KR101321861B1 - Interferometer and method for measuring characteristics of optically unresolved surface features - Google Patents

Abstract

Comparing the information generateable from the plurality of scanning interferometry signals corresponding to different surface locations of the test object with information corresponding to the plurality of models for the test object; And outputting information about the unanalyzed surface shape based on the comparison, wherein the plurality of models are parameterized by a series of features associated with one or more unanalyzed lateral shapes of the test object. A method is disclosed.

Description

INTERFEROMETER AND METHOD FOR MEASURING CHARACTERISTICS OF OPTICALLY UNRESOLVED SURFACE FEATURES}

Surface topography of objects with complex surface structures, such as thin film (s), discrete structures of dissimilar materials, or discrete structures not analyzed by the optical resolution of an interference microscope And / or using a scanning interferometer to measure other features. These measurements relate to the characterization of flat panel display components, semiconductor wafer metrology, and analysis of thin films and dissimilar materials in-situ.

Cross reference of related application

This application claims the benefit of 35 U.S.C. 119 (e), filed Nov. 15, 2005 under the name "INTERPEROMETER AND METHOD FOR MEASURING CHARACTERISTICS OF OPTICALLY UNRESOLVED SURFACE FEATURES" Priority claims based on U.S. Provisional Patent Application No. 60 / 737,016, and the present application also claims benefit by 35 USC120, "PROFILING COMPLE SURFACE STRUCTURES USING HEIGHT SCANNING INTERFEROMETRY), filed on March 8, 2004, filed in part of US Patent Application Ser. No. 10 / 795,579. US Patent Application No. 10 / 795,579, supra, discloses 35 U.S.C. U.S. Provisional Patent Application No. 60, filed March 6, 2003, entitled "PROFILING COMPLEX SURFACE STRUCTURES USING HEIGHT SCANNING INTERFEROMETRY" by 119 (e). / 452,615; United States Provisional Patent Application 60 / 452,465, filed March 6, 2003, entitled "PROFILING COMPLEX SURFACE STRUCTURES USING SIGNALS FROM HEIGHT SCANNING INTERFEROMETRY". number; And US Provisional Patent Application No. 60 / 539,437, filed Jan. 26, 2004 entitled "SURFACE PROFILING USING AN INTERFERENCE PATTERN MATCHING TEMPLATE." do. All of the above related applications are incorporated herein by reference.

Interferometry techniques are commonly used to measure the profile of an object's surface. To do so, the interferometer combines reference wavefronts reflected from the reference plane and measurement isophase planes reflected from the surface of interest to produce an interferogram. The fringe of the interference image represents the spatial change between the object plane and the reference plane.

Typically, a scanning interferometer has an optical path length between the reference leg and the measurement leg of the interferometer over a range similar to or wider than the coherence length of the interference isophase plane. length difference, OPD), to generate a scanning interferometry signal for each camera pixel used to measure the interfering image. For example, a white light source can be used to generate a limited coherence length, called scanning white light interferometry (SWLI). A typical scanning white light interferometry (SWLI) signal is a small number of fringes localized near the zero optical path length difference (OPD) position. This signal is typically characterized by sinusoidal carrier modulation ("fringe") with a bell shaped fringe-contrast envelope. The conventional idea underlying SWLI metrology is the use of fringe localization to measure surface profiles.

SWLI processing technology includes two principal trends. The first approach is to place the peak or center of the envelope assuming a position corresponding to the zero optical path length difference (OPD) of two beam interferometers, where one beam reflects off the object surface. The second approach inevitably assumes that the linear slope is directly proportional to the position of the object, converting the signal into the frequency domain and calculating the rate of change of phase with respect to the wavelength. See, for example, US Pat. No. 5,398,113 to Peter de Groot. This second approach is called frequency domain analysis (FDA).

Unfortunately, this assumption can be broken when applied to a test object having a thin film due to reflection by the thin film top surface and the film / substrate interface below it. Recently, there has been a method for this structure. W. S. W. Kim and G. H. U.S. Patent No. 6,545,763 to G. H. Kim. This method fits the frequency domain phase profile of the SWLI signal for the thin film structure to the estimated frequency domain phase profile for various film thicknesses and plane heights. Simultaneous optimization was used to determine calibrated film thickness and face height.

Composite surface structures, for example patterned semiconductor wafers, may be comprised of features of dissimilar materials of various sizes ranging from mm to tens of nm. Quantitative measurement of surface topography is now very important in some industries, including certain semiconductor industries. Since typical chip shapes are small in size, the equipment used to measure them should generally have high spatial resolution for both the horizontal and vertical directions of the chip surface. Engineers and scientists use surface topography measurement systems to control the process and to detect defects that may occur as a result of manufacturing processes, especially processes such as etching, polishing, cleaning, and patterning.

In the semiconductor industry, non-optical measurements such as top down critical dimensions (CD) scanning electron microscopy (SEM) and atomic force microscopy (AFM) are used to obtain pattern and topography information. Non-optical metrology tools are widely used. All of these techniques required horizontal resolution, which is very slow, requiring considerable time to collect data across a large area of wafers. This is especially true of atomic force microscopy (AFM). Top-down critical dimension scanning electron microscopy (CD SEM) can be programmed to automatically collect data from specific areas of the wafer set, but even with this feature the time required to collect complete wafer data is enormous.

Conventional optical surface profilers, such as confocal sensors, interferometeric sensors, or slope sensors, overcome some of these problems, but surface features It is not generally available if it is too small for proper analysis, too narrow a gap, or both, and the surface height change is not accurate.

Scanning white light interference microscopes, also known as coherence probe microscopes, laser radars, and vertical scanning interferometers, provide a surface profile using limited coherence of white light (or more generally broadband light) illumination. To aid in discrete surface morphology, rough surface structure, and narrow line surface profiling. Scanning white light interference microscopes generally have a lateral resolution that is limited to approximately one wavelength of the illumination light source. Some of these scanning white light interference microscopes can be configured to measure the thickness of the film.

Scatterometer determines the surface characteristic by contrasting the distribution of scattered or diffracted light with a library pre-calculated for scattering and diffraction distributions from the nominal structure. The scatterometer does not directly measure the surface profile with respect to the reference, as in an interferometer. Scatterometers also generally work with only a limited set of 2D structures.

A confocal microscope uses a limited depth of focus to vertically partition an object, for example to measure the surface profile.

Nomarski microscopy and other differential techniques measure the difference in surface height by comparing the surface heights with each other.

Ellipsometers use Fresnel reflection coefficients and high incident angle polarization to measure heterogeneous material structures and thin films. In general, the feature of interest is greatly compared to the source wavelength, and the ellipsometer does not provide surface profile information.

The inventors found that the scan interferometric signal is rich in information, many of which are ignored in conventional processing. However, complex surface structures, such as thin films or under-resolved surface shapes (i.e., transverse surface shapes smaller than the spatial resolution of an interference microscope), can be used to identify peak positions of fringe contrast envelopes or to determine frequency domain phase profiles. While errors can occur in conventional processing techniques based on calculating the slope for, the new processing techniques disclosed herein can extract surface height information and / or information about composite face structures.

For example, although it is not assumed that plane height information is directly related to the peak of the fringe contrast envelope, some embodiments of the present invention change the plane height change to a scan interferometry signal for a reference scan location, but otherwise scan Preserve the shape of the interferometry signal. Thus, the shape of the scan interferometry signal is particularly useful for characterizing composite surface structures because it is independent of plane height (independent). Similarly, in the frequency domain, some embodiments assume a change in plane height that introduces a linear term in the frequency domain phase profile, although the frequency domain profile itself may not be linear. However, the change in plane height leaves the frequency domain amplitude profile unchanged. Thus, the frequency domain amplitude profile is particularly useful for characterizing composite surface structures.

After characterizing the composite surface structure, the plane height can be determined efficiently. For example, cross-correlation between a scan interferometry signal and a model signal having a shape corresponding to the composite surface structure may produce a peak in scan coordinates corresponding to the plane height. Likewise, in the frequency domain, the phase contribution due to the composite surface structure can be subtracted from the frequency domain phase file, and the plane height can be extracted using conventional FDA analysis.

Examples of composite surface structures include simple thin films (in this case, for example, an important variable parameter may be the film thickness, the refractive index of the film, the refractive index of the substrate, or any combination thereof); Multilayer thin film; Sharp edges and surface shapes that produce diffraction or otherwise complex interference effects; Unanalyzed surface roughness; Unresolved surface features such as, for example, sub-wavelength width grooves on other smooth surfaces; Heterogeneous materials (e.g., the surface may comprise a combination of thin film and solid metal, in which case the library may comprise both surface structure forms and automatically match the membrane or solid metal by matching the corresponding frequency domain spectrum). Can be identified); Surface structures that are the source of light activity, such as fluorescence; Spectroscopic properties such as color and wavelength dependent reflectivity; Polarization dependent properties of the surface; And deflectable surface shapes or surface deflections, vibrations or movements that result in perturbation of the interference signal.

In some embodiments, the limited coherence length of the light used to generate the scan interferometry signal is based on a white light source or more generally a broadband light source. In other embodiments, the light source may be monochromatic, and the limited coherence length may be due to the use of high numerical apertures (NAs) for light directed to the test object and / or light received from the test object. have. High NA causes the light rays to contact the test surface over an angular range and produce different spatial frequency components in the recorded signal when the OPD is scanned. In another embodiment, the limited coherence may be due to the combination of both effects.

The origin of the limited coherence length is also the physical basis for the information present in the scan interferometry signal. Specifically, the scan interferometry signal contains information about the composite surface structure since it is generated by light rays that contact the test surface at many different wavelengths and / or many different angles.

In the processing technique described herein, information that can be generated from a scanning interferometry signal (including the scanning interferometry signal itself) for the first surface position of the test object is compared with information corresponding to a plurality of models for the test object. do. Multiple models are parameterized by a series of features of the test object. For example, the test object may be shaped as a thin film, and the set of features may be a set of values for the thickness of the thin film. The information compared may include, for example, information about the frequency domain phase profile, while it may include information about the shape of the scan interferometric data and / or information about the frequency domain amplitude profile. Also, in order to focus the comparison on the composite surface structure rather than the surface height at the first surface location, the plurality of models may all correspond to a fixed surface height of the test object at the first surface location. The comparison itself may be based on the calculation of a merit function indicating the similarity between the information from the actual scan interferometry signal and the information from each model. For example, the merit function may represent a fit between information that can be generated from the scan interferometric data and a function parameterized by a series of features.

Further, in some embodiments, a series of features, including, for example, diffractive surface structures that contribute to the interfacial signal at the first surface location, correspond to the features of the test object at a second location different from the first location. . Thus, often the composite surface structure is something other than the surface height at the first surface location corresponding to the scanning interferometric signal, but the composite surface structure is spaced from the first surface location corresponding to the scanning interferometric signal. It may correspond to a surface height shape.

In another embodiment, the information that can be generated from the scan interferometric signal is an estimate of the surface height relative to the first location. This information indicates that when measuring different values of the actual surface height profile with an interferometry system (considering the unresolved shape at multiple surface locations), the apparent height of the first surface location is determined. It is compared with models that calculate how much it is. Based on this comparison, the accuracy of the surface height measurement is improved, for example by selecting the actual surface height used in the model that produced the apparent height that is closest to the actual measured height.

More generally, information can be generated from the scanning interferometry signals at a plurality of surface locations (even if at least some of the surface shapes present at these locations have not been analyzed), and the information can be retrieved from an important shape that has not been analyzed. It can be compared with models that compute information that should be a function of different values of parameterization. For example, the generateable information may correspond to the observed surface profile measured using conventional scanning interferometric algorithms. Important shapes that are not analyzed will become obscure in this surface profile, but will still contribute to the observed surface profile, and information about the unanalyzed shape can be extracted from the observed surface profile by comparing the observed surface profile with other models. have.

For example, individual lines of the non-analyzed grating structure will become obscure in the surface profile observed from the processing of conventional interferometric signals at multiple surface locations. Nevertheless, for example, the inventors have found that the surface height of the grating structure of the apparent cavity in the surface profile obtained from the scanning interferometry signal of which the modulation depth of the actual grating structure is processed in a conventional manner. Has been found to correlate with.

The observed surface height of the cavity lattice structure is an example of the "apparent" property of the test surface. In other words, because the test surface includes shapes that have not been analyzed, this is an example of a characteristic related to how the test surface appears in the measurement equipment. These apparent properties can be compared with different models by expecting the expected response of the measuring instrument by parameterizing with different values characterizing the unanalyzed shape of the test surface. This comparison reveals which of the different values produces the expected response that is closest to the observed response, thereby providing information about the unanalyzed shape of the test subject based on the apparent characteristics generated from the scan interferometric signal (s). can do.

The measurement techniques and post-measurement analytical methods described herein are applicable to several semiconductor processing steps. By the use of optical proximity correction and a phase shift mask, the dimension of the patterned object may be smaller than the wavelength used by the optical lithography instrument. For example, in today's high volume manufacturing equipment, a 193 nm lithographic apparatus patterns an object at 65 nm; The use of the etching bias step and the hard mask structure can make the lower limit to 45 nm or less. The use of the ability to print shapes of sub wavelengths requires monitoring of these shapes and the associated etching and deposition steps. The disclosed embodiments allow the measurement of nested patterned structures, where the nested form is defined as a repeating surface structure of known shape with periodicity similar to the transverse dimension of the structure. In particular, this netted structure can be used to monitor the following process steps: isolation patterning and etching, polysilicon gate electrode patterning and etching, source / drain etching and deposition, and a plurality of front end metallizations (front). end metalization) patterning, etching, and polishing processes. Additional applications include the measurement of resist on some films / substrates. Exposure and focus curves are characterized by changes in line width and line depth that can be measured with the apparatus and method of the present invention.

One example of such in-process metrology measurement of semiconductor chips is the use of scanning interferometer measurements for non-contact surface topography measurements of semiconductor wafers during chemical mechanical polishing (CMP) of dielectric layers on the wafer. It involves doing. CMP is used to smooth the surface of dielectric layers and is suitable for precision optical lithography. Based on the results of the interferometric topography, the process conditions of the CMP (eg, pad pressure, polishing slurry composition, etc.) to maintain surface non-uniformities within acceptable limits. Can be adjusted.

The various aspects and features of the present invention are now outlined.

In general, in one aspect, (i) comparing information that can be generated from a plurality of interferometric signals corresponding to different surface locations of a test object with information corresponding to a plurality of models for the test object; And (ii) outputting information about the unanalyzed surface shape based on the comparison, wherein the plurality of models are characterized by a series of features related to one or more unanalyzed lateral shapes of the test object. A parameterized method is disclosed.

Embodiments of the method may include any of the following features.

The one or more unanalyzed lateral shapes of the test object may correspond to one or more of the pitch, modulation depth, and element width of the unanalyzed patterned lateral structure on the test object. For example, the series of features may include different values for the modulation depth. In addition, the plurality of models may be represented by a correlation that maps possible results of information that can be generated from a plurality of interferometric signals to corresponding ones of the different values of the modulation depth, wherein the comparing comprises modulating And determining among the different values of the depth that best matches information that can be generated from the plurality of interferometric signals.

The modulation depth can be expressed in terms of the bias offset value.

At least some of the interferometric signals may be generated from illumination of the test object whose polarization is oriented with respect to the components of the patterned transverse structure. For example, the polarization may be linear polarization aligned in a direction orthogonal to the length of the individual components defining the patterned transverse structure (referred to herein as "x polarization").

The one or more unanalyzed lateral shapes of the test object may be one or more of the position and height of the step of the test object. For example, the series of features may include different values of the position or height of the stairs.

The information that can be generated from the plurality of interferometric signals may include one or more values extracted from the height profile of the test object generated from the plurality of interferometric signals, wherein the unanalyzed surface shape is the extracted height. Ambiguous or unclear within the profile. For example, the test object may include a patterned transverse structure in which individual components are not ambiguous or apparent in the extracted height profile.

The information that can be generated from the plurality of interferometric signals may be a value for the height of the collection of unanalyzed components in the patterned transverse structure extracted from the height profile. The information about the unanalyzed surface shape may correspond to one or more of the width and modulation depth of the components of the patterned transverse structure.

Different surface locations for the interferometric signals may include a reference portion of the test object that provides a reference height value of the extracted height profile. For example, the test object may be etched to produce the patterned structure, and the reference portion of the test object may be a portion that is known to be unetched of the test object.

At least some of the interferometric signals that determine the height profile may be generated from illumination of the test object, whose polarization is oriented with respect to the components of the patterned transverse structure. For example, the polarization can be linear polarization aligned in a direction orthogonal to the length of the individual component defining the patterned transverse structure (x polarization).

The height profile may be obtained from frequency domain analysis of the interferometric signal. Alternatively, the height profile can be obtained from the relative position of the coherence peaks in each interferometric signal. The height profile may also be obtained using other methods.

The unanalyzed surface shape of the test object may have a shape size smaller than 400 nm, smaller than 200 nm, or smaller than 100 nm.

The models can be generated computationally using rigorous coupled wave analysis (RCWA).

The models can be empirically generated from test objects having known properties.

Information about the unshaved surface shape may be output to the user.

Information about the unanalyzed surface shape can be output to an automated process control system for semiconductor manufacturing.

The interferometric signal may be a scan interferometric signal. For example, the scanning interferometry signal may image test light from the test object to cause interference with reference light on a detector, and at the common source between the interference portions of the test light and the reference light. Varying the optical path length difference to the detector, wherein the test light and the reference light are obtained from the common source, and the scanning interferometry signal is generated by the detector when the optical path length difference changes. Corresponds to the measured interference intensity. The method may further comprise generating the scan interferometry signal.

Such a scan interferometry signal may be a low coherence scan interferometry signal. For example, the test light and the reference light may have a spectral bandwidth greater than 5% of the center frequencies of the test light and the reference light, and the optical path length difference is such that the spectrum is generated to generate the scanning interferometry signal. Vary over a wider range than the coherence length.

The low coherence may also allow optics used to deliver test light onto the test object to image the detector so that the numerical aperture for the test light is greater than 0.8. To reduce the coherence length, the common source may be a spatially extended source.

In another related aspect, a processor in a computer may be configured to compare information generated from a plurality of scanning interferometry signals corresponding to different surface locations of a test object with information corresponding to a plurality of models for the test object. Apparatus comprising a computer readable medium having a program is disclosed, wherein the plurality of models are parameterized by a series of features related to one or more unanalyzed lateral shapes of the test object, the comparison Outputs information about the unanalyzed surface shape based on.

In yet another aspect, (i) a scanning interferometry system configured to generate a plurality of scanning interferometry signals corresponding to different surface locations of a test object; And (ii) coupled to the scanning interferometry system to receive the scanning interferometry signal, and programmed to compare information generated from the plurality of scanning interferometry signals with information corresponding to a plurality of models for the test object. An apparatus comprising an electronic processor is disclosed, wherein the plurality of models are parameterized by a series of features associated with one or more unanalyzed lateral shapes of the test object, and the unanalyzed lateral direction based on the comparison. Outputs information about the surface shape.

Embodiments of both devices may include any of the features described above in connection with the corresponding method.

In another aspect, a method of determining one or more spatial peroperties for a grating structure on a test object is disclosed, wherein the grating structure is a line of width narrower than 400 nm that is not fully analyzed by an interference microscpe. Contains components. The method includes (i) determining an apparent height to collect at least some of the grating lines from interference signals at different locations of the test object measured by the interference microscope; (ii) providing an expected response of the interference microscope for different possible values of the properties of the grating structure, wherein the expected response includes contributions of unanalyzed line components of the grating structure. Providing the expected response; (iii) comparing the apparent height with an expected response of the different possible values to determine information about the spatial characteristics of a lattice structure; And (iv) outputting the determined information about the spatial characteristics of the lattice structure.

In addition to the above features, embodiments of the method may include any of the following features.

The apparent height may be determined with reference to the reference portion of the test object.

The interference microscope can illuminate the grating structure with polarized light (x polarization) in a direction orthogonal to the length of each grating line when determining the apparent height.

The determined information about the spatial characteristics of the grating structure may correspond to the modulation depth of the grating structure.

The grating structure may be a series of periodically spaced lines formed at least in part by etching portions between the lines of the test object.

The interference signal is configured to image test light coming from the test object to cause interference with reference light on a detector, and an optical path from the common source to the detector between the interference portions of the test light and the interference light. Scan interference signals generated by varying the length difference, wherein the test light and the reference light are generated from the common source, and each scan interferometry signal is measured by the detector when the optical path length difference is changed. Corresponding interference intensity. The method may further comprise generating the scan interferometry signal. For example, the optical path length difference may vary over a wider range than the coherence length of the interference microscope.

In a related aspect, an apparatus for determining one or more spatial peroperties for a lattice structure on a test object is disclosed, wherein the lattice structure is a line of width narrower than 400 nm that is not fully analyzed by an interference microscpe. Components comprising: 1) determining, by the interference microscope, an apparent height to collect at least some of the grating lines from interference signals measured at different locations of the test object; 2) provide an expected response of the interference microscope for different possible values of the properties of the grating structure, wherein the expected response comprises contributions of the unresolved line components of the grating structure, and 3) the apparent height of the Determine information about the spatial characteristic of the lattice structure compared to the expected response of different possible values; 4) output the determined information about the spatial characteristics of the lattice structure.

In another related aspect, an apparatus for determining one or more spatial characteristics for a grating structure on a test object is disclosed, wherein the grating structure includes line components of widths less than 400 nm that are not fully analyzed by an interference microscope, and The apparatus includes the interference microscope; And an electronic processor coupled to the interference microscope, the electronic processor comprising: 1) an apparent height to collect at least some grating lines from the interference signals measured at different locations of the test object by the interference microscope; Determine; 2) provide an expected response of the interference microscope for different possible values of the properties of the grating structure, wherein the expected response includes contributions of unanalyzed line components of the grating structure; 3) compare the apparent height with the expected response of the different possible values to determine information about the spatial characteristics of the lattice structure; 4) is programmed to output the determined information about the spatial characteristics of the lattice structure.

Embodiments of both devices may include any of the features described above in connection with the corresponding method.

In another aspect, (i) determining one or more apparent characteristics of a test surface from an interferometric signal generated by an interferometric system; (ii) comparing the apparent characteristic determined from the interferometric signal with an expected response of the interferometric system for different possible values of one or more unanalyzed shapes of the test surface; (iii) based on the comparison, a method comprising outputting information about the one or more unanalyzed shapes of the test surface.

In addition to the features described above, embodiments of the method may include any of the following features.

The interferometric system may be a scanning interferometric system.

The apparent characteristics of the test surface can be determined from the interferometric signal based on a change in any of the interference phase, interference contrast, and surface reflectivity.

The expected response can be calculated for one or more changes in surface height and surface composition.

The test surface may comprise a patterned structure having components having modulation depth, periodicity, and width, and the expected response may be calculated for a change in one or more of the modulation depth, periodicity, and width of the component. And the expected response can be calculated for the change in modulation depth.

The correspondence between apparent modulation and actual modulation depth calculated for the expected response is a positive correlation over a first range of actual modulation depth and a second range of actual modulation depth. It may include negative correlation over.

The information about the one or more unanalyzed shapes may include one or more semiconductor processing steps, such as isolation patterning and etching, polysilicon gate electrode patterning and etching, source / drain etching and deposition, and metallization patterning, etching, and polishing processes. Can be used to monitor.

In another related aspect, a processor in a computer is configured to: 1) determine one or more apparent characteristics of a test surface from an interferometric signal generated by an interferometric system; 2) compare the apparent characteristic determined from the interferometric signal with an expected response of the interferometric system for different possible values of one or more unanalyzed shapes of the test surface; 3) An apparatus is disclosed that includes a computer readable medium having a program for outputting information about the one or more unanalyzed shapes of the test surface based on the comparison.

In another related aspect, an interferometric system configured to generate a plurality of interferometric signals corresponding to different surface locations of a test object; And an electronic processor coupled to the interferometric system to receive the interferometric signals, the electronic processor comprising: 1) determining one or more apparent characteristics of a test surface from the interferometric signals; 2) comparing the apparent characteristic determined from the interferometric signals with the expected response of the interferometric system to different possible values of one or more unanalyzed shapes of the test surface; 3) is programmed to output information about the one or more unanalyzed shapes of the test surface based on the comparison.

Embodiments of both devices may include any of the features described above in connection with the corresponding method.

In another aspect, (i) determining one or more apparent characteristics of a test surface from an interferometric signal (eg, a scanning interferometric signal) generated by an interferometric system; (ii) providing an expected response of the interferometric system to different possible values of the feature of the test surface (eg, a change in one or more of surface height and surface configuration), wherein the expected response is the test surface. Providing said contribution from the unanalyzed shape of; And iii) comparing the apparent characteristic determined from the interferometric signal with the expected response to the different possible values of the characteristics to improve the accuracy of the determined characteristics.

Embodiments of the method may include any of the following features.

The apparent characteristics of the test surface can be determined from the interferometric signal based on a change in any of the interference phase, interference contrast, and surface reflectance.

The expected response can be calculated for one or more changes in surface height and surface composition. For example, the test surface can include a patterned structure (eg, a grating) having components having modulation depth, periodicity, and width, wherein the expected response is the modulation depth, periodicity, and component width. Can be calculated for one or more changes.

The correspondence of the apparent characteristic with the interferometric signal and the actual value of the characteristic used to generate the model is such that a positive correlation over a first range of actual modulation depth and a negative correlation over a second range of actual modulation depth It may include. For example, in a particular embodiment, the expected response is calculated for changes in the modulation depth. In this case, the correspondence between the actual modulation depth and the apparent modulation calculated for the expected response may include a positive correlation over a first range of actual modulation depths and a negative correlation over a second range of actual modulation depths. Can be.

In other embodiments, the unanalyzed shape may be a single trench, step, or protrusion, rather than a series of components as in the patterned structure. In such examples, the information for the models from the actual signal may correspond to either the depth (or height in the case of steps or projections) and the location or width of the shapes.

The method may also include monitoring semiconductor processing steps using the improved accuracy of the measured properties of the test surface. For example, such process steps may include any of isolation patterning and etching, polysilicon gate electrode patterning and etching, source / drain etching and deposition, and metallization patterning, etching, and polishing processes.

In another aspect, a processor in a computer causes the processor to compare an apparent characteristic of a test object determined from an interferometric signal generated by an interferometric system with an expected response of the interferometric system to different possible values of the characteristics of the test surface. An apparatus is disclosed that includes a computer readable medium having a program for causing the expected response to include contributions from unanalyzed shapes of the test surface, and to determine the accuracy of the determined characteristics based on the comparison. To improve.

In another aspect, an interferometric system configured to generate an interferometric signal; And the expected response of the interferometric system to different possible values of the test surface, the apparent characteristics of the test surface being determined from the interfering signals generated by the interferometric system, the interferometric signals being connected to the interferometric system. An apparatus is disclosed that includes an electronic processor programmed to compare with and wherein the expected response includes contributions from unanalyzed shapes of the test surface and improves the accuracy of the properties determined based on the comparison.

Embodiments of such an apparatus may also include features corresponding to any of the features mentioned in connection with the corresponding method described above. In general, in another aspect, the present invention includes comparing information that can be generated from a scanning interferometry signal at a first surface location of a test object with information corresponding to a plurality of models for the test object, wherein the plurality of The model is characterized by a method in which the model is parameterized by a series of features of the test object.

Embodiments of the present invention may include any of the following features.

The method may further comprise determining an exact feature of the test object based on the comparison.

The method may further comprise determining a relative surface height of the first surface location based on the comparison. In addition, determining the relative surface height may include determining which model corresponds to an accurate feature of the feature of the test object based on the comparison, and corresponding to the exact feature to calculate the relative surface height. And using the model.

For example, using a model corresponding to the correct feature may include compensating data from the scan interferometry signal to reduce the contribution by the correct feature. Compensating the data may include removing a phase contribution by the exact feature from the phase component of the transformation of the scanning interferometry signal of the test object, using a model corresponding to the exact feature. The step may further comprise calculating the relative surface height from the phase component of the transformation after removing the phase contribution by the exact feature.

In another example, calculating the relative surface height using a model corresponding to the correct feature comprises: a peak in a correlation function used to compare the information of the test object with the information of the model corresponding to the correct feature. Determining the position of may include.

The method may further comprise comparing information that can be generated from the scanning interferometry signal for additional surface location with information corresponding to the plurality of models. In addition, the method may further comprise determining a surface height profile for the test object based on the comparison.

The comparing step may include calculating one or more merit fuctions indicating a similarity between the information that can be generated from the scan interferometry signal and the information corresponding to each of the models.

The comparing step may include fitting information generated from the scan interferometry signal with an expression for the information corresponding to the models.

The information corresponding to the plurality of models may include information about at least one amplitude component of a transform (eg, a Fourier transform) of a scanning interferometry signal corresponding to each of the models of the test object. Similarly, the information that can be generated from the scan interferometry signal includes information about at least one amplitude component of the transformation of the scan interferometry signal of the test object.

The comparing step includes comparing the relative intensity of the at least one amplitude component of the test object with the relative intensity of the at least one amplitude component of each of the model.

The information corresponding to the plurality of models may be a function of coordinates for the transformation. For example, the information corresponding to the plurality of models may include an amplitude profile of the transform of each of the models. The comparing may also include comparing an amplitude profile of the transformation of the scanning interferometry signal of the test object to each of the amplitude profiles of the model.

The comparing may also include comparing information in the phase profile of the transform of the scan interferometry signal of the test object with information in the phase profile of the transform of each of the models. For example, the information in the profile may include information about nonlinearity of the phase profile with respect to the transform coordinates and / or information about a phase gap value.

Information that can be generated and compared from the scanning interferometry signal can be a number. Alternatively, the information that can be generated and compared from the scan interferometry signal can be a function. For example, the information may be a function of scan position or a function of spatial frequency.

Information about the test object may be generated from the scan interferometry signal of the test object into a spatial frequency domain (eg, a Fourier transform). The information about the test object may include information about the amplitude profile of the transform and / or the phase profile of the transform.

The information about the test object may be related to the shape of the scanning interferometry signal for the test object at the first location. For example, the information about the test object may be related to fringe contrast magnitude in the shape of the scan interferometry signal. The information about the test object may also be related to the relative distance between zero-crossings in the shape of the scanning interferometry signal. Information about the test object can also be expressed as a function of scanning position, which function is obtained from the shape of the scanning interferometry signal.

The comparing may include calculating a correlation function (eg, a complex correlation function) between the information about the object and the information about each of the models. The comparing may further comprise determining one or more peak values in each of the correlation functions. The method may then further comprise determining an accurate feature for the test object based on parameterization of the model corresponding to the largest peak value. Alternatively or in addition, the method may further comprise determining a relative surface height for the test object at the first surface location based on the coordinates for at least one peak value of the correlation function. .

The plurality of models may correspond to a fixed surface height of the test object at the first position.

The series of features may comprise a series of values for at least one physical parameter for the test object. For example, the test object may comprise a thin film layer with a thickness, and the physical parameter may be the thickness of the thin film at a first location.

The series of features may include a series of features of the test object at a second surface location that is different from the first surface location. For example, the test object may include a structure at the second surface location that diffracts light to contribute to the scanning interferometric signal relative to the first surface location. In one example, the series of features at the second surface location may include permutations of the step height at the second location and the magnitude of the position of the second location. In another example, the series of features at the second surface location may include substitution of a modulation depth of the grating and an offset point of the grating, wherein the grating extends beyond the second location.

Further, the information that can be generated from the interferometric signal may correspond to an estimate of the relative surface height relative to the first surface location. For example, an estimate of the surface height relative to the first location may be based on a frequency domain analysis of the interferometric signal, or an estimate of the surface height relative to the first location is determined by a coherent signal in the interferometric signal. It may be based on the relative position of the highest peak.

The method further comprises comparing the information that can be generated from an interferometric signal for one or more additional surface locations including the second surface location of the test object with information corresponding to the plurality of models of the test object. It may include.

For example, information that can be generated from the interferometric signal for the first surface location and the additional surface location may include a surface of the test object with respect to a range of surface locations including the first surface location and the additional surface location. May correspond to a height profile. Further, the information corresponding to the plurality of models is used to generate the interferometric signal when the interferometric signal for each of the plurality of models of the test object is processed using a conventional processing method. And a surface height profile that is expected to be generated by an interferometric system, wherein the expected surface height profile is such that an interference signal of the first surface location is associated with test object shapes at the second surface location. Include contributions from unanalyzed shapes to include. For example, the test object may include a structure at the second surface location that diffracts light to contribute to the scanning interferometric signal relative to the first surface location.

In one example, the test object includes a patterned structure extending over the first surface location and the additional surface location, the information that can be generated from the scanning interferometry signal for the first surface location and the additional surface location. Includes an estimate for at least one of the modulation depth of the patterned structure, the periodicity of the patterned structure, and the width of each of the components of the patterned structure.

Similarly, as an example of the patterned structure, the series of features at the second surface location include the modulation depth of the patterned structure, the periodicity of the patterned structure, and the patterned structure at the first surface location. It may include different values for each of at least one of the widths of each of the components of.

In other embodiments, the unanalyzed shape may be a single trench, step, or protrusion rather than a series of components as in the patterned structure. In such examples, the information for the models from the actual signal may correspond to either the depth (or height in the case of the steps or projections) and the location or width of the shapes.

The information generateable from the interferometric signal may correspond to an estimate of the relative surface height of the first surface location, and the series of features at the second surface location may be relative to the relative surface height at the second surface location. For a series of values.

For example, the test object may comprise a patterned structure spanning the first surface location and the second surface location, wherein an estimate of the relative surface height of the first surface location is a modulation of the patterned structure. And a series of values for the relative surface position at the second surface position correspond to the modulation depth of the patterned structure. The plurality of models may correlate different estimates of the modulation depth to corresponding ones of the values.

In some embodiments, the plurality of models may correlate information that can be generated from the interferometric signal to a corresponding value of the information for the plurality of models, wherein the correspondence is negative in a positive correlation. Change until correlation

For example, in the case of the patterned structure, the correlation between the different estimates for the modulation depth and the corresponding values for the modulation depth from the models varies from positive to negative correlation. .

The method may further include determining an exact feature of the test object, such as relative surface height of the first surface location, based on the comparison. For example, determining the relative surface height may include determining which model corresponds to the correct one of the features of the test object based on the comparison, and determining the relative surface height to determine the relative surface height. Using the corresponding model. The series of features may be a series of surface materials for the test object.

The series of features may be a series of surface layer configurations for the test object.

The scan interferometry signal may be generated by a scan interferometry system, and the comparing includes considering systematic contributions to the scan interferometry signal generated from the scan interferometry system. can do. For example, the systematic contribution may include information regarding dispersion in phase shift of reflections from components of the scanning interferometry system. The method may also include comparing information generateable from the scanning interferometry signal for additional surface locations with information corresponding to the plurality of models, in which case the systematic contribution is the surface locations It can be broken down into multiple contributions to. The method may further comprise calibrating the systematic contribution of the scanning interferometry system using another test object having known properties.

The scanning interferometry signal is configured to image test light from the test object to cause interference with reference light on a detector, and from the common source to the detector between the interference portions of the test light and the interference light. By varying the optical path length difference of the test light and the reference light can be obtained from the common source (e.g., spatially extending source), and the scanning interferometric signal is obtained by the optical path length It corresponds to the interference intensity measured by the detector when the difference is changed.

The test light and the reference light may have a spectral bandwidth greater than 5% of a center frequency of the test light and the reference light.

The common source may have a spectral coherence length, and the optical path length difference is varied over a wider range than the spectral coherence length to produce the scan interferometry signal.

Optics used to deliver test light onto a test object to form an image on the detector can set the numerical aperture for the test light to about 0.8.

The method may also further comprise generating a scan interferometry signal.

In another aspect, the invention provides a program that allows a processor in a computer to compare information that can be generated from a scanning interferometry signal for a first surface location of a test object with information corresponding to a plurality of models for the test object. And a plurality of models are parameterized by a series of properties of the test object.

The apparatus may include any of the features described above in connection with the method.

In another aspect, the present invention provides a scanning interferometry system configured to generate a scanning interferometry signal; And compare information that is coupled to the scanning interferometry system to receive the scanning interferometry signal and that is generateable from the scanning interferometry signal for a first surface location of a test object with information corresponding to a plurality of models for the test object. It is characterized by an apparatus comprising a programmed electronic processor, wherein the plurality of models are parameterized by a series of characteristics of the test object.

The apparatus may include any of the features described above in connection with the method.

In general, in another aspect, the present invention provides a method of chemical mechanical polishing of a test object; Collecting scanning interferometry data for surface topography of the test object; And adjusting the process conditions of the chemical mechanical polishing of the test object based on information obtained from the scanning interferometry signal. For example, the process conditions can be pad pressure, and / or polishing slurry configuration. In a preferred embodiment, adjusting the process conditions based on information obtained from the scan interferometry signal comprises: generating information from the scan interferometry signal for at least a first surface location of a test object, wherein Comparing the information corresponding to the model, wherein the plurality of models are parameterized by a series of characteristics for the test object. The analysis of the scan interferometric signal may further comprise any of the features described in the first mentioned method.

Unless defined otherwise, all technical and scientific terms herein have the same meaning as commonly known to one of ordinary skill in the art to which this invention belongs. In case of conflict with the foregoing publications, patent applications, patents, and other references incorporated herein by reference, the present specification will include definitions.

For example, most often the scanning interferometer involves mechanically scanning the relative optical path length between the reference interval and the measurement interval, but as used herein, the scanning interferometer adds an additional ratio to change the relative optical path length. It is intended to include non-mechanical means. For example, an interferometric signal can be generated by varying the center frequency of the light source over a wavelength range of a floating path length interferometer (ie, the nominal optical path lengths of the reference path and the nominal path of the measurement path). . The different wavelengths have different optical path lengths, resulting in different phase shifts for the reference path and the measurement path, thus changing the phase shifts between the paths.

Further, the term "light" may sometimes be understood to be limited to electromagnetic radiation in the visible spectrum, but as used herein, the term "light" refers to an ultraviolet spectral region, a visible spectral region, a near infrared spectral region. It is intended to include any of, and infrared spectrum regions.

Other features, objects, and advantages of the invention will be apparent from the following detailed description.

1 is a flowchart of an interferometry method.

FIG. 2 is a flowchart illustrating a modification of the interferometric method of FIG. 1.

3 is a schematic diagram of a Linnik-type scanning interferometer.

4 is a schematic diagram of a Mirau-type scanning interferometer.

5 shows the illumination of the test sample through the objective lens.

6 shows the theoretical Fourier amplitude spectrum for scanning interferometric data within two limits.

Figure 7 shows two surface types, with and without a thin film.

FIG. 8 shows a merit function search procedure for simulating a Si0 ₂ film on a Si substrate with thin film thickness of zero.

9 shows a merit function search procedure for simulating a Si0 ₂ film on a Si substrate with a thin film thickness of 50 nm.

10 shows a merit function search procedure for simulating a Si0 ₂ film on a Si substrate with a thin film thickness of 100 nm.

11 shows a merit function search procedure for simulating a Si0 ₂ film on a Si substrate with a thin film thickness of 300 nm.

12 shows a merit function search procedure for simulating a Si0 ₂ film on a Si substrate with a thin film thickness of 600 nm.

13 shows a merit function search procedure for simulating a Si0 ₂ film on a Si substrate with a thin film thickness of 1200 nm.

FIG. 14 shows the surface and substrate profiles determined for the simulation of Si0 ₂ on a Si thin film whose top surface is always 0 (zero), the film thickness constantly varying from 0 to 1500 nm in 10 nm increments per pixel.

FIG. 15 shows the surface substrate profile determined for the same simulation as FIG. 14 except for the addition of random noise (2 bits rms out of the average 128 intensity bits).

FIG. 16 is a library disclosed herein for a surface height profile (FIG. 16 (a)) determined using a conventional FDA assay and 2400 line gratings per mm with an actual peak-to-valley modulation depth of 120 nm. The search method (Fig. 16 (b)) is shown.

FIG. 17 illustrates distortions caused by unanalyzed step heights in the scanning interferometry signal of pixels corresponding to various surface locations near the step height.

FIG. 18 shows nonlinear distortion in the frequency domain phase spectrum for pixels corresponding to the left (FIG. 18 (a)) and right (FIG. 18 (b)) surface positions of the unanalyzed step height of FIG. 17.

FIG. 19 shows the library search method disclosed herein (FIG. 19 (b)) for surface height profiles (FIG. 19 (a)) determined using conventional FDA assays, and step heights that were not analyzed.

20 shows the actual scanning interferometry signal of the base Si substrate without the thin film.

21 and 22 show interference template patterns for a bare Si substrate and a thin film structure with 1 micron SiO ₂ on Si, respectively.

23 and 24 show the merit function as a function of the scan position for the template function of FIGS. 21 and 22, respectively.

25 is a schematic diagram of an object having an unanalyzed surface shape (lattice pattern).

FIG. 26 is a graph of the predicted surface profile for the grating pattern of FIG. 25 based on conventional interferometer analysis. FIG.

FIG. 27 is a graph of the apparent modulation depth of the grating of FIG. 25 as a function of the actual modulation depth of the grating in conventional interferometer analysis.

FIG. 28 is a graph showing the theoretical and experimental instrument transfer function of a typical white light scanning interferometer.

29A-29C are schematic diagrams of one embodiment and grating structure of the unanalyzed surface measurement techniques disclosed herein.

30A is a schematic representation of the model structure of five pure Si lattice lines (without upper layer) with a width W = 120 nm and a pitch L = 320 nm.

Fig. 30B is a diagram showing the rigorous combined wave analysis (RCWA) of the scanning interferometry signal (z direction) for the model structure. This is for y polarization parallel to the line and is more sensitive to the top of the line than to the top of the area.

31A and 31B are diagrams showing x-polarized light and y-polarized light of the scanning interferometry signal with respect to the center pixel of FIG. 30A, respectively.

32A and 32B show surface height profiles obtained from FDA analysis of the signal of FIG. 30A for x polarization (orthogonal to grid line) and y polarization (parallel to grid line), respectively.

33A and 33B show apparent etch depths E obtained from simulated scanning interferometry data, which is a function of different actual etch depths E for x polarization (orthogonal to grid line) and y polarization (parallel to grid line), respectively. RCWA analysis of '= H'-E. E '= E line is for reference only.

34A and 34B are graphs showing signal intensities corresponding to different actual etch depths E for x polarization (orthogonal to the grid line) and y polarization (parallel to the grid line), respectively.

FIG. 35 is another graph for the same data as in FIG. 33B, showing the measurement bias or offset E−E ′ as a function of the measured step height −E ′ of the silicon grating.

FIG. 36 is a graph showing the RCWA prediction of measured etch depth E '= H'-E as a function of the actual etch depth E of the silicon grating, using circular polarization, compared to a one-to-one corresponding line.

37A and 37B show the measured (apparent) etch depth E '= H'-E, for etch depth E = 100 nm and pitch L = 320 nm, respectively, for a 5-line silicon grating for y polarization and x polarization. Is a graph showing the RCWA prediction as a function of shape width W. In both graphs the solid mark is the reference etch depth.

38 is a schematic diagram of an interferometric measurement system illustrating a method for automatically controlling various components of an interferometric system.

 Like reference numerals in the different drawings indicate common elements.

1 is a flow diagram illustrating an embodiment of the invention as a whole performing analysis of scanning interferometric data in the spatial frequency domain.

Referring to FIG. 1, to measure data from the test object surface, an interferometer is used to mechanically or electro-optically scan the optical path length difference (OPD) between the reference path and the measurement path. The measuring path goes straight to the object surface. The optical path length difference at the start of scanning is a function of the local height of the object surface. The computer records the interference intensity signal while scanning the optical path length difference for each of the plurality of camera pixels corresponding to different surface positions of the object surface. After storing the interference intensity signal, which is a function of the optical path length differential scanning position for each of the different surface locations, the computer performs a transform (eg, a Fourier transform) to generate a frequency domain spectrum of that signal. The frequency domain spectrum includes both magnitude and phase information as a function of the spatial frequency of the signal within the scanning dimension. For example, suitable frequency domain analysis methods for generating spectra, etc., are owned by Peter de Groot and are entitled "Methods and Apparatuses for Measuring Surface Topography by Spatial Frequency Analysis of Interfering Images (Method) and Apparatus for Surface Topography Measurements by Spatial-Frequency Analysis of Interferegrams, "US Pat. No. 5,398,113, the entire disclosure of which is incorporated herein by reference.

In a separate step, the computer generates a library of theoretical predictions for the frequency domain spectrum of the various surface parameters and the model of the interferometer. These spectra extend, for example, to the range of possible thin film thicknesses, surface materials and surface textures. In a preferred embodiment, the computer generates a library spectrum for a constant surface height (eg, height = 0). Thus, in this embodiment, the library does not contain information about surface topography, but the surface structure when generating distinctive features of the frequency domain system and the type of interaction of the surface structure, the optical system Include only information relevant to lighting, detection and detection systems. Alternatively, prediction libraries can be empirically generated using sample artifacts. Alternatively, the prediction library can be used to reduce the number of unknown surface parameters, such as information from previous further measurements of the object surface, provided by other equipment, such as ellipsometers, and known properties of the object surface. The user's other inputs can be used. Any of these techniques for augmented theories by library generation, theoretical modeling, empirical data, or additional measurements, to generate intermediate values in real time as part of library generation or during library searches. It can be extended by interpolation.

In the next step, a library search is used to compare the experimental data with the prediction library. In the example where the thickness of the film is unknown, a library of single surface type (eg, SiO ₂ on Si) will span many possible film thickness ranges where the top surface height is always zero. In another example, the adjustable parameter would be surface roughness, which can be a roughness depth and / or spatial frequency. Library searches may be performed, for example, with changes in magnitude as a function of spatial frequency in a monochromatic high numerical aperture system related to the mean value of the magnitude spectrum related to the total reflectance of the surface or to the scattering angle of the reflected light. It matches the characteristics of the FDA spectrum independent of the surface height.

This analysis may also include system characterization. System characterization can be used to determine parameters such as efficiency, dispersion, and system wavefront error that cannot be included in the theoretical model, for example, of one or more reference workpieces with known surface structures and surface topography. Include measurement

In addition, this analysis may include a full calibration. This calibration is used to determine the calibration between measured surface parameters, such as film thickness as determined by library search, and the value of these parameters as determined independently, for example, by an ellipsometry assay. For example, measurement of one or more reference workpieces.

Based on the comparison of the experimental data and the prediction library, the computer identifies the surface model corresponding to the best match. The surface parameter results can then be displayed or transmitted numerically or graphically to the user or to a host computer for further analysis or data storage. Using the surface parameter results, the computer can then determine the features identified by the library search in addition to the surface height information. In some embodiments, the computer produces a compensated phase spectrum, for example by subtracting the corresponding theoretical phase spectrum directly from the experimental phase spectrum. The computer then determines the local surface height for one or more surface points by analysis of the compensated phase as a function of spatial frequency, for example by analysis of the coefficients produced by a linear fit. Thereafter, the computer generates a completed three-dimensional image consisting of the height data and the corresponding image plane coordinates, along with a graphical or numerical representation of the surface features as determined by library search.

In some cases, library searches and data collection may be repeated to further improve results. In detail, the library search may be subdivided on a pixel-by-pixel or regional basis. For example, if a preliminary library search found that the surface had a thin film of approximately 1 micron, the computer could generate a fine-grain library of exemplary values close to 1 micron to further refine the search. have.

In other embodiments, the user may only be interested in the surface features modeled by the prediction library, not the surface height, in which case the step of determining the surface height is not performed. Conversely, a user may only be interested in the surface height, not the surface features modeled by the prediction library, in which case the computer compensates the experimental data for the contributions of the surface features using the comparison of the test data and the prediction library. However, the surface height is more accurately determined, but there is no need to explicitly determine or display the surface features.

This assay is a simple thin film (in this case, for example, an important variable parameter can be the film thickness, the reflectance of the film, the reflectance of the substrate, or some combination thereof); Multilayer thin film; Sharp edge and surface shapes that produce diffraction or other complex interference effects; Unresolved surface roughness; Unanalyzed surface shapes such as, for example, grooves of sub-wavelength width on other smooth surfaces; Heterogeneous materials (eg, the surface may include thin films and metals, in which case the library may include both automatic identification of the film or solid metal against the frequency structure spectrum corresponding to the surface structure type). ); Optical activity such as fluorescence; Spectral characteristics of the surface, such as color and wavelength dependent reflectance; Polarization dependent properties of the surface; Surface deflection, vibration, or movement or deformation of the surface resulting in perturbation of the interference signal; And may be applied to a variety of surface analysis problems, including, for example, data distortions associated with data collection procedures such as data acquisition windows that do not completely contain interference intensity data.

The interferometer includes: a spectrally narrow-band light source having a high numerical aperture (NA) objective lens; Spectral broadband light sources; The combination of a high numerical aperture objective and a spectral broadband light source; Interference microscope objective lens; Oil / water immersion and solid immersion types, for example in the Michelson, Mirau or Linnik geometries; Series of measurements at multiple wavelengths; Unpolarized light; And any of the features, such as polarized light, including linear, circular or structured. For example, organized polarized light can be used to reveal polarization dependent optical effects that can be attributed to surface features, such as polarization masks that produce different polarizations for different segments of illumination or imaging pupils. ) May be included. The interferometer may also include calibration of the entire system as described above.

When comparing theoretical and experimental data, the library search includes the product of the average magnitude and the average phase, the average magnitude itself, and the average phase itself, or the difference between them, or the product of the magnitude and / or phase data of the frequency spectrum, or car; Slope, width, and / or height of the size spectrum; Interference contrast; Data in the frequency spectrum at DC or zero spatial frequency; Nonlinearity or shape of the magnitude spectrum; Zero frequency intercept of phase; Nonlinearity or shape of the phase spectrum; And any combination of these criteria. Note that magnitude and amplitude may be used interchangeably as used herein.

FIG. 2 is a flow diagram generally describing another embodiment of an analysis of scanning interferometric data. FIG. This analysis method is similar to that described for FIG. 1 except that the comparison of experimental data and prediction library is based on information in the scan coordinate area. The experimental signal may be characterized by a quasi-periodic carrier modulated with amplitude by an envelope function for scan coordinates. When comparing the theoretical and experimental signals, the library search includes the average signal strength; The shape of the signal envelope, including deviations from some ideal or reference shapes such as, for example, Gaussian; Phase of the carrier signal with respect to the envelope function; The relative spacing of zero crossings and / or signal maximums and minimums; After adjustment of the optimal relative scan position, the peak value of the correlation between the library and the measured signals; And any combination of these criteria.

Also in further embodiments, information compared to library models may be obtained from experimental signals at a plurality of surface locations. This may be particularly useful if the test object includes an unanalyzed surface shape because experimental information compared to library models may correspond to the collective surface response of the test surface in interferometric measurement. . For example, the information obtained from the experimental signals at the plurality of surface positions may be a surface profile of a test surface obtained from a conventional method of processing an interferometric signal, or information obtained from a surface profile. This treatment method can only obtain the apparent properties of the test surface since the unanalyzed shapes will not be apparent. Nevertheless, the unanalyzed shapes may be correlated with more accurate information about the unanalyzed shapes by comparing the apparent properties with the corresponding models for the test object parameterized by the values characterizing the unanalyzed shapes. It can leave a signature within the apparent surface profile that can.

In addition, the test object may also include certain reference structures that may be involved in the comparison of apparent shapes and library models. For example, when comparing the apparent surface height profile of an etched grating pattern in which individual lines are not analyzed with several models, the portion of the test object that is known to not be etched is compared to the apparent surface height in the grating portion of the test object. A reference point can be provided.

Thus, in certain embodiments, an interferometric profiler having both a measuring beam and a reference beam, such as a wideband or low coherence interferometer, may not be as complex or analyzed as can be found on a patterned semiconductor wafer. Used to measure the characteristics of surface structures. This profiler interprets the change in interference phase, contrast and / or surface reflectance as the measured apparent surface height change. In a separate step, the data processing means calculates the expected response of the profiler for possible changes in the actual surface height and / or surface composition of the surface, including the unanalyzed surface structure. The data processing means then compares the true surface characteristics by comparing the measured apparent surface height with the expected response of the interference profile to possible changes in the actual surface height, unanalyzed surface structure, and / or surface composition. Decide

Before acquiring, during, or after acquiring the data, the data processor may determine the profile of the instrument for possible changes in the actual surface height and / or surface composition of the surface, including the unstructured surface structure. Calculate the expected response. 25 is an example of an object having a surface shape that was not analyzed for visible wavelength (400-700 nm wavelength) interferometry. Specifically, FIG. 25 shows a device array (also referred to herein as a patterned structure or a grating structure). Unanalyzed surface shapes have a height H, a spacing l, and a width d on adjacent surfaces S. The height H is also referred to herein as the modulation depth of the patterned structure. "Unresolved" means that the individual shapes are not fully distinguished in the surface profile image, so that the individual shapes have an incorrect profile and / or have a height H because the instrument's lateral resolution is limited. Is incorrect.

FIG. 26 is a visible wavelength interferometer (560 nm center frequency, 110 nm bandwidth full width half maximum (FWHM)) using the Railleigh Hypothesis technique (described below) and an objective lens numerical aperture ( When NA) is 0.8, it represents the predicted response to the unanalyzed shapes of the object of FIG. 25. This surface structure is unanalyzed due to the measured apparent surface profile that does not resemble the actual surface structure at all. For this calculation, height H = 200 nm, spacing l = 200 nm, and width d = 120 nm. FIG. 27 shows the unanalyzed measured profile of FIG. 26 as a function of the actual height H of the shapes. It is noteworthy that the relationship between measured height and true height is complex and even has a negative correlation above 40 nm. This negative correlation can be explained by the difficulty in coupling light into a narrow, sub-wavelength trench.

After data collection and calculation of the expected system response, the data processor compares the measured apparent surface height with the expected response of the device of the present invention for possible changes in actual surface height, unanalyzed surface structure and / or surface composition. Determine true surface characteristics. The following example of FIGS. 25-27 is as shown in FIG. 27, determining the relationship between the actual height H and the measured apparent height, and using the information known in this relationship to determine the actual height from the measurement results. Steps. Other modalities include determining width d and spacing l.

As will be described further below, polarizing the objective lens can also adjust the sensitivity range by changing the illumination wavelength using another light source, thereby improving sensitivity to specific parameters such as etch depth.

Many processing techniques can be used to extract an apparent characteristic or properties (eg, an apparent surface profile) from scanning interferometric data. For example, the prior art includes the identification of a position corresponding to the highest or center of the fringe contrast envelope for each pixel or the use of frequency domain analysis (FDA) for each pixel, and a wavelength directly proportional to the surface height. It is associated with the rate of change of phase with (see, eg, US Pat. No. 5,398,133). METHODS AND SYSTEMS FOR INTERFEROMETRIC ANALYSIS OF SURFACES, filed by Peter de Groot, which is hereby incorporated by reference, and entitled "Methods and Systems for Interferometric Analysis of Surfaces and Related Applications." AND RELATED APPLICATIONS ", co-owned by US Patent Publication No. 2005-0078318 Al, or Peter de Groot, filed May 18, 2006, and entitled" Information on Thin Film Structure. " METHOD AND SYSTEM FOR ANALYZING LOW-COHERENCE INTERFEROMETRY SIGNALS FOR INFORMATION ABOUT THIN FILM STRUCTURES, "Interference Measurement, It is also possible to use more advanced processing techniques that attempt to remove thin film effects from the signal. Of course, the model library used for comparison should take into account the processing techniques used to extract experimentally obtained information. The following provides a detailed mathematical description and examples of the analysis. First, an example of a scanning interferometer will be described. Second, a mathematical model of the scan interferometric data is determined. Third, describe the optical properties of the surface and describe how to use this information to generate accurate models of scanning interferometry data for different surface features. Fourth, we describe how experimental interference data can be compared with prediction libraries to provide information about test objects. First, thin film applications are described, and later, other composite surface structures, particularly step heights and grating patterns that are not optically analyzed, are described. It also focuses initially on analysis in the spatial frequency domain and later on analysis in the scan coordinate domain. Subsequently, further examples of techniques for extracting information about the unresolved surface shape from the interferometric signal as in the example of FIGS. 25 to 27 are described.

3 shows a scanning interferometer of the linic type. Illumination light 102 from a source (not shown) is partially transmitted by beam splitter 104 to define a reference light 106, and is partially reflected by beam splitter 104 to measure measurement light ( measurement light) 108. The measurement light is focused on the test sample 112 (eg, a sample comprising a single thin film layer or multiple thin film layers of one or more heterogeneous materials) by the measuring objective lens 110. Similarly, the reference light is focused on the reference mirror 116 by the reference objective lens 114. Preferably the measurement objective lens and the reference objective lens have common optical properties (eg matched numerical aperture). The measurement light reflected (or scattered or diffracted) from the test sample 112 propagates back through the measuring objective lens 110, transmitted by the beam splitter 104, and on the detector 120 by the imaging lens 118. It is formed. Similarly, the reference light reflected from the reference mirror 116 propagates back through the reference objective lens 114, is reflected by the beam splitter 104, and is imaged on the detector 120 by the imaging lens 118, where the measurement is made. Cause interference with light.

For brevity, FIG. 3 shows that the measurement light and the reference light are focused at specific points on the test sample and the reference mirror, respectively, and then interfere at corresponding points on the detector. This light corresponds to the portion of the illumination light that propagates perpendicular to the pupil planes of the measurement interval and reference interval of the interferometer. The other parts of the illumination light eventually illuminate the other parts of the test sample and the reference mirror, and then are imaged at the corresponding points on the detector. In FIG. 3, this is shown by dashed line 122 and corresponds to chief rays coming from different points on the test sample, which are imaged at corresponding points on the detector. The principal rays intersect at the center of the pupil plane 124 of the measurement zone, which is the back focal plane of the measurement objective lens 110. Light exiting the test sample at an angle different from that of the main beam intersects at different locations on the pupil plane 124.

In a preferred embodiment, the detector 120 comprises a plurality of devices for independently measuring the interference between the measurement light and the reference light corresponding to different points on the test sample and the reference mirror (i.e. to provide spatial resolution of the interference pattern). (For example, a plurality of pixels).

Scanning stage 126 associated with test sample 112 scans the position of the test sample relative to measurement objective lens 110, as indicated by scan coordinate ζ in FIG. 3. For example, the scanning stage can be based on a piezoelectric transducer (PZT). The detector 120 measures the intensity of optical interference at one or more pixels of the detector when the relative position of the test sample is scanned and transmits the information to the computer 128 for analysis.

Since the scanning takes place in the region where the measurement light is focused on the test sample, the scanning causes the optical path of the measurement light from the source to the detector to be measured by the angle of the measurement light incident on the test sample and the angle of the measurement light exiting the test sample. Change it differently. As a result, the optical path length difference OPD from the source to the detector between the interference portions of the measurement light and the reference light differs in scan coordinate ζ depending on the angle of the measurement light incident on the test sample and the angle of the measurement light coming out of the test sample. In another embodiment of the present invention, the same result can be obtained by scanning the position of the reference mirror relative to the reference objective lens 114 instead of scanning the test sample 112 against the measurement objective lens 110.

This difference in the method of varying the optical path length difference (OPD) according to the scan coordinate ζ introduces a limited coherence length of the interference signal measured at each pixel of the detector.

For example, an interference signal that is a function of scan coordinates is typically modulated by an envelope having a spatial coherence length of approximately λ / 2 (NA) ² , where λ is the nominal (nominal) wavelength of the illumination light, and NA Is the numerical aperture of the measurement objective lens and the reference objective lens. As further described below, the modulation of the interfering signal provides angle dependent information regarding the reflectance of the test sample. In order to increase the limited spatial coherence, it is desirable to set the objective lenses of the scanning interferometer to a large numerical aperture, for example, about 0.7 or more (or more preferably about 0.8 or more or about 0.9 or more). The interfering signal can also be modulated by the limited time correlation length associated with the spectral bandwidth of the illumination source. Depending on the configuration of the interferometer, one or the other of these limited coherence length effects may take the lead, or they may all contribute substantially to the overall coherence length.

Another example of a scanning interferometer is the Mirau type interferometer shown in FIG.

Referring to FIG. 4, source module 205 provides illumination light 206 to beam splitter 208 and delivers illumination light to Mirae interference objective lens assembly 210. Assembly 210 includes an objective lens 211, a reference flat 212 with a reflective coating on a small central portion defining a reference mirror 215, and a beam splitter 213. In operation, objective lens 211 focuses illumination light to test sample 220 through reference plane 212. The beam splitter 213 reflects the first portion of the focused light to the reference mirror 215 to define the reference light 222, and transmits the second portion of the focused light to the test sample 220 to measure the measurement light ( 224). The beam splitter 213 then recombines the measurement light reflected (or scattered) from the test sample 220 with the reference light reflected from the reference mirror 215, and the objective lens 211 and the imaging lens 230 The combined light is imaged to cause interference on detector (eg, multi-pixel camera) 240. As in the system of FIG. 3, the measurement signal (s) from the detector are sent to a computer (not shown).

Scanning in the embodiment of FIG. 4 includes a piezoelectric transducer (PZT) 260 coupled to the Mirau interference objective lens assembly 210, which scans the interferometric data I (ζ, h) at each pixel of the camera. And to scan the assembly 210 as a whole relative to the test sample 220 along the optical axis of the objective lens 211 to provide. Alternatively, the piezoelectric transducer PZT may be connected to the test sample, not the assembly 210, to provide relative movement there between, as represented by the piezoelectric transducer actuator 270. In another embodiment, scanning may be provided by the movement of one or both of the reference mirror 215 and the beam splitter 213 relative to the objective lens 211 along the optical axis of the objective lens 211.

The source module 205 includes a source 201 that is spatially expanded, a telescope formed of the lenses 202 and 203, and an aperture located at the front focusing surface of the lens 202 (which coincides with the rear focusing surface of the lens 203). (stop) 204. The device forms a spatially expanding source at the pupil plane 245 of the Mirae interference objective lens assembly 210, which is an example of Koehler imaging. The size of the aperture controls the size of the illumination field on the test sample 220. The system may also include an aperture stop (not shown) positioned between the beam splitter 208 and the interference objective lens assembly 210. In another embodiment, the source module includes an apparatus for imaging the spatially expanding source directly on the test sample, which is known as critical imaging. Any type of source module can be used with the linic type scanning interferometry system of FIG.

In another embodiment of the present invention, a scanning interferometry system can be used to determine angle dependent scattering or diffraction information, ie scatterometry, for a test sample. For example, a scanning interferometry system examines a test sample with test incidence (eg, substantially vertical incidence or otherwise collimated) over only very narrow incidence angles that can be scattered or diffracted by the test sample. It can be used to Light from the sample is imaged in the camera to interfere with the reference light as described above. The spatial frequency of each component of the scanning interferometry signal will vary with the angle of the test light coming from the test sample. Thus, the Fourier transform following vertical scanning (i.e., scanning along the optical axis of the objective lens) can measure diffracted and / or scattered light as a function of the exiting angle without directly accessing or imaging the rear focusing plane of the objective lens. To make it possible. In order to provide substantially perpendicular incident illumination, for example, the source module may be configured to form a point source on the pupil plane, or otherwise reduce the extent to which the illumination light fills the numerical aperture of the measuring objective lens. Can be configured. Scatterometry techniques may be useful for analyzing the discrete structure of a sample surface, such as grating lines, edges, or general surface roughness that diffracts and / or scatters light at greater angles.

In many of the analyses herein, it is assumed that the polarization state of light at the pupil plane is random, i.e., made up of approximately the same amount as both s-polarized light (orthogonal to the incidence plane) and p-polarized light (orthogonal to the incidence plane). . As can be realized using a radiating polarizer located at the pupil plane (e.g., the rear focusing surface of the measuring objective lens in the case of a nic interferometer, and the rear focusing surface of the common objective lens to the Mirau interferometer). Other polarizations are also possible, including pure s polarization. Other possible polarizations include radiated p polarization, circular polarization, and modulated polarization (eg, two states where one state follows another) for ellipsometric measurements. In other words, the optical properties of the test sample can be analyzed not only for angle or wavelength dependence, but also for polarization dependence or selected polarization. This information can also be used to improve the accuracy of thin film structure characterization.

For such elliptical polarization measurements, the scanning interferometry system may include fixed or variable polarizers in the pupil plane. Referring again to FIG. 4, a Mirau-type interferometry system includes, for example, a polarization optic in the pupil plane to select the desired polarization for light incident on the test sample and light exiting the test sample. ) 280. In addition, polarizing optics can be reconfigured to change the selected polarization. The polarizing optics can include one or more elements, including a polarizer, a wave plate, an apodization aperture, and / or a modulation element for selecting a given polarization. In addition, the polarizing optics can be fixed, organized or reconstructed to produce data similar to the data of an elliptical polarimeter. For example, a first measurement using a pupil polarized in the radial direction for s polarization involves a pupil polarized in the radial direction for p polarization. In another example, use an apodized pupil plane with a linearly polarized light that can rotate within the pupil plane to direct any desired linear polarization state to the object, eg, slit or wedge shaped. Or a reconfigurable screen such as a liquid crystal display may be used.

In addition, polarizing optics can provide variable polarization across the pupil plane (eg, by including multiple polarizers or spatial modulators). Thus, for example, by providing different polarizations for angles of incidence greater than shallow angles, it is possible to " tag " the polarization state according to spatial frequency.

In yet another embodiment, the selectable polarization can be combined with a phase shift as a function of polarization. For example, polarizing optics, which may include linear polarizers, are located in the pupil plane and carry two waveplates (eg, eight waveplates) in opposing quadrant pupil planes. Linear polarization results in a full range of polarization angles with respect to the plane of incidence of the objective lens. For example, if the waveplate is aligned such that the s polarized light has a predominantly fixed phase shift, both the radiated s polarized light and the p polarized light are provided simultaneously, but in phase with respect to each other, for example π (pi) Moved so that the interferometer effectively detects the difference between these two polarization states as the fundamental signal.

In other embodiments, the polarizing optics may be located elsewhere in the device. For example, linearly polarized light can be obtained anywhere in the system.

The physical model of the scan interferometric signal will now be described.

The object surface has a height feature h to be profiled over the area indicated by the coordinates x, y. The stage provides a smooth and continuous scan ζ of either the interfering objective lens or the object itself as shown. During scanning, the computer records the intensity data I _{ζ, h} for each of the camera pixels or image points of successive camera frames. Note that the major dependence and surface height of the intensities I _{ζ and h} on the scan position are indicated by subscripts, and this notation is used throughout the specification.

Appropriate physical models of optics, taking into account the interplay of partial coherence of light sources, polarization mixing in interferometers, imaging properties of high numerical aperture objectives, and the presence of large incidence field vectors and discrete surface shapes Can be very complex.

For example, in order to strictly predict an interferometer signal from a given structure, it is necessary to solve the Maxwell's Equation for that structure. White light interferometry requires a set of sufficiently dense wavelengths to cover the bandwidth of the illumination. There are many approaches to solving Maxwell's equations in 2D. In 2D, a particularly convenient approach is based on the rail hypothesis. In this approach, the structure is treated as a thin film stack, but with an interface between each layer in the stack with a particular topography. The rail hypothesis is said to be able to extend in plane waves where the electromagnetic field propagates up and down in each layer, and solutions can be found by selecting coefficients so that the electromagnetic fields meet standard boundary conditions at the interface. This approach is easy to implement, fairly fast, and generates a scattering matrix or full optical transfer function of the surface for one polarization at one wavelength in just one calculation. This approach is limited to obtaining a valid solution only if the topography of each interface is applied less than approximately half wavelength in the layers representing the interface boundary. Thus, high index materials, such as silicon, make topography at visible wavelengths much smaller than 100 nm.

3D modeling techniques include finite difference time domain (FDTD), finite element (Finite Element), and rigorous coupled wave analysis (RCWA). For example, M. G. M. G. Moharam and T. K. Paper of T. K. Gaylord ["Diffraction analysis of dielectric surface-relief gratings" J. Opt. Soc. Am., 72, 1385-1392, (1982), and M. M. Totzeck's paper ["Numerical simulation of high-NA quantitative polarization microscopy and corresponding near-fields", Optik, 112 (2001) 381- [390]. In addition, the Institute of Technical Optics (ITO) at Stuttgart University, Based on the work of M. Totzeck, he developed software to perform RCWA called Microsim. These techniques are powerful, although often limited to a small amount, ie, a few wavelengths on one side, to maintain both proper execution time and memory requirements. Nevertheless, because these techniques can be used prior to creating a suitable library, the large amount of computation time used to generate the library does not prevent the application of the techniques disclosed herein in the process.

For convenience and to illustrate certain aspects of the present invention, the model is simplified by assuming an extended source of random polarization and diffusion, low coherence. The modeling of the interfering signal is simplified by adding the contribution of all ray bundles that pass through the pupil plane of the objective lens and reflect from the object surface at the angle of incidence ψ as shown in FIG. 5.

The contribution of interference of a single ray bundle through the optical system is proportional to the following equation:

In the above formula, Z _{β, k} is the effective object intensity reflectance including the effect of the beam splitter, for example, R _{β, k} is the effective reference reflectance including both the beam splitter and the reference mirror. The refractive index of the ambient medium is n ₀ , and the directional cosine for the angle of incidence (ψ)

The frequency of the source light is

to be.

The sign convention of phases results in an increase in surface height corresponding to a positive change in phase. The phase term has contributions ω _{β, k} for the object path in the interferometer including the thin film effect from the object surface and contributions ν _{β, k} for the reference path including the reference mirror and other optics in the objective lens.

The total interfering signal integrated through the pupil plane is proportional to

Where U _β is the light distribution of the pupil plane and V _k is the light spectral distribution. The weighting factor β in equation (4) follows the cos (ψ) term due to the projection angle, and the sin (ψ) term for the diameter of the annulus with the width ψ of the pupil plane is given by 5) Follows:

Here, it is assumed that the objective lens follows the Abbe sine condition as shown in FIG. 5. This fairly simple weighting is possible for irregularly polarized and spatially incoherent illumination, in which all light bundles are independent of one another. Finally, the integration limit across all angles of incidence includes 0 ≦ β ≦ 1, and the spectral integration over all waves is 0 ≦ k ≦ ∞.

In the frequency domain analysis (FDA), first the Fourier transform of the interference intensity signal I _{ζ, h} is calculated. For literal (non-numerical) analysis, we will use the unnormalized Fourier integration of Eq. (6).

In the above equations, K is the spatial frequency and is for example multiple cycles per μm. The frequency domain value qK, h has a unit of inverse wavenumber, for example μm. From this the power spectrum of formula (7),

The phase spectrum of equation (8) comes.

의 이중 프라임은 주사 시작 지점에 대해 픽셀에서 픽셀로 그리고 전체 양쪽에서, 프린지 차원에 이중의 불확실성(two-fold uncertainty)이 존재하는 것을 의미한다. 종래의 FDA는 그런 다음 파워 스펙트럼 Q_{K ,h} 에 의해 가중된 위상 스펙트럼

에 선형 맞춤으로 표면 토포그래피의 결정으로 바로 진행된다. 선형 맞춤은 각각의 픽셀에 식 (9)의 기울기와,

Double prime of means that there is a two-fold uncertainty in the fringe dimension, from pixel to pixel and on both sides relative to the scan starting point. The conventional FDA then added the phase spectrum weighted by the power spectrum Q _{K , h}

With a linear fit on, the process proceeds directly to the determination of surface topography. The linear fit is the slope of equation (9) for each pixel,

Provide the intercept of equation (10).

Note that the intercept or "phase gap" A "is independent of height h, but has a double prime inherited from the fringe order uncertainty of the phase data. The slope σ is free from this uncertainty. From the section A" and the slope σ , For a particular mean or nominal spatial frequency K 0, the " coherence profile "

The "phase profile" of equation (12) can be defined.

In the simple ideal case of a fully uniform and homogeneous dielectric surface that is free of thin film and dissimilar material effects and a fully balanced optical system, the phase profile and coherence pile may be applied to the surface height as shown in equations (13) and (14). Linearly proportional to:

이 더욱 정확하지만, 단색성 간섭측정의 프린지 차수 특성에서의 불확실성이 있다. 고 해상도의 경우, 이러한 불확실성을 없애기 위해 모호하지는 않지만 덜 정확한 코히어런스에 기초한 높이 값

를 사용하여 최종 값 h_θ를 얻을 수 있다.Of the two height calculations, the height value based on the phase

This is more accurate, but there is uncertainty in the fringe order characteristics of monochromatic interferometry. For high resolutions, the height value based on less accurate but less accurate coherence to eliminate this uncertainty.

The final value h _θ can be obtained by using.

는 여전히 공간 주파수의 선형 함수 가까이에 있다고 가정한다. 하지만, 본 실시예의 경우, 막 두께와 같은 표면 구조의 핵심 파라미터들을 실험적인 데이터와, 고도의 비선형 위상 스펙트럼 및 파워 스펙트럼의 관련된 변조를 포함할 수 있는 이론적인 예측을 비교하여 결정한다.Conventional FDA has found interference phases even in less idealized situations.

Assumes that it is still near the linear function of the spatial frequency. However, for this example, key parameters of the surface structure, such as film thickness, are determined by comparing experimental data with theoretical predictions that may include highly nonlinear phase spectra and associated modulation of the power spectrum.

For this reason, the definition of the Fourier transform of equation (6) having an interference signal of equation (4) is combined into the following equation for the predicted FDA spectrum:

In order to improve the computational efficiency, the partial character value of the triple integration of equation (15) can be obtained.

The analysis of the letters in equation (15) begins with a change in the order of integration to first obtain the individual interference signals g _{β, k, ζ, h} over all scan positions ζ at fixed β and k:

After expanding the cosine term of g _{β, k, ζ, h} in the usual way using equation (17),

The inner integral over ζ is given by equation (18).

From the stomach

Was used.

The function δ has the inverse of the physical unit of the argument, in this case the inverse of the wave number.

These δ (delta) functions confirm the equivalence between the spatial frequency K and the product βkn ₀ . Therefore, the logical change of the variables for the next integral

to be.

는 공간 주파수 K와 동일한 의미를 갖지만, 자유로운 적분 변수로 사용될 것이다. 식 (18)은 In the above equation

Has the same meaning as the spatial frequency K, but will be used as a free integral variable. Equation (18) is

Can be written as

와 k에 의존하게 된다는 것에 유의하기 바란다.By changing the variable, the β dependence of the R, Z, υ, ω term in equation (23)

Note that it depends on and k.

In the next step, first

Where H is the heavy side step function without units defined by equation (28),

f is an arbitrary function of K and k. Using equations (25) to (27), equation (23) becomes:

Now the expression

To get the final result:

Since there is little integration, equation (33) is much more computationally efficient than the original triple integration of equation (15).

Some limited cases are interested in solving analytically. For example, if the phase contribution (υ _{K, k} -ω _{K, k} ) = 0 and the reflectances R, Z are not dependent on the angle of incidence and wavelength, equation (33) is simplified to equation (34),

Only integrals containing the weighting factors Γ _{K, k} specified in equation (24) need to be processed. This idealized case simplifies the calculation for two additional limitations of equation (34), namely near-monochromatic illumination with high numerical aperture objectives and low numerical aperture broadband illumination.

Normalized spectrum for near monochromatic light sources with narrow spectral bandwidth k _Δ

Has

Where k0 is the nominal source frequency. The integral of equation (34) now takes the form:

Assuming U _{K , k} is inherently constant over a narrow bandwidth k _Δ ,

In the calculation of the integral

Is used and is valid for a narrow bandwidth k _Δ ≦ k 0. In particular, the positive, nonzero portion of the spectrum

Decreases.

은 없으며,Thus, in this particular case of a narrow spectral bandwidth light source, the reflectivity R, Z is constant and the phase contribution

There is no

In this particular case, the phase is linearly proportional to the surface height, consistent with conventional FDA. Spatial frequencies also correspond directly to directional cosines:

의 K 가중치가 식 (41)로부터 계산된다는 것에 유의하기 바란다. 이것은 도 6 (a)에 나타낸 스펙트럼 예의 증거이며, 수직 입사에서 시작하여 최대 대물 렌즈 개구수(NA)에 의해 부가된 지향성 코사인 한계까지의 범위에 걸쳐 동공면의 완전히 균일한 쌓기(filing)에 대한 이론적인 예측을 나타낸다:Thus, there is a one-to-one correspondence between the spatial frequency coordinates of the FDA spectrum and the angle of incidence. Fourier size also

Note that the K weight of is calculated from equation (41). This is evidence of the spectral example shown in FIG. 6 (a) and is for full uniform filing of the pupil plane over a range starting from normal incidence up to the directional cosine limit added by the maximum objective lens numerical aperture (NA). Theoretical predictions are shown:

.

.

As a second example, consider the case of broadband illumination with illumination limited to a narrow range β _Δ of directional cosine near normal incidence. The normalized pupil plane distribution is then

After changing the variable,

to be.

The finite integration of equation (34) in this case is of the form

This is calculated as follows:

In the above equation

Was used.

The positive non-zero portion of the spectrum when near this broadband source illumination and near normal incidence is thus:

가 소스 스펙트럼 분포

에 비례하는 친숙한 결과에 엄밀히 대응한다. 식 (52)는 또한 선형 위상 전개의 가정 For example, for a Gaussian spectrum centered at the nominal or average wavelength k 0, as shown in FIG. 6 (b), this is a Fourier magnitude.

Source Spectrum Distribution

Respond strictly to familiar results proportional to. Equation (52) also assumes linear phase evolution

Please note that, in accordance with the conventional FDA.

및 위상

이 간섭 세기 I_ζ,h의 푸리에 변환으로부터 구해지기 때문에, 그 역변환은 실제 간섭 신호Fourier size

And phase

Since this interference is obtained from the Fourier transform of the interference intensity I _{ζ, h} , the inverse transform is the actual interference signal.

Go back to the domain of,

를 다시 한번 사용하였다. 따라서, 세기 신호를 계산하는 한 방법은 식 (33)으로 푸리에 성분 qK,h를 생성하고 식 (54)를 사용하여 I_ζ,h로 변환하는 것이다.As far as spatial frequency is concerned, to emphasize that it is a free variable of the integral of equation (54)

Was used once again. Thus, one way to calculate the intensity signal is to generate the Fourier component qK, h with equation (33) and convert it to I _{ζ, h} using equation (54).

Assume random polarization of the source light in this model. However, this does not mean that the polarization effect should be ignored. Rather, the above calculation assumes an incoherent superposition of equally weighted results from the two orthogonal polarization states s and p defined by the plane of incidence of the illumination.

If you use superscript notation for polarization,

Thus, the average phase angle of the unpolarized light at this β, k is:

Note that unless the magnitude is the same for the two polarization contributions, very often:

및

이 복소 평면(complex plane)에 완전히 평행하지않는 한,Also

And

As long as it is not completely parallel to this complex plane,

및

에 각각 적용하며, 위상이 동일하지않는 한 이들을 직접 가산할 수 없다.Same observations of system and object reflectance

And

Are applied to each other, and they cannot be added directly unless they are in the same phase.

Properly removing the polarization effect from the calculation of the object surface reflectivity makes the modeling considerably simpler and flexible enough to handle more interesting cases of more complete polarized light.

The next step is to move to discrete numerical formulas, taking into account software development. The Discrete Fourier Transform is used to redefine the relationship between the interference signal I _{ζ, h} and the Fourier spectrum as follows:

는

의 복소 공액(complex conjugate)이고, 간섭 신호 I_ζ,h 에는 N개의 이산 샘플이 있다. 식 (60) 및 이하에서, 식의 유도에 있어서는 중요하였지만 공간 주파수 K를 대신할 때에는 더 이상 필요하지 않은 자유 변수

의 사용을 유보한다. 그러면 예측된 포지티브 주파수 FDA 복소 스펙트럼(complex spectrum)은 In the above equation,

The

It is a complex conjugate of, and there are N discrete samples in the interference signal I _{ζ, h} . In equations (60) and below, free variables that are important for derivation of equations but are no longer needed when substituting spatial frequency K

Reserve the use of The predicted positive frequency FDA complex spectrum is then

ego,

In the above equation, the normalized height dependence coefficient is

Lt;

In the above equation, the normalization of the integral range is

to be.

The heavyside step function H of equation (62) prevents unnecessary contribution to the sum. The weighting factor Γ _{K, k} is as defined in equation (24).

To compare the experiment with the theory, equation (61) is used to generate an experimental FDA spectrum, and equation (62) is used to perform a transformation that returns to the spatial domain of the theoretical prediction of I _{ζ, h} . This is most efficiently performed by fast Fourier transforms (FFTs).

The nature of the FFT determines the range of K values. If N discrete samples for I _{ζ, h} are spaced by increments ζ _step , start at 0 and increment

There will be N / 2 +1 positive spatial frequencies that are incremented by N / 2 cycles per per data trace.

To facilitate phase unwrapping in the frequency domain, the zero position of the scan is adjusted to be near the signal peak, reducing the phase slope in the frequency domain. Since the FFT always assumes the first data point at scan time as zero, the signal must be properly offset.

We will now focus on the modeling of sample surfaces with thin films.

7 shows two surface types with and without a thin film. In both cases, the effective amplitude reflectance Z _{β, k} is defined according to equation (66),

Where Z _{β, k} is the intensity reflectance and ω _{β, k} is the phase change in reflection.

Subscript β, k is the directional cosine of illumination

Stressing the dependence on, where ψ ₀ is the angle of incidence and

to be.

Is the wavelength of the light source. Note that the subscript β indicates the first incident directional cosine β ₀ .

The surface is partially characterized by its refractive index. The refractive index of the surrounding medium (generally the atmosphere) is n ₀ . In the case of the simple surface of Fig. 7 (a), there is only one refractive index n ₁ . In the case of the thin film of Fig. 7 (b), there are two surface deflections, n ₁ for transparent or partially transparent foil and n ₂ for substrate. Most commonly, these refractive indices are complex numbers characterized by real and imaginary parts. For example, in the case of chromium having a typical refractive index, for example, lambda = 550 nm, n ₁ = 3.18 + 4.41 i, and the imaginary part here adopts a rule defined as positive.

The refractive index of a material depends on the wavelength. Dispersion in air of refractive index n ₀ is not very important, but important for the surface of many samples, especially metals. If it changes near nominal k0 above a small wavelength, most materials have a nearly linear dependence on wavenumber,

Can be written as

,

는 명목상 파수 k0에서의 굴절률 n1인 경우에 각각 절편과 기울기이다.In the above equation,

,

Is the intercept and the slope, respectively, for the refractive index n1 at the nominal frequency k0.

The most commonly used index of refraction is Snell's law. Referring to FIG. 7 (b), the refracted beam angle inside the film is

이고,

ego,

In the above equation, ψ ₀ is the angle of incidence on the upper surface of the medium having refractive index n ₁ in the medium having refractive index n ₀ , and ψ _{1, β, k} is the refractive angle.

If the refractive index shows evanescent propagation, it is possible to take these angles at a complex value.

The complex amplitude reflectance of the boundary between the two media depends on the polarization, wavelength, angle of incidence and refractive index. The s and p polarized light reflectivities of the top surface of the film of FIG. 7 (b) were Fresnel formulae

.

The dependence on _{β, k} is due to the angles ψ ₀ and ψ _{1, β, k} , and the exit angle ψ _{1, β, k} introduces k dependence through the refractive indices n _{1 , k} . Similarly, the substrate-film interface reflectance is

to be.

Note that if the angle of incidence and refraction angle are the same in the Fresnel equation, the reflectivity for both polarizations is zero.

For simple samples without thin films, the sample surface reflectance is the same as the top surface reflectance.

Therefore, the phase change on reflection (PCOR) is different from the reflection caused by surface reflection.

In order to meet the boundary conditions, it should be noted that s polarization upon reflection is “flip” (= π phase shift for dielectric), whereas p polarization is not.

In any case resulting in division by zero in the Fresnel equation, the distinction between polarization states becomes meaningless at exactly normal incidence, but other equations treat this as limiting the case.

When the plus sign rule is used in the complex portion of the refractive index, the larger the absorption (complex portion), the larger PCOR ω _{β, k} . In other words, the larger absorption coefficient is equivalent to the reduction of the effective surface height. This in an intuitive sense assumes not just complete reflection and transmission at the boundary, but absorption as the penetration of the light beam into the material prior to reflection.

In the following general promise that an increase in surface height corresponds to a positive change in the phase difference between the reference plane and the measurement plane, the positive surface PCOR is subtracted from the interferometer phase.

Thin films are a special case of parallel plate reflections. Light passing through the top surface is partially reflected (see FIG. 7) and continues to the substrate surface where a second reflection with phase retardation occurs with respect to the first reflection. But this is not the end. The light reflected from the substrate is once again partially reflected when passing back through the top surface, resulting in an additional reflected beam that is directed back to the substrate. This in theory continues forever with each additional reflection slightly weaker than the previous one. Assuming all these multiple reflections remain and contribute to the final surface reflectivity, the infinite series

Can be obtained as

As explained, the dependence of β _{i, β, k indicates} the dependence on the incident directional cosine β ₀ in the air medium with refractive index n ₀ . The same equation (77) can be applied to all polarization states with a corresponding single surface reflectance.

Examination of these equations shows why conventional FDA treatment failed in the presence of thin films. Conventional FDA determines surface height with a linear fit to the Fourier phase spectrum weighted by the Fourier power spectrum, which uses broadband (white) light to produce Fourier spatial frequency spreading. This idea is that the phase evolution comes from the expected linear phase dependency on the surface height. Any other constant offset or linear coefficients (eg, "dispersion") related to the surface properties are removed by simply ignoring the phase contribution that does not change by system characterization or field position.

This work is perfect for simple surfaces. For unpolarized light, and for the most part circularly polarized light, the wavelength dependence of the PCOR is close to linear for the wavenumber and constant for a given material. However, with the presence of thin films, conventional assays are frustrated. The phase becomes nonlinear, the phase slope becomes sensitive to film thickness, and may be changing throughout the field of view. Thus, this assay uses knowledge of how the thin film changes the reflectance of the surface, for example, to determine key parameters of the surface structure, such as film thickness, by comparing experimental data with theoretical predictions.

We now describe how to compare experimental data with a library of theoretical predictions that provide surface structure parameters such as phase change for film thickness and reflection (PCOR). If the thickness of the film is unknown, the library for a single surface type (eg, SiO ₂ on Si) will range over many possible film thicknesses. In a frequency domain embodiment, the idea is to search this library to check for features of the FDA spectrum independent of surface topography, such as structures specific to the size spectrum due to thin film interference effects. The computer then uses the library spectrum to compensate the FDA data and allows for more accurate surface topography maps.

In one embodiment, the library includes sample FDA spectra for the surface structure, each providing a series of complex coefficients ρ _k representing the Fourier coefficients as a function of spatial frequency K. These spectra are the Fourier transform of the intensity data I _{ζ, h} collected during the scan ζ of the optical path of the interferometer. The spatial frequency K is proportional to the segment of the source light spectrum, the refractive index n ₀ of the air medium, and the angular wavenumber k = 2π / λ for the directional cosine β = cos (ψ), where ψ is transmitted to the object surface. The angle of incidence of the bundle of rays is:

In addition to the surface height, the ρ _k coefficients of the prediction library include the optical properties of the surface, which may affect the appearance of the FDA spectrum.

The prediction of the FDA spectrum includes an integral that represents the incoherent sum of the bundles of rays over a range of angles of incidence φ and angle wave k of the source light. As mentioned above, the numerical integration can be reduced to a computationally efficient single sum over the N each wave number k weighted by the factor Γ _{K, k} :

The weighting factor is

는 가중 인자의 전체 공간 주파수 전체의 합이다:Where V _k is the source spectrum and U _{K , k} is the light distribution of the pupil plane. Corresponding normalization

Is the sum of all the total spatial frequencies of the weighting factors:

은 간단하게 규정되는 정규화이고, H는 헤비사이드 계단 함수이다.In the above equation,

Is simply defined normalization and H is a heavyside step function.

Particularly for thin films, the inherent properties of the object surface structure are the object path phases ω _{K, k} and reflectance as detailed above. Z _{K , k} enters the spectrum ρ _k . Equally important is the reference path phase υ _{K, k} and reflectance _, which depend on the scanning interferometer itself R _{K , k} . These factors can be determined by theoretically modeling a scanning interferometer or by calibrating it with a test sample having known properties, which is described in more detail below.

A typical prediction library for thin films is a series of spectra ρ _k indexed by film thickness L. The memorized spectrum typically spans only 15 or 16 narrow spatial frequency regions of interest (ROI) in 256 frame intensity data collections, with the remaining values outside this region of interest being zero. The limit of the region of interest comes from the definition of spatial frequency:

US Lau 100X objective lens and the narrow bandwidth, the range of a typical spatial frequency of the scanning interferometer based on the 500nm light source is 2.7㎛ 4.0㎛ ^-1 to ^{- 1.} For computational efficiency, instead of an analytical search routine involving several recalculations for each pixel using equations (80) to (83), a compact lookup table indexed at 0.5-5 nm between sample spectra Can be used.

The library search includes the following steps: (1) selecting predicted FDA spectra from a library corresponding to a particular surface type, and (2) using merit functions to calculate how closely these spectra match experimental data. Then, (3) iterate over several or all library data sets to determine which spectra best match. It is to find "signatures" in the frequency domain that are inherently related to surface properties such as thin films, dissimilar materials, staircase structures, roughness, and interferometer interactions with optical systems. Thus, this comparison explicitly removes the linear rate of change of phase at spatial frequency, which is a property of the FDA spectrum that changes directly with surface topography and is thus independent of library searches.

When comparing the spectra, it is advantageous to separate the contributions of phase and magnitude to the merit calculation. So in theory

In the above equation, connect _K is a function that removes 2-π steps from the spatial frequency dependence of φ _{K, h} . For experimental data,

의 이중 프라임은 주사 시의 시작점에 대한 픽셀 대 픽셀, 및 전체 둘 다로부터의 프린지 차수(fringe order)의 불확실성을 나타낸다. 실험적인 데이터는 국부적인 표면 높이에 대한 기울기 항을 반드시 포함하는데, 이것이 ρ심볼 대신에 q 심볼을 사용하는 이유이다.

The double prime of represents the uncertainty of the fringe order from both pixel to pixel, and the whole, to the starting point in scanning. The experimental data must include the slope term for the local surface height, which is why we use q symbols instead of ρ symbols.

In a particular set of experimental surface parameters, the phase difference

는 보상된 FDA 위상이며, 실험적인 파라미터가 정확하다고 가정한다. 실험에 대한 이론의 양호한 일치는 원칙적으로 절편이 0인(즉, 위상 갭 0) 공간 주파수 K의 단순 선형 함수인 위상

를 초래한다. 따라서, 잘 보상된 위상

은 결국 종래의 FDA 분석에 다운스트림을 공급해야 하는 것이고, 주파수 공간에서의 위상의 기울기는 표면 높이에 직접 비례한다.Can be calculated. Phase difference

Is the compensated FDA phase and assumes the experimental parameters are correct. A good agreement of the theory for the experiment is that, in principle, phase is a simple linear function of spatial frequency K with intercept zero (i.e. phase gap 0).

Brings about. Thus, a well compensated phase

Will eventually feed downstream to conventional FDA analysis, and the slope of the phase in frequency space is directly proportional to the surface height.

에는 표면 높이에 독립한 실험에 대한 이론의 일치를 평가할 수 있도록 해주는 두 가지 중요한 특성이 있다. 그 첫 번째는 위상 갭 A" 또는 선형 맞춤으로 얻은 K = 0 절편 값

이고, 두 번째는 선형 맞춤 후의 파수에 대한 잔류 비선형성(residual nonlinearity)이다. 대응하는 메리트 함수는, 예를 들어Based on the observations in the previous paragraph, the compensated phase

Has two important properties that allow us to evaluate the agreement of the theory for experiments that are independent of surface height. The first is the K = 0 intercept value obtained by phase gap A "or linear fit

And the second is residual nonlinearity with respect to wavenumber after linear fit. The corresponding merit function is for example

에 대한 (가중된 크기) 선형 맞춤의 기울기이다. 식 (91)에서 round ( ) 함수는 위상 갭 A"를 ±π 범위로 제한한다.Σ _h is the compensated phase

The slope of the (weighted size) linear fit for. In equation (91), the round () function limits the phase gap A "to the range ± π.

및/또는

중 하나 또는 둘 다를 최소화함으로써, 라이브러리 검색을 할 수 있지만, 또한 푸리에 크기에 중요하고 유용한 서명을 가지고 있다. 이 크기는 특히 표면 높이에 비간섭적으로 독립하는 점에서 특히 흥미롭다. 따라서, 예를 들어 위상 메리트들(phase merits)을 사용하여 근사 유추법으로 다음의 크기 메리트 함수Although using only phase information, i.e. the value of the merit function

And / or

By minimizing either or both, you can do a library search, but also have a signature that is important and useful for Fourier size. This size is particularly interesting in that it is incoherently independent of the surface height. Thus, the following magnitude merit function is approximated using, for example, phase merits.

Can be defined,

Where Ω is the empirical scaling factor:

는 파수 의존성에 독립한, 물체의 전체 반사율에 가장 근사하게 관계되는 데 반해,

는 형상에 있어 이론적인 크기 선도(plot)와 실험적인 선도 가 얼마나 잘 일치하는지를 나타낸다.merit

Is most closely related to the total reflectivity of an object, independent of its wave dependence,

Shows how well the theoretical size plot matches the experimental plot in shape.

및/또는

는 위상 메리트

및/또는

에 더하거나, 이를 대신하기도 한다. 일반적인 라이브러리 검색 메리트는 함수는 따라서 Size merit function

And / or

Phase merit

And / or

In addition to, or in place of it. A common library search merit is that the function

를 선택한다.Where w is a weighting factor. In principle, the weight can be determined in equation (96), which knows the standard deviation for the various parameters. A more empirical approach is to test various weights on real and simulated data to see if they work well. In the following example, equal weights for all merit contributions

.

8-13 show merit function search procedures for film thicknesses 0, 50, 100, 300, 600, and 1200 nm of SiO ₂ on six Si, respectively. The single library for all examples includes a range of 0-1500 nm at 2 nm intervals. The data is a noise-free simulation. As explained in all the examples herein, the scan step is 40 nm, the source wavelength is 498 nm, and the source Gaussian FWHM is 30 nm (quasi-monochromatic).

,

메리트 함수가 실제 박막에 특히 유용하고, 시스템 특징화에 중요한 위치를 차지하며, 위상 갭 및 크기 결과에 직접 결합한다는 것을 보여준다.The most interesting aspect of this simulated search is the action of the four merit functions. In general, it has been found that including these four functions helps to reduce the ambiguity of the final merit value, and that there is a strong periodicity for the individual merit values that are a function of the film thickness. Another general observation is that the merit based on nonlinearity in both phase and size is most effective above 30 nm, while the phase gap and mean size are prominent at film thicknesses below 30 nm. this is

,

It shows that the merit function is particularly useful for real thin films, occupies an important place in system characterization, and directly couples to phase gap and magnitude results.

를 사용한다. 원칙적으로, 모델링이 성공적이면,

는 비선형성에 무관하여야 하고 위상 갭은 0이어야 한다. 따라서, 다음 단계는 위상 스펙트럼

에 대한 선형 맞춤이다. 이것은 제곱된 크기 대신에 크기 스펙트럼 P_K를 사용하기 위해 고 개구수 FDA에 더욱 효과적인 것으로 보인다. 선형 맞춤은 각 픽셀에 대한 기울기First determine the thickness of the film (or other uses for material identification or algorithms), and the FDA process proceeds in the usual manner, but the calibrated FDA phase instead of the original experimental phase data.

Lt; / RTI > In principle, if modeling is successful,

Should be independent of nonlinearity and the phase gap should be zero. Thus, the next step is the phase spectrum

For linear alignment. This appears to be more effective for high numerical aperture FDA to use the size spectrum P _K instead of the squared size. Linear fit is the slope for each pixel

Intercept, phase gap

.

Note that the phase gap A "has a double prime inherited from the fringe order uncertainty of the phase data. The slope σ _h is independent of this uncertainty. From the intercept A" and the slope σ _h , a specific mean or nominal spatial frequency "Coherence Profile" for K0

And, "phase profile"

 To regulate.

에서 픽셀 대 픽셀 프린지 차수 불확실성을 제거한다:Then, phase

Remove pixel-to-pixel fringe order uncertainty from:

Where α 'is an approximation to the original phase gap A ", independent of pixel-to-pixel 2π steps.

Finally, the height profile

.

을 생성할 때 이미 감산하였기 때문에, 위상 오프셋

를 감산할 필요가 없다는 것에 유의하여야 한다.Compensated phase

Phase offset because we have already subtracted when

Note that there is no need to subtract.

The first example of surface topography measurement (FIG. 14) is pure simulation. The surface topography is zero anywhere, but there is an underlying film layer that increases from 0 to 1500 nm in 10 nm increments. Using the same prediction library of FIGS. 8-13, this test demonstrates an explicit determination of the film thickness across the range of the complete but noisy prediction library.

The next example (Fig. 15) is also a simulation, but with noise added. The random additional noise is a Gaussian with a standard deviation of 2 bits out of the average 128th century bits, and is likely to illustrate the actual data. The result is clearly sufficient despite the significant difference (4% to 45%) of the reflectance between SiO ₂ and Si.

The system characterization will now be described.

과 선형 분산

을 규정한다. 시스템 특징화 데이터를 포함하기 위해, 라이브러리 검색에 이전에, 그리고 픽셀 단위로 임의의 다른 FDA 처리 이전에Phase offset using data collected during system characterization

And linear dispersion

To define. Prior to library search, and before any other FDA processing on a pixel-by-pixel basis to include system characterization data

를 보정한다.Fourier transformed experimental data

Calibrate

K0 in the above equation is the nominal spatial frequency and represents the nominal spectral frequency for the FDA data set, for example, as identified by being located at the midpoint of the region of interest (ROI). Note that the theoretical library remains unchanged. Scanning coefficient M (Greece uppercase letter "M") introduces a new system feature that allows the use of object surface reflectance as a parameter in library searches.

과 시스템 위상 갭 A_sys은 필드 위치의 함수로서 저장될 수 있으며, 진짜 시스템 편차를 다음 식에 따라 계산한다:Phase offset as a function of field position

And the system phase gap A _sys can be stored as a function of field position, and the true system deviation is calculated according to the following equation:

The magnitude factor M is also field dependent.

The generation of system characterization data is done in the same manner as described above for object samples. System characterization is determined by moving to a workpiece with known characteristics, measuring it, and examining how the results differ from what one would expect for a complete system. Specifically, using known samples with a predetermined calibration library entry, create a phase gap A "as in equation (98) and a final height h 'as in equation (102). Assume phase offset

And, system phase gap

Where connect _xy () is pixel-to-pixel phase continuity. The size map looks like this:

In some embodiments, for the last application (eg, SiO ₂ on Si) over a range of sample types, workpieces with possibly similar surface structures may be used to average several system characterizations.

In many of the descriptions and simulations above, the focus is on thin film surface structures, but this method can be applied to other types of complex surface structures. Next, how the scanning interferometry signal can be analyzed to explain the surface structure smaller than the optical resolution of the scanning interferometer microscope will be described. Optical resolution is ultimately limited by the wavelength of the light source and the numerical aperture (NA) of the light collecting optics.

Figure 16 (a) is a height profile determined from actual scanning interferometric data of 2400 lines (1 pmm) grating with a peak-to-valley (PV) modulation depth of 120 nm using a 500 nm nominal wavelength light source. It is shown. The top profile in FIG. 16 (a) shows the height profile determined using a conventional FDA assay. Conventional assays show only a PV modulation depth of about 10 nm, with very little estimate of the actual modulation depth. This inaccuracy occurs because the gratings are characterized at the limits of the optical resolution of 500-nm devices. This is true even if the device's camera's pixel resolution is more than sufficient to accurately analyze the grid.

One way to think about this effect is that when additional surface locations have sufficiently sharp surface properties proportional to the light wavelength of the diffracted light for the first pixel, the first method generally corresponds to the first surface location. Scanning interferometry signals of one camera pixel also include contributions from adjacent surface locations. Surface height features at these adjacent surface locations corrupt the conventional analysis of the scanning interferometry signal corresponding to the first surface location.

At the same time, however, this means that the scanning interferometry signal corresponding to the first surface position contains information about the near composite surface feature. FIG. 17 illustrates this by showing a scanning interferometry signal from pixels corresponding to various positions for the step height feature. The signal of (a) is for the step height is to the right of the pixel and higher, the signal of (b) is for the step to pass directly through the pixel, and the signal of (c) is for the step to the left and low of the pixel. One signature immediately visible in the signal is the reduction of the fringe contrast in (b) for (a) and (c). For example, if the step height was equal to a quarter of the pixel position exactly corresponding to the position of the wavelength and the step height, the fringe contrast in (b) would be because interference from the two sides of the step would exactly cancel each other out. It must disappear completely. There is also a lot of information in the signals shown in (a) and (c). For example, FIG. 18 shows the nonlinear distortion of the frequency domain phase spectrum for signals (a) and (c) of FIG. 17. Without the step height, the frequency domain phase spectrum would be linear. Thus the nonlinear feature of the frequency domain phase spectrum for the pixels corresponding to the surface location adjacent to the step height nevertheless contains information about the step height.

To more accurately measure the surface profile of a test surface where such an unanalyzed surface shape is present, a library search technique as described above for thin films can be used. For example, in the case of a test surface with an unanalyzed grating, a series of model FDA spectra are generated for different values of PV modulation depth and offset position. As in the thin film example, the surface height of the model spectrum is still fixed. This analysis is then parameterized by the modulation depth and offset position, as in the example of the thin film above, except that the model spectrum is parameterized by the film thickness. The comparison between the signature of the FDA spectrum against the actual test surface and the other model spectrum can then be used to determine a match. Based on the match, the distortion of the actual FDA spectrum for each pixel generated due to the presence of the grating is removed, so that the surface height for each pixel can be determined using conventional processing. The results of this analysis using the same merit functions as described above for the thin film are shown in FIGS. 16 (b) and 19 (b).

FIG. 16 (b) shows a height profile determined by using a library search analysis on a grid of 2400 lines per mm described above with reference to FIG. 16 (a). Although the same data was used in Figures 16 (a) and (b), the library search assay modulated the PV modulation depth of the grating, rather than the actual 120-nm modulation than the 10-nm result determined by the conventional FDA treatment of Figure 16 (a). It was determined to be 100 nm much closer to the depth. 19 (a) and (b) show similar analytical methods for simulations with discrete step heights, assuming nominally a 500-nm light source. Figure 19 (a) shows the height profile determined using conventional FDA treatment compared to the actual height profile (dotted line) of the simulation. 19 (b) shows the height profile determined using the library search method (solid line) compared to the actual height profile (dashed line) of the simulation. Parameters for the model spectrum of the library search are position and step height magnitudes. As shown, library search analysis improves lateral resolution from about 0.5 microns to about 0.3 microns.

In the example of Figures 19 (a), (b), the library is expressed in terms of the equation for the thin film except that Theoretically generated using an equation similar to), the first and second terms of the molecule are weighted according to the lateral distance of the measuring point from the actual step height position, and the parameter L corresponds to the step height itself, not the film thickness. do. Thus, this theoretical model is based on the complex summing of the light rays appearing from both sides of the step height. As the transverse position of the pixel being inspected moves away from the step height at that point, the signal tends to change to a signal of a simple flat surface.

 In the example of Figures 16 (a) and (b), the library was created experimentally by observing the signal generated for the 2400 line grating by the interferometric device. Based on this experimental data, a library was constructed by correlating the signatures in the scanning interferometric data with the corresponding transverse position in the lattice period. The experimental data for each pixel of the test sample (which in this case was a 2400 line grid) was compared with the library to determine the best transverse position within the period of the pixel.

In the above detailed analysis, a comparison between the information in the actual data and the information corresponding to the different models was made in the frequency domain. In other embodiments, the comparison may be in the scan coordinate domain. For example, a change in the absolute position of the fringe contrast envelope generally indicates a change in surface height at the first surface position corresponding to that signal, and the shape of the signal (independent of its absolute position) is determined by the first surface position. Information of the composite surface structure, such as the surface structure of the underlying layer and / or adjacent locations.

One simple case is to consider the size of the fringe contrast envelope itself. For example, when the thickness of a thin film is very thin compared to the range of wavelengths generated by a light source, the interference effect generated by the thin film becomes wavelength independent, in which case the thin film thickness directly modulates the size of the fringe contrast envelope. do. Thus, in order to identify a match for a particular thin film thickness (considering the systematic contribution of the interferometer itself), the fringe contrast size can generally be compared with that of models corresponding to different thin film thicknesses.

Another simple case is to examine the relative spacing of zero crossings of the fringes in the fringe contrast envelope. In simple surface structures illuminated with symmetrical frequency distributions, the relative spacing between different zero crossings should be nominally equal. Thus, the variation in relative spacing (when considering the systematic contribution of the interferometer itself) represents a composite surface structure and can be compared with models of different composite surface structures to confirm consistency with a particular surface structure.

Another case is to perform a correlation between the frequency domain signal and the scan domain signal corresponding to a different model of the test surface. The match generally corresponds to the correlation with the highest peak value, and represents a model in which the scan domain signal has the shape that most closely resembles the shape of the actual signal. It should be noted that this analysis is generally independent of the surface height because of the difference between the surface height of the actual sample and the surface height of each model that just moved the position of the peak of the correlation function, but generally does not affect the peak value itself. . On the other hand, once the calibration model is identified, the position of the highest point of the correlation function of the calibration model yields the surface height of the test sample without requiring further analysis (such as conventional FDA).

Similar to the analysis in the spatial frequency domain, the analysis in the scan coordinate domain can be used for many different types of composite surfaces, including thin films as well as other composite surface structures such as the unanalyzed surface height features as described above. .

The scan coordinate library search analysis involving the correlation between the signal of the test sample and the signal corresponding to various models of the test sample is now described in detail.

This approach is other than all the pixels in the data set that correspond to surface locations with the same underlying, localized interference pattern, and surface locations with the same complex surface characteristics, including only shifted (and maybe resized) positionally for each pixel. Discard all assumptions about the interference pattern. It does not matter whether the signal looks real, has a Gaussian envelope or linear phase pattern in the frequency domain, or whatever. The technical idea is to generate a sample signal or template representing this localized interference pattern for different models of the composite surface structure of the test object, and then the localized interference pattern for each pixel is the actual localized interference pattern. Find the model that most closely matches the shape of and find the scan location in the data set that provides the best match between the interference pattern template and the observed signal for that model. One approach is to correlate each template with data mathematically. For each model, two profiles are recovered using a complex (i.e. real plus imaginary) template function, one closely associated with the envelope of the signal and the other associated with the underlying phase.

In another embodiment, the analysis for each pixel, for example, comprises: (1) selecting a test template from a template library calculated or recorded for a particular value of an adjustable parameter such as film thickness; (2) obtaining a local surface height using a selected template and a correlation technique (described later for this example); (3) recording the peak merit function value for the selected test template based on the correlation technique; (4) repeating steps 1 to 3 for every template or subset template in the library; (5) determining which test template best matches (= highest peak merit function value); (6) recording the value of the adjustable parameter for the best matching template (eg thin film thickness); And (7) recalling the height value providing the highest point coincidence position in the data trace.

A suitable correlation technique will now be described based on complex correlation. Template interference pattern for each model of the test surface

및

는 복합 표면 구조의 특성을 기술하는 것이지만, 신호에 대응하는 위치에서의 표면 높이에 무관하며, 0(영)으로 설정된다. 바람직한 실시예에서,

및

는 또한 간섭계로부터의 계통적인 기여분을 고려한다. 이후 템플릿 패턴에 대한 복소 표현식을 사용한다:In this equation, the index j represents a specific model of the template pattern. function

And

Describes the characteristics of the composite surface structure, but is set to 0 (zero), independent of the surface height at the location corresponding to the signal. In a preferred embodiment,

And

Also considers the systematic contribution from the interferometer. We then use a complex expression for the template pattern:

We also use a window function that selects a specific part of the complex template function:

For example, the appropriate window

는 수동으로 설정될 수 있다.Window width in the above formula

Can be set manually.

를 가지고, 이를 실제 데이터 세트와 비교할 준 비가 되었다. 이에 대한 준비로, 실제 실험 데이터 세트Interference pattern template

Now we are ready to compare this with the actual data set. In preparation for this, the actual experimental data set

를 생성하기 쉬울 것이다.Complex signal starting from

It will be easy to generate.

The Fourier transform of this signal

In the above equation

Then construct a partial spectrum with the positive frequency portion of the spectrum:

Then the inverse transformation

to be.

의 실수부는 원래의 실험적인 데이터 I_ex이다. 또한 위상과 포락선은 단순한 연산에 의해 분리할 수 있으며, 예를 들어 복소 함수

의 크기를 사용한 신호 세기 AC_ex(x)와 포락선 m_ex의 곱으로 접근할 수 있다:In the above formula, a complex function

The real part of is the original experimental data I _ex . In addition, phase and envelope can be separated by simple calculations, for example complex functions

We can access the signal strength AC _ex (x) by the magnitude of the envelope m _ex :

와 같은 일반적인 형상을 가지기 위한 m_ex의 적어도 의미있는 부분을 예상하여, 유일한 차이는 선형 오프셋 h_ex와 스케일링 인자 AC_ex(x)이다. 또한 정확한 모델에 대해 높이 h_ex에 선형적으로 비례되도록, 실험적 및 간섭 패턴 템플릿의 위상 오프셋

사이의 차이를 각각 예상한다.According to the theory that underlies this technique, we can

In anticipation of at least a significant portion of m _ex to have a general shape such as, the only difference is the linear offset h _ex and the scaling factor AC _ex (x). In addition, the phase offsets of the experimental and interference pattern templates are linearly proportional to the height h _ex for the correct model.

Expect the difference between each.

에 의해 표현되는 특정 신호 패턴을 실험적인 데이터 세트

내에 배치하고, 각각의 상이한 모델 j에 대해 얼마나 잘 일치하는지를 결정하는 것이다. 이하에서, 지수 j를 버리고 각각의 모델에 대해 일치 분석을 진행하는 것에 유의하여야 한다.The challenge we faced is the interference pattern template

Experimental data set of specific signal patterns represented by

And how well they match for each different model j. In the following, it should be noted that discarding the index j and proceeding concordance analysis for each model.

의 형상이 가장 일치되는 주사 위치

를 찾는 것이다. 실행 가능한 접근법은 창 w에 의해 규정된 주사의 세그먼트 내의 신호와 함께 간섭 패턴의 정규화된 상관에 기초한 메리트 함수이다:The first step is the envelope m _ex , m _pat and

Scanning position where the shape of

. A viable approach is a merit function based on the normalized correlation of the interference pattern with the signal in the segment of the scan defined by window w:

 In the above equation,

 Is a complex correlation function,

의 사용은 동시성의 선형 위상 항

를 상쇄하고,

가 일치하는 경우에 Π를 최대화한다. 상관의 절대값 ∥은 임의의 잔류 복소 위상을 제거한다.Is a normalization that makes the merit function Π independent of signal strength. Complex conjugate of the template

The use of the linear phase term of concurrency

Offsets,

Π maximizes if The absolute value of the correlation removes any residual complex phase.

를 방지하기 위해, 다음과 같이 최소값을 공통요소(denominator)에 가산하는 것이 현명하고,From the generation of false high values or the encounter of singularity at low signal levels

To avoid this, it is wise to add the minimum value to the denominator as

에 걸쳐 신호 세기

의 최대값을 돌려주고, MinDenom은 메리트 함수 검색에서 유효한 것으로 간주하는 상대적인 신호 세기의 최소값이다. MinDenom의 값은 5% 또는 어떤 다른 작은 값으로 정해질 수 있고(hard coded), 또는 조정 가능한 파라미터로 남을 수 있다.Where max () is the total scan length

Signal strength across

Returns the maximum value of, and MinDenom is the minimum value of relative signal strength that is considered valid in merit function search. The value of MinDenom can be hard coded to 5% or some other small value, or left as an adjustable parameter.

은 상관 이론을 사용하여 주파수 도메인에서 수행될 수 있다:Correlation integral

Can be performed in the frequency domain using correlation theory:

In the above formula, I is

Was created using

를 얻고, Π의 값은 완전한 일치에 대응하는 것을 하나 가지는 0에서 1까지의 범위의 일치의 품질의 측정값이다. 메리트 함수의 최고점 값은 어느 모델이 최고 일치하는지를 결정하기 위해 상이한 모델 각각에 대해 계산되며, 그 모델에 대한 최고 일치 위치

는 표면 높이를 제공한다.Search through Π to find the highest value finds the best match

The value of [pi] is a measure of the quality of the match in the range of 0 to 1 with one corresponding to a perfect match. The peak value of the merit function is calculated for each of the different models to determine which model is the best match, and the highest match location for that model.

Gives the surface height.

20-24 show an example of this technique. 20 shows the actual scan interferometry signal of the base Si substrate without the thin film. 21 and 22 illustrate interference template patterns of a bare Si substrate and a thin film structure each having 1 micron of SiO ₂ on Si. 23 and 24 illustrate merit functions that are functions of the scan position for the template functions of FIGS. 21 and 22. The merit function shows that the interference template pattern (peak value 0.92) of the bare substrate matches much better than that of the thin film template pattern (peak value 0.76), thus indicating that the test sample is a bare substrate. In addition, the position of the highest point in the merit function of the correct template pattern provides the relative surface height position of the test sample.

Some examples of techniques related to unanalyzed shapes, such as patterned structures in front end semiconductor manufacturing processes, are now described.

Although patterned by optical lithography, semiconductor manufacturing processes at the front end or transistor level include shapes that are well below the resolution limit of visible wavelength microscopy. The minimum shape, such as the transistor gate, is approximately 45 nm wide, and the instrument transfer function of a typical scanning interferometer is zero for a 400 nm periodic structure. For example, FIG. 28 shows the theoretical and experimental instrument transfer functions of a white light interference microscope using a Mirau objective lens of 100 ×, 0.8 NA and noncoherent illumination. Separation of gates, shallow trench isolation (STI), wirings and vias are often close to the lower limit; Thus, some surface structures are seen, but not all. Therefore, these unanalyzed shapes cannot be measured directly as height objects in the usual way using white light interference microscopy. However, monitoring of these shape parameters (e.g., depth and width) is still possible if one knows what is the height change below the optical resolution that usually affects the generation of the scan interferometric data.

29A shows a simple cross-sectional model of a symmetric grating with unpatterned areas on both sides. The y coordinate is parallel to the lines and in the figure the x coordinate is from left to right. The z = 0 vertical position corresponds to the top of the lines. The simple scalar diffraction model and the principle of Abbe confirm the case where the grating line width L and the spacing W are below the resolution limit implied by FIG. 28, the grating lines are overall blurry, and NewView is transverse Dimensions L and W and height H cannot also be measured directly. However, this same scalar analysis shows that the apparent height of the etched regions actually depends on the height, width and spacing of the lines. The exact dependence is predicted to some extent by modeling and can be refined by experimental evidence.

As mentioned above, certain embodiments disclosed herein utilize observation that the apparent height of the patterned areas is related to important shape parameters.

Apparent surface height profiles are created using conventional processing interferometric techniques such as the FDA. If the pitch is less than half wavelength, these profiles show no grating lines at all, or show simulated measured surface profiles superimposed over the grating structure, as shown in FIG. It shows some echoes of the lines with the averaged "height H '. Thus, the height H 'between the area on the line and the area on the unpatterned bare substrate is measured. This result can be referenced to the zero-etch height by subtracting the etch depth E to find E '= H'-E. Alternatively, if a measured profile of etch free is available as in FIG. 29C (also representing a simulated measured surface file superimposed on the lattice structure), the measured etch depth E 'may be directly referenced to the etch free height. Can be. If the dimensions L and W in the transverse direction are known in advance, the exact line height H depends on the modeling of the sensitivity of the whole measurement procedure for the critical parameters. Alternatively, if we know H and L, we can estimate some other parameters, such as W or line shape.

The use of more stringent modeling to account for the polarization effect indicates that the simulated profiles shown in FIGS. 29B and 29C qualitatively represent the behavior when polarization is orthogonal (defined in the x direction) to the gate lines. Shows. In contrast, parallel or y polarization is much more strongly influenced by the top of the line in some cases where there seems to be no gap between the lines. This has a physical meaning in that the polarization aligned with the gate lines produces a current that increases the effect of the top of the lines at the expense of the areas between the lines. Since the sensitivity of the measurement for various structural parameters varies with polarization, it may be possible to separate certain parameters, such as etch depth, while minimizing sensitivity for other parameters, such as line width.

Accordingly, embodiments of this measurement technique include: 1) the use of standard interference microscopy with circular polarization, and the presence of neighboring unpatterned regions in a field of view (FOV) of step height and known height for silicon before etching. compare; 2) as above, but using x (orthogonal to the lines) polarization to improve sensitivity to deep (> 20 nm) trenches; 3) use of linear polarization and comparison of heights of neighboring regions arranged in a direction orthogonal to the grating lines; 4) comparison of the measured heights for both the x and y polarization states within the same field of view against a common reference that is not dependent on polarization, such as a smooth and flat area; And 5) a comparison of the measured heights for both the x and y polarization states within the same field of view with respect to each other, for example by simultaneously acquiring data of x and y polarizations or directly interfering polarization states. . This approach can eliminate all the need for individual criteria on the object surface.

It should also be noted that the basic measurement principle is not limited to lattice but can be extended to other structures.

Scalar or Abbe models provide some important insight into fundamental measurement problems, but complex (multiple materials) unanalyzed surface structures, based on the rail hypothesis 2D approach or more rigorous RCWA approach, all described above As such, one can solve the Maxwell equation over a range of bandwidths and angles of incidence to benefit from more rigorous modeling of the interferometric signal.

For example, the 2D rail approach describes the inversion result shown in FIG. 27 for circularly polarized light incident on the sample wafer on the patterned structure. Specifically, there is contention between the x polarization state and the y polarization state that results in inverse correlation in non-polarized or circularly polarized light. This inversion may be due to the high sensitivity to the etch depth of the reflection intensity of the x-polarized light, causing modulation in the contribution of the x-polarized light relative to the y-polarized light. Since these two polarizations report different depths, the balance between their reflected intensities can result in a nonlinear correlation between the apparent grating modulation depth and the actual grating modulation depth. The 2D rail calculations showed that this inversion could occur due to the grid lines not analyzed.

In another example, a rigorous RCWA approach was used to model a pure silicon 5 line grating without a top layer, with lines having a width W = 120 nm and a pitch L = 320. The simulation output shown in FIG. 30B is a SWLI signal (z direction) simulated for each line (x direction) of pixels. FIG. 30B shows only the results for y polarization parallel to the lines, which is more sensitive to the tops of the lines than the area or trench between the lines. Examining the output for one pixel for x polarized light and y polarized light as shown in FIGS. 31A and 31B, respectively, will recognize a well-known white light interference pattern, approximating a carrier fringe pattern modulated by an envelope or fringe contrast function. Can be. Noteworthy is a slightly distorted x polarized signal (FIG. 31A), which indicates contention between the unanalyzed top of the grating lines and the trenches between them, shifting generally left with an envelope structure, corresponding to a low height. Results in a weakened signal.

FDA analysis of these signals produces the apparent surface profile shown in FIGS. 32A and 32B, respectively, for x polarization (orthogonal to the grid line) and y polarization (parallel to the grid line). To relate this to the step height measurement, the center pixel and edge pixel are compared to determine H ', which edge pixel will probably represent the true etch depth E of the substrate. The measured substrate etch depth is then E '= H'-E. This is the value that would have been measured when referring to a substrate area that was not etched. Based on modeling or empirical data, the apparent etch depth is changed to the actual etch depth.

33A and 33B are apparent when extracted from simulated scanning interferometry data that is a function of different actual etch depth E for y polarization (parallel to grid line) and x polarization (orthogonal to grid line), respectively. Results of RCWA analysis for etch depth E '= H'-E are shown. E '= E is for reference. 34A and 34B show corresponding signal intensities of different actual etch depths for y polarization (parallel to grid line) and x polarization (orthogonal to grid line), respectively. These results show the expected behavior for both polarization states. Assuming that the minimum value of the reflected intensity at the 145 nm etch depth shown in FIG. 34B can be avoided, apparently the preferred configuration for etch depth sensitivity is x polarization.

The difference between the measured height E 'and the etch depth E is in some sense a bias or offset due to the unanalyzed grating line comprising the top film layer. In an ideal case, assuming that the substrate etch depth was the only important parameter, the device would simply ignore the presence of the lines. FIG. 35 shows the same data as in FIG. 33B in another graph, showing that the bias is not very large in the case of x polarization and does not change rapidly depending on the etching depth. Specifically, Figure 35 shows the measurement bias or offset E-E 'as a function of the step height -E' measured for the silicon grating.

36 shows the expected RCWA results for circular polarization in the case of pure silicon lattice. Specifically, FIG. 36 shows the RCWA prediction for the measured etch depth E '= H'-E, which is a function of the actual etch depth E, in the case of a silicon grating, compared to a one-to-one corresponding line using circular polarization. Noted is the response of the poor system as a whole, in particular the inverse correlation of the measured etch depth with the actual etch depth, especially between 60 nm and 170 nm. This is the same phenomenon as that predicted from the hypothesis of the rail shown in FIG. 27 and described above. If the etch exceeds 100 nm, the measured depth is negative, meaning that the etched silicon appears to have risen over the unetched silicon. This area looks like a protrusion, not a recess.

37A and 37B show RCWA predictions for measured (apparent) etch depths E '= H'-E, respectively, with etch depths E = 100 nm and pitch, for y polarization and x polarization for a 5-line silicon grating. It shows as a function of the shape width W in the case of L = 320 nm. Solid lines in both graphs represent the etch depth for reference. These graphs show some interesting behavior, including relatively insensitive to line width over a wide range. These results mean that for line widths between 100 and 180 nm, for example at least for a combination of these parameters (compared to FIG. 33), x polarization measurements are much more sensitive to etch depth than line width. In either case, however, although the sensitivity is weak, y polarization shows higher sensitivity to line width than to etch depth.

Qualitatively, the reflectance minimum in FIG. 33B and the inverse correlation in FIG. 36 correspond to the cause of the phase filp where the depth of the grating is related to a quarter-vave antireflective condition. It may be understood that this is due to the situation. Thus, if one wants to move to a more straight portion of the correlation curve, the product of the grating depth and the refractive index of the material between the lines (e.g., in the present structure, n = 1 for air) is a quarter wavelength of light. When the quarter wavelength condition is met when the sum of any integer multiple of half the wavelength of light (including O (zero)) is met, the wavelength of the light used in the interferometer is proportional to the quarter wavelength condition. Can be adjusted. On the other hand, the presence of a reflectance minimum (or more generally, reflectance information obtained from the intensity of the interferometric signal) compares this information with the expected information for the different models of the sample to provide information about the unresolved surface shapes. In the determination, it can be used alone or in combination with the apparent surface profile.

While the above examples are particularly directed to one-dimensional patterned structures, the general principles of this technique can be extended to other types of unanalyzed surface patterns, such as discrete step heights and two-dimensional patterned structures. The basic principle is that although such surface shapes may not be analyzed, they contribute both to the interferometric signals for individual pixels and to the collective information extracted from the interferometric signals from the plurality of pixels. Thus, the information obtained by the experiment can be compared with the model parameterized by different values for the important unanalyzed shapes, with the best comparison to obtain the corresponding values of the important shapes.

Also, in other examples, other portions of the grating may have different modulation depths or may be constructed of different material compositions. For example, the lattice structure may be formed on a silicon substrate comprising regions of silicon dioxide. Also, the regions of the grating between the silicon disides can be etched to different depths. In such embodiments, the information compared with the models may be, for example, the difference in apparent surface height between different analyzed regions of the grating. In other words, for example, individual grating lines cannot be analyzed, but the portion of the grating formed over silicon dioxide may be distinguishable from other parts of the grating.

FIG. 38 is a schematic diagram illustrating a method for automating various components of an interferometric system 900 used to generate an interferometric signal under the control of an electronic processor 970, and in the embodiments described herein, mathematical A processor 972 for analysis to perform the analysis (eg, comparison with model libraries); A device controller 974 for controlling various components of the interferometric system 900; User interface 976 (eg, keyboard and display); And a storage medium 978 that stores information (eg, library model and calibration information), data files, and automated protocols. Interferometric system 900 generally irradiates test light 922 to a test object 926 that is fixed to mount 940 on stage 950.

First, interferometric system 900 may include a motorized turret 910 configured to support a plurality of objective lenses 912 and to introduce a selected objective lens into the path of the input light. One or more of the objective lenses may be interfering objective lenses, providing different magnifications using different interference objective lenses. Further, in certain embodiments, an interfering objective lens (eg, to illuminate a grating pattern with light polarized in a direction orthogonal to the grating line) is attached to one or more polarizing elements (eg, to Linear polarizers). In this case, the orientation of the polarizing element can also be under automatic control, for example to align the polarization of the illumination light with respect to the grating lines of the patterned structure. In addition, since one or more of the objective lenses may each be non-interfering objective lenses having different magnifications (ie, no interference intervals), the interferometric system 900 may also be a conventional microscope to collect optical images of the test surface. Can operate in mode. The turret 910 is under the control of the electronic processor 970 and selects the desired objective lens according to the user's input or some automated protocol.

The interferometric system 900 then includes a motorized stage 920 (eg, tube lens holder) that supports the relay lenses 936, 938. This stage 920 is an interferometer for determining the characteristics of the surface of the object, such as the profiling mode as planned throughout this application, or as filed by Colonna de Lega et al. CHARACTERISTICS OF AN OBJECT SURFACE, "selection of ellipsometry or reflectometry mode, in which a pupil plane is imaged on a detector, as disclosed in US Patent Publication No. US-2006-0158659-A1. And the entire contents of the patent document are incorporated herein by reference. The motorized stage 920 is under the control of the electronic processor 970 and selects the desired relay lens according to the user's input or some automated protocol. In other embodiments, the translation stage is moved to adjust the position of the detector and switched to either the first mode or the second mode, and the transition is under the control of the electronic processor. Further, in embodiments having a plurality of detection channels, each detector is connected to an electronic processor 970 for analysis.

The interferometric system 900 may also include motorized openings 930 and 932 that control the dimensions of the field stop and aperture stop under the control of the electronic processor 970. In addition, the motorized openings 930 and 932 select the desired settings under the control of the electronic processor 970 according to the user's input or some automated protocol.

Also under control of the electronic processor 970 is a transition stage 980 used to change the relative optical path length between the test interval and the reference interval of the interferometer. The transition stage 980 may be connected to adjust the position of the interference objective lens relative to the mount 940 supporting the test object 926. Alternatively, in other embodiments, the transition stage may adjust the position of the interferometric system globally relative to the mount, or the transition stage may be connected to the mount, so this is a mount that moves to change the optical path length difference.

In addition, the lateral shift stage 950 can also be connected to a mount 940 that supports the test object under the control of the electronic processor 970 to shift the area of the test surface laterally under light irradiation. In certain embodiments, the transverse transition stage 950 can also orient (eg, provide tips and tilts) the mount 940 such that the test surface is aligned perpendicular to the optical axis of the interfering objective lens.

Finally, object handling station 960 under control of electronic processor 970 may also be coupled to mount 940 to provide automated introduction and removal of test samples to interferometric system 900 for measurement. . For example, automated wafer handling systems known in the art can be used for this purpose. Also, if desired, the interferometric system 900 and the object handling system 960 may be housed under vacuum or in a clean room to minimize contamination of the test object.

The resulting system provides excellent flexibility in providing a variety of measurement modalities and procedures. For example, the system may first be configured in microscope mode with one or more selected magnifications to obtain an optical image of the test object for various transverse positions of the object. Such an image may be analyzed by a user or an electronic processor 970 (using a machine vision technique) to identify certain areas within the object (eg, specific structures or shapes, landmarks, reference markers, defects, etc.). Can be. Based on this identification, selected regions of the sample can then be closely examined in an elliptical polarization measurement mode to determine sample properties (eg, refractive index, underlayer thickness, material identification, etc.).

When using the automated object handling system 960 together, the measurement procedure can be automatically repeated for a series of samples. This may be useful for various process control methods, such as monitoring, testing, and / or optimizing one or more semiconductor process steps.

For example, the system can be used in semiconductor processes for tool specific monitoring or for control of the process flow itself. In process monitoring applications, monolayer / multilayer films are grown, deposited, polished or etched on unpatterned Si wafers (monitor wafers) by corresponding process tools, and then thickness and optical properties are described using the interferometry system disclosed herein. Is measured.

Within wafer uniformity, the average of the thickness (and / or optical properties) of this monitor wafer is determined whether the relevant process equipment is operating in accordance with a targeted specification or retargeting, adjusting, or using the product. It is used to determine if it should be considered.

In process control applications, the latter may be grown, deposited, polished, or etched on Si, patterned by corresponding process tools, product wafers, and then thickness and optical properties may be incorporated into the interferometric system disclosed herein (e.g., For example, using an elliptical polarization measurement mode, a profiling mode, or both). Product measurements used in process control generally include the ability to align the measurement instrument to small measurement sites and critical sample areas. This site can consist of a multilayer stack (which can itself be patterned), requiring complex mathematical modeling to extract the relevant physical properties. Process control measurements determine the safety of the integrated process flow and determine whether the integrated process should be continued, retargeted, redirected to other equipment, or shut down completely. Decide

Specifically, for example, the interferometry system disclosed herein can be used to monitor the following equipment: diffusion, rapid thermal annealing, chemical vapor deposition apparatuses (both low and high pressure), dielectric etching, chemical mechanical polishing machines Plasma deposition, plasma etching, lithography tracks, and lithographic exposure apparatuses. In addition, the interferometric systems disclosed herein can be used to control the following processes: trench and isolation, transistor formation, and interlayer dielectric formation (such as dual damascene):

The methods and systems described above are particularly useful for semiconductor applications. Additional embodiments of the present invention include the application of any of the foregoing measurement methods to process any of the semiconductor applications described below, and include a system that implements both measurement techniques and semiconductor applications.

Quantitatively measuring surface topography is of great importance in the semiconductor industry. Because typical chip shapes are compact, the instrument used to measure them should typically have high spatial resolution both parallel and perpendicular to the chip surface. Engineers and scientists use surface topography measurement systems for process control and to detect defects that occur during manufacturing, particularly as a result of processes such as etching, polishing, cleaning and patterning.

In order for process control and defect detection to be particularly useful, they must have lateral resolution that can be compared with the lateral size of a typical surface shape and vertical resolution that can be compared with the minimum surface step height allowed. Typically, this requires lateral resolutions less than microns and vertical resolutions less than 1 nanometer. It is also desirable that such a system does not contact the surface of the chip, or otherwise apply a potentially damaging force to the surface, and take measurements to avoid deformation of the surface or to cause defects. In addition, it is well known that the effects of many processes used in chip fabrication are strongly dependent on local factors such as pattern density and edge proximity, so that surface topography measurement systems can produce high measurement yields and one or many important surface geometries. It is very important to have the ability to sample densely over large areas in a range that can be included.

To make electrical interconnects between different parts of the chip, the use of so-called 'dual damascene copper' processes is becoming common among chip manufacturers. This is an example of a process that can be effectively characterized using a suitable surface topography system. The dual damascene process involves the deposition of five parts, namely (1) an interlayer dielectric (ILD), in which a layer of dielectric material (such as polymer or glass) is deposited on the surface of the wafer (including a plurality of individual chips). (2) chemical mechanical polishing (CMP), which polishes the dielectric layer to create a smooth surface suitable for precise optical lithography, (3) a narrow trench running parallel to the wafer surface, and from the bottom of the trench A combination of lithographic patterning and reactive ion etching steps resulting in a complex network comprising small vias leading to a defined) electrically conductive layer, (4) a combination of metal deposition steps overfilling trenches and vias with copper, (5) Final chemical mechanical polishing (CM) to remove excess copper and leave a network of copper-filled trenches (and possibly vias) surrounded by dielectric material Can be considered to have a step P).

Typically, the thickness of copper (ie, trench depth) in the trench region, and the thickness of the surrounding dielectric is in the range of 0.2 to 0.5 microns. The width of the final trench may be in the range of 100 to 100,000 nanometers, and the copper region within each chip may form a regular pattern, such as an array of parallel lines in some regions, and may not have a distinct pattern in other regions. Similarly, within some regions the surface may be densely covered with copper regions, while others may be sparse. The polishing rate, and therefore the thickness of copper (and dielectric) remaining after polishing, is strongly dependent on polishing conditions (such as pad pressure and polishing slurry compound) and local detail arrangements of copper and surrounding dielectric regions (ie, orientation, proximity and shape). It is important to recognize that you depend on it in a complex way.

It is known that this 'position dependent polishing rate' can change surface topography on various side length scales. For example, this means that chips located closer to the wafer edge on the aggregate are polished more quickly than those located closer to the center, creating areas of copper thinner than desired near the edges and thicker than desired at the center. This is an example of the non-uniformity of the 'wafer scale' process, i.e. what occurs at the length scale that can be compared with the wafer diameter. It is also known that areas with high density copper trenches are polished at a higher rate than near areas with low copper line density. This results in a phenomenon known as 'CMP induced corrosion' in high copper density regions. This is an example of nonuniformity in the 'chip scale' process, i.e., nonuniformity that occurs on a length scale that can be compared (and sometimes smaller) to the linear dimensions of a single chip. Another type of chip scale non-uniformity known as 'dishing' occurs within a single trench region filled with copper (which tends to be polished at a higher rate than the surrounding dielectric material). For trenches larger than a few microns in width, the dishing can be serious, as the affected line will later exhibit excessive electrical resistance resulting in chip failure.

Unevenness in CMP induced wafer and chip scale processes is inherently unpredictable and subject to changes over time depending on conditions within the CMP processing system. In order to ensure that all non-uniformities are kept within acceptable limits, it is often necessary for process engineers to measure surface topography in a noncontact manner frequently for chips in large numbers and a wide range of locations in order to effectively monitor and appropriately adjust process conditions. It is important. This is possible using the embodiment of the interferometric technique described above.

The interferometer embodiments described above include an interference objective lens of the Mirau type and the Linic type. In the Mirau type, the beam splitter in the interference objective lens carries the reference light back along the optical axis of the test light. In the clinic type, the beam splitter is positioned in front of the objective lens with respect to the test surface (for incident light) to transmit the test light and the reference light along different paths. Individual objective lenses are used to focus the reference light onto the reference lens. In other words, the beam splitter separates the input light into the test light and the reference light, and then an individual objective lens focuses the test light and the reference light on the test plane and the reference plane, respectively. Ideally, the two objective lenses match each other, so that the test light and the reference light have similar aberrations and light paths.

In other embodiments, an interferometry system may include a beam splitter that transmits reference light away from the optical axis of the test light (eg, the beam splitter may be oriented at 45 degrees relative to the input light, such that the test light and the reference light move perpendicular to each other). Other types of interference objective lenses can be used, such as the Michelson objective lens. In this case, the reference plane may be located outside the path of the test light.

Other interferometer configurations are possible. For example, the system can be configured to collect test light that passes through the test sample and then combines with the reference light. For these embodiments, for example, the system may implement a Mach-Zehnder interferometer with dual microscope objective lenses on each leg.

The light source in the interferometer can be any of the following: an incandescent source such as a halogen bulb or a metal halide lamp with or without a spectral bandpass filter; Broadband laser diodes; Light emitting diodes; A combination of several light sources of the same or different type; Arc lamps; All sources in the visible spectral range; All sources within the infrared (IR) spectral range, especially for the use of rough surface viewing & phase profiles; And all sources within the ultraviolet (UV) spectral range, especially for improved lateral resolution. For wideband applications, the source preferably has a net spectral bandwidth that is greater than 5% of the average wavelength. Or the source more preferably has a net spectral bandwidth of greater than 10%, 20%, 30% or even 50% of the average wavelength. For adjustable narrowband applications, the range of adjustment is preferably wide to provide information over a wide range of wavelengths (e.g., greater than 50 nm, greater than 100 nm, or even greater than 200 nm for visible light). The spectral width at any particular setting is preferably narrow to optimize resolution, for example as small as 10 nm, 2 nm, or 1 nm. The source may also include one or more diffuser elements to increase the spatial extent of the input light emitted from the source.

In addition, various transition stages of a system such as transition stage 150 can be driven by any of a piezoelectric device, a stepper motor, and a voice coil; Optomechanical, not by pure translation (e.g., by the use of liquid crystals, electro-optical effects, strained fibers, and rotating waveplates) to introduce a change in optical path length. Or photoelectronically; It can be any driver with a flex mount and any driver with mechanical stages, such as roller bearings or air bearings. As mentioned above, phase shifting of the scan interferometry signal is usually performed using a mechanical transition stage, but the wavelength of the source is reduced when the optical path difference between the test section and the reference section is non-zero. By changing, it is also possible to change the phase of either the test section and the reference section of the interferometer.

The electronic detector can be any type of detector for measuring optical interference patterns with spatial resolution, such as multi-element CCD or CMOS detectors.

Any of the computer analysis methods described above may be implemented in hardware or software, or a combination thereof. These methods can be implemented as computer programs using standard programming techniques that follow the methods and numerical values described herein. Program code is applied as input data to perform the functions described herein and generate output information. Output information is applied to one or more output devices, such as a display monitor. Each program is implemented in a high level procedural or object oriented programming language to communicate with a computer system. However, the program may be implemented in assembly or machine language, if desired. In any case, a language can be a language that is compiled or interpreted. The program can also be executed on a dedicated integrated circuit preprogrammed for this purpose.

Each such computer program is preferably stored on a storage medium or device (eg, a ROM or a magnetic disk) readable by a general purpose or special purpose programmable computer, so that when the storage medium or device is read by the computer, Configure and operate the computer to perform the described procedure. The computer program can reside in cache or main memory while executing the program. The analysis method may also be embodied as a computer readable storage medium composed of a computer program, which allows the computer to operate in a specific and predefined manner to perform the functions described herein. The above specific description refers to a scanning interferometry signal in which limited coherence in the system causes localization of the interference fringes; For many embodiments, it is also possible to extract information about the composite surface shape from such interferometric signal (s) without fringe localization.

For example, interferometric signals from different locations of the test object, without fringe localization, can still be used to generate the apparent surface profile of the test object, and the apparent surface profile, or information generated therefrom, Compare the information about such unanalyzed shape as described above for the low coherence scan interferometry signal, compared to models of expected response to different values of the transverse surface shapes of the test object that are not analyzed or ambiguous in the profile. You can decide. Techniques for extracting surface profile information from such "high" coherence interferometry signals are commonly referred to as phase shifting interferometry (PSI) algorithms and are well known in the art. For example, the invention is entitled "(METHOD AND SYSTEM FOR PROFILING OBJECTS HAVING MULTIPLE REFLECTIVE SURFACES USING WAVELENGTH-TUNING PHASE- See the background and contents of US Pat. No. 6,359,692, which is incorporated herein by reference. For such PSI assays, given to measure interferometric data, An interferometric signal for a pixel may be generated by mechanically varying the optical path length difference between the reference and measurement intervals, or by varying the wavelength of light for a fixed, non-zero optical path length difference between the reference and measurement intervals. Can be.

Many embodiments of the invention have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the invention.

Claims (56)

Comparing the information that can be generated from a plurality of scanning interferometry signals corresponding to different surface locations of a test object with information corresponding to a plurality of models for the test object, wherein the plurality of models comprise A comparison step, parameterized by a series of features associated with one or more unanalyzed lateral shapes; And Outputting information about the unanalyzed lateral shape based on the comparison Including, Information that can be generated from the plurality of interferometric signals includes one or more values extracted from a height pofile of the test object generated from the plurality of interferometric signals, The unanalyzed transverse shape is not ambiguous or apparent within the extracted height profile. The method of claim 1, The at least one unanalyzed lateral shape of the test object is one of the pitch, modulation depth, and element width of the unanalyzed patterned lateral structure on the test object. Corresponding to one or more. The method of claim 1, And the one or more unanalyzed lateral shapes of the test object correspond to a modulation depth of at least an unanalyzed patterned lateral structure on the test object. The method of claim 3, And the series of features are different values for the modulation depth. 5. The method of claim 4, The plurality of models are represented by a correlation that maps possible results of information that can be generated from a plurality of interferometric signals to corresponding ones of different values of the modulation depth, And the comparing step includes determining among the different values of the modulation depths that best matches information that can be generated from the plurality of interferometric signals. The method of claim 3, And the modulation depth can be expressed in terms of a bias offset value. 3. The method of claim 2, At least some of the interferometric signals are generated from illumination of the test object whose polarization is oriented with respect to the components of the patterned transverse structure. The method of claim 7, wherein Wherein the polarization is linear polarization aligned orthogonally to the longitudinal direction of the individual components defining the patterned transverse structure. The method of claim 1, The one or more unanalyzed transverse shapes of the test object correspond to one or more of the height and position of a step on the test object. 10. The method of claim 9, The series of features include different values for the location or height of the stairs. delete The method of claim 1, Wherein the test object comprises a patterned transverse structure in which individual components are not ambiguous or apparent in the extracted height profile. The method of claim 12, The information generateable from the plurality of interferometric signals is a value for a height of a collection of unanalyzed components in the patterned transverse structure extracted from the height profile. 14. The method of claim 13, The information about the unanalyzed transverse shape corresponds to one or more of the width and modulation depth of a component of the patterned transverse structure. 14. The method of claim 13, The different surface locations for the interferometric signals include a reference portion of the test object that provides a reference height value of the extracted height profile. 16. The method of claim 15, The test object is etched to produce the patterned structure, The reference portion of the test object is known to be unetched of the test object. The method of claim 12, At least some of the interferometric signals that determine the height profile are generated from illumination of the test object with its polarization oriented with respect to the components of the patterned transverse structure. 18. The method of claim 17, Wherein the polarization is linearly polarized light aligned in a direction orthogonal to the longitudinal direction of the individual component defining the patterned transverse structure. The method of claim 1, The height profile is obtained from frequency domain analysis of the interferometric signal. The method of claim 1, And the height profile is obtained from the relative position of the coherence peak in each interferometric signal. The method of claim 1, And the unanalyzed lateral shape of the test object has a feature size of less than 400 nm. The method of claim 1, And the unanalyzed lateral shape of the test object has a shape size of less than 200 nm. The method of claim 1, And the unanalyzed lateral shape of the test object has a shape size of less than 100 nm. The method of claim 1, Wherein the models are generated computationally using rigorous coupled wave analysis (RCWA). The method of claim 1, The models are empirically generated from test objects having known properties. The method of claim 1, Information about the unanalyzed lateral shape is output to the user. The method of claim 1, Information about the unanalyzed lateral shape is output to an automated process control system for semiconductor manufacturing. The method of claim 1, And the interferometric signal is a scanning interferometric signal. The method of claim 28, The scanning interferometry signal comprises imaging light from the test object to interfere with reference light on a detector, and light from the common source to the detector between the interference portions of the test light and the reference light. Generated by varying the path length difference, The test light and the reference light are generated from the common source, And the scanning interferometric signal corresponds to an interference intensity measured by the detector when the optical path length difference is changed. 30. The method of claim 29, Generating the scanning interferometry signal. 30. The method of claim 29, Wherein the test light and the reference light have a spectral bandwidth greater than 5% of a center frequency of the test light and the reference light. The method of claim 31, wherein The common source has a spectral coherence length, And the optical path length difference varies over a wider range than the spectral coherence length to produce the scan interferometry signal. 30. The method of claim 29, Optics used to transfer test light onto the test object to form an image on the detector, the numerical aperture for the test light being greater than 0.8. 34. The method of claim 33, The common source is a spatially extended source. delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete An apparatus comprising a computer readable medium, The computer readable medium includes: And having a processor in the computer to compare information that can be generated from the plurality of scanning interferometry signals corresponding to different surface positions of the test object with information corresponding to the plurality of models for the test object, The plurality of models are parameterized by a series of features related to one or more unanalyzed lateral shapes of the test object, The program causes the processor in the computer to output information about the unanalyzed lateral shape based on the comparison, The information generateable from the plurality of interferometric signals includes one or more values extracted from a height profile of the test object generated from the plurality of interferometric signals, Wherein the unanalyzed lateral shape comprises a computer readable medium that is not ambiguous or apparent in the extracted height profile. A scan interferometry system configured to generate a plurality of scan interferometry signals corresponding to different surface locations of the test object; And An electronic processor coupled to the scan interferometry system to receive the scan interferometry signal and programmed to compare information generated from the plurality of scan interferometry signals with information corresponding to a plurality of models for the test object, The plurality of models are parameterized by a series of features associated with one or more unanalyzed lateral shapes of the test object, and programmed to output information about the unanalyzed lateral shapes based on the comparison. Electronic processor Including, The information generateable from the plurality of interferometric signals includes one or more values extracted from a height profile of the test object generated from the plurality of interferometric signals, Wherein the unanalyzed transverse shape is not ambiguous or apparent within the extracted height profile. delete delete delete delete

Images