JP2011108801A - Method for evaluation of work-affected layer - Google Patents

Abstract

The present invention provides a nondestructive inspection that enables simple and quantitative evaluation of a work-affected layer of a single crystal.
An analysis method of an X-ray rocking curve obtained from a single crystal for detecting a work-affected layer of the single crystal, wherein the work-affected layer is evaluated based on a ratio of the intensity of the skirt portion to the peak intensity. Is the method. At this time, the position of the skirt portion is set to a position where the X-ray analysis intensity is attenuated to the background level or a position away from the peak position by ± 5000 seconds.
[Selection] Figure 1

Description

  The present invention relates to a method for detecting a work-affected layer of a single crystal such as a GaN single crystal substrate used for manufacturing a blue light emitting diode or the like.

  Gallium nitride bulk wafers have been put into practical use as blue laser substrate materials, and their development as high-brightness and high-efficiency LED substrates and power device substrates has become active.

  Polishing is indispensable for use as a substrate for these devices. The purpose of the polishing process is to accurately mold the wafer into a predetermined shape, but the intended performance cannot be achieved by simply molding the wafer. That is, if a damaged layer (processed altered layer) due to processing remains on the surface of the wafer, the original physical property value of the wafer may be deteriorated or a functional layer produced on the wafer may be adversely affected. Accordingly, when polishing the gallium nitride wafer, it is desirable that the work-affected layer is removed after the final polishing process is completed.

  Patent Document 1 (WO2005 / 041283) discusses a work-affected layer of gallium nitride. Gallium nitride has a Ga surface and an N surface that have different physical and chemical properties, the easiness of polishing and etching are different, the work-affected layer is generated by polishing, and the work-affected layer is dry. It is disclosed that it can be removed by etching. On the other hand, it is not described how to determine the presence or absence of a work-affected layer.

  Although there is no literature that clearly describes a method for directly measuring a work-affected layer on the surface of a gallium nitride single crystal, X-ray irradiation is performed on the gallium nitride single crystal layer, and the obtained X-ray rocking curve is half value. A method for measuring the crystal quality of a gallium nitride single crystal layer from the full width is known (Patent Document 2: Japanese Patent No. 3184717).

  Patent Document 3 (Patent No. 2894154) discloses a method for measuring a work-affected layer of a silicon single crystal wafer. A silicon wafer is wet etched, and the number of etch pits generated on the surface is measured. In the work-affected layer, more etch pits are generated than in other portions, and therefore the depth of the work-affected layer can be estimated.

  In Patent Document 4 (Japanese Patent Laid-Open No. 2001-296118), a part of the surface of a silicon wafer including a work-affected layer is removed in an arc shape and treated with an acid solution, whereby a work-affected layer that appears in this part is disclosed. A method for observing the film with a microscope and estimating its depth is disclosed.

WO2005 / 041283 Japanese Patent No. 3184717 Patent No. 2894154 JP2001-296118

  In Patent Document 2 (Patent 3184717), the crystal quality of the gallium nitride single crystal layer is measured from the full width at half maximum of the X-ray rocking curve. The presence and properties of the work-affected layer on the surface of the bulk gallium nitride single crystal are measured. It is not specified to measure.

  In Patent Document 3 (Patent No. 2894154) and Patent Document 4 (Japanese Patent Laid-Open No. 2001-296118), the work-affected layer on the silicon surface is measured, both of which are based on chemical etching. The method using chemical etching is a destructive inspection in which etch pits are generated on a silicon wafer, and an actual product cannot be directly measured. Moreover, since it is judgment by visual observation, there exists a possibility of depending on an observer. Moreover, since it is necessary to heat and use a strong acid, it is bad for the environment.

  An object of the present invention is to provide a nondestructive inspection that can easily and quantitatively evaluate a work-affected layer of a single crystal.

  The present invention detects a work-affected layer of a single crystal based on the ratio of the intensity of the skirt portion to the peak intensity of the X-ray rocking curve obtained from the single crystal. It concerns the method.

  The half width of the X-ray rocking curve obtained by irradiating the surface of the single crystal with X-rays reflects the crystal quality of the single crystal. If there is a work-affected layer on the surface of the single crystal, the surface is usually mosaic. That is, an amorphous layer or a polycrystalline layer is generated, and a mosaic layer remains inside thereof. If a polycrystalline layer is formed, the X-ray rocking curve becomes broad and the half-value width seems to increase.

  On the other hand, as a result of further studies by the present inventors, it has been found that there is an information source different from the information originating from the amorphous layer and the polycrystalline layer in the X-ray rocking curve. This is the discovery of the inventor.

  That is, the present inventor introduced a work-affected layer on the surface by subjecting a single crystal to surface grinding. At this stage, the full width at half maximum of the X-ray rocking curve is increased, and the base of the X-ray rocking curve is greatly expanded. When considered normally, the X-ray rocking curve represents the crystallinity of the surface of the single crystal. Therefore, it is considered that if the half width of the X-ray rocking curve is reduced, the spread of the skirt is reduced.

The inventor tried to lapping the single crystal after this surface grinding with diamond abrasive grains to reduce the surface irregularities and remove the polycrystalline layer on the surface. When the half width of the X-ray rocking curve was measured after the lapping, it was smaller than before the lapping. This is presumably because the amorphous layer or polycrystalline layer introduced to the surface by the surface grinding process was partially removed, so that the crystal quality was improved and the half width was reduced.

  However, it has been found that the bottom of the peak is rather high in spite of the fact that the full width at half maximum of the X-ray rocking curve is reduced. In other words, despite the improved crystallinity of the surface of the single crystal, a peak due to another defect has occurred, and the X-ray rocking curve includes information other than the crystal quality of the surface. It means that

  As a result of further study of this phenomenon by using the reciprocal lattice map, the present inventor has found that the increase in the skirt height of the X-ray rocking curve introduces and increases the strain layer below the polycrystalline layer rather than the crystal quality itself. I found out that it was caused by

  That is, as shown in FIG. 1, an amorphous layer, a polycrystalline layer, and a mosaic layer are generated on the surface of the work-affected layer due to disorder of crystallinity caused by application of mechanical stress. Since these are accompanied by deviations in crystal orientation when viewed microscopically, the diffraction angle of X-rays deviates from the original angle, and tends to be directly reflected in the half-value width of the X-ray rocking curve. On the other hand, a stress transition layer (strained layer) may be generated and remain between the mosaic layer and the complete crystal layer. When this strain is large, a crack is generated on the strained layer.

  It has been found that when such a strained layer is introduced, the skirt height of the X-ray rocking curve increases. This is considered to be because the crystal lattice is deformed by the stress applied during processing such as lapping, and the diffraction conditions are slightly changed due to slight variations in the lattice constant. For example, in the above-described example, it is considered that a strained layer having a slightly changed c-axis length is introduced due to the stress applied in the lateral direction in order to remove the polycrystalline layer by lapping. Therefore, in the same crystal system, if the height of the bottom of the X-ray rocking curve is measured, the existence and degree of the strained layer can be quantitatively measured. By using this method, it is possible to non-destructively inspect the presence and degree of the work-affected layer on the single crystal surface.

It is a schematic diagram which shows the structure of a typical process-affected layer. The arrow represents the specific orientation of the crystal. There is actually no clear boundary between each layer. It is a graph which shows the relationship between the surface processing method of a gallium nitride single crystal, and the X-ray rocking curve obtained from the gallium nitride single crystal. The diffraction intensity (Intensity) is normalized with the peak intensity set to 1. It is a graph which shows the relationship between the surface processing method of a gallium nitride single crystal, and the X-ray rocking curve obtained from the gallium nitride single crystal. The diffraction intensity (Intensity) is normalized with the peak intensity set to 1.

(Single crystal)
As single crystals targeted by the present invention, semiconductor single crystals such as silicon, GaAs, GaP, InP, SiC, and group III metal nitrides, oxide single crystals such as lithium niobate, tantalum niobate, quartz, and sapphire Halide single crystals such as CaF2 and NaI are preferred. Particularly preferable examples include gallium nitride, aluminum nitride, indium nitride, and mixed crystals (AlInGaN) made of these.

(X-ray rocking curve)
An X-ray rocking curve is a curve plotting diffraction intensity obtained when an X-ray diffractometer is used to rotate the single crystal around the diffraction center axis while irradiating the single crystal with X-rays.
In the present invention, preferably, an X-ray diffraction method using four crystals is used. In this method, the lattice constant of a single crystal is accurately evaluated, and the completeness of the crystal is evaluated from the half width. Specifically, X-rays are made highly monochromatic by a monochromator composed of four (two sets) crystals and irradiated to a sample, and an X-ray peak diffracted from this sample is obtained.

As the X-ray source, a characteristic X-ray using a Cu, W, Mo, Co, Fe, or Cr tube as a light source is applicable, but a CuK ray or a MoK ray is preferable, and Cukα ₁ is particularly preferable.

  Although tube voltage is not limited, 20-60 kV is preferable, for example, 40 kV.

  The tube current is not limited, but is preferably 30 to 50 mA, for example 40 mA.

  As a monochromator crystal for monochromatization, Ge (220), Ge (440), and Ge (111) are preferable, and Ge (220) is particularly preferable.

  Moreover, what diffraction peak to measure among the single crystals to be measured may be appropriately determined using knowledge of X-ray diffraction.

  For example, in the case of a silicon single crystal, it is preferable to measure diffraction peaks of (100) and (111). In the case of a nitride single crystal such as GaN, AlN, or AlGaN, it is preferable to measure the diffraction peaks of (0002) (10-12) (10-10).

(Half width of X-ray rocking curve)
The half width in the present invention is the full width at half-maximum (FWHM) of the obtained X-ray rocking curve. The full width at half maximum is the full width at half maximum at 50% of the peak intensity.

(Ratio of the intensity of the tail to the peak intensity of the X-ray rocking curve)
The peak intensity is the intensity at the peak position of the X-ray rocking curve. The position at which the intensity of the skirt portion is measured is preferably a position where the diffraction intensity attenuates to the background level when an X-ray rocking curve of a single crystal in which a work-affected layer is not generated is measured. In a preferred embodiment, the intensity of the skirt portion of the X-ray rocking curve is an average value of the intensity at a position that is +4900 to 5100 seconds away from the peak position and the intensity at a position that is −4900 to 5100 seconds away. Typically, it is an average value of the intensity at a position separated by +5000 seconds from the peak position and the intensity at a position separated by -5000 seconds.

  It can be seen that when this ratio is high, a work-affected layer, particularly a work strain layer, is generated. Further, the degree of the processed strain layer can be known by comparing the numerical values of the ratios.

  X-ray rocking curve measurement (0002 reflection) of a GaN wafer (unpolished) by a commercially available vapor phase method was performed. After that, surface grinding with a horizontal grinding machine (grinding stone # 600), lapping with a diamond lapping machine (abrasive grains 3μm, 0.5μm), CMP processing (colloidal alumina), dry etching processing (chlorine-based gas) are performed for each process. X-ray rocking curve was measured. The measurement results are shown in FIGS. In FIG. 2, the vertical axis is plotted as a linear axis, and in FIG. 3, the vertical axis is plotted as a logarithmic axis.

  Table 1 shows the X-ray rocking curve half-width and the reflection intensity at the bottom. The half width of the X-ray rocking curve is the half width at 50% of the peak intensity, and the intensity of the skirt portion is an average value of the intensity at a position ± 5000 seconds away from the peak position. From FIG. 3, the X-ray diffraction intensity in the absence of the work-affected layer and the stress-strained layer is at the same level as the background at a position away from the peak position by ± 5000 seconds.

A schematic diagram of a work-affected layer that is expected to occur is shown in FIG.
The full width at half maximum once increased by surface grinding, but decreased as the process progressed, and it was found that CMP polishing returns to the same state as unpolished. This suggests that surface grinding introduces a polycrystalline layer and a mosaic layer that affect the half width, but is almost removed by diamond wrap and CMP. On the other hand, the strength of the hem increased with diamond wrap rather than surface grinding. This was different from the previous expectation, and it was considered that a compressive stress was introduced in the in-plane horizontal direction by the diamond wrap, and a stress-strained layer with a slightly increased c-axis was generated. This strain was not recovered by CMP, but was finally recovered by dry etching.

  As described above, when the strength of the skirt portion of the X-ray rocking curve is increased, it indicates that the processing strain layer is introduced. When the strength is decreased, the introduced processing strain layer is removed or decreased. I understand that.

Claims (5)

  A method for detecting a work-affected layer of a single crystal, comprising: detecting the work-affected layer of the single crystal based on a ratio of the intensity of the skirt portion to the peak intensity of an X-ray rocking curve obtained from the single crystal.   The method according to claim 1, wherein the work-affected layer is a strained layer.   The position of the skirt portion is a position where the X-ray diffraction intensity is attenuated to a background level when an X-ray rocking curve of a single crystal without the work-affected layer is measured. the method of.   The method according to claim 1, wherein the position of the skirt portion is a position that is ± 5000 seconds away from the peak position.   The method according to claim 1, wherein the single crystal is a group III nitride single crystal.

Images