Evaluation method, manufacturing method of semiconductor device, and evaluation system

Abstract

To provide an evaluation method suitable for making an original edition sealing condition appropriate.SOLUTION: An evaluation method includes: generating a first spectrum and a second spectrum by applying first conversion processing to first two-dimensional data indicating defect distribution of a first substrate where a pattern is formed by sealing an original edition to a photosensitive agent on the first substrate and second two-dimensional data indicating predicted defect distribution of a second substrate; applying filter processing to the generated first spectrum and the generated second spectrum; restoring the first two-dimensional data and the second two-dimensional data by applying second conversion processing which is inverse to the first conversion processing, to the processed first spectrum and the processed second spectrum; applying threshold processing to the restored first two-dimensional data and the restored second two-dimensional data; and computing a degree of coincidence by applying a comparison function to the threshold-processed first two-dimensional data and the threshold-processed second two-dimensional data.SELECTED DRAWING: Figure 10

Description

[Technical field]

[0001] The present embodiment relates to an evaluation method, a manufacturing method for a semiconductor device, and an evaluation system. [Background technology]

[0002] In NIL (Nano Imprint Lithography), a master is prepared, a photosensitive agent is applied onto a substrate, and the master is imprinted onto the photosensitive agent on the substrate. The shape of the master is transferred onto the photosensitive agent, which is then exposed and hardened. The master is then released from the substrate, and the substrate is processed using the pattern formed on the photosensitive agent as a mask, forming a pattern on the substrate. In NIL, it is desirable for the conditions for imprinting the master to be appropriate. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 3870052 [Patent Document 2] JP 2020-113767 A [Patent Document 3] JP 2022-518156 A [Patent Document 4] JP 2015-200658 A Summary of the Invention [Problem to be solved by the invention]

[0004] An object of one embodiment is to provide an evaluation method, a manufacturing method for a semiconductor device, and an evaluation apparatus that are suitable for optimizing conditions related to imprinting of a master. [Means for solving the problem]

[0005] According to one embodiment, there is provided an evaluation method. The evaluation method includes performing a first conversion process on first two-dimensional data showing a defect distribution of a first substrate on which a pattern is formed by imprinting an original onto a photosensitive agent on a first substrate, and second two-dimensional data showing a predicted defect distribution of a second substrate, respectively, to generate a first spectrum and a second spectrum. The evaluation method includes performing a filter process on the generated first spectrum and the generated second spectrum, respectively. The evaluation method includes performing a second conversion process, which is the inverse of the first conversion process, on the processed first spectrum and the processed second spectrum, respectively, to restore the first two-dimensional data and the second two-dimensional data. The evaluation method includes performing a threshold process on the restored first two-dimensional data and the restored second two-dimensional data. The evaluation method includes applying a comparison function to the threshold process-processed first two-dimensional data and the threshold process-processed second two-dimensional data to calculate a degree of agreement. [Brief description of the drawings]

[0006] [Figure 1] FIG. 2 is a plan view showing a configuration of an original and a substrate in the embodiment. [Diagram 2] FIG. 4 is a cross-sectional view showing imprinting of an original in the embodiment. [Diagram 3] 4A to 4C are cross-sectional views showing the formation of a pattern on a substrate by imprinting a master in an embodiment. [Figure 4] FIG. 1 is a diagram showing a functional configuration of an evaluation system that executes an evaluation method according to an embodiment. [Diagram 5] FIG. 1 is a diagram showing a hardware configuration of an evaluation system that executes an evaluation method according to an embodiment. [Figure 6] 1 is a flowchart showing a method for manufacturing a semiconductor device including an evaluation method according to an embodiment. [Figure 7] 5 is a flowchart showing a calibration process in the embodiment. [Figure 8] FIG. 4 is a data flow diagram showing the operation of the evaluation system in the embodiment. [Figure 9] 1 is a flowchart showing the flow of an evaluation method according to an embodiment. [Figure 10]11 is a flowchart showing the flow of a data extraction process in the embodiment. [Figure 11] FIG. 4 is a diagram showing two-dimensional data in the embodiment. [Figure 12] 11 is a flowchart showing a layout optimization process according to the embodiment. [Figure 13] FIG. 4 is a data flow diagram showing the operation of the evaluation system in the embodiment. [Figure 14] 11 is a flowchart showing a condition optimization process according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0007] The evaluation method according to the embodiment will be described in detail below with reference to the accompanying drawings. Note that the present invention is not limited to the embodiment.

[0008] (Embodiment) The evaluation method according to the embodiment performs evaluation related to NIL (Nano Imprint Lithography). In NIL, an original plate is prepared, a photosensitive agent is applied onto a substrate, and the original plate is imprinted onto the substrate. The photosensitive agent is exposed to light and hardened, the original plate is released from the substrate, and the substrate is processed using the pattern of the photosensitive agent as a mask, thereby forming a pattern on the substrate. In the evaluation method, evaluation is performed for defects such as pattern defects on the substrate caused by NIL.

[0009] For example, in NIL, an original OG as shown in Fig. 1(a) is prepared. In the following, the direction perpendicular to the back surface OGb of the original OG is defined as the Z direction, and two directions perpendicular to each other in a plane perpendicular to the Z direction are defined as the X direction and the Y direction. Fig. 1 is an XY plan view showing the configuration of the original OG and the substrate SB. Fig. 1(a) is a view of the original OG as seen from the front surface OGa.

[0010] The original OG is a plate-like member and may have a substantially rectangular shape in the XY plane. The original OG is also called a mold or template. The original OG may be made of a light-transmitting material such as glass, quartz, or transparent plastic. The original OG has a pedestal portion MS near the center in the XY directions. The pedestal portion MS includes a pattern region PT and a peripheral region PH. The pattern region PT extends in the XY directions and has a pattern including recesses PTa and protrusions PTb. The peripheral region PH is disposed outside the pattern region PT in the XY directions and surrounds the pattern region PT.

[0011] The original OG may be imprinted on a substrate SB as shown in Fig. 1(b). Fig. 1(b) is a view of the substrate SB as seen from the surface SBa side.

[0012] The substrate SB is a plate-like member and may have a circular shape in the XY plane view. The substrate SB is also called a wafer. The substrate SB may be made of a material containing a semiconductor (for example, silicon) as a main component.

[0013] The original OG is imprinted onto the substrate SB as shown in FIG. 2, and a pattern is formed on the substrate SB as shown in FIG. 3. FIG. 2 is a YZ cross-sectional view showing the imprinting of the original OG. FIGS. 2(a) and 2(b) show schematic diagrams of each step of the imprinting of the original OG. FIG. 3 is a YZ cross-sectional view showing the formation of a pattern on the substrate SB by imprinting the original OG. FIGS. 3(a) to 3(f) show in detail each step of the imprinting of the original OG.

[0014] Figures 2(a) and 2(b) show a cross section of the original OG taken along line AA in Figure 1. As shown in Figure 2(a), in the original OG, the pedestal portion MS is a portion that protrudes in a pedestal shape from the surface OGa to the -Z side in the YZ cross section.

[0015] In the steps shown in FIG. 2(a) and FIG. 3(a), a photosensitive agent RG is applied onto a substrate SB. The substrate SB has a base region BS and a film FM laminated in order in the Z direction. The photosensitive agent RG is made of a photosensitive resin, and may be made of, for example, a resin having photocuring properties. The photosensitive agent RG is also called a resist. The photosensitive agent RG may be applied onto the substrate SB by a spin coating method, or may be applied onto the substrate SB by an inkjet method. FIG. 3(a) illustrates an example in which the photosensitive agent RG is applied onto the substrate SB by a spin coating method.

[0016] In the steps shown in FIGS. 2(a) and 3(b), the XY positions of the original OG and the substrate SB are adjusted, and the original OG and the substrate SB are disposed so as to face each other in the Z direction.

[0017] In the steps shown in Figures 2(b) and 3(c), the original OG and the substrate SB are brought relatively close to each other in the Z direction until the pedestal portion MS comes into contact with the photosensitive agent RG. The original OG is imprinted onto the substrate SB. Once the pedestal portion MS comes into contact with the photosensitive agent RG and a predetermined time has elapsed, the recesses PTa of the original OG are filled with the photosensitive agent RG, and light such as ultraviolet light is irradiated onto the photosensitive agent RG. The photosensitive agent RG is exposed and hardened, and a photosensitive agent pattern RP is formed.

[0018] 3(d), the original OG and the substrate SB are moved away from each other in the Z direction. The original OG is released from the substrate SB. The photosensitive agent pattern RP includes recesses RPa corresponding to the protrusions PTb.

[0019] 3(e), the substrate SB is processed using the photosensitive agent pattern RP as a mask. The photosensitive agent pattern RP is transferred to the film FM, and recesses FMa corresponding to the recesses RPa are formed in the film FM.

[0020] 3(f), the photosensitive agent pattern RP is removed from the substrate SB, resulting in the substrate SB having the pattern of the recesses FMa formed in the film FM.

[0021] Thereafter, the XY position of the original OG relative to the substrate SB is shifted, and the steps shown in FIG. 2(a) and FIG. 3(b) and subsequent steps are repeated. As a result, as shown in FIG. 1(b), a plurality of shot regions SH-1 to SH-n (n is any integer of 2 or more) are formed on the substrate SB. The plurality of shot regions SH-1 to SH-n may be arranged in the XY direction on the surface SBa of the substrate SB. Each shot region SH includes a device region DV and a kerf region. The device region DV corresponds to the pattern region PT of the original OG. The device region DV is a region to which the pattern of the pattern region PT is transferred. The kerf region KF corresponds to the peripheral region PH of the original OG. The kerf region KF is disposed outside the device region DV in the XY direction and surrounds the device region DV in the XY plane view. The plurality of shot regions SH-1 to SH-n are partitioned by an outer peripheral region DL. The outer peripheral region DL is disposed outside each shot region SH in the XY direction. The outer peripheral region DL is also called a dicing line.

[0022] In NIL, defects such as pattern defects on the substrate SB may occur. Causes of defects include: (1) Impurities or defects in the original pattern (e.g., collapse of the convex parts of the original OG), (2) Malfunction of the pattern RP of the photosensitive material RG (e.g., collapse or peeling of the pattern RP), (3) A defect regarding the imprinting of the original OG onto the substrate SB (for example, insufficient filling of the photosensitive material RG into the recesses of the original OG or leakage of the photosensitive material RG outside the shot area SH, etc.) It is possible that:

[0023] Among these, (3) is a phenomenon specific to a direct pattern transfer method such as NIL, and can occur depending on the conditions related to the imprinting of the master OG. The conditions related to the imprinting of the master OG include, for example, the coating conditions of the photosensitive agent PG and the imprinting conditions of the master OG.

[0024] The application conditions of the photosensitive agent RG may include the material of the photosensitive agent RG. When the photosensitive agent RG is applied by a spin coating method, the application conditions of the photosensitive agent RG may include the rotation speed of the substrate SB during application, the supply speed of the photosensitive agent RG, etc. When the photosensitive agent RG is applied by an inkjet method, the application conditions of the photosensitive agent RG may include the size of the droplets of the photosensitive agent RG, the supply speed of the droplets of the photosensitive agent RG, the supply density of the droplets of the photosensitive agent RG, etc.

[0025] The stamping conditions of the original plate OG may include the magnitude of pressure applied to the original plate OG when stamping, a predetermined time for filling the photosensitive agent RG into the recess PTa of the original plate OG, and the magnitude of force used to peel the original plate OG from the substrate SB when releasing.

[0026] For example, the occurrence distribution of defects due to (3) is determined by the complex interactions of various NIL parameters related to the imprinting of the master OG. The various NIL parameters include the type of photosensitive material RG, the underlayer, various physical properties of the atmospheric gas, the mechanical structure of the master OG, various conditions when the master OG is pressed, the distribution and size of the imprint pattern, and the initial state of the photosensitive material RG.

[0027] The seepage of the photosensitive agent RG outside the shot area SH is unlikely to immediately cause a defect, but insufficient filling of the photosensitive agent RG in the recesses PTa of the original OG is likely to immediately cause a defect, with some exceptions. These exceptions include cases where the position where the defect will occur is known in advance and the relevant location is a pattern that allows defects, or where the relevant location is a part where the defect can be recovered.

[0028] In the initial stage of adjusting the conditions for imprinting the master OG, defects with various distributions occur. This defect distribution is thought to reflect various NIL parameters, so it can be a hint for obtaining appropriate values ​​for the NIL parameters. In other words, if the NIL parameters are adjusted and it can be determined by some comparison method that the defect distribution has "improved" from the previous time, then optimization is possible.

[0029] In reality, however, there are a great number of NIL parameters to be adjusted, and it is difficult to adjust all of them through experiments. In addition, if the NIL parameters are adjusted by increasing or decreasing the number of defects, there is a high possibility that the solution will be a locally appropriate solution (a false solution).

[0030] Therefore, it is effective to obtain a set of NIL parameters that is closer to the true appropriate solution by performing a highly accurate simulation (incorporating a complex physical model) and fitting the defect distribution pattern.

[0031] The evaluation method quantitatively evaluates the two-dimensional defect distribution in the substrate pattern created by NIL. For the two-dimensional defect distribution, actual measurement defect distribution and simulation defect distribution are prepared, each is converted into a probability density distribution, and the similarity between the two is numerically evaluated using a comparison function. This quantitative evaluation makes it possible to verify the accuracy of the simulation and to adjust the NIL parameters.

[0032] The evaluation method can be executed by an evaluation system 1 as shown in Fig. 4. Fig. 4 is a diagram showing the functional configuration of the evaluation system 1 that executes the evaluation method.

[0033] When the evaluation system 1 receives a start request, it can start the evaluation program 81 and execute the evaluation method in accordance with the evaluation program 81. The evaluation method will be described in detail later.

[0034] The evaluation system 1 functionally comprises a generation unit 2, a conversion unit 3, a filter unit 4, a restoration unit 5, a threshold processing unit 6, a calculation unit 7, a storage unit 8, and a display control unit 9. The storage unit 8 stores an evaluation program 81.

[0035] The generator 2 generates two-dimensional data TD1. The two-dimensional data TD1 indicates a defect distribution of the substrate SB1 on which a pattern is formed by imprinting the original OG1 onto the photosensitive agent RG2 on the substrate SB1.

[0036] For example, as shown in FIG. 2 and FIG. 3, the original OG1 is imprinted on the photosensitive agent RG on the substrate SB1 to form a pattern on the substrate SB1. The substrate SB1 is carried into a defect inspection device 200 (see FIG. 8). As shown in FIG. 3(g), the defect inspection device 200 optically acquires an image (surface image) of the surface of the substrate SB1. The defect inspection device 200 analyzes the surface image and generates measurement data 83 according to the analysis result. The measurement data 83 includes the level and two-dimensional position of the defect for one or more defects. The two-dimensional position may include coordinates with the origin being a reference position in the shot area SH (for example, the center of the shot area SH). The defect inspection device 200 supplies the measurement data 83 to the evaluation system 1. The evaluation system 1 receives the measurement data 83 of the defect inspection device 200 and stores it in the memory unit 8. The generating unit 2 reads out the measurement data 83 from the storage unit 8, and generates two-dimensional data TD1 (see FIG. 11(a)) in which defect levels of 1 or more are mapped within the shot area SH according to the measurement data 83.

[0037] The generator 2 generates two-dimensional data TD2. The two-dimensional data TD2 indicates a predicted distribution of defects in the substrate SB2.

[0038] For example, the defect occurrence simulation device 300 (see FIG. 8) predicts NIL parameters using a physical model equivalent to the processing shown in FIG. 2 and FIG. 3. The defect occurrence simulation device 300 calculates the distribution of defects that occur in the pattern formation on the substrate SB2 by NIL using the predicted NIL parameters. The defect occurrence simulation device 300 generates simulation data 84 as a calculation result. The simulation data 84 includes the level and two-dimensional position of the defect for one or more defects. The two-dimensional position may include coordinates with a reference position (for example, the center of the shot area SH) in the shot area SH as the origin. The evaluation system 1 receives the simulation data 84 of the defect occurrence simulation device 300 and stores it in the storage unit 8. The evaluation system 1 may receive the simulation data 84 in advance for a plurality of different sets of NIL parameters. The evaluation system 1 may configure the simulation data 84 as a library in association with information identifying the set of NIL parameters and store it in the storage unit 8. The generator 2 generates two-dimensional data TD2 (see FIG. 11(b)) in which the levels and two-dimensional positions of one or more defects are mapped within the shot area SH in accordance with the simulation data 84.

[0039] The conversion unit 3 acquires the two-dimensional data TD1 from the generation unit 2. The conversion unit 3 performs a first conversion process on the two-dimensional data TD1 to generate a spectrum SP1.

[0040] The first transformation process may be any transformation that allows inversion or any series expansion that allows inverse series expansion. The first transformation process may be a transformation corresponding to a series expansion using orthogonal basis functions. The first transformation process may be a discrete cosine transform corresponding to a Fourier series expansion, a series expansion using Legendre polynomials (hereinafter referred to as Legendre series expansion), or a series expansion using Chebyshev polynomials (hereinafter referred to as Chebyshev series expansion).

[0041] The first transformation process may be a transformation that is invertible and suitable for the defect distribution, or a series expansion that is invertible and suitable for the defect distribution. For example, when defects tend to be distributed along the kerf region KF (see FIG. 1(b)), the first transformation process may be a discrete cosine transformation.

[0042] When the defects tend to be distributed along the outer circumferential region DL (see FIG. 1(b)), the first transformation process may be a Legendre series expansion.

[0043] When the defects tend to be distributed along the kerf region KF (see FIG. 1(b)) and along the outer circumferential region DL (see FIG. 1(b)), the first transformation process may be a discrete cosine transformation.

[0044] The conversion unit 3 acquires the two-dimensional data TD2 from the generation unit 2. The conversion unit 3 performs a first conversion process on the two-dimensional data TD2 to generate a spectrum SP2.

[0045] The filter unit 4 obtains the spectrum SP1 from the conversion unit 3. The filter unit 4 performs a filtering process on the spectrum SP1.

[0046] The filtering process may be any process suitable for converting the defect distribution represented by the two-dimensional data TD1 before conversion into a probability density distribution. The filtering process may be a process for smoothing the defect distribution represented by the two-dimensional data TD1 before conversion. The filtering process may be a low-pass filter process, or a process for leaving components of a predetermined order or less among the order components expanded in series.

[0047] The filter unit 4 obtains the spectrum SP2 from the conversion unit 3. The filter unit 4 performs a filtering process on the spectrum SP2.

[0048] The restoration unit 5 acquires the filtered spectrum SP1 from the filter unit 4. The restoration unit 5 performs a second conversion process on the filtered spectrum SP1 to restore two-dimensional data TD1a. The two-dimensional data TD1a is data in which the original two-dimensional data TD1 is probability density distributed. The two-dimensional data TD1a can be regarded as data in which the probability density levels of defects are two-dimensionally mapped within the shot area SH.

[0049] The second transformation process is an inverse transformation process of the first transformation process. If the first transformation process is a discrete cosine transformation, the second transformation process may be an inverse discrete cosine transformation. If the first transformation process is a Legendre series expansion, the second transformation process may be an inverse Legendre series expansion. If the first transformation process is a Chebyshev series expansion, the second transformation process may be an inverse Chebyshev series expansion. Here, the inverse series expansion refers to a method of restoring the original distribution using the spectrum obtained by series expansion and the basis function used in the expansion.

[0050] The restoration unit 5 acquires the filtered spectrum SP2 from the filter unit 4. The restoration unit 5 performs a second conversion process on the filtered spectrum SP2 to restore two-dimensional data TD2a. The two-dimensional data TD2a is data in which the original two-dimensional data TD2 is probability density distributed. The two-dimensional data TD2a can be regarded as data in which the probability density levels of defects are two-dimensionally mapped within the shot area SH.

[0051] The threshold processing unit 6 acquires the restored two-dimensional data TD1a from the restoration unit 5. The threshold processing unit 6 performs threshold processing on the restored two-dimensional data TD1a. The threshold processing unit 6 replaces the level of data that is below a threshold level among the multiple data included in the restored two-dimensional data TD1a with the threshold level.

[0052] Threshold processing is a process of replacing the levels of data below a threshold level with the threshold level to raise the minimum level. By raising the minimum level, the influence of data corresponding to minute defects or errors can be mitigated, and the accuracy of the comparison process to be performed after threshold processing can be improved.

[0053] The threshold level may be experimentally determined in advance. The threshold level may be experimentally determined in advance based on the maximum level of data when there is one defect. The threshold level may be determined by multiplying the maximum level of data when there is one defect by a coefficient greater than 0 and less than 1. The coefficient may be a number greater than 0 and less than 1 (e.g., 0.5) that is experimentally determined in advance.

[0054] Alternatively, the threshold level may be experimentally determined in advance based on the maximum level of the data when the defect diameter is approximately the thickness of the photosensitive agent RG. The threshold level may be determined by multiplying the maximum level of the data when the defect diameter is approximately the thickness of the photosensitive agent RG by a coefficient greater than 0 and less than 1. The coefficient may be a number greater than 0 and less than 1 (e.g., 0.5) that is experimentally determined in advance.

[0055] Alternatively, the threshold level may be experimentally determined in advance based on the maximum level of data when the area of ​​the defect is approximately the square of the thickness of the photosensitive agent RG. The threshold level may be determined by multiplying the maximum level of data when the area of ​​the defect is approximately the square of the thickness of the photosensitive agent RG by a coefficient greater than 0 and less than 1. The coefficient may be a number greater than 0 and less than 1 (e.g., 0.5) that is experimentally determined in advance.

[0056] For example, when the level of the data included in the two-dimensional data TD1a is D1 and the threshold level is Δ, the threshold processing unit 6 performs the process shown in the following formula 1.

number

[0057] In Formula 1, max() is a max function that returns the maximum value of the arguments in parentheses.

[0058] As a result, the threshold processor 6 can obtain the two-dimensional data TD1b after the threshold processor.

[0059] The threshold processing unit 6 acquires the restored two-dimensional data TD2a from the restoration unit 5. The threshold processing unit 6 performs threshold processing on the restored two-dimensional data TD2a. The threshold processing unit 6 replaces the level of data that is below a threshold level among the multiple data included in the restored two-dimensional data TD2a with the threshold level.

[0060] For example, when the level of the data included in the two-dimensional data TD2a is D2 and the threshold level is Δ, the threshold processing unit 6 performs the process shown in the following equation 2.

number

[0061] In Equation 2, max() is a max function that returns the maximum value of the arguments in parentheses, allowing the threshold processing unit 6 to obtain the two-dimensional data TD2b after threshold processing.

[0062] The calculation unit 7 obtains the threshold-processed two-dimensional data TD1b and the threshold-processed two-dimensional data TD2b from the threshold processing unit 6. The calculation unit 7 applies a comparison function to the threshold-processed two-dimensional data TD1b and the threshold-processed two-dimensional data TD2b to calculate the degree of match. The degree of match may be a value between 0 and 1, and is closer to 1 as the distribution shapes become more similar.

[0063] The calculation unit 7 applies a first comparison function to the processed first two-dimensional data and the processed second two-dimensional data to calculate a first degree of agreement. The first comparison function may be a function suitable for evaluating the similarity (trend agreement) of distribution shapes. The first comparison function may be, for example, a comparison function based on mode coupling theory. The first degree of agreement is a value between 0 and 1, which is closer to 1 as the distribution shapes become more similar.

[0064] For example, when the level of the data included in the two-dimensional data TD1b is D1 and the level of the data included in the two-dimensional data TD2b is D2, the calculation unit 7 calculates the degree of match ξ shown in the following formula 3.

number

[0065] In Equation 3, the denominator on the right side represents the product of the convolution integral of the square of the data level D1 and the convolution integral of the square of the data level D2. The numerator on the right side represents the square of the convolution integral of the product of the data level D1 and the data level D2. The degree of agreement ξ is a value between 0 and 1, and is closer to 1 as the distribution of the two-dimensional data TD1b and the distribution of the two-dimensional data TD2b become more similar.

[0066] In response to a predetermined condition being satisfied, the calculation unit 7 may switch the comparison function to be used from the first comparison function to the second comparison function. After switching, the calculation unit 7 applies the second comparison function to the processed first two-dimensional data and the processed second two-dimensional data to calculate a second degree of coincidence.

[0067] The second comparison function may be a function suitable for evaluating the absolute degree of similarity. For example, the second comparison function may be a comparison function that applies the Jaccard index. The second degree of similarity is a value between 0 and 1, which is closer to 1 as the distribution shapes become more similar.

[0068] For example, when the level of the data included in the two-dimensional data TD1b is D1 and the level of the data included in the two-dimensional data TD2b is D2, the calculation unit 7 calculates the degree of coincidence Σ shown in the following formula 4.

number

[0069] In Equation 4, max() is a max function that returns the maximum value of the arguments in parentheses. min() is a min function that returns the minimum value of the arguments in parentheses. The denominator on the right side represents the convolution integral of the larger of the data levels D1 and D2. The numerator on the right side represents the convolution integral of the smaller of the data levels D1 and D2. The degree of agreement Σ is a value between 0 and 1, and is closer to 1 as the distribution of the two-dimensional data TD1b and the distribution of the two-dimensional data TD2b become more similar.

[0070] The specified conditions include the first degree of match exceeding a specified value (e.g., 0.5) and the number of times that the first degree of match is a specified percentage or more (e.g., 10% or more) of the maximum value in the past reaches a specified number of times (e.g., 10 times).

[0071] Alternatively, the calculation unit 7 may calculate the second degree of coincidence by applying the second comparison function to the processed first two-dimensional data and the processed second two-dimensional data from the beginning.

[0072] Alternatively, the calculation unit 7 may calculate the first degree of agreement using the first comparison function and the second degree of agreement using the second comparison function in parallel. The calculation unit 7 may calculate a score according to the first degree of agreement and the second degree of agreement. The score may be obtained by multiplying the first degree of agreement and the second degree of agreement and taking the square root.

[0073] For example, when the degree of match calculated by the first comparison function is ξ and the degree of match calculated by the second comparison function is Σ, the calculation unit 7 may calculate the score SC as shown in the following formula 5.

number

[0074] The calculation unit 7 may switch the calculation result from the score to the second degree of agreement when the number of times that the score SC is equal to or greater than a predetermined percentage (e.g., 10% or more) of the past maximum value reaches a predetermined number (e.g., 10 times). After the switching, the calculation unit 7 may calculate the second degree of agreement using the second comparison function without calculating the first degree of agreement using the first comparison function.

[0075] After calculating the degree of agreement, the calculation unit 7 may compare the calculated degree of agreement with a previously calculated degree of agreement to obtain a correlation between the direction of change with the set of NIL parameters and the direction of change in the degree of agreement.

[0076] For example, the calculation unit 7 compares the previous degree of agreement with the current degree of agreement. The calculation unit 7 compares the previous NIL parameter with the current NIL parameter to identify the change direction of the NIL parameter. If the current degree of agreement is higher than the previous degree of agreement, the calculation unit 7 identifies the change direction of the NIL parameter as the change direction that increases the degree of agreement. The calculation unit 7 may store information on the change direction of the NIL parameter that increases the degree of agreement in the storage unit 8 as correlation information 82.

[0077] Alternatively, the calculation unit 7 compares the previous degree of agreement with the current degree of agreement to identify a change direction (for example, a change direction that increases the degree of agreement or a change direction that decreases the degree of agreement). The calculation unit 7 compares the previous NIL parameter with the current NIL parameter to identify a change direction of the NIL parameter. The calculation unit 7 may store information in which the change direction of the NIL parameter and the change direction of the degree of agreement are associated with each other as correlation information 82 in the storage unit 8.

[0078] This allows the degree of agreement between the defect distribution obtained by actual measurement and the defect distribution obtained by simulation to be quantitatively evaluated, and the degree of agreement can be used as a guideline for improving the accuracy of the simulation. For example, if the degree of agreement is equal to or greater than a predetermined value, it can be determined that the simulation satisfies the required accuracy. In addition, by performing Bayesian estimation while using the correlation information 82, the degree of agreement can be effectively improved, and the total processing time required for improving the accuracy of the simulation can be shortened.

[0079] The storage unit 8 stores an evaluation program 81, correlation information 82, measurement data 83, and simulation data 84.

[0080] The display control unit 9 displays predetermined information on the display screen. The display control unit 9 may display the two-dimensional data TD1, TD2 on the display screen. The display control unit 9 may display the restored two-dimensional data TD1a, TD2a on the display screen. The display control unit 9 may display the threshold processed two-dimensional data TD1b, TD2b on the display screen. The display control unit 9 may display the calculation result of the degree of similarity on the display screen. This allows the user to be notified of information regarding the evaluation of the evaluation system 1.

[0081] The evaluation system 1 may be configured with hardware as shown in Fig. 5. Fig. 5 is a diagram showing the hardware configuration of the evaluation system 1.

[0082] The evaluation system 1 has a CPU (Central Processing Unit) 101, a non-volatile storage unit 102, a volatile storage unit 103, a display unit 104, an input unit 105, and a bus 106. The CPU 101, the non-volatile storage unit 102, the volatile storage unit 103, the display unit 104, and the input unit 105 are communicably connected to each other via the bus 106. The non-volatile storage unit 102 is a ROM (Read Only Memory), a HDD (Hard Disk Drive), a SSD (Solid State Drive), or the like. The volatile storage unit 103 is a RAM (Random Access Memory), or the like. The display unit 104 is a liquid crystal display, or the like. The input unit 105 is a keyboard, a mouse, a touch panel, or the like.

[0083] Each of the generating unit 2, the converting unit 3, the filter unit 4, the restoring unit 5, the threshold processing unit 6, the calculating unit 7, and the display control unit 9 (see FIG. 4) may be implemented in hardware by a circuit or the like in the CPU 101. Each of the generating unit 2, the converting unit 3, the filter unit 4, the restoring unit 5, the threshold processing unit 6, the calculating unit 7, and the display control unit 9 may be implemented in software as a functional module that is deployed in a buffer area of ​​the volatile storage unit 103 collectively when the program is compiled or sequentially according to the progress of the processing by the program. Alternatively, each of the generating unit 2, the converting unit 3, the filter unit 4, the restoring unit 5, the threshold processing unit 6, the calculating unit 7, and the display control unit 9 may be implemented in hardware in part and in software in the rest. The storage unit 8 (see FIG. 4) may be implemented in hardware such as the non-volatile storage unit 102 or the volatile storage unit 103.

[0084] The evaluation program 81 may be stored in the non-volatile storage unit 102 .

[0085] Alternatively, the evaluation program 81 may be provided as a file in an installable or executable format recorded on a recording medium readable by the evaluation system 1, such as a CD-ROM, a flexible disk (FD), a CD-R, or a digital versatile disk (DVD).

[0086] Alternatively, the evaluation program 81 may be configured to be stored on a computer connected to a network such as the Internet and to be provided in the evaluation system 1 by being downloaded via the network. Also, the program executed in the evaluation system 1 of this embodiment may be configured to be provided or distributed via a network such as the Internet.

[0087] Next, a method for manufacturing a semiconductor device including an evaluation method will be described with reference to FIG. 6. FIG. 6 is a flow chart showing a method for manufacturing a semiconductor device including an evaluation method. FIG. 6 focuses on the manufacture of a second device. Consider a situation in which a lot including a first device and a lot including a second device are processed in sequence on a semiconductor device manufacturing line. In this situation, the first device that is processed before the second device is used to manufacture the second device.

[0088] The evaluation system 1 performs a calibration process using a first device (S1). The evaluation system 1 adjusts the NIL parameters of the defect occurrence simulation device 300 to match the target (characteristics of the real device) so that the degree of agreement falls within an allowable range. In this way, the evaluation system 1 identifies the correlation between the change direction of the NIL parameters of the defect occurrence simulation device 300 and the change direction of the defect distribution while making the defect distribution of the real device (first device) and the defect distribution of the simulator coincide with each other to some extent, and generates correlation information 82.

[0089] The evaluation system 1 performs a layout optimization process for the second device (S2). The evaluation system 1 optimizes the NIL layout for the real device (second device) while adjusting the NIL parameters of the defect occurrence simulation device 300 using the correlation information 82 generated in S1.

[0090] The evaluation system 1 performs a condition optimization process for the second device (S3). The evaluation system 1 optimizes the conditions related to imprinting of the original OG for the real device (second device) while adjusting the NIL parameters of the defect occurrence simulation apparatus 300 using the correlation information 82 generated in S1. Then, using the optimized conditions, a pattern is formed on the substrate SB as shown in Figs. 2 and 3 to manufacture the real device (second device).

[0091] Next, the calibration process (S1) of the first device will be described in detail with reference to Fig. 7. Fig. 7 is a flowchart showing the calibration process (S1).

[0092] The circuit design apparatus performs a circuit design for the first device based on predetermined design information and / or instructions from a user (S11), and generates schematic data and supplies it to the layout design apparatus.

[0093] The layout design device performs layout design based on the schematic data and / or instructions from the user (S12), and generates layout data.

[0094] The layout design device performs layout design of the dummy patterns (S13). The layout design device generates multiple layout data in which the presence or absence of addition of dummy patterns and the number of added dummy patterns are changed with respect to the layout data of S12, and supplies the generated layout data to the defect occurrence simulation device 300. When the layout design of the dummy patterns is completed, the layout design device notifies the evaluation system 1 of the completion.

[0095] In response to the notification from the layout design device, the evaluation system 1 causes the defect occurrence simulation device 300 to repeat the simulation based on the design of S13, and optimizes the pattern design including the presence or absence of a dummy pattern (S14).

[0096] The evaluation system 1 supplies the proposal of the NIL parameters to the defect occurrence simulation device 300. The defect occurrence simulation device 300 calculates the distribution of defects that occur in pattern formation on the substrate SB by NIL according to the layout data, while changing the NIL parameters for each of the multiple layout data generated in S13 according to the proposal from the evaluation system 1. The defect occurrence simulation device 300 supplies the calculation result to the evaluation system 1 as simulation data 84.

[0097] The evaluation system 1 selects layout data that satisfies a first criterion from among the multiple layout data as appropriate layout data based on the simulation data 84. The first criterion may be, for example, that the average number of defects for multiple NIL parameters is the smallest, or that the minimum number of defects for multiple NIL parameters is the smallest. The evaluation system 1 supplies the appropriated layout data to the defect occurrence simulation device 300.

[0098] The evaluation system 1 repeats simulations in the defect occurrence simulation device 300 based on the design in S14, and optimizes various processing conditions during NIL imprinting (S15).

[0099] The evaluation system 1 supplies the proposal of the NIL parameters to the defect occurrence simulation device 300. For the layout data selected in S14, the defect occurrence simulation device 300 calculates a distribution of defects that will occur in pattern formation on the substrate SB by NIL according to the layout data while changing the NIL parameters among a plurality of NIL parameters according to the proposal from the evaluation system 1. The defect occurrence simulation device 300 supplies the calculation result to the evaluation system 1 as simulation data 84.

[0100] The evaluation system 1 selects, as appropriate NIL parameters, NIL parameters that satisfy a second criterion from among the multiple NIL parameters based on the simulation data 84. The second criterion may be, for example, that the number of defects is minimal. The evaluation system 1 supplies the appropriate NIL parameters to the defect occurrence simulation device 300.

[0101] The evaluation system 1 performs final adjustments by a NIL experiment using the actual original OG (S16). The evaluation system 1 starts the evaluation program 81 and executes the evaluation method. Details of S16 will be described later.

[0102] Processing including NIL processing is performed with the pattern and processing conditions optimized in S11 to S16, and a first device is manufactured (S17).

[0103] For example, pattern information of the original OG is generated according to the pattern optimized in S14 and finally adjusted in S16. The original OG is manufactured according to the pattern information of the original OG. A photosensitive agent is applied onto the substrate SB. The original OG is imprinted onto the substrate SB under the conditions optimized in S15 and finally adjusted in S16. The photosensitive agent is exposed and hardened, the original OG is released from the substrate SB, and the substrate SB is processed using the pattern of the photosensitive agent as a mask, forming a pattern on the substrate SB. In this manner, a first device is manufactured.

[0104] Next, S16 will be described in detail.

[0105] In S16, the evaluation system 1 operates as shown in Fig. 8, and executes the evaluation method shown in Fig. 9. Fig. 8 is a data flow diagram showing the operation of the evaluation system 1. Fig. 9 is a flowchart showing the evaluation method.

[0106] In Figure 8, for the sake of simplicity, in the evaluation system 1, the components corresponding to the generation unit 2, conversion unit 3, filter unit 4, restoration unit 5, threshold processing unit 6, and calculation unit 7 (see Figure 4) are designated as an arithmetic unit 1a, a part of the memory unit 8 is designated as a memory device 1b, and the other part is designated as a memory device 1c.

[0107] The storage device 1b is disposed between the defect inspection device 200 and the arithmetic device 1a. The storage device 1b can store measurement data 83 supplied from the defect inspection device 200. The arithmetic device 1a can read out the measurement data 83 from the storage device 1b.

[0108] The storage device 1c is disposed between the defect occurrence simulation device 300 and the arithmetic device 1a. The storage device 1c can store simulation data 84 supplied from the defect occurrence simulation device 300. The arithmetic device 1a can read the simulation data 84 from the storage device 1c.

[0109] In S16, the evaluation system 1 performs the following process.

[0110] The evaluation system 1 acquires measurement data 83 as target data from the defect inspection device 200 (S21) and stores it in the storage device 1a.

[0111] The evaluation system 1 determines the NIL parameters optimized in S15 as initial parameters (S22).

[0112] The evaluation system 1 causes the defect occurrence simulation device 300 to perform a simulation (S23). The evaluation system 1 supplies the defect occurrence simulation device 300 with a proposal for NIL parameters (see FIG. 8). If it is the first time, the evaluation system 1 supplies the defect occurrence simulation device 300 with the proposal for the NIL parameters determined in S22. If it is the second or subsequent time, the evaluation system 1 supplies the defect occurrence simulation device 300 with a proposal for NIL parameters different from the already proposed NIL parameters.

[0113] The defect occurrence simulation device 300 performs a simulation using NIL parameters according to the proposal of the evaluation system 1, and calculates the distribution of defects that occur during pattern formation on the substrate SB by NIL. The defect occurrence simulation device 300 generates simulation data 84 as a calculation result. The simulation data 84 includes the defect level and two-dimensional position for one or more defects. The two-dimensional position may include coordinates with a reference position in the shot area SH (e.g., the center of the shot area SH) as the origin. The evaluation system 1 receives the simulation data 84 from the defect occurrence simulation device 300 and stores it in the storage device 1c.

[0114] The evaluation system 1 performs a data extraction process (S24). In the data extraction process, as shown in Fig. 10, data suitable for comparison is extracted from the measurement data 83 acquired in S21 and the simulation data 84 acquired in S23. Fig. 10 is a flowchart showing the flow of the data extraction process (S24).

[0115] In the data extraction process (S24), the processes of S31 to S33 and the processes of S34 to S36 are carried out in parallel.

[0116] In S31, the evaluation system 1 generates two-dimensional data TD1 in accordance with the measurement data 83 acquired in S21.

[0117] The evaluation system 1 generates two-dimensional data TD1 as shown in FIG. 11(a) according to information on the defect level and two-dimensional position contained in the measurement data 83. FIG. 11 is a diagram showing two-dimensional data TD1 and TD2. The two-dimensional data TD1 is data in which defect levels of 1 or more are mapped within a shot area SH. In the example of FIG. 11(a), the defect level is indicated by the shade of color, and the closer the color is to white, the greater the defect level is. In the two-dimensional data TD1, the boundary between the white area and the black area is relatively clear. That is, the two-dimensional data TD1 has a large change in defect level between areas with and without defects, and is vulnerable to variations in the positions of defective areas, and is not suitable for comparison as it is.

[0118] Therefore, in S32, the evaluation system 1 converts the two-dimensional data TD1 into a probability density distribution.

[0119] The evaluation system 1 performs a first conversion process on the two-dimensional data TD1 to generate a spectrum SP1.

[0120] The first transformation process may be any transformation that is invertible. The first transformation process may be a transformation corresponding to a series expansion using orthogonal basis functions. The first transformation process may be a discrete cosine transform corresponding to a Fourier series expansion, a series expansion using Legendre polynomials, or a series expansion using Chebyshev polynomials.

[0121] The first transformation process may be an invertible transformation process suitable for the defect distribution. For example, if the defects tend to be distributed along the kerf region KF (see FIG. 1(b)), the first transformation process may be a discrete cosine transformation.

[0122] When the defects tend to be distributed along the outer circumferential region DL (see FIG. 1(b)), the first transformation process may be a Legendre series expansion.

[0123] When the defects tend to be distributed along the kerf region KF (see FIG. 1(b)) and along the outer circumferential region DL (see FIG. 1(b)), the first transformation process may be a discrete cosine transformation.

[0124] The evaluation system 1 applies filtering to the spectrum SP1.

[0125] The filtering process may be any process suitable for converting the defect distribution represented by the two-dimensional data TD1 before conversion into a probability density distribution. The filtering process may be a process for smoothing the defect distribution represented by the two-dimensional data TD1 before conversion. The filtering process may be a low-pass filter process, or a process for leaving components of a predetermined order or less among the order components expanded in series.

[0126] The evaluation system 1 performs a second conversion process on the spectrum SP1 after the filtering process to restore the two-dimensional data TD1a. The two-dimensional data TD1a is data in which the original two-dimensional data TD1 is probability density distributed. The two-dimensional data TD1a can be regarded as data in which the probability density level of defects is two-dimensionally mapped within the shot area SH.

[0127] The second transformation process is an inverse transformation of the first transformation process. If the first transformation process is a discrete cosine transformation, the second transformation process may be an inverse discrete cosine transformation. If the first transformation process is a Legendre series expansion, the second transformation process may be an inverse Legendre series expansion. If the first transformation process is a Chebyshev series expansion, the second transformation process may be an inverse Chebyshev series expansion.

[0128] The obtained two-dimensional data TD1a includes minute defects and a noise level, but the minute defects and the noise level may hinder a comparison of the overall trend.

[0129] Therefore, in S33, the evaluation system 1 performs threshold processing on the restored two-dimensional data TD1a. The evaluation system 1 replaces the level of data below a threshold level among the multiple data included in the restored two-dimensional data TD1a with the threshold level.

[0130] Threshold processing is a process of replacing the levels of data that are below a threshold level with the threshold level to raise the minimum level. By raising the minimum level, the effects of minute defects and noise levels can be mitigated, and the accuracy of the comparison process that is to be performed after threshold processing can be improved.

[0131] The threshold level may be experimentally determined in advance. The threshold level may be experimentally determined in advance based on the maximum level of data when there is one defect. The threshold level may be determined by multiplying the maximum level of data when there is one defect by a coefficient greater than 0 and less than 1. The coefficient may be a number greater than 0 and less than 1 (e.g., 0.5) that is experimentally determined in advance. The threshold level may be experimentally determined in advance based on the maximum level of data when the defect diameter is approximately the thickness of the photosensitive agent RG. The threshold level may be determined by multiplying the maximum level of data when the defect diameter is approximately the thickness of the photosensitive agent RG by a coefficient greater than 0 and less than 1. The coefficient may be a number greater than 0 and less than 1 (e.g., 0.5) that is experimentally determined in advance.

[0132] For example, when the level of the data included in the two-dimensional data TD1a is D1 and the threshold level is Δ, the evaluation system 1 performs the process shown in Equation 1.

[0133] In Formula 1, max() is a max function that returns the maximum value of the arguments in parentheses.

[0134] This enables the evaluation system 1 to obtain the two-dimensional data TD1b after threshold processing.

[0135] On the other hand, in S34, the evaluation system 1 generates two-dimensional data TD2 in accordance with the simulation data 84 acquired in S23.

[0136] The evaluation system 1 generates two-dimensional data TD2 as shown in FIG. 11(b) according to the defect level and two-dimensional position information contained in the simulation data 84. The two-dimensional data TD2 is data in which defect levels of 1 or more are mapped within the shot area SH. In the example of FIG. 11(b), the defect level is indicated by the shade of color, and the closer the color is to white, the greater the defect level is. In the two-dimensional data TD2, the boundary between the white area and the black area is relatively clear. In other words, the two-dimensional data TD2 has a large change in defect level between areas with and without defects, is vulnerable to variations in the positions of defective areas, and is not suitable for comparison as it is.

[0137] Therefore, in S35, the evaluation system 1 converts the two-dimensional data TD2 into a probability density distribution.

[0138] The evaluation system 1 performs a first conversion process on the two-dimensional data TD2 to generate a spectrum SP2. The evaluation system 1 performs a filter process on the spectrum SP2. The evaluation system 1 performs a second conversion process on the filtered spectrum SP2 to restore two-dimensional data TD2a. The two-dimensional data TD2a is data in which the original two-dimensional data TD2 is probability density distributed. The two-dimensional data TD2a can be regarded as data in which the probability density levels of defects are two-dimensionally mapped within the shot area SH.

[0139] The obtained two-dimensional data TD2a includes minute defects and a noise level, but the minute defects and the noise level may hinder a comparison of the overall trend.

[0140] Therefore, in S36, the evaluation system 1 performs threshold processing on the restored two-dimensional data TD2a. The evaluation system 1 replaces the level of data below the threshold level among the multiple data included in the restored two-dimensional data TD2a with the threshold level.

[0141] Threshold processing is a process of replacing the levels of data that are below a threshold level with the threshold level to raise the minimum level. By raising the minimum level, the effects of minute defects and noise levels can be mitigated, and the accuracy of the comparison process that is to be performed after threshold processing can be improved.

[0142] For example, when the level of the data included in the two-dimensional data TD2a is D2 and the threshold level is Δ, the evaluation system 1 performs the process shown in Equation 2.

[0143] In Equation 2, max() is a max function that returns the maximum value of the arguments in parentheses, allowing the evaluation system 1 to obtain the two-dimensional data TD2b after threshold processing.

[0144] When the two-dimensional data TD1b of S33 and the two-dimensional data TD2b of S36 are both obtained, the evaluation system 1 performs a quantitative comparison process (S25), as shown in FIGS.

[0145] Specifically, the evaluation system 1 applies a comparison function to the two-dimensional data TD1b after the threshold processing and the two-dimensional data TD2b after the threshold processing to calculate the degree of match. The degree of match may be a value between 0 and 1, and the closer to 1 the distribution shapes are to be similar, the closer the value is to 1.

[0146] The evaluation system 1 may calculate a first degree of agreement by applying a first comparison function to the processed first two-dimensional data and the processed second two-dimensional data. The first comparison function may be a function suitable for evaluating the similarity (trend agreement) of distribution shapes. The first comparison function may be, for example, a comparison function based on mode coupling theory. The first degree of agreement is a value between 0 and 1, which is closer to 1 as the distribution shapes become more similar.

[0147] For example, when the level of the data included in the two-dimensional data TD1b is D1 and the level of the data included in the two-dimensional data TD2b is D2, the calculation unit 7 calculates the degree of match ξ shown in Equation 3.

[0148] In Equation 3, the denominator on the right side represents the product of the convolution integral of the square of the data level D1 and the convolution integral of the square of the data level D2. The numerator on the right side represents the square of the convolution integral of the product of the data level D1 and the data level D2. The degree of agreement ξ is a value between 0 and 1, and is closer to 1 as the distribution of the two-dimensional data TD1b and the distribution of the two-dimensional data TD2b become more similar.

[0149] In response to a predetermined condition being satisfied, the evaluation system 1 may switch the comparison function used from the first comparison function to the second comparison function. After switching, the calculation unit 7 applies the second comparison function to the processed first two-dimensional data and the processed second two-dimensional data to calculate a second degree of agreement. The second comparison function may be a function suitable for evaluating an absolute degree of agreement. The second comparison function may be, for example, a comparison function that applies the Jaccard index. The second degree of agreement is a value between 0 and 1, inclusive, which is closer to 1 as the distribution shapes become more similar.

[0150] For example, when the level of the data included in the two-dimensional data TD1b is D1 and the level of the data included in the two-dimensional data TD2b is D2, the evaluation system 1 calculates the degree of match Σ shown in Equation 4.

[0151] In Equation 4, max() is a max function that returns the maximum value of the arguments in parentheses. min() is a min function that returns the minimum value of the arguments in parentheses. The denominator on the right side represents the convolution integral of the larger of the data levels D1 and D2. The numerator on the right side represents the convolution integral of the smaller of the data levels D1 and D2. The degree of agreement Σ is a value between 0 and 1, and is closer to 1 as the distribution of the two-dimensional data TD1b and the distribution of the two-dimensional data TD2b become more similar.

[0152] Alternatively, the evaluation system 1 may from the beginning calculate a second degree of match by applying a second comparison function to the processed first two-dimensional data and the processed second two-dimensional data.

[0153] Alternatively, the evaluation system 1 may calculate the first degree of agreement using the first comparison function and the second degree of agreement using the second comparison function in parallel. The calculation unit 7 may calculate a score according to the first degree of agreement and the second degree of agreement. The score may be obtained by multiplying the first degree of agreement and the second degree of agreement and taking the square root.

[0154] For example, when the degree of match calculated by the first comparison function is ξ and the degree of match calculated by the second comparison function is Σ, the evaluation system 1 may calculate a score SC as shown in Equation 5.

[0155] The evaluation system 1 may switch the calculation result from the score to the second degree of agreement when the number of times that the score SC is equal to or greater than a predetermined percentage (e.g., 10% or more) of the past maximum value reaches a predetermined number (e.g., 10 times). After the switching, the calculation unit 7 may calculate the second degree of agreement using the second comparison function without calculating the first degree of agreement using the first comparison function.

[0156] The evaluation system 1 judges whether the result of S25 satisfies the criteria (S26).

[0157] For example, if it is a first degree of match calculated in S25, the evaluation system 1 may determine that the result of S25 does not satisfy the criterion if the first degree of match is equal to or less than a first reference value, and may determine that the result of S25 satisfies the criterion if the first degree of match exceeds the first reference value. The first reference value may be determined in advance experimentally.

[0158] Alternatively, if it is the second degree of match calculated in S25, the evaluation system 1 may determine that the result of S25 does not satisfy the criterion if the second degree of match is equal to or less than a second reference value, and may determine that the result of S25 satisfies the criterion if the second degree of match exceeds the second reference value. The second reference value may be determined in advance by experiment.

[0159] Alternatively, if the degree of match calculated in S25 is switched from the first degree of match to the second degree of match midway, the evaluation system 1 may determine that the result of S25 does not satisfy the criterion until the switch. After the switch, the evaluation system 1 may determine that the result of S25 does not satisfy the criterion if the second degree of match is equal to or less than a second reference value. The evaluation system 1 may determine that the result of S25 satisfies the criterion if the second degree of match exceeds the second reference value. The second reference value may be determined in advance experimentally.

[0160] Alternatively, if it is a score calculated in S25, the evaluation system 1 may determine that the result of S25 does not satisfy the criterion if the score is equal to or less than a third reference value, and may determine that the result of S25 satisfies the criterion if the score exceeds the third reference value. The third reference value may be determined in advance experimentally.

[0161] After calculating the degree of match, the evaluation system 1 may compare the calculated degree of match with a previously calculated degree of match to find a correlation between the direction of change with the set of NIL parameters and the direction of change in the degree of match.

[0162] For example, the evaluation system 1 compares the previous degree of agreement with the current degree of agreement. The evaluation system 1 compares the previous NIL parameter with the current NIL parameter to identify the change direction of the NIL parameter. If the current degree of agreement is higher than the previous degree of agreement, the evaluation system 1 identifies the change direction of the NIL parameter as the change direction that increases the degree of agreement. The evaluation system 1 may store information on the change direction of the NIL parameter that increases the degree of agreement as correlation information 82 in the storage device 1c.

[0163] Alternatively, the evaluation system 1 compares the previous degree of agreement with the current degree of agreement to identify the direction of change (for example, the direction of change that increases the degree of agreement or the direction of change that decreases the degree of agreement). The evaluation system 1 compares the previous NIL parameter with the current NIL parameter to identify the direction of change of the NIL parameter. The evaluation system 1 may store information in which the direction of change of the NIL parameter and the direction of change of the degree of agreement are associated with each other as correlation information 82 in the storage device 1c.

[0164] If the result of S25 does not satisfy the criteria (No in S26), the evaluation system 1 changes the NIL parameters (S27). The evaluation system 1 performs Bayesian estimation while referring to the correlation information 82 to identify the change direction of the NIL parameters that will increase the degree of match, and determines the changed NIL parameters according to the identified change direction. After this, the evaluation system 1 performs the processes from S23 onwards.

[0165] That is, the evaluation system 1 repeats the loop process of S23 to S27 until the result of S25 satisfies the criterion (No in S26), and ends the process when the result of S25 satisfies the criterion (Yes in S26). That is, since the change of the NIL parameters is repeated until the result of S25 satisfies the criterion, the NIL parameters can be adjusted to match the target data acquired in S21 to a certain extent. Also, the correlation between the direction of change in the set of NIL parameters and the direction of change in the degree of agreement can be obtained as correlation information 82. This makes it possible to easily improve the accuracy of the simulation.

[0166] Next, details of the layout optimization process (S2) will be described with reference to Fig. 12. Fig. 12 is a flowchart showing the layout optimization process (S2).

[0167] The circuit design apparatus performs a circuit design for the second device based on predetermined design information and / or instructions from the user (S51), and generates schematic data and supplies it to the layout design apparatus.

[0168] The layout design device performs layout design based on the schematic data and / or instructions from the user (S52), and generates layout data.

[0169] The layout design device performs layout design of the dummy patterns (S53). The layout design device generates multiple layout data in which the presence or absence of addition of dummy patterns and the number of added dummy patterns are changed with respect to the layout data of S52, and supplies the generated layout data to the defect occurrence simulation device 300. When the layout design of the dummy patterns is completed, the layout design device notifies the evaluation system 1 of the completion.

[0170] In response to the notification from the layout design device, the evaluation system 1 causes the defect occurrence simulation device 300 to repeat the simulation based on the design in S53, and optimizes the pattern design including the presence or absence of a dummy pattern (S54).

[0171] In S54, the rating system 1 may operate as shown in Fig. 13. Fig. 13 is a data flow diagram showing the operation of the rating system 1.

[0172] In the evaluation system 1 shown in FIG. 13, the storage device 1b can store simulation data 84 supplied from the defect occurrence simulation device 300. The storage device 1c stores correlation information 82 and two-dimensional data 85. The two-dimensional data 85 is data in which the defect distribution is two-dimensionally mapped in advance within the shot area SH based on the measurement data acquired from the defect inspection device 200. The two-dimensional data 85 may be data generated based on the measurement data of the first device acquired in S16, or may be data generated based on the measurement data of another device. Other points are the same as those of the evaluation system 1 shown in FIG. 8.

[0173] In S54, the evaluation system 1 supplies the proposal of NIL parameters to the defect occurrence simulation device 300. The evaluation system 1 may change the NIL parameters while referring to the correlation information 82, and supply the proposal of the changed NIL parameters to the defect occurrence simulation device 300. The defect occurrence simulation device 300 calculates the distribution of defects that occur in pattern formation on the substrate SB by NIL according to the layout data while changing the NIL parameters according to the proposal from the evaluation system 1 for each of the multiple layout data generated in S53. The defect occurrence simulation device 300 supplies the calculation result to the evaluation system 1 as simulation data 84.

[0174] The evaluation system 1 may select, as appropriate layout data, layout data that satisfies a first criterion from among the multiple layout data, based on the simulation data 84. The first criterion may be, for example, that the average number of defects for multiple NIL parameters is the smallest, or that the minimum number of defects for multiple NIL parameters is the smallest.

[0175] Alternatively, the evaluation system 1 may execute an evaluation method similar to the evaluation method shown in FIG. 9 using simulation data 84 and two-dimensional data 85 for each of the multiple layout data. In this case, the two-dimensional data 85 corresponds to two-dimensional data TD1 corresponding to the target data, and the evaluation method omits acquisition of the target data (S21). In other respects, the evaluation system 1 is similar to the evaluation method of FIG. 9 executed in S16. The evaluation system 1 may select layout data that satisfies a fourth criterion from the multiple layout data as appropriate layout data. The fourth criterion may be, for example, the highest degree of match.

[0176] The evaluation system 1 supplies the optimized layout data to the defect occurrence simulation device 300 .

[0177] Next, details of the condition appropriateness process (S3) will be described with reference to Fig. 14. Fig. 14 is a flowchart showing the condition appropriateness process (S3).

[0178] In the condition appropriateness process (S3), the process of S61 and the process of S62 are carried out in parallel.

[0179] In S61, the evaluation system 1 causes the defect occurrence simulation device 300 to repeat simulation based on the design in S54, and optimizes various processing conditions during NIL imprinting.

[0180] In S61, the evaluation system 1 may operate as shown in FIG.

[0181] The evaluation system 1 supplies the proposal of the NIL parameters to the defect occurrence simulation device 300. For the layout data selected in S54, the defect occurrence simulation device 300 calculates a distribution of defects that will occur in pattern formation on the substrate SB by NIL according to the layout data while changing the NIL parameters among a plurality of NIL parameters according to the proposal from the evaluation system 1. The defect occurrence simulation device 300 supplies the calculation result to the evaluation system 1 as simulation data 84.

[0182] The evaluation system 1 selects, as an appropriate NIL parameter, an NIL parameter that satisfies a second criterion from among the multiple NIL parameters, based on the simulation data 84. The second criterion may be, for example, that the number of defects is minimal.

[0183] Alternatively, the evaluation system 1 may execute an evaluation method similar to the evaluation method shown in FIG. 9 using simulation data 84 and two-dimensional data 85 for each of the multiple layout data. In this case, the two-dimensional data 85 corresponds to two-dimensional data TD1 corresponding to the target data, and the evaluation method omits acquisition of the target data (S21). In other respects, the evaluation system 1 is similar to the evaluation method of FIG. 9 executed in S16. The evaluation system 1 may select layout data that satisfies a fourth criterion from the multiple layout data as appropriate layout data. The fourth criterion may be, for example, the highest degree of match.

[0184] The evaluation system 1 supplies the optimized NIL parameters to the defect occurrence simulation device 300 .

[0185] Meanwhile, in S62, an original OG for a second device is manufactured (S62). According to the pattern optimized in S54, information on the pattern of the original OG is generated. According to the information on the pattern of the original OG, the original OG is manufactured.

[0186] When the processes of S61 and S62 are both completed, the evaluation system 1 performs final adjustments by NIL experiments using the actual master OG (S63). The evaluation system 1 starts the evaluation program 81 and executes the evaluation method. In S63, the evaluation system 1 operates as shown in Fig. 8 and executes the evaluation method shown in Fig. 9. Details of the evaluation method are the same as those of S16, except that the first device is replaced with the second device, and therefore will not be described.

[0187] Processing including NIL processing is performed with the pattern and processing conditions optimized in S51 to S63, and a second device is manufactured (S64).

[0188] For example, a photosensitive agent is applied onto the substrate SB. The original OG is imprinted onto the substrate SB under conditions that are optimized in S61 and finally adjusted in S63. The photosensitive agent is exposed and hardened, the original OG is released from the substrate SB, and the substrate SB is processed using the pattern of the photosensitive agent as a mask, forming a pattern on the substrate SB. In this way, a second device is manufactured.

[0189] As described above, in this embodiment, the evaluation system 1 converts the two-dimensional defect distribution by actual measurement and the two-dimensional defect distribution by simulation into probability density distributions, and numerically evaluates the similarity between the two using a comparison function. The evaluation system 1 calculates a degree of match indicating the similarity between the two. This makes it possible to quantitatively evaluate the two-dimensional defect distribution in the substrate pattern by NIL, and makes it possible to verify the accuracy of the simulation and adjust the NIL parameters. As a result, the accuracy of the simulation can be easily improved.

[0190] Although some embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, and are included in the scope of the invention and its equivalents described in the claims. [Explanation of symbols]

[0191] 1 Evaluation system, 2 Generation unit, 3 Conversion unit, 4 Filter unit, 5 Restoration unit, 6 Threshold processing unit, 7 Calculation unit, 8 Memory unit, 9 Display control unit.

Claims

1. applying a first conversion process to first two-dimensional data indicating a defect distribution of a first substrate on which a pattern has been formed by imprinting an original onto a photosensitive agent on the first substrate and second two-dimensional data indicating a predicted defect distribution of a second substrate, respectively, to generate a first spectrum and a second spectrum; filtering the generated first spectrum and the generated second spectrum, respectively; performing a second conversion process that is the inverse of the first conversion process on the processed first spectrum and the processed second spectrum, respectively, to restore the first two-dimensional data and the second two-dimensional data; performing a threshold process on the restored first two-dimensional data and the restored second two-dimensional data; calculating a degree of match by applying a comparison function to the thresholded first two-dimensional data and the thresholded second two-dimensional data; Evaluation methods including:

2. The step of performing the threshold processing includes: replacing a level of data that is below a threshold level among a plurality of data included in the restored first two-dimensional data with the threshold level; replacing a level of data that is below the threshold level among a plurality of data included in the restored second two-dimensional data with the threshold level; Includes The evaluation method according to claim 1 .

3. The threshold level is predetermined based on the maximum level of data when there is one defect. The evaluation method according to claim 2.

4. The threshold level is predetermined by multiplying the maximum level of data for a single defect by a coefficient greater than 0 and less than 1. The evaluation method according to claim 3.

5. The degree of agreement calculated by the comparison function is a value between 0 and 1, and the closer to 1 the distribution shapes are, the more similar they are. The evaluation method according to claim 1 .

6. The application of the comparison function calculating a first degree of agreement by applying a first comparison function to the thresholded first two-dimensional data and the thresholded second two-dimensional data; calculating a second degree of agreement by applying a second comparison function to the first two-dimensional data after the threshold processing and the second two-dimensional data after the threshold processing in response to the first degree of agreement satisfying a predetermined condition; Includes The evaluation method according to claim 1 .

7. The application of the comparison function and calculating a score according to the first degree of match and the second degree of match. The evaluation method according to claim 6.

8. applying the evaluation method according to any one of claims 1 to 7, calculating a degree of agreement between first two-dimensional data indicating a defect distribution of a first substrate on which a pattern has been formed by imprinting an original onto a photosensitive agent on the first substrate and second two-dimensional data indicating a defect distribution of a second substrate predicted using a parameter indicating a condition related to the imprinting of the original while changing the parameter, and generating correlation information indicating a correlation between a direction of change in the parameter and a direction of change in the degree of agreement; predicting a defect distribution of a third substrate while changing the parameters by using the correlation information, and optimizing a pattern to be formed on the third substrate; predicting a defect distribution of a third substrate while changing the parameters using the correlation information, and optimizing conditions for imprinting the original; imprinting the master onto the third substrate using the optimized pattern and the optimized conditions to form a semiconductor device; A method for manufacturing a semiconductor device comprising the steps of:

9. a conversion unit that applies a first conversion process to first two-dimensional data indicating a defect distribution of a first substrate on which a pattern has been formed by imprinting an original onto a photosensitive agent on the first substrate and second two-dimensional data indicating a predicted defect distribution of a second substrate, to generate a first spectrum and a second spectrum; a filter unit that performs a filtering process on each of the generated first spectrum and the generated second spectrum; a restoration unit that performs a second conversion process that is the inverse of the first conversion process on each of the processed first spectrum and the processed second spectrum to restore the first two-dimensional data and the second two-dimensional data; a processing unit that performs threshold processing on the restored first two-dimensional data and the restored second two-dimensional data; a calculation unit that calculates a degree of match by applying a comparison function to the first two-dimensional data after the threshold processing and the second two-dimensional data after the threshold processing; A rating system with.

Images