Performance enhancements for overlay metrology

Abstract

To provide an improved device and method for semiconductor circuit metrology.SOLUTION: At least first and second patterned layers are successively deposited on a device, a first target feature is included in a first patterned layer, a second target feature is included in a second patterned layer, and at least one illumination beam is directed at a semiconductor wafer having a second target feature overlying the first target feature. A sequence of images of the first and second target features is captured while varying one or more imaging parameters throughout the sequence. The images in the sequence are processed to identify centers of symmetry for each of the first and second target features in the images and to measure variations in the centers of symmetry as a function of the varying imaging parameters, and the variations are applied in measuring overlay error between the first and second patterned layers.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates generally to semiconductor device manufacturing, and more particularly to an apparatus and method for semiconductor circuit metrology. [Background technology]

[0002] Semiconductor circuits are commonly fabricated using photolithographic methods. In photolithography, a thin layer of photosensitive polymer (photoresist) is deposited over a semiconductor wafer and then patterned using optical or other radiation, leaving the portions of the wafer covered by the photoresist. After patterning, the wafer can be modified by etching, ion bombardment, or other methods to alter the wafer's material properties and topography while leaving the portions of the wafer covered by the photoresist unaffected.

[0003] Semiconductor circuit metrology is widely used to measure the properties of patterned photoresist, such as the topography and location of patterned features. Accurate location of patterned features in photoresist relative to previous process layers is critical to ensuring high yield in photolithographic processes. Any error (misregistration) in the alignment (registration) of patterned photoresist relative to the underlying process layer is called "overlay error." As an example, for a typical semiconductor circuit with a minimum linewidth of 10-14 nm (so-called 10-nm design rule), the maximum allowable overlay error is 2-3 nm. As linewidths in cutting-edge semiconductor circuits are shrinking to 5 nm, the maximum allowable overlay error is decreasing accordingly.

[0004] Because optical radiation in the visible and near-infrared wavelengths can penetrate the photoresist layer and the dielectric layer below the photoresist, overlay error is typically measured using optical overlay metrology tools. Optical overlay metrology tools, such as the Archer™ series tools from KLA Corporation (Milpitas, California, USA), image proxy targets (e.g., KLA's AIM™ overlay targets) located within the scribe lines (the lines separating adjacent semiconductor chips) of a semiconductor wafer. Image analysis algorithms are applied to the acquired images to locate the centers of symmetry (CoS) of a set of target features in the process layer and the CoS of corresponding target features in the patterned photoresist layer. The overlay error is calculated as the distance between the centers of symmetry of the target features in the two layers.

[0005] The terms "optical ray," "optical radiation," "light," and "beam of radiation," as used herein and in the claims, generally refer to visible, infrared, and ultraviolet radiation in general. Summary of the Invention [Problem to be solved by the invention]

[0006] In the embodiments of the present invention described below, improved apparatus and methods for semiconductor circuit metrology are provided. [Means for solving the problem]

[0007] That is, one embodiment of the present invention provides a metrology method in which a semiconductor wafer having at least first and second patterned layers deposited thereon, the semiconductor wafer having a first target feature in the first patterned layer and a second target feature in the second patterned layer, the second target feature overlying the first target feature, is illuminated by directing at least one illumination beam. A sequence of images of the first and second target features is captured while varying one or more imaging parameters throughout the sequence. The images in the sequence are processed to identify centers of symmetry for the first and second target features in the images, and variations in the centers of symmetry are measured as a function of the varying imaging parameters. The measured variations are then applied to measure overlay error between the first and second patterned layers.

[0008] In one disclosed embodiment, the first patterned layer includes a process layer and the second patterned layer includes a resist layer deposited above the process layer.

[0009] In certain embodiments, capturing the sequence of images involves capturing first and second images of the target feature using first and second cameras in mutual registration, and processing the images to measure the variation in the center of symmetry by comparing the first and second images. In one embodiment, capturing the first and second images involves projecting a registration image toward the first and second cameras, and aligning the first and second cameras with the registration image. In one exemplary embodiment, projecting the registration image involves generating a grid pattern aligned with the image of the target feature and projecting it onto the detector arrays of the first and second cameras, respectively.

[0010] Additionally or alternatively, when capturing the first and second images, the one or more imaging parameters are set to a first setting for the first image and to a second setting for the second image. In some disclosed embodiments, when setting the one or more imaging parameters, the first and second cameras are set to respective first and second focal positions, and the variation of the center of symmetry is measured as a function of focal position by stepping the first and second cameras through respective first and second sequences of the first and second focal positions. In some exemplary embodiments, the first and second focal positions are separated by a constant focal length Δz, and both the first and second focal positions are incremented by Δz at each step in the first and second sequences.

[0011] Additionally or alternatively, the at least one illumination beam may include first and second illumination beams having respective first and second polarization states, and a polarizing beam splitter may be applied when capturing the first and second images such that light reflected from the wafer in the first polarization state is directed towards a first camera and light reflected from the wafer in the second polarization state is directed towards a second camera.

[0012] In one embodiment, images of the target feature are captured at various focus settings when capturing a sequence of images, and when processing the images, the variation in the center of symmetry is measured as a function of focus setting.

[0013] Additionally or alternatively, images of the target feature are captured at a plurality of different wavelengths when capturing a sequence of images, and when processing the images, the variation in the center of symmetry is measured as a function of wavelength.

[0014] Additionally or alternatively, when capturing a sequence of images, images of the target feature are captured at a plurality of different polarization states, and when processing the images, the variation in the center of symmetry is measured as a function of polarization state.

[0015] Additionally or alternatively, when capturing a sequence of images, images of the target feature are captured with at least one aperture of at least one illumination beam at a plurality of different offsets, and when processing the images, the variation in the center of symmetry is measured as a function of the aperture offset.

[0016] Additionally, in one embodiment, images of the target feature are captured using a camera and at various angular orientations of the semiconductor wafer relative to the camera when capturing a sequence of images, and when the images are processed, tool-induced shifts of the center of symmetry are measured as a function of angular orientation.

[0017] In certain embodiments, applying the measured variations involves searching for an optimal range for one or more imaging parameters in response to the measured variations, and setting the one or more imaging parameters to values within the optimal range to generate a recipe for measuring overlay errors. In one disclosed embodiment, capturing the sequence of images involves capturing images of a plurality of target features at a plurality of distinct locations on the semiconductor wafer, and searching for the optimal range involves applying the variations measured at the plurality of distinct locations to select an optimal range across an area of the semiconductor wafer.

[0018] Additionally or alternatively, when processing the image, the asymmetry of at least one of the target features is measured.

[0019] An embodiment of the present invention also provides a metrology method in which a semiconductor wafer having at least one patterned layer deposited thereon and comprising a grating having a plurality of bars oriented parallel to a predetermined axis is illuminated by directing at least one illumination beam, one or more images of the grating are captured and processed to resolve asymmetry of one or more of the bars about that axis, and the resolved asymmetry is applied in a metrological evaluation of the patterned layer.

[0020] In one disclosed embodiment, a sequence of images of the grating is captured at various focus settings when capturing one or more images, and when processing the one or more images, a variation in the center of symmetry of the grating in the images is measured as a function of focus setting, and asymmetry is determined based on the measured variation. Additionally or alternatively, when processing the one or more images, a correlation is calculated between the image of the one or more bars and a reflected version of that image, and a measure of asymmetry is derived based on the calculated correlation.

[0021] Additionally, one embodiment of the present invention provides an optical inspection apparatus including an illumination assembly configured to direct at least one illumination beam to illuminate a semiconductor wafer having at least first and second patterned layers deposited thereon, the semiconductor wafer having a first target feature in the first patterned layer and a second target feature in the second patterned layer, the second target feature overlying the first target feature; an imaging assembly configured to capture a sequence of images of the first and second target features; and a controller configured to vary one or more imaging parameters of the apparatus over the sequence, process the images in the sequence to identify centers of symmetry of the first and second target features in the images, measure variations in the centers of symmetry as a function of the varying imaging parameters, and apply the measured variations in measuring overlay error between the first and second patterned layers.

[0022] Additionally, an embodiment of the present invention provides an optical inspection apparatus including an illumination assembly configured to direct at least one illumination beam to illuminate a semiconductor wafer having at least one patterned layer deposited thereon and including a grating having a plurality of bars oriented parallel to a predetermined axis, an imaging assembly configured to capture one or more images of the grating, and a controller configured to process the one or more images to resolve asymmetry of one or more of the bars about the axis and apply the resolved asymmetry in performing a metrological evaluation of the patterned layer.

[0023] The present invention will be more fully understood from the following drawings in conjunction with the following detailed description of various embodiments thereof. [Brief explanation of the drawings]

[0024] [Figure 1] 1 is a schematic side view of an optical inspection apparatus for measuring optical properties of patterned thin film layers on semiconductor wafers according to an embodiment of the present invention; [Figure 2] FIG. 2 is a schematic side view of a grating projector used in the optical inspection device according to one embodiment of the present invention. [Figure 3] FIG. 2 is a schematic representation of an image captured by a camera in an optical inspection apparatus according to an embodiment of the present invention. [Figure 4] 1 is a flow chart that schematically illustrates a process for measuring CoS variations of process layer and resist layer features of an overlay metrology proxy target, according to one embodiment of the present invention. [Figure 5] FIG. 5 is a plot that schematically illustrates the focus separation between the two cameras in the process of FIG. 4, according to one embodiment of the present invention. [Figure 6A] 1 is a flow chart that schematically illustrates a process for measuring CoS variation of features of an overlay metrology proxy target in association with illumination wavelength, according to an embodiment of the present invention. [Figure 6B]1 is a flow chart that schematically illustrates a process for measuring CoS variation of features of an overlay metrology proxy target in association with illumination wavelength, according to an embodiment of the present invention. [Figure 7A] 1 is a flow chart that schematically illustrates a process for generating CoS landscapes and evaluating sensitivity across focus settings and wavelengths, according to one embodiment of the present invention. [Figure 7B] 1 is a flow chart that schematically illustrates a process for generating CoS landscapes and evaluating sensitivity across focus settings and wavelengths, according to one embodiment of the present invention. [Figure 8A] FIG. 1 is a schematic representation of the landscape of tool-induced shift of CoS (CoS_TIS) in a resist layer according to an embodiment of the present invention. [Figure 8B] FIG. 2 is a schematic representation of a corrected CoS (CoS_COR) landscape in a resist layer according to an embodiment of the present invention. [Figure 8C] FIG. 2 is a schematic representation of a CoS_TIS landscape at the process layer according to one embodiment of the present invention. [Figure 8D] FIG. 2 is a schematic representation of a CoS_COR landscape at the process layer according to one embodiment of the present invention. [Figure 9A] FIG. 1 is a schematic representation of the precision landscape at a resist layer according to an embodiment of the present invention. [Figure 9B] FIG. 1 is a schematic representation of the precision landscape at the process layer according to an embodiment of the present invention. [Figure 10] FIG. 10 is a plot illustrating the variation of CoS_TIS as a function of aperture offset, according to one embodiment of the present invention. [Figure 11A] FIG. 2 is a schematic representation of a proxy target image acquired by a camera in an optical inspection apparatus according to an embodiment of the present invention. [Figure 11B] FIG. 9B is a schematic cross-sectional view of a grating bar in the proxy target of FIG. 9A according to one embodiment of the present invention. [Figure 12]10A-10C are a series of plots that schematically illustrate the use of image signal correlation to monitor feature asymmetry in overlay proxy targets, according to one embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0025] [Overview] Overlay metrology proxy targets are commonly used for precision measurement of overlay between successive patterned layers on a semiconductor wafer, which may include, for example, a process layer and a resist layer, or may be used between two process layers in post-etch applications (i.e., although exemplary embodiments are described below in the context of a process layer and a resist layer, the principles of these embodiments can be applied mutatis mutandis to a first and second process layer).

[0026] However, features in proxy targets (both resist layer target features and process layer target features) differ from their corresponding features in the device area: proxy target features typically have wider lines than their counterparts in the device, allowing them to be resolved by metrology tools operating with light in the visible or near-infrared spectrum; and because these targets are symmetrical by design, overlay values can be calculated by implementing powerful symmetry-based image processing algorithms. Furthermore, proxy targets are typically placed within the scribe line of a semiconductor wafer to avoid taking up valuable "real estate" in the device area. The optical distortions of the photolithographic exposure system (scanner) are different within the scribe line than within the device area, leading to a spatially varying differential shift between the pattern in the proxy target and the corresponding pattern in the device.

[0027] These design and metrology considerations cause the proxy target features to respond differently to lithographic and process effects than the device features in the chip area, and the overlay error measured based on the proxy target may have an offset relative to the overlay error of the real device features. By applying a calibration function, an accurate measure of the overlay error in the device area may be derived from the overlay error measured based on the proxy target. However, accurate calibration requires stable and repeatable overlay measurements based on the proxy target, which in turn are subject to process-induced effects, such as feature asymmetry.

[0028] Furthermore, because each proxy target includes both target features in the photoresist and target features in the previous process layer, the two sets of target features can be separated by a distance of up to several microns along a direction perpendicular to the semiconductor wafer. The two sets of target features are then imaged by focusing the metrology tool separately on the resist layer and the process layer and acquiring images at their respective focus settings. However, the combination of process variation effects and topography makes it difficult to find an optimal metrology "recipe," i.e., a set of metrology conditions (e.g., focus, illumination numerical aperture, and wavelength) that produces stable and repeatable overlay measurement results from the proxy targets.

[0029] The embodiments of the invention described herein address these issues by providing an optical metrology tool and method capable of independently resolving two sets of target features on an optical overlay proxy target. In certain embodiments, the metrology tool's two imaging cameras are aligned with each other and focused on the proxy target at a fixed height difference. By comparing the images captured by the two cameras, variations in the optical properties of the proxy target layer, such as center of symmetry (CoS) variations due to imaging parameters such as focus variations, spectral response, and polarization, are calibrated and corrected.

[0030] In certain embodiments, a projector projects a registration image, e.g., an image of a two-dimensional grid, onto the two cameras. The metrology tool's focus is stepped through successive steps, each equal to the fixed height difference, so that one camera (called CAM1) always arrives at the focus setting where the other camera (CAM2) was during the previous step. When CAM1 arrives at its new focus position, it is aligned with the image acquired by CAM2 when CAM2 was at that focus position. At each focus position, each camera acquires an image of its proxy target. The series of acquired images of each of the two sets of target features on the proxy target are aligned with each other through various focus settings. Based on the two image sequences, the variation of CoS with focus can be calculated for each of the two sets of target features. The stability of CoS with focus indicates the focus settings at which the images of the two sets of target features should be acquired for stable overlay measurements. Stable overlay metrology in turn allows for stable and repeatable calibration of overlay errors in the semiconductor circuit.

[0031] Additionally or alternatively, the variability of the CoS for each set of target features can be mapped as a function of illumination wavelength and / or polarization, and the mapping can be used to achieve stable overlay metrology. That is, in current embodiments, the optimal measurement range is identified in two-dimensional space by wavelength and focus.

[0032] [Explanation of optical inspection equipment] FIG. 1 is a schematic, pictorial illustration of an optical inspection system 10 for measuring optical properties of patterned thin film layers on a semiconductor wafer 12 according to one embodiment of the present invention.

[0033] Optical inspection system 10 includes an imaging assembly 14, an illumination assembly 16, and an optical relay assembly 18. The optical inspection system further includes a grating projector 20, a camera assembly 22, a controller 24, a memory 25, and a table 26 on which semiconductor wafer 12 is placed. The orientation of system 10 and its components is defined according to Cartesian coordinates 28. The same Cartesian coordinates 28 are shown in the corresponding orientation in the following figures. Hereinafter, lowercase letters x, y, and z will be used to represent the three Cartesian coordinate axes, and uppercase letters X, Y, and Z will be used to represent coordinates on those axes.

[0034] Imaging assembly 14 is shown schematically as a single objective lens 30. Alternatively, assembly 14 may include an interference objective (e.g., a Linnik interferometer), a dark field objective, a phase contrast objective, or any other suitable type of objective lens or combination of lenses and / or mirrors.

[0035] The objective lens 30 is typically a compound lens of very high optical quality, with a high numerical aperture (NA), for example, an NA of 0.7 or greater. According to an alternative embodiment, the objective lens 30 has a variable NA, which can be controlled by the controller 24.

[0036] In the illustrated embodiment, the illumination assembly 16 is controlled by the controller 24 and includes two illuminators 15 and 17, each with its own light source 32 and 33, which independently emit optical radiation in respective beams 34 and 35 at one or more discrete, tunable wavelengths, or across one or more continuous spectra in continuous wave (CW) or pulsed form. The light sources 32 and 33 may also emit optical radiation in various polarization states, such as unpolarized, linearly polarized, or circularly polarized radiation.

[0037] Illuminators 15 and 17 further include two separate aperture assemblies 36 and 37 coupled to respective light sources 32 and 33. Aperture assemblies 36 and 37 are actuated by respective actuators 38 and 39, which bring the various apertures in assembly 36 into beam 34 and the various apertures in assembly 37 into beam 35. Actuators 38 and 39 also allow fine adjustment of the individual apertures of each assembly within the plane of the individual apertures. Beams 40 and 41 emitted from illuminators 15 and 17, respectively, are collinearly combined by beam splitter 42 into beam 43. This type of dual illumination assembly enhances the flexibility of apparatus 10 by providing two illuminators to provide independent illumination conditions (e.g., wavelength, polarization, and / or NA) for process and resist layers on wafer 12.

[0038] Alternatively, the illumination assembly 16 may comprise a single illuminator, e.g., illuminator 15, and the illumination conditions for the process layer and the resist layer may be selected by appropriate adjustment of the light source 32 and aperture assembly 36. Still alternatively, the illumination assembly may comprise more than two illuminators, e.g., three or four illuminators, and the beams emerging from the individual illuminators may be combined using appropriate optical devices, e.g., beam splitters.

[0039] Optical relay assembly 18 includes beam splitters 44 and 45, beam splitter assembly 46, and lenses 50 and 52. Beam splitters 47 and 48 of beam splitter assembly 46 can be moved in and out of the optical path of apparatus 10 using actuator 49, as described in more detail below. Camera assembly 22 includes two detector arrays 54 and 56, which may be referred to as "cameras," and are labeled CAM1 and CAM2, respectively. Camera assembly 22 also includes two actuators 58 and 60 that move CAM1 and CAM2, respectively, along the z-axis. While lenses 50 and 52 are shown as single lenses in this illustration, they may instead include multiple lenses and / or mirrors.

[0040] Grid projector 20 is configured to project grid images into CAM1 and CAM2, as will be described in further detail with reference to FIG. 2. Controller 24 is coupled to grid projector 20, memory 25, table 26, light sources 32 and 33, and actuators 38, 39, 49, 58, and 60. Controller 24 typically includes a programmable processor programmed in software and / or firmware to perform the functions described herein, and appropriate digital and / or analog interfaces for connection to other elements of device 10. Alternatively or additionally, controller 24 may include hardwired and / or programmable hardware logic circuitry that performs at least a portion of its functions. While controller 24 is shown in FIG. 1 as a single, monolithic functional block for simplicity, in practice the controller may include multiple interconnected control units and appropriate interfaces for receiving and outputting signals depicted and described in the figures.

[0041] Prior to operation of the optical inspection system 10, the semiconductor wafer 12 is placed on a table 26. During operation, under the control of a controller 24, the table 26 allows the wafer 12 to move along the x-, y-, and z-axes, and to rotate about the z-axis. Movement along the z-axis is referred to as "focusing."

[0042] To illuminate wafer 12, illumination assembly 16 emits beam 43 of optical radiation toward beam splitter 44, which reflects the beam into objective lens 30. Objective lens 30 then focuses beam 43 onto wafer 12. The cross-section that beam 43 exhibits upon exiting illumination assembly 16 in a plane perpendicular to its propagation (the yz-plane) is modified by suitably positioned and aligned apertures in aperture assemblies 36 and 37. These apertures define the shape of beam 43's cross-section—for example, circular, square, or anamorphic—as well as the dimensions of that cross-section. As will be described in more detail below, beam 43 may include two beams having different wavelengths and / or polarization states, the cross-sections of which can be independently controlled by aperture assemblies 36 and 37.

[0043] The apertures in assemblies 36 and 37 are typically conjugate to (and imaged onto) the entrance pupil of objective 30 by additional optics, omitted from the drawing for simplicity; therefore, the cross-section of beam 43 emerging from illuminator assembly 16 defines the numerical aperture (NA) of the optical radiation that illuminates wafer 12. That is, the shape of the illumination can be defined in angular space as, for example, circular, square, or anamorphic, and can vary between the full NA of objective 30 and a fraction of that full NA. In some configurations, the illumination can be limited to NA values above the collection NA of objective 30, enabling dark-field imaging of features on wafer 12.

[0044] Optical radiation illuminating wafer 12 is reflected back by the wafer toward objective lens 30, which images features on the wafer toward camera assembly 22. The reflected radiation is received by objective lens 30 and projected through beam splitters 44 and 45 into beam splitter assembly 46, where it strikes either beam splitter 47 or beam splitter 48, depending on which of the two beam splitters is positioned in its path by actuator 49. Beam splitter 47 in this example is a wavelength-neutral beam splitter, i.e., its reflection and transmission coefficients exhibit the same spectral behavior. Beam splitter 48 is a dichroic beam splitter configured to transmit one spectral band Δλ1, e.g., 380-550 nm, and reflect another (non-overlapping) spectral band Δλ2, e.g., 560-800 nm. That is, when beam splitter 47 is in the optical path, each of cameras CAM1 and CAM2 receives a portion of the reflected radiation across its entire spectrum, while when beam splitter 48 is in the optical path, the radiation spectrum is split so that CAM1 receives radiation within spectral band Δλ1 and CAM2 receives radiation within spectral band Δλ2. Having light source 32 emit optical radiation within spectral band Δλ1 and light source 33 emit optical radiation within spectral band Δλ2 allows for independent control of illumination (including illumination NA) for each of the two layers.

[0045] Alternatively or additionally, one of beamsplitters 47 and 48 may be a polarizing beamsplitter that transmits one polarization state and reflects the orthogonal polarization state. That is, for example, if light sources 32 and 33 emit optical radiation with orthogonal polarization states, radiation from light source 32 would be directed toward CAM 1 and radiation from light source 33 would be directed toward CAM 2. Similar to the spectral splitting of illumination described above, controlling the polarization of the illumination allows for independent control of illumination for each of the two layers. In some embodiments, beamsplitter 48 is a combination dichroic and polarizing beamsplitter.

[0046] Optical radiation transmitted and reflected by the beam splitters in selected assembly 46 is focused onto CAM1 by lens 50 and onto CAM2 by lens 52, respectively. Images of wafer 12 are thus captured by CAM1 and CAM2 and are read out and processed by controller 24.

[0047] 2 is a schematic, pictorial representation of a grating projector 20 according to one embodiment of the present invention. Grating projector 20 projects a grating image toward cameras CAM1 and CAM2 that serves as a position reference between the two cameras. Grating projector 20 includes a light source assembly 80, a single-mode optical fiber 82, a diffraction assembly 84, and a spatial filter 86.

[0048] In this example, light source assembly 80 includes two superluminescent light emitting diodes (sLEDs) 88 and 90, with sLED 88 emitting optical radiation at wavelength λ1 = 450 nm and sLED 90 emitting optical radiation at wavelength λ2 = 750 nm. Light source 80 also includes lenses 92, 94, and 96, and a dichroic beam splitter 98. Other types and wavelengths of light sources may alternatively be used.

[0049] Diffraction assembly 84 includes a high-contrast transmission diffraction grating assembly 100, such as a chrome-on-glass grating, positioned between two lenses 102 and 104. Diffraction grating assembly 100 includes orthogonal gratings that diffract light in both the y and z directions, which together create portions of the grating pattern projected by projector 20.

[0050] The optical radiation emitted by sLEDs 88 and 90 is projected by respective lenses 92 and 94 toward a dichroic beam splitter 98. Beam splitter 98 is configured to pass the optical radiation emitted by sLED 88 and to reflect the optical radiation emitted by sLED 90, so that the radiation emitted by the two sLEDs is combined into a single beam 106. Beam 106 is focused by lens 96 into an input end 108 of single-mode optical fiber 82. The optical radiation transmitted through fiber 82 exits the fiber via its output end 110 into diffraction assembly 84 and is projected by lens 102 as beam 112 toward diffraction grating 100. Because output end 110 is located at the focal plane of lens 102, beam 112 is collimated. Beam 112 is diffracted by grating assembly 100 into parallel diffracted orders 114 and focused by lens 104 onto focal plane 116 .

[0051] The spatial filter 86 is positioned within the focal plane 116 and is configured to transmit only the ±1st orders (among the multi-order diffracted beams 114) resulting from diffraction by the grating assembly 100. This function is detailed in the inset 118, which shows the spatial filter 86 in a yz view, i.e., along the x-axis. The spatial filter 86 comprises a transparent ring 120 on an opaque base 122, formed, for example, by removing chrome from a chrome-on-glass base. The ±1st orders are shown within the ring 120 as squares 124 and 126, respectively, for the radiation emitted by the sLED 88 and sLED 90. The zeroth order is blocked by a central portion 128 of the spatial filter, while orders greater than ±1 are blocked by a peripheral portion 130 of the spatial filter 86.

[0052] After passing through spatial filter 86, the ±1st order diffracted light forms expanding beam 132. Interference between these beams results in a propagating sinusoidal grating (as an interference pattern between the ±1st order diffracted light) that is reflected by beam splitter 45 (FIG. 1) into beam 134. Sinusoidal gratings are described in more detail below in connection with FIG. 3. Beam 134 propagates collinearly with the optical radiation reflected from wafer 12 (FIG. 1) toward cameras CAM1 and CAM2, enabling mutual registration of the two cameras, as described in more detail below.

[0053] The spectral content of the sinusoidal grating propagating as beam 134 depends on which one or both of sLEDs 88 and 90 are excited to emit optical radiation. By matching the emission wavelengths λ1 and λ2 to the spectral characteristics of dichroic beamsplitter 48, one of the wavelengths can be reflected by the beamsplitter while the other is transmitted.

[0054] 3 is a schematic representation of an image 150 acquired by one of cameras CAM1 and CAM2 in accordance with one embodiment of the present invention. Image 150 in this example includes an AIM™ proxy target 152 and four gratings 154, 156, 158, and 160 projected by grating projector 20 along the image of the proxy target. Target features within AIM™ proxy target 152 include four resist gratings 162 oriented in pairs along the x and y axes, and four process layer gratings 164 oriented in pairs along the x and y axes. For clarity, only two of the resist gratings 162 and two of the process layer gratings 164 are shown framed, one oriented along the x axis and one oriented along the y axis.

[0055] Other types of proxy targets may instead use other forms of target features, for example the target feature in a so-called frame-in-frame proxy target comprises a square frame made up of bars.

[0056] To calculate the resist layer-to-process layer overlay error, the controller 24 calculates the X and Y coordinates of the CoS (CoS X,R ,CoS Y,R ) are calculated, and similarly the X and Y coordinates of the CoS of the process layer grid 164 (CoS X,P ,CoS Y,P ) are calculated. The difference between the X and Y coordinates of the symmetry centers (projected back onto the wafer 12) gives the overlay error OVL for each of the X and Y axes. X =(CoS X,R -CoS X,P ),OVL Y =(CoS Y,R -CoS Y,P For simplicity, we will use CoS to represent the two-dimensional vector (CoS X ,CoS Y ) is written.

[0057] Gratings 154, 156, 158, and 160 are projected by grating projector 20 as previously described (FIG. 2). Spatial division and alignment of the gratings is achieved, for example, by dividing the gratings in diffraction grating assembly 100 into two pairs of orthogonal gratings. Gratings 154, 156, 158, and 160 are "seen" by both cameras CAM1 and CAM2 and used by controller 24 (FIG. 1) to align the two cameras with respect to the gratings in both the x and y directions. As will be described in more detail below, this alignment (registration) is an essential part of the process of accurately registering successive images of target 152.

[0058] [Uncovering resist and process target features in overlay metrology proxy targets] The following figures illustrate a variety of methods for measuring CoS variations of target features of an overlay proxy target as a function of various imaging parameters, such as focus and illumination wavelength. For simplicity and clarity, these methods are described in the context of the system structures and components described above and shown in the preceding figures, and in the context of a particular type of proxy target and its associated target features. However, those skilled in the art who have read this specification will recognize that the principles of these methods can be similarly applied, mutatis mutandis, to other overlay metrology systems and using other types of proxy targets. Furthermore, elements of these various methods may be combined to provide multi-factor CoS measurement and calibration. All such alternative implementations are considered to be within the scope of the present invention.

[0059] Figure 4 is a flow chart 200 that schematically illustrates a process for measuring CoS variation of process and resist layer target features of an overlay metrology proxy target as a function of focus, according to one embodiment of the present invention. This process references the optical inspection tool 10 illustrated in Figure 1, with additional reference to Figures 2 and 3. The objective of the process illustrated in Figure 4 is to individually resolve the CoS location of target features of each layer of an overlay proxy target, such as an AIM™ target 152, in relation to the focus setting of the wafer 12 in the tool 10.

[0060] The process begins at start step 202. In focus step 204, cameras CAM1 and CAM2 are set by controller 24 to focus on wafer 12 by moving table 26 along the z direction and / or by moving the cameras with actuators 58 and 60. The cameras are set to a focus difference of ΔZ by differential movement of actuators 58 and 60. (Focus setting herein refers to the Z coordinate in wafer space. For example, focus difference ΔZ refers to the xy planes on which cameras CAM1 and CAM2 are focused being separated by ΔZ on or near wafer 12.) Further details of the focus separation and focusing of cameras CAM1 and CAM2 are provided below in conjunction with FIG. 5.

[0061] In a first grating registration step 206, both cameras CAM1 and CAM2 are aligned with the gratings 154, 156, 158, and 160 projected by the grating projector 20. To perform this registration, the controller 24 processes the acquired images of the gratings to determine the relative position of each camera along the x and y directions with respect to the gratings. The cameras can be aligned with each other by physically moving them with respect to the gratings 154, 156, 158, and 160, or by calculating a camera-to-grating offset for use in subsequent processing. Because of the periodic nature of the gratings and the repeating structure of the pixels in CAM1 and CAM2, the controller 24 can align each camera with respect to the gratings with an accuracy of better than 0.1 nm (in terms of x and y coordinates in wafer space). Furthermore, because the same grating is projected onto each camera CAM1 and CAM2, any spatial shift or vibration of the projected grating will occur in common mode in both cameras. Because each camera is aligned to the same common mode grating, the cameras are aligned to one another with an accuracy of better than 0.1 nm. In a first acquisition step 208, images of proxy target 152, specifically of gratings 162 and 164, are read from cameras CAM1 and CAM2 by controller 24 and stored in memory 25.

[0062] In a refocusing step 210, CAM1 is brought to the Z coordinate where CAM2 was located prior to step 210 by moving wafer 12 by a distance ΔZ along the z direction using table 26. In a second grating registration step 212, controller 24 again aligns cameras CAM1 and CAM2 with respect to gratings 154, 156, 158, and 160, as in first grating registration step 206. The purpose of this step is to ensure persistent registration between the two cameras. In a CAM1 registration step 214, CAM1 is aligned with an image acquired by CAM2 at the previous focus position, which, together with second grating registration step 212, establishes the positions of the two cameras in the x-y plane relative to the previous focus position. In a second acquisition step 216, similar to first acquisition step 208, images of proxy target 152 are read by controller 24 through cameras CAM1 and CAM2 and stored in memory 25.

[0063] In decision step 218, controller 24 determines whether another focus step is required given the preset sequence of focus steps. If the answer is yes, the process returns to refocus step 210 and continues from there. Once all preset focus steps have been taken, in calculation step 220, controller 24 processes the images stored in memory 25 to calculate the CoS of each of the gratings 162 and 164 as a function of the focus setting through the ΔZ focus steps. The process ends in end step 222.

[0064] In certain embodiments, adjusting the direction in which illumination impinges on wafer 12 compensates for residual optical errors in the metrology tool's imaging optics. For example, in optical inspection system 10, controller 24 can adjust the position of aperture assembly 36 in the y-z plane via actuator 38 to compensate for residual optical errors in objective lens 30. Because the CoS of each of gratings 162 and 164 depends on the position of aperture assembly 36, measuring CoS as a function of focus for multiple positions of aperture assembly 36 can gather more comprehensive data. In one embodiment, the process described in flowchart 200 is performed for a set of Y and Z coordinates of the aperture assembly, e.g., (Y ± n * ΔY, Z ± n * ΔZ), where Y ± n * ΔY and Z ± n * ΔZ represent the nominal center position of aperture assembly 36, ΔY and ΔZ represent incremental increments of the aperture assembly, and n is an integer index that assumes values from 0 to a maximum value of N. As discussed below in the section entitled "Selection of Measurement Conditions," the data obtained can be used to further improve the quality of the overlay measurements.

[0065] FIG. 5 is a plot that schematically illustrates the focus spacing between CAM1 and CAM2 and the focus steps taken in the process of FIG. 4, according to one embodiment of the present invention.

[0066] The positions of the two cameras CAM1 and CAM2 in terms of Z coordinates in wafer space, i.e., Z coordinates relative to the wafer 12, are shifted through N focus steps. As indicated on the Z coordinate axis 251, in a first step 250, CAM1 is focused on plane Z=Z0 and CAM2 is focused on plane Z=Z0+ΔZ. In a second step 252, the wafer focus is incremented by ΔZ, bringing CAM1 to Z=Z0+ΔZ and CAM2 to Z=Z0+2ΔZ, so that CAM1 is now at the same focus position as CAM2 was in the first step 250. In a third step 254, the wafer focus is again incremented by ΔZ, bringing CAM1 to Z=Z0+2ΔZ and CAM2 to Z=Z0+3ΔZ. This process continues until stage N 256, where CAM1 has a location Z=Z0+(N-1)ΔZ and CAM2 has a location Z=Z0+NΔZ.

[0067] In other words, at each focus step, CAM1 is positioned at the focus where CAM2 was at the previous step, allowing registration between successive steps. This sequence of steps, in combination with grid registration steps 206 and 212 (FIG. 4), allows controller 24 to precisely align each camera along the x and y directions through focus, and to calculate the true CoS of each of grids 162 and 164 through focus.

[0068] 6A-6B are flowcharts 300 that schematically illustrate a process for measuring CoS variation of target features of an overlay metrology proxy target as a function of illumination wavelength, according to one embodiment of the present invention. The process uses the optical inspection system 10 shown in FIG. 1 and additionally references FIGS. 2 and 3. The purpose of the process illustrated in FIGS. 6A-6B is to individually resolve the CoS location of each target feature of an overlay proxy target, e.g., AIM™ target 152, as a function of the wavelength used by the system 10. Flowchart 300 includes first and second portions 301 and 302 in which the CoS of gratings 164 and 162, respectively, is calculated as a function of illumination wavelength.

[0069] The process begins at start step 303. In place beamsplitter step 304, controller 24 activates actuator 49 to bring dichroic beamsplitter 48 into the optical path of apparatus 10, thereby splitting the optical radiation in the optical path of apparatus 10 so that the portion of the radiation within spectral band Δλ1 is received by CAM1 and the portion of the radiation within spectral band Δλ2 is received by CAM2.

[0070] The process then enters a first portion 301. In a first illumination step 306, semiconductor wafer 12 is illuminated with wavelengths λ1 and λ2, where λ1 is within spectral band Δλ1 and wavelength λ2 is within spectral band Δλ2, and light source 32 emits optical radiation at wavelength λ1 and light source 33 emits optical radiation at wavelength λ2. In a first focusing step 308, controller 24 uses table 26 and actuators 58 and 60 to focus camera CAM1 to a contrast focus on grating 162 and camera CAM2 to a contrast focus on grating 164. The term "contrast focus" refers to the focus position where the grating image on each camera exhibits maximum contrast C. Contrast C also refers to the maximum and minimum intensity I of the grating image on a given camera. max and I min Based on this, C=(I max -Imin ) / (I max +I min ) In a first grating registration step 310, cameras CAM1 and CAM2 are aligned with respect to the projected gratings 154, 156, 158, and 160, similar to first grating registration step 206 (FIG. 4). Then, in a first acquisition step 312, images of gratings 162 and 164 are acquired by cameras CAM1 and CAM2, read out from the cameras by controller 24, and stored in memory 25.

[0071] In a first wavelength change step 314, controller 24 changes the wavelength of illumination reaching CAM2 from λ2 to λ2 + Δλ by incrementing the wavelength of optical radiation emitted by light source 33 in spectral band Δλ2 by Δλ. In a first refocus step 316, CAM2 is refocused to a contrast focus at the incremented wavelength. In a second grating registration step 318, cameras CAM1 and CAM2 are aligned with gratings 154, 156, 158, and 160, as in first grating registration step 310. In a re-registration step 320, CAM1 is refocused and re-aligned with the same image of grating 162 acquired through CAM1 in acquisition step 312. That is, it continues to be aligned with the same physical grating 162 at the same wavelength and focus, thereby establishing CAM1 as an "anchor" for portion 301. In a second acquisition step 322, an image of the grid 164 is acquired by CAM2, read out from that camera by controller 24 and stored in memory 25.

[0072] In a first determination step 324, controller 24 determines whether another wavelength step is needed within spectral band Δλ2, given the preset sequence of wavelength steps. If the answer is yes, the process returns to first wavelength change step 314, where the wavelength of illumination reaching CAM2 is again incremented by Δλ, and the process continues. Once all preset focus steps have been exhausted, controller 24 in a first calculation step 326 calculates the CoS of grating 164 as a function of wavelength through the wavelength steps Δλ within spectral band Δλ2, based on the images stored in memory 25.

[0073] The process continues in a second portion 302, which is detailed to clarify the differences between the first and second portions. Steps 328, 330, 332, 334, 340, and 346 in the second portion 302 are identical to corresponding steps 306, 308, 310, 312, 318, and 324 in the first portion 301. However, steps 336, 338, 342, and 344 in the second portion 302 differ from corresponding steps 314, 316, 320, and 326 in the first portion 301 in that in those steps of the second portion, a spectral scan is performed across the spectral range Δλ1 rather than across the spectral range Δλ2 as in the first portion.

[0074] In a second illumination step 328, semiconductor wafer 12 is illuminated with wavelengths λ1 and λ2. In a second focusing step 330, controller 24 focuses camera CAM1 on the contrast focal point on grating 162 and camera CAM2 on the contrast focal point on grating 164. In a third grating registration step 332, cameras CAM1 and CAM2 are aligned with gratings 154, 156, 158, and 160. In a third acquisition step 334, images of gratings 162 and 164 are acquired by cameras CAM1 and CAM2, read out from the cameras by controller 24, and stored in memory 25.

[0075] In a second wavelength change step 336, controller 24 increments the wavelength of optical radiation emitted by light source 32 within spectral band Δλ by Δλ. In a second refocusing step 338, CAM1 is refocused to the contrast focus at the incremented wavelength Δλ+Δλ. In a fourth grating registration step 340, cameras CAM1 and CAM2 are aligned with gratings 154, 156, 158, and 160. In a second reregistration step 342, CAM2 is refocused and realigned with the same image of grating 164 read from CAM2 in the third acquisition step 334, thereby establishing CAM2 as the anchor for second portion 302. In a fourth acquisition step 344, an image of grating 162 is acquired by camera CAM1, read from that camera by controller 24, and stored in memory 25.

[0076] In a second determination step 346, controller 24 determines whether another wavelength step is needed within spectral band Δλ1, given the preset sequence of wavelength steps. If the answer is yes, the process returns to second wavelength change step 336, where the wavelength of illumination reaching CAM1 is again incremented by Δλ, and the process continues. Once all preset wavelength steps have been taken, controller 24 calculates the CoS of grating 162 as a function of wavelength through the wavelength steps Δλ in spectral band Δλ1 in a second calculation step 348. The process ends in end step 350.

[0077] Similar to the measurement of the CoS variation with focus, the obtained data on the CoS variation with wavelength can be used to further improve the quality of the overlay measurement, as described below in the section entitled "Selection of Measurement Conditions."

[0078] Measurement of the variation of CoS with illumination wavelength may be performed using the polarization state of the optical radiation as an additional parameter. In one embodiment, the variation of CoS with wavelength is measured for various polarization states of illumination impinging on wafer 12. That is, upon command from controller 24, light sources 32 and 33 emit optical radiation in two orthogonal polarization states, and the variation of CoS with illumination wavelength is measured separately for each polarization state. In an alternative embodiment, wafer 12 is illuminated with unpolarized optical radiation such that a specific state of polarization reaches each of two cameras CAM1 and CAM2, the specific state being determined by dichroic beamsplitter 48, which also functions as a polarizer, or by polarizers appropriately positioned between the dichroic beamsplitter and the two cameras.

[0079] 7A-7B are a flowchart 500 that schematically illustrates the process of generating CoS landscapes and evaluating sensitivity across focus settings and wavelengths, according to one embodiment of the present invention.

[0080] To capture scanner-induced overlay errors, such as misplacement and rotation of the wafer 12 in the scanner, and scanner field distortion, overlay errors are typically measured at several metrology sites on the wafer. The process illustrated in flowchart 500 is for measurements on N sites on the wafer 12, where the sites are numbered n (n=1, 2, ...N). Furthermore, the process is performed for both resist and process layers.

[0081] The CoS is measured in relation to the focal point in two wafer 12 orientations, and for the second orientation, the wafer is rotated 180° around the Z axis. The CoS associated with the focal point in the first orientation (arbitrarily set to 0° orientation) is called CoS0(Z), and the CoS in the second orientation is called CoS 180 (Z), the tool induced shift (TIS) of CoS calculated by the controller 24, i.e., CoS_TIS, is given by CoS_TIS(Z)=(CoS(Z)+CoS 180(Z)) / 2. The CoS error described by CoS_TIS is due to asymmetries in the optical components in the imaging path of the apparatus 10, such as the asymmetry of the objective lens 30. The corrected CoS, i.e., the measured CoS minus CoS_TIS, can be calculated by the controller 24 as CoS_COR(Z); where CoS_COR(Z)=[CoS(Z)-CoS 180 (Z)] / 2, and "COR" stands for "corrected." As mentioned above, CoS is a two-dimensional vector (CoS X ,CoS Y ), and the notation "CoS" includes both the X and Y coordinates.

[0082] Inaccuracies in the CoS measurement due to mechanical vibrations of the device 10 can be reduced by acquiring multiple images of the proxy target 152 and averaging the measurements.

[0083] The process begins at start step 502. At polarization selection step 504, the polarization state(s) of the illumination emitted by illumination assembly 16 is selected. At site selection step 506, a site n on wafer 12 is selected. At wavelength selection step 507, a wavelength λ is selected. At 0° CoS through focus step 508, CoS is measured across focus Z as described in connection with FIGS. 4 and 5. At 180° CoS through focus step 510, the above measurement is repeated, but the orientation of wafer 12 is rotated 180° relative to that at step 508. At first CoS_TIS step 512, CoS_TIS is calculated by controller 24 based on the individual CoS values at 0° and 180° orientations for each focus setting Z, as follows: CoS_TIS=(CoS_TIS+CoS_TIS) 180 ) / 2. For simplicity, the explicit dependencies on focus Z, wavelength λ, polarization P, and site n have been omitted from these equations. In a first CoS_COR step 514, CoS_COR is calculated by the controller 24 from the individual CoS values at 0° and 180° for each focus setting Z as follows: CoS_COR=(CoS0-CoS 180) / 2 (again, explicit reference to variables is omitted). In wavelength determination step 516, controller 24 determines, based on a preset list of wavelengths, whether steps 507-514 should be re-executed. If the answer is affirmative, wavelength λ is incremented in wavelength increment step 517, and the process continues from step 507.

[0084] Once all preset wavelengths have been exhausted, the process continues by measuring CoS versus wavelength at the best contrast focus, as described in connection with Figures 6A-6B. Although CoS versus focus has been measured at all desired wavelengths in the previous step, drift in the apparatus 10 may cause some of the CoS versus focus measurements to drift along the focus coordinate Z. As described in more detail below, the CoS versus wavelength measurements can be used to correct for such drift.

[0085] In a 0° CoS thru wavelength step 518, CoS is measured across a preset spectrum of wavelengths at the best contrast focus. In a 180° CoS thru wavelength step 520, the above measurement is repeated, but the orientation of the wafer 12 is rotated 180° relative to that in step 516. In a second CoS_TIS step 522 and a second CoS_COR step 524, CoS_TIS and CoS_COR are calculated by the controller 24 from the data acquired in steps 518 and 520, as described above in connection with steps 512 and 514, respectively.

[0086] In a CoS_TIS stitching step 526, controller 24 compares, for each wavelength λ, the focus-through CoS_TIS results from steps 507-516, with focus Z as the best contrast focus, to the CoS_TIS results from step 522. If there is a discrepancy between the two results, the focus-through CoS_TIS results at wavelength λ are shifted along focus coordinate Z to eliminate the discrepancy. This "stitches" the focus-through CoS_TIS results at adjacent wavelengths into a consistent representation of CoS_TIS in two-dimensional Zλ space, as described in further detail below with respect to FIG. 8A. In a CoS_TIS landscape step 528, controller 24 collects this representation into a collection of CoS_TIS values for the two variables Z and λ. This collection is referred to as the CoS_TIS landscape.

[0087] In a CoS_TIS derivative step 530, the controller 24 calculates the second derivative ∂, which indicates the sensitivity of CoS_TIS to variations in the variables Z and λ. 2 The value of CoS_TIS / ∂Z∂λ is calculated. In a minimum CoS_TIS step 532, the controller 24 identifies the two-dimensional area(s) in the CoS_TIS landscape that are in the (Z,λ) plane and where the absolute value of CoS_TIS is less than a predetermined limit, thereby indicating the area of minimum tool-induced shift in CoS, i.e., minimum CoS error (separate limits may be assigned to the X and Y components of CoS_TIS). In a minimum CoS_TIS derivative step 533, the controller 24 identifies the two-dimensional area(s) in the CoS_TIS landscape that are in the (Z,λ) plane and where the absolute value of CoS_TIS is less than a predetermined limit, thereby indicating the area of minimum tool-induced shift in CoS, i.e., minimum CoS error (separate limits may be assigned to the X and Y components of CoS_TIS). 2 The two-dimensional area(s) where the absolute value of CoS_TIS / ∂Z∂λ is less than another predetermined limit are identified, thus indicating the area of greatest stability of CoS_TIS.

[0088] Similar to the CoS_TIS stitching step 526, in a CoS_COR stitching step 534, the controller 24 compares, for each wavelength λ, the cross-focus CoS_COR results from steps 507-516, with focus Z as the best contrast focus, to the CoS_COR results from step 524. If there is a discrepancy between the two results, the cross-focus CoS_TIS result for wavelength λ is shifted along the focus coordinate Z to remove the discrepancy. This "stitches" the cross-focus CoS_COR results for adjacent wavelengths together to form a consistent representation of CoS_COR in Zλ space.

[0089] Similar to the CoS_TIS landscape step 528, in a CoS_COR landscape step 536, the controller 24 collects this representation into a collection of CoS_COR values with respect to the two variables Z and λ, referred to herein as the CoS_COR landscape. In a CoS_COR derivative step 538, the controller 24 calculates a second derivative ∂ that indicates the sensitivity of CoS_COR to changes in the variables Z and λ. 2 The value of CoS_COR / ∂Z∂λ is calculated.

[0090] In a minimum CoS_COR derivative step 540, the controller 24 determines, within its CoS_COR landscape, ∂ 2 Two-dimensional area(s) where the absolute value of CoS_COR / ∂Z∂λ is less than another predetermined limit are identified, thereby indicating areas where CoS_COR is most stable (since CoS_COR can be assumed to have an arbitrary offset from 0, there is no theoretical justification for identifying areas where CoS_COR is less than a predetermined limit).

[0091] In site determination step 542, controller 24 determines whether another site n on wafer 12 needs to be measured. If so, the process returns to step 506, where the next site is selected. Once all sites have been measured (for the current polarization), controller 24 determines whether measurements need to be performed with illumination in an additional polarization state in polarization determination step 544. If so, the process returns to step 506, where all N sites are measured again using illumination in the new polarization state. Once all required polarization states have been exhausted, the process ends in end step 546.

[0092] 8A-8D are schematic representations of CoS_TIS and CoS_COR landscapes associated with resist and process layers, according to one embodiment of the present invention. FIG. 8A shows a CoS_TIS landscape 600 associated with a resist layer, FIG. 8B shows a CoS_COR landscape 602 associated with a resist layer, FIG. 8C shows a CoS_TIS landscape 604 associated with a process layer, and FIG. 8D shows a CoS_COR landscape 606 associated with a process layer. Each landscape 600, 602, 604, and 606 is associated with a given site n and a given polarization P. Similar landscapes are generated for all N sites and for illumination with all polarization states. Each landscape generates a CoS_TIS landscape 600 associated with a given wavelength range (λ min ,λ max ) and a certain focal range (Z min ,Z max ) are plotted. Curves 608, 610, 612, and 614 show the focus position of the best contrast focus as a function of wavelength for each of the landscapes 600, 602, 604, and 606.

[0093] The areas identified in steps 532, 533, and 540 of Figure 7A are shown in landscapes 600-606. That is, in area 616, ∂ 2CoS_TIS / ∂Z∂λ is less than a predetermined limit L1, and in area 618 CoS_TIS is less than a predetermined limit L2. 2 CoS_COR / ∂Z∂λ is less than a predetermined limit L3. Within area 622, ∂ 2 CoS_TIS / ∂Z∂λ is less than a predetermined limit L4, and in area 624 CoS_TIS is less than a predetermined limit L5. 2 CoS_COR / ∂Z∂λ is less than the predetermined limit L6. Thus, areas 616 and 622 indicate areas where CoS_TIS is highly stable for the resist and process layers, respectively, and areas 618 and 624 indicate areas where CoS_TIS is low, i.e., where metrology tool error is low, for those layers. Areas 620 and 626 indicate areas where CoS_COR is highly stable for the resist and process layers, respectively.

[0094] The "stitching" concept introduced in steps 526 and 534 of Figure 7A is depicted diagrammatically in Figure 8A. Three lines 628, 630, and 632 show the relationship between CoS_TIS as a function of focus Z and the individual wavelengths λ i-1 , λ i and λ i+1 7B represents the three paths measured according to steps 507-516 of FIG. 7A. Three points 634, 636, and 638 on curve 608 indicate where CoS_TIS as a function of wavelength λ was measured in steps 518-522 of FIG. 7B. The CoS_TIS value on line 628 where that line intersects curve 608 is compared to the CoS_TIS value at point 634. If the two values are the same, line 628 is not shifted. Conversely, if the values do not match, line 628 (and its CoS_TIS value) is shifted along the Z direction until the value on line 628 where it intersects curve 608 matches the value at point 634. A similar process is repeated for line 630 relative to point 636 and for line 632 relative to point 638.

[0095] The three lines 628, 630, and 632 are shifted as necessary, and when the values where each line intersects with the curve 608 match the values at the individual points 634, 636, and 638, the lines are "stitched" together, so to speak. This process calculates λ for all similar lines. min From λ max , thereby correcting for any drift that may occur in the device 10 during the measurement of the CoS_TIS values across focus. A similar stitching operation is applied to the landscapes 602, 604, and 606.

[0096] 9A and 9B are schematic representations of precision landscapes for resist and process layers according to an embodiment of the present invention. In this embodiment, the precision of the CoS measurement is evaluated. The precision measurement involves several consecutive CoS measurements, e.g., 5, 10, or 15 measurements, and a precision metric is calculated. A common metric is the 3σ value (three times the standard deviation) of the measurements. In one embodiment, the precision of the CoS (labeled CoS_Prec) is measured during the measurements described above (FIGS. 5-7) and correlated with two variables, focus Z and wavelength λ, to generate a precision landscape.

[0097] Landscape 700 in Figure 9A represents the precision of the CoS of the resist layer, and landscape 702 in Figure 9B represents the precision of the CoS of the process layer. Each landscape, like landscapes 600, 602, 604, and 606 in Figures 8A, 8B, 8C, and 8D, respectively, is plotted over a wavelength range (λ min ,λ max ) and a certain focal range (Z min ,Z max) in a given landscape. Landscapes 700 and 702 have respective curves 704 and 706 showing the focus position of the best contrast focus relative to wavelength. Based on the CoS_Prec values associated with the resist layer within landscape 700, controller 24 has identified areas 708 and 710 where CoS_Prec has a value below a preset limit L7. Similarly, based on the CoS_Prec values associated with the process layer within landscape 702, controller 24 has identified areas 712, 714, and 716 where CoS_Prec has a value below a preset limit L8.

[0098] [Measurement condition selection] The settings of various parameters of the tool 10 (e.g., focus, wavelength, polarization) during a metrology recipe, i.e., overlay metrology, can significantly affect the quality of the metrology results. As described in more detail below, a user of the tool 10 can choose to vary the measurement conditions to achieve a desired balance between factors such as stability and accuracy. The layer-by-layer metrology result interpretation depicted in Figures 4-9 provides the user with a sophisticated set of tools to select measurement conditions with the user's specific goals in mind. Two exemplary embodiments are presented below that demonstrate the use of these tools.

[0099] Embodiment 1 - In this embodiment, the index M1 is calculated by the controller 24 for all measurement sites and all polarizations, layer by layer, based on the landscapes 600, 602, 604 and 606 (Figures 8A-8D) and the landscapes 700 and 702 (Figures 9A and 9B).

number

[0100] The variables in this index and its components include the wavelength of illumination λ, the focal coordinate Z, the polarization state P, and the layer L (resist layer or process layer). N is the average over N measurement sites, and 3σ Nis three times the standard deviation across those N sites, and AVG N and 3σ N The effect of process variations across the landscape across sites and polarizations is included. For simplicity, precision is denoted as "Prec."

[0101] The metric M1 includes CoS_TIS as its contributor and therefore places emphasis on assessing metrology accuracy (tool-induced errors). Let LIMIT(M1) be the predetermined limit for M1, M1(λ,Z,P,L) <LIMIT(M1) By finding the measurement condition where ≡ ...

[0102] For an apparatus 10 (FIG. 1) with two illuminators 15 and 17, the metrology conditions can be optimized independently for each layer through the use of a layer-specific index M1. If the apparatus 10 only has one illuminator, a compromise between the metrology conditions for the two layers would have to be found. For example, a potential requirement for a single illuminator apparatus is: M1(λ,Z,P,L resist )+M1(λ,Z,P,L process ) <LIMIT’(M1) However, L resist ,L process are the resist and process layer respectively, and LIMIT'(M1) is a (different) predetermined limit.

[0103] If CoS_TIS were a constant correction factor for the tool, it might be possible to calibrate it using a one-time calibration procedure. However, because of the coupling between the local geometry of the proxy target's target features and the optical response of the metrology tool optics, CoS_TIS can vary from target to target when multiple locations on a wafer are measured. This type of CoS_TIS variation can occur as a result of process variations across the wafer.

[0104] The contribution of CoS_TIS to the index M1 can be reduced layer by layer by lateral shifting of the apertures of each aperture assembly 36 and 37.

[0105] Embodiment 2 - In this embodiment, the index M2 is calculated by the controller 24 for each layer using the same landscapes as those in the index M1, except that the formula for the index M2 is different from that of the index M1.

number

[0106] Since the index M2 contains the (second) derivatives of both CoS_COR and CoS_TIS, it represents the stability of CoS during overlay error measurement. That is, let LIMIT(M2) be a predetermined limit for M2, M2(λ,Z,P,L) <LIMIT(M2) By requiring that ≡ ...

[0107] The same variables as those in M1 are used for the indicator M2, and AVG N and 3σ N applies to the N sites in the same way as for index M1. Similar considerations as for M1 also apply to index M2, depending on whether the lighting assembly 16 has a single illuminator or a two illuminator configuration.

[0108] Alternatively or additionally, other metrics may be generated using data from landscapes 600, 602, 604 and 606 and from landscapes 700 and 702 to reflect different user-specific requirements.

[0109] FIG. 10 is a plot 800 that schematically illustrates the variation of CoS_TIS as a function of aperture offset, according to one embodiment of the present invention.

[0110] Plot 800 shows the variation in CoS_TIS measured by tool 10 for four metrology sites (n=1, ..., 4) as a function of the lateral offset AO of one of the apertures in aperture assembly 36. The individual CoS_TIS values for the various sites are shown as lines 802. Line 802 exhibits different offsets and slopes due to variations across the four sites, such as layer contrast, feature topology, focus variation, site tilt, and process variation. The average of CoS_TIS across the four sites, i.e., AVG(CoS_TIS) as a function of AO, is shown as line 804.

[0111] The optimal (smallest) CoS_TIS variation across the four measurement sites, i.e., 3σ(CoS_TIS)1, is found at a certain aperture offset AO1, where CoS_TIS = CoS_TIS1. A non-zero value of AO1 indicates a global angular alignment error between the wafer 12 and imaging assembly 14. Choosing an aperture offset larger than AO1 increases 3σ(CoS_TIS) but decreases AVG(CoS_TIS), indicating an opportunity for optimization between AVG(CoS_TIS) and 3σ(CoS_TIS). Using independent light source and aperture assemblies, such as apparatus 10 of Figure 1, allows for independent optimization between AVG(CoS_TIS) and 3σ(CoS_TIS) for each layer.

[0112] 11A and 11B are schematic illustrations of the application of CoS as a function of focus to resolve sidewall asymmetry of target features in an AIM™ proxy target 152, according to one embodiment of the present invention.

[0113] 11A is an image of the AIM™ proxy target 152 (shown in FIG. 3) with outlines around the resist grating 162 and process layer grating 164 oriented along the x-axis. The individual grating bars 902 and 904 of the gratings 162 and 164 are oriented along the y-axis.

[0114] Figure 11B is a schematic cross-sectional view of grating bars 904a and 904b taken from process layer grating 164 along lines 908a and 908b in Figure 11A. To show the two cross-sectional views together, the grating bars and their spacing from one another along the x-axis are not shown to scale.

[0115] Asymmetric process effects in semiconductor fabrication processes, such as asymmetric etching, result in the grating bars 904 having an asymmetric topographical structure, as shown in FIG. 11B : the left sidewall 910a of grating bar 904a is perpendicular to the xy plane, while the right sidewall 912a meets the xy plane at an oblique angle. Similarly, the left sidewall 910b of grating bar 904b is perpendicular to the xy plane, while the right sidewall 912b of that grating bar meets the xy plane at an oblique angle. Because the area occupied by the proxy target 152 is typically small, with linear dimensions of tens of microns or less, all of the bars 904 in the grating 164 exhibit the same asymmetry shown in FIG. 11B . The asymmetry in the grating bars causes the CoS of the grating 164 to shift with focus, and the CoS variation as a function of focus can be used to resolve this asymmetry, as described in more detail below.

[0116] To resolve the asymmetry of grating bars 904a and 904b (and therefore all grating bars 902 and 904), controller 24 reads three images of grating 164 acquired by cameras CAM1 and CAM2 at three focus steps 920, 922, and 924 at focus locations labeled on Z axis 926. Controller 24 focuses cameras CAM1 and CAM2 on wafer 12 using the contrast focus (FIGS. 6A-6B) by moving table 26 along the z direction and / or by moving the cameras via actuators 58 and 60. In all three focus steps 920, 922, and 924, CAM1 is focused at a fixed x-y location Z=Z0 on grating 162 and is aligned with grating 162 at that fixed focus. That is, Z coordinate Z0 forms the "anchor focus" for measurements.

[0117] In the three focus steps 920, 922, and 924, CAM2 is focused in the xy plane, having Z coordinates Z1, Z2, and Z3, respectively. During image acquisition, both cameras CAM1 and CAM2 are aligned with the projection gratings 154, 156, 158, and 160, as in the first grating registration step 206 (FIG. 4), ensuring a known lateral registration (registration in the xy plane) between the cameras. Based on the image read from CAM1, controller 24 calculates the CoS of grating 164, which is diagrammatically labeled point 928 in FIG. 11B, and stores it in memory 25. Based on the three images read from CAM2 at the three focus positions, controller 24 calculates three individual CoS values, diagrammatically labeled points 930, 932, and 934, and stores them in memory 25. A curve 936 is fitted by controller 24 to points 928, 930, 932, and 934 and is indicative of the shift in CoS with focus, i.e., a measure of the cross-sectional asymmetry of grating bars 902 and 904. Curve 936 can be a straight line or a higher order curve.

[0118] The above method can be similarly applied to process layer gratings 164 oriented along the y direction, or resist gratings 162 oriented along both the x and y directions, to account for their cross-sectional asymmetries, and actual overlay measurements made on production wafers can then be corrected to address the apparent CoS that may result from these asymmetries.

[0119] In an alternative embodiment, only a single camera, e.g., CAM1, is used to measure CoS relative to focus. In this type of measurement, CAM1 is focused through four focus positions Z0, Z1, Z2, and Z3, and an image of grating 164 is read from CAM1 by controller 24 at each focus position and stored in memory 25. Controller 24 then calculates individual CoS values from the images stored in memory 25 and calculates CoS relative to focus as described above for two-camera measurement. This single-camera measurement method is more sensitive to the mechanical stability of apparatus 10 than the two-camera method described above, due to the lack of anchoring to fixed features and co-registration of the cameras.

[0120] Although the methods described above are implemented using four focus positions (an "anchor" position Z0 and three focus positions Z1, Z2, and Z3), fewer or more focus positions may alternatively be used.

[0121] 12 is a series of plots that schematically illustrate the use of image signal correlation to monitor asymmetries of target features in an overlay proxy target, according to one embodiment of the present invention. As previously explained, asymmetric process effects in semiconductor manufacturing processes, such as asymmetric etching, can cause target features in an overlay proxy target to have asymmetric cross-sections. In this embodiment, monitoring of these asymmetries (without even evaluating them) is utilized to monitor the semiconductor manufacturing process.

[0122] 12 shows the cross-sectional appearance of grating bar 904a, as in FIG. 11B. Controller 24 reads an acquired image of bar 904a, for example, from CAM1, converts the image into an image signal, and stores it in memory 25. The portion of the image signal along curve 608 (FIG. 8A) is shown as curve 1002 in FIG. 12. Based on the image signal of curve 1002, controller 24 generates a reflected image signal by reflecting it around the z-axis, i.e., a reflected image signal shown as curve 1004.

[0123] Controller 24 calculates two correlation curves 1006 and 1008: correlation curve 1006, which is the autocorrelation (correlation with itself) of curve 1002, and correlation curve 1008, which is the cross-correlation between curves 1002 and 1004. Curve 1006 is C auto max The curve 1008 has a maximum value of C cross max Since the curve 1002 is asymmetric, C cross max <C auto max The maximum value C of the autocorrelation curve 1006 auto max Maximum value C of the cross-correlation curve 1008 for cross max can be used as a measure of asymmetry in the image signal 1002, and thus as a measure of process effects that cause asymmetric cross-sectional profiles of target features in the overlay proxy target.

[0124] Images of target features of the overlay proxy target, such as images of bar 904a, can be acquired under various operating conditions of the apparatus 10, such as by varying the wavelength and / or polarization of the illumination of the wafer 12 and / or with various focus settings. cross max / C auto max The variation of σ results in a "correlation landscape" that can be further utilized to monitor the semiconductor manufacturing process during the execution of the process. For example, the ratio of two correlation maxima C cross max / C auto max If σ falls below a preset limit, say 0.8, this can be used to indicate unacceptable process variation.

[0125] The above method can be applied equally to process layer gratings 164 oriented along the y direction and to resist gratings 162 oriented along both the x and y directions, thereby revealing their cross-sectional asymmetries.

[0126] The above-described embodiments can be used individually or in combination to determine an optimal overlay metrology recipe, i.e., a recipe that provides robust overlay metrology conditions when performing measurements of overlay errors. Such a recipe specifies optimal focus settings, wavelength, polarization, illumination conditions, and objective pupil control. Separate conditions can also be applied separately for process layers and resist layers.

[0127] Additionally or alternatively, information provided by the above metrology techniques, such as information regarding the variation of CoS with focus, may be used by controller 24 to enhance the overlay metrology algorithms. The above methods can also be generalized to additional illumination and collection channels to simultaneously measure overlay error between multiple pairs of layers.

[0128] It is understood that the above-described embodiments have been cited by way of example, and that the present invention is not limited to what has been particularly shown and described above. Rather, the scope of the present invention includes both combinations and subcombinations of the various features described above, as well as variations and modifications thereof that would occur to one skilled in the art upon reading the foregoing description and that are not disclosed in the prior art.

Claims

1. 1. An optical device, comprising: an illumination assembly configured to direct one or more illumination beams onto the semiconductor wafer such that the first target feature and the second target feature at least partially overlap; an imaging assembly; Equipped with the imaging assembly a first detector configured to image the semiconductor wafer according to a first imaging parameter; a second detector configured to image the semiconductor wafer with a second imaging parameter different from the first imaging parameter; a controller including one or more processors; and the one or more processors: receiving a sequence of images from the first detector and the second detector, wherein at least one of the first imaging parameter or the second imaging parameter is varied over the sequence of images, thereby providing varied imaging parameters at one or more sites of the semiconductor wafer; providing an index indicating a variation in center of symmetry (COS) of the first target feature and the second target feature as a function of the varied imaging parameter based on the sequence of images; identifying a landscape of values of the changed imaging parameter in which the index is below one or more predetermined limits; generating a recipe for metrology measurements in which the varied imaging parameters are set to values within the landscape; an optical device configured to generate one or more metrology measurements associated with one or more product wafers based on the recipe;

2. 10. An optical device according to claim 1, The optical device, wherein the one or more processors are further configured to calibrate and / or correct the one or more metrology measurements based on variations in the COS.

3. 10. An optical device according to claim 1, An optical apparatus, wherein the one or more metrology measurements include one or more overlay measurements.

4. 10. The optical device according to claim 1, further comprising: a grid projector configured to project a grid image onto the first detector and the second detector; An optical arrangement in which the one or more processors of the controller are further configured to align the sequence of images based on the projected grid image.

5. 10. An optical device according to claim 1, The optical system, wherein the varied imaging parameter is at least one of a focus position of the semiconductor wafer, a wavelength, a polarization, an angular offset of an aperture of the imaging assembly, or an angular position of at least one of the first detector or the second detector.

6. 10. An optical device according to claim 1, an optical system wherein the varied imaging parameter includes a focal position of the semiconductor wafer, the first detector and the second detector are configured to simultaneously image the semiconductor wafer at different values of the focal position separated by a fixed value ΔZ, and the sequence of images is generated by successive steps in which the values of the focal position associated with the first detector and the second detector are shifted by the fixed value ΔZ.

7. 7. An optical device according to claim 6, further comprising: a grid projector configured to project a grid image onto the first detector and the second detector; The one or more processors of the controller further registering the images from the first detector and the second detector at each of the successive steps; an optical device configured to align an image from the first detector associated with a current step in the successive steps with an image from the second detector associated with a previous step in the successive steps at the common value of focus position.

8. 10. An optical device according to claim 1, the varied imaging parameter comprises a wavelength of at least one of the one or more illumination beams; the first detector and the second detector are configured to simultaneously image the semiconductor wafer based on different values of the wavelength; the sequence of images a first image sequence in which the wavelength value associated with the first detector is held constant and the wavelength value associated with the second detector is successively changed in a first sequence of successive steps; a second image sequence, wherein the wavelength value associated with the second detector is held constant and the wavelength value associated with the first detector is sequentially changed in a second sequence of sequential steps.

9. 9. An optical device according to claim 8, further comprising: a grid projector configured to project a grid image onto the first detector and the second detector; The one or more processors of the controller further aligning the first sequence of images from the first detector and the second detector at each step of the first series of successive steps; aligning an image from the first detector associated with a current step in the first sequence of successive steps with an image from the second detector associated with a previous step in the first sequence of successive steps; registering the second sequence of images from the first detector and the second detector at each step of the second series of successive steps; an optical device configured to align an image from the first detector associated with a current step in the second sequence of steps with an image from the second detector associated with a previous step in the second sequence of steps.

10. 10. An optical device according to claim 1, the sequence of images from the first detector and the second detector; a first sequence of images from the first detector and the second detector when the semiconductor wafer is in a first orientation at which the varied imaging parameter is varied; a second sequence of images from the first detector and the second detector when the semiconductor wafer is in a second orientation at which the varied imaging parameter is varied; the second orientation is an orientation rotated 180 degrees from the first orientation, the varied imaging parameters include at least one of a focus position, a wavelength, or a polarization of the semiconductor wafer; The optical system, wherein the indication is based on a variation in COS from the first image sequence and the second image sequence associated with the first orientation and the second orientation of the semiconductor wafer.

11. 11. The optical device according to claim 10, COS as a function of the changed imaging parameters from the first image sequence is COS ０ and COS as a function of the changed imaging parameter from the second image sequence is COS １８０ and The index is COS_TIS=(COS ０ +COS １８０ ) / 2, the second derivative of COS_TIS with respect to at least one of the varied imaging parameters, COS_COR=(COS ０ -COS １８０ ) / 2, or the second derivative of COS_COR with respect to at least one of the varied imaging parameters.

12. 11. An optical device according to claim 10, The optical device, wherein the index is further based on an accuracy index related to the accuracy of the COS based on multiple measurements.

13. 11. An optical device according to claim 10, An optical apparatus, wherein the indication is based on a measurement of asymmetry of one or more features on the semiconductor wafer based on the sequence of images.

14. 1. A method comprising: generating a sequence of images of the semiconductor wafer from a first detector and a second detector of an imaging assembly, wherein the first detector is configured to image the semiconductor wafer with a first imaging parameter and the second detector is configured to image the semiconductor wafer with a second imaging parameter different from the first imaging parameter, and at least one of the first imaging parameter or the second imaging parameter is varied over the sequence of images to thereby provide varied imaging parameters at one or more sites of the semiconductor wafer; providing an index based on the sequence of images that indicates a variation in center of symmetry (COS) of a first target feature and a second target feature as a function of the varied imaging parameter; identifying a landscape of values of the changed imaging parameter in which the index is below one or more predetermined limits; generating a recipe for metrology measurements in which the varied imaging parameters are set to values within the landscape; generating one or more metrology measurements associated with one or more product wafers based on the recipe.

15. 15. The method according to claim 14, further comprising: calibrating and / or correcting the one or more metrology measurements based on the variation in the COS.

16. 15. A method according to claim 14, comprising: The method, wherein the one or more metrology measurements include one or more overlay measurements.

17. 15. The method according to claim 14, further comprising: projecting a grid image onto the first detector and the second detector; The method, wherein one or more processors of a controller are further configured to align the sequence of images based on the projected grid image.

18. 15. A method according to claim 14, comprising: The method, wherein the varied imaging parameter is at least one of a focus position of the semiconductor wafer, a wavelength, a polarization, an angular offset of an aperture of the imaging assembly, or an angular position of at least one of the first detector or the second detector.

19. 15. A method according to claim 14, comprising: wherein the varied imaging parameter includes a focal position of the semiconductor wafer, the first detector and the second detector are configured to simultaneously image the semiconductor wafer at different values of the focal position separated by a fixed value ΔZ, and the sequence of images is generated by successive steps in which the values of the focal position associated with the first detector and the second detector are shifted by the fixed value ΔZ.

20. 20. The method according to claim 19, further comprising: projecting a grid image onto the first detector and the second detector; registering the images from the first detector and the second detector at each of the successive steps; and aligning an image from the first detector associated with a current one of the successive steps with an image from the second detector associated with a previous one of the successive steps at the common value of focus position.

21. 15. A method according to claim 14, comprising: the varied imaging parameter comprises a wavelength of at least one illumination beam of the one or more illumination beams; the first detector and the second detector are configured to simultaneously image the semiconductor wafer based on different values of the wavelength; the sequence of images a first image sequence in which the wavelength value associated with the first detector is held constant and the wavelength value associated with the second detector is successively changed in a first sequence of successive steps; a second image sequence in which the wavelength value associated with the second detector is held constant and the wavelength value associated with the first detector is successively changed in a second sequence of successive steps.

22. 22. The method according to claim 21, further comprising: projecting a grid image onto the first detector and the second detector; aligning the first sequence of images from the first detector and the second detector at each step of the first series of successive steps; aligning an image from the first detector associated with a current step in the first sequence of successive steps with an image from the second detector associated with a previous step in the first sequence of successive steps; registering the second sequence of images from the first detector and the second detector at each step of the second series of successive steps; and aligning an image from the first detector associated with a current step in the second sequence of successive steps with an image from the second detector associated with a previous step in the second sequence of successive steps.

23. 15. A method according to claim 14, comprising: the sequence of images from the first detector and the second detector; a first sequence of images from the first detector and the second detector when the semiconductor wafer is in a first orientation at which the varied imaging parameter is varied; a second sequence of images from the first detector and the second detector when the semiconductor wafer is in a second orientation at which the varied imaging parameter is varied; the second orientation is an orientation rotated 180 degrees from the first orientation, the varied imaging parameters include at least one of a focus position, a wavelength, or a polarization of the semiconductor wafer; The method, wherein the indicator is based on a variation in COS from the first image sequence and the second image sequence associated with the first orientation and the second orientation of the semiconductor wafer.

24. 24. The method of claim 23, COS as a function of the changed imaging parameters from the first image sequence is COS ０ and COS as a function of the changed imaging parameter from the second image sequence is COS １８０ and The index is COS_TIS=(COS ０ +COS １８０ ) / 2, the second derivative of COS_TIS with respect to at least one of the varied imaging parameters, COS_COR=(COS ０ -COS １８０ ) / 2, or a second derivative of COS_COR with respect to at least one of the varied imaging parameters.

25. 15. A method according to claim 14, comprising: A method wherein the index is further based on an accuracy index related to the accuracy of the COS based on multiple measurements.

26. 15. An optical device according to claim 14, The method, wherein the indicator is based on a measurement of asymmetry of one or more features on the semiconductor wafer based on the sequence of images.

Images