JP5239346B2 - Stress evaluation method using Raman spectroscopy and semiconductor device manufacturing method - Google Patents

Description

  The present invention relates to a stress evaluation method for evaluating stress in the depth direction (or strain: hereinafter the same) in a micron or submicron region in a sample using Raman spectroscopy, and more particularly to a stress or lattice in an electronic device manufacturing process. The present invention relates to a stress evaluation method suitable for strain monitoring and a semiconductor device manufacturing method using the stress evaluation result.

  When stress is applied to the crystal, the lattice is distorted, the band structure is changed, and the electrical characteristics are also changed. By utilizing this property, the electrical characteristics of an electronic device (semiconductor element) formed of silicon crystal, gallium arsenide crystal, or the like can be improved, or deterioration of the electrical characteristics can be suppressed. Since electronic devices generally have a three-dimensional structure, it is important to control the three-dimensional stress distribution (strain distribution) of the crystal in order to improve the characteristics of the electronic device by applying stress. It becomes.

  When developing an electronic device using stress or lattice distortion, the manufacturing technique is of course important, but the technique of detecting and evaluating the stress is also important. Conventionally, as a method for evaluating stress in a crystal, a method using X-ray diffraction, a method using electron beam diffraction, and a method using Raman spectroscopy (Raman spectroscopy) are known.

  Although the method using X-ray diffraction has the advantage that the stress in the crystal can be measured relatively easily, the X-ray beam diameter is as large as about 1 mm, so that in recent miniaturized electronic devices (semiconductor elements). It is difficult to apply to the evaluation of stress distribution.

  On the other hand, the method using electron diffraction has the advantage of high spatial resolution because the beam diameter of the electron beam is as small as about 1 nm. However, since the transmittance of the electron beam is low, it is necessary to make the sample (crystal) thin. When cutting or polishing is performed in order to make the sample thin, the stress applied to the measurement surface is released, so that actual strain cannot be measured. In addition, when using electron beam diffraction, there is a method for more accurately evaluating distortion by theoretically and experimentally devising, but since it is an indirect evaluation, it cannot be said that reliability is sufficient.

  On the other hand, Raman spectroscopy has a relatively small laser beam diameter of the order of submicron (0.2 to 1 μm), and therefore can be applied to the evaluation of stress distribution in recent miniaturized electronic devices. . In general stress measurement by Raman spectroscopy, laser light is incident from the surface (or cross section) of the sample, and Raman scattered light emitted from the sample is detected by a detector arranged on the sample surface side (or cross section). Then, the stress is evaluated by comparing the peak position of the spectrum of the incident light with the peak position of the spectrum of the Raman scattered light (Raman spectrum).

  Patent Document 1 describes an example of a stress distribution measurement method using Raman spectroscopy. In Patent Document 1, the spectrum obtained by scanning a laser or X-ray with a known spot diameter and intensity distribution with a moving amount equal to or less than the spot diameter and the stress within the spot diameter obtained by the same measurement system are uniform. In this case, it is described that the stress or stress distribution of the minute portion below the spot diameter is measured by combining the relationship between the stress and spectrum of the same material as the sample to be measured and the intensity distribution data of the laser or X-ray. .

In addition, there exist some which were described in patent document 2, 3 as a prior art considered to be related to this invention. Patent Document 2 describes that a Raman spectrum is measured by irradiating excitation light from the back side of a substrate. Further, in Patent Document 3, the substrate is cleaved in a direction intersecting the surface thereof, laser light is incident from one of the cleaved surfaces, and Raman scattered light emitted from the opposite cleaved surface is detected. It has been proposed to evaluate distortion.
Japanese Patent No. 2544428 JP 2002-124702 A JP 7-294436 A

  Usually, in Raman spectroscopy, laser light is incident from the surface or cross section of a sample as described above. However, for example, if a wiring or a via contact is formed on a portion to be measured, the laser light is reflected by the wiring or the via contact, so that the stress cannot be evaluated.

  In recent years, it has become necessary to examine not only the stress distribution in the direction parallel to the surface but also the stress distribution in the depth direction. Since the penetration depth of the laser beam is related to the wavelength, it is necessary to change the laser wavelength in various ways to investigate the stress distribution in the depth direction, which causes a problem of deterioration in throughput and variation in resolution. When laser light is irradiated from the back surface side of the substrate as described in Patent Document 2, a similar problem occurs when an attempt is made to examine the stress distribution in the depth direction.

  When examining the stress distribution in the depth direction, it is conceivable to cut (cleave) the sample and inspect the cut surface as described in Patent Document 3. However, since the stress applied to the cross section at the time of cutting the sample is released, the stress distribution before cutting cannot be measured accurately.

  An object of the present invention is to provide a stress evaluation method using Raman spectroscopy that can measure stress distribution in the depth direction in a sample with high accuracy, and a semiconductor device manufacturing method using the stress evaluation result.

  According to one aspect of the present invention, there is provided a method for measuring a stress distribution in a depth direction of a sample in which a member serving as a stress source is provided on a first surface side, the reverse of the first surface of the sample. Polishing step for polishing the second surface on the side, thickness measuring step for measuring the thickness of the sample, and detecting the Raman scattered light by making light incident on the evaluation position set on the second surface side A stress evaluation method for obtaining a stress distribution in the depth direction by calculating the stress at the evaluation position from the spectrum of the Raman scattered light is repeated.

  In the present invention, a polishing step for polishing a surface (second surface) opposite to the first surface provided with a member (film or the like) serving as a stress source, and a thickness measurement for measuring the thickness of the sample. The process and the Raman scattered light detection process in which light (laser light) is incident on the evaluation position set on the second surface side and Raman scattered light is detected are repeated. That is, every time the second surface side is polished, the thickness of the sample is measured and Raman scattered light is detected. Thereby, the relationship between the thickness of the sample and the stress is clarified, and a stress distribution in the depth direction is obtained.

  In the Raman scattered light detection step, when light is sequentially incident on a plurality of evaluation positions set on the second surface side and Raman scattered light at each evaluation position is detected, a two-dimensional surface in a plane parallel to the first surface is obtained. Stress distribution can be obtained. In addition, a three-dimensional stress distribution can be obtained by obtaining a two-dimensional stress distribution every time the second surface side is polished. If the three-dimensional stress distribution is known in this way, the stress distribution on an arbitrary surface can be easily obtained. Therefore, in the Raman scattered light detection step, it is preferable that light is sequentially incident on a plurality of evaluation positions set on the second surface side to detect Raman scattered light at each evaluation position.

  According to another aspect of the present invention, there is provided a method for measuring a stress distribution in a depth direction of a sample in which a member serving as a stress source is disposed on a first surface side, in a direction intersecting the first surface. Polishing a second surface opposite to the first surface of the sample so as to form a surface, and sequentially applying light to a plurality of evaluation positions set on the second surface side of the sample. Is provided to obtain a spectrum of Raman scattered light at each evaluation position, and to obtain a stress distribution in the depth direction by calculating stress from the spectrum at each evaluation position.

  For example, when a single layer film or a multilayer film is uniformly formed on a silicon substrate having a uniform surface state, the stress at the interface between the silicon substrate and the film should not depend on the location. Therefore, in the present invention, the opposite side of the first surface is formed so that the surface is formed in the direction intersecting with the first surface (for example, the cross section in the direction orthogonal to the first surface has a wedge shape). The surface (second surface) is polished. Then, light is sequentially incident on a plurality of evaluation positions set on the second surface side, and a spectrum of Raman scattered light at each evaluation position is acquired. Thereby, the stress distribution in the depth direction can be obtained in one polishing process.

  According to another aspect of the present invention, after the stress distribution in the depth direction is measured by the above-described method, the manufacturing conditions (for example, the film formation temperature or the film formation time) are controlled based on the measurement result, and the semiconductor Manufacture equipment. Thereby, a semiconductor device having desired characteristics can be manufactured.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

(First embodiment)
FIG. 1 is a block diagram showing a configuration of a stress measuring apparatus used in the stress measuring method according to the first embodiment of the present invention.

  As shown in FIG. 1, the stress measuring device includes a stage 11, a microscopic optical unit 12, a sample thickness measuring device 13, a laser scanning control unit 14, a spectrometer 15, a detector 16, and a main control. The unit 17, the stage control unit 18, and the display unit 19 are configured.

A sample 10 to be examined for stress distribution in the depth direction is mounted on a stage 11. The sample 10 is composed of, for example, a silicon substrate (silicon wafer) 10a and a film 10b serving as a stress source for applying stress to the crystals constituting the silicon substrate 10a, with the film 10b side down, that is, the back surface of the sample 10 Placed on stage 11 with the side up. Hereinafter, in the present embodiment, the description will be made on the assumption that the film 10a for applying a stress to the crystal constituting the silicon substrate 10a is made of silicon oxide (SiO ₂ ). In the present embodiment, the element formation surface of the silicon substrate 10a is referred to as a front surface, and the opposite surface is referred to as a back surface.

  The microscopic optical unit 12 generates laser light in accordance with a signal from the main control unit 17 and irradiates the sample 10. Further, the microscopic optical unit 12 detects the Raman scattered light emitted from the sample 10, converts it into an electrical signal, and outputs it.

  Based on the signal from the main control unit 17, the laser scanning control unit 14 controls the optical system of the microscopic optical unit 12 so that the laser beam scans the back surface side (the upper surface in FIG. 1) of the sample 10. The spectroscope 15 receives the electrical signal output from the microscopic optical unit 12 and performs spectral processing. That is, the spectrometer 15 classifies and outputs the electricity output from the microscopic optical unit 12 for each energy (wave number or wavelength). The detector 16 integrates the signals output from the spectrometer 15 for each energy to obtain a Raman spectrum. The detector 16 identifies the peak position of the Raman spectrum and detects the shift amount from the peak position of the spectrum of the incident light (hereinafter also referred to as “reference peak position”).

  The sample thickness measuring device 13 measures the thickness of the sample 10. A micrometer may be used as the sample thickness measuring device 13. Further, as the sample thickness measuring device 13, a thickness measuring device using near infrared rays or a film thickness measuring device using optical interferometry may be used. Further, the back surface of the sample and a specific portion inside the sample may be focused, and the difference in the focal position may be measured as the sample thickness. The sample thickness data measured by the sample thickness measuring device 13 is input to the main controller 17 manually or automatically.

  The stage control unit 18 moves the stage 11 in the horizontal direction (XY direction) so that the laser light output from the microscopic optical unit 12 is incident on a predetermined position of the sample 10 in accordance with a signal from the main control unit 17. Moving. The display unit 19 displays a Raman spectrum, a stress measurement result, and the like according to a signal from the main control unit 17.

  Hereinafter, before describing the stress evaluation method according to the present embodiment, preliminary items for facilitating understanding of the present invention will be described.

  Stress evaluation by conventional Raman spectroscopy is performed by making laser light incident on a sample, detecting Raman scattered light to identify the peak of the Raman spectrum, and examining the shift amount from the standard peak position. The laser beam in the infrared wavelength region is transmitted through a silicon substrate (wafer) having a thickness of 1 mm. However, the depth at which information useful as Raman spectroscopy can be obtained is not large. The distance at which the light intensity is 1 / e (e is the base of the natural logarithm), the so-called penetration depth d, is expressed by the following equation (1).

d = λ / 4πk (1)
Where λ is the wavelength of the incident light and k is the imaginary part of the refractive index of the complex refractive index n + ik.

  From the complex refractive index of silicon (n = 4.367, k = 0.079), it can be seen that the penetration depth of light with a wavelength λ of 488 nm into the silicon substrate is as small as about 500 nm. Therefore, the laser beam having this wavelength cannot detect the stress generated on the front surface side of the silicon substrate from the back surface side of the substrate. Therefore, in this embodiment, a step of polishing the back side of the sample horizontally (polishing parallel to the substrate surface) (polishing step) and a step of measuring the effective thickness (remaining thickness) of the sample (effective sample thickness) The distribution in the depth direction of the stress is obtained by sequentially repeating the measurement step) and the step of detecting the Raman scattered light (Raman scattered light detecting step) by entering laser light from the back side of the sample.

  When the laser beam is scanned parallel to the sample surface (two-dimensional scan), a two-dimensional stress distribution on the sample surface can be obtained. Therefore, the polishing step, the effective sample thickness measurement step, and the Raman scattered light detection step are sequentially performed. By repeating, a three-dimensional stress distribution is obtained. From this three-dimensional stress distribution, a two-dimensional stress distribution in an arbitrary plane can be obtained. Instead of performing two-dimensional scanning with laser light, the stage 11 may be moved in the horizontal direction (XY) direction. Further, by repeating the above steps without moving the stage 11, a stress distribution in the depth direction at a specific position can be obtained.

  By the way, if the polishing is performed to the vicinity of the stress source, the stress is released, so that the stress cannot be measured. Since stress release is independent of the direction of polishing, stress release can occur, for example, in cross-section polishing or planar polishing of electronic devices. In order to evaluate the stress more accurately, it is important to know in advance how much the stress is released in the measurement sample after polishing.

The extent to which the stress is released by polishing depends on the type of sample. Therefore, it is necessary to examine in advance how much the stress is released by polishing for each sample. The method will be described below with reference to FIGS. 2 (a) and 2 (b). Here, as shown in FIGS. 2A and 2B, it is assumed that the sample 10 includes a silicon substrate 10a and a SiO ₂ film 10b serving as a stress source.

First, as shown in FIG. 2A, an SiO ₂ film 10b serving as a stress source is formed on the silicon substrate 10a with a uniform thickness. Thereafter, as shown in FIG. 2B, the silicon substrate 10a is cut into a wedge shape when viewed from a direction orthogonal to the substrate surface, and the cut surface is polished. In this way, a sample having a measurement surface whose thickness d changes continuously is obtained.

Next, laser light is scanned along the interface between the silicon substrate 10a and the SiO ₂ film 10b, and the spectrum of the Raman scattered light (Raman spectrum) is detected at each location (evaluation position). Then, the peak position of the spectrum is specified, and the stress is calculated from the shift amount from the standard peak position. Thereafter, the sample thickness at which the stress is released is obtained from the relationship between the sample thickness d and the stress.

  FIG. 3 is a diagram showing the results of actual examination of the relationship between the sample thickness d and the stress, with the sample thickness d on the horizontal axis and the stress on the vertical axis. From FIG. 3, it can be seen that the stress detected greatly changes when the sample thickness d is about 200 nm. That is, when the sample thickness d exceeds about 200 nm, the stress is large, and when the sample thickness d is about 200 nm or less, the stress is small.

Since the SiO ₂ film 10b serving as a stress source is uniformly formed, the stress applied to the interface with the SiO ₂ film 10b should not depend on the location unless the stress is released. However, as a result of the experiment, as shown in FIG. 3, when the sample thickness is about 200 nm or less, the detected stress is remarkably reduced. This indicates that the stress is released (stress relaxation) when the sample thickness reaches about 200 nm. With respect to this sample, stress is released to two surfaces (inclined surfaces) in the thickness d direction. Therefore, when the sample is polished only on one surface side, about 100 nm, which is half of the thickness, can be said to be a thickness at which the stress is released.

Hereinafter, the stress measurement method according to the present embodiment will be described with reference to the flowchart shown in FIG. 4 and the block diagram showing the apparatus configuration of FIG. However, here, a case will be described in which the stress distribution in the depth direction is measured in the sample 10 in which the SiO ₂ film 10b is formed on the silicon substrate 10a (lower side in FIG. 1) as shown in FIG. Here, from FIG. 3, when the SiO ₂ film (stress source) is formed on the silicon substrate, the thickness at which the stress is released by polishing is 100 nm. That is, when the distance from the stress source to the polished surface is 100 nm or less, the stress applied by the stress source is released, and there is no point in measuring the stress. For this reason, when the distance from the stress source to the polished surface becomes 100 nm or less, it is determined that the effective sample thickness has become 0, and the measurement is terminated.

First, a sample 10 is prepared. That is, the SiO ₂ film 10b is formed as a stress source on the silicon substrate 10a. Then, the back side of the substrate 10a is polished (horizontal polishing) in parallel with the substrate surface (step S11).

Next, the sample 10 is mounted on the stage 11 with the SiO ₂ film 10b on the lower side. Then, the stage 11 is moved while observing the sample 10 with an optical microscope, and an evaluation position (laser beam incident position) is set (step S12).

Next, the sample thickness is measured (step S13). Here, it is assumed that the light reflected at the interface between the substrate 10a and the SiO ₂ film 10b is wavelength-dispersed and the sample thickness is theoretically calculated. FIG. 5 is a diagram showing an example of the result of measuring the wavelength dependence of the reflection intensity together with the theoretical waveform, with the wavelength on the horizontal axis and the reflection intensity on the vertical axis. From a comparison between the theoretical waveform and the measured value, in this example, the sample thickness (distance between the interface between the substrate 10a and the SiO ₂ film 10b and the back surface (polishing surface) of the substrate 10a) is set to 2.0 μm.

Next, the effective sample thickness (remaining thickness) is calculated from the measured value of the sample thickness (step S14). As described above, when the SiO ₂ film 10b is formed on the silicon substrate 10a, the thickness at which the stress is released by polishing is in the range of 100 nm or less from the stress source, and thus the sample thickness (2) measured in step S13. The value obtained by subtracting the thickness (100 nm) from which stress is released by polishing, that is, 1.9 μm, is set as the effective sample thickness (remaining thickness).

  Next, the microscopic optical unit 12 is controlled by the laser scanning control unit 14 to generate laser light, enters the evaluation position of the sample 10 (step S15), and the Raman scattered light is detected by the microscopic optical unit 12 (step S16). ). The detection signal of the Raman scattered light output from the microscopic optical unit 12 is spectrally processed by the spectroscope 15 and input to the detector 16.

  Next, in the detector 16, the output of the spectroscope 15 is integrated for each energy (wave number or wavelength) to obtain a Raman spectrum, and the peak position is determined. Then, the shift amount is calculated from the difference between the peak position of the Raman spectrum and the reference peak position (step S17). Thereby, the stress at the evaluation position can be calculated. Thereafter, the process proceeds to the next step (step S18).

  When two-dimensional scanning is performed (YES in step S18), the microscopic optical unit 12 is controlled by the laser scanning control unit 14 to change the evaluation position, and then the process returns to step S15 to repeat the above-described processing. When the two-dimensional scanning is not performed or when the two-dimensional scanning is completed (NO in step S18), the process proceeds to step S19.

  In step S19, it is determined whether or not the effective sample thickness is zero. When the effective sample thickness is not 0 (in the case of NO), after polishing the back surface of the sample, the process returns to step S13 and the above-described processing is repeated. On the other hand, when the effective sample thickness is determined to be 0 in step S19 (in the case of YES), the process proceeds to step S20.

  In step S20, the stress is calculated from the shift amount of the peak position of the Raman spectrum calculated in step S17. For example, in the case of a silicon substrate having a surface orientation (001), the stress σ is calculated by the following equation (2).

σ = −0.25 × 10 ⁹ × Δν (2)
However, Δν is the shift amount of the peak position. Then, the main control unit 17 accumulates the calculation results of the stress at each evaluation position, and displays the stress distribution in the depth direction at the predetermined position on the display unit 19 based on the result (step S21). In this way, a stress distribution in the depth direction of the silicon substrate 10a is obtained.

  The stress applied to the silicon substrate varies depending on the film forming conditions (film forming temperature or film forming time). After measuring the stress distribution in the depth direction of the silicon substrate 10a by the method described above, a semiconductor device having desired characteristics can be manufactured by appropriately determining the film forming conditions based on the measurement result.

  FIG. 6 is a schematic cross-sectional view showing a sample obtained by actually measuring the stress distribution in the depth direction by the method described above. However, the sample shown in FIG. 6 is a silicon substrate 20 on which a MOS transistor is formed. The MOS transistor includes a gate electrode 21, a gate oxide film 22, and a pair of source / drain regions 23.

In FIG. 6, reference numeral 24 denotes a via contact electrically connected to the source / drain region 23, 25 denotes a silicide film for ohmic connection between the source / drain region 23 and the via contact 24, and 26 denotes SiO ₂ covering the MOS transistor. A film (interlayer insulating film) 27 is an element isolation film that electrically isolates element regions. The silicon substrate 20 is mounted on the stage 11 of the stress measuring device with the element formation surface (the surface on which the MOS transistor 20 is formed) facing down.

7, the horizontal axis indicates the distance from the interface between the silicide film 25 and the silicon substrate 20 (source / drain region 23), the vertical axis indicates the stress measured by the method described above, and the arrow in FIG. It is a figure which shows the measurement result of the stress distribution of the depth direction in the evaluation position shown. However, here, the sample thickness is measured by the optical interference method with reference to the interface between the silicide film 25 and the silicon substrate 20 (source / drain region 23). Further, a position (indicated by a broken line in FIG. 6) that is 100 nm away from the interface between the silicide film 25 and the silicon substrate 20 (source / drain region 23) in the vertical direction is a position where the effective sample thickness is zero. From this FIG. 7, it can be seen that in the case of this sample, there is an influence of stress by the SiO ₂ film 26 from the interface between the silicide film 25 and the silicon substrate 20 (source / drain region 23) to a position of about 1500 nm.

  As described above, in the present embodiment, the stress by repeating measured beforehand whether stress Polishing advance far from being released, the polishing process, the effective sample thickness measuring step and Raman scattered light detecting step in the order Therefore, the stress distribution in the depth direction can be measured with high accuracy while avoiding the release of the stress.

  In this embodiment, since laser light is incident from the back surface side of the substrate, stress can be measured even if an element such as a transistor, wiring, or via contact is formed on the front surface side of the substrate.

  In the case of a semiconductor substrate on which an element is formed, the stress generally increases as it approaches the surface of the semiconductor substrate. Therefore, when the stress is measured from the surface side of the substrate, it is difficult to detect a small stress in a deep region with high accuracy. . However, in this embodiment, since the stress is measured while polishing the back surface side of the substrate, a small stress in a deep region away from the surface of the substrate can be measured with high accuracy. A semiconductor device having desired characteristics can be manufactured by appropriately determining manufacturing conditions in accordance with the measurement result.

(Second Embodiment)
Hereinafter, a second embodiment of the present invention will be described. The present embodiment relates to a method for measuring a stress distribution in the depth direction when a stress source is uniformly formed on a member having a uniform surface state.

  When a single layer film or a multilayer film is uniformly formed on a silicon substrate having a uniform surface state, the stress at the interface between the silicon substrate and the film should not depend on the location. Therefore, in this embodiment, the silicon substrate is polished so that the cross section has a wedge shape, and stress at each evaluation position on the inclined surface is measured to obtain a stress distribution in the depth direction. Hereinafter, the stress measurement method according to the present embodiment will be described with reference to the schematic cross-sectional views shown in FIGS. 8A to 8C and the flowchart shown in FIG. Since the configuration of the stress measuring device is basically the same as that described in the first embodiment, the block diagram showing the configuration of the stress measuring device in FIG. 1 is also referred to here.

First, as shown in FIG. 8A, a sample 30 is prepared, that is, an SiO ₂ film 30b is formed as a stress source on a silicon substrate 30a having a uniform surface state. Then, as shown in FIG. 8B, the back surface of the silicon substrate 30a is obliquely polished (inclined polishing) so that the cross section has a wedge shape (step S31).

Next, as shown in FIG. 8C, the sample 30 is mounted on the stage 11 with the surface on which the SiO ₂ film 30b is formed facing down. Then, the stage 11 is moved while observing the sample 30 with an optical microscope, and an evaluation position (laser beam incident position) is set (step S32). In FIG. 8C, each arrow indicates an evaluation position.

  Next, the sample thickness is measured (step S33). Here, it is assumed that the thickness at each evaluation position on the measurement surface is measured using an optical interference method. Thereafter, the effective sample thickness at each evaluation position is calculated (step S34). That is, as described above, the thickness at which the stress is released by polishing is in the range of 100 nm or less from the stress source. Therefore, 100 nm is subtracted from the sample thickness at each evaluation position measured in step S33 to obtain the effective sample thickness. .

  Next, laser light is generated by controlling the microscopic optical unit 12 by the laser scanning control unit 14, and is incident on the first evaluation position of the sample 30 (step S35). Raman scattering light is detected by the microscopic optical unit 12 (step S36). A detection signal of Raman scattered light output from the microscopic optical unit 12 is input to the spectroscope 15, subjected to spectral processing, and output to the detector 16.

  The detector 16 integrates the output of the spectrometer 15 for each energy (wave number or wavelength) to acquire a Raman spectrum, and searches for a peak position. Then, the shift amount is calculated from the difference between the peak position of the Raman spectrum and the reference peak position (step S37).

  In step S38, it is determined whether to scan with laser light based on the effective sample thickness. That is, when it is determined that the effective sample thickness is not 0 (in the case of YES), the laser beam is scanned to change the evaluation position, and then the process returns to step S35 and the above-described processing is repeated. On the other hand, when it is determined in step S38 that the effective sample thickness is 0 (in the case of NO), the process proceeds to step S39.

In step S39, the stress is calculated from the shift amount of the peak position of the Raman spectrum calculated in step S37. For example, in the case of a silicon substrate whose surface orientation is (001), the stress σ is calculated by the above-described equation (2). The main control unit 18 accumulates these calculation results inside and displays them on the display unit 19 as a stress distribution in the depth direction (step S40). In this way, a stress distribution in the depth direction of the sample 30 composed of the silicon substrate 30a and the SiO ₂ film 30b is obtained.

FIG. 10 is a diagram illustrating a stress measurement result with the horizontal axis indicating the distance from the interface between the silicon substrate and the SiO ₂ film and the vertical axis indicating the stress measured by the method described above. From this FIG. 10, it can be seen that in the case of this sample, there is an influence of the stress due to the SiO ₂ film from the interface between the silicon substrate and the SiO ₂ film as the stress source to about 200 nm.

  In the present embodiment, when a film serving as a stress source is uniformly formed on a substrate having a uniform surface state, the stress distribution in the depth direction can be measured with high accuracy by one polishing (gradient polishing). .

  In the above-described embodiment, the example in which the present invention is applied to the measurement of the stress distribution in the depth direction of the semiconductor substrate (silicon wafer) has been described. By this, the present invention measures the stress distribution in the depth direction of the semiconductor substrate. The present invention is not limited to the method, and the present invention can be applied to the measurement of stress distribution in the depth direction of various members.

  Similarly to the first embodiment, after measuring the stress distribution according to the present embodiment, a semiconductor device having desired characteristics can be manufactured by appropriately determining the film formation conditions based on the measurement result. it can.

FIG. 1 is a block diagram showing a configuration of a stress measuring apparatus used in the stress measuring method according to the first embodiment of the present invention. FIGS. 2A and 2B are schematic views showing a method for examining the thickness at which stress is released. FIG. 3 is a diagram showing the results of examining the relationship between the sample thickness d and the stress. FIG. 4 is a flowchart showing the stress measurement method according to the first embodiment. FIG. 5 is a diagram illustrating an example of a result of measuring the wavelength dependence of the reflection intensity together with a theoretical waveform. FIG. 6 is a schematic cross-sectional view showing a sample obtained by actually measuring the stress distribution in the depth direction. FIG. 7 is a diagram showing the measurement result of the stress distribution in the depth direction at the evaluation position indicated by the arrow in FIG. 8A to 8C are cross-sectional views showing a stress measurement method according to the second embodiment of the present invention. FIG. 9 is a flowchart showing a stress measurement method according to the second embodiment. FIG. 10 is a diagram illustrating a stress measurement result according to the second embodiment.

Explanation of symbols

10, 30 ... sample,
10a, 20, 30a ... silicon substrate,
10b, 26, 30b ... SiO ₂ film,
11 ... stage,
12 ... Microscopic optical part,
13 ... Sample thickness measuring device,
14 ... Laser scanning controller,
15 ... Spectroscope,
16 ... detector,
17 ... main control unit,
18 ... Stage control unit,
19 ... display part,
21 ... Gate electrode,
22 ... Gate oxide film,
23 ... Source / drain region,
24 ... via contact,
25. Silicide film,
27: Element isolation film.

Claims (7)

A method for measuring a stress distribution in a depth direction of a sample provided with a member serving as a stress source on the first surface side,
A polishing step for polishing a second surface opposite to the first surface of the sample, a thickness measurement step for measuring the thickness of the sample, and an evaluation position set on the second surface side Repeat the Raman scattered light detection process of detecting the Raman scattered light by entering the light,
A stress evaluation method characterized by obtaining a stress distribution in the depth direction by calculating a stress at the evaluation position from a spectrum of Raman scattered light.   2. The Raman scattered light detection step according to claim 1, wherein in the Raman scattered light detection step, light is sequentially incident on a plurality of evaluation positions set on the second surface side, and Raman scattered light at each evaluation position is detected. Stress evaluation method.   3. The method according to claim 1, wherein the polishing step, the thickness measurement step, and the Raman scattered light detection step are repeated until the stress applied to the sample by the stress source is released by polishing. Stress evaluation method. A method for measuring a stress distribution in a depth direction of a sample in which a member serving as a stress source is disposed on a first surface side,
Polishing a second surface of the sample opposite to the first surface so as to form a surface in a direction intersecting the first surface;
Injecting light sequentially into a plurality of evaluation positions set on the second surface side of the sample to obtain a spectrum of Raman scattered light at each evaluation position;
And calculating a stress from a spectrum at each evaluation position to obtain a stress distribution in the depth direction.   The stress evaluation method according to claim 4, wherein the evaluation position is set only in a region other than a region where stress applied to the sample by the stress source is released by polishing. A method for manufacturing a semiconductor device, comprising:
A polishing step of polishing a second surface opposite to the first surface of the semiconductor substrate having a semiconductor element formed on the first surface; a thickness measuring step of measuring a thickness of the semiconductor substrate; Repeating the Raman scattered light detection step of detecting the Raman scattered light by making the light incident on the evaluation position set on the second surface side,
Obtain stress distribution in the depth direction by calculating the stress at the evaluation position from the spectrum of Raman scattered light,
A manufacturing method of a semiconductor device, wherein manufacturing conditions for manufacturing the semiconductor element on another semiconductor substrate are adjusted according to the calculated stress distribution in the depth direction. A method for manufacturing a semiconductor device, comprising:
Polishing a second surface of the semiconductor substrate opposite to the first surface so as to form a surface in a direction intersecting the first surface on which the semiconductor element is formed; and the second surface of the semiconductor substrate. A process of acquiring light in order to obtain a spectrum of Raman scattered light at each evaluation position by sequentially entering light at a plurality of evaluation positions set on the surface side, and calculating stress from the spectrum at each evaluation position to obtain stress in the depth direction. And obtaining a distribution,
Obtain stress distribution in the depth direction by calculating stress from the spectrum of each evaluation position,
A manufacturing method of a semiconductor device, wherein manufacturing conditions for manufacturing the semiconductor element on another semiconductor substrate are adjusted according to the calculated stress distribution in the depth direction.

Images