Method for evaluating silicon wafers and method for removing layers affected by processing of silicon wafers
The silicon wafer evaluation method using Raman spectroscopy and peak shift analysis allows precise estimation of process-affected layer depth, ensuring complete removal and improved productivity by optimizing machining allowances.
Patent Information
- Application Number
- JP2021185360
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-11-15
- Publication Date
- 2024-11-13
- Estimated Expiration
- 2041-11-15
AI Technical Summary
Existing methods fail to accurately estimate the maximum depth of process-affected layers on silicon wafers, leading to incomplete removal and reduced productivity due to insufficient machining allowances.
A silicon wafer evaluation method using Raman spectroscopy to generate a peak shift histogram, calculate average and standard deviation, and estimate the maximum depth of process-affected layers, followed by etching or polishing beyond this depth to ensure complete removal.
Accurately estimates the maximum depth of process-affected layers, enabling reliable removal and improving productivity by avoiding excessive machining, thus meeting defect standards and enhancing yield.
Smart Images

Figure 0007582161000001 
Figure 0007582161000002
Abstract
Description
[Technical field]
[0001] The present invention relates to a method for evaluating a silicon wafer and a method for removing a process-affected layer from a silicon wafer. [Background technology]
[0002] In the manufacture of silicon wafers, the process of processing using a grinding mechanism such as slicing and lapping has the main role of forming the shape of the wafer, such as warpage and thickness, while it also leaves a processing-induced layer (damage) in the bare silicon wafer, which is a layer in which stress due to processing remains, generating cracks and streaks. These processing-induced layers are removed by subsequent processes such as etching and polishing, but it was difficult to confirm the required removal amount other than by actually etching or polishing and evaluating the quality. For example, as described in Patent Document 1, SEM observation from angle polishing is listed as a conventional method, but it was not possible to accurately estimate the sample depending on the position of the wafer from which the sample was cut, changes occurred due to the pretreatment of the sample, and the observation field was narrow, so that deep damaged areas were overlooked. Therefore, a single and local collection method such as the method described in Patent Document 1 was not desirable.
[0003] Meanwhile, a method for non-destructively analyzing a target sample using Raman spectroscopy has been proposed.
[0004] For example, Patent Document 2 describes a method for increasing sensitivity in analyzing foreign matter adhering to the surface of a wafer using micro-Raman spectroscopy by shifting the focal position of the objective lens above or below the sample to be analyzed.
[0005] Patent Document 3 describes a method for determining the pass / fail level for continued use by measuring and comparing the Raman spectra of the surface, crystalline layer, and non-crystalline layer of a quartz crucible.
[0006] Patent Document 4 describes a method for measuring the width of terraces on a flat surface by using a peak shift of Raman scattered light.
[0007] Patent Document 5 describes a method for measuring strain (damage) in a silicon thin film on an SOS (silicon-on-sapphire) substrate using a peak shift method of Raman scattered light after peeling and polishing, and describes the correlation between strain (damage) and peak shift.
[0008] Patent Document 6 describes a method for measuring the strain level in a strained silicon thin film on a strained silicon-on-insulator (SSOI) substrate by using a peak shift by Raman spectroscopy. [Prior art documents] [Patent documents]
[0009] [Patent Document 1] JP 2003-062740 A [Patent Document 2] JP 2017-083369 A [Patent Document 3] JP 2019-119618 A [Patent Document 4] JP 2008-109012 A [Patent Document 5] JP 2011-138932 A [Patent Document 6] Special Publication No. 2009-503907 Summary of the Invention [Problem to be solved by the invention]
[0010] However, the above-mentioned method is not a method capable of accurately estimating the maximum depth of the process-affected layer remaining in a bare silicon wafer.
[0011] Therefore, there has been a need to develop an evaluation method that can accurately estimate the maximum depth of the process-induced damage layer remaining in a bare silicon wafer.
[0012] The present invention has been made to solve the above problems, and has an object to provide a silicon wafer evaluation method capable of accurately estimating the maximum depth of a process-damaged layer remaining in a bare silicon wafer, and a silicon wafer process-damaged layer removal method capable of reliably removing the process-damaged layer from a bare silicon wafer on which the process-damaged layer remains. [Means for solving the problem]
[0013] In order to solve the above problems, the present invention provides a method for evaluating a silicon wafer, comprising the steps of: Obtaining first-order Raman peak positions of silicon using a Raman spectroscopic microscope at multiple points within a bare silicon wafer surface on which a processing-affected layer remains; generating a histogram of peak shifts from the first-order Raman peak positions of the silicon obtained at the multiple locations; Calculating an average value A and a standard deviation S from the histogram; From the average value A and the standard deviation S, a maximum depth D of a process-affected layer remaining in the silicon wafer is estimated. The present invention provides a method for evaluating a silicon wafer, comprising:
[0014] Such a silicon wafer evaluation method of the present invention makes it possible to accurately and non-destructively estimate the maximum depth of a process-affected layer remaining in a bare silicon wafer (particularly a wafer after lapping or grinding). Based on this estimate, an appropriate amount of etching and / or polishing removal can be determined.
[0015] It is preferable to estimate the maximum depth D of the process-affected layer using A+3S as an index.
[0016] By using the above index, the maximum depth of the work-affected layer can be estimated more accurately.
[0017] For example, the silicon wafers having a surface roughness of 1 nm or more in a 100 μm×100 μm field of view can be evaluated.
[0018] The object to be evaluated is not particularly limited as long as it is a bare silicon wafer with a residual processing-affected layer. For example, a silicon wafer with a surface roughness of 1 nm or more in a 100 μm×100 μm field of view can be evaluated.
[0019] The wavelength of the light source used in the Raman spectroscopic microscope is preferably 532 nm.
[0020] The maximum depth D of the process-affected layer can be estimated more accurately based on a histogram based on the first-order Raman peak position of silicon obtained using a Raman spectroscopic microscope with a light source wavelength of 532 nm.
[0021] It is preferable that the number of points from which the first-order Raman peak positions of silicon are obtained in order to calculate the average value A and the standard deviation S is 200 or more.
[0022] By obtaining the first-order Raman peak position of silicon from 200 or more points, the maximum depth D of the process-affected layer can be estimated more accurately.
[0023] The present invention also provides a method for removing a process-affected layer from a silicon wafer, comprising the steps of: The present invention provides a method for removing a process-affected layer of a silicon wafer, comprising the steps of: etching and / or polishing the silicon wafer with a removal amount exceeding a maximum depth D estimated by the silicon wafer evaluation method of the present invention.
[0024] By subjecting the silicon wafer to etching and / or polishing with a removal amount exceeding the maximum depth D estimated by the silicon wafer evaluation method of the present invention, it is possible to reliably remove the process-affected layer remaining on the bare silicon wafer. By reliably removing the process-affected layer, it is possible to obtain a silicon wafer that satisfies a standard for the number of localize light scatter (LLS) defects after etching and / or polishing, for example. Effect of the Invention
[0025] As described above, the method for evaluating a silicon wafer of the present invention makes it possible to accurately estimate the maximum depth of a process-affected layer remaining in a bare silicon wafer.
[0026] Furthermore, the method for removing a process-affected layer from a silicon wafer according to the present invention can reliably remove the process-affected layer from a bare silicon wafer on which the process-affected layer remains. In addition, since there is no need to remove more than necessary, this contributes to improving productivity and yield. [Brief description of the drawings]
[0027] [Figure 1] 1 is an example of a histogram used in the silicon wafer evaluation method of the present invention. [Diagram 2] 1 is a graph showing the relationship between the polishing stock removal rate and the average number of defects in Examples 1 and 2 and Comparative Examples 1 and 2. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0028] As described above, there has been a need for the development of a silicon wafer evaluation method capable of accurately estimating the maximum depth of a process-damaged layer remaining in a bare silicon wafer, and a silicon wafer process-damaged layer removal method capable of reliably removing the process-damaged layer from a bare silicon wafer on which the process-damaged layer remains.
[0029] As a result of intensive investigations into the above-mentioned problems, the inventors have found that it is possible to accurately estimate the maximum depth D of a processing-affected layer remaining in a bare silicon wafer to be evaluated by obtaining first-order Raman peak positions of silicon using a Raman spectroscopic microscope at multiple locations within the surface of a bare silicon wafer on which a processing-affected layer remains, generating a histogram of peak shifts from the first-order Raman peak positions of silicon obtained at the multiple locations, calculating the average value and standard deviation by regarding the generated histogram of peak shifts as a normal distribution, and estimating the maximum depth D of the processing-affected layer based on the calculated average value and standard deviation, thereby completing the present invention.
[0030] That is, the present invention is a method for evaluating a silicon wafer, comprising the steps of: Obtaining first-order Raman peak positions of silicon using a Raman spectroscopic microscope at multiple points within a bare silicon wafer surface on which a processing-affected layer remains; generating a histogram of peak shifts from the first-order Raman peak positions of the silicon obtained at the multiple locations; Calculating an average value A and a standard deviation S from the histogram; From the average value A and the standard deviation S, a maximum depth D of a process-affected layer remaining in the silicon wafer is estimated. The present invention relates to a method for evaluating a silicon wafer, the method comprising the steps of:
[0031] The present invention also provides a method for removing a process-affected layer from a silicon wafer, comprising the steps of: A method for removing a process-affected layer of a silicon wafer, comprising: subjecting the silicon wafer to etching and / or polishing with a removal amount exceeding a maximum depth D estimated by the silicon wafer evaluation method of the present invention.
[0032] The present invention will be described in detail below with reference to the drawings, but the present invention is not limited thereto.
[0033] <Silicon wafer evaluation method> [About the wafers to be evaluated] The silicon wafer evaluated by the silicon wafer evaluation method of the present invention is a bare silicon wafer with a process-affected layer remaining. More specifically, it is a bare silicon wafer in a process before etching, for example, a bare silicon wafer after slicing, lapping, grinding, etc. The process-affected layer is a layer in which stress due to these processes remains in the silicon wafer. Quantitatively expressing this, for example, it can be expressed as a silicon wafer with a surface roughness of 1 nm or more in a 100 μm×100 μm field of view. Note that, if etching is performed, the process-affected layer necessary for estimating the damage depth is removed, so a wafer after etching is not desirable.
[0034] [About measurements] In the present invention, a Raman spectroscopic microscope is used to obtain the first-order Raman peak positions of silicon at multiple locations within the surface of a bare silicon wafer on which a processing-affected layer remains.
[0035] For example, from a spectrum obtained using a laser Raman spectroscopic microscope with a light source wavelength of 532 nm, the first-order Raman peak position of silicon (520 cm -1 The peak shift value can be calculated by fitting the Lorentzian function to the area around the peak, and the peak shift value can be used as data for one location.
[0036] In the research of this invention, the peak shift value was measured at 1000 points on bare silicon wafers after various processes such as free abrasive slicing, fixed abrasive slicing, lapping, and grinding with a grindstone. The smallest standard deviation S was 0.031 cm -1 If there are 180 or more measurement points, this value can be calculated with a 99% confidence interval of ±0.005 cm. -1 By measuring at 200 or more points, this value can be obtained with a higher degree of reliability.
[0037] [About generating histograms and calculating the average value A and standard deviation S] In the silicon wafer evaluation method of the present invention, a histogram of peak shifts is generated from the first-order Raman peak positions of silicon obtained at multiple locations on a bare silicon wafer to be evaluated, and an average value A and a standard deviation S are calculated from the generated histogram.
[0038] 1 shows an example of a histogram of peak shifts that can be obtained by the silicon wafer evaluation method of the present invention with a solid line, and a histogram that is approximated to a normal distribution with a dashed line.
[0039] For example, the histogram of peak shifts shown by the solid line can be regarded as a normal distribution shown by the dashed line, and the average value A and the standard deviation S can be calculated from this normal distribution. In the histogram of peak shifts in FIG. 1, the average value A is 0.989 cm -1 and the standard deviation S is 0.030 cm -1 It is.
[0040] [Estimation of maximum depth D of the processing-affected layer] In the method for evaluating a silicon wafer of the present invention, the maximum depth D of the process-affected layer remaining in the silicon wafer is estimated from the average value A and standard deviation S calculated as described above.
[0041] For example, it is preferable to estimate the maximum depth D of the work-affected layer using A+3S as an index.
[0042] More specifically, it is preferable to estimate the maximum depth D of the work-affected layer using an index A+3S and a coefficient K according to the formula: D=K×(A+3S).
[0043] Here, the coefficient K is a coefficient that differs depending on the machine difference of the measuring device used, the number of local light scattering (LLS) defects on the wafer surface that is desired, and the like, and can be determined, for example, by experience and / or experiment.
[0044] By estimating the maximum depth D of the process-affected layer remaining in the bare silicon wafer to be evaluated from the average value A and standard deviation S calculated from the histogram of peak shifts described above, it is possible to determine an appropriate removal amount that can reliably remove the remaining process-affected layer.
[0045] On the other hand, local observation such as angle polishing and TEM, and the examples of Patent Document 5, etc., do not measure at enough locations to generate a histogram of peak shifts, but are equivalent to measuring several points from this histogram. With such measurements, it is statistically difficult to estimate the maximum depth D of the work-affected layer.
[0046] <Method for removing layers affected by processing on silicon wafers> In the method for removing a process-damaged layer from a silicon wafer of the present invention, the silicon wafer is etched and / or polished with a removal amount exceeding the maximum depth D estimated by the silicon wafer evaluation method of the present invention.
[0047] This makes it possible to reliably remove the process-affected layer remaining on the bare silicon wafer. By reliably removing the process-affected layer, it is possible to obtain a silicon wafer that satisfies a standard for the number of localized light scattering (LLS) defects after, for example, etching and / or polishing.
[0048] If a removal amount exceeding the maximum depth D estimated by the silicon wafer evaluation method of the present invention can be achieved, a plurality of processes may be used in combination.
[0049] On the other hand, if the estimated maximum depth D is exceeded, processing is not performed with a removal allowance greater than necessary, and therefore the productivity and yield of the wafer can be improved. EXAMPLES
[0050] The present invention will be specifically described below using examples and comparative examples, but the present invention is not limited to these.
[0051] <Examples 1 and 2 and Comparative Examples 1 and 2> [About wafers] The wafers to be evaluated were 10 sample wafers A cut from a 300 mm diameter P-type silicon single crystal using the free grain method with a grit of 1000, and 10 sample wafers B lapped with a grit of 1500.
[0052] The average surface roughness of sample wafer A was 351 nm in a 100 μm × 100 μm field of view, and the average surface roughness of sample wafer B was 226 nm in a 100 μm × 100 μm field of view.
[0053] [Measurement and histogram generation] Three wafers were extracted from the above sample wafer A, and the first-order Raman peak positions of silicon were obtained using a Raman spectroscopic microscope at a total of 200 points on the wafer surface. The light source wavelength used was 532 nm. Using the same procedure, the first-order Raman peak positions of silicon were obtained at a total of 200 points on sample wafer B.
[0054] From the first-order Raman peak positions of silicon obtained at a total of 200 points, histograms of peak shifts for each of sample wafers A and B were generated.
[0055] Next, the average value A and the standard deviation S were calculated from the generated histogram. As a result, for sample wafer A that was sliced with free abrasive grains, the average value A was 1.021 cm -1 and the standard deviation S=0.052 cm -1 In addition, for sample wafer B that was lapped, the average value A was 0.784 cm -1 and the standard deviation S=0.042 cm -1 It was.
[0056] [Estimation of maximum depth D of processing-affected layer] A reference wafer was prepared by surface grinding using a grinding wheel of No. 2000. For the prepared reference wafer, the first-order Raman peak positions of silicon were obtained at a total of 200 points on the surface using the same procedure as that performed for sample wafers A and B, and a histogram of peak shifts was generated from the first-order Raman peak positions of silicon obtained at a total of 200 points. The average value A0 and standard deviation S0 were calculated from the generated histogram. The average value A0 was 0.381 cm -1 and the standard deviation S0 is 0.065 cm -1 It was.
[0057] Next, the ground surface of the reference wafer was polished in increments of 0.5 μm in thickness.
[0058] The polishing machine used was a Fujikoshi Machinery DSP-20B equipped with a foamed polyurethane pad, the carrier was a titanium substrate with an insert made of FRP glass fiber impregnated with epoxy resin, and the slurry was a KOH-based slurry containing silica abrasive grains with an average particle size of 35 nm.
[0059] Every time 0.5 μm was polished, the number of local light scattering (LLS) defects on the polished surface was measured using a KLA particle size analyzer SP1. The total removal amount was defined as D0 when the number of local light scattering (LLS) defects with a particle size of 500 nm or more measured each time fell below 10. The total removal amount D0 was 8.6 μm.
[0060] Using the A0, S0, and D0 obtained above, the coefficient K was calculated from the formula: K = D0 / (A0 + 3S0). The coefficient K was 14.9 μm / cm -1 It was.
[0061] The maximum depth D of the process-affected layer of sample wafers A and B was estimated from the above average value A, standard deviation S, and coefficient K according to the formula: D=K×(A+3S).
[0062] As a result, the maximum depth D of the process-affected layer of sample wafer A, which was subjected to free abrasive slicing, was estimated to be D = 17.5 μm. Also, the maximum depth D of the process-affected layer of sample wafer B, which was subjected to lapping, was estimated to be D = 13.6 μm. Therefore, as explained below, polishing was performed with a polishing allowance, and the number of local light scattering (LLS) defects of 500 nm or more was measured using particle measurement device SP1 after cleaning, and the average number of defects for 10 wafers was confirmed.
[0063] Example 1 In Example 1, the sample wafer A was polished with removal amounts of 18 μm, 20 μm, and 22 μm, which exceeded the previously estimated maximum depth D (D=17.5 μm) of the processing-affected layer. The polishing conditions were the same as those for calculating the coefficient K.
[0064] Comparative Example 1 In Comparative Example 1, the sample wafer A was polished with removal rates of 12 μm, 14 μm, and 16 μm, which were less than the previously estimated maximum depth D (D=17.5 μm) of the processing-affected layer. The polishing conditions were the same as those for calculating the coefficient K.
[0065] Example 2 In Example 2, sample wafer B was polished with removal amounts of 14 μm, 16 μm, 18 μm, 20 μm, and 22 μm, which exceeded the previously estimated maximum depth D (D = 13.6 μm) of the processing-affected layer. The polishing conditions were the same as those for calculating the coefficient K.
[0066] Comparative Example 2 In Comparative Example 2, sample wafer B was polished with removal amounts of 10 μm and 12 μm, which are less than the previously estimated maximum depth D (D=13.6 μm) of the process-affected layer. The polishing conditions were the same as those for determining the coefficient K.
[0067] FIG. 2 is a graph showing the relationship between the polishing stock removal rate and the average number of defects in Examples 1 and 2 and Comparative Examples 1 and 2.
[0068] 2, in Example 1, in which sample wafer A sliced with free abrasive grains was polished with a removal amount exceeding the estimated maximum depth D (D=17.5 μm) of the process-damaged layer, the average number of defects achieved the standard of 10 or less. On the other hand, in Comparative Example 1, in which sample wafer A was polished with a removal amount equal to or less than the estimated maximum depth D (D=17.5 μm) of the process-damaged layer, the average number of defects failed to achieve the standard of 10 or less.
[0069] As is clear from the circle plots and dashed line graph in FIG. 2, in Example 2, in which the lapped sample wafer B was polished with a removal allowance exceeding the estimated maximum depth D (D = 13.6 μm) of the process-affected layer, the average number of defects achieved the standard of 10 or less. =1 In Comparative Example 2, in which polishing was performed with a removal stock of 3.6 μm or less, the average number of defects could not achieve the standard of 10 or less.
[0070] The present invention is not limited to the above-described embodiment. The above-described embodiment is merely an example, and anything that has substantially the same configuration as the technical idea described in the claims of the present invention and exhibits similar effects is included in the technical scope of the present invention.
Claims
1. A method for evaluating a silicon wafer, comprising the steps of: Obtaining first-order Raman peak positions of silicon using a Raman spectroscopic microscope at multiple points within a bare silicon wafer surface on which a processing-affected layer remains; generating a histogram of peak shifts from the first-order Raman peak positions of silicon obtained at the multiple locations; Calculating an average value A and a standard deviation S from the histogram; A maximum depth D of a processing-affected layer remaining in the silicon wafer is estimated from the average value A and the standard deviation S. A method for evaluating a silicon wafer, comprising:
2. 2. The method for evaluating a silicon wafer according to claim 1, wherein the maximum depth D of the process-affected layer is estimated using A+3S as an index.
3. 3. The method for evaluating a silicon wafer according to claim 1, further comprising the step of evaluating the silicon wafer having a surface roughness of 1 nm or more in a 100 μm×100 μm visual field.
4. 4. The method for evaluating a silicon wafer according to claim 1, wherein a light source wavelength used in the Raman spectroscopic microscope is 532 nm.
5. The method for evaluating a silicon wafer according to any one of claims 1 to 4, characterized in that the number of points from which the first-order Raman peak positions of silicon are obtained in order to calculate the average value A and the standard deviation S is 200 or more.
6. A method for removing a processed damaged layer from a silicon wafer, comprising the steps of: A method for removing a processed-degraded layer of a silicon wafer, comprising: subjecting the silicon wafer to etching and / or polishing with a removal amount exceeding a maximum depth D estimated by the silicon wafer evaluation method according to any one of claims 1 to 5.
Citation Information
Patent Citations
Method for evaluating crystaline characteristic
JP1996114548A
High-speed mirror surface polishing method of silicon wafer
JP2001358099A
Manufacturing method for mirror-polished wafer
JP2003062740A
Wafer supporting substrate
JP2004296912A
Evaluation method and evaluation apparatus for semiconductor wafer
JP2008109012A
