Raman spectroscopy-based measurements in patterned structures
By applying optimized Raman spectroscopy and modeling tools in semiconductor devices, the measurement problem of complex patterned structures is solved, and accurate measurement of material composition, stress and doping properties is achieved, which meets high sensitivity and high throughput The demand for measurement performance has been significantly improved.
Patent Information
- Application Number
- CN202111225747.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2015-12-15
- Filing Date
- 2016-12-15
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2036-12-15
AI Technical Summary
Existing technologies are difficult to effectively measure and control the material composition, stress and doping properties of patterned structures in complex semiconductor devices, especially in the case of increased device design complexity and performance requirements, which lack the ability to meet stringent sensitivity and high throughput. Required metrology solutions.
Using Raman spectroscopy-based measurement methods, the sensitivity of specific parameters is highlighted and isolated by optimizing illumination and light acquisition conditions, combined with modeling tools, and differentiating different material parameters, the peak positions in the Raman spectroscopy are used to extract material composition and stress. The information of the doping level is monitored and the accuracy of the patterned structural characteristics is achieved.
Multi-property measurement of semiconductor structures is achieved, the sensitivity and accuracy of measurement is improved, and the manufacturing process can be effectively controlled, especially in metrological applications in nanostructured devices.
Smart Images

Figure CN113933286B_ABST
Abstract
Description
[0001] This application is a divisional application of the Chinese national phase application of the PCT application with the application date of December 15, 2016, the international application number of PCT / IL2016 / 051349, and the invention name of “Raman spectroscopy-based measurements in patterned structures”, all contents of which are incorporated herein by reference. Technical Field
[0002] The present invention belongs to the field of metrology technology and relates to a method and system for measuring various parameters / properties of patterned structures such as semiconductor wafers using Raman spectroscopy based measurements. Background Art
[0003] The growing complexity of semiconductor device design in advanced technology nodes involves not only the reduction of structure size and higher complexity of device design, but also the utilization of new materials. Device yield and performance become more sensitive to material properties such as composition, stress, crystallinity, and doping, which in turn requires appropriate metrology solutions for process control.
[0004] Various optical measurement techniques have been developed for measuring strain and other properties of materials.
[0005] US 7,274,440 describes a system and method for measuring stress in a specimen. A system includes an optical subsystem configured to measure stress-induced birefringence in a patterned structure formed on a specimen. In some embodiments, the optical subsystem may be configured as a spectroscopic ellipsometer, a multi-angle laser ellipsometer, a polarimeter, a polarization reflectometer, or some combination thereof. The system also includes a processor coupled to the optical subsystem. The processor is configured to determine stress in a material of the patterned structure using stress-induced birefringence measurements. A method includes measuring stress-induced birefringence in a patterned structure formed on a specimen using optical techniques. The method also includes determining stress in a material of the patterned structure using stress-induced birefringence measurements. Summary of the invention
[0006] The present invention provides a novel solution for measuring various structure parameters, which is particularly useful for controlling the process of structure manufacturing. This is based on the understanding of the inventors that, although some techniques such as X-ray diffraction (XRD), high resolution X-ray diffraction (HRXRD), X-ray fluorescence (XRF), X-ray photoelectron spectroscopy (XPS), low energy electron induced X-ray emission spectroscopy (LEXES) provide relevant information about the parameters of the patented structure, however, a good solution is needed that fully meets the stringent sensitivity and throughput requirements for controlling the manufacturing of patterned structures (particularly, semiconductor process control).
[0007] The present invention provides a novel technique for advanced characterization of properties of semiconductor structures (usually patterned structures). Examples of such properties are material composition, stress, and doping.
[0008] The present invention is based on Raman spectroscopy, typically using a specified measurement configuration (eg, polarization configuration) in combination with suitable modeling capabilities to highlight and isolate sensitivities to parameters of interest, as well as to distinguish sensitivities to different material parameters.
[0009] Raman spectroscopy is an established technique with a large literature describing its use for characterizing the properties of a wide variety of materials. However, as described in more detail below, proper control of the illumination and acquisition channel properties, as well as accompanying signal processing and modeling tools, are critical to enabling accurate measurements using this method.
[0010] Raman spectra carry information about various properties of the sample being probed. Most notably, different peaks in the spectrum correspond to different materials. When the measurement target contains a material compound (e.g., SiGe), a given peak in the Raman spectrum will correspond to different atomic pairs (e.g., Si-Si, Si-Ge, and Ge-Ge).
[0011] In this context, reference Figure 1 , Figure 1 The Raman spectrum from a thin SiGe layer deposited on Si (Graph S1) and from bulk Si as a reference (Graph S2) are illustrated. In the SiGe measurement, four peaks are clearly observed. At 520 cm -1 The strong peaks correspond to Si-Si vibrations in the substrate. The three additional peaks correspond to Si-Si pairs, Si-Ge pairs, and Ge-Ge pairs in the SiGe film. In the pure Si reference spectrum, only the substrate Si-Si peak is observed. Thus, the presence of the SiGe layer produces three additional peaks associated with the vibrations of different atomic pairs in the layer.
[0012] Methods for extracting information about concentration and stress from the positions of these peaks are well known in the literature. For example, a set of equations for the positions of three SiGe peaks with a germanium composition and layer stress are presented in the following publication: TS Perov et al., Composition and strain in thin Si1—xGex virtual substrates measured by micro-Raman spectroscopy and x-ray diffraction, J. App. Phys. 109, 033502 (2011).
[0013] Doping is another feature that affects Raman spectra. The carrier concentration caused by the dopant distribution affects the Raman signal and causes the additional / shift of the Raman peak. Therefore, the level of doping can be incorporated into the fitting program, and the doping level, as well as stress and composition can be evaluated simultaneously by monitoring the peak position (e.g., see A. Perez-Rodriguez et al., Effect of stress and composition on the Raman spectra of etch-stop SiGeBlayers, J. Appl. Phys. 80, 15 (1996)).
[0014] The present invention provides a novel metrology method and device configured to allow optimized Raman metrology for sample features of interest. The method allows access to multiple properties of the sample and has particular application also to metrology of nanostructured devices.
[0015] Furthermore, the present invention provides a set of modeling solutions (methods and systems) that allow the correct exploitation of the degrees of freedom of these method and device configurations, as well as accurate interpretation of Raman measurements.
[0016] The above two aspects of the present invention can be used separately, and have potential great benefits to either aspect. On the contrary, as will be discussed below, they can lead to significantly improved metering performance when used together.
[0017] Therefore, according to a broad aspect of the present invention, there is provided a method for use in measuring one or more features of a patterned structure. The method comprises:
[0018] providing measurement data including data indicative of at least one Raman spectrum obtained from the patterned structure under measurement using at least one selected optical measurement scheme, the at least one selected optical measurement scheme each having a predetermined configuration of at least one of illumination and collection light conditions corresponding to the one or more features to be measured;
[0019] processing the measurement data and determining, for each of said at least one Raman spectrum, a distribution of Raman contribution efficiencies over at least a portion of the structure under measurement, the distribution being dependent on said predetermined configuration of characteristics of the structure and at least one of illumination and collection light conditions in a corresponding optical measurement scheme;
[0020] The distribution of the Raman contribution efficiencies is analyzed and one or more characteristics of the structure are determined.
[0021] The one or more characteristics of the structure to be measured include at least one of the following: size, material composition, stress, crystallinity.
[0022] The predetermined configuration of illumination and collection light conditions is characterized by selecting at least one of the following (possibly illumination and collection individually): excitation wavelength, polarization, retardation, beam shape, angular distribution of the illumination light, and wavefront of the light.
[0023] In some embodiments, the measurement data indicates a number n (n>l) of Raman spectra obtained from the patterned structure under measurement using a number n of optical measurement schemes with n different configurations of illumination and collection light conditions. Processing of the measurement data includes: for each i-th Raman spectrum of the n Raman spectra, calculating a distribution of Raman contribution efficiencies RCE over at least a portion of the structure under measurement i (x, y, z); and selecting one or more distributions of Raman contribution efficiencies corresponding to one or more features to be measured; and determining the characteristics of the structure from the selected distributions. For clarity, RCE represents the spatial distribution of the contribution of the Raman signal. It depends on the coupling of electromagnetic radiation into the structure, the excitation of the Raman signal in the structure, and the coupling of the excitation radiation to the detection system.
[0024] Measurement data including n number of Raman spectra are obtained in n measurement periods using n optical measurement schemes respectively. The n measurement schemes may be performed consecutively; or at least some of the measurement schemes may be performed simultaneously.
[0025] In another broad aspect of the invention, there is provided a method for use in measuring one or more features of a patterned structure, the method comprising:
[0026] applying n number of two or more different optical measurement schemes to the patterned structure and determining measurement data including corresponding n Raman spectra from the patterned structure, the n optical measurement schemes differing from each other in at least one condition of one or both of illumination light and collection light;
[0027] Processing the measurement data and determining the one or more characteristics of the structure, the processing comprising determining, for the n Raman spectra, n Raman contribution efficiency distributions RCE1(x,y,z), RCE2(x,y,z), ... RCE on at least a portion of the structure, respectively. n (x, y, z), each of the Raman contribution efficiency distributions depends on a feature of the structure and a corresponding optical measurement scheme, thereby enabling the determination of the one or more features of the structure.
[0028] In some embodiments, the method of the present invention further comprises: selecting an optimal measurement scheme for determining one or more features of interest in the patterned structure under measurement. This operation is achieved by interpreting the Raman contribution efficiency corresponding to one or more schemes in the previous measurement schemes and optimizing the configuration of at least one of the illumination and collection light conditions (i.e., optimizing the measurement scheme) for one or more schemes in the consecutive measurement schemes applied to at least a portion of the patterned structure.
[0029] In yet another broad aspect, the present invention provides a control system for use in measuring one or more features of a patterned structure. The control system comprises:
[0030] A processor unit configured to receive and process measurement data including data indicating n Raman spectra obtained from a patterned structure under measurement using n optical measurement schemes, each of the n optical measurement schemes having a different configuration of illumination and collection light conditions corresponding to the one or more features to be measured, wherein the processing of the measurement data includes determining, for the n Raman spectra, n Raman contribution efficiency distributions RCE1(x, y, z), RCE2(x, y, z), ... RCE on at least a portion of the structure, respectively. n (x, y, z), each of the Raman contribution efficiency distributions depends on the characteristics of the structure and the corresponding optical measurement scheme, thereby one or more characteristics of the structure can be determined from one or more distributions selected from the n Raman contribution efficiency distributions.
[0031] The control system may further include at least one of an illumination controller and a collection controller configured and operable to controllably change at least one of the illumination and collection light conditions by changing at least one of the following: excitation wavelength, polarization, delay, beam shape, angular spread of the illumination light, angular spread of the collected light, and wavefront of the light.
[0032] The present invention also provides a system for use in measuring one or more features of a patterned structure, the system comprising: an optical measurement system, configured and operable to perform a number n of different optical measurement schemes on the patterned structure and determine measurement data including corresponding n Raman spectra from the patterned structure, the n optical measurement schemes being different from each other in at least one condition of one or both of illumination light and collection light; and the above-mentioned control system, configured to perform data communication with the optical measurement system to receive and process the measurement data and determine one or more features of the patterned structure under measurement. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] For a better understanding of the subject matter disclosed herein and to illustrate how it can be accomplished in practice, embodiments will now be described, by way of non-limiting examples only, with reference to the accompanying drawings, in which:
[0034] Figure 1 Raman spectra from two samples are illustrated, one sample is a thin SiGe layer deposited on Si, and the other sample is pure bulk Si;
[0035] Figure 2 The polarization effect on the relative intensity of the single-phonon peak and the double-phonon peak in Si is illustrated;
[0036] Figure 3a to Figure 3c illustrates the position dependence of the contribution of the main Si peak in the Raman signal for a simple grating structure, where Figure 3c schematically shows a modeled structure formed by a SiO2 layer on a patterned Si layer; and Figure 3a and Figure 3b Measured Raman signals corresponding to wavelength λ=405 nm and wavelength λ=532 nm, respectively, from the Si grating are shown;
[0037] Figure 4 A flow chart showing a Raman spectroscopy model-based method of the present invention for use in the measurement of patterned structures;
[0038] Figure 5 A block diagram of a system of the present invention is shown, which is used to perform Figure 4 a method for interpreting Raman metrology measurement data (and possibly also managing Raman metrology measurements); and
[0039] Figure 6 Shows that it is allowed to Figure 5 A specific but non-limiting example of the configuration of an optical measurement system for use with a control system in conjunction with considerable flexibility in measurement schemes. DETAILED DESCRIPTION
[0040] As described above, the present invention is based on Raman spectroscopy using a specified configuration of at least one of illumination and light collection conditions (e.g., polarization, excitation wavelength, delay, beam shape, angular propagation of illumination / collection light, etc.), and preferably combined with suitable modeling capabilities. The technique provides the ability to highlight and isolate sensitivity to parameters of interest and to distinguish between sensitivities of different material parameters.
[0041] The following is a description of an optimized Raman metrology scheme according to some embodiments of the present invention.
[0042] Raman metrology represents a unique type of light-matter interaction. Different parts of the Raman spectrum respond differently to changes in the optical scheme, specifically: changes in illumination and collection polarization / retardation, illumination and collection angles of incidence and pupil shape, wavefront, and changes in focus.
[0043] As an example of the benefits that may be derived from proper manipulation of one or more of these parameters, we consider the problem of two phonon backgrounds. As described previously, the Raman spectrum from pure Si has a peak at about 520 cm -1 There is a sharp peak at 230cm -1 Up to 380cm -1 An additional, very broad and significantly weaker spectral peak is observed at . This weak Raman signal arises from a 2-phonon process.
[0044] In most cases, metrology is not concerned with the weak Raman signal associated with the 2-phonon process. However, this signal serves as a background signature that can significantly affect (and confuse) the interpretation of the Raman spectrum. The relative intensity between the 1-phonon peak and the 2-phonon peak can be modified by several orders of magnitude through proper control of the illumination and collection polarization.
[0045] Figure 2 This dependency is shown in Figure 2 The polarization effect of the relative intensity of the single-phonon peak and the double-phonon peak in Si is illustrated. Two graphs P1 and P2 are shown, where two polarization configurations are presented, namely, the illumination and collection polarization P ill ||P coll (Graph P1) and P ill ⊥P coll (Graph P2). As shown in the figure, when the illumination and collection polarization P ill and P coll When tuned to be perpendicular to each other and both oriented at 45° to the lattice of the specimen, the 2-phonon peak was observed to be suppressed by 3 orders of magnitude. Thus, by aligning the illumination polarization P at 45° to the crystalline structure, the 2-phonon peak is suppressed by 3 orders of magnitude compared to having the polarizations co-aligned. ill , and collect polarization P coll Aligned perpendicular to the illumination polarization, the 2-phonon signal is significantly reduced.
[0046] This is just one example of how proper polarization manipulation can greatly improve signal quality and the ability to isolate important measurement components from those that are not (or less) of interest. As will be described below, the same principles can be used to highlight sensitivity to specific parameters of interest.
[0047] It is impractical (and may even be impossible) to apply such a simple method to measurement targets that are not planar films but structured films. The electromagnetic field distribution inside the structure can be very complex and depends on the structure size as well as the material characteristics.
[0048] The present invention provides for achieving such sensitivity with outstanding background suppression not only for coverage targets but also for structural measurements. This can be achieved through hardware control and by correctly optimizing the measurement scheme using modeling tools.
[0049] Below are some specific but non-limiting examples of such measurement scheme optimization. These examples include: (i) Profile of measured features across a structure: Profile information across z (depth profile) can be obtained, or even the entire profile of measured properties at different locations (x, y, and z) across the structure; (ii) Metrological measurement of independent stress components inside a structure; (iii) Use of modeling / algorithmic tools during measurement feedback; (iv) Dimensional metrology.
[0050] (i) Profile of measured features across the structure
[0051] Raman spectroscopy provides an integrated measurement of the detected target: the measured signal is averaged over the entire measurement point and over the penetration depth in the sample. There is usually great interest in being able to identify the distribution of the measured features on the structure as a function of depth and lateral position ("profile").
[0052] In order to vary the penetration depth, it is common practice to use multiple wavelengths. The dependence of the penetration depth on wavelength is very pronounced, giving this analytical method good resolving power. Table 1 shows an example of the penetration depth into Si at normal incidence.
[0053] Table 1
[0054]
[0055]
[0056] The values in the table are due to field absorption along the illumination and acquisition paths.
[0057] The method is relatively easy to interpret for planar films, however, its application to structured specimens is highly misleading. The electromagnetic (EM) field distribution inside the structure can have a very complex distribution, depending on the complex way the electromagnetic field interacts with the structural features. In these cases, the measured spectrum represents the material properties at those locations in the structure that contribute most to the Raman signal. These do depend on the wavelength, but in no way can be easily related to the simple concept of 'penetration depth'.
[0058] In this context, reference is made to illustrate the position dependence of the contribution of the main Si peak in the Raman signal for a simple grating structure. Figure 3a to Figure 3c . Figure 3c A modeled structure formed by a SiO2 layer on a patterned Si layer is schematically shown. Figure 3a and Figure 3b Measured Raman signals corresponding to wavelengths λ=405 nm and λ=532 nm, respectively, from Si gratings are shown in . It should be noted that the Raman signal originates only from the grating structure (where Si is present) and not from the SiO2 surrounding it.
[0059] According to the present invention, a model-based approach is used to achieve a highly flexible optical arrangement that allows acquisition of multiple different information channels, allowing comprehensive and structural profiling capabilities (both with respect to depth and / or lateral position). Figure 4 , a flow chart 10 illustrating the model-based method of the present invention. For simplicity, the solution can be divided into fewer steps; these are not all necessary for all implementations, and some uses that do not involve all of these steps are described below. However, these steps represent the main building blocks of the method of the present invention.
[0060] First, according to this method, the 'Raman Contribution Efficiency' (RCE) is defined to specify the position-dependent contribution to the Raman signal. This property depends on the measured structural features (size, material), the excitation wavelength, and the characteristics of the illumination and collection channels (as described below).
[0061] Calculate the Raman contribution efficiency for the measured structure (step 12). Optionally, this calculation can be assisted by information about the structure obtained / measured from other metrology tools and / or test sites (e.g., OCD\SEM\TEM) (step 14) to provide dimensional features, ellipsometry\XPS\SIMS, etc. for material characterization.
[0062] Data about n different measurement conditions are provided (step 16), and Raman contribution efficiency distributions RCE1(x,y,z), RCE2(x,y,z), ... RCE on the structure are calculated for each of the n measurement conditions (usually illumination / collection conditions). n (x,y,z). These can include different angles of incidence (AOI), wavelengths, polarization, pupil shaping options, etc. Each different i-th configuration provides a different distribution RCE in the structure i (x,y,z).
[0063] Then, select a subset of the configurations that have been computed for RCE n-j -RCE n-k, where j and k are integers, j ≥ k (step 18), to obtain information about the distribution of the measured parameters. Deriving the distribution of parameters within the structure from the set of measurements (step 20) can be accomplished using standard algorithms (eg, deconvolution methods).
[0064] As an example, a simple method to achieve this derivation can be based on a linear scheme: collect a set of measured Raman intensities I i , it is known that each I i It is related to the parameter distribution within the structure through its RCE, namely:
[0065] I i =∫RCE i (x, y)P(x, y)dxdy.
[0066] By restricting the measurement structure to some spatial sampling, this relationship can be written in matrix form:
[0067] I i =M i,j P j or equivalently
[0068] Here, the subscript j is associated with different spatial locations and the subscript i is associated with different measurements. Since I and M are known (through measurements and the corresponding modeling engine), the spatial distribution of the parameters can be directly obtained using the RMS scheme:
[0069]
[0070] Many other algorithmic approaches are available, allowing for more stable and more easily controlled solutions.
[0071] The method can be applied to any measurable property such as stress, composition, crystallinity, to name a few non-limiting examples.
[0072] It should be noted that the above exemplified technique does not necessarily have to be applied to a set of measurements. In fact, the method can be used to find a single, optimized measurement scheme that provides outstanding sensitivity to the parameter of interest. The principles in the present invention do not involve using a single measured Raman spectrum to obtain all the cross-sectional information on the structure, but in this way the most important information can be obtained in a short acquisition time. Alternatively, the simultaneous acquisition of Raman spectra for multiple AOIs can be measured using k-space imaging techniques.
[0073] A block diagram of a measurement system 100 is shown that utilizes a flexible Raman metrology measurement scheme. Figure 5This is exemplified. As described, this flexibility, used in conjunction with advanced modeling tools, allows for improved performance as well as entirely new capabilities (e.g., profiling). System 100 includes a control system 106 configured to receive input measurement data indicative of a Raman spectrum (or spectra) measured on at least a portion of a specimen 105 using one or more measurement schemes implemented in an optical system; and to process the measurement data by performing the above-described techniques of the present invention.
[0074] Typically, the measurement data may be processed and analyzed in real time, in which case the control system 106 may receive the measurement data directly from the output of the optical measurement system, or the control system 106 may perform the analysis off-line, in which case the measurement data source is an external storage device. For example, measurement data from a specified type of structure may be analyzed off-line, and the pre-calculated data may form a parameter-dependent 'library' of theoretical spectra to select a predetermined (optimal) measurement scheme for measuring a structural parameter of interest used in controlling a manufacturing process for the structure of the type in question.
[0075] The optical measurement system comprises a light source system 102 defining an illumination channel IC and a detection system 104 defining a collection channel CC, and further comprises a light affecting unit located in / associated with at least one of the illumination channel and the collection channel. In this non-limiting example, the system comprises a light illumination affecting unit 108 and a light collection affecting unit 110. The illumination and collection affecting units are configured for affecting illumination and collection conditions. Such units may comprise light propagation affecting optics for affecting conditions of propagation of light along the respective channels. The illumination affecting unit 108 may comprise a controller (e.g., outside of the light propagation affecting optics) for controlling the operation of the light source system; alternatively or in addition, such a controller may be part of the control system 106.
[0076] The control system 106 is typically a computer system configured to communicate with a measurement data provider (this may be wired and / or wireless communication using any known suitable technology); and may also be configured to manage / control measurements using different measurement schemes, in which case the control unit is also configured to communicate with at least some elements of the optical measurement system (this may be wired and / or wireless communication using any known suitable technology). As shown in the present example, the control system 106 may include an illumination controller 106A associated with the illumination influencing unit 108 and / or the light source system 102; an acquisition controller 106B associated with the acquisition influencing unit 110; and a data processor entity 106C. The latter is configured (programmed as a specially designed software product) to carry out the above-described method for processing data indicative of the measured Raman spectrum (or spectra) and determining the parameter distribution within the sample 105.
[0077] The efficiency of the above approach depends on the measurement flexibility allowed by the measurement system. The degrees of freedom added to the optical path allow for improved fine tuning of the measurement to the required metrology target. Figure 6 Specific but non-limiting examples of configurations of the optical system 200 that allow for such considerable flexibility are shown in a self-illustrative manner, including polarization control (including measurement of the full Raman-Mueller matrix), pupil shaping at illumination and acquisition, multiple wavelengths, k-space imaging, etc. For ease of understanding, the same reference numerals are used to denote Figure 5 and Figure 6 The main functional components of the system.
[0078] exist Figure 6 In the system 200, the light source system 102 includes n light sources (e.g., lasers) each for generating a light beam of a different wavelength; and includes / associated with n beam adjustment units respectively located on the optical paths of the n light beams; and includes a laser beam selection unit (optical and / or electronic unit) for selecting an excitation (illumination) wavelength. Beam adjustment includes spectral filtering, and / or polarization, and / or spatial filtering, and / or beam expansion, and / or collimation. The beam adjustment unit is part of the illumination influencing unit 108. In this example, the illumination influencing unit 108 further includes a polarization rotation component (e.g., a variable λ / 2 plate) and an aperture variation component. In other words, the illumination influencing unit 108 is configured to enable various illumination wavelengths, polarization conditions, and angular propagation of illumination. The acquisition influencing unit 110 is configured to provide various polarization conditions, spectral filtering, and angular propagation of light collected from the sample.
[0079] (ii) Measurement of stress distribution within the structure, allowing separation of different stress components
[0080] Stress metrology is a standard target of Raman spectroscopy. Crystalline strain accompanied by internal stress affects the forces between atoms, resulting in corresponding changes in vibrational frequencies, which are directly detected by Raman spectroscopy through shifts in the Raman peak positions. There are different methods to separate stress / strain from other features that affect the peak positions (e.g., composition, doping).
[0081] It has been shown that Raman spectroscopy at normal incidence is primarily sensitive to the z-component of crystallographic strain (GH Loechelt et al., Polarized off-axis Raman spectroscopy: A technique for measuring stress tensors in semiconductors, J. App. Phys. 86, 6164 (1999)). For simple, unpatterned films, some sensitivity to the normal stress component can be obtained using tilted illumination and / or collection channels, or very high digital aperture measurement settings. However, these methods are completely irrelevant when measuring in patterned structures.
[0082] Similar to the metrology described above with respect to spatial profiles, the present invention provides for the use of modeling tools in conjunction with the selected measurement model (illumination / acquisition AOI and pupil shaping, wavelength, polarization, etc.) to identify a preferred metrology scheme (or combination of schemes) to provide the best sensitivity to the strain orientation of interest. Alternatively, the best metrology scheme may be identified experimentally, first by utilizing a large set of variable measurement information channels, and then identifying the most advantageous set (the set that provides the best precision, accuracy, and / or any other attribute).
[0083] From a set of such measurements using standard algorithms, the full vector strain distribution within the structure can be derived. For clarity, possible schemes are illustrated below, but it should be understood that many other algorithmic approaches can be implemented. The same method can be used in the same way for profiling of structures for any other properties (e.g., composition, doping) to which the Raman signal is sensitive.
[0084] Each measurement data set of the RCE depends on the strain distribution by some known weight on different parts of the structure (which is known based on the modeling tool). Explicitly, we can write:
[0085]
[0086] Here, C i represents a measurement channel / measurement scheme defined by the configuration of illumination and acquisition light conditions (AOI, polarization, azimuth, etc.); (x, y, z) are spatial coordinates; RCE Ciis the Raman contribution efficiency of the channel defined above; P is a position-dependent parameter (in this case, strain); and S[P(x,y,z)] is the associated Raman signal obtained by modeling for this distribution of parameters.
[0087] The rationale behind this expression is that the Raman signal sums individual contributions from different parts of the structure, each contribution depending on the local strain and weighted according to the local RCE. It should be noted that this expression is an approximation of a more rigorous derivation and is used here for simplicity to illustrate the proposed method for strain distribution characterization.
[0088] In matrix form, this expression can be written as:
[0089]
[0090] Here, the subscript i represents the measurement channel, and vectorization is performed in the entire spatial domain (ie, different entries in the vector correspond to different locations).
[0091] Now consider a set of measurement channels, which results in a set of measurement signals We can write:
[0092]
[0093] in, is a matrix known from the model that holds the weights for the structure of the individual measurement channels.
[0094] Using the measurement signal and the calculated RCE, the spatial strain distribution can be estimated by using the following formula:
[0095]
[0096] (iii) Use of modeling / algorithmic tools in the measurement feedback process
[0097] The above solution involves using modeling capabilities to predefine the measurement sequence of the application of interest, i.e., what channel combination to use, how to optimize the measurement sequence, etc. This solution can be extended to allow the measurement process to be modified based on the measurement results. For this implementation, several variations are possible, as follows:
[0098] Results from previous Raman measurement interpretations can be used. More specifically, a first set of Raman measurements (one or several) provides information about the features measured. Then, using modeling capabilities, the measurement features (wavelength, AOI, polarization, etc.) can be changed to provide improved measurement performance. Successive measurements performed on other locations, or repeated measurements performed on the same location, can interpret the method.
[0099] An example of specific interest for this use of metrology feedback is the identification of stress relaxation. When growing crystalline layers (e.g., epitaxial growth of SiGe or GaAs) on substrates made of different materials (e.g., Si), crystalline strain develops due to different lattice constants. Depending on the growth conditions and layer thickness, strain relaxation can potentially occur in the form of alternating relaxed and strained areas. This strain relaxation is catastrophic to the performance of the fabricated device and requires appropriate monitoring. Based on the concept of the measurement feedback process, the use of Raman spectroscopy can provide this metrology as follows: Local strain and composition can be inferred from Raman measurements. In extreme cases, this measurement alone is sufficient to identify obvious strain relaxation. However, if the relaxation is not excessively severe, the measurement can only identify the suspected relaxation. Using model-based tools, the strain readings can be analyzed to identify the suspected relaxation situation. Modeling is particularly important for measurements on structures when different strain components can confound the measurement. When a suspected strain-relaxed specimen is identified, another set of Raman measurements can be performed at adjacent locations. As noted, it is a typical feature of a strain-relaxed specimen when the strain becomes inhomogeneous, expressing areas of high and low strain. If such variability is indeed identified, the specimen is classified as strain relaxed.
[0100] (iv) Dimensional measurement
[0101] One capability of unique interest that is enabled by adding modeling capabilities to Raman spectroscopy is dimensional metrology. Indeed, this capability requires a multifaceted modeling tool / approach involving a comprehensive characterization of the penetration of the electromagnetic field into and out of the structure, as well as modeling of the Raman signal within the structure. This approach can provide highly sensitive information about the measured structure.
[0102] Evidence that the size factor affects the measured Raman signal is known in the literature, for example, from the following publications: A.K. Arora et al., Raman spectroscopy of optical phonon confinement in nanostructured materials, J. of Raman Spectroscopy 38, 604 (2007); B. Kaleli et al., Strain characterization of FinFETs using Raman spectroscopy, Thin Solid Films 31497 (2013); T. Nuytten et al., Edge-enhanced Raman scattering in narrows Ge fin field-effect transistor channels, App. Phys. Lett. 106, 033107 (2015). In specific cases (e.g., nanowires), the Raman signal was found to provide size information about the dimensional features of the structure (e.g., nanowire diameter [J. Liu et al., Raman spectrum of array-ordered crystalline silicon nanowires, Physica E 23, 221 (2004); RP Wang et al., Raman spectral study of silicon nanowires: High-order scattering and phonon confinement effects, Phys. Rev. B 61, 16827 (2000)]).
[0103] However, similar to the modelling used in OCD metrology, through general modelling capabilities, metrology similar to OCD can be used to solve the inverse problem of inferring dimensional properties from measurements. In this approach, the measured signal is compared to the signal calculated from the modelling tool for some assumed properties of the test structure (size, material). When a good fit is obtained between the measured and calculated signals, it is inferred that the measured structure has similar characteristics to the corresponding calculated structure. Similar to common practice in OCD metrology, the theoretical Raman signal can be calculated in real time ('real-time regression') or pre-calculated to form a parameter-dependent 'library' of theoretical spectra.
Claims
1. A method for use in measuring one or more features of a patterned structure, the method comprising: applying n optical measurement schemes of two or more different optical measurement schemes to the patterned structure and determining measurement data including corresponding n Raman spectra from the patterned structure, the n optical measurement schemes differing from each other in at least one condition of one or both of illumination and acquisition; and The measurement data is processed and the one or more characteristics of the patterned structure are determined, where n is an integer greater than one, and the processing includes determining, for the n Raman spectra, n Raman contribution efficiency distributions RCE1(x, y, z), RCE2(x, y, z), ... RCE on at least a portion of the patterned structure, respectively. n (x, y, z), each of the Raman contribution efficiency distributions is different from a feature of the patterned structure and depends on a feature of the patterned structure and a corresponding optical measurement scheme, thereby enabling determination of the one or more features of the patterned structure, wherein a set of measured Raman intensities is the integral of the product of the parameter distribution within the patterned structure multiplied by the Raman contribution efficiency distribution, The calculation of the distribution of the Raman contribution efficiency for each of the n Raman spectra comprises processing additional measurement data obtained by one or more measurements of a different type than Raman spectroscopy applied to the patterned structure or a corresponding test structure. 2 . The method of claim 1 , comprising selecting a subset of the n Raman contribution efficiency distributions, and determining the one or more characteristics of the patterned structure based on the subset.
3. The method according to claim 1, wherein: The one or more characteristics of the patterned structure to be measured include at least one of the following: size, material composition, and stress.
4. The method according to claim 1, wherein: The one or more characteristics of the patterned structure to be measured is stress relaxation.
5. The method according to claim 1, wherein: The one or more characteristics of the patterned structure to be measured is crystallinity.
6. The method according to claim 1, wherein: The predetermined configuration of at least one of the illumination and collection light conditions is characterized by selecting at least one of: excitation wavelength; polarization; delay; beam shape; angular spread of the illumination light; angular spread of the collected light; and wavefront of the light.
7. The method according to claim 1, wherein: The at least one condition includes pupil shape.
8. The method according to claim 1, wherein: At least some of the n measurement segments are performed simultaneously.
9. A system for use in measuring one or more features of a patterned structure, the measurement system comprising: an optical measurement system configured and operable to perform a number n of different optical measurement schemes on a patterned structure and determine measurement data including corresponding n Raman spectra from the patterned structure, the n optical measurement schemes differing from one another in at least one condition of one or both of illumination and collection light; wherein n is an integer greater than one; and A control system, comprising a processor unit, configured to receive and process the measurement data, wherein the processing of the measurement data comprises determining, for the n Raman spectra, n Raman contribution efficiency distributions RCE1(x, y, z), RCE2(x, y, z), ... RCE on at least a portion of the patterned structure, respectively. n (x, y, z), each of the Raman contribution efficiency distributions being different from a feature of the patterned structure and depending on a feature of the patterned structure and a corresponding optical measurement scheme, whereby the one or more features of the patterned structure can be determined from one or more selected ones of the n Raman contribution efficiency distributions, A set of measured Raman intensities is the integral of the product of the parameter distribution within the patterned structure multiplied by the Raman contribution efficiency distribution, Therein, calculating the distribution of the Raman contribution efficiency for each of the n Raman spectra comprises processing additional measurement data obtained by one or more measurements of a different type than Raman spectroscopy applied to the patterned structure or a corresponding test structure.
10. The system according to claim 9, wherein: The processor is configured to select a subset of the n Raman contribution efficiency distributions, and the one or more characteristics of the patterned structure are determined based on the subset.
11. The system according to claim 9, wherein: The one or more characteristics of the patterned structure to be measured include at least one of the following: size, material composition, and stress.
12. The system according to claim 9, wherein: The one or more characteristics of the patterned structure to be measured is stress relaxation.
13. The system according to claim 9, wherein: The one or more characteristics of the patterned structure to be measured is crystallinity.
14. The system according to claim 9, wherein: The predetermined configuration of at least one of the illumination and collection light conditions is characterized by selecting at least one of: excitation wavelength; polarization; delay; beam shape; angular spread of the illumination light; angular spread of the collected light; and wavefront of the light.
15. The system according to claim 9, wherein: The at least one condition includes pupil shape.
16. The system of claim 9, wherein: At least some of the n measurement segments are performed simultaneously.
