Multiple-height, spectral-based metrology

Abstract

Systems and methods for OCD metrology are provided including directing an illumination beam towards a sample, capturing a first spectroscopic measurement at a first height that best focuses a first spectral sub-band and capturing a second spectroscopic measurement at a second height that best focuses a second spectral sub-band. Spectral data from the two measurements are extracted and combined into a stitched spectrogram spanning the full illumination spectrum. The stitched spectrogram is then output for subsequent analysis, such as comparison with modeled spectral data.

Description

MULTIPLE-HEIGHT, SPECTRAL-BASED METROLOGYFIELD OF THE INVENTION

[0001] The present invention relates generally to the field of optical inspection of integrated circuit wafer patterns, and in particular to methods for optical critical dimension (OCD) metrology.BACKGROUND

[0002] Integrated circuits (ICs) are produced on semiconductor wafers through multiple steps of depositing, altering, and removing thin layers, which build up into stacked structures on the wafers. These stacked structures, also referred to as “stacks” or “features,” may be formed in patterns, which, like diffraction gratings, have optical properties.

[0003] Optical critical dimension (OCD) metrology typically applies methods of spectroscopy to assess whether these structures are being fabricated with the intended dimensions and parameters. In particular, spectroscopic methods of scatterometry are used to measure reflected radiation from structures formed during IC production. Common scatterometry methods include spectral reflectometry (SR), spectral ellipsometry (SE) and spectral interferometry (SI). During manufacturing operations, scatterometric measurements are compared to modeled signals that are calculated for the intended structures, to determine whether a significant change is present in the measured structure. A significant change can be typically, but not limited to, a change of greater than 0.5% in the spectrum at any wavelength, across the spectrum. Often parameters are “interpreted” (extracted) from the spectrogram by use of a dedicated “interpreting” modeling software. In this respect, a significant change can be, but is not limited to, more than 0.1% of the nominal value of the parameter, or a change of more than 10% of the allowed process control variation range corresponding to the specific parameter at hand. For example, if the nominal value is 1000Angstrom, the first limit set above would be 1 Angstrom but working according to the second limit (process control), if the controlled allowed range for the parameter was for example 1015 Angstrom to 990 Angstrom, the allowed change in the value of the interpreted parameters would be 10%* 25Angstrom= 2.5 Angstrom

[0004] OCD typically uses a focused spectrum of light directed towards small targets of a wafer, for example, targets of less than 30 pm. Chromatic aberration hinders refractive focusing, which often means that more expensive optics must be employed to reduce chromatic aberration. However, even with high quality optical elements, tradeoffs must typically be made between desired optical characteristics, such as chromatic aberration, transmission, resolution, scatter, stray light, thermal stability, and polarization sensitivity. Reducing chromatic aberration without impinging on other desired characteristics is an ongoing goal for spectroscopic metrology.SUMMARY

[0005] Embodiments of the present invention provide systems and methods for use in optical critical dimension (OCD) metrology. The methods and novel system typically include evaluating illumination spot size as a function of wavelength, at multiple heights of a sample under inspection, to determine a set of optimal heights for acquiring spectrographic measurements, where each height results in optimal spot sizes for a sub -band of the full spectral range used for inspection.

[0006] The approach maintains a reduced spot size across the full spectrum, or across the “significant part of the spectrum for the application at hand”, thereby improving the resolution that can be inspected by OCD metrology, particularly when using optics that exhibit a focal-shift variation across ultraviolet, visible, and infrared wavelengths.BRIEF DESCRIPTION OF DRAWINGS

[0007] For a better understanding of various embodiments of the invention and to show how the same may be carried into effect, reference is made, by way of example, to the accompanying drawings. Structural details of the invention are shown to provide a fundamental understanding of the invention, the description, taken with the drawings, making apparent to those skilled in the art how the several forms of the invention may be embodied in practice.

[0008] Fig. 1 is a schematic diagram of a spectroscopic system for optical critical dimension (OCD) metrology, configured to acquire spectral measurements at multiple sample heights and to generate a stitched spectrogram from the acquired measurements, in accordance with an embodiment of the present invention.

[0009] Figs. 2A and 2B are graphs illustrating focal shift behavior of an optics assembly as a function of wavelength, showing variation in optimal focal height across ultraviolet, visible, and infrared spectral regions.

[0010] Fig. 3 is a graph showing illumination spot size as a function of wavelength for multiple heights relative to a reference height, illustrating wavelength-dependent crossover points at which different heights provide improved focusing performance.

[0011] Fig. 4 is a graph illustrating spot size versus wavelength for spectrograms formed by stitching spectral measurements acquired at two different sample heights, with stitching points selected to minimize a maximum spot size across the spectral range.

[0012] Fig. 5 is a graph illustrating spot size versus wavelength for spectrograms formed by stitching spectral measurements acquired at three different sample heights, demonstrating further reduction of maximum spot size relative to two-height stitching.

[0013] Fig. 6 is a graph of spectrograms, comparing reflectivity measurements acquired without stitching with measurements stitched together from multiple sample heights, thereby showing the enhanced spectral resolution from stitched measurements.

[0014] Fig. 7 is a graph illustrating a merit function versus spatial position when scanning a target feature on a sample, comparing stitched and non-stitched measurements and demonstrating improved metrology performance using stitched spectrograms.

[0015] Figs. 8A and 8B are graphs illustrating spot size variation versus wavelength for an optics assembly exhibiting a monotonically varying focal shift, and showing extension of the usable spectral range through multi-height spectral acquisition and stitching.

[0016] Fig. 9 is a flow chart of a process for multiple-height, spectral acquisition for spectroscopic metrology, illustrating the analysis of spectral sensitivity, the capture of measurements at multiple heights, and the formation of a stitched spectrogram for significant change detection and sample characterization, in accordance with an embodiment of the present invention.DETAILED DESCRIPTION

[0017] Embodiments of the present invention provide systems and methods for use in optical critical dimension (OCD) metrology. Instead of capturing a single metrological image at a given planar location of a sample (e.g., a semiconductor wafer), multiple images are captured at the planar location by adjusting the distance of the sample from the objective lens of the OCD system. The distance is adjusted so that different ranges of the spectrum are in better focus, to maintain a focal spot size within each range to a preset maximum value. A single “stitched” spectrogram is then generated from appropriately cut measurements corresponding to the wavelength ranges obtained at the different heights.

[0018] Fig. 1 is a schematic diagram of a system 20 for spectral interferometry metrology, as is typical for spectral reflectometry. The system typically includes a broadbandor tunable light source 22 that generates light 24 that is focused by an optics assembly 26 into a focused illumination (or “incident”) beam 28, directed toward a target location 30 of a sample 32 (e.g., a semiconductor wafer). As described further herein, the system is configured to acquire spectrographic measurements of the same target location while the relative distance between the sample and the optics assembly may be varied, such that each measurement corresponds to a different sample height and a different focal condition for at least a portion of the illumination spectrum.

[0019] The system further includes a stage 40 configured to position the sample, both in a planar direction, with movement according to x, y coordinates, in the vertical (z) direction, moving the height of the sample higher (i.e., closer to the objective) or lower (i.e., farther from the illumination optics). Alternatively, or additionally, the position of the illumination optics may be controlled to alter the distance at which the focused incident beam (indicated as beam 30) illuminates the sample. Note that the angle of incidence of the beam may be normal or oblique. Typically, for normal incidence, the same optics assembly is used for both illumination and collection.

[0020] The system also includes collection optics configured to direct the signal 42 that is reflected or scattered from the sample to a sensor 44 (e.g., a spectrometer or detector array), which is configured to measure wavelength dependent characteristics of the received light. A control and analysis module, executed by a processor 46, may be configured to process the measured data; it typically compares the measured spectra to modeled or previously measured data to determine whether significant changes indicative of manufacturing variations are present at the illuminated location of the sample. Typically, the processor also controls the mechanical operation of the system, positioning the sample according to a pre-determined program. Typically, the sensor 44 is a spectrometer.

[0021] Due to chromatic aberration of the illumination optics, there is no single, optimal height at which the incident beam may be focused on the sample, but, rather, each wavelength of light will have an optimal focus when the sample is at a different height, that is, a height at which the illumination spot size of that wavelength is minimum. For example, for illumination optics that would generally be regarded as “achromatic,” the optimal focal height can vary by microns to tens of microns over the designated operating range.

[0022] Fig. 2 is a focal shift graph that shows this focal height variation for an optics assembly that would be considered nominally achromatic in the visible region of 380-780 nm, but shows more significant focal shifts towards the extremes of a spectrum indicated as extending from 200 nm (near UV) to 1000 nm (near IR). As indicated in the graph, the zero, reference height is set to provide optimal focus at 380 nm and at approximately 850 nm . It should be noted that the reference height is often set as an average of the focal distance for wavelengths across the desired spectrum. However, as can be seen in the graph, there is a rapid reduction in focal distance in the near IR range, while there is also a drop and then a subsequent rise for wavelengths in the near UV range between 200 and 350 nm. That is, the focal shift curve that is shown in Fig. 2 is non-monotonic. It should be noted that this type of focal shift is merely one example; depending on the illumination optics, the focal shift curve may be monotonic, unimodal, or multi-modal with one or more inflection points. Fig. 8, for example, described further hereinbelow, provides an example of a monotonic focal shift curve.

[0023] Returning to the graph of Fig. 2, due to the significant focal shifts (i.e., defocus) outside of the visible range, in both the near UV and near IR ranges, when the sample is set at the reference height (i.e., z=0), spot sizes of wavelengths outside of the visible range are relatively larger, meaning that the resolution provided by these wavelengths is lower. This resolution difference is not an issue when a pad size under investigation is larger than thespot size of wavelengths across the desired spectral range. However, the spot size difference becomes an issue for smaller pad sizes, because spots that are larger than the targeted pad at certain wavelengths are reflected not only from the pad under investigation, but also from “out-of-pad” elements that surround the pad, which may have very different reflection properties and therefore are effectively noise in the spectroscopic measurement. For some types of measurements, the spectrum of interest can simply be narrowed to exclude parts of a spectrum that are out of focus. However, for many types of spectrographic measurements, the reflected near UV or near IR signals, or both, provide useful spectroscopic information. Consequently, improving the focus, i.e., reducing the spot sizes of these wavelength ranges, can have significant value.

[0024] Fig. 3 is a graph of three curves, showing the variation in spot size by wavelength when the sample is set at three different respective z-axis heights, +40, 0, and -40, relative to the z=0 reference height for the optics assembly described above (the height, as well as the spot size diameter, being measured in microns). It is to be understood that the z-axis height (or “sample height”) is typically measured with respect to the optics assembly’s height when imaging the sample surface, or with respect to the stage height setting, and may alternatively, or additionally, be adjusted by movement of the optics assembly or the stage. More generally, the term “sample height,” or “sample distance” as used herein refers to a distance between an objective lens of the optics assembly and a reference distance to the sample (e.g., the semiconductor wafer under inspection).

[0025] The spot sizes shown in the three spot size curves were calculated for a range of 250 to 950 nm, in increments of 100 nm. In practice, the spot size versus wavelength relationship of the illumination optics may be obtained by various methods that may include, for example, optical modeling of the optics assembly, empirical measurements taken during system characterization, calibration carried out during installation or maintenance, or fromany combination thereof. Based on the modeled, measured, or calibrated spot size, the intersection points between spot size curves for different sample heights are identified. Each such intersection corresponds to a wavelength at which the spot size is the same at the respective heights and marks the transition at which spectral data from one height may be preferentially selected over spectral data from another height. More generally, wavelength sub-bands of the spectrum are defined based on a comparison of illumination spot sizes at different heights, such that for each sub-band one height provides smaller spot sizes than the other height.

[0026] For good spectroscopic resolution across the spectrum, the smallest spot sizes are preferable. As shown in the graph, throughout the visible light range, the spot sizes are smallest with the height set to +40. By contrast, the spot sizes at the two ends of the spectrum, in the UV and IR ranges, are smallest with a height set to -40. Between the middle range and the extremes, the spot sizes are smallest in two small ranges identified as “Ref Min” (i.e., the z=0 reference height minimum). As the curves indicate, different measurement heights provide optimal spot sizes for different sub-bands of the spectrum.

[0027] Cross-over points between the curves of spot size for different measurement heights are marked in the figure as “A”, “B”, and “C ” The intersections marked “A” indicate points where the preference between the zero height and the +40 height switches. The +40 height results in smaller spot sizes for all wavelengths in a central sub-band of the spectrum of approximately 430 nm to 840 nm. The zero height (“reference height”) is preferable to the +40 height in the dual region “sub-band” outside this central sub-band, that is, in the region to the left of the “A” point at 430 nm and in the region to the right of the “A” point at 840 nm.

[0028] Similarly, the intersections marked “B” indicate points where the preference between the -40 height and the +40 height switches, these being points farther to the extremesof the spectrum than the “A” points. Intersections marked “C” indicate points where the preference between the -40 height and the 0 height switches, which are points farther to the extremes than the “A” and “B” points. As described further with reference to Figs. 4 and 5, these points of intersection between the curves indicate where spectrograms derived from spectrographic (i.e., scatterometric) measurements at different heights can be “stitched” together to create combined, higher resolution spectrograms. That is, based on the comparison and combination of spot size curves, boundaries between sub-bands of the spectrum are determined, where each sub-band is associated with the measurement height that provides the smallest illumination spot sizes over that wavelength range. These sub-band boundaries then define how spectral data from different spectrographic measurements are later selected and combined, such that each portion of a “stitched” spectrogram is derived from the measurement height that provides superior focusing for that portion of the spectrum.

[0029] Fig. 4 shows a graph indicating the spot size vs wavelengths that can be achieved when making measurements at the two heights z=40 and z=-40 described above. The combined curve with the preferred height measurements is indicated with the wider black line. The points of intersection of the spot size curves, indicated as points “Bl” and “B2” (both marked as “B” in Fig. 3, described above) are then used as the points for “stitching” a spectrogram that derives its measurements from the superior focusing of the dual height measurement. Combining at the points of intersection is preferrable to other possible stitching points where the spot sizes would not match and where the spot size along one curve would therefore be greater than at the intersection.

[0030] For example, the intersection “Bl” at the left side of the curve is at a point where the wavelength is approximately 400 nm and the spot size is approximately 20 pm along both curves. Splicing at 450 nm would cause the maximum spot size (along the -40 curve) to increase to 22 pm.

[0031] Similarly, the largest spot size of the combined spot size curve is at point “B2” (where the curves intersect at the right side of the curve). Between “Bl” and “B2,” where the z=+40 curve is preferred, the spot sizes are smaller than along the z=-40 curve. At larger wavelengths to the right of “B2” the z= -40 curve is preferred.

[0032] It should be noted that while measurements at multiple heights can generally achieve lower spot sizes across a spectrum than measurements at a single height, the benefit must be balanced with the disadvantage of the additional time required to reposition the sample height and acquire an additional spectrographic measurement. In addition, it should be noted that in the above example the spot size curves for the two heights, z=40 and z=-40, had two intersections, such that the sub-band of the curve at the lower height had two sections, one in the UV range and the other in the IR range. In a further example described below with respect to Fig. 8B, the focal shift curve for the optics assembly is monotonic, and the spot size curves of the two selected heights have only one point of intersection.

[0033] Fig. 5 shows a graph indicating the spot size of wavelengths that can be achieved by combining or “stitching” of spectra measured at all three heights described above, z=40, z=0, and z=-40. As for the two height scenario, the stitching points are chosen as the points where the curves intersect, which minimizes the maximum spot size over the extent of the spectrum. By stitching sections taken from all three heights, the areas referred to above, with respect to Fig. 3, as “Ref Min,” that is, spot sizes at the reference z=0 height, are also used (areas between the “A” and “C” splices). In this manner, spectral data obtained at a third height may be incorporated into the stitched spectrogram by replacing corresponding wavelength portions that would otherwise be selected from the first and second heights. That is, the third height provides improved spot size performance for a defined sub-band. For example, with the RefMin portion of the third sub-band, the maximum spot size in the range of 350-450 nm is reduced from 20 pm to approximately 19 pm. When there is a need toachieve a higher resolution across the spectrum, measurement may also be made at additional heights (that is, four or more) to further reduce remaining peaks in the spot size curve.

[0034] Fig. 6 shows spectrogram curves of reflectivity vs. wavelength for three measurements: first, a measurement of a 15 pm pad with a single-height spectroscopic measurement, second, a measurement of the 15 pm pad with a dual -height spectroscopic measurement, with the spectra stitched as described above, and third, a measurement of a 40 pm pad with the same pattern as the 15 pm pad, over a larger area. Because the 40 pm pad has the same pattern as the 15 pm pad, the reflected signal for both size pads would be the same if the spot sizes for all wavelengths were smaller than 15 pm. However, as shown in Figs. 3-5, described above, spot sizes across the spectrum, at any of the three heights described, are significantly less than 40 pm, but greater than 15 pm. Consequently, at any of these heights, there are no “out-of-pad” reflections from a 40 pm pad, but there are out-of- pad reflections at 15 pm, and these are significantly greater for the single-height measurement. By contrast, in the example shown, the effect of out-of-pad reflections is particularly noticeable in the UV and IR ranges for the single height measurement of a 15 pm pad. Due to the larger spot sizes in the UV range, the resultant reflection from the pad is significantly increased, as the out-of-pad area, in this instance silicon based, is more reflective than the pad itself, in this wavelength range, where the pad is copper based. At the same time, the reflection of the pad is decreased in the IR range, as the silicon based out-of- pad elements are less reflective than the pad itself, copper based, in this wavelength range (IR). Therefore, in this example of Fig 6, one sees that the stitched spectrogram is closer to the theoretical spectrum than the single height, z=0, spectrogram. Hence, conclusions drawn on process variations performed on the 15 micron pad would be more accurate (less flawed) if they were based on the analysis of the stitched 15 pm spectrogram, not the z=0 single focus based spectrogram.

[0035] Larger spot sizes at certain wavelengths cause reflections to be measured from areas adjacent to the target pad that are not part of the original modelling. Consequently, these out-of-pad reflections contribute noise to the measurement of the target location and reduce the sensitivity to manufacturing changes. For example, if a portion of the spectrum has a 30 micron spot size, then, when measuring a 20 micron pad, this portion will result in reflective signals that may appear anomalous with respect to a modelled or predetermined spectral signal (e.g., a spectrogram) against which the stitched spectrogram is compared for quality control during manufacturing. Analysis (interpretation) of the spectrogram with a predetermined spectral signal may also indicate structural parameters, such as layer height, at the target location. Thus, comparison with a “noisy,” large spot imaging a small pad, spectrogram may lead to incorrect conclusions of process variations

[0036] Fig. 7 provides another view of the improved measurement achieved with stitching, in this case when scanning a 20 micron pad, in both the x and y directions. The graph shows a merit function for the Euclidian norm of error vs the movement along the (transverse) y axis. The merit function, indicative of the error (i.e., the theoretical spectrum of the pad minus the measured spectrum- where the measured spectrum may contain portions from the “out of pad”), is lower, particularly when the pad center is the focus of illumination.

[0037] Figs. 8A and 8B show graphs of spot size variation vs wavelength for a different optics assembly. In this example, the optics assembly has a monotonically increasing focal shift, shown as the hash-mark line. More generally, the optics assembly may exhibit various patterns of focal shift, such as the concave focal shift of Fig. 2, or the monotonically increasing focal shift of Figs. 8A and 8B. The focal shift behavior in turn influences the number of measurement heights that may be selected and the allocation of wavelength subbands.

[0038] In Fig. 8A, there is only a single minimum spot size for any given height. The solid line curve of Fig. 8A shows the spot size variation for a single measurement at a height set at a 500 nm reference. Note that below 350 nm the spot size rises rapidly, effectively limiting the appropriate spectrum to a lower wavelength of 350 nm for pad sizes of 30 pm (though lower wavelengths, with larger spot sizes, can be used for larger pad sizes).

[0039] For a two-height measurement, adding an additional measurement at focal height optimal for the 300 nm wavelength, as shown in Fig. 8B, allows extension of the spectrum measured into the UV range, providing a spot size of no more than 30 pm at a wavelength as low as 280 nm. Adding a third height measurement at the focal height 250 nm wavelength (not shown) would extend the useful spectrum (for pads of 20 pm or less) down to approximately 200 nm. Again, the number of heights can be increased, but with a proportional increase in the time required to perform the measurements.

[0040] Fig. 9 shows a flow chart of a process 900 for multiple-height, spectral acquisition for spectroscopic metrology. At a step 902, the relationship between illumination spot size, sensitivity to changes of parameters of interest, and wavelength is analyzed to evaluate spectral sensitivity to changes in sample height. This analysis may be based on modeled, measured, or calibrated spot size versus wavelength responses of the optics assembly at various vertical positions. By evaluating these relationships, the system determines how height or focal length adjustments influence focusing performance for specific wavelengths, particularly in terms of minimizing spectral error or parameter drift based on the specific product or layer being inspected. This step facilitates the selection of the optimal number of heights — which may be one, two, or three, or more — and the specific vertical coordinates required for the tool and the layer at hand. This can also be done for different wafers of interest, where each wafer has a different process variation that is to be measured and interpreted by a different range of sub-bands. The process can also becustomized so that sub-bands that are particularly useful for specific types of spectral analysis have smaller spot sizes. That is, if quality control limitations limit the number of heights for reasons such as throughput rate, the heights can be optimized for sub-bands of interest. For example, if a particular inspection test is particularly sensitive to UV wavelengths, spectral measurements can be performed with heights that enhance the spot size resolution in the UV range.

[0041] At a step 904, wavelength sub-bands and their corresponding stitching points are defined by identifying intersection points between the spot size curves. These intersections occur at wavelengths where the illumination spot sizes are substantially identical for at least two different sample heights. Alternatively, they may split the “important part of the spectrum” into these sub bands, ignoring areas of the spectrum that are less important to analysis. These boundaries divide the full illumination spectrum, or the “important part of the spectrum” so that if, for example, there are no reflections in the UV range, the above analysis would be applied only to the VIS+IR region. The process divides the spectrum into distinct sub-bands, where each sub-band is assigned to the specific measurement height that provides the smallest illumination spot size for that wavelength range. This selection process ensures that the maximum “effective” spot size of the eventually combined spectrogram is minimized compared to any single-height measurement.

[0042] At a step 906, a first spectrographic measurement is captured while the sample is positioned at a first height relative to the optics assembly. The illumination beam, which contains a spectrum ranging from a lower wavelength to an upper wavelength, is directed toward a target location on the sample. The reflected or scattered signal is collected by a sensor, such as a spectrometer or detector array, to measure wavelength-dependent characteristics while the sample is at this initial focal condition.

[0043] At a step 908, additional spectrographic measurements, such as a second and potentially a third or more measurements, are captured at different sample heights. The distance between the sample and the optics assembly is adjusted using a stage or positioning mechanism to move the sample to the heights selected during step 902. Each subsequent measurement focuses a different portion of the spectrum, such as the ultraviolet or infrared extremes, ensuring that every wavelength sub-band is captured at a height that provides superior focus for that specific range.

[0044] At a step 910, a stitched spectrogram is formed by extracting and combining spectral data from the different captured measurements. The processor utilizes the previously defined sub-band boundaries and stitching points to select the optimal data from each height. For example, spectral data from a second or third height may be used to replace corresponding portions of a first measurement where the alternative heights provide improved spot size performance. The resulting combined spectrogram maintains a reduced spot size across the entire spectral range, thereby enhancing spectral resolution.

[0045] At a step 912, the stitched spectrogram may be provided to spectral analysis applications known in the industry, such as applications to detect “process variations’7anomalies at the target location. For example, the stitched spectrogram may be compared with a previously determined control spectrogram, which may be modeled or determined empirically, to determine whether a sample anomaly is present. Detecting such anomalies may include determining at least one dimensional or structural parameter of the sample, though the comparison with the reference spectrogram is always performed to identify manufacturing faults. By using high-resolution data that minimizes out-of-pad reflections and noise, the system achieves greater sensitivity to manufacturing faults in small targets compared to traditional single-height acquisition methods.

[0046] It is to be understood that processing elements within the metrology system are implemented using one or more computers or processing devices, which may include central processing units (CPUs), graphics processing units (GPUs), or other specialized processing circuitry. These processors execute computer-readable program instructions, which may be written in various programming languages or provided as microcode, and which perform the described metrology operations. In certain embodiments, the system may utilize programmable logic circuitry, such as field-programmable gate arrays (FPGAs) or programmable logic arrays (PLAs), which are personalized by state information to execute the specific logical functions required for spectral stitching and anomaly detection.

[0047] Instruction storage is provided by non-transitory, computer-readable storage media that are capable of retaining program instructions, such as random access memory (RAM), read-only memory (ROM), flash memory, and solid-state or magnetic storage devices. Such media store the instructions necessary to control mechanical system operations, process measured spectral data, and facilitate the comparison of stitched spectrograms against quality control models to identify manufacturing faults.

[0048] Where aspects of the invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention, it will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented or operated by computer readable program instructions.

[0049] It is to be understood that each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions. The functions noted in the blocks may occur out of the order shown herein. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may be executed in the reverseorder, depending upon the functionality involved. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.

[0050] EXAMPLES

[0051] Examples of the present invention may include the following configurations.

[0052] An example 1 of the present invention is a method for spectral-based metrology. The method includes directing an illumination beam toward a target location of a sample and capturing a first spectrographic measurement when the sample is positioned at a first height with respect to an optics assembly. The illumination beam typically includes a spectrum ranging from a lower wavelength to an upper wavelength. The method further includes directing the illumination beam toward the target location of the sample and capturing a second spectrographic measurement when the sample is positioned at a second height with respect to the optics assembly. First and second sub-bands of the spectrum are selected such that illumination spot sizes of wavelengths in the first sub-band, when the sample is at the first height, are less than illumination spot sizes in the first sub-band when the sample is at the second height, and illumination spot sizes of wavelengths in the second sub -band, when the sample is at the second height, are less than illumination spot sizes in the second subband when the sample is at the first height. The method further includes forming a stitched spectrogram that combines spectral data of the first sub-band captured in the first spectrographic measurement with spectral data of the second sub-band captured in the second spectrographic measurement. The stitched spectrogram may then be output to spectral analysis applications, such as applications to compare the spectrogram with a previously determined spectral signal to determine a structural parameter of the sample or presence of a sample anomaly or process variation.

[0053] An example 2 of the present invention includes the features of example 1 and further includes selecting the first and second sub-bands by identifying points of intersection at one or more wavelengths of the spectrum. At these wavelengths, illumination spot sizes are substantially the same at both the first height and the second height.

[0054] In an example 3, which may include the features of either one of examples 1 or 2, the first height and the second height are selected such that a maximum spot size of a combined spot size curve, formed from the first sub-band measured at the first height and the second sub-band measured at the second height, is less than a maximum spot size of the spectrum measured at any single height.

[0055] An example 4 of the present invention includes the features of and one of examples 1 to 3 and further includes determining and optimizing the illumination spot sizes based on a modeled, measured, or calibrated spot size versus wavelength response of the optics assembly at each height. This can be based on a general standard silicon wafer, or optimized for the layer measured, as described above with respect to step 902 in Fig 9.

[0056] An example 5 of the present invention includes the features of any one of examples 1-4 and further includes capturing a third spectrographic measurement when the sample is positioned at a third height with respect to the optics assembly. The third height provides illumination spot sizes that, over wavelengths in a third sub-band, are less than illumination spot sizes in the third sub-band when the sample is at the first height and the second height. The stitched spectrogram is formed using spectral data of the third sub-band from the third spectrographic measurement in place of corresponding portions of the first and second sub-bands. The same process may also be performed for additional heights, that is, for four or more heights.

[0057] In further examples, the optics assembly may, for example, have a focal shift range of less than 200 pm over a wavelength range of approximately 200 to 1000 nm. Insome implementations, the optics assembly may exhibit a concave focal shift or a monotonically increasing focal shift.

[0058] It is also to be understood that a further example of the present invention may be a metrology system configured to implement any of the above method examples. In addition, a further example of the present invention is non-transitory computer readable medium storing instructions to implement any of the above method examples.

Claims

CLAIMS1. A method for spectral-based metrology comprising: directing an illumination beam toward a target location of a sample and capturing a first spectrographic measurement when the sample is positioned at a first height with respect to an optics assembly, wherein the illumination beam includes a spectrum ranging from a lower wavelength to an upper wavelength; directing the illumination beam toward the target location of the sample and capturing a second spectrographic measurement when the sample is positioned at a second height with respect to the optics assembly; selecting first and second sub-bands of the spectrum, such that illumination spot sizes of wavelengths in the first sub-band, when the sample is at the first height, are less than illumination spot sizes in the first sub-band when the sample is at the second height, and illumination spot sizes of wavelengths in the second sub-band, when the sample is at the second height, are less than illumination spot sizes in the second sub-band when the sample is at the first height; forming a stitched spectrogram combining spectral data of the first sub-band captured in the first spectrographic measurement with spectral data of the second sub-band of the spectrum captured in the second spectrographic measurement; and providing the stitched spectrogram for one or more applications of spectral analysis.

2. The method of claim 1, wherein selecting the first and second sub-bands further comprises selecting their points of intersection at one or more wavelengths of the spectrum at which spot sizes at those wavelengths are the same at both the first and second heights.

3. The method of claim 1, further comprising selecting the first and second heights such that the maximum spot size of a combined spot size curve of the first sub -band measured at thefirst height and of the second sub-band measured at the second height is less than the maximum spot size of the spectrum for any single height.

4. The method of claim 1, further comprising determining the spot sizes by a modeled, measured, or calibrated spot size versus wavelength response of the optics assembly at each height.

5. The method of claim 1, further comprising capturing one or more additional spectrographic measurements when the sample is positioned at respective one or more additional heights with respect to the optics assembly, the one or more additional heights providing illumination spot sizes that, over wavelengths in respective one or more sub-bands, are less than illumination spot sizes in those one or more sub-bands when the sample is at the first and second heights, and further comprising forming the stitched spectrogram with spectral data of the one or more sub-bands from the one or more additional spectrographic measurements in place of those sub-bands in the first and second sub-bands.

6. The method of claim 1, wherein the optics assembly has a focal shift range of less than 200 pm over a wavelength range of 200 to 1000 nm.

7. The method of claim 1, wherein the optics assembly has a concave or monotonically increasing focal shift.

8. The method of claim 1, wherein the one or more applications of spectral analysis include: comparing the stitched spectrogram with a previously determined spectral signal to determine one or more structural parameters of the sample or presence of a sample anomaly.

9. A system for spectral-based metrology, comprising: an illumination source configured to generate an illumination beam including wavelengths of a spectrum ranging from a lower wavelength to an upper wavelength;an optics assembly configured to direct the illumination beam toward a target location of a sample; a stage and / or an optics positioning mechanism configured to adjust a distance between the sample and the optics assembly so as to position the sample at a first height with respect to the optics assembly during a first spectrographic measurement and to position the sample at a second height with respect to the optics assembly during a second spectrographic measurement; and a processor coupled to a sensor receiving reflected signals from the sample to make the first and second spectrographic measurements, the processor executing instructions that: select first and second sub-bands of the spectrum, such that illumination spot sizes of wavelengths in the first sub-band, when the sample is at the first height, are less than illumination spot sizes in the first sub-band when the sample is at the second height, and illumination spot sizes of wavelengths in the second sub-band, when the sample is at the second height, are less than illumination spot sizes in the second subband when the sample is at the first height; and form a stitched spectrogram combining spectral data of the first sub -band of the spectrum captured in the first spectrographic measurement with spectral data of the second sub-band of the spectrum captured in the second spectrographic measurement.

10. The system of claim 9, wherein first and second heights are selected such that the maximum spot size of a combined spot size curve of the first sub-band measured at the first height and of the second sub-band measured at the second height is less than the maximum spot size of the spectrum for any single height.

11. The system of claim 9, wherein the processor is further configured to determine the spot sizes from a modeled, measured, or calibrated spot size versus wavelength response of the optics assembly at each height.

12. The system of claim 9, wherein the configuration of the stage and / or optics positioning mechanism is set to adjust the distance to a one or more additional heights for capturing respective one or more spectrographic measurements, the one or more additional heights providing illumination spot sizes that, over wavelengths in respective one or more sub-bands, are less than illumination spot sizes in those sub-bands when the sample is at the first and second heights, and wherein the processor is further configured to form the stitched spectrogram with spectral data of the one or more sub-bands from the one or more additional spectrographic measurements in place of those sub-bands in the first and second sub-bands.

13. The system of claim 9, wherein the optics assembly has a focal shift range of less than 200 pm over a wavelength range of 200 to 1000 nm.

14. The system of claim 9, wherein the optics assembly has a concave or monotonically increasing focal shift.

15. The system of claim 9, wherein the processor is further configured to output the stitched spectrogram to one or more applications of spectral analysis include comparing the stitched spectrogram with a previously determined spectral signal to determine one or more structural parameters of the sample or presence of a sample anomaly.

16. A non-transitory computer readable medium that stores instructions for spectral -based metrology, for characterizing a semiconductor sample, the instructions configured to execute steps of:receiving a first spectrographic measurement from a sample when the sample is positioned at a first height and receiving a second spectrographic measurement from the sample when the sample is positioned at a second height; selecting first and second sub-bands of the spectrum, such that illumination spot sizes of wavelengths in the first sub-band, when the sample is at the first height, are less than illumination spot sizes in the first sub-band when the sample is at the second height, and illumination spot sizes of wavelengths in the second sub-band, when the sample is at the second height, are less than illumination spot sizes in the second sub-band when the sample is at the first height; forming a stitched spectrogram combining spectral data of the first sub -band of the spectrum captured in the first spectrographic measurement with spectral data of the second sub-band of the spectrum captured in the second spectrographic measurement.

17. The non-transitory computer readable medium of claim 16, wherein the first and second heights are selected such that the maximum spot size of a combined spot size curve of the first sub-band measured at the first height and of the second sub-band measured at the second height is less than the maximum spot size of the spectrum for any single height.

18. The non-transitory computer readable medium of claim 16, wherein the instructions further comprise capturing one or more additional spectrographic measurements when the sample is positioned at respective one or more additional heights with respect to the optics assembly, the one or more additional heights providing illumination spot sizes that, over wavelengths in respective one or more sub-bands, are less than illumination spot sizes in those one or more sub-bands when the sample is at the first and second heights, andfurther comprising forming the stitched spectrogram with spectral data of the one or more sub-bands from the one or more additional spectrographic measurements in place of those sub-bands in the first and second sub-bands.

Images