As semiconductor device scaling continues, the critical dimension CD has continued to decrease. The required CD for advanced device nodes is now beyond the resolution limit of deep ultraviolet (DUV) lithography and extreme ultraviolet lithography (EUVL) is now widely used to meet resolution requirements. Despite continuous improvements in the performance of photoresists, masks, and post-lithography processes, stochastic defects, or stochastic failures (5) - space bridges and line breaks - are still a major factor of yield loss in production(14). Detecting, characterizing, and repairing stochastic defectivity using experimental methods alone is expensive, requiring a lot of wafers, metrology resources and time (9, 15). We discuss a cost-effective predictive failure probability model - a virtual inspector - and demonstrate how it is an essential tool for the study and reduction of stochastic defectivity in EUVL. We use rigorous probabilistic lithography modeling to construct accurate failure probability models FPMs for EUVL. The computing rate of the FPM admits virtual inspection of full layouts. We confirm that the defectivity predictions of the FPM align well with large-area e-beam inspection results. We demonstrate that the FPM can be used to find the optimal patterning condition and the optical proximity correction OPC that minimize stochastic failures in an EUVL process.

Background: Natural physical phenomena occurring at length scales of a few nm produces variation in many aspects of the EUV photoresist relief image: edge roughness, width roughness, feature-tofeature variability, etc. 1,2,3,4. But the most damaging of these variations are stochastic or probabilistic printing failures 5, 6. Stochastic or probabilistic failures are highly random with respect to count and location and occur on wafers at spectra of unknown frequencies. Examples of these are space bridging, line breaking, missing and merging holes. Each has potential to damage or destroy the device, reducing yield 6, 10. Each has potential to damage or destroy the device, reducing yield 6, 10. The phenomena likely originates during exposure where quantized light and matter interact1 . EUV lithography is especially problematic since the uncertainty of energy absorbed by a volume of resist is much greater at 13.5 nm vs. 248 nm and 193 nm. Methods: In this paper, we use highly accelerated rigorous 3D probabilistic computational lithography and inspection to scan an entire EUV advanced node layout, predicting the location, type and probability of stochastic printing failures.

Optical lithography is rapidly becoming very complex. As the limits of resolution are pushed to achieve feature sizes on the order of the wavelength of light or smaller, many phenomena must be understood. Technology innovation to extend optical lithography is also introducing many implementation options that must be assessed. Together, the increased concern for physical effects and the introduction of innovations have greatly increased the number of parameters whose effects must be characterized and balanced. Modeling offers a solid foundation for efficient characterization and a way to systematically quantify relationships and quickly investigate new innovations. The ultimate test is, of course, producing the desired features on product wafers. Yet a little time spent in understanding the models or in making a few simulation runs at a computer terminal can make working in the fabrication facility much more effective. It is also true that observational feedback from the fabrication facility can make modeling and simulation much more effective. This chapter is designed to provide information about modeling and simulation at four distinct levels. It begins with an overview of the phases and nature of modeling and simulation. Then the underlying basic physical models and phenomena of optical imaging, substrate interactions, and resist dissolution are considered. The usefulness of modeling and simulation in concert with conventional characterization methods for determining the practical performance of lithography is then illustrated. Uses of modeling and simulation in assessing technology innovation in materials, exposure tools, masks, etc., are considered. Finally, a summary of available simulators is provided.

Stochastic effects are the ultimate limiters of optical lithography and are a major concern for next-generation technology nodes. In previous work, we compared the performance of several types of EUV resists on dense patterns or brightfield mask SRAM cells across technology nodes. It was shown that due to low photon/chemical shot noise and reduced blur, metal-oxide resists could potentially reduce lithography failures at the 5nm technology node though even at 7nm technology node failures may be noticeable if process variations are considered. Following up on work published the last three years, in this paper we study how different OPC strategies and photoresist properties might affect failure rates for a darkfield mask SRAM cell at the 5nm technology node. Four cases are considered: Aerial image optimization by mask biasing; stochastic simulations are performed with an organic chemically amplified resist model. Aerial image optimization by model-based OPC; stochastic simulations are performed with an organic chemically amplified and a metal-oxide resist model. Aerial image model-based OPC enhanced by rigorous stochastic modeling; stochastic simulations are performed with an organic chemically amplified resist model. In all cases, a numerical aperture of 0.33 is used. Process windows are generated averaging similar to 2150 (3.5s) stochastic simulations for each focus-dose combination, while best focus-dose target CDs are found by analyzing failure rates across focus and dose. Roughly 1.8 million (5 sigma) trials are then run at best condition for all cases to quantify part per million failures.

Stochastics effects are the ultimate limiter of optical lithography technology and are a major concern for next-generation technology nodes in EUV lithography. Following up on work published last year, we compare the performance of organic chemically-amplified and condensed metal-oxide resists exposed at different sizing doses using a proxy 2D SRAM layout. For each combination of material, technology node, and lithographic approach, we perform 550,000 physics based Monte-Carlo simulations of the SRAM cell. We look at many performance data, including stochastic process variation bands at fixed, nominal conditions assuming no variation in process parameters vs. the stochastic process variation bands obtained by inclusion of process parameters. Perturbations are applied to exposure dose, focus, chief-ray azimuthal angle, mask CD, stack thicknesses, and PEB temperature. We study stochastic responses for three technology nodes: • An SRAM cell for 7 nm technology node, with Numerical Aperture = 0.33 and patterned with organic chemically amplified resist • An SRAM cell for 5 nm technology node, with Numerical Aperture = 0.33 and patterned with: o Organic chemically amplified resist o Fast photospeed organic chemically amplified resist o Metal-oxide resist • An SRAM cell for 3 nm technology node, patterned with organic chemically amplified resist and: o Numerical Aperture = 0.33 in single exposure o Numerical Aperture = 0.33 with double exposure o Numerical Aperture = 0.55 with anamorphic pupil For each case, we optimize mask bias, source illumination and process conditions across focus to maximize the optical contrast. We did not apply optical proximity correction to the mask. The purpose of the work is to evaluate the stochastic behavior of different features as a function of material strategy, technology node, and lithographic approach.

Edge position variation in EUV patterns is significantly affected by stochastic phenomena that occur during the EUV exposure and the chemical processes in photoresist. Hence, it is important to understand and quantify the contribution of each of the stochastic effects to the edge roughness. In this work, various computational approaches are used based on the rigorous stochastic resist model in order to assess the stochastic contribution of photon absorption and random chemical reactions in EUV photoresist. The simulation results are presented for both the traditional chemically amplified EUV resists and resists utilizing alternative mechanisms of image formation, such as metal based-resists.

We evaluate through simulations and experimental data the impact of process non-idealities with a particular attention to mask CD uniformity for 44 nm pitch DRAM contact hole array. Several millions of contact holes are simulated with PROLITH after full-physical stochastic process calibration. Process Windows, LCDU and failure rates are compared at nominal conditions, assuming no variation in process parameters vs. the stochastic process variation obtained by inclusion of perturbations of process parameters. The simulations are repeated including Gaussian distributed mask CD variations. Skewness, kurtosis, and failure rates are calculated..

Correlation length has nowadays become of common use in lithographic applications. Together with the Hurst parameter, the power spectral density curve and the PSD(0), the correlation length ξ enables a full comprehension of the roughness along the edges of the features after lithography and after process (i.e. etch). The correlation length can be intuitively defined by how much different points along the same feature’s edge know each other’s position. Behind this simple definition, ξ wraps multiple physical properties and parameters of the whole lithographic process: 1. Source or exposure dose, which set the lowest roughness values – like the PSD(0) – in the lower frequency ranges, before the optical frequency cut-off and resist reaction-diffusion mechanisms. This source of noise is also called incident Photon Shot Noise, and it is a white-type of noise 2. Mask roughness, composed by absorber roughness (i.e. mask line edge roughness) and mask surface roughness which can form speckle patterns 3. Optical system and illumination which fix the minimum printable pitch, but also the maximum roughness frequency transmittable by the exposure tool 4. Photoresist, mainly split in three components: a. Extinction coefficient k, which determine the minimum absorbed PSN which affects low-frequency roughness b. Physical/chemical reaction-diffusion mechanisms such as electron blur and yield (for EUV lithography), and acid-quencher motion in the mid-high frequency range c. Development dynamics, which can change correlation length accordingly to dissolution properties and development time, forming self-affine structures along the feature’s edges 5. Metrology, which affects the whole spectrum, and can lead to non-negligible roughness bias In this work, we study how the correlation length ξ, and more in general the power spectral density curve, changes considering variations of process conditions for both experimental and simulated features. Controlled process perturbations are applied to all the elements composing the lithographic step: source, mask, optical system, and photoresists. Experiments are carried out at imec, simulations are run with PROLITH, and metrology is performed with Fractilia MetroLER. The purpose of this study is to better understand which information can be extrapolated by a thorough roughness analysis in the frequency domain, and how these can be used to limit the variability and failure rates of the printed features.

BACKGROUND: The ionizing wavelength in extreme ultraviolet (EUV) resist exposure leads to photoelectron scattering and uncertainty in the resulting acid image, producing line-edge roughness (LER) and poor CD uniformity of the printed features. GOALS: Try to determine how photoelectron and acid exposure blur effects affect EUV lithography and how they might be better controlled. Try to determine whether or not, and if so under what conditions, high resist quantum yields are beneficial to EUV lithography. METHODS: Using a stochastic resist simulator, we study the effects of resist properties upon photoelectric scattering, the uncertainty in the acid release and the properties of the after-development photoresist image in high NA EUV lithography. Uncertainty in the release of acids is the fundamental cause of LER and the ultimate limiter of optical lithography technology.

Minimization and control of line-edge roughness (LER) and contact-edge roughness (CER) is one of the current challenges limiting EUV line-space and contact hole printability. One significant contributor to feature roughness and CD variability in EUV is photon shot noise (PSN); others are the physical and chemical processes in photoresists, known as resist stochastic effect. Different approaches are available to mitigate each of these contributions. In order to facilitate this mitigation, it is important to assess the magnitude of each of these contributions separately from others. In this paper, we present and test a computational approach based on the concept of an 'ideal resist'. An ideal resist is assumed to be devoid of all resist stochastic effects. Hence, such an ideal resist can only be simulated as an 'ideal resist model' (IRM) through explicit utilization of the Poisson statistics of PSN2 or direct Monte Carlo simulation of photon absorption in resist. LER estimated using IRM, thus quantifies the exclusive contribution of PSN to LER. The result of the simulation study done using IRM indicates higher magnitude of contribution (60%) from PSN to LER with respect to total or final LER for a sufficiently optimized high dose 'state of the art' EUV chemically amplified resist (CAR) model.

The optimization problem of reducing EUV line edge roughness (LER) of a given feature, subject to the tolerance constraints on a CD of this feature at nominal EUV process conditions and several off-nominal conditions, is formulated. A stochastic rigorous Monte-Carlo EUV resist model is employed to solve this stochastic optimization problem. Several options for optimization algorithms, suitable for the solution of the formulated EUV LER optimization problem, are presented and discussed, along with the results of their tests.

Alternative photoresist platforms are being developed with the goal of meeting Resolution, Roughness and Sensitivity requirements for EUV lithography. Metal-based materials appear promising due to the high etch resistance, high absorption, and high resolution. However, the exposure mechanism of these materials is quite different from that of organic chemically amplified resists. The current electron-scattering model built into PROLITHTM X6.0 allows a direct comparison of the exposure mechanisms for different resist platforms: in particular, it is now possible to estimate the intrinsic resist uncertainty by evaluating electron, acid shot noise and spatial blurring, while forcing the photon shot noise contribution to zero. A comparison between organic resists and metal-based platforms reveals how the denser nature of the latter help containing the electron scattering in a much closer radius around the absorption event. The consequent electron-reaction (acid generation for photo-active-generator-containing organic materials, ligand dissociation for the metal-oxides) reflects the electron shot noise of the different platforms. The higher absorption combined with lower blur of the metaloxide materials seem to become of crucial importance for the 5 nm technology node and beyond.

Metal-based photoresists are appealing for use in EUV lithography due to their improved etch resistance and absorption compared with organic resists, and due to their resolving power demonstrated with 13.53 nm exposures using synchrotron light. Recently imec has started a new project to study novel photoresists for EUV lithography, with particular attention to metal containing materials, in order to explore alternative approaches that may offer superior characteristics in photoresist imaging and etching performance compared with more mature chemically amplified resists. In order to model these novel resists it is mandatory to understand both the optical properties and the electronic response to photon absorption. As in previous experiments on organic materials, some of the optical properties can be determined by merging analysis from high-energy electron scattering models (e.g. CXRO website), X-ray absorption spectroscopy, and DUV spectroscopic ellipsometry. Dispersion curves can be used to calculate the electronic inelastic and elastic mean-free paths; convolved with the expected spectrum at wafer level it is possible to estimate the electron yield and the secondary electron blur of the photoresist. These material properties can be used to modify the physical models currently used to simulate organic photoresist performance in computational lithography software.

We present a framework to analyze the performance of optical imaging in a hyper numerical aperture (NA) immersion lithography scanner. We investigate the method to quantify imaging performance by computing upperand lower-bounds on the threshold normalized image log-slope (NILS) and the depth of focus (DOF) in conjunction with the traditional image quality metrics such as the mask error enhancement factor (MEEF) and the linearity for various different pitches and line to space (LS) duty cycles. The effects of the interaction between the light illumination and the feature size are extensively characterized based on the aerial image (AI) behavior in particular for the extreme dipole illumination that is one of the commonly used off-axis illuminations for sub-100nm logic and memory devices, providing resolution near the physical limit of an optical single patterning step. The proposed aerial imaging-based DOF bounds are compared to the results obtained from an experimentally calibrated resist model, and we observed good agreement. In general, the extreme dipole illumination is only optimal for a single particular pitch, therefore understanding the through-pitch imaging performance bound, which depends on the illumination shape, pattern size, and process conditions, is critically important. We find that overall imaging performance varies depending upon the number of diffracted beams passing through the scanner optics. An even number of beams provides very different trends compared to the results from an odd-number of beams. This significant non-linear behavior occurs in certain pitch regions corresponding to 3 beam interference imaging. In this region the imaging performance and the pattern printability become extremely sensitive to the LS duty cycle. In addition, there is a notable tradeoff between the DOF and the NILS that is observed in the problematic 3-beam region and this tradeoff eventually affects the achievable process window (PW). Given the practical real world constraints such as the design rules and target design restrictions, computing upper- and lower-bounds of the through-pitch DOF and NILS will be especially useful for both lithographers and metrology target designers in understanding this complex behavior, as well as helping in the design of optimal targets used for applications including alignment, overlay control, and process control in high volume semiconductor manufacturing.

The main roadblock for EUV lithography to be successfully adopted for high-volume manufacturing is the current lack of source power. One way to help mitigate this problem is to optimize the photoresist by increasing both absorbance and quantum yield. The latter represents the ratio between the sums of generated acids and absorbed photons. Yield is also thought to be limited by the number of generated electrons per absorbed photon, or electron yield, that may be generated after a photo absorption event. While absorbance is relatively easy to measure, yields are extremely difficult to quantify, and the debate on upper limits is far from settled. In this paper, we present how, using synchrotron light with tunable energy, we directly measured dispersion curves and electron yield for ArF, KrF and EUV photoresists using X-ray Absorption Spectroscopy. Knowing the electron yield allowed us to better model organic EUV materials: stochastic simulations show how both electron yield and blur are very similar for organic materials, and how the electron blur is not a fixed property of the material, but may vary spatially, depending on a combination of photoresist formulation and local photon absorption density.

Scanning Electron Microscopy (SEM) is widely used to measure Critical Dimensions (CD) in semiconductor lithography processes. Correlation between the CD-SEM metrology and target profile has drawing attention from metrology community [1]. In this paper, we use a recently developed CD-SEM simulator [2-3] to investigate some artifacts of SEM metrology. The simulation consists of two parts. First part is a stochastic resist modeling for lines and spaces through pitch, exposure dose and focus. Second part is CD-SEM simulation. Both CD and LWR extracted from experimental CD-SEM images were used to train the SEM model. Two types of artifacts were found to be metrology dependent: the first artifact is that a CD-SEM measures CD at various heights across pitch for the same SEM threshold. The second artifact is a misleading CD measurement for trenches not fully developed. By overlapping the CD-SEM simulation with 3D lithography simulation, correlation between CD-SEM metrology and target 3D profile is studied. Finally, a Process Window (PW) analysis based on both experiment and simulation is presented, using the simulated features and SEM images to correct the experimental PW.

Scanning Electron Microscopy (SEM) is widely used to measure Critical Dimensions (CD) in semiconductor lithography processes. As the size of transistors keeps shrinking, the uncertainty associated with CD-SEM accounts for a fast growing contributor to the entire manufacturing error budget. Capability to predict the metrology results from a CDSEM is highly desirable to quantify the uncertainty of metrology. Simulation has proven to be a valuable means of studying both SEM metrology and photolithography. Monte-Carlo based simulators are generally used to model the detailed image formation process of a CD-SEM, while physics-based photolithography simulations, such as PROLITH™ are commonly used for lithography modeling. However, the high computational cost limits the application of Monte- Carlo based CD-SEM simulations in conjunction with lithography simulation. We present here a compact physical CDSEM simulator which simplifies the image formation process while preserving many essential SEM imaging mechanisms. Several applications of our CD-SEM simulator are presented to demonstrate the predicting capability compared with experiments.

We have shown that the dissolution properties can be successfully modified to improve the line/space profile and LWR of a low diffusion EUV CA resist. The surface roughness is a function of hot spots in the nominally unexposed regions of the resist material. We conjecture that the photoacid hot spots are formed due to DC flare present in the optical train of the exposure system. We also have shown that the PAGs can be further improved for out-of-band radiation (OOB) response. The improvement can be as much as 557% for 193nm exposure, and 838% by 248nm exposure. The improved OOB response leads to better contact hole performance. We also shared our continued improvement in resist witness plate performance with the majority of our resists passing for carbon growth, and all samples passing for non-cleanables. There does appear to be a site-to-site bias which we attribute to differences between e-beam and EUV exposure and/or substrate working distance from the source. Lastly, we show outstanding lithographic process window for 24 nm contact arrays on an NXE 3300 stepper as well as 15 nm half pitch lines and spaces on the PSI interferometric tool.

Implementation of EUV Lithography (EUVL) for device high volume manufacturing (HVM) requires advanced photoresists capable of meeting the criteria of advanced logic and memory design rules. To achieve the level of performance required, resists must show excellent performance in terms of resolution, LWR (or CDU) and sensitivity. In addition, resists must meet the outgassing criteria required for HVM on the NXE toolset. Lastly, it is anticipated that resists with low OOB sensitivity will also be required.In this paper, we describe our progress in all of these areas. Based on our results, we believe we are on track to deliver production worthy resists for the EUVL era.

Patterning uncertainty in EUV lithography arises from each lithographic component: the source, the photomask, the optical system, and the photoresist. All contribute to line roughness and contact disuniformity. In extreme cases, feature variability can result in patterning failures such as line microbridging or random missing contact holes. Historically, redundant contact holes (or vias) were placed to overcome the effects of a missing contact. Due to the aggressive CD shrink of feature size, the use of redundant contacts has been progressively decreased. For some types of devices, almost every contact of the billions found on the chip must be electrically active in order for the device to function. In such scenario, lithographic printing failures may cause catastrophic loss of yield, considering that closed contacts can hardly be corrected by smoothing techniques or etching. In this paper, the minimum contact CD which prints without failure – the contact hole printability limit – is studied for 54nm and 44nm pitch dense arrays. We find that the same resist may show dramatically different printability limits depending upon sizing dose and illumination conditions. This analysis will be implemented to estimate, through simulation-assisted experiments, the required exposure dose and aerial image to safely print sub-30nm contact holes.