KR102381348B1 - Non-destructive analysis for TSD and TED defect of silicon carbide wafers - Google Patents

Abstract

The present invention relates to a nondestructive analysis method for TSD and TED defects of silicon carbide wafers. The nondestructive analysis method for TSD and TED defects of silicon carbide wafers comprises: a step of scanning a silicon carbide wafer by using a laser of a PL device; a step of image-processing a PL detection signal acquired by the laser scanning with a band pass filter; and a step of matching a PL intensity distribution graph and an image to determine TSD and TED defects. The steps control laser scanning step intervals in a-A um (0＜a＜A) when the laser line width is A um to adjust the image resolution of TSD and TED defects. According to the present invention, the nondestructive analysis method for TSD and TED defects of silicon carbide wafers can control acquisition image resolution while classifying crystal detects existing on an epitaxial layer by using silicon carbide and identifying distributions and density in a nondestructive manner.

Description

Non-destructive analysis for TSD and TED defect of silicon carbide wafers

The present invention relates to a non-destructive analysis of TSD (threading screw dislocation) and TED (threading edge dislocation) defects of silicon carbide wafers, and more particularly, to silicon carbide (SiC), a wideband gap semiconductor material. A simple, non-destructive method for performing TSD and TED material defects.

Silicon carbide (SiC), a wideband gap semiconductor material, has difficulties in realizing theoretical properties due to material defects present in silicon carbide despite its excellent physical properties.

Silicon carbide material quality can be divided into several categories: 1) Wafer shape (bow, warp, total thickness variation (TTV), etc.) and scratches during wafer processing in single crystal ingots, 2) Concentration and thickness uniformity during epi layer formation It can be broadly divided into gender and 3) crystal defects, and each of 1) and 2) can be improved through process optimization.

Crystal defects of silicon carbide are generated from the single crystal ingot growth stage to the epitaxial layer growth.

Since crystal defects are a major cause of yield reduction due to deterioration in characteristics during device fabrication, it is essential to fabricate devices using high-quality silicon carbide.

Also, in the device manufacturing stage, material quality inspection is essential to predict device yield based on material quality. Crystal defect analysis of silicon carbide can be largely divided into destructive analysis and non-destructive analysis.

The most widely used destructive analysis method is the etching method.

When silicon carbide is etched using an alkali molten salt such as potassium hydroxide (KOH), there is a difference in the etching rate between the defective and non-defective areas.

At this time, an etch pit is formed where there is a crystal defect, the type of the defect can be identified through the shape, and the defect density can be known through the number of etch pits.

On the other hand, types of defects and Burgers vectors can be analyzed using transmission electron microscopy (TEM), but only micro-regions of the entire wafer can be analyzed.

The non-destructive analysis method is an analysis method that can be used efficiently before device fabrication because the specimen can be reused after crystal defect analysis.

When the wafer is observed using a polarization microscope, it is possible to analyze the stress and high stress defects present in the wafer, but detailed observation of the defects causing small stress is difficult.

In addition, by using X-ray topography (XRT), it is possible to analyze the types of defects and Burgers vector in the wafer. In the case of the XRT method, the efficiency of analysis is not high because it is analyzed using a radiation accelerator.

It can be analyzed using lab source equipment, but it has a disadvantage in that it takes a relatively long time to measure.

Commercially available silicon carbide material analysis equipment optically observes the wafer surface and mainly analyzes surface defects having a shape. It is easy to observe defects such as morphological defects and triangular defects containing 3C polytypes.

However, it is not possible to analyze defects such as stacking fault (SF), basal plane dislocation (BPD), threading screw dislocation (TSD), and threading edge dislocation (TED) that do not show geometry.

Photoluminescence (PL) is a commercially available material analysis equipment, and PL equipment is specialized for SF type analysis.

Recently, the PL function is added to the surface observation and analysis equipment to perform the analysis.

Many evaluations have been made on the effect of the defects in the surface shape and the SF defects on the characteristics of the device.

Material growth technology to reduce these defects is also being developed. In order for silicon carbide devices to be applied to places closely related to our lives, such as electric vehicles and railroads, device characteristics and device reliability are very important.

In the case of TSD and TED defects, it is known that there is no significant effect on the diode device.

However, it is pointed out as a cause of increasing the leakage current of devices such as metal-oxide semiconductor field-effect transistor MOSFETs, which ultimately deteriorates device reliability.

International publication WO2016121628 The defect inspection method and defect inspection apparatus for wide-gap semiconductor substrates are inventions proposed as an effort to solve such problems using PL devices, but they leave technical problems with respect to the resolution of defect images. there is an invention

Republic of Korea Application No. 10-2019-7001018 Defect inspection apparatus for wide gap semiconductor substrate is a technology proposed to solve the resolution problem of the international publication WO2016121628, but adopts a configuration that improves the acquired image using an optical system, and the original image There was a problem that could not be solved if it had a resolution problem at the source.

International publication WO2016121628 Defect inspection method and defect inspection apparatus for wide-gap semiconductor substrates Korean Application No. 10-2019-7001018 Defect Inspection Device for Wide Gap Semiconductor Substrate

The present invention is to solve the above problems, and uses silicon carbide to classify crystal defects existing in the epitaxial layer and non-destructively grasp the distribution and density while controlling the acquired image resolution TSD and It aims to provide a non-destructive analysis method for TED defects.

In order to achieve the above object, the present invention uses a laser of a PL device on a silicon carbide wafer, and when the laser line width is Aum (0<A), the laser scanning step interval is set from a to A um (0<a<A). controlling the laser scanning; image processing the PL detection signal obtained by the laser scanning with a bandpass filter; The technical gist is a non-destructive analysis of TSD and TED defects of a silicon carbide wafer, characterized in that the TSD and TED defects are determined by matching the PL intensity distribution graph and the image;

delete

In addition, the laser scanning is performed by scanning with a laser scanning step interval of 1 um using a 355 nm, 25 mW, 5 um width laser of the PL device, and the band pass filter consists of a 382 ± 15 nm band pass filter. TSD and TED defect image resolution is judged as a TSD defect when the PL intensity is in the range of 69.0±5% compared to the non-defect area, and a TED defect when the PL intensity is in the range of 78.0±5% compared to the non-defect area. It is preferable to become a TSD and TED defect non-destructive analysis method of the silicon carbide wafer characterized in that it is.

In addition, the laser scanning is performed by scanning at a laser scanning step interval of 5 μm using a 355 nm, 25 mW, 5 μm wide laser of the PL device, and the band pass filter consists of a 382 ± 15 nm band pass filter. TSD and TED defect image resolution is determined as a TSD defect when the PL intensity is in the range of 86.8 to 92.0% compared to the defect-free area, and as a TED defect when the PL intensity is in the range of 92.6 to 96.7% compared to the non-defect area. It is preferable to become a TSD and TED defect non-destructive analysis method of the silicon carbide wafer characterized in that it is.

In addition, the TSD and TED defect non-destructive analysis method of the silicon carbide wafer of the present invention is a TSD and TED defect non-destructive analysis method of a silicon carbide wafer, characterized in that the TSD and TED defects are displayed and output on the matched image after TSD and TED defects are determined. It is preferable to be

According to the present invention, a non-destructive analysis method for TSD and TED defects of a silicon carbide wafer that can control the acquired image resolution while classifying crystal defects existing in the epitaxial layer using silicon carbide and non-destructively grasping the distribution and density is provided. There is an advantage to be

A comparison diagram of a defect image obtained with a 1um laser scanning step interval in the method of the present invention in FIG. 1 and a defect image obtained by X-ray topography (XRT) analysis ((a) is a 382±15nm bandpass of silicon carbide epitaxial layer defects TSD and TED type analysis image of PL result in filter (BPF), (b) is TSD and TED classification through X-ray topography (XRT) analysis of the same area (TSD: red circle, TED: blue circle)
2 is a PL intensity distribution graph according to FIG. 1 (TSD: red, TED: blue, background (defect-free region): gray)
Comparison of defect images acquired at 5um laser scanning step interval in the method of the present invention in FIG. 3
4 is a PL intensity distribution graph according to FIG. 3 (TSD: red, TED: blue, background (defect-free region): gray)
5 is a graph of resolution change according to the laser scanning step interval

Hereinafter, the present invention will be described with reference to the drawings, and in the description of the present invention, if it is determined that a detailed description of a related known technology or configuration may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. will be.

And, the terms described below are terms defined in consideration of functions in the present invention, which may vary depending on the intention or custom of a user or operator, and thus definitions should be made based on the content throughout this specification describing the present invention.

The present invention is a method for determining the density of crystal defects present in an epitaxial layer in a non-destructive way to improve device reliability.

In general, wafer suppliers provide information by identifying crystal defects of silicon carbide on a substrate using an etching method. BPD, TSD, TED obtained by chemically etching one of the substrate wafers obtained from a single crystal ingot through molten salt. Find the defect density through chippit.

In addition, it is determined that the wafer obtained from the same ingot is identical to the defect density obtained from the etching result of one substrate.

However, the distribution of defects in the ingot may vary during growth, and defects having epitaxial shapes such as droplet, particle, and triangular defects may occur during epi layer growth. It should be controlled during growth.

In addition, in order to improve the reliability of a device using silicon carbide, it is important to understand the BPD, TSD, and TED present in the epitaxial layer.

More than 95% of BPD is converted to TED when the epitaxial layer is grown, and most of TSD and TED are propagated to the epitaxial layer.

In this way, knowing the BPD, TSD, and TED defect levels of the substrate, the level of defects in the epitaxial layer can also be predicted.

However, the dislocation density provided by the wafer supplier is the defect density of the substrate of the same ingot, not the defect density of the substrate manufactured on the device.

In addition, the shape of the distribution of defects cannot be known only by the defect density.

Therefore, in order to know the effect on the characteristics and reliability of the device, it is important to know the defect information at the location where the device is to be manufactured in the epitaxial layer.

The present invention relates to a method of grasping defect information at a location where a device in an epitaxial layer is to be manufactured in order to know the effect on the characteristics and reliability of the device.

Hereinafter, the present invention will be described.

The present invention provides a step of scanning a silicon carbide wafer using a laser of a PL device, processing an image of a PL detection signal obtained by laser scanning with a bandpass filter, and matching the image with a PL intensity distribution graph for TSD and TED defects is made up of steps to determine

Among the non-destructive analysis methods of silicon carbide, the PL analysis method generates electron-hole pairs by irradiating energy (laser or ultraviolet) larger than the band gap of silicon carbide, and analyzes the energy emitted when recombination to determine the type of defect. It is an analysis method to determine

The PL analysis method was mainly used to analyze the SF and triangular defects present in the silicon carbide epitaxial layer. The method to obtain the energy emitted during recombination as a spectrum through a spectrometer and to observe only a specific energy using a filter that absorbs or transmits a specific wavelength There are two ways to do it.

When obtaining spectral results through a spectrometer, it takes a lot of time for analysis, but the PL mapping method, which detects specific energy using a filter, has a short analysis time and can obtain results for the entire wafer quickly. It is an effective method to determine the defect density.

However, looking at the result of the PL intensity at a wavelength including 390 nm, which corresponds to the bandgap energy of silicon carbide, when the wavelengths indicated by the defects are different, the PL intensity appears low, and through this, the defect distribution can be grasped.

In addition, the types of detailed defects can be distinguished by understanding the wavelength band.

The present invention is characterized in that the TSD and TED defects are displayed and output on the mapped image using the point that the distribution and types of the defects can be identified.

In detail, since TSD and TED defects are not crystal defects having different polytypes, they are not confirmed in the filter results of wavelengths other than those including 390 nm.

The presence of TSD and TED defects in the epi layer causes stress around them, which is proportional to the square of the size of the Burgers vector of the defect.

Since the Burgers vector of the silicon carbide TSD defect is 1c (c=10.053 Å) and the Burgers vector of the TED defect is 1/3<11-20> (3.70 Å), the surrounding stress is about 7 times or more.

The present invention is a method of distinguishing between TSD and TED existing in the epitaxial layer through PL analysis and calculating the density using this principle.

According to the present invention, there is an advantage in that the influence on device characteristics and performance can be grasped before device fabrication through the location of TSD and TED.

Meanwhile, in the present invention, when the laser line width of the PL device is Aum, the laser scanning step interval is controlled from a to A um (0 < a < A) to adjust the TSD and TED defect image resolution.

Hereinafter, the present invention will be described in detail with reference to Examples.

<Example 1>

A silicon carbide wafer with an epitaxial layer thickness of 40 um and a doping concentration of 8E14 cm ^-3 was analyzed using a 382±15 nm bandpass at 355 nm with a laser line width of 5 μm, with a 1 μm laser scanning step interval of a 25 mW PL device. The results are shown in Figure 1 (a).

FIG. 1( b ) is an XRT analysis result for confirming a TSD defect and a TED defect in the same region to confirm the reliability of the result of FIG. 1( a ).

XRT analysis is a well-known method for analyzing TSD defects and TED defects in the prior art. As shown in FIG. 1(b), the stress caused by the defect is observed as a bright color in a round shape. A TSD defect having a larger size according to the observed size is a TSD defect, and TED The flaw is that it is small in size.

Comparative analysis of the present invention In FIG. 1(a), TSD is indicated by a red circle and TED is indicated by a blue circle. Compared with FIG. .

In addition, it can be seen that FIG. 1(a) is more precise in the size of the defect (which can be interpreted as a resolution indicating the exact location of the defect) and the overall resolution of the screen compared to FIG. 1(b).

FIG. 2 is a graph showing the PL intensity at defect locations identified by TSD and TED using the XRT result.

In FIG. 2 , it can be seen that the background is the PL intensity of the region without defects, and the PL intensity of the TSD and TED defects is lower than that of the region without the defect.

According to FIG. 2, it can be seen that the two defects are clearly distinguished by the difference in PL intensity. Dislocation densities of TSD and TED in the region of FIG. 1(a) were 8,739 and 11,959/cm ² .

As described above, it can be seen that when the PL device is used, the distribution and density of crystal defects can be obtained by TSD and TED using non-destructive analysis.

<Example 2>

PL analysis of a silicon carbide wafer with an epi layer thickness of 40 um and a doping concentration of 8E14 cm ^-3 with a laser line width of 5 um at 355 nm, with a 5 um laser scanning step interval of a 25 mW PL device, using a 382±15 nm bandpass did.

3 is a result obtained using a bandpass filter of 382±15 nm. The PL intensity is higher where the light is brighter, and where the dark color is lower.

Based on Example 1, the PL analysis region of FIG. 3(a) was divided into TSD and TED as shown in FIG. 3(b) based on the PL intensity.

In Fig. 3(b), the red circle is the TSD defect, and the blue circle is the TED defect.

4, the PL intensity distribution of each defect is shown for each defect type.

Also in FIG. 4, it can be seen that the analysis of the two defects is possible through the difference in PL intensity between TSD and TED, and the dislocation densities of TSD and TED were 1,900 and 18,900/cm ² .

Table 1 shows the results of comparing the PL intensities indicated by TSD and TED with the PL intensities of regions without defects according to the measurement steps during PL measurement based on the above two examples.

Referring to FIG. 4, the laser scanning step interval is an important factor affecting the analysis time. When the laser scanning step interval is 5 μm, the TSD defect has a range of 86.8 to 92.0% compared to the PL intensity of the area without the defect, TED defects range from 92.6 to 96.7%.

When the laser scanning step interval is 1um, TSD and TED are 69.0% and 78.0%, respectively. The measurement time is related to the throughput of the equipment, and it is possible to clearly distinguish crystal defects even when measuring with a 5um laser scanning step interval.

It is understood that this is because, when the laser scanning step interval is smaller than the laser pulse width, the difference in PL intensity between the background and the defect becomes larger as the overlap scanning section is formed. The larger the PL intensity difference appears, the clearer the acquired defect image is against the background.

This defect image sharpness means that it is expressed as a precise dot image closer to the actual defect location as shown in FIG. 5 .

The drawings shown above for the purpose of explanation of the present invention are one embodiment in which the present invention is embodied, and as shown in the drawings, it can be seen that various types of combinations are possible in order to realize the gist of the present invention.

Therefore, the present invention is not limited to the above-described embodiments, and as claimed in the following claims, anyone with ordinary skill in the art to which the invention pertains can implement various modifications without departing from the gist of the present invention. It will be said that there is the technical spirit of the present invention to the extent possible.

Claims (5)

laser scanning by controlling the laser scanning step interval from a to A um (0 < a < A) when the laser line width is Aum (0<A) using a laser of a PL device on a silicon carbide wafer;
image processing the PL detection signal obtained by the laser scanning with a bandpass filter;
Matching the PL intensity distribution graph and the image to determine the TSD and TED defects; as
TSD and TED defect non-destructive analysis method of silicon carbide wafer, characterized in that the TSD and TED defect image resolution is adjusted. The method of claim 1, wherein the laser scanning
It is made by scanning with a laser scanning step interval of 1 μm using a 355 nm, 25 mW, 5 μm wide laser of the PL device,
The bandpass filter consists of a 382±15nm bandpass filter when processing the image,
TSD and TED defects image resolution is
If the PL intensity is in the range of 69.0±5% compared to the non-defective area, it is judged as a TSD defect,
TSD and TED defect non-destructive analysis method of silicon carbide wafer, characterized in that it is judged as a TED defect when the PL intensity is in the range of 78.0±5% compared to the defect-free area. The method of claim 1, wherein the laser scanning
It is made by scanning with a laser scanning step interval of 5um using a 355 nm, 25 mW, 5um width laser of the PL device,
The bandpass filter consists of a 382±15nm bandpass filter when processing the image,
TSD and TED defects image resolution is
If the PL intensity is in the range of 86.8~92.0% compared to the area without defects, it is judged as a TSD defect,
TSD and TED defect non-destructive analysis method of silicon carbide wafer, characterized in that it is judged as a TED defect when the PL intensity is in the range of 92.6 to 96.7% compared to the defect-free area. 4. The method of claim 2 or 3, wherein the TSD and TED defect non-destructive analysis of the silicon carbide wafer is
After judging TSD and TED defects
A non-destructive analysis method of TSD and TED defects of a silicon carbide wafer, characterized in that the TSD and TED defects are displayed on the matched image and output. delete

Images