JPS6258138B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a method for manufacturing a semiconductor substrate, and more particularly to a method for reducing minute defects found in an epitaxial surface layer during epitaxial growth of a silicon single crystal thin film by a vapor phase reaction method. Silicon epitaxial layers are essential as substrates for ICs and LSIs, as well as for VLSIs, and are used for bipolar devices. Crystal defects such as stacking faults, dislocations, and micro defects (shallow pits) are found in this epitaxial layer. It is well known that these crystal defects deteriorate the electrical characteristics of devices. Minimizing these crystal defects can contribute to improving the yield of devices. Various methods have been proposed to suppress crystal defects. Currently, stacking faults in epitaxial layers can be reduced to 10/cm ² or less by gas etching before epitaxial growth, and dislocations can be reduced by improving uniform heating methods during epitaxial growth. Similarly, the defect density is
It is becoming less than 10 pieces/ ^cm2 . The problem here is minute defects found in the epitaxial surface layer. Micro defects grow as the device process progresses, becoming large stacking faults and causing a decrease in yield. This microdefect density is 10 ⁶ to 10 ⁷ per epitaxial layer using conventional untreated wafers.
^cm2 , and the correlation with device characteristics, such as a drop in breakdown voltage, is being clarified. It is presumed that the cause of these microdefects is heavy metal contamination, which has a high diffusion rate. It is currently unclear whether the heavy metals are due to contamination of the wafer surface during epitaxial growth pretreatment, or whether they come flying from the inner walls of the surrounding equipment during epitaxial growth. The methods for gettering heavy metals include (1) forming mechanically strained debris on the back surface of the substrate, (2) forming strained debris by ion implantation, and (3) forming a Si ₃ N ₄ film and introducing misfit dislocations. (4) Formation of strained debris using a laser has been proposed. The effectiveness of the gettering effect caused by these backside treatments has been confirmed for MOS devices, but the process for bipolar devices is more complex, and it is believed that the effectiveness will gradually be lost as high-temperature processes are repeated. It is said. The present invention heat-treats a silicon single crystal substrate in two steps, one in a dry oxygen atmosphere and one in a wet oxygen atmosphere, to generate micro-defects consisting of precipitates and dislocations inside, and then use the resulting strain field to destroy the micro-defects or their nuclei. This is an attempt to reduce micro defects in the epitaxial film by the so-called intrinsic gettering (IG) effect. For example, in the process of epitaxially growing a silicon single-crystal thin film on a silicon single-crystal substrate by vapor phase reactivity, the present invention can be applied by pre-processing the substrate in a dry oxygen atmosphere at a temperature range of 650°C to 1000°C for 16 hours.
Two types of heat treatment steps are carried out in succession: removing the oxide film after heat treatment for more than an hour, and then heat treatment in a wet oxygen atmosphere at a temperature range of 1050°C to 1200°C for at least 2 hours, especially after this, epitaxial growth is performed. This is a method for manufacturing a semiconductor substrate characterized by the following. Another feature is that epitaxial growth is performed before heat treatment in a wet oxygen atmosphere. Regarding the effectiveness of IG effect, for example, GA
Rozgonyi et ol: Appl.Phys.Lett.vol.32, 747~
749 (1978), improvements in the lifetime of MOS devices have been reported. The present inventors considered that the IG effect is particularly effective in reducing microdefects in the epitaxial layer. A detailed explanation will be given below based on examples. Example 1: A substrate for epitaxial use was heat treated in advance in a dry oxygen atmosphere at a temperature range of 600°C to 1200°C for 16 to 64 hours to remove the oxide film.
Grown epitaxially. In order to evaluate minute defects in the epitaxial film, it was heat treated at 1140°C for 2 hours in a wet oxygen atmosphere, etched for 30 seconds with dilt etch solution, and observed using a Nomarski interference microscope. As a result, in dry oxygen, 700℃~
The number of minute defects in the epitaxial film grown epitaxially using a substrate heat-treated in a temperature range of 1000° C. was significantly reduced to 5×10 ³ defects/cm ² or less. Example 2: Substrates having various oxygen concentrations were heat-treated in the above dry oxygen atmosphere, and the occurrence of internal defects and the amount of interstitial oxygen were measured by etching and infrared spectroscopy. As a result, it was found that the reduction in interstitial oxygen significantly depends on the oxygen concentration of the starting wafer and the heat treatment temperature and time. The amount of interstitial oxygen decrease is proportional to the amount of precipitated internal defects, and etching observations clearly showed such a correlation at temperatures below 1000°C. An example of the measurement results of the amount of decrease in interstitial oxygen is shown in Table 1.

[Table] The rate of decrease in interstitial oxygen is as follows: 1) The longer the heat treatment time is, the greater it is. 2) Under the same heat treatment conditions, the higher the oxygen concentration of the starting wafer, the greater the rate of decrease. 3)
It was found that it exhibits characteristic behavior, such as the rate of decrease reaching its maximum at a specific temperature. Example 3: Next, as shown in Example 2 above, the oxide film of the wafer heat-treated in dry oxygen was removed, and then heat-treated in wet oxygen at 1100°C to 1200°C for 2 hours to remove interstitial oxygen. The reduction rate and the state of precipitation of internal defects were measured. As a result, 650℃~1000℃
It has been found that when a substrate that has been heat-treated in dry oxygen is subsequently heat-treated in wet oxygen, the rate of decrease in interstitial oxygen is even more remarkable, and the increase in internal defects is also remarkable. Table 2 shows the results when the samples shown in Table 1 were further heat treated in wet oxygen at 1140°C for 2 hours.

[Table] As shown above, when heat treatment is performed in dry oxygen in advance at an appropriate temperature range and then heat treatment is performed in wet oxygen, the rate of decrease in interstitial oxygen concentration is more than one order of magnitude higher than in the first treatment. It was found that there are cases where it increases. Since the second treatment only takes a short time, by performing these two heat treatment steps in succession, it has become possible to shorten the heat time as a whole. For example, to achieve a reduction rate of 40% or more, it takes 64 hours for sample E with the first treatment alone, whereas 18 hours is sufficient for sample A with the first and second treatments. . In other words, by adding the second process, the first process time can be significantly shortened. Furthermore, by adding the second treatment to wafers with a low oxygen concentration of about 13×10 ¹⁷ /cm ³ , a reduction rate of 17.1% could be achieved, and the IG effect was recognized. Example 4: In order to confirm the effectiveness of successive first and second treatments, the interstitial oxygen concentration did not decrease even if heat treatment was performed only in wet oxygen in the second treatment without performing the first treatment. No internal defects were observed. Example 5: Therefore, when epitaxial growth was performed using a wafer that had been subjected to the heat treatment steps of the first and second treatments described above, a remarkable IG effect was observed, and with an appropriate combination of heat treatment conditions, The microdefect density in the epitaxial layer was less than 5×10 ³ /cm ² . As is clear from the above examples, by successively performing dry and wet heat treatment in oxygen,
The heat treatment time required to produce the IG effect can be significantly shortened, and even on low oxygen concentration substrates.
It has been found that this method has significant effects when manufacturing semiconductor devices, such as being able to bring about the IG effect.

Claims (1)

[Claims] 1 In the process of epitaxially growing a silicon single crystal thin film on a silicon single crystal substrate by vapor phase reactivity, after the epitaxial growth, the substrate is heat treated in a dry oxygen atmosphere at a temperature range of 700°C to 1150°C, and then wet oxygen 1100℃~1200℃ in atmosphere
1. A method for manufacturing a semiconductor substrate, characterized by performing heat treatment in a temperature range of .