Optical and X-ray metrology methods for patterned semiconductor structures with randomness
Methods and systems for determining random variation in one or more structures on a specimen are provided. One method includes determining characteristic(s) of output generated by an output acquisition subsystem for structure(s) formed on a specimen and simulating the characteristic(s) of the output with initial parameter values for the structure(s). The method also includes determining parameter values of the structure(s) formed on the specimen as the initial parameter values that resulted in the simulated characteristic(s) that best match the determined characteristic(s). The determined parameter values are responsive to random variation in parameter(s) of the structure(s) on the specimen.
1 . A system configured for determining random variation in one or more structures formed on a specimen, comprising:
an output acquisition subsystem configured for generating output for one or more structures formed on a specimen, wherein the output acquisition subsystem comprises at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to the specimen by the output acquisition subsystem, and wherein the detector is configured to detect energy from the specimen and to generate the output responsive to the detected energy; and
a computer subsystem configured for:
determining one or more characteristics of the output generated for the one or more structures;
simulating the one or more characteristics of the output with parameter values for the one or more structures;
determining parameter values of the one or more structures formed on the specimen as the parameter values that resulted in the simulated one or more characteristics that best match the determined one or more characteristics, wherein the determined parameter values are responsive to random variation in one or more parameters of the one or more structures on the specimen, and wherein the random variation comprises randomness in geometric parameters, material parameters, or geometric and material parameters of the one or more structures on the specimen; and
altering a process that was or will be performed on the specimen based on the determined parameter values.
2 . The system of claim 1 , wherein the one or more characteristics of the output are responsive to the random variation in the one or more parameters of the one or more structures as formed on the specimen according to an arbitrary statistical distribution of random degrees of freedom parameterized by a user a priori and sampled by a quasirandom or pseudorandom distribution defining one supercell structure larger than the one or more structures or an ensemble thereof.
3 . The system of claim 1 , wherein the computer subsystem is further configured for determining an arbitrary statistical distribution of aperiodic degrees of freedom comprising arbitrary characteristics, wherein the arbitrary characteristics comprise one or more of correlations and non-Gaussian distributions, by sampling the arbitrary statistical distribution by a quadrature defined by quasirandom or pseudorandom numbers.
4 . The system of claim 1 , wherein the random variations in the one or more parameters are parameterized by one number per each of the one or more parameters of the one or more structures, and wherein the one number is an arbitrary single parameter.
5 . The system of claim 4 , wherein a probability distribution that describes the random variation is a Gaussian distribution, dual-Gaussian distribution, uniform distribution, skewed Gaussian distribution, or Poisson distribution.
6 . The system of claim 4 , wherein a probability distribution that describes the random variation comprises a correlation predetermined by a user a priori such that one value of the arbitrary single parameter for each geometric critical dimension parameter of the one or more parameters is sufficient for describing a correlated multi-parameter distribution.
7 . The system of claim 1 , wherein the simulating is performed with a super-cell model having an arbitrary number of unit cells in x and y directions, and wherein x and y are nominal periodic directions of the structure.
8 . The system of claim 7 , wherein the simulating is further performed with an ensemble of independent instances of the super-cell model, and wherein the simulating comprises averaging results of the simulating for the ensemble according to a weighted average formula.
9 . The system of claim 8 , wherein values of the arbitrary number of the unit cells in the x and y directions and a size of the ensemble are automated to reach a predetermined accuracy, a predetermined speed, a predetermined robustness, or a combination thereof of the results of the simulating.
10 . The system of claim 8 , wherein the super-cell model or the ensemble is designed to describe long-range correlations in the random variation in the one or more parameters of the one or more structures formed on the specimen.
11 . The system of claim 1 , wherein said simulating comprises simulating the output with a forward model configured for performing a regression technique.
12 . The system of claim 1 , wherein the computer subsystem is further configured for identifying a first of the one or more characteristics of the output that is more responsive to at least one of the parameter values than a second of the one or more characteristics of the output, and wherein determining the one or more characteristics, simulating the one or more characteristics, and determining the parameter values are performed with only the first of the one or more characteristics.
13 . The system of claim 1 , further comprising one or more components executed by the computer subsystem, wherein the one or more components comprise an electromagnetic solver configured for performing said simulating.
14 . The system of claim 1 , further comprising one or more components executed by the computer subsystem, wherein the one or more components comprise a model-based machine learning model configured for fitting the one or more characteristics of the output as a function of the parameter values with which said simulating is performed.
15 . The system of claim 1 , wherein said simulating is performed with a machine learning model trained with synthetic and real spectra, and wherein the real spectra are collected with a sampling selected so that a statistical distribution describing the random variation is determined per die on a training specimen.
16 . The system of claim 1 , wherein the computer subsystem is further configured for determining a probability distribution that describes the random variation by collecting electrical testing results from multiple devices within one die on an additional specimen and generating an electrical testing Gaussian distribution from the electrical testing results.
17 . The system of claim 1 , wherein the output is further responsive to x-rays from the specimen, and wherein the one or more characteristics of the output comprise diffuse scattering and diffraction order intensities.
18 . The system of claim 17 , wherein said simulating comprises calculating diffuse and specular scattering responsive to the random variation by averaging results of electromagnetic simulations over an ensemble of supercell profiles, determining a diffuse scattering from the averaged results, and determining a diffuse scattering detector signal by interpolation of the determined diffuse scattering.
19 . The system of claim 18 , wherein said simulating further comprises determining a diffraction detector signal from the averaged results and determining a full detector signal by combining the diffuse scattering detector signal and the diffraction detector signal.
20 . The system of claim 17 , wherein determining the one or more characteristics comprises removing diffraction orders from the output to thereby extract the output responsive to only the diffuse scattering in the output.
21 . The system of claim 17 , wherein the output comprises a full x-ray signal, and wherein determining the one or more characteristics comprises matching the full x-ray signal with a beam shape describing both the diffraction order intensities and the diffuse scattering.
22 . The system of claim 17 , wherein the computer subsystem is further configured for determining additional parameter values of the one or more structures from additional output generated by the output acquisition subsystem for the one or more structures, wherein the additional parameter values comprise values of one or more of the geometric parameters of the one or more structures that are not responsive to the random variation in the one or more structures, and wherein the output acquisition subsystem further comprises a mask blocking one or more regions of the detector between diffraction orders in the x-rays from the specimen during generation of the additional output.
23 . The system of claim 1 , wherein the output generated by the output acquisition subsystem is further responsive to light from the specimen, and wherein the one or more characteristics comprise depolarization in the light.
24 . The system of claim 23 , wherein said simulating comprises simulating the output generated for the one or more structures with the parameter values, wherein the simulated output is a Mueller Matrix as a function of wavelength, and wherein said simulating further comprises calculating an averaged Mueller Matrix function from the Mueller Matrix and calculating the depolarization in the light from the averaged Mueller Matrix function.
25 . The system of claim 24 , wherein said simulating is performed by regression with a cost function based on a weighted average of depolarization cost and Mueller Matrix cost.
26 . The system of claim 23 , wherein said simulating is performed by regression with a cost function based on only depolarization cost.
27 . A non-transitory computer-readable medium, storing program instructions executable on a computer system for performing a computer-implemented method for determining random variation in one or more structures on a specimen, wherein the computer-implemented method comprises:
determining one or more characteristics of output generated by an output acquisition subsystem for one or more structures formed on a specimen, wherein the output acquisition subsystem comprises at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to the specimen by the output acquisition subsystem, and wherein the detector is configured to detect energy from the specimen and to generate the output responsive to the detected energy;
simulating the one or more characteristics of the output with parameter values for the one or more structures;
determining parameter values of the one or more structures formed on the specimen as the parameter values that resulted in the simulated one or more characteristics that best match the determined one or more characteristics, wherein the determined parameter values are responsive to random variation in one or more parameters of the one or more structures on the specimen, and wherein the random variation comprises randomness in geometric parameters, material parameters, or geometric and material parameters of the one or more structures on the specimen; and
altering a process that was or will be performed on the specimen based on the determined parameter values.
28 . A computer-implemented method for determining random variation in one or more structures on a specimen, comprising:
determining one or more characteristics of output generated by an output acquisition subsystem for one or more structures formed on a specimen, wherein the output acquisition subsystem comprises at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to the specimen by the output acquisition subsystem, and wherein the detector is configured to detect energy from the specimen and to generate the output responsive to the detected energy;
simulating the one or more characteristics of the output with parameter values for the one or more structures;
determining parameter values of the one or more structures formed on the specimen as the parameter values that resulted in the simulated one or more characteristics that best match the determined one or more characteristics, wherein the determined parameter values are responsive to random variation in one or more parameters of the one or more structures on the specimen, and wherein the random variation comprises randomness in geometric parameters, material parameters, or geometric and material parameters of the one or more structures on the specimen; and
altering a process that was or will be performed on the specimen based on the determined parameter values, wherein said determining the one or more characteristics, simulating the one or more characteristics, determining the parameter values, and altering the process are performed by a computer subsystem coupled to the output acquisition subsystem.
29 . A system configured for determining random variation in one or more structures formed on a specimen, comprising:
an output acquisition subsystem configured for generating output for one or more structures formed on a specimen, wherein the output acquisition subsystem comprises at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to the specimen by the output acquisition subsystem, and wherein the detector is configured to detect energy from the specimen and to generate the output responsive to the detected energy; and
one or more components executable on a computer subsystem coupled to the output acquisition subsystem, wherein the one or more components comprise a machine learning model configured for determining random variation in one or more parameters of the one or more structures formed on the specimen based on the generated output, wherein the random variation comprises randomness in geometric parameters, material parameters, or geometric and material parameters of the one or more structures on the specimen, and wherein the computer subsystem is configured for altering a process that was or will be performed on the specimen based on the determined random variation.
30 . The system of claim 29 , wherein the machine learning model is a model-based machine learning model.
31 . The system of claim 29 , wherein the machine learning model is a model-free machine learning model.
32 . The system of claim 29 , wherein the machine learning model is trained with synthetic and real spectra, and wherein the real spectra are collected with a sampling selected so that a statistical distribution describing the random variation is determined per die on a training specimen.
33 . The system of claim 29 , wherein the computer subsystem is further configured for determining a probability distribution that describes the random variation by collecting electrical testing results from multiple devices within one die on an additional specimen and generating an electrical testing Gaussian distribution from the electrical testing results.
34 . The system of claim 29 , wherein the output is responsive to x-rays from the specimen, wherein the machine learning model is further configured for determining one or more characteristics of the output generated for the one or more structures, and wherein the one or more characteristics of the output comprise diffuse scattering and diffraction order intensities.
35 . The system of claim 34 , wherein determining the one or more characteristics comprises removing diffraction orders from the output to thereby extract the output responsive to only the diffuse scattering in the output.
36 . The system of claim 34 , wherein the machine learning model is further configured for determining additional parameter values of the one or more structures from additional output generated by the output acquisition subsystem for the one or more structures, wherein the additional parameter values comprise values of one or more of the geometric parameters of the one or more structures that are not responsive to the random variation in the one or more structures, and wherein the output acquisition subsystem comprises a mask blocking one or more regions of a detector between diffraction orders in the x-rays from the specimen during generation of the additional output.
37 . The system of claim 29 , wherein the output generated by the output acquisition subsystem is further responsive to light from the specimen, wherein the machine learning model is further configured for determining one or more characteristics of the output generated for the one or more structures, and wherein the one or more characteristics comprise depolarization in the light.
38 . A computer-implemented method for determining random variation in one or more structures formed on a specimen, comprising:
generating output for one or more structures formed on a specimen with an output acquisition subsystem, wherein the output acquisition subsystem comprises at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to the specimen by the output acquisition subsystem, and wherein the detector is configured to detect energy from the specimen and to generate the output responsive to the detected energy;
determining random variation in one or more parameters of the one or more structures formed on the specimen based on the generated output with a machine learning model, wherein the machine learning model is included in one or more components executable on a computer subsystem coupled to the output acquisition subsystem, and wherein the random variation comprises randomness in geometric parameters, material parameters, or geometric and material parameters of the one or more structures on the specimen; and
altering a process that was or will be performed on the specimen based on the determined random variation, wherein said altering is performed by the computer subsystem.
39 . A non-transitory computer-readable medium, storing program instructions executable on a computer system for performing a computer-implemented method for determining random variation in one or more structures on a specimen, wherein the computer-implemented method comprises:
generating output for one or more structures formed on a specimen with an output acquisition subsystem, wherein the output acquisition subsystem comprises at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to the specimen by the output acquisition subsystem, and wherein the detector is configured to detect energy from the specimen and to generate the output responsive to the detected energy;
determining random variation in one or more parameters of the one or more structures formed on the specimen based on the generated output with a machine learning model, wherein the machine learning model is included in one or more components executable on the computer system coupled to the output acquisition subsystem, and wherein the random variation comprises randomness in geometric parameters, material parameters, or geometric and material parameters of the one or more structures on the specimen; and
altering a process that was or will be performed on the specimen based on the determined random variation.