Method of manufacturing device

Abstract

A method for analyzing a process, the method comprising: obtaining a multidimensional probability density function representing an expected distribution of values ​​of a plurality of process parameters; obtaining a performance function relating the values ​​of the process parameters to a performance measure of the process; and using the performance function to map the probability density function to a performance probability function with the process parameters as independent variables.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Application No. 62 / 774,480 filed on December 3, 2018 and U.S. Application No. 62 / 814,544 filed on March 6, 2019, the entire contents of which are incorporated herein by reference. Technical Field

[0003] The present description relates to methods of manufacturing devices, particularly using photolithographic processes, and to analysis and control of the manufacturing process. Background Art

[0004] There is a continuing need to fabricate devices, such as integrated circuits, with smaller features. Integrated circuits and other micro-devices are typically fabricated using optical lithography, but other fabrication techniques such as imprint lithography, electron beam lithography, and nanoscale self-assembly are also known. Summary of the invention

[0005] When printing a pattern using, for example, photolithography, one or more various parameters of the exposure, such as focus and dose, can be changed to affect the characteristics of the printed pattern, such as CD and CDU. Similarly, one or more parameters of the process steps (e.g., etching time) can also be varied. Conventionally, a process matching scheme (which specifies the values ​​of controllable parameters to be used during the exposure and subsequent process steps of a particular pattern) is optimized to maximize yield. Yield is typically defined as the proportion of correctly printed devices. In some cases, the exposure matching scheme can be optimized to maximize output (i.e., the exposure rate of the substrate). In some cases, the exposure matching scheme is optimized to maximize the process window (i.e., the tolerance within which one or more control parameters can be varied from their nominal values ​​while still obtaining effective exposure).

[0006] However, optimizing process recipes often involves various rules of thumb or best practices, since fully characterizing the effects of varying all variable parameters of the exposure and subsequent process steps involves extremely large numbers of trial exposures and measurements or simulations.

[0007] Improved methods for optimizing and controlling device fabrication processes are needed.

[0008] According to one embodiment, a method for analyzing a process is provided, the method comprising: obtaining a multidimensional probability function representing an expected distribution of values ​​of a plurality of process parameters; obtaining a performance function relating the values ​​of the process parameters to a performance metric of the process; and using the performance function to map the multidimensional probability function to a performance probability function with the process parameters as independent variables.

[0009] According to one embodiment, a device manufacturing method is provided, the method comprising: exposing a radiation-sensitive layer on a substrate to form a latent image therein; developing the radiation-sensitive layer to fix the latent image; and transferring the fixed image to the substrate, wherein the exposure, development and / or transfer are performed using a set of nominal process settings determined by the above method.

[0010] According to one embodiment, a device manufacturing method is provided, the method comprising: exposing a radiation-sensitive layer on a substrate to form a latent image therein; developing the radiation-sensitive layer to fix the latent image into a developed image; transferring the fixed image onto a substrate to form a device feature or a measurement target; measuring characteristics of the latent image, the developed image, the device feature and / or the measurement target; and determining process settings for subsequent exposure, development and / or transfer steps using a performance probability function determined according to the above method and the measured characteristics.

[0011] According to one embodiment, a computer program is provided, the computer program comprising instructions which, when executed by one or more computers and / or lithography tools, instruct the computers and / or lithography tools to perform the methods described herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Embodiments of the invention will now be described, by way of example only, with reference to the accompanying schematic drawings, in which:

[0013] Figure 1 The lithographic unit is depicted;

[0014] Figure 2 depicts the lithographic apparatus;

[0015] Figure 3 A schematic overview of the measurement equipment is depicted;

[0016] Figure 4 The overall photolithography approach is depicted;

[0017] Figure 5 The effect of varying the etching time during the etching step on the critical dimension (CD) is depicted;

[0018] Figure 6(a) , 6(b) , 6(c) and 6(d) depict device fabrication methods;

[0019] Figure 7 is a graph showing an exemplary multivariate normal distribution of overlap and CD values ​​in a fabricated device having a particular covariance matrix;

[0020] Figure 8 is a graph showing the relationship between edge position error (EPE), overlay (OVL), and CD in an exemplary device pattern;

[0021] Fig. 9 is a graph showing the relationship between edge position error and overlay at three different CDs in an exemplary device pattern; the lines in the graph represent Figure 8 A cross section of the diagram in ;

[0022] Figures 10 to 12 is a graph showing exemplary multivariate normal distributions of overlay and CD values ​​in fabricated devices for different amounts of correlation between overlay and CD;

[0023] Fig.13 is a graph showing the distribution of edge position errors for different amounts of correlation between set overlap and CD distributions;

[0024] Fig.14 and 15 are plots of the distribution of edge position errors for different values ​​of mean CD and overlap, respectively, where the predicted probability of an edge position error < 6 nm for an individual event is denoted by FR (“Failure Rate”);

[0025] Fig.16 is a plot of device failure rate as a function of focus and dose, which indicates the feasible process window;

[0026] Fig.17 is a graph showing measured device failure rate as a function of overlap;

[0027] Fig.18 is a flowchart of a method for determining a process window according to an embodiment of the present invention;

[0028] Fig.19 is a flow chart of a device manufacturing method according to an embodiment of the present invention;

[0029] Fig. 20 is a flow chart of a device manufacturing method according to an embodiment of the present invention;

[0030] Fig.21 is a diagram showing the derivation of a control window from synthetic data; and

[0031] Fig. 22 is a graph showing the distribution of false positive and false negative estimates of yield. DETAILED DESCRIPTION

[0032] exist Figure 1 A lithography unit is schematically depicted in FIG. A lithography unit is a device comprising a lithography apparatus (e.g., as in Figure 2 ), one or more substrate processing devices or apparatuses, and inspection equipment (e.g., as schematically depicted in Figure 3 The lithography unit can be configured to perform an overall lithography process using a plurality of lithography units. Figure 4 The strict control loop depicted in FIG. The lithographic apparatus, the lithographic unit and the overall lithographic process are described below. The lithographic apparatus, the lithographic unit and the inspection apparatus can be used in a device manufacturing method according to an embodiment of the present invention, which will be further described below.

[0033] Figure 2 A lithographic apparatus LA is schematically depicted. The lithographic apparatus LA comprises an illumination system (also referred to as an illuminator) IL configured to condition a radiation beam B (e.g., UV radiation, DUV radiation, or EUV radiation), a patterning device support (e.g., a mask stage) MT configured to support a patterning device (e.g., a mask) MA and connected to a first positioner PM configured to accurately position the patterning device MA according to certain parameters, a substrate support (e.g., a wafer stage) WT configured to hold a substrate (e.g., a wafer coated with a resist) W and connected to a second positioner PW configured to accurately position the substrate support according to certain parameters, and a projection system (e.g., a refractive projection lens system) PS configured to project a pattern imparted to the radiation beam B through the patterning device MA onto a target portion C (e.g., comprising one or more dies) of the substrate W.

[0034] In operation, the illumination system IL receives a radiation beam from a radiation source SO, for example, via a beam delivery system BD. The illumination system IL may include various types of optical components, such as refractive, reflective, magnetic, electromagnetic, electrostatic and / or other types of optical components, or any combination thereof, for directing, shaping and / or controlling the radiation. The illuminator IL may be used to condition the radiation beam B so that it has a desired spatial and angular intensity distribution in its cross-section at the plane of the patterning device MA.

[0035] The term "projection system" PS as used herein should be broadly interpreted as covering various types of projection systems, including refractive, reflective, catadioptric, anamorphic, magnetic, electromagnetic and / or electrostatic optical systems, or any combination thereof, which projection system is appropriate for the exposure radiation used and / or other factors (such as the use of immersion liquid or the use of vacuum). Any use of the term "projection lens" herein may be considered synonymous with the more general term "projection system" PS.

[0036] The lithographic apparatus LA may be of a type in which at least part of the substrate may be covered by a liquid having a relatively high refractive index (e.g., water) to fill the space between the projection system PS and the substrate W, which is also referred to as immersion lithography. More information on immersion technology is given in U.S. Pat. No. 6,952,253, which is incorporated herein by reference in its entirety.

[0037] The lithographic apparatus LA may also be of a type having two or more substrate supports WT (also referred to as "dual stage"). In such a "multi-stage" machine, the substrate supports WT may be used in parallel, and / or steps for preparing a substrate W for subsequent exposure may be performed on a substrate W located on one of the substrate supports WT, while another substrate W on another substrate support WT is used to expose a pattern on another substrate W.

[0038] In addition to the substrate support WT, the lithographic apparatus LA may comprise a measuring table. The measuring table is arranged to hold sensors and / or cleaning means. The sensors may be arranged to measure properties of the projection system PS or properties of the radiation beam B. The measuring table may accommodate a plurality of sensors. The cleaning means may be arranged to clean a part of the lithographic apparatus, for example a part of the projection system PS or a part of a system for providing immersion liquid. The measuring table may be moved under the projection system PS when the substrate support WT is away from the projection system PS.

[0039] In operation, a radiation beam B is incident on a patterning device (e.g. a mask) MA, which is held on a support MT, and is patterned by a pattern (design layout) present on the patterning device MA. After passing through the patterning device MA, the radiation beam B passes through a projection system PS, which focuses the beam onto a target portion C of the substrate W. With the aid of a second positioner PW and a position measurement system IF, the substrate support WT can be accurately moved, for example, in order to position different target portions C at focus and alignment positions in the path of the radiation beam B. Similarly, the first positioner PM and possibly another position sensor (in Figure 1 The patterning device MA and the substrate W may be aligned using patterning device alignment marks M1, M2 and substrate alignment marks P1, P2. Although the substrate alignment marks P1, P2 as shown occupy dedicated target portions, they may be located in spaces between target portions. When the substrate alignment marks P1, P2 are located between target portions C, they are referred to as scribe line alignment marks.

[0040] like Figure 1As shown, the lithography apparatus LA may form part of a lithography unit LC, which is sometimes also referred to as a lithography cell or (lithography) cluster, which typically also includes equipment for performing pre-exposure and post-exposure processes on a substrate W. Conventionally, these include, for example, a spin coater SC for depositing a resist layer, a developer DE for developing the exposed resist, a cooling plate CH and a baking plate BK, for example for regulating the temperature W of the substrate, for example for regulating the solvent in the resist layer. A substrate handler or robot RO picks up the substrate W from the input / output ports I / O1, I / O2, moves the substrate W between different processing equipment, and transports the substrate W to a loading station LB of the lithography apparatus LA. The devices in the lithography unit (also usually collectively referred to as tracks) are typically controlled by a track control unit TCU, which itself may be controlled by a supervisory control system SCS, which may also control the lithography apparatus LA, for example via the lithography control unit LACU.

[0041] In order for the substrate W exposed by the lithographic apparatus LA to be correctly and consistently exposed, it is desirable to inspect the substrate to measure one or more properties of the patterned structure, such as overlay error values ​​between subsequent layers, line thickness values, critical dimension values ​​(CD), etc. To this end, an inspection tool (not shown) may be included in the lithography cell LC. If an error is detected, for example, adjustments may be made to the exposure of subsequent substrates or other process steps to be performed on the substrate W, particularly if the inspection is performed before other substrates W of the same batch or lot are still to be exposed or processed.

[0042] The inspection apparatus (which may also be referred to as a metrology apparatus) is used to determine one or more properties of a substrate W, in particular to determine how the values ​​of one or more properties vary from one substrate to another W or how the values ​​of one or more properties associated with different layers of the same substrate W vary between layers. The inspection apparatus may alternatively be configured to identify defects on the substrate W and may, for example, be part of a lithography cell LC or may be integrated into a lithography apparatus LA or may even be a stand-alone device. The inspection apparatus may measure one or more properties on a latent image (the image in the resist layer after exposure), a semi-latent image (the image in the resist layer after a post-exposure bake step PEB) or a developed resist image (wherein exposed or unexposed parts of the resist have been removed) or even an etched image (after a pattern transfer step such as etching).

[0043] Inspection tools are often referred to as metrology tools. Different types of metrology tools for making such measurements are known, including scanning electron microscopes or various forms of scatterometer metrology tools. A scatterometer is a versatile instrument that can measure parameters of a lithography process by mounting a sensor in the pupil or conjugate plane of the objective of the scatterometer, the measurements being often referred to as pupil-based measurements, or mounting the sensor in the image plane or a plane conjugate to the image plane, in which case these measurements are often referred to as image-based or field-based measurements. Such scatterometers and associated measurement techniques are further described in U.S. Patent Application Publication Nos. US 2010-0328655, US 2011-102753, US 2012-0044470, US 2011-0249244, and US 2011-0026032, and European Patent Application Publication No. EP1628164. Each of the foregoing patent application publications is incorporated herein by reference in its entirety. The aforementioned scatterometers can measure gratings using radiation from soft X-rays as well as radiation in the visible to near IR wavelength range.

[0044] In one embodiment, the scatterometer is an angle-resolved scatterometer. In such a scatterometer, reconstruction methods can be applied to the measured signal to reconstruct or calculate one or more properties of the periodic structure or grating. For example, such a reconstruction can be performed by simulating the interaction of the scattered radiation with a mathematical model of the target structure and comparing the simulation results with the measured results. The parameters of the mathematical model are adjusted until the simulated interaction produces a diffraction pattern similar to that observed from an actual target.

[0045] In one embodiment, the scatterometer is a spectroscopic scatterometer. In such a spectroscopic scatterometer, radiation emitted by a radiation source is directed onto a target, and reflected or scattered radiation from the target is directed to a spectrometer detector, which measures the spectrum of the specularly reflected radiation (i.e., a measure of intensity as a function of wavelength). From this data, the structure or profile of the target that produced the detected spectrum can be reconstructed, for example, by rigorous coupled wave analysis and nonlinear regression or by comparison with a library of simulated spectra.

[0046] In one embodiment, the scatterometer is an ellipsometer. An ellipsometer allows one or more parameters of a lithography process to be determined by measuring the scattered radiation in each polarization state. Such a metrology device emits polarized radiation (such as linear, circular or elliptical polarization) by using, for example, one or more appropriate polarization filters in the illumination portion of the metrology device. A source suitable for the metrology device may also provide polarized radiation. Various embodiments of existing ellipsometers are described in U.S. Patent Application Publication Nos. US 2007 / 0296960, US 2008 / 0198380, US 2009 / 0168062, US 2010 / 0007863, US 2011 / 0032500, US 2011 / 0102793, US 2011 / 0188020, US 2012 / 0044495, US 2013 / 0162996, and US 2013 / 0308142 (each of which is incorporated herein by reference in its entirety).

[0047] In one embodiment, the scatterometer is adapted to measure the overlap of two misaligned gratings or periodic structures by measuring the reflection spectrum and / or the asymmetry of the detection configuration, which asymmetry is related to the degree of overlap. The two (usually overlapping) grating structures can be applied in two different layers (not necessarily continuous layers) and can be formed at approximately the same position on the substrate. The scatterometer can have a symmetrical detection configuration, such as described in European Patent Application Publication No. EP1,628,164 (which is incorporated herein by reference in its entirety), so that any asymmetry is clearly distinguishable. This provides a simple method of measuring misalignment in gratings. Other examples of measuring the overlay error between two layers containing a periodic structure as a target by the asymmetry of the periodic structure can be found in PCT Patent Application Publication No. WO 2011 / 012624 and U.S. Patent Application Publication No. US 2016 / 0161863 (each of which is incorporated herein by reference in its entirety).

[0048] Other parameters of interest may be focus and dose. As described in U.S. Patent Application Publication No. US2011 / 0249244 (which is incorporated herein by reference in its entirety), focus and dose may be determined simultaneously by scatterometry (or alternatively by scanning electron microscopy). A single structure may be used that has a unique combination of critical dimension and sidewall angle measurements for each point in the focus energy matrix (FEM - also called focus exposure matrix). If these unique combinations of critical dimension and sidewall angle are available, then focus and dose values ​​may be uniquely determined from these measurements.

[0049] The metrology target may be a collection of composite gratings, which are formed by a lithography process, mainly in resist, but may also be formed after, for example, an etching process. Typically, the pitch and line width of the structures in the grating strongly depend on the measurement optics, in particular the numerical aperture (NA) of the optics, in order to be able to capture the diffraction orders from the metrology target. As previously mentioned, the diffraction signal may be used to determine the offset between two layers (also called "overlay"), or may be used to reconstruct at least part of the original grating produced by the lithography process. This reconstruction may be used to provide guidance on the quality of the lithography process, and may be used to control at least part of the lithography process. The target may have smaller sub-segments, which are configured to mimic the size of a functional part of the design layout in the target. Due to this segmentation, the target will behave more similarly to the functional part of the design layout, so that the overall process parameter measurement better resembles the functional part of the design layout. The target may be measured in an underfill mode or an overfill mode. In the underfill mode, the measurement beam generates a spot that is smaller than the entire target. In the overfill mode, the measurement beam generates a spot that is larger than the entire target. In this overfill mode, different targets can also be measured simultaneously and thus different process parameters can be determined simultaneously.

[0050] The overall measurement quality of a lithography parameter using a particular target is determined at least in part by the measurement profile used to measure the lithography parameter. The term "substrate measurement profile" may include one or more parameters of the measurement itself, one or more parameters of the one or more patterns measured, or both. For example, if the measurement used in the substrate measurement profile is a diffraction-based optical measurement, the one or more parameters measured may include the wavelength of the radiation, the polarization of the radiation, the angle of incidence of the radiation relative to the substrate, the orientation of the radiation relative to the pattern on the substrate, and the like. One of the criteria for selecting a measurement profile may, for example, be the sensitivity of one of the measurement parameters to process variations. More examples are described in U.S. Patent Application Publication Nos. US2016 / 0161863 and US 2016 / 0370717 (each of which is incorporated herein by reference in its entirety).

[0051] Typically, the patterning process in the lithographic apparatus LA is an important step in the process and involves high accuracy in the sizing and placement of structures on the substrate W. To help ensure this high accuracy, three systems can be combined in a controlled environment, such as Figure 4As shown. One of these systems is a lithography apparatus LA that is (virtually) connected to a metrology tool MET (second system) and a computer system CL (third system). The purpose of such a control environment is to optimize the cooperation between these three systems to enhance the overall process window and to provide a tight control loop to help ensure that the patterning performed by the lithography apparatus LA remains within the process window. The process window defines a range of process parameters (e.g., dose, focus, coverage) within which a specific manufacturing process will produce a defined result (e.g., a functional semiconductor device), and within which process parameters in the lithography process or patterning process are usually allowed to vary.

[0052] The computer system CL can use (a portion of) the design layout to be patterned to predict which resolution enhancement technique to use, and perform computational lithography simulations and calculations to determine which patterning device (e.g., mask) layout and lithography equipment settings can achieve a larger or maximum overall process window (in terms of resolution enhancement techniques) for the patterning process. Figure 3 Typically, the resolution enhancement technique is arranged to match the patterning possibilities of the lithographic apparatus LA. The computer system CL may also be used to detect at which position within the process window the lithographic apparatus LA is currently operating (e.g. using input from a metrology tool MET) in order to predict whether defects may be present due to, for example, suboptimal processing (e.g., at the process window in which the lithographic apparatus LA is currently operating). Figure 3 , represented by an arrow pointing to “0” of the second scale SC2).

[0053] The metrology tool MET may provide input to the computer system CL to enable accurate simulations and predictions, and may provide feedback to the lithographic apparatus LA to identify possible drifts, e.g., in the calibration state of the lithographic apparatus LA (in Figure 3 SC3).

[0054] Figure 3 A schematic diagram of a metrology apparatus 302 is depicted, wherein radiation having a wavelength ranging from 0.1 nm to 100 nm may be used to measure one or more parameters of a structure on a substrate. Figure 3 The metrology device 302 presented in FIG. 3 is suitable for the soft X-ray (SXR) or EUV domain.

[0055] Just as an example, Figure 3 A schematic physical arrangement of a metrology apparatus 302 is shown, the metrology apparatus 302 comprising a spectroscopic scatterometer using EUV and / or SXR radiation.An alternative form of inspection apparatus may be provided in the form of an angle-resolved scatterometer.

[0056] Inspection apparatus 302 includes a radiation source 310 , an illumination system 312 , a substrate support 316 , detection systems 318 , 398 , and a metrology processing unit (MPU) 320 .

[0057] In this example, source 310 includes a generator of EUV or soft X-ray radiation based on high harmonic generation (HHG) technology. An alternative to the HHG source is a discharge produced plasma (DPP) source, a free electron laser (FEL) source, an inverse Compton scattering (ICS) source, or a compact synchrotron. The HHG source is available from KMLabs in Boulder, Colorado, USA. The main components of the radiation source are a drive laser 330, a LAS, and a HHG gas unit 332 (HHG). A gas supply 334 (GS) supplies a suitable gas to the gas unit, where the gas can be optionally ionized by a power supply 336. The drive laser 300 can be, for example, a fiber-based laser with an optical amplifier, which generates infrared radiation pulses as needed, each pulse duration can be less than 1 ns (1 nanosecond), and the pulse repetition frequency is up to several megahertz. The wavelength of the infrared radiation can be, for example, in the range of 1 μm (1 micrometer). The laser pulses are transmitted as a first radiation beam 340 to the HHG gas cell 332 where a portion of the radiation is converted in the gas to a higher frequency than the first radiation to form a beam 342 of coherent second radiation including one or more desired wavelengths.

[0058] The second radiation may contain multiple wavelengths. If the radiation is monochromatic, the measurement calculations (e.g., reconstruction of the structure of interest) may be simplified. The volume of the gas in the gas cell 332 defines the HHG space, but the space does not need to be completely enclosed and the flow of the gas may be used instead of the static volume. The gas may be, for example, a rare gas, such as neon (Ne) or argon (Ar). N2, O2, He, Kr, Xe gas or a mixture thereof (including Ne or Ar) may be used. These are all issues of design choice and may even be selectable options within the same device. For example, when imaging structures of different materials, different wavelengths will provide different levels of contrast. In order to inspect metal structures or silicon structures, for example, a wavelength different from that used to image the features of (carbon-based) resists or to detect contamination of such different materials may be selected. One or more filtering devices 344 may be provided. For example, filters such as thin films of aluminum (Al) may be used to prevent fundamental IR radiation from further entering the inspection device. A grating (not shown) may be provided to select one or more specific harmonic wavelengths from the wavelengths generated in the gas cell. Part or all of the optical path may be contained within a vacuum environment, noting that EUV and SXR radiation are absorbed when traveling through air. Various components of the radiation source 310 and illumination optics 312 may be adjustable to enable different measurement "options" within the same device. For example, different wavelengths and / or polarizations may be made selectable.

[0059] Depending on the material of the structure being inspected, different wavelengths may provide the desired level of penetration into the underlying layers. In order to address the smallest device features and defects in the smallest device features, short wavelengths may be preferred. For example, one or more wavelengths in the range of 1-20nm, or alternatively in the range of 1-10nm, or alternatively in the range of 10-20nm may be selected. Wavelengths shorter than 5nm have very low critical angles when reflecting off materials of general interest in semiconductor manufacturing. Therefore, selecting a wavelength greater than 5nm will provide a stronger signal at a higher angle of incidence. On the other hand, if the inspection task is to detect the presence of a certain material (e.g., detecting contamination), wavelengths up to 50nm may be useful.

[0060] From the radiation source 310, the filtered beam 342 enters the inspection chamber 350, in which the substrate W including the structure of interest is held in a measurement position for inspection by the substrate support 316. The structure of interest is marked as T. The pressure in the inspection chamber 350 is maintained near vacuum by a vacuum pump 352 (VAC), so that, for example, EUV or SXR radiation can pass through the focusing system without being unduly attenuated. The illumination system 312 has the function of focusing the radiation into a focused beam 356, and may include, for example, a two-dimensional curved mirror or a series of one-dimensional curved mirrors, as described in U.S. Patent Application Publication No. US2017 / 0184981 (which is incorporated herein by reference in its entirety). When projected onto the structure of interest, focusing is performed to obtain a circular or elliptical spot S with a diameter of less than about 10 μm. This can be achieved by imaging the source directly onto the image plane or by imaging a small aperture illuminated by the source onto the image plane. The substrate support 316 includes, for example, an XYZ translation stage and a rotation stage, by which any portion of the substrate W can be brought to the focus of the beam in a desired orientation. Thus, a radiation spot S is formed on the structure of interest. Alternatively or additionally, the substrate support 316 comprises, for example, a tilting stage, which can tilt the substrate W at a certain angle to control the incident angle of the focused beam on the structure of interest T.

[0061] Optionally, the illumination system 312 provides a reference radiation beam to a reference detector 314, which can be configured to measure the spectrum and / or intensity of different wavelengths in the filtered beam 342. The reference detector 314 can be configured to generate a signal 315 that is provided to the measurement processing unit 320 and can include information about the spectrum of the filtered beam 342 and / or the intensity of different wavelengths in the filtered beam.

[0062] Reflected radiation 360 is captured by detector 318, and spectrum 382 is provided to processing unit 320 for calculating properties of target structure T. Illumination system 312 and detection system 318 thus form an inspection apparatus. The inspection apparatus may include a soft X-ray and / or EUV spectroscopic reflectometer as described in U.S. Patent Application Publication No. 2016 / 282282, which is incorporated herein by reference in its entirety.

[0063] If the target T has a certain periodicity, the radiation of the focused beam 356 may also be partially diffracted. The diffracted radiation 397 follows another path at a well-defined angle relative to the incident angle and the reflected radiation 360. Figure 3 In , the plotted diffracted radiation 397 is plotted in a schematic manner, and the diffracted radiation 397 may follow many other paths besides the plotted path. The inspection device 302 may also include one or more additional detection systems 398 that detect and / or image at least a portion of the diffracted radiation 397. Figure 3, a single further detection system 398 is drawn, but embodiments of the inspection device 302 may comprise a plurality of further detection systems 398, which are arranged at different positions to detect and / or image the diffracted radiation 397 in a plurality of diffraction directions. In other words, the (higher) diffraction orders of the focused radiation beam incident on the target T are detected and / or imaged by one or more further detection systems 398. The one or more detection systems 398 generate a signal 399 which is provided to the metrology processing unit 320. The signal 399 may comprise information of the diffracted radiation 397, and / or may comprise an image obtained from the diffracted radiation 397.

[0064] To help align and focus the spot S with the desired product structure, the inspection device 302 may provide auxiliary optical elements that use auxiliary radiation under the control of the measurement processor 320. The measurement processor 320 may also communicate with a position controller 372, which operates a translation stage, a rotation stage, and / or a tilt stage. The processor 320 receives highly accurate feedback about the position and orientation of the substrate via one or more sensors. The one or more sensors 374 may include, for example, one or more interferometers or encoders that can measure the position of the substrate support 316 with an accuracy in the picometer (pm) range. In operation of the inspection device 302, the spectral data 382 captured by the detection system 318 is transmitted to the measurement processing unit 320.

[0065] In one embodiment, the inspection device uses soft X-rays and / or EUV radiation at any suitable angle of incidence (e.g., normal incidence, near normal incidence, or grazing incidence), for example, to perform diffraction-based asymmetry measurements. The angle of incidence can also be in the range of 20 to 40 degrees, for example, 30 degrees.

[0066] Inspection equipment may be provided in a hybrid metrology system. Performance parameters to be measured may include overlay (OVL, OV), critical dimension (CD), via contact landing (combination of CD and OVL), and / or CD uniformity (CDU). Measurement techniques may include coherent diffraction imaging (CDI) and resolution overlay (ARO) metrology.

[0067] The soft X-ray and / or EUV radiation may, for example, have a wavelength of less than 100 nm, for example, radiation in the range of 5-30 nm or alternatively in the range of 10 nm to 20 nm is used. The radiation may be narrow-band or broadband in character. The radiation may have discrete peaks in a particular wavelength band, or may have a more continuous character.

[0068] Inspection device 302 can be used to measure structures in resist material that has been processed in a lithography unit (e.g., after development inspection (ADI)), and / or can be used to measure structures after the structures have been formed in a harder material (e.g., after etching inspection (AEI)). For example, inspection device 302 can be used to inspect a substrate after it has been processed by a development device, an etching device, an annealing device, and / or other devices (including thin layer deposition and ion implantation (doping) devices).

[0069] When developing a new process for forming one or more layers of a device using a photolithographic process, a great deal of work must be done to determine the appropriate settings for exposure (e.g., focus and dose) as well as the appropriate settings for subsequent process steps (e.g., etch time and deposition thickness). A common goal when developing a process is to improve or maximize the yield of the process, that is, the proportion of manufactured devices that meet specification requirements. A process for a particular layer can be improved or optimized by reference to specific performance metrics that are most closely related to yield (e.g., edge position error (EPE)) and by reference to specific features that are important or critical to the proper operation of the device.

[0070] Such a performance metric can be related to one or more other characteristics of the pattern printed on the substrate. For example, the edge position error is related to the critical dimension (CD) and the overlap (pattern position error). CD and overlap are affected by one or more tool control parameters (e.g., dose and / or focus). Therefore, when printing a pattern or feature multiple times on each substrate and on multiple substrates, the actual value of one or more characteristics will be distributed around the corresponding target value. One or more tool control parameters often exhibit strong correlations. For example, the edge of the substrate is usually not clamped to the substrate holder with the same force as the center of the substrate, resulting in substrate bending. The lithographic equipment can try to compensate for this using a focus control loop that leaves a significant residual error. In other words, there may be a significant focus error at the edge of the substrate. This focus error leads to a CD error. In addition, the non-telecentricity of the projection system means that focus errors may also lead to overlap errors.

[0071] Additionally, further processing of the geometries transferred to the resist is often non-linear, e.g. by virtue of etch loading, non-conformal behavior of deposition and / or recesses in chemical mechanical polishing, depending on e.g. the local density of the pattern. Strong correlations between measured values ​​of one or more parameters (e.g. CD and overlay), as well as non-linear transfer of geometries of actual interest towards device yield, can lead to non-normal distributions.

[0072] Typical fabricated devices have a large number of geometric features (>10 9). If one critical feature fails, the device will malfunction. Therefore, it is desirable to measure and inspect a very large number of features to help ensure that the proper settings for lithography and processing equipment (OVL, CD, etch time, etc.) are fully optimized to achieve good yield.

[0073] Therefore, to fully determine the impact of changes in one or more process settings on yield, extensive testing and measurement is required.Therefore, it is desirable to have a better method for optimizing the lithography process, for example, a method that requires less time and labor and / or provides improved results.

[0074] Therefore, an embodiment of the present invention provides a method 2100, such as Fig.18 As shown, the method 2100 is used to analyze a process, for example, so that the process is improved or optimized, or for process control in production. The method includes obtaining 2101 a multidimensional probability density function (as an example of a multidimensional probability function), which represents the expected distribution of values ​​of multiple process parameters. The process parameters can be process settings, that is, control values ​​applied to a process tool, such as focus or dose, or characteristics of a pattern, such as CD, which are stable and controlled by one or more process settings. Characteristics such as CD are related to focus and dose in a stable manner, so that considering the distribution of CD rather than the underlying one or more process settings can simplify the methods described herein. The values ​​of the process parameters can be measured directly on one or more substrates, and a model suitable for the distribution type is used. For example, these values ​​can be fitted using a multivariate normal distribution:

[0075]

[0076] where the variables have a specific correlation which can be determined by its unbiased estimator through the measurement of the individual parameters on the substrate:

[0077] ∑ xy =sum ij (X i -mean(X))*(Y j -mean(Y)) / sqrt(n-1) (2) where n is the number of measurements. These will then populate the covariance matrix Sigma described previously.

[0078] By applying a defect-based process window approach in the exposure matrix (e.g., focus layer 1 - dose layer 1 - ovl - focus layer 2 - dose layer 2), not only the tool scale (approximately mm 2 and larger), but also changes in the distribution of process parameters at local scales (feature-to-feature) can be considered.

[0079] The method also includes obtaining 2102 a performance function that relates the value of a process parameter to a performance metric of the process. The performance function may be referred to herein as a transfer function or transfer curve. The performance function is then used to map 2103 the probability density function of the individual tool parameters to a performance probability function with the process parameters as independent variables. The performance probability function may then be used to analyze 2104 the process, for example, to determine the probability of normal operation of a manufactured device for a given set of values ​​of the process parameters. The probability of correct operation may be considered equivalent to the probability that the performance metric is within a desired limit, for example, the EPE value of the device is less than 6 nm.

[0080] Embodiments of the invention are advantageous because fewer measurements and / or simulations are required to obtain the probability density function and the performance function than to obtain directly the function linking the process parameters to the yield. In order to qualify the overall distribution function related to the yield (particularly the tail behavior), a large number of measurements (e.g., about 10 in a typical plant) are required. 9 ). If the transfer curve or one parameter changes, it is necessary to redo the entire qualification check. Tool parameters (OVL / CD) can be measured at a higher throughput than the throughput limit in the lithography tool (e.g., requiring TEM), and distribution types of tool parameters generally perform better, i.e., require fewer parameters to model.

[0081] The desired process parameter is selected based on how the values ​​of the process parameter are distributed, for example, based on some characteristic of its associated statistical distribution. For example, the characteristic is based on similarity to a normal distribution (sometimes called a Gaussian distribution). A normal distribution can be characterized (i.e., to determine the mean and variance) using fewer data points than other distributions. A normal distribution can typically be characterized using 10 2 to 10 4 To fully characterize by direct measurement of key performance metrics, -6 A failure rate of magnitude of 10 9 If the parameter of interest does not have a normal distribution, one or more base parameters that have a normal distribution can be identified and used instead. Alternatively, one can rely on parameters that are commonly measured by lithography and processing equipment and use their distribution directly.

[0082] In one embodiment, the selected process parameters are at least partially correlated. The correlation between process parameters may have a significant impact on the failure rate, so it needs to be properly considered. Selecting correlated process parameters can improve the accuracy of the results without increasing the number of measurements.

[0083] Embodiments of the present invention provide particular advantages where the performance function is nonlinear, e.g., edge location error. When the performance function is nonlinear, small changes in conditions can lead to large changes in failure rate. Therefore, conventional techniques for characterizing a process involve a large number of measurements, especially when the nonlinear location of the performance function is not known.

[0084] In one embodiment of the invention, the method further includes identifying a subspace of process parameter values ​​for which the performance probability function meets a criterion (e.g., failure rate is below a selected threshold), e.g., one or more ranges of process parameters. Depending on the selected process parameters, the subspace (defined by one or more ranges, which define a closed space embedded within the space generated by the dimensions associated with the individual process parameters) can be a process window, or can allow a process window to be derived. Thus, a set of nominal process settings for a lithography process that falls within the subspace can be manually or automatically selected for use in a device manufacturing method.

[0085] Embodiments of the present invention may be used in method 2200 for process validation, such as Fig.19 As shown. First, an initial selection is obtained 2201. The substrate is exposed and measured 2202 or simulated (tested). As described above, the measurement or simulation results are analyzed 2203 to derive a performance probability function and an optional process window. If the analysis shows that the selection is defective, for example, the process window is insufficient, the selection is improved 2204. Steps 2202 and 2203 can be repeated as needed until the process is satisfactory. The production substrate is exposed 2205.

[0086] In an embodiment of the present invention, the performance probability function may be Fig. 20 2300 for process control. The substrate is exposed 2301 and optionally processed to transfer the pattern formed in the exposure step to the substrate (a layer of the substrate). Then, a characteristic selected from at least one of the following is measured 2302: a latent image, a developed image, a device feature and / or a measurement target. The performance probability function determined as described above and the measured characteristics can be used to determine 2303 process settings for subsequent exposure, development or transfer steps 2304. For example, if the CD or overlap measurement indicates that the substrate is near the edge of the process window, the etching time can be compensatory adjusted, or a more reliable etching tool can be selected. Alternatively, if the analysis 2303 indicates that the probability of failure is too high, the substrate can be reprocessed 2305. In this way, scrap and rework rates can be reduced, thereby improving yield.

[0087] Embodiments of the present invention may be embodied in a computer program comprising instructions that, when executed by one or more lithography tools, instruct the lithography tools to perform all or part of the above method. The steps of the method may be performed in a computer integrated in a lithography apparatus and / or other tools (such as a metrology tool or a processing tool) or in a separate computing system.

[0088] Embodiments of the present invention will be further described below with reference to exemplary device patterns, but it will be readily appreciated that the principles of the present invention are equally applicable to other patterns.

[0089] Figure 5 is an example of a nonlinear transfer function. In the top figure, the bottom CD measured at the interface of two layers (e.g., resist and an underlying hard mask to which the pattern is initially transferred before further processing) is shown as a function of etch time. In the bottom figure, the CD measured in the underlying hard mask is shown as a function of the CD measured in the resist. It can be seen that the CD at the interface has a strong nonlinear characteristic: it is negligible until the etch time reaches a certain level, and then rises rapidly before stabilizing. In addition, after etching, the relationship between the CD at the interface and the CD of the underlying hard mask is nonlinear, so that the relationship between the process settings that affect the CD in the resist and the CD in the underlying hard mask is complex.

[0090] Another example of a nonlinear process is shown in Figures 6(a) to (d). Figure 6(a) shows two metal lines (upper 601, lower 602) that need to be connected to each other through a via 603. In this example, the y-direction of the via 603 is self-aligned relative to the upper line 601 through a dual damascene process flow. The bottom of the via 603 in the horizontal / x direction is not self-aligned with the via, but has a certain designed taper, as shown in Figure 6(b). The bottom of the via 603 should have a minimum spacing (EPE) compared to adjacent lines to avoid short circuits or electromigration related reliability issues. Figure 6(c) is a graph of EPE versus OVL for different ΔCD values, which shows that the minimum distance (EPE) is a nonlinear function of both CD and OVL, due to etch loading and related via buckling (etching tends to attract the closest line in the layer below). FIG6( d ) is a graph showing the probability density versus EPE for different CD and OVL values, which shows that even for normally distributed OVL and CD distributions over the entire substrate, the final EPE distribution depends on the nature of the transfer function and the correlation between the two parameters.

[0091] To address these issues, embodiments of the present invention provide a process for fully characterizing a multi-dimensional window and reducing or minimizing the measurements involved.

[0092] exist Figures 7 to 9The basic steps of this embodiment are depicted in . Instead of measuring the EPE distribution directly, the variation of the process parameters (which may also be referred to as input parameters) is measured. Input parameters that are known to be stable or can be controlled and ideally normally distributed are selected to limit the number of measurements involved. If the parameter of interest is not normally distributed (e.g., EUV local CDU), it is desirable to select the basis parameters of the normal distribution to limit the number of measurements. The normal distribution typically involves about 100 to 10,000 measurements to estimate the average parameter and the degree of variation. For example, consider measuring CD using a CD-SEM tool. Although many parameters can improve accuracy, only the repeatability of the average CD and its variation is considered, that is, the local CDU classified by 3-sigma variation. Currently, random behavior typically has a 3-sigma local CD error of about 3nm. In order to obtain a 3-sigma repeatability of 0.1nm in CD, approximately 1000 measurements are required: 3 / sqrt(1000) ~ 0.1nm. The repeatability of local CD is 3 / sqrt(2×1000). The number of measurements can usually be captured in one image, but characterizing the tail behavior will require orders of magnitude more (e.g., 10 9 Measurements to predict 10 -6 failure rate behavior in the range with a reliability of approximately 5-sigma).

[0093] Figure 7 An example of a variation of a process parameter (CD vs. OVL) characterized by measurement is depicted. Configurable variations can also be used, which may include measured or known correlations between parameters, especially if simulation is to be used.

[0094] The transfer function of the yield-limiting construct is measured (e.g., its sensitivity matrix) or simulated. This can be done during a set-get experiment using reference CD and OVL measurements versus nominal parameters. Figure 8 The transfer function between the selected process parameters and the critical yield limiting parameter (e.g., the bottom EPE of the via) is depicted. Different metrology tools can be used for each measurement (e.g., optical measurement tools for OVL (e.g., using UV radiation), scanning electron microscope type tools (SEM) for CD, transmission electron microscope type tools (TEM) for the bottom of the EPE). Fig. 9 is another depiction of the relationship between EPE, overlap, and CD in an exemplary device pattern. It should be noted that the measurements of the transfer functions and the distribution of the input parameters can be performed sequentially or in parallel.

[0095] Individual measurements of the relevant dimensions corresponding to the failure mode make it possible to estimate the distribution of this parameter. In practice, the measurements can be made using a variety of techniques, such as:

[0096] • TEM or X-section SEM, which is usually the default measurement for process tools.

[0097] • High-pressure SEM (eg, HMI NanoScan tool).

[0098] Profilometry (e.g. for measuring the conformity of deposits)

[0099] Optical CD reconstruction, e.g. scatterometry.

[0100] Based on the distribution function and transfer function of the process (input) parameters, a multidimensional performance probability function can be calculated, which indicates the probability of success (or failure) of the process as a function of the nominal values ​​of the process parameters. For example, for two parameters CD and OVL, the performance probability function can be given by:

[0101] PDF EPE ΔEPE=∑ f(CD，OVL)＝EPE PDF CD，OVL ΔCDΔOVL (3) For a given bonding probability function, a variety of numerical methods can be used to estimate the probability density of EPE. For example, a PDF for predicting the value of EPE is EPE The method of (EPE) is to solve OVL, CD = f -1 (EPE-dEPE / 2) and OVL, CD = f -1 The inverse equation for the underlying region between (EPE+dEPE / 2) is then summed over the discrete joint probability functions in this region.

[0102] The performance probability function can be estimated for different nominal parameters, such as expected values ​​of CD, OVL, deposition thickness, etch time, one or more CMP settings, etc. The performance probability function can be derived as a function of more than two process parameters. One advantage is that the number of measurements involved when there are changes will be greatly reduced. For example, if a new tool or control strategy is introduced, only the process parameters need to be revalidated. If a new process is adopted, only the transfer curve needs to be remeasured. In both cases, there is no need to define the complex tail behavior and how the distribution changes throughout the process setup. Embodiments of the present invention can make it possible to analyze the impact of manufacturing tool process changes and the nonlinear transfer process itself separately.

[0103] In addition, various approximation methods allow parameters to be added and severely limit the number of measurements required. For example, if no correlation exists between the existing process parameters and any new parameters (e.g., CMP pressure), the CMP pressure variation can be measured for a nominal set of substrates (through some indirect measurement of CD or simply by recording control residuals on the CMP tool), and then measurements can be performed on one or more deflected substrates where all other parameters are nominal and the CMP pressure is varied. For greater accuracy or if there is a correlation between the parameters, additional substrates can be measured and the results added to the original measurements.

[0104] The nonlinearity of the transfer function and the correlation between parameters can have a substantial impact on the failure rate, which can be revealed by the performance probability function. Figures 10 to 15 Shown in. Fig.10 , 11 and 12 depict the probability density functions of CD and coverage for different amounts of correlation between these two process parameters: Fig.10 depicts the case of a smaller positive correlation, p = +0.3; Fig.11 depicts the case of no correlation, p = 0; Fig.12 A small negative correlation is depicted, p = -0.3. Figure 8 The transfer function shown is Fig.13 , 14 The performance probability functions for different nominal settings are shown in and 15 . Fig.13 The total EPE for different correlation factors is shown, which shows that the difference in failure rate is about 1.5 orders of magnitude.The predicted probability of an individual event having an EPE < 6 nm is indicated by FR ("Failure Rate"). Fig.14 The performance probability function at different average CD values ​​is shown, which shows that the failure rate varies by 2 to 3 orders of magnitude with a CD variation of ±2nm. Fig.13 As shown, the predicted probability of an EPE < 6 nm for an individual event is indicated by the FR (“Failure Rate”). Fig.15 The performance probability functions at different average overlap values ​​are shown, which show that the failure rate varies by 7 orders of magnitude with different values ​​of overlap up to 4 nm.

[0105] A particularly advantageous use of embodiments of the present invention is to derive a process window. For a given expected failure rate, the nominal process parameters can be searched to find an acceptable process window. In other words, the performance probability function provides a multi-dimensional window that gives the values ​​of the relevant process parameters (e.g., CD, OVL, etc.) that meet the criterion that the failure rate is less than the target.

[0106] If necessary, the process window can be calibrated by performing test exposures at the edge of the process window and measuring the obtained failure rate. The measured failure rate is compared with the expected failure rate obtained by calculating the cumulative performance probability function and obtaining the failure rate for the process parameter range for the test exposure. The process parameters of the test exposure can be selected so that the expected failure rate is high enough to enable the failure rate to be estimated with a sufficient confidence interval (e.g., 10 -3 ) to identify these rates, thereby using a relatively small number of tests. Failure rates under off-target conditions can be measured by the above measurements or other methods, such as:

[0107] Electric vehicles

[0108] Bright field inspection

[0109] Voltage comparison check

[0110] This calibration method can be used to verify the derivation of the performance probability function and eliminate any potential inaccuracies in the measurement of measurement-related parameters (e.g., EPE). The error can be a deviation of the measurement tool on an absolute nanometer scale or even a smaller linear correction within a reasonable range. By solving the opposite problem, these yield (or reliability) measurements can also be used to set reasonable specifications for EPE. If the failure rate under a specific EPE distribution behavior is known, the EPE limit can be found by looking up the relevant failure rate value in the cumulative distribution function.

[0111] According to one embodiment, the performance probability function can be used in process control, for example, in high-volume production. The performance probability function characterizes the nonlinear behavior of the process, so by monitoring one or more individual parameters, the process margin can be calculated or one or more parameters can be controlled to stay at or near the center of the process window. For example, if it is known that the processed substrate has a large OVL and a large CD, it can be compensated and the chance of "snap-on" can be reduced or minimized by using a shorter etch time. The performance probability function provides information for determining how much the etch time should be reduced (i.e., the OVL / CD process window as a function of etch time).

[0112] It will be appreciated that embodiments of the present invention may provide a novel method for processing relevant parameters to achieve control of process parameters.

[0113] An advantage of an embodiment of the present invention is that it can be used to reduce or minimize the number of simulations required to determine a process window. For example, the number of measurements used to characterize a process is reduced from 10 for each condition to 10 for each condition. 9 Reduced to 10 per condition 3 (characterizes the measurable failure rate and reduces the 10 required for noise 3High throughput SEM / TEM can screen multiple parameters in a reasonable amount of time which would otherwise be virtually impossible.

[0114] If a known or calibrated process model is available in the simulation software, the sensitivity calculations together with the estimation of the initial parameters can be used in the method described in this article to create a complete model that predicts the probability distribution behavior and failure rates (based on certain geometric rules or external calibration). This allows improving the design of the process stack and improving, for example, the way the OVL / CD process window depends on it.

[0115] The performance probability function can also be used to estimate the electrical performance and the variance therein. For example, if the die has a smaller OVL, then the dose can be adjusted to select a larger CD for all vias in the die, thereby reducing the average resistance of the vias, i.e., improving power consumption, and even adjusting the expected flip-flop timing / performance.

[0116] If certain parameters have been measured for the process, the calibrated performance probability function can be used to define and use additional control “knobs” in the lithography or other processing tool to stay at or near the center of the process window (the large OVL / large CD described above can be compensated by limiting the etch time to create a smaller bottom CD).

[0117] In one embodiment, various options for controlling the lithography process may be presented to the lithography operator in a performance dashboard. Such a dashboard may present the budget or target values ​​for the process (e.g., Table 1 below) and the actual achieved values ​​(e.g., Table 2 below).

[0118] Table 1

[0119]

[0120] Table 2

[0121]

[0122] Ideally, the report includes at least some selected from: optimal targets for one or more or all contributors; measurement or target windows; and / or final process capability (one or more targets). The actual state of the process window or alternative budget ideally includes one or more updated targets resulting from one or more new windows and / or updated process capability states. The dashboard can then allow: optimizing one or more ideal process parameters (e.g., optimal thickness); trad-in one or more individual parameter windows (e.g., etch time / OVL windows); and / or reporting on the process capability of a given toolset.

[0123] Another aspect of the invention is directed to improving the indirect control of lithographic equipment based on multiple sensor readings and interpretation of the readings. Conventional control strategies are directed to minimizing the measured value of the error. For example, an overlap-based control strategy is directed to minimizing the average overlap locally or over a field or substrate. However, the metrology measurement on which such control is based may be subject to systematic error because it is measured on a target that may behave differently from product characteristics when undergoing a pattern transfer process (e.g., etching) or a device manufacturing process (e.g., CMP). Methods for measuring and correcting such errors have been proposed, but there is still a need for improvement.

[0124] According to this aspect, it is proposed to determine a process window function to be used for controlling the lithographic apparatus to improve or optimize the yield rather than to minimize the error measured by the metrology apparatus. Typically, the measured error will vary across the substrate or field. Fig.21 An example of measurement error of overlay error across a substrate is depicted, with arrows indicating the direction and relative amount of overlay measurement error at different points on the substrate. Fig.21 A method for calibrating such an error is further shown. It shows how the center of a process window optimized for one location on a substrate needs to vary depending on the location on the substrate to improve or maximize yield. It also shows the uncertainty of such calibration error. This embodiment proposes a model for determining adjustments to be applied in the control of a lithography process, in particular to the controlled center of a process window, based on an established relationship between measurement error and actual yield performance.

[0125] Using the proposed calibration method for process window adjustment as a function of position on the substrate in, for example, lithographic apparatus control will improve yield. This function uses each substrate measurement to estimate adjustments to one or more control devices (e.g., one or more actuators) to improve yield:

[0126]

[0127] in is a vector of correction basis functions containing overlap (OV) and focus (F). In the above equation, yield represents the calibrated process window function whose center varies with position on the substrate or field. The control method searches for the correction The yield function (in this example) is maximized. Likewise, the method will attempt to jointly optimize different control schemes, such as focus and coverage control.

[0128] Embodiments employ a two-step approach to derive new control methods. First, a per-substrate prediction of one or more device performance indicators (eg, overlay / defocus) is obtained using all available sensor data (including indirect measurements and calculations of actuation adjustments).

[0129] Secondly, sensor readings (e.g., readings from lithographic apparatus sensors, such as substrate alignment mark measurements) are used to generate control functions. The control functions can control various controllable functions of a lithographic apparatus or other tool. For example, displacement errors can be compensated by stage position adjustments during scanning of a field. However, any other available control function can be used. The sensor readings are used together with data-driven training or simulations to distinguish yield loss contributors that may or may not be corrected by the control functions of the lithographic apparatus or other tool. For example, deposited thickness variations cannot be corrected in the lithographic apparatus, but can be corrected in the deposition tool. The results of the training or simulation can be used to correct the measurement signal to avoid crosstalk between errors that are correctable in one device (e.g., the lithographic apparatus) and errors that can be corrected elsewhere (e.g., in the deposition tool).

[0130] The measurement signal can be derived from any sensor that measures the substrate, for example, a level sensor (which obtains a height map of the substrate surface), a CD measurement tool, or other measurement. In one embodiment, a CD measurement is used. Desirably, a combination of CD and overlap is used to allow calculation of tradeoffs between different product printing key performance indicators (KPIs).

[0131] Knowledge of how the sensor measures the sensor target can also help with predictions. In one embodiment, the measurement results are decoded. For example, an alignment sensor can use multiple wavelengths of radiation (color) to measure the position of a single target. Color-to-color differences in alignment measurements made on the same target can provide an indication of asymmetry errors, and therefore suggest measurement errors. This information and reconstruction of the target enables the data model to characterize errors as correctable in the lithographic apparatus, or uncorrectable in some other apparatus. The measurement signal strength can also be varied to clarify errors that cannot be corrected in, for example, the lithographic apparatus.

[0132] Data mining methods can be used to find relationships between other sensor measurements (e.g., humidity or temperature) and yield. This method allows compensation for missing data during calibration of the yield function, since not all substrates or not all parts of a substrate will be sampled equally by metrology measurements, and the sensor readings can be treated as independent continuous measurements. These measurements, together with the data model, can be used as an interpolation of direct sensor readings, enabling better control.

[0133] In one embodiment, sensor measurements as described above are used to create estimates for control adjustments (eg, substrate table displacements) that will depend on the order of substrates in a batch or the position on the substrate.

[0134] Once the relationship between sensors and yield has been established, this function can be used for control or excursion detection. If one of the sensors gives a different reading, the model will help flag this as a problem if it affects yield. Thus, yield losses can be prevented or reduced.

[0135] Models derived according to embodiments of the present invention can be used for different products produced using the same process. Adjustment and offset control settings can be derived for each product.

[0136] Another aspect of the invention is to determine whether a measurement with a large variance (i.e., a value that is far from the mean or at the limit of the measurement range) is correct. Such a measurement may be referred to as an outlier. An outlier may be due to a measurement error on a correct target or a correct measurement on a target with an extreme value of the parameter being measured. It is generally not possible to know which case applies just by measuring. It is common practice to simply disregard the outlier, assuming that the measurement error is more likely than the parameter being measured being an extreme value. However, disregarding a correct measurement in a control loop may be as harmful as including an incorrect measurement.

[0137] In lithography, control errors can have a lasting impact if a static or "run-to-run" control strategy is employed. Such a control strategy uses the results of a series of measurements or lithography parameters, such as overlay, CD, and / or focus, performed on a set of substrates to correct for a future set of substrates. In a run-to-run strategy, measurements are taken at a regular frequency, while in a static strategy, measurements are taken as needed. In both cases, a single measurement can have a large impact - a single measurement may affect tens, hundreds, or even thousands of fields, depending on the correction strategy used.

[0138] Therefore, embodiments of the present invention aim to identify metrology measurements, such as overlapping measurements, which are unlikely to be correct, or which may be correct despite being outliers. In practice, to identify incorrect measurements, an external reference is required; this embodiment uses yields.

[0139] As described above, given the local overlap of a field, the yield probability function (YPF) can be used to predict whether the field will pass or fail. These failed dies can also be identified by a "good die" approach, where dies with overlap greater than a yield calibration threshold are predicted to fail.

[0140] Embodiments of the present invention examine fields that were predicted to fail a yield test but actually passed. In particular, when a field is predicted to fail but actually passes, this is an indicator that the overlap range has been locally estimated to be larger than it actually is. In this approach, a second parameter or combination of parameters is searched for that can provide additional evidence to confirm the inclusion or exclusion of the data point. For example, if a performance indicator such as target signal strength is high, it may indicate that the result should be included because the target performed as expected. The greatest utilization of such a model occurs when parameters are identified that are separate from the original measurement, such as measurements using alignment sensor performance indicators to validate overlap sensor measurements, where one or more candidate second parameters are selected by a search process. In Fig. 22 A simple one-dimensional model is given as an example to illustrate this concept.

[0141] like Fig. 22 As shown, for a given lithography parameter, different field groups can be seen to have different distributions depending on whether they are correctly classified or misclassified. For dies where the lithography parameter magnitude in the figure is less than about 0.4, the probability of the die passing is much greater than the probability of a die failing. Conversely, for dies where the lithography parameter magnitude is greater than 0.5, the probability of observing a die failing is much greater than a die passing. Embodiments of the present invention use methods such as Fig. 22 The distribution shown is used to train a model to assess the confidence of metrology measurements. A simple model makes an initial guess from the overlay model to identify whether a die is considered to pass, then looks at the auxiliary parameters shown (e.g., lithography parameters) and checks whether it is in region 1 or region 2. If it is in region 1, it is likely a correct measurement, while if it is in region 2, it is likely an incorrect measurement. If a measurement is marked as untrustworthy, it should be excluded from control. Likewise, if it is marked as trustworthy, it can be kept in the control loop even if it is an outlier.

[0142] Fig. 22 The setting parameter may be overlay, CD, focus, leveling, or some other parameter from a processing tool or a metrology tool. This includes possible combinations of such parameters, as well as one or more performance KPIs from a metrology tool (e.g., color-to-color variation, stack sensitivity, or intensity) and / or a process tool (e.g., temperature sensor, alignment target substrate quality, or color-to-color variation).

[0143] Embodiments of the present invention may also use a model that takes into account the correlation between one or more lithography parameters and overlay, or the correlation across multiple locations on a substrate. This includes, for example, creating a Bayesian estimate of the true overlay present on a substrate, given an indication of the reliability of available overlay measurements.

[0144] Thus, this aspect of the invention provides a method for using one or more other parameters (eg, one or more lithography parameters) to evaluate measurements including outliers that may be excluded based on, for example, mean 3-sigma filtering.

[0145] This aspect may provide a run-to-run control loop that uses probabilistic decisions for control within a lithography context by making probabilistic inclusion decisions or by making weighted overlap estimates based on the confidence of the estimates.

[0146] Embodiments of this aspect may provide more accurate metrology estimates and may therefore produce higher yields or equal yields using fewer or faster metrology measurements. Further, information about one or more other (e.g., lithography) parameters may be used to determine whether a measurement is likely to produce useful information—e.g., if a measurement is implausible, there is no reason to do it, thereby reducing costs.

[0147] In one embodiment, a method for analyzing a process is provided, the method comprising: obtaining a multidimensional probability density function representing an expected distribution of values ​​of a plurality of process parameters; obtaining a performance function relating the values ​​of the process parameters to a performance metric of the process; and using the performance function to map the probability density function to a performance probability function with the process parameters as independent variables.

[0148] In one embodiment, the process is a lithography process and the process parameters are two or more process parameters selected from the group consisting of: critical dimension, focus, dose, aberration, pattern position, pupil intensity distribution, source bandwidth and etching time. In one embodiment, obtaining a multidimensional probability density function includes measuring the process parameters on a plurality of processed substrates. In one embodiment, obtaining a multidimensional probability density function includes multiple simulations of the execution of the process. In one embodiment, the process parameters are selected based on the distribution of the process parameters being a normal distribution. In one embodiment, the process parameters are at least partially correlated. In one embodiment, the performance function is nonlinear. In one embodiment, the performance metric is selected from one or more of the following: edge position error, yield, output, overlap margin and / or electrical performance. In one embodiment, the method further includes identifying process parameter values ​​for which the performance probability function meets a criterion. In one embodiment, the method further includes selecting a set of nominal process settings for the lithography process that falls within the subspace. In one embodiment, a device manufacturing method is provided, the method comprising: exposing a radiation-sensitive layer on a substrate to form a latent image therein; developing the radiation-sensitive layer to fix the latent image; and transferring the fixed image to the substrate, wherein at least one of the exposing, developing and / or transferring steps is performed using a set of nominal process settings determined by the method as described above.

[0149] In one embodiment, a device manufacturing method is provided, the method comprising: exposing a radiation-sensitive layer on a substrate to form a latent image therein; developing the radiation-sensitive layer to fix the latent image into a developed image; transferring the fixed image onto a substrate to form a device feature or a measurement target; measuring characteristics of at least one of the latent image, the developed image, the device feature and / or the measurement target; and using the above-mentioned performance probability function and the measured characteristics to determine process settings for subsequent exposure, development or transfer steps.

[0150] In one embodiment, a computer program is provided, the computer program comprising instructions which, when executed by one or more lithography tools, instruct the lithography tools to perform the methods described herein.

[0151] Additional embodiments are disclosed in the following numbered item list:

[0152] 1. A method for analyzing a process, the method comprising:

[0153] obtaining a multidimensional probability function representing an expected distribution of values ​​of a plurality of process parameters;

[0154] obtaining a performance function relating the value of the process parameter to a performance measure of the process; and

[0155] The probability density function is mapped to a performance probability function with the process parameter as an independent variable using the performance function.

[0156] 2. A method according to item 1, wherein the multidimensional probability function is a probability density function or a yield probability function.

[0157] 3. A method according to clause 1 or clause 2, wherein the process is a lithography process and the process parameters are selected from the group consisting of: focus, dose, overlap, aberrations, parameters related to the movement of the stage of the lithography equipment, pupil intensity distribution and source bandwidth.

[0158] 4. A method according to any one of clauses 1 to 3, wherein the process comprises transferring a pattern to the substrate and the process parameters are selected from the group consisting of: RF power (per frequency), substrate temperature, (partial) gas pressure in the plasma, composition of the plasma, CMP pressure, CMP polishing routine, etching time and deposition thickness.

[0159] 5. A method according to any one of clauses 1 to 4, wherein obtaining the multidimensional probability function comprises measuring values ​​of the process parameter on a plurality of processed substrates.

[0160] 6. A method according to any one of clauses 1 to 5, wherein obtaining a multidimensional probability function comprises simulating the execution of the process multiple times.

[0161] 7. A method according to any one of items 1 to 6, wherein obtaining a multidimensional probability function comprises training a machine learning algorithm.

[0162] 8. A method according to any one of clauses 1 to 7, wherein the process parameter has a normal distribution.

[0163] 9. A method according to any one of clauses 1 to 8, wherein the process parameters are at least partially correlated.

[0164] 10. A method according to any one of clauses 1 to 9, wherein the performance function is non-linear.

[0165] 11. The method of any of clauses 1 to 10, wherein the performance metric is selected from the group consisting of: edge position error, critical dimension, and critical dimension uniformity.

[0166] 12. A method according to any of clauses 1 to 11, further comprising identifying a subspace of process parameter values ​​for which the performance probability function meets a criterion.

[0167] 13. The method of clause 12, further comprising calibrating the process window by performing test exposures at edges of the subspace and measuring the resulting failure rate.

[0168] 14. The method of clause 12 or clause 13, further comprising selecting a set of nominal process settings for a lithography process that falls within the subspace.

[0169] 15. A method according to any one of clauses 1 to 14, further comprising analysing the performance probability function to determine changes to the performance function and / or the probability density function that will improve yield.

[0170] 16. A device manufacturing method comprising:

[0171] exposing a radiation sensitive layer on a substrate to form a latent image therein;

[0172] developing the radiation sensitive layer to fix the latent image; and

[0173] transferring the fixed image onto the substrate,

[0174] Wherein at least one selected from the exposing, developing and / or transferring steps is performed using a set of nominal process settings determined by the method according to clause 14.

[0175] 17. A device manufacturing method comprising:

[0176] exposing a radiation sensitive layer on a substrate to form a latent image therein;

[0177] developing the radiation-sensitive layer to fix the latent image into a developed image;

[0178] transferring the fixed image onto the substrate to form a device feature or a measurement target;

[0179] measuring a characteristic of at least one selected from the latent image, the developed image, the device feature, and / or the measurement target; and

[0180] Using the performance probability function determined according to any of clauses 1 to 15 and the measured characteristic, process settings for a subsequent exposure, development or transfer step are determined.

[0181] 18. A method of evaluating a field of a substrate that has been patterned using a photolithographic process, the method comprising:

[0182] measuring a first lithography parameter of the field to obtain a measurement value;

[0183] using said measured value of said first lithographic parameter to obtain an estimate of said probability that said field has been correctly patterned; and

[0184] A confidence level of the estimate is determined based on a second lithography parameter different from the first lithography parameter.

[0185] 19. The method of clause 18, further comprising controlling a lithographic apparatus, a process step or a process tool based on the measurement value only in response to the estimate being determined to be credible.

[0186] 20. The method of clause 18, further comprising controlling a lithographic apparatus, a process step, or a process tool based on a weighted value of the measurement value only in response to the estimate being determined to be credible.

[0187] 21. A computer program comprising instructions which, when executed by one or more computers and / or one or more processing tools, direct the one or more computers and / or one or more processing tools to perform a method according to any one of clauses 1 to 20.

[0188] 22. A method for analyzing a process, the method comprising: obtaining a multidimensional probability density function representing an expected distribution of values ​​of a plurality of process parameters; obtaining a performance function relating the values ​​of the process parameters to a performance metric of the process; and using the performance function to map the multidimensional probability density function to a performance probability function with the process parameters as independent variables.

[0189] 23. The method of clause 22, further comprising identifying one or more ranges of the process parameters for which the performance probability function meets a criterion.

[0190] 24. The method according to item 23 also includes calibrating a process window associated with the process parameter by performing exposure at one or more edges of the one or more ranges of the process parameter and measuring a failure rate associated with a feature formed as a result of the performed exposure.

[0191] 25. The method of clause 23, further comprising selecting process settings for a lithography process based on the identified one or more ranges of the process parameters.

[0192] 26. A method according to clause 22, wherein the step of obtaining a multidimensional probability density function is preceded by the step of selecting the process parameters based on characteristics of an expected statistical distribution of values ​​of the process parameters.

[0193] 27. A method according to clause 26, wherein the characteristic is the expected similarity to a normal distribution.

[0194] Although specific reference may be made herein to the use of metrology equipment in IC manufacturing, it should be understood that the equipment described herein may have other applications. Possible other applications include the manufacture of integrated optical systems, guidance and detection patterns for magnetic domain memories, flat panel displays, liquid crystal displays (LCDs), thin film magnetic heads, etc.

[0195] Although specific reference is made herein to "measurement equipment" or "inspection equipment," both terms may also refer to inspection equipment or inspection systems. For example, an inspection or measurement equipment including embodiments of the present invention may be used to determine characteristics of structures on a substrate (e.g., a wafer). For example, an inspection or measurement equipment including embodiments of the present invention may be used to detect defects in a substrate or defects in a structure on a substrate. In such embodiments, the characteristic of interest of a structure on a substrate may be related to a defect in the structure, the absence of a particular portion of the structure, or the presence of an unwanted structure on the substrate.

[0196] Although embodiments of the invention may be specifically referenced herein in the context of a metrology device, embodiments of the invention may be used in other devices. Embodiments of the invention may form part of a patterning device (e.g., mask) inspection device, a lithography device, or any device that measures or processes an object such as a wafer (or other substrate) or a mask (or other patterning device). These devices may generally be referred to as lithography tools. Such lithography tools may use vacuum conditions or ambient (non-vacuum) conditions.

[0197] Although specific reference may have been made above to the use of embodiments of the present invention in the context of optical lithography, it will be appreciated that embodiments of the present invention are not limited to optical lithography and may be used in other applications such as imprint lithography where the context permits.

[0198] To the extent that certain U.S. patents, U.S. patent applications, or other materials (e.g., articles) are incorporated by reference, the text of such U.S. patents, U.S. patent applications, and other materials is incorporated by reference only to the extent that there is no conflict between such materials and the statements and drawings set forth herein. In the event of such a conflict, any such conflicting text in such incorporated by reference U.S. patents, U.S. patent applications, and other materials is not expressly incorporated by reference herein.

[0199] In the block diagram, the components shown are depicted as discrete functional blocks, but the embodiments are not limited to systems in which the functions described herein are organized as shown. The functions provided by each component may be provided by software or hardware modules organized differently than currently depicted, for example, such software or hardware may be mixed, combined, copied, decomposed, distributed (e.g., within a data center or geographically), or otherwise organized differently. The functions described herein may be provided by one or more processors of one or more computers executing code stored on a tangible, non-transitory machine-readable medium. In some cases, a third-party content delivery network may host some or all of the information transmitted over the network, in which case, to the extent that information (e.g., content) is referred to as being supplied or otherwise provided, the information may be provided by sending instructions for retrieving the information from the content delivery network.

[0200] Although specific embodiments of the present invention have been described above, it should be understood that the present invention may be practiced in a manner other than that described. The above description is intended to be illustrative rather than restrictive. Therefore, it will be clear to those skilled in the art that modifications may be made to the described invention without departing from the scope of the claims set forth below.

Claims

1. A method for analyzing a process, the method comprising: include: obtaining a multidimensional probability density function representing an expected distribution of values ​​of a plurality of process parameters of a manufacturing process; obtaining a performance function relating a value of the process parameter to a performance measure of the process; as well as The multidimensional probability density function is mapped to a performance probability function with the process parameters as independent variables using the performance function.

2. The method of claim 1, wherein the process is a lithography process and the process parameters are selected from the group consisting of: focus, dose, overlay, aberrations, parameters related to movement of a stage of a lithography apparatus, pupil intensity distribution, and source bandwidth.

3. The method of claim 1 , wherein the process comprises transferring a pattern to a substrate and the process parameters are selected from the group consisting of: RF power (per frequency), substrate temperature, (partial) gas pressure in the plasma, composition of the plasma, CMP pressure, CMP polishing routine, etching time and deposition thickness.

4. The method according to claim 1, wherein a multidimensional probability density function is obtained include: The values ​​of the process parameters are measured on a plurality of processed substrates, or the execution of the process is simulated a plurality of times.

5. The method of claim 1, wherein obtaining a multidimensional probability density function comprises training a machine learning algorithm. The method according to claim 1 , wherein the process parameters have a normal distribution and / or are at least partially correlated. The method of claim 1 , wherein the performance function is non-linear.

8. The method of claim 1, wherein the performance metric is selected from the group consisting of: edge location error, critical dimension, and critical dimension uniformity.

9. The method of claim 1, further comprising identifying one or more ranges of the process parameters for which the performance probability function meets a criterion.

10. The method of claim 9, further comprising calibrating a process window associated with the process parameter by performing exposures at one or more edges of the one or more ranges of the process parameter and measuring failure rates associated with one or more features formed as a result of the exposures. 11 . The method of claim 9 , further comprising selecting a process setting based on the identified one or more ranges of the process parameters.

12. The method of claim 1, further comprising analyzing the performance probability function to determine changes to the performance function and / or the multidimensional probability density function that will increase yield.

13. The method of claim 1, wherein the step of obtaining a multidimensional probability density function is preceded by the step of selecting the process parameter based on characteristics of an expected statistical distribution of values ​​of the process parameter. The method of claim 13 , wherein the characteristic is an expected similarity to a normal distribution.

15. A non-transitory machine-readable medium having stored thereon a computer program configured to analyze a process, the computer program comprising instructions that when executed by one or more computers perform the following steps: obtaining a multidimensional probability function representing an expected distribution of values ​​of a plurality of process parameters; obtaining a performance function relating a value of the process parameter to a performance measure of the process; as well as The multidimensional probability density function is mapped to a performance probability function with the process parameters as independent variables using the performance function.

16. The non-transitory machine-readable medium of claim 15, wherein the instructions for obtaining a multi-dimensional probability density function include instructions for measuring values ​​of the process parameters on a plurality of processed substrates or simulating the performance of the process a plurality of times.

17. The non-transitory machine-readable medium of claim 15, further comprising instructions for identifying one or more ranges of the process parameters for which the performance probability function meets a criterion.

18. The non-transitory machine readable medium of claim 17, further comprising instructions for selecting a process setting based on one or more identified ranges of the process parameters.

19. The non-transitory machine-readable medium of claim 15, further comprising instructions configured to analyze the performance probability function to determine changes to the performance function and / or the multidimensional probability density function that will improve yield.

20. The non-transitory machine-readable medium of claim 15, further comprising instructions configured to, prior to the step of obtaining a multi-dimensional probability density function, perform the step of selecting the process parameter based on characteristics of an expected statistical distribution of values ​​of the process parameter.

Images