JP6881387B2 - DZ layer measurement method - Google Patents

Description

The present invention relates to a method for measuring a DZ layer of a silicon wafer.

As a silicon wafer for a semiconductor device, a silicon single crystal grown by the Czochralski method (CZ method) is mainly used. Furthermore, for advanced devices, the wafer surface layer, which is the device formation region, is defect-free, and high-density oxygen precipitates (Bulk Micro Defects: BMD) that contribute to gettering of metal impurities are present in the bulk. Is required to be.

However, in the silicon single crystal grown by the CZ method, there is a crystal defect called a Green-in defect introduced during crystal growth. Therefore, the wafer for advanced devices may be manufactured by subjecting it to a predetermined heat treatment. By this heat treatment, Green-in defects disappear on the wafer surface layer, a defect-free layer (Depended Zone: DZ) is formed, and BMD is formed at a high density in the bulk.

At the time of device manufacturing, this DZ layer is required to have a predetermined width or more, which is an important quality item at the time of wafer manufacturing.

Conventionally, when measuring the DZ layer width, a wafer is cleaved or diagonally polished, the cleaved surface or the polished surface is selectively etched, and then a visual inspector observes the wafer and / or photographs it via a microscope. It was measured from the image. For example, the distance from the wafer surface to the predetermined BMD is adopted as the DZ layer width.

Japanese Unexamined Patent Publication No. 2002-100631 Japanese Unexamined Patent Publication No. 2000-068280

However, the above-mentioned method has a problem that the measurement accuracy is high because the inspector actually observes the defect, but the throughput is poor because it is a visual inspection.

On the other hand, as in Patent Document 1, the DZ layer width can be obtained from the oxygen concentration distribution measured by SIMS, but SIMS also detects oxygen due to extremely small oxygen precipitates. However, such microscopic oxygen precipitates cannot contribute to gettering.

According to Patent Document 2, the BMD size that contributes to gettering depends on the BMD density. For example, when the BMD density is 1 × 10 ⁹ / cm ³ , a BMD of about 40 nm or more contributes to gettering. On the other hand, in microscopic observation, BMD of 50 nm or more is detected, and only BMD that contributes to gettering can be detected. Therefore, the DZ layer width evaluated by microscopic observation is important.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a method for measuring a DZ layer, which can obtain a DZ layer width obtained by microscopic observation with good throughput.

In order to achieve the above object, the present invention is a method for measuring a DZ layer formed on the surface layer of a silicon wafer, based on the DZ layer width obtained from the depth distribution of oxygen concentration measured by SIMS and microscopic observation. A calibration curve with the obtained DZ layer width is created, and the DZ layer width obtained by microscopic observation is obtained from the DZ layer width obtained by SIMS of the measurement sample based on the prepared calibration curve. A method for measuring the DZ layer is provided.

With such a method for measuring the DZ layer, the width of the DZ layer obtained by microscopic observation can be obtained with good throughput. That is, since the visual inspection is performed only when the calibration curve is created, the throughput can be improved. In addition, the DZ layer width obtained by microscopic observation, which is more important than the DZ layer width obtained by SIMS, can be obtained.

At this time, in the SIMS measurement, the shallowest depth at which the difference between the moving average value of the oxygen concentration and the measured value of the oxygen concentration is within ± 3% of the moving average value of the oxygen concentration is defined as A, which is deeper than the A. The depth B in which the moving average value of the oxygen concentration and the measured value of the oxygen concentration deviate by ± 5% or more from the moving average value of the oxygen concentration in the region can be defined as the DZ layer width of SIMS.

By such a procedure, the DZ layer width can be suitably obtained from the depth distribution of the oxygen concentration measured by SIMS.

At this time, in the SIMS measurement, the shallowest depth at which the difference between the moving average value of the oxygen concentration and the measured value of the oxygen concentration is within ± 3% of the moving average value of the oxygen concentration is defined as A, which is deeper than the A. The moving average value of oxygen concentration and the measured value of oxygen concentration in the region deviate by ± 5% or more from the moving average value of oxygen concentration. A curve can be created, and the depth C at which the moving average value of the oxygen concentration and the approximate curve deviates from the approximate curve by ± 5% or more in a region deeper than A can be defined as the DZ layer width of SIMS.

Also by such a procedure, the DZ layer width can be suitably obtained from the depth distribution of the oxygen concentration measured by SIMS.

At this time, the moving average value of the oxygen concentration is preferably the average value of the measured values of the oxygen concentration of 5 or more and 10 or less before and after the measurement depth.

The SIMS measurement interval is preferably 0.1 μm or less, and if there are 5 or more measurement points for calculating the moving average value, the measured values are sufficiently smoothed to prevent erroneous determination of the DZ layer. it can. Further, if the number of measurement points for calculating the moving average value is 10 or less, the smoothing is not performed too much, and it is not impossible to determine the point where the oxygen concentration suddenly increases.

As described above, according to the method for measuring the DZ layer of the present invention, the width of the DZ layer obtained by microscopic observation can be obtained with good throughput.

It is a flow chart which shows the measuring method of the DZ layer of this invention. It is a figure which shows the oxygen concentration distribution of a heat-treated wafer. It is a figure which shows the error rate of the measured value of oxygen concentration and the moving average value of oxygen concentration. It is a figure which shows the oxygen concentration distribution of another heat-treated wafer. It is a figure which shows the error rate between the moving average value and the approximate value of oxygen concentration. It is a figure which shows the oxygen concentration distribution of the sample 3 of an Example. It is a figure which shows the calibration curve of the DZ layer width obtained by SIMS measurement, and the DZ layer width obtained by microscopic observation.

Conventionally, when measuring the DZ layer width, the wafer is cleaved or diagonally polished, the cleaved surface or the polished surface is selectively etched, and then a visual inspector determines the distance from the wafer surface to the predetermined BMD. It was used as the DZ layer width.
However, since this method is a visual inspection, there is a problem that the throughput is poor.

On the other hand, the DZ layer width can be obtained from the oxygen concentration distribution measured by SIMS, but SIMS also detects oxygen due to extremely small oxygen precipitates. However, such microscopic oxygen precipitates cannot contribute to gettering.
The BMD size that contributes to gettering depends on the BMD density. For example, when the BMD density is 1 × 10 ⁹ / cm ³ , a BMD of about 40 nm or more contributes to gettering. On the other hand, in microscopic observation, BMD of 50 nm or more is detected, and only BMD that contributes to gettering can be detected. Therefore, the DZ layer width evaluated by microscopic observation is important.

Therefore, the present inventor repeated diligent studies and found that there is a correlation between the DZ layer width obtained from the oxygen concentration distribution measured by SIMS and the DZ layer width obtained from microscopic observation, and both DZ layers. We have found that the DZ layer width by microscopic observation can be quickly obtained from the DZ layer width obtained from the oxygen concentration distribution measured by SIMS by using the width calibration curve, and completed the present invention.

Hereinafter, the present invention will be described in detail with reference to the drawings as an example of an embodiment, but the present invention is not limited thereto.

Hereinafter, the method for measuring the DZ layer of the present invention will be described with reference to FIG.

In SIMS, even a small BMD is detected as a change in the measured value of oxygen concentration, and the DZ layer width is underestimated. Therefore, it is necessary to convert it into the DZ layer width observed by a microscope.

Therefore, first, the DZ layer width is measured by both SIMS and microscopic observation methods, and a calibration curve for both is prepared (see S11 in FIG. 1). At this time, since the DZ layer width is underestimated in SIMS, the two values do not match, but as the present inventor has found, there is a correlation between the two.

Specifically, a plurality of heat-treated wafers are prepared, and a calibration curve of the DZ layer width obtained by SIMS and the DZ layer width obtained by microscopic observation is created.

In the oxygen concentration measurement by SIMS, only interstitial oxygen in the wafer is detected in the region where BMD is not formed (that is, the DZ layer), whereas in the region where BMD is formed, in addition to interstitial oxygen, Since the oxygen contained in the BMD is also detected, the oxygen concentration is higher than that of the DZ layer. Therefore, it can be determined that the DZ layer has a depth to which the oxygen concentration rapidly increases and the variation in the oxygen concentration increases.

At this time, in the BMD forming region, the oxygen concentration value fluctuates depending on the size and density of the BMD. Therefore, when the measurement interval is 0.1 μm or less, the moving average of the measured values of 5 or more and 10 or less before and after the measurement depth. The determination accuracy can be further improved by using the value.

If there are 5 or more measurement points for calculating the moving average value, the measured value is sufficiently smoothed, it is possible to prevent the DZ layer from being erroneously determined, and the moving average value is calculated. If the number of measurement points is 10 or less, it is possible to prevent the points from being over-smoothed and being unable to determine the points where the oxygen concentration suddenly increases.

Further, in the surface layer of the wafer, the oxygen concentration value of SIMS is different from the oxygen concentration in the wafer due to the influence of oxygen and impurities adhering to the surface, so it is necessary to exclude this region.

Therefore, in order to obtain the DZ layer width by SIMS, for example, the oxygen concentration distribution from the wafer surface to a depth of 10 μm is measured, and the moving average value of each measurement depth is calculated. After that, in order to exclude the influence of the surface, the shallowest depth at which the difference between the measured value of oxygen concentration and the moving average value of oxygen concentration is within ± 3% of the moving average value of oxygen concentration is set as A, and this depth is set. In a region deeper than A, the depth B in which the measured value of oxygen concentration and the moving average value of oxygen concentration deviate by ± 5% or more from the moving average value of oxygen concentration can be defined as the DZ layer width.

The width of the DZ layer by microscopic observation is determined by cleaving or diagonally polishing the sample according to the conventional method, revealing BMD by selective etching, and then observing with a microscope, for example, the depth from the wafer surface to the third BMD. Just ask.

For example, consider the oxygen concentration distribution of the heat-treated wafer as shown in FIG. The error rate ((measured value-moving average value) / moving average value) between the measured value of oxygen concentration and the moving average value of oxygen concentration shown in FIG. 3 deviates by 10% or more in the surface layer. This is because the oxygen concentration value of SIMS is different from the oxygen concentration in the wafer due to the influence of oxygen and impurities adhering to the surface of the wafer surface layer, and it is necessary to exclude this region. Therefore, the shallowest depth A (0.21 μm in this example) at which the error rate is ± 3% or less is obtained.

Next, in a region deeper than depth A, the measured value of oxygen concentration and the moving average value of oxygen concentration deviate by ± 5% or more from the moving average value of oxygen concentration, and depth B (5.16 μm in this example). May be the DZ layer width. This is because in the region deeper than A, only interstitial oxygen is detected in the region where BMD is not formed (DZ layer). On the other hand, in the region where the BMD is formed, in addition to the interstitial oxygen, the oxygen contained in the BMD is also detected, so that the oxygen concentration is detected higher than that of the DZ layer. Therefore, it can be determined that the DZ layer is up to the depth at which the oxygen concentration rapidly increases and the variation in the oxygen concentration increases. Since the oxygen concentration value fluctuates depending on the size and density of the BMD, the calculation accuracy can be improved by using the moving average value.

After creating the calibration curve as described above, the DZ layer width by SIMS of the sample whose DZ layer width is to be measured is obtained, and by applying the calibration curve, the DZ layer width obtained by microscopic observation is obtained (S12 in FIG. 1). reference).
In this way, the DZ layer width obtained by microscopic observation with poor throughput can be easily obtained.
At this time, the DZ layer widths obtained by both methods have a correlation, but the measured values are different. This is because only BMD that can be visually confirmed by microscopic observation is detected together with BMD that cannot be visually confirmed by SIMS.

Further, the depth C in which the approximate curve created from the measured value of the oxygen concentration and the moving average value of the oxygen concentration deviate from the approximate curve by ± 5% or more can be set as the DZ layer width of SIMS. At that time, oxygen contained in BMD is also detected in the BMD forming region, and the oxygen concentration is higher than that in the DZ layer region. Therefore, linear approximation, exponential approximation, logarithmic approximation, polynomial approximation, exponentiation approximation, etc. are performed from the measured values from the depth A to the depth B 3/4 of the DZ layer region that is not affected by the surface. The approximate curve with the highest squared value (R ^{2) of the relational number may be used.}

Further, any DZ layer width calculation method may be used, which is a comparison between the measured value of the oxygen concentration and the moving average value of the oxygen concentration, and a comparison between the approximate curve and the moving average value of the oxygen concentration.

For example, as shown in FIG. 4, an approximate curve is created from the measured oxygen concentration in the region of 3/4 (3.57 μm in this example) of depth A to depth B. Specifically, linear approximation, exponential approximation, logarithmic approximation, polynomial approximation, and power approximation may be performed, and the approximation curve having the highest ^{correlation coefficient squared value (R 2) may be used.} After that, the depth C (4.79 μm in this example) deviated by ± 5% or more from the error rate between the moving average value of the oxygen concentration and the approximate value (value obtained from the approximate curve) in FIG. 5 is defined as the DZ layer width. It's fine.

The DZ layer of SIMS can also be determined by creating an approximate curve from the measured values of oxygen concentration and observing the difference between the approximate curve and the moving average value of oxygen concentration. However, since the oxygen concentration in the BMD forming region is higher than that in the DZ layer forming region, it is necessary to create an approximate curve only in the DZ layer region. Therefore, it is necessary to create an approximate curve from the measured values of the oxygen concentration in the region from the depth A to the depth B 3/4.

Hereinafter, the present invention will be described in more detail with reference to Examples and Comparative Examples, but the present invention is not limited thereto.

(Example)
Nitrogen concentration [N ₂ ] = 3 × 10 ¹³ / cm ³ , oxygen concentration [Oi] = 1.1 × 10 ¹⁸ atoms / cm ³ (ASTM'79) diameter 300 mm silicon wafer, defect-free layer (DZ layer) Was heat-treated in an Ar atmosphere. Specifically, the maximum temperature / holding time was set to three levels of 1200 ° C./10 min (sample 1), 1200 ° C./20 min (sample 2), and 1200 ° C./30 min (sample 3). After that, SIMS measurement was performed at a measurement interval of 0.03 μm. FIG. 6 shows the oxygen concentration distribution of Sample 3, and it can be seen that the oxygen concentration increases sharply from the vicinity of 5 μm.

Next, the moving average value of the oxygen concentration was calculated from the measured values of the oxygen concentration at 10 points before and after each measurement depth. Since the oxygen concentration in the surface layer is different from the oxygen concentration in the wafer due to the influence of the surface, this region needs to be excluded from the evaluation target. Therefore, the shallowest depth A at which the difference between the moving average value of the oxygen concentration and the measured value of the oxygen concentration is within ± 3% of the moving average value of the oxygen concentration was calculated for each sample. When the depth A was determined, it was found to be 0.24 μm in sample 1, 0.20 μm in sample 2, and 0.16 μm in sample 3 (see Table 1).

Subsequently, in the region deeper than the depth A, the depth B (DZ layer width) in which the moving average value of the oxygen concentration and the measured value of the oxygen concentration deviate by ± 5% or more from the moving average value of the oxygen concentration is set. I asked. Sample 1 was 1.94 μm, Sample 2 was 3.74 μm, and Sample 3 was 5.16 μm (see Table 1).

Further, linear approximation, exponential approximation, logarithmic approximation, polynomial approximation, and exponentiation approximation were performed at a depth from the above depth A to the above depth B to 3/4. Then, samples 1, 2 and 3 together, the linear approximation is most ^{R 2} values (square value of the correlation coefficient) was high. Then, the depth C (DZ layer width) in which the linear approximation value and the moving average value of the oxygen concentration deviated from the linear approximation value by ± 5% or more was also calculated. When the depth C was determined, sample 1 was found to be 1.79 μm, sample 2 was found to be 3.72 μm, and sample 3 was found to be 4.79 μm (see Table 1).

Next, the DZ layer width was determined by microscopic observation. Following the conventional method, the sample was diagonally polished to reveal the BMD by selective etching, and then observed with a microscope to determine the depth (DZ layer width) from the wafer surface to the third BMD (see Table 2). .. Sample 1 was 17.0 μm, Sample 2 was 41.0 μm, and Sample 3 was 64.7 μm.

Then, a calibration curve was prepared with the DZ layer widths B and C of SIMS and the DZ layer width of microscopic observation (see FIG. 7). As a result, although the measured values were different, a very good correlation was obtained.

Next, as a measurement sample, a heat-treated wafer having an unknown DZ layer width (Sample 4: maximum temperature reached / holding time: 1200 ° C./15 min) was prepared. SIMS measurement was performed, and the depth B was found to be 2.96 μm and the depth C was found to be 2.85 μm, and these were applied to the calibration curve in FIG. As a result, the width of the DZ layer observed under a microscope was determined to be 31.2 μm from B and 31.7 μm from C.
The throughput per chip was about 3 hours from vacuuming to measurement.

(Comparison example)
The sample 4 used in the examples was obliquely polished to reveal the BMD by selective etching, and then observed with a microscope to determine the depth (DZ layer width) from the wafer surface to the third BMD. As a result, it was 31.6 μm. The obtained DZ layer width was the same as that of the example, but it took about 4 hours and 30 minutes per chip from oblique polishing to microscopic observation, and the throughput was poor as compared with the example.

The present invention is not limited to the above embodiment. The above embodiment is an example, and any one having substantially the same configuration as the technical idea described in the claims of the present invention and exhibiting the same effect and effect is the present invention. It is included in the technical scope of the invention.

Claims (4)

A method for measuring the DZ layer formed on the surface layer of a silicon wafer.
A calibration curve of the DZ layer width obtained from the depth distribution of the oxygen concentration measured by SIMS and the DZ layer width obtained by microscopic observation is prepared, and based on the prepared calibration curve, it is obtained by SIMS of the measurement sample. A method for measuring a DZ layer, which comprises obtaining the DZ layer width obtained by microscopic observation from the obtained DZ layer width. In the SIMS measurement, the shallowest depth at which the difference between the moving average value of oxygen concentration and the measured value of oxygen concentration is within ± 3% of the moving average value of oxygen concentration is defined as A, and oxygen is oxygenated in a region deeper than A. The DZ according to claim 1, wherein a depth B in which the moving average value of the concentration and the measured value of the oxygen concentration deviate from the moving average value of the oxygen concentration by ± 5% or more is defined as the DZ layer width of SIMS. How to measure layers. In the SIMS measurement, the shallowest depth at which the difference between the moving average value of the oxygen concentration and the measured value of the oxygen concentration is within ± 3% of the moving average value of the oxygen concentration is defined as A, and oxygen is oxygenated in a region deeper than A. An approximate curve is created from the measured values of the oxygen concentration in the region up to 3/4 of the depth B where the moving average value of the concentration and the measured value of the oxygen concentration deviate by ± 5% or more from the moving average value of the oxygen concentration. The DZ layer width of SIMS is defined as a depth C in which the moving average value of the approximate curve and the oxygen concentration deviates from the approximate curve by ± 5% or more in a region deeper than A. The method for measuring the DZ layer according to. The DZ layer according to claim 2 or 3, wherein the moving average value of the oxygen concentration is an average value of the measured values of 5 or more and 10 or less oxygen concentrations before and after the measurement depth. Measuring method.

Images