Wavelet Systems and Methods for Improving Misalignment and Asymmetry in Semiconductor Devices

Abstract

The present invention discloses a wavelet analysis system and method for manufacturing a semiconductor device wafer. The system includes: a misalignment metrology tool operable to measure at least one measurement site on a wafer to generate an output signal; and a wavelet-based analysis engine operable to generate at least one wavelet-transformed signal by applying at least one wavelet transform to the output signal and to generate a quality metric by analyzing the wavelet-transformed signal.

Description

[0001] Cross - Reference to Related Applications

[0002] Reference is hereby made to U.S. Provisional Patent Application No. 63 / 043,828, filed on June 25, 2020, and titled "WAVELET BASED OVERLAY (OVL) CALCULATIONS AND ASYMMETRY EXTRACTION", the disclosure of which is hereby incorporated by reference herein and for which priority is hereby claimed.

[0003] Reference is also made to the following patents and patent applications of the applicant related to the subject matter of the present application, the disclosures of which are hereby incorporated by reference herein:

[0004] U.S. Patent No. 7,656,512, titled "METHOD FOR DETERMINING LITHOGRAPHIC FOCUS AND EXPOSURE";

[0005] U.S. Patent No. 7,804,994, titled "OVERLAY METROLOGY AND CONTROL METHOD";

[0006] U.S. Patent No. 9,490,182, titled "MEASUREMENT OF MULTIPLE PATTERNING PARAMETERS";

[0007] U.S. Patent No. 9,927,718, titled "MULTI-LAYER OVERLAY METROLOGY TARGET AND COMPLIMENTARY OVERLAY METROLOGY MEASUREMENT SYSTEMS";

[0008] U.S. Patent No. 10,415,963, titled "ESTIMATING AND ELIMINATING INTER-CELL PROCESS VARIATION INACCURACY";

[0009] U.S. Patent No. 10,527,951, entitled "Compound Imaging Metrology Targets";

[0010] PCT Application No. PCT / US2019 / 035282, filed on June 4, 2019, and entitled "Misregistration Measurements Using Combined Optical and Electron Beam Technology"; and

[0011] PCT Patent Application No. PCT / US2019 / 051209, filed on September 16, 2019, and entitled "Periodic Semiconductor Device Misregistration Metrology System and Method". TECHNICAL FIELD

[0012] The present invention generally relates to the measurement of misregistration in semiconductor device manufacturing. BACKGROUND ART

[0013] Various methods and systems are known for the measurement of misregistration in semiconductor device manufacturing. SUMMARY OF THE INVENTION

[0014] The present invention seeks to provide improved methods and systems for the measurement of misregistration in semiconductor device manufacturing.

[0015] Accordingly, in a preferred embodiment of the present invention, there is provided a wavelet analysis system for manufacturing a semiconductor device wafer, the system comprising: a misregistration metrology tool operable to measure at least one measurement site on the wafer to generate an output signal; and a wavelet-based analysis engine operable to generate at least one wavelet-transformed signal by applying at least one wavelet transform to the output signal, and to generate a quality metric by analyzing the wavelet-transformed signal.

[0016] According to a preferred embodiment of the present invention, the misregistration metrology tool is an electron beam misregistration metrology tool. Preferably, the analysis includes correlating a particular portion of the wavelet-transformed signal with a particular position within the measurement site.

[0017] According to a preferred embodiment of the present invention, the quality metric includes an indication of at least one individual structure formed within the measurement site, at least one group of structures formed within the measurement site, and the asymmetry of at least one of the measurement sites.

[0018] According to another preferred embodiment of the present invention, the quality metric includes an indication of misalignment between a first layer and a second layer formed on the wafer.

[0019] Preferably, the quality metric is operable for the generation of at least one adjusted manufacturing parameter. Preferably, the at least one adjusted manufacturing parameter is used in manufacturing the semiconductor device wafer.

[0020] According to another preferred embodiment of the present invention, there is also provided a wavelet analysis method for manufacturing a semiconductor device wafer, the method comprising: providing a first wafer; forming at least a first layer on the wafer using a first set of manufacturing parameters; forming at least a second layer on the wafer using a second set of manufacturing parameters; subsequently generating an output signal by measuring a measurement site on the first wafer using a misalignment metrology tool; generating at least one wavelet-transformed signal by applying at least one wavelet transform to the output signal; generating a quality metric by analyzing the wavelet-transformed signal; generating at least one adjusted set of manufacturing parameters by at least partially adjusting at least one manufacturing parameter based on the quality metric, the at least one manufacturing parameter being selected from at least one of the first set of manufacturing parameters and the second set of manufacturing parameters; and subsequently forming at least one layer on at least one of the first wafer and the second wafer using the adjusted set of manufacturing parameters.

[0021] According to a preferred embodiment of the present invention, the wavelet transform is a continuous wavelet transform. Preferably, the analysis includes correlating a specific portion of the wavelet-transformed signal with a specific location within the measurement site.

[0022] According to a preferred embodiment of the present invention, the quality metric includes an indication of misalignment between the first layer and the second layer.

[0023] According to a preferred embodiment of the present invention, the generating of the wavelet-transformed signal further includes selecting a frequency range, the wavelet-transformed signal varying according to the frequency range, and assigning zero values to the coefficients of the wavelet transform corresponding to frequency values outside the frequency range.

[0024] Alternatively, according to a preferred embodiment of the present invention, the generating of the wavelet-transformed signal further includes selecting a frequency range, the wavelet-transformed signal varying according to the frequency range, and multiplying the coefficients of the wavelet transform corresponding to frequency values outside the frequency range by a weighting factor.

[0025] According to a preferred embodiment of the present invention, the frequency range includes a frequency corresponding to at least one dimension of at least one structure formed in at least one of the first layer and the second layer.

[0026] Alternatively, according to a preferred embodiment of the present invention, the frequency range does not include a frequency corresponding to at least one dimension of at least one structure formed on the wafer. According to a preferred embodiment of the present invention, the at least one structure is formed together with at least one of the first layer and the second layer. Alternatively, according to a preferred embodiment of the present invention, the at least one structure is formed together with a structure layer, which is a layer other than the first layer and the second layer.

[0027] According to a preferred embodiment of the present invention, the quality metric includes an indication of the asymmetry of at least one individual structure formed within the measurement site, at least one group of structures formed within the measurement site, and at least one of the measurement sites.

[0028] According to a preferred embodiment of the present invention, the analyzing the wavelet-transformed signal includes identifying a general symmetry boundary within the wavelet-transformed signal and identifying at least one pair of asymmetric portions of the wavelet-transformed signal, each of the at least one pair of asymmetric portions including a pair of portions of the wavelet-transformed signal that are asymmetric with respect to the general symmetry boundary.

[0029] According to a preferred embodiment of the present invention, the analyzing the wavelet-transformed signal includes analyzing a signal having the same unit as the unit of the wavelet-transformed signal. Alternatively, according to a preferred embodiment of the present invention, the analyzing the wavelet-transformed signal further includes analyzing a signal having the same unit as the unit of the output signal.

[0030] According to a preferred embodiment of the present invention, the method further includes removing at least one of the first layer and the second layer from the wafer, and the at least one layer replaces at least one of the first layer and the second layer. Alternatively, according to a preferred embodiment of the present invention, the at least one layer does not replace any of the first layer and the second layer. Description of the Drawings

[0031] The present invention will be more fully understood and appreciated from the following detailed description taken in conjunction with the accompanying drawings, in which:

[0032] Figure 1 is a simplified schematic diagram of a wavelet analysis system for manufacturing semiconductor devices on a wafer;

[0033] Figure 2A is applicable to Figure 1Simplified description of a conventional measurement site of a wavelet analysis system including a conventional target;

[0034] Figure 2B is a simplified description of a conventional output signal generated from a part of the target shown in Figure 2A and is an amplification corresponding to the enlarged circle D in Figure 2A ;

[0035] Figure 2C is from Figure 1 the wavelet analysis system and is a simplified description of a wavelet-transformed signal generated from a part of the target shown in Figure 2A and is an amplification corresponding to the enlarged circle E in Figure 2A ;

[0036] Figure 3 illustrates Figures 1 to 2C a simplified flowchart of the wavelet analysis method used by the wavelet analysis system;

[0037] Figure 4A and 4B illustrates Figure 3 a simplified flowchart of an embodiment of a part of the wavelet analysis method; and

[0038] Figure 5 illustrates Figure 3 a simplified flowchart of an additional embodiment of a part of the wavelet analysis method. Detailed Description of the Invention

[0039] Now refer to Figure 1 , which is a simplified schematic diagram of a wavelet analysis system 100 for manufacturing semiconductor devices on a wafer 102 or a similar wafer, and refer to Figure 2A , which is a simplified description of a measurement site in the wavelet analysis system applicable to Figure 1 , refer to Figure 2B , which is a simplified description of a conventional output signal generated from a part of the measurement site shown in Figure 2A , and refer to Figure 2C , which is a simplified description of a wavelet-transformed signal generated from a part of the measurement site shown in Figure 2A .

[0040] In particular, as shown in Figure 1 and 2AAs seen, the wavelet analysis system 100 is preferably used in combination with a manufacturing tool 110 having adjustable manufacturing parameters. The manufacturing tool 110 preferably uses at least a first set of manufacturing parameters to at least partially form a first layer 112 on a wafer 102. Additionally, the wavelet analysis system 100 is preferably used in combination with a manufacturing tool 114 having adjustable manufacturing parameters. The manufacturing tool 114 preferably uses at least a second set of manufacturing parameters to at least partially form a second layer 116 on the wafer 102. It should be understood that the first layer 112 and the second layer 116 may be adjacent layers, but need not be so, and may be separated by a height in the range from 100 nm to over 10 μm.

[0041] In a preferred embodiment of the present invention, examples of the manufacturing parameters in the first set and the second set of manufacturing parameters specifically include the manufacturing parameters identified by a modeling kit (such as the K-T Analyzer platform) available from KLA Corporation, Milpitas, California, USA.

[0042] The manufacturing tools 110 and 114 can be embodied as any suitable manufacturing tools, particularly including lithography scanners, etching tools, and polishing tools. In a preferred embodiment of the present invention, the manufacturing tools 110 and 114 are lithography scanners and are embodied as one or both of an immersion scanner and an extreme ultraviolet (EUV) scanner. A typical immersion scanner that can be used as the manufacturing tool 110 or 114 is the NSR-S635E available from Nikon Corporation, Tokyo, Japan.

[0043] It should be understood that in one embodiment of the present invention, the manufacturing tool 110 and the manufacturing tool 114 are a single tool. In another embodiment of the present invention, the manufacturing tool 110 and the manufacturing tool 114 are separate tools. In an embodiment where the manufacturing tools 110 and 114 are separate tools, the manufacturing tools 110 and 114 can be of the same type of tool (e.g., two lithography tools), or different types of tools (e.g., a lithography tool and a nanoimprint lithography tool).

[0044] Generally, in addition to the first layer 112 and the second layer 116, additional layers are also formed on the wafer 102. Preferably, the structure is formed with additional layers. In some embodiments of the present invention, the wavelet analysis system 100 is operable to measure the misalignment between at least one of the additional layers and at least one other layer formed on the wafer 102, where the at least one other layer is embodied as the first layer 112, the second layer 116, or another one of the additional layers formed on the wafer 102.

[0045] The wavelet analysis system 100 preferably includes a misalignment metrology tool 120 that measures at least one of a plurality of measurement sites 122 on the wafer 102. The misalignment metrology tool 120 can be any suitable misalignment metrology tool, particularly including an electron beam misalignment metrology tool. A typical electron beam misalignment metrology tool that can be used as the misalignment metrology tool 120 is the eDR7380 available from KLA-Tencor Corporation of Milpitas, California, USA. TM .

[0046] Preferably, any material pair between the misalignment metrology tool 120 and each of the first layer 112 and the second layer 116 is at least partially transparent to the electromagnetic radiation used by the misalignment metrology tool 120 to measure the measurement site or sites 122 on the wafer 102.

[0047] In one embodiment of the present invention, each measurement site 122 on the wafer 102 includes a structure formed therein that is intended to be the same as the structures formed in the other ones of the measurement sites 122 on the wafer 102. In another embodiment of the present invention, each measurement site 122 on the wafer 102 includes a structure formed therein that is intended to be different from the structures formed in the other ones of the measurement sites 122 on the wafer 102. In yet another additional embodiment of the present invention, each of the measurement sites 122 in at least a first group of measurement sites on the wafer 102 includes a structure formed therein that is intended to be the same as the structures formed in the other ones of the measurement sites 122 on the wafer 102, while each of the measurement sites 122 in at least a second group of measurement sites on the wafer 102 includes a structure formed therein that is intended to be different from the structures in the other ones of the measurement sites 122 on the wafer 102.

[0048] Specifically as seen in the embodiment of one measurement site 122 shown in Figure 2A , the measurement site 122 preferably includes at least one target 124 formed therein. The target 124 is preferably suitable for being measured by the misalignment metrology tool 120. The target 124 generally includes a first structure 126 formed together with the first layer 112 and a second structure 128 formed together with the second layer 116.

[0049] In a preferred embodiment of the present invention, the target 124 is particularly suitable for being measured by a misalignment metrology tool 120 that is not aligned. For example, the target 124 can be embodied as a target particularly suitable for being measured by an electron beam misalignment metrology tool, and the misalignment metrology tool 120 can be embodied as an electron beam misalignment metrology tool. However, in additional embodiments of the present invention, the target 124 is particularly suitable for being measured by a misalignment metrology tool other than the misalignment metrology tool 120. For example, the target 124 can be embodied as a target particularly suitable for being measured by an imaging misalignment metrology tool, and the misalignment metrology tool 120 can be embodied as an electron beam misalignment metrology tool. Preferably, in this case, the misalignment metrology tool 120 can even generate a meaningful measurement output from this target 124.

[0050] For simplicity, the target 124 is shown in Figure 2A as an advanced imaging metrology (AIM) target. However, the target 124 can be embodied as any suitable target, particularly for example: a box-in-box target, such as a target similar to the target described in U.S. Patent No. 7,804,994; an AIM die (AIMid) target, such as a target similar to the target described in U.S. Patent No. 10,527,951; a micro blossom target, such as a target similar to the target described in C.P. Ausschnitt, J. Morningstar, W. Muth, J. Schneider, R.J. Yerdon, L.A. Binns, N.P. Smith's "Multilayer overlay Meterlogy", Proc. SPIE 6152, Metrology, Inspection, and Process Control for Microlithography XX, 615210 (March 24, 2006); a combined optical and electron beam target, such as a target similar to the target described in PCT Application No. PCT / US2019 / 035282; and a target that can be used to measure misalignment between three or more layers, such as a target similar to the target described in U.S. Patent No. 9,927,718. Additionally, the target 124 can be embodied as a complete or partial semiconductor device intended to be a functional semiconductor device, such as the device described in PCT Patent Application No. PCT / US2019 / 051209.

[0051] Each of the first structures 126 is generally intended to have the same minimum dimension J, but in some embodiments, different ones of the first structures 126 are intended to have a minimum dimension J that is intentionally different in value from the minimum dimension J of the other ones of the first structures 126. Figure 2A , the minimum dimension J is preferably the width of the structure 126 measured from a top plan perspective. Similarly, the second structures 128 are generally intended to each have the same minimum dimension K, but in some embodiments, different ones of the second structures 128 are intended to have a minimum dimension K that is intentionally different in value from the minimum dimension K of the other ones of the second structures 128. Figure 2A As seen in FIG, the minimum dimension K is preferably the width of the structure 128 measured from a top plan perspective.

[0052] In some embodiments of the present invention, as particularly seen in the enlarged circle A, each first structure 126 is formed from a first substructure 136, each first substructure 136 having a minimum dimension L. As seen in the enlarged circle A, the minimum dimension L is preferably the width of the substructure 136 measured from a top plan perspective. In other embodiments of the present invention, the first structure 126 is generally a unitary structure and does not include substructures.

[0053] Similarly, in some embodiments of the present invention, as particularly seen in the enlarged circle A, each second structure 128 is formed from a second substructure 138, each second substructure 138 having a minimum dimension M. As seen in the enlarged circle A, the minimum dimension M is preferably the width of the substructure 138 measured from a top plan perspective. In other embodiments of the present invention, the second structure 128 is generally a unitary structure and does not include substructures.

[0054] In general, structures 126 and 128 and substructures 136 and 138 are generally intended to be symmetrical; however, some or all of structures 126 and 128 and substructures 136 and 138 may exhibit unintentional asymmetry. Types of unintentional asymmetry of structures 126 and 128 and substructures 136 and 138 particularly include angular asymmetry within the structures.

[0055] exist Figure 2A In the embodiment illustrated in FIG, structures 126 and 128 and substructures 136 and 138 of target 124 are intended to be symmetrical structures, such as symmetrical structure 142. As particularly seen in magnified circles B and C, each symmetrical structure 142 preferably includes a pair of angles θ and a pair of angles Preferably, in each symmetrical structure 142, the angles θ are generally equal to each other, and the angles Usually equal to each other.

[0056] However, due to unintentional angular asymmetries, at least some of structures 126 and 128 of target 124 and sub-structures 136 and 138 are asymmetric structures, such as asymmetric structure 144. As particularly seen in enlarged circles B and C, each asymmetric structure 144 includes at least a pair of angles α and β, where angle α is generally not equal to angle β.

[0057] It should be understood that, in the embodiments shown in Figure 2A , particularly as seen in enlarged circles B and C, the cross-sections of structures 126 and 128 are intended to be isosceles trapezoids having two identical upper corners and two identical lower corners. Similarly, in the embodiments shown in Figure 2A , the cross-sections (not shown) of sub-structures 136 and 138 are intended to be isosceles trapezoids having two identical upper corners and two identical lower corners. Thus, the pairs of angles θ and and the pairs of angles α and β are respectively embodied as two lower corners or two upper corners within structure 142 or 144. However, in other embodiments of the present invention, other angles may be intended to be the same, and the pairs of angles θ and and the pairs of angles α and β are embodied as any angles intended to be the same, such as two side angles.

[0058] Similarly, even when target 124 is intended to be symmetric, target 124 may exhibit unintentional symmetries. The types of unintentional symmetries of target 124 particularly include pitch walk, as described in U.S. Patent No. 9,490,182; height, width, or other variations between structures in target 124, as described in U.S. Patent No. 7,656,512; and pad-to-pad variations, as described in U.S. Patent No. 10,415,963. Additionally, measurement site 122 may contain one or more asymmetric structures, such as a portion of groove 146 or foreign material 148.

[0059] When measuring measurement site 122, as seen in enlarged circle D, misaligned metrology tool 120 generates output signal 152. It should be understood that, for ease of understanding and for simplicity, output signal 152 is shown as a two-dimensional signal; however, output signal 152 may have any suitable number of dimensions.

[0060] Further, it should be understood that Figure 2A the embodiment of output signal 152 shown in contains data related to only a single asymmetric structure 144. However, in other typical embodiments of the present invention, output signal 152 may contain data related to the whole of measurement site 122 or any part thereof.

[0061] In one embodiment of the present invention, the output signal 152 is in the form of raw data, which is three-dimensional data in a Euclidean x-y-z coordinate system. The three-dimensional data is preferably characterized by an easily understandable pair of correspondence characteristics between each position in the x-y plane of the x-y-z coordinate system and the positions within the measurement site 122.

[0062] In Figure 2A the embodiment of the present invention shown in the enlarged circle D, the output signal 152 is in the form of adjusted data. This adjusted data can in particular be a kernel, which is preferably generated from three-dimensional data in a Euclidean x-y-z coordinate system. For each point on the x-axis of the x-y-z coordinate system, the values along the y-axis at that x value are combined to produce a single y value for that x value. The y-axis values can be combined in particular by averaging the y-axis values. The data in the kernel is preferably characterized by an easily understandable correspondence between each position on the x-axis and a plurality of positions within the measurement site 122.

[0063] In a preferred embodiment of the present invention, the output signal 152 generated by the misalignment metrology tool 120 is communicated to a wavelet-based analysis engine 160, which applies at least one wavelet transform to the output signal 152, thereby generating at least one wavelet-transformed signal 162, an example of which is shown in the enlargement E. It should be understood that, for ease of understanding, the wavelet-transformed signal 162 is shown as a three-dimensional signal; however, the wavelet-transformed signal 162 can have any suitable number of dimensions.

[0064] Further, it should be understood that Figure 2A the embodiment of the wavelet-transformed signal 162 shown in

[0065] contains data related only to a single asymmetric structure 144. However, in other typical embodiments of the present invention, the wavelet-transformed signal 162 can contain data related to the whole or any part of the measurement site 122.

[0066] As is known in the art, a wavelet transform is a convolution of a signal with a series of wavelet functions. Since each wavelet function in the series of wavelet functions returns a non-zero value only within a finite interval, a particular part of the wavelet-transformed signal 162 can be easily associated with the corresponding part of the output signal 152. Thus, a particular part of the wavelet-transformed signal 162 can be easily associated with a particular position within the measurement site 122.

[0067] In one embodiment of the present invention, for example, particularly in the embodiments seen in Figure 2B and 2C the relationship between a particular portion of the wavelet-transformed signal 162 and a particular location within the measurement site 122 can be different from the relationship between a particular portion of the output signal 152 and a particular location within the measurement site 122. In another embodiment of the present invention, the relationship between a particular portion of the wavelet-transformed signal 162 and a particular location within the measurement site 122 can be the same as the relationship between a particular portion of the output signal 152 and a particular location within the measurement site 122.

[0068] The wavelet-based analysis engine 160 preferably analyzes the wavelet-transformed signal 162 to generate a quality metric. The wavelet-based analysis engine 160 preferably communicates the quality metric to at least one of the manufacturing tools 110 and 114 for adjusting at least one of a first set of manufacturing parameters used in forming the first layer 112 and a second set of manufacturing parameters used in forming the second layer 116.

[0069] A particular feature of the present invention is that since a particular portion of the wavelet-transformed signal 162 can be easily associated with a corresponding location within the measurement site 122, the quality metric generated by the wavelet-based analysis engine 160 can be easily associated with a corresponding location within the measurement site 122. Thus, the wavelet analysis system 100 identifies the contribution of a particular one of the particular locations or structures 126 and 128 and sub-structures 136 and 138 within the measurement site 122 to the quality metric and accordingly adjusts at least one parameter of at least one of the first set of manufacturing parameters and the second set of manufacturing parameters.

[0070] In one embodiment of the present invention, the quality metric generated by the wavelet-based analysis engine 160 is a misalignment value, which preferably indicates a misalignment between the first layer 112 and the second layer 116. In another embodiment of the present invention, the quality metric generated by the wavelet-based analysis engine 160 is an indication of the asymmetry of at least one of the structure 126, the structure 128, the sub-structure 136, the sub-structure 138, the target 124, and the misalignment site 122. The asymmetry indicated by this indication of asymmetry can particularly include: angular asymmetry within the structure; pitch walk; height, width, or other variations between structures within the target 124; pad-to-pad variation; and asymmetric structures within the measurement site 122, such as one or more trenches 146 or foreign material 148.

[0071] In a preferred embodiment of the present invention, the quality metric generated by the wavelet-based analysis engine 160 is used to adjust at least one of a first set of manufacturing parameters used by the manufacturing tool 110 and a second set of manufacturing parameters used by the manufacturing tool 114 in the production of semiconductor devices formed on the wafer 102 or on different wafers.

[0072] In an embodiment where the quality metric is a misalignment value, the adjustment of at least one of the manufacturing parameters preferably results in improved alignment between the layers 112 and 116 fabricated using the adjusted manufacturing parameters as compared to the layers 112 and 116 fabricated using the unadjusted manufacturing parameters.

[0073] In an embodiment where the quality metric is an indication of asymmetry, the adjustment of at least one manufacturing parameter preferably results in better symmetry of at least one of the structures 126, 128, sub-structures 136, 138, measurement sites 122, and targets 124 as compared to the structures 126, 128, sub-structures 136, 138, measurement sites 122, and targets 124 fabricated using the unadjusted manufacturing parameters.

[0074] Additionally or alternatively, the indication of asymmetry can be used to adjust the misalignment value generated by the wavelet analysis system 100. For example, an indication of asymmetry at a particular measurement site 122 can cause the wavelet analysis system 100 to measure the wafer 102 at different measurement sites 122 using the misalignment metrology tool 120, thereby generating data that can be used to generate the misalignment value. Similarly, the indication of asymmetry can be used to select a particular portion of the output signal 152 generated by the misalignment metrology tool 120 for use in generating the misalignment value. Additionally, the indication of asymmetry can be used to adjust the output signal 152 from the misalignment metrology tool 120, and the adjusted output signal can be used to generate the misalignment value.

[0075] Additionally or alternatively, the indication of asymmetry can be used to adjust the misalignment value generated by a suitable misalignment metrology tool that is not part of the wavelet analysis system 100. Examples of suitable misalignment metrology tools specifically include eDR7380 TM 、Archer TM 750 or ATL100 TM, all of which are available from KLA-Tencor Corporation, Milpitas, California, USA. For example, an indication of the asymmetry at a particular measurement site 122 can facilitate measuring the wafer 102 at different measurement sites 122 with a suitable misalignment metrology tool, thereby generating data that can be used to generate a misalignment value. Similarly, an indication of the asymmetry can be used to select a particular portion of the output signal 152 generated by a suitable misalignment metrology tool for generating a misalignment value. Additionally, an indication of the asymmetry can be used to adjust the output signal 152 from a suitable misalignment metrology tool, and the adjusted output signal can be used to generate a misalignment value.

[0076] In a preferred embodiment of the present invention, the wavelet-transformed signal 162 generated by the wavelet analysis system 100 and its corresponding misalignment value are relatively noise-free compared to the output signal and misalignment value generated by a conventional tool.

[0077] It should be understood that, in one embodiment of the present invention, the wavelet analysis system 100 is used to measure misalignments at a plurality of misalignment sites 122 that are intended to be the same as each other on the same wafer 102 or different wafers 102.

[0078] Furthermore, the misalignment values generated by the wavelet analysis system 100 for different ones of a plurality of misalignment sites 122 that are intended to be the same as each other are generally more similar to each other than the misalignment values generated by a conventional system for different ones of a plurality of misalignment sites 122 that are intended to be the same as each other. The greater similarity between the misalignment values generated by the wavelet analysis system 100 relative to the misalignment values generated by a conventional misalignment metrology system indicates a relatively greater accuracy of the misalignment values generated by the wavelet analysis system 100 when compared to a conventional system.

[0079] As specifically referenced below Figure 2B and 2C described in more detail, the wavelet-transformed signal 162 generated by the wavelet analysis system 100 and the corresponding asymmetry indication have a relatively greater sensitivity to asymmetry compared to the typical output signal and corresponding asymmetry indication generated by a conventional tool.

[0080] Now refer to Figure 2B and 2C , which are magnifications of the enlarged circles D and E in Figure 2A respectively, and show the output signal 152 and the wavelet-transformed signal 162 respectively.

[0081] In particular, as seen in Figure 2B , the output signal 152 is generally symmetric about a general axis of symmetry 172. The output signal further includes an asymmetric signal portion 174 that is asymmetric about the general axis of symmetry 172; however, the asymmetry of the portion 174 is not obvious.

[0082] In contrast, in particular, asFigure 2C As seen in Figure 2C , the wavelet-transformed signal 162 includes a generally symmetric plane 182 and an asymmetric signal portion 184. As is readily appreciated from a visual inspection of the wavelet-transformed signal 162, the asymmetric signal portion 184 is asymmetric with respect to the generally symmetric plane 182.

[0083] Accordingly, the wavelet-transformed signal 162 and the corresponding asymmetry indication generated by the wavelet analysis system 100 have greater sensitivity to asymmetry within the measurement site 122 than the conventional output signal 152 and the corresponding asymmetry indication generated by conventional tools.

[0084] Now additionally refer to Figure 3 , which is a simplified flowchart illustrating the wavelet analysis method 200 used by the wavelet analysis system 100.

[0085] As Figure 3 seen in Figure 3 , in a first step 202 of the wavelet analysis method 200, the fabrication tool 110 preferably uses at least a first set of fabrication parameters to at least partially form a first layer 112 on the wafer 102. In the next step 204, the fabrication tool 114 preferably uses at least a second set of fabrication parameters to at least partially form a second layer 116 on the wafer 102.

[0086] As mentioned above, the fabrication tool 110 and the fabrication tool 114 may be embodied as a single tool or separate tools. Also as mentioned above, the first layer 112 and the second layer 116 may be adjacent layers, but need not be so, and may be separated by a height in the range from 100 nm to more than 10 μm.

[0087] In the next step 206, the misalignment metrology tool 120 preferably measures at least one measurement site 122 on the wafer 102, thereby generating an output signal 152.

[0088] In the next step 208, the output signal 152 generated in step 206 is communicated to the wavelet-based analysis engine 160, which preferably applies at least one wavelet transform to the output signal 152, thereby generating at least one wavelet-transformed signal 162. In a preferred embodiment of the present invention, the wavelet transform performed in step 208 is a continuous wavelet transform. In another embodiment of the present invention, the wavelet transform performed in step 208 is a discrete wavelet transform.

[0089] In the next step 210, the wavelet-based analysis engine 160 preferably analyzes the wavelet-transformed signal 162 generated in step 208, thereby generating a quality metric. In a preferred embodiment of the present invention, the quality metric is particularly an indication of misalignment between the first layer 112 and the second layer 116, as referred to below in Figure 4A and 4BMore specifically described, or an indication of the asymmetry present within measurement site 122, as described in more detail below with reference to Figure 5 More specifically described.

[0090] In the next step 212, the wavelet-based analysis engine 160 preferably generates at least a set of adjusted manufacturing parameters by adjusting at least one manufacturing parameter based at least in part on the quality metric generated in step 210. Preferably, at least one manufacturing parameter is selected from at least one of the first set of manufacturing parameters used by the manufacturing tool 110 in step 202 and the second set of manufacturing parameters used by the manufacturing tool 114 in step 204.

[0091] In the next step 214, the at least a set of adjusted manufacturing parameters generated in step 212 are preferably used in the production of semiconductor devices formed on the wafer 102 or a different wafer. Preferably, the adjusted parameters are used to form at least one layer on at least one of the wafer 102 and the additional wafers. In one embodiment of the present invention, at least one of the first layer 112 and the second layer 116 is removed from the wafer 102, and at least one layer formed using the adjusted parameters replaces at least one of the first layer 112 and the second layer 116. In another embodiment of the present invention, neither the first layer 112 nor the second layer 116 is removed from the wafer 102, and at least one layer formed using the adjusted parameters is formed above or below the first layer 112 and the second layer 116.

[0092] In an embodiment where the quality metric is a misalignment value, compared with the layers 112 and 116 manufactured using unadjusted manufacturing parameters, the adjustment of at least one of the manufacturing parameters in step 212 preferably results in improved alignment between the layers 112 and 116 manufactured using the adjusted parameters.

[0093] In an embodiment where the quality metric is an indication of asymmetry, compared with the structures 126, structure 128, sub-structures 136, sub-structures 138, measurement site 122 and target 124 manufactured using unadjusted manufacturing parameters, the adjustment of at least one manufacturing parameter in step 212 preferably results in better symmetry of at least one of the structures 126, structure 128, sub-structures 136, sub-structures 138, measurement site 122 and target 124 manufactured using the adjusted manufacturing parameters. Additionally or alternatively, as described in more detail below with specific reference to Figure 5 described, the indication of asymmetry can be used to adjust the misalignment value.

[0094] Now additionally referring to Figure 4A , which is a simplified flowchart illustrating an embodiment of step 208 of the wavelet analysis method 200, where the quality metric generated in step 210 is an indication of the misalignment between the first layer 112 and the second layer 116.

[0095] AsFigure 4A As seen in Figure 4A , in the first sub-step 402, during the wavelet transform that generates the wavelet-transformed signal 162 in step 208, a decision is made as to whether to apply one or more wavelet-based filters. This wavelet-based filter is preferably embodied as a frequency range, and the wavelet-transformed signal 162 varies according to the frequency range. As is known in the art, the wavelet-transformed signal 162 contains inputs from a plurality of coefficients of the wavelet transform used in step 208.

[0096] In a preferred embodiment of the present invention, when applying a wavelet-based filter or a plurality of wavelet-based filters, zero values are assigned to all coefficients of the wavelet transform corresponding to frequency values outside the frequency range of the wavelet-based filter. Thus, when applying one or more wavelet-based filters, all non-zero values of the wavelet-transformed signal 162 generated in step 208 are only associated with frequencies within the frequency range of the wavelet-based filter or the plurality of wavelet-based filters.

[0097] In an additional embodiment of the present invention, when applying a wavelet-based filter or a plurality of wavelet-based filters, all coefficients of the wavelet transform corresponding to frequency values outside the frequency range of the wavelet-based filter are multiplied by a weighting factor. Preferably, the weighting factor is greater than zero and less than one. Thus, when applying one or more wavelet-based filters, the contribution from frequencies not included in the frequency range of the wavelet-based filter or the plurality of wavelet-based filters to the wavelet-transformed signal 162 generated in step 208 is suppressed.

[0098] If no wavelet-based filter is applied, then the method may proceed to the next sub-step 404, and an unfiltered wavelet-transformed signal is generated. Then, the method proceeds to Figure 3 step 210. If one or more wavelet-based filters are to be applied, then the method alternatively proceeds to the next sub-step 406. In sub-step 406, a decision is made as to whether one or more wavelet-based filters contain a structure of interest.

[0099] If the wavelet-based filter or the plurality of wavelet-based filters will contain a structure of interest, then the method proceeds to the next sub-step 408, where one or more wavelet-based filters containing the structure of interest are selected and applied. In a preferred embodiment of the present invention, the wavelet-based filter containing the structure of interest is embodied as a frequency range corresponding to at least one of the dimensions J, K, L, and M of the respective structures 126 and 128 and the sub-structures 136 and 138. In another preferred embodiment of the present invention, especially if the structure of interest is part of a periodic structure of a group, then the wavelet-based filter containing the structure of interest is embodied as a frequency range corresponding to the pitch of the periodic structure.

[0100] Preferably, the relationship between a given frequency within the frequency range and a pitch or dimension within the measurement site 122 (e.g., one of the dimensions J, K, L, and M of the respective structures 126 and 128 and the sub-structures 136 and 138) is an inverse relationship. It should be understood that there may be many suitable values for the scaling factor that relates a given frequency in the frequency range to the dimension within the measurement site 122. For example, suitable frequencies for the frequency range of the wavelet-based filter included in sub-step 408 may specifically be equal to 1 / J, 10 / J, 0.5 / K, 4 / K, 0.1 / M, 3.14159 / M, 6 / L, 0.7 / L, or 1 / L.

[0101] If the wavelet-based filter or filters do not include the structure of interest, the method proceeds to the next sub-step 410, where one or more wavelet-based filters that exclude specific structures are selected and applied. In a preferred embodiment of the present invention, the wavelet-based filter that excludes specific structures suppresses the contribution to the wavelet-transformed signal 162 from at least one of the structures 126 and 128 and the sub-structures 136 and 138. In this case, the wavelet-based filter is embodied as at least one frequency range that does not correspond to at least one of the dimensions J, K, L, and M of the respective structures 126 and 128 and the sub-structures 136 and 138. In another preferred embodiment of the present invention, particularly if the wavelet-based filter applied in sub-step 410 is intended to suppress the contribution to the wavelet-transformed signal 162 from at least one structure that is part of a group of periodic structures, the wavelet-based filter or filters that exclude specific structures are embodied as frequency ranges, each frequency corresponding to the pitch of the periodic structure to be excluded.

[0102] Additionally or alternatively, the wavelet-based filter that excludes specific structures suppresses the contribution to the wavelet-transformed signal 162 from one or more structures formed on the wafer 102 together with layers other than layer 112 or 116. In this case, the wavelet-based structure is embodied as at least one frequency range that does not correspond to at least one pitch or dimension of at least one structure formed on the wafer 102 together with layers other than layer 112 or 116.

[0103] Preferably, the relationship between a given frequency within the frequency range and a pitch or dimension within the measurement site 122 (such as one of the dimensions J, K, L, and M of the respective structures 126 and 128 and the sub-structures 136 and 138) is an inverse relationship. It should be understood that there may be many suitable values for the scaling factor that relates a given frequency in the frequency range to the dimension within the measurement site 122. For example, suitable frequencies for the frequency range of the wavelet-based filter included in sub-step 410 may specifically be equal to 1 / J, 10 / J, 0.5 / K, 4 / K, 0.1 / M, 3.14159 / M, 6 / L, 0.7 / L, or 1 / L.

[0104] Regardless of whether one or more filters including or excluding specific structures are applied in sub-steps 408 and 410 respectively, the method proceeds to the next sub-step 412, where a filtered wavelet-transformed signal is generated. It should be understood that the filtered wavelet-transformed signal generated in sub-step 412 emphasizes, in the case of sub-step 408, or removes, in the case of sub-step 410, the contribution of a specific structure of interest to the output signal 152 of step 206. After sub-step 412, the method proceeds Figure 3 to step 210.

[0105] Now also refer to Figure 4B , which is a simplified flowchart illustrating an embodiment of step 210 of the wavelet analysis method 200, where the quality metric generated in step 210 is an indication of misalignment between the first layer 112 and the second layer 116.

[0106] As Figure 4B seen, in a first sub-step 420, an evaluation is made to confirm whether the wavelet-transformed signal 162 generated in step 208 is a filtered wavelet-transformed signal (such as the wavelet-transformed signal generated in sub-step 412), or an unfiltered wavelet-transformed signal (such as the wavelet-transformed signal generated in sub-step 404).

[0107] If the wavelet-transformed signal 162 generated in step 208 is a filtered wavelet-transformed signal, the method proceeds to the next sub-step 422, where a decision is made whether to generate a filtered signal having the same units as the units of the output signal 152. If a decision is made to generate a filtered signal having the same units as the units of the output signal 152, the method proceeds to the next sub-step 424. In sub-step 424, an inverse transform is applied to the filtered wavelet-transformed signal generated in step 208 to generate a filtered signal that can be used to generate misalignment values. In a preferred embodiment of the present invention, the inverse transform applied in sub-step 424 is the mathematical inverse of the wavelet transform applied in step 208. In another embodiment of the present invention, the inverse transform applied in sub-step 424 is not the mathematical inverse of the wavelet transform applied in step 208.

[0108] If in sub-step 420, the wavelet-transformed signal 162 is determined to be an unfiltered wavelet-transformed signal, or if in sub-step 422, a decision is made not to generate a filtered signal having the same units as the units of the output signal 152, or after sub-step 424, the method proceeds to the next sub-step 426. In sub-step 426, a misalignment value indicating a misalignment between the first layer 112 and the second layer 116 is generated. After sub-step 426, the method proceeds Figure 3 to step 212. It should be understood that in one embodiment of the present invention, the misalignment value generated in sub-step 426 is generated by analyzing a signal having the same units as the units of the wavelet-transformed signal 162 (e.g., the wavelet-transformed signal 162 generated in step 208). In another embodiment of the present invention, the misalignment value generated in sub-step 426 is generated by analyzing a signal having the same units as the units of the output signal 152 (e.g., the filtered signal generated in sub-step 424).

[0109] Now referring additionally to Figure 5 , which is a simplified flowchart illustrating an embodiment of step 210 of the wavelet analysis method 200, where the quality metric generated in step 210 is an indication of the asymmetry present within the measurement site 122.

[0110] As Figure 5 can be seen, in a first sub-step 502, a frequency and position range within which the wavelet-transformed signal 162 generated in step 208 is examined is selected. For example, for a position range from approximately 5.5 μm to 6.7 μm and a normalized frequency range from 0.1 to 0.3 cycles / sample, the Figure 2C wavelet-transformed signal 162 is examined, even though the wavelet-transformed signal 162 may contain non-zero data at position and frequency values outside of the range.

[0111] In the next sub-step 504, identify general symmetry boundaries, such as Figure 2C the general symmetry plane 182 of Figure 2C and 5 Although the general symmetry plane is shown in

[0112] it should be understood that the identified general symmetry boundaries can have any suitable form and any suitable number of dimensions. Next, in subsequent sub-step 506, identify at least a pair of asymmetric portions of the wavelet-transformed signal, such as the asymmetric portion 184 of the wavelet-transformed signal 162. In subsequent sub-step 508, generate an indication of the asymmetry. It should be understood that in one embodiment of the present invention, the indication of the asymmetry generated in sub-step 508 is generated by analyzing a signal having the same cells as the cells of the wavelet-transformed signal 162 (such as the wavelet-transformed signal 162 generated in step 208). In another embodiment of the present invention, the indication of the asymmetry generated in sub-step 508 is generated by analyzing a signal having the same cells as the cells of the output signal 152, such as by performing an inverse transform on the wavelet-transformed signal 162 in a manner similar to the operation performed in sub-step 424.

[0113] In the next sub-step 510, make a decision as to whether the indication of the asymmetry generated in sub-step 508 is to be used in the adjustment of the misalignment value generated by the wavelet analysis system 100 or a different suitable misalignment measurement system, as described above with reference to Figures 1 to 2C described.

[0114] If the indication of the asymmetry generated in sub-step 508 is to be used in the adjustment of the misalignment value, then the method proceeds to sub-step 512 and adjusts the misalignment value. For example, the indication of the asymmetry generated in sub-step 508 can facilitate the measurement of the wafer 102 at different measurement sites 122, thereby generating data that can be used to generate an adjusted misalignment value. Similarly, the indication of the asymmetry generated in sub-step 508 can be used to select a specific portion of the output signal 152 generated in step 206 for use in generating an adjusted misalignment value. Additionally, the indication of the asymmetry generated in sub-step 508 can be used to adjust the output signal 152 generated in step 206, and the adjusted output signal can be used to generate an adjusted misalignment value.

[0115] After sub-step 512, or immediately after sub-step 510, if the indication of the asymmetry generated in sub-step 508 is not used in the adjustment of the misalignment value, then the method proceeds to Figure 3 step 212.

[0116] Those skilled in the art will appreciate that the present invention is not limited to what has been specifically shown and described above. The scope of the present invention includes both combinations and sub - combinations of the various features described above, as well as modifications thereof, all of which are not in the prior art.

Claims

1. A wavelet analysis system for manufacturing semiconductor device wafers, the system comprising: A misalignment metrology tool operable to measure at least one measurement site on a wafer, thereby generating an output signal; And A wavelet-based analysis engine operable to: Generate at least one wavelet-transformed signal by applying at least one wavelet transform to the output signal; and Generate a quality metric by analyzing the wavelet-transformed signal, wherein the analysis includes associating a particular portion of the wavelet-transformed signal with a particular location within the measurement site, and wherein the analysis further associates the contribution of one or more structures formed within the measurement site with the quality metric.

2. The wavelet analysis system according to claim 1, and wherein the misalignment metrology tool is an electron beam misalignment metrology tool.

3. The wavelet analysis system according to claim 1 or claim 2, and wherein the quality metric includes an indication of the asymmetry of at least one of the following: At least one individual structure among the structures formed within the measurement site; At least one group among the structures formed within the measurement site; and The measurement site.

4. The wavelet analysis system according to claim 1 or claim 2, and wherein the quality metric includes an indication of misalignment between a first layer and a second layer formed on the wafer.

5. The wavelet analysis system according to claim 1 or claim 2, and wherein the quality metric is operable for the generation of at least one adjusted manufacturing parameter.

6. The wavelet analysis system according to claim 5, and wherein the at least one adjusted manufacturing parameter is for the manufacturing of the semiconductor device wafer.

7. A wavelet analysis method for manufacturing semiconductor device wafers, the method comprising: Providing a first wafer; Forming at least a first layer on the wafer using a first set of manufacturing parameters; Forming at least a second layer on the wafer using a second set of manufacturing parameters; Subsequently generating an output signal by measuring a measurement site on the first wafer using a misalignment metrology tool; Generating at least one wavelet-transformed signal by applying at least one wavelet transform to the output signal; Generating a quality metric by analyzing the wavelet-transformed signal, wherein the analysis includes associating a particular portion of the wavelet-transformed signal with a particular location within the measurement site, and wherein the analysis further associates the contribution of one or more structures formed within the measurement site with the quality metric; Generating at least an adjusted set of manufacturing parameters by at least partially adjusting at least one manufacturing parameter based on the quality metric, the at least one manufacturing parameter being selected from at least one of the first set of manufacturing parameters and the second set of manufacturing parameters; And Subsequently forming at least one layer on at least one of the first wafer and a second wafer using the adjusted set of manufacturing parameters.

8. The wavelet analysis method according to claim 7, and wherein the wavelet transform is a continuous wavelet transform.

9. The wavelet analysis method according to claim 7 or claim 8, and wherein the quality metric includes an indication of misalignment between the first layer and the second layer.

10. The wavelet analysis method according to claim 9, and wherein generating the wavelet-transformed signal further comprises: selecting a frequency range, wherein the wavelet-transformed signal varies according to the frequency range; and assigning zero values to the coefficients of the wavelet transform corresponding to frequency values outside the frequency range.

11. The wavelet analysis method according to claim 9, and wherein generating the wavelet-transformed signal further comprises: selecting a frequency range, wherein the wavelet-transformed signal varies according to the frequency range; and multiplying the coefficients of the wavelet transform corresponding to frequency values outside the frequency range by a weighting factor.

12. The wavelet analysis method according to claim 10, and wherein the frequency range includes a frequency corresponding to at least one dimension of at least one structure formed in at least one of the first layer and the second layer.

13. The wavelet analysis method according to claim 10, and wherein the frequency range does not include a frequency corresponding to at least one dimension of at least one structure formed on the wafer.

14. The wavelet analysis method according to claim 13, and wherein the at least one structure is formed together with at least one of the first layer and the second layer.

15. The wavelet analysis method according to claim 13, and wherein the at least one structure is formed together with a structure layer, the structure layer being a layer other than the first layer and the second layer.

16. The wavelet analysis method according to claim 7 or claim 8, and wherein the quality metric includes an indication of the asymmetry of at least one of the following: at least one individual structure among the structures formed within the measurement site; at least one group among the structures formed within the measurement site; and the measurement site.

17. The wavelet analysis method according to claim 16, and wherein analyzing the wavelet-transformed signal comprises: identifying a general symmetry boundary within the wavelet-transformed signal; and identifying at least one pair of asymmetric portions of the wavelet-transformed signal, each of the at least one pair of asymmetric portions comprising a pair of portions of the wavelet-transformed signal that are asymmetric with respect to the general symmetry boundary.

18. The wavelet analysis method according to claim 7 or claim 8, and wherein analyzing the wavelet-transformed signal includes analyzing a signal having the same units as the units of the wavelet-transformed signal.

19. The wavelet analysis method according to claim 7 or claim 8, and wherein analyzing the wavelet-transformed signal further includes analyzing a signal having the same units as the units of the output signal.

20. The wavelet analysis method according to claim 7 or claim 8, and further comprising removing at least one of the first layer or the second layer from the wafer, and wherein the at least one layer replaces at least one of the first layer and the second layer.

21. The wavelet analysis method according to claim 7 or claim 8, and wherein said at least one layer does not replace either the first layer or the second layer.

Images