main heating source for heating the crucible from the side follows a prescribed curve, and is preferably readjusted when necessary to reduce the load on the ring resistance heater. Preferably, the pulling speed v _{p is} used as a manipulation variable to control the diameter of the first part of the seed cone to the end of the seed cone.

In order to measure the current diameter (actual diameter) of the single crystal, it is preferable to analyze the images of the camera system by means of automatic image processing. These images depict the pale ring visible on the melt surrounding the growing single crystal. . Starting from the diameter of the shallow ring, conclusions can be drawn about the diameter of the single crystal. In principle, any known method can be used to measure the current diameter. The controller of the feedback control loop of the present invention is designed as a PID controller, preferably a PD controller.

The desired defect distribution of intrinsic point defects preferably includes a distribution in which, although these point defects are present in the single crystal, in terms of their number, they are still below the concentration that exists in an agglomerated state. In the presence of a better distribution, especially in the form of COP defects (grains derived from crystals), OSF defects (stacking faults caused by oxidation), B-belt defects and Lpit defects (large pits) in the form of agglomeration Body, can not be detected. The distribution of intrinsic point defects is preferably such that the silicon semiconductor wafer obtained from the cylindrical portion of the single crystal includes only silicon interstitial lattice atoms (interstitials) or only vacancy as point defects, but does not include a size greater than 5 nm Of detectable agglomerates. Correspondingly, the curve of the pulling speed v _p is preferably specified so that the concentration of holes and gaps is kept below the critical concentration. The critical concentration can be determined by comparing with the incidence of larger aggregates.

The silicon single crystal pulled by the method of the present invention preferably has a diameter of at least 300 mm.

The present invention will be further described below with reference to the drawings.

The device according to Fig. 1 includes: a pulling chamber 1, which contains a crucible 2 for containing the melt 3, and a camera system 4, which is used to observe the phase between the melt 3 and the grown single crystal 5 along the optical axis 12 World. The crucible 2 is carried by a shaft 6, which can be raised, lowered and rotated. Arranged around the crucible 2 is a resistance heater 7 which is used to melt solid silicon to form a melt 3. The single crystal 5 grown from the melt is pulled by the pulling mechanism 8. Arranged on the crucible 2 are a heat shield 9 and a ring-shaped resistance heater 10, which surround the grown single crystal 5. A heat shield (not shown) that restricts the upward transfer of heat may be provided above the annular resistance heater 10.

Figures 2 and 3 are intended to illustrate _{how fluctuations in the pulling speed v p} can change the concentration C of point defects in the single crystal. _{Fig. 2 shows the jump Δv p} of the two pulling speeds as a function of the axial position P in the single crystal in a continuous curve and a dashed curve. The position P ₀ represents the starting point of the cylindrical portion of the single crystal. Figure 3 shows the direct effect of jumps with significant changes in point defect concentration ΔC. It can be seen that after the jump, a considerable amount of crystal length must be pulled to reach the concentration of point defects that existed before the jump. Because controlling the diameter of the single crystal with the pulling speed as a manipulated variable certainly includes such fluctuations in the pulling speed, despite its advantages, considering the effects shown, its use will not be expected, at least when the single crystal is not only When the diameter and the distribution of point defects remain within the specified range.

The suggestion explained as the state of the art is to use the heating power of a ring resistance heater arranged above the melt to control the diameter of the single crystal when pulling the cylindrical part of the single crystal. 3 shows the effect of yield loss.

Therefore, according to the present invention, it is proposed to start this control before the completion of pulling the seed cone of the single crystal. The difference is shown in Figure 4. In the example of the present invention shown (solid line), before reaching the crystal position P ₀ , in other words, during the pulling of the seed cone, the diameter is switched from controlling the diameter _{by the pulling speed v p as the manipulated variable to the ring shape} The heating power of the resistance heater is controlled, and after switching, the _{pulling speed v p is controlled according to the regulation that the ratio v p} /G is placed within the imaginary value range. On the contrary, according to the counter example (dashed line), _{the pulling speed fluctuation system due to the control of the pulling speed v p} as the manipulation variable is maintained up to the position P ₀ in the single crystal, and the result is shown in FIG. 3.

Regarding the deviation of the diameter of the single crystal from the target diameter, the method according to the present invention is also advantageous. Figure 5 compares the target diameter (solid line) and the diameter according to the example (dash-dotted line) and counterexample (dashed line). The diameter ds represents the target diameter in the cylindrical portion of the single crystal. When the method of the present invention is applied, the deviation from the target diameter is relatively small even when the cylindrical portion of the single crystal is initially pulled.

The above description of the exemplary embodiments should be understood as illustrative. Its completed disclosure enables the technical personnel to understand the present invention and its related advantages on the one hand, and on the other hand, within the scope of the technical personnel's understanding, it also includes obvious modifications and changes to the described structure and method. Therefore, the present invention and all such modifications and changes and equivalents will be covered by the protection scope of the patent application.

1: Lifting room 2: Crucible 3: melt 4: Camera system 5: Single crystal 6: axis 7: Resistance heater 8: Lifting mechanism 9: Heat shield 10: Ring resistance heater 12: Optical axis

Figure 1 shows an apparatus for pulling silicon single crystals through the CZ method.

Figure 2 shows a curve of the deviation of the pulling speed as a function of the position in the single crystal after two jumps of the pulling speed.

FIG. 3 shows the change in the concentration of point defects in the single crystal as a result of the jump shown in FIG. 2.

Fig. 4 shows a graph of the pulling speed as a function of the crystal position when the method of the present invention is used and when the single crystal diameter control that is not the present invention is used.

FIG. 5 shows a curve of the deviation of the diameter of the single crystal from the target diameter as a function of the position in the single crystal when the method of the present invention and a counterexample that is not the present invention are used.

Claims (5)

A method for pulling silicon single crystal from the melt by the Czochralski method, which includes: Pull at least one end portion and cylindrical portion of the seed cone of the single crystal at a pulling speed that follows the prescribed curve, and repeat the following steps (a) to (c), and steps (a) to (c) are used as control Part of the diameter of the end of the seed cone and the cylindrical part of the single crystal: (a) Measure the current diameter of the single crystal; (b) Determine the deviation of the current diameter from the target diameter of the single crystal; and (c) Use the heating power of a ring resistance heater as a control variable for controlling the diameter of the single crystal toward the target diameter of the single crystal, the ring resistance heater having a diameter larger than the current diameter of the single crystal, and The seed cone is arranged concentrically above the melt. The method according to claim 1, which includes: When a seed cone with a length of at least 15 mm needs to be pulled, the heating power of the ring resistance heater is used as a manipulation variable to start to control the diameter of the single crystal seed cone. The method according to claim 2, which includes using a PD controller for controlling the diameter of the single crystal using the heating power of the ring resistance heater as a manipulation variable. The method according to claim 1 or 2, which includes using the pulling speed v _p as a manipulation variable to control the diameter of the first part of the seed cone to the end of the seed cone. The method according to claim 1 or 2, wherein the annular resistance heater is disposed above the melt at a distance of not less than 1 mm and not more than 200 mm.

Images