Image Placement Error Research Articles

Aperiodic multilayer masks for M3D mitigation in high- and hyper-NA extreme ultraviolet lithography

As extreme ultraviolet lithography tools with higher image numerical apertures (NAi) are introduced, the range of angles at the multilayer mask stack is also increased. Lithography systems are designed to fulfill the “abbe sine rule,” where NAm is related to NAi by the reduction factor. As a result, increases in NAi will increase NAm. High-NA and hyper-NA systems will be implemented with anamorphic optics, 4× in “X,” and 8× in “Y” to reduce the necessary angles. Even so, hyper-NA masks may see illumination angles as high as 13.4°, up from 10.8° for 0.33 NA. This represents a challenge for maintaining through-angle mask reflection using the current periodic mask multilayer structures. In addition to the reflectance amplitude, the phase of reflected light, which plays an important role in imaging, is also strongly influenced by increasing angles. The propagation of light through each bilayer in the stack imparts a phase shift based on the incidence angle. This is then accumulated over many layers, inducing phase effects, which are unique to each illumination point. This will become especially true for high-NA and hyper-NA mask applications. While adjustments to the multilayer period are sufficient to achieve acceptable reflectance, periodic multilayers may suffer in normalized image log slope (NILS) and image placement error (IPE) metrics as a result of strong oblique multilayer M3D phase effects. In this work, we present a first principles methodology for the systematic reduction of oblique multilayer M3D effects through the use of aperiodic multilayer design. We find that when used for low-k1, hyper-NA patterning NILS is improved by 40+%, while IPE through pitch is similarly improved. Under high-NA conditions, a 5%–25% NILS gain is found alongside the enhanced depth of focus.

Overlay error statistics for multiple-exposure patterning

Background: The mathematical equations that explain overlay error of multiple-exposure patterning schemes have not been fully described in the literature and some commonly accepted methods lead to inaccurate estimated and/or measured overlay error. Aims: Develop the proper mathematical framework, using a first principles statistical approach, so that engineers using multiple-exposure patterning can determine the overlay impact and overlay controls needed. Alert patterning community that grouped overlay metrology of multiple-exposures undermeasures the true overlay error. Approach: Use image placement error and population-based statistics to enable a mathematical framework to be established that predicts the actual overlay error for an overlaying pattern that minimizes overlay error back to a pattern that is patterned with multiple-exposure patterning. Results: The overlay error between two patterns is usually less than the root sum square of the two overlay error values of the patterns individually measured to a common prior pattern. Overlay error for a pattern minimizing back to multiple-prior patterns increases quickly as systematic overlay error between the prior patterns increases. Conclusions: Controlling systematic overlay error between patterns of a multipatterned layer is important for subsequent patterns that need to minimize overlay error back to the composite multipatterned layer. The ratio between the overlay error determined with metrology and true overlay can be calculated.

Characterizing and solving imaging challenges in thick resists for wafer and panel based lithography applications

Abstract Increasing volume using larger substrates with decreasing process margins continues to create new challenges for advanced packaging applications. Key step and repeat camera technology continues evolving for the mass production of microstructures used for 2.5D and 3D technologies. Printing dense arrays of smaller features with high aspect ratios requires higher sidewall angles in thick photoresists and polyimides. To help solve these imaging challenges we have leveraged resist modeling software and guided the adjustment of optical parameters needed for better performance. Higher contrast films have also been evaluated to help achieve the improvements in performance needed. Resist models that can include the effects of flare have been critical to understand the requirement for printing in thick negative resists and has aided in printing features on varying topography and film thicknesses. Special chucks help improve the flatness of warped wafers and real-time auto-focusing provides good image fidelity. Printing microstructures over larger formats with higher throughput has been accomplished using large magnification adjustment for improving overlay and validated by characterizing image placement errors over large substrates. Examples of resist models that are created using resist parameters and optimized using SEM images of printed features are compared. Extrapolations of the resist models have shown that guide improvements can be achieved by varying optical parameters. SEM images confirm that the modeled result of the optimal solution was achieved. Reduction in large substrate overlay error was achieved after the stage corrections were applied. Examples show that topographical errors of warped wafers can be reduced and how real-time auto-focusing for each exposure minimizes focus errors.

Improving mask registration and wafer overlay control using an ultrashort pulsed laser

Requirements for control of mask registration errors and wafer overlay errors become more demanding as the integrated circuit (IC) feature size specifications become tighter and tighter. Registration control, also known as RegC, is a well-known laser-based process in the IC industry that has proven to be robust, repeatable, and efficient in adjusting mask registration and wafer overlay. Using an ultrashort pulsed laser, microscopic deformation elements, strain centers, which are also known as pixels, are generated deep inside the fused silica mask substrate causing strain. The induced strain moves the mask pattern in predicted calculated directions to reduce the measured image placement error (IPE). RegC corrections can be applied in two different ways: as a periphery-mask correction, which places the strain centers in the mask frame area outside the exposure field, or as a full-mask correction, that places the strain centers across the entire mask, including the exposure field. Relative pattern displacements up to 1 nm can be achieved between sites located at a distance of 1 mm, reducing the magnitude of the residual IPE by an average of 20% (10% to 25% range) for a periphery correction or 40% (20% to 50% range) for a full-mask correction. The input data for the RegC process are typically taken from mask registration information, which is measured by registration metrology tools or measured from wafer overlay (the latter is not within the scope of this paper). This paper reviews the RegC process flow for improving mask registration in mask shops.

General analytical expressions for the impact of polarization aberration on lithographic imaging under linearly polarized illumination.

With the shrinking of the critical dimension, the impact of polarization aberration on lithographic imaging becomes increasingly prominent. In this paper, the linear relationships between the image placement error and odd Pauli-Zernike polarization aberrations, as well as those between the best focus shift and even Pauli-Zernike polarization aberrations, are established by analyzing the imaging of the alternating phase-shifting mask. The relational expressions of the polarization aberration sensitivities (PAS) and the polarization angle of illumination are obtained based on these linear relationships. Then the expressions for the zero-value points and extremum points of the PAS are derived, and the impact of the polarization angle of illumination on the PAS is analyzed. The derived analytical expressions match the simulation results well; these can be used to analyze the detrimental impact of polarization aberration on lithographic imaging and provide a theoretical basis for exploring polarization aberration measurement and control techniques.

Intrafield overlay correction for illumination-based distortion

The use of different illumination source shapes and multipatterning processes used to generate sub–40-nm images can create image placement errors level to level, resulting in significant intrafield overlay errors. These errors arise because of the impact of the different pupil shapes on lens aberrations resolving into image placement errors as well as because different tools will react differently to the same pupil shapes. We compare the impact of two extreme illumination sources on intrafield image placement and its effect on overall pattern overlay. We also discuss a method to empirically isolate and measure the amount of intrafield overlay distortion relative to a reference illumination source and then use the results to correct the resultant image placement errors.

直视合成孔径激光成像雷达滑动聚束模式下图像信噪比研究

A measuring method of polarization aberration based on aerial image of alternating phase-shift mask is proposed.The polarization aberration is represented by Pauli-Zernike coefficients.The image placement error and best focus shift are measured by image sensor at multiple illumination settings,combined with X and Ylinearly polarized illuminating light.The Pauli-Zernike coefficients are retrieved using the calibrated sensitivity matrix of polarization aberration.The validity of this proposed method is verified by numerical simulations,and the results indicate that the measuring accuracy is less than 3.07 mlambda.

Analytical analysis of the impact of polarization aberration of projection lens on lithographic imaging

In high-numerical aperture (NA) and hyper-NA lithography systems, the polarization aberration of the projection lens leads to imaging degradations. Typically, commercial simulators, which usually entail relatively higher computing cost and lack sufficient theoretical support, are used to explore the relationship. Analytical analysis of the impact of polarization aberration of the projection lens on the aerial image is performed. In the analysis process, an alternating phase-shift mask is used, and different components of the linear polarized illumination light vector are considered. The analytical expressions of image placement error (IPE) and best focus shift (BFS) caused by polarization aberration are derived from the intensity of the aerial image. The linear relationships between IPE and odd Pauli-Zernike polarization aberrations as well as that between BFS and even Pauli-Zernike polarization aberrations are established. Moreover, the polarization aberration sensitivities are given and compared when different components of the linear polarized illumination light vector are adopted. All derived expressions match simulation results well and can be used to understand more fully the detrimental impact of polarization aberration on lithographic imaging. The accuracy of the linear relationships is assessed by the least square method.

Improved compensation for a reduction stepper to meet the challenges for advanced packaging applications

Advanced packaging manufacturers require steppers that will provide solutions for the challenges encountered with new advances in wafer-level packaging technologies such as TSV, eWLB, silicon and glass interposers being utilized in leading edge mobile devices. Step and repeat photolithography systems capable of finer imaging with tighter overlay are being introduced to meet the challenging manufacturing requirements associated with the mix and match needed for volume production on larger wafers. A 2X reduction stepper with unique features incorporated that extend the range of compensation is necessary to achieve the tighter specifications needed for many advanced packaging applications printed on 300 to 450mm wafers. A high throughput projection optical system is used to expose circuit patterns from a reticle mask onto a substrate to image features with the optimal fidelity required for advanced packaging technologies. The camera incorporates 350–450nm light from a mercury arc lamp that is transmitted through the mask containing circuit patterns. The imaging field prints a large 52mm × 66mm area in a single exposure. These features enable a system to process wafers in fewer shots which result in higher throughput using lower power. Substrates are positioned with a precise X, Y, Θ stage by locating marks using an off-axis, bright field alignment system with fully trainable mark feature capability. The approach results in precisely placed features within a layer and from layer to layer without directly referencing the reticle. The integrated metrology and precision positioning subsystem technologies are combined with a low distortion projection lens and a wide range of adjustments, allowing the stepper to be integrated into a production line in a mix and match setup with other lithography systems. This equipment can be used to image critical layers on substrates while ensuring grid registration and alignment with other lithography systems that are also printing images in the same process line. Several important global and intra-field image placement relationships for devices requiring multiple layer patterning have been combined in the stepper matching correction software. Further adjustment to the tool can be made to improve overlay when incorporated with fab-wide yield management software for automated, real-time process control. The types of adjustments needed and techniques that can be applied to compensate for image placement errors over large areas are discussed.

Particle contamination effects in extreme ultraviolet lithography: enhanced theory for the analytical determination of critical particle sizes

Existing analytical and numerical methodologies are discussed and then extended in order to calculate critical contamination-particle sizes, which will result in deleterious effects during extreme ultraviolet lithography E-chucking in the face of an error budget on the image-placement-error (IPE). The enhanced analytical models include a gap-dependent clamping pressure formulation, the consideration of a general material law for realistic particle crushing and the influence of frictional contact. We present a discussion of the defects of the classical de-coupled modeling approach where particle crushing and mask/chuck-indentation are separated from the global computation of mask bending. To repair this defect we present a new analytic approach based on an exact Hankel transform method which allows a fully coupled solution. This will capture the contribution of the mask-indentation to the image-placement-error (estimated IPE increase of 20%). A fully coupled finite element model is used to validate the analytical models and to further investigate the impact of a mask back-side CrN-layer. The models are applied to existing experimental data with good agreement. For a standard material combination, a given IPE tolerance of 1 nm and a 15 kPa closing pressure, we derive bounds for single particles of cylindrical shape (radius×height<44 μm2) and spherical shape (diameter<12 μm).

Analytical approach to the impact of polarization aberration on lithographic imaging

An analytical approach to the impact of polarization aberration on lithographic imaging is proposed. The linear relationship between image placement error (IPE) of alternating phase-shifting mask (Alt-PSM) and odd aberration items of polarization aberrations, as well as that between best focus shift (BFS) of Alt-PSM and even aberration items of polarization aberrations are established by analytical equations, respectively. The validity of the linear relationships is demonstrated by numerical results. The differences and connections between scalar aberration and polarization aberration are briefly discussed based on these linear relationships.

Assessing the mask clamping ability of a low thermal expansion material chuck

The successful implementation of extreme ultraviolet lithography (EUVL) for patterning in the sub-32 nm regime will require significant improvements in image placement (IP) accuracy over the next few years. The IP error budget for the mask system is extremely stringent; consequently, the reduction or elimination of all sources of error is essential. One potential source of IP error is the effect of the residual nonflatness of the patterned surface of the mask during exposure scanning. The focus of this article is to characterize the clamping ability of a Coulombic electrostatic pin-type chuck and to assess its ability to adequately flatten an EUVL mask. The chuck was fabricated from low thermal expansion material, with a nonflatness over the pin area of approximately 74 nm. Experimental results illustrate that a highly bowed substrate (1149 and 1047 nm for the frontside and backside nonflatness, respectively) could be chucked flat to less than 100 nm. Numerical models were also used to simulate electrostatic chucking and to predict the final nonflatness of the pattern surface of the mask. Excellent agreement was found between the experimental and numerical results. In addition, the modeling tools developed here can be used to optimize the chuck design parameters and to establish the requirements on nonflatness of both the EUVL substrate and the surface of the chuck.

Experimental determination of image placement accuracy in extreme ultraviolet lithography

Image placement and overlay error specifications in the International Technology Roadmap for Semiconductors continue to get tighter with each successive technology node. Some of the primary contributors to this error in extreme ultraviolet lithography are reticle and chuck surface nonflatness and chucking flatness nonuniformity. In this article, the authors report on results from a set of experiments that were designed to identify and separate chuck and mask contributions to image placement error (IPE). In the first set of experiments, a reticle with large peak-to-valley (P-V) nonflatness was printed to study reticle flattening and reticle thickness variation contributions to the final IPE. The experiments were designed to partition the total IPE into tool correctable and noncorrectable components. A second set of experiments was performed with a flat reticle that can be imaged in four different orientations (0°, 90°, 180°, and 270°). Both reticles have arrays of image placement fiducials that can be printed on the wafer and then used to measure the IPE.

STRONG PHASE SHIFT MASK MANUFACTURING ERROR IMPACT ON THE 65 NM POLY LINE PRINTABILITY

A methodology and Monte Carlo simulation flow with the integrated LSI Logic package, Molotof, was applied to the 65 nm poly line sensitivity analysis. Strong phase shift mask (sPSM) manufacturing process specification was optimized to obtain image critical dimensions (CDs) and image placement errors (IPEs) complying with technology process tolerances. Reticle manufacturing statistical variarions of phase depth, phase width, and phase intensity imbalance were used to generate a virtual sPSM for imaging poly lines on a wafer. A criterion for qualifying reticle specification is to obtain all latent image CDs and IPEs within a photo process module allowed range for a given mask specification. The approach allows for computing reticle and litho budgets into CD imaging performance. Sensitivity to a single parameter variation and full statistical analysis of the 65 nm poly line imaging performance affected by reticle manufacturing errors is presented. The optimum reticle specification, yielded 100% of critical dimensions and image placement errors, was found in simulation and confirmed by empirical data.

Analysis of pattern-dependent image placement of single-membrane stencil masks for electron-beam lithography

Large single-membrane stencil masks have been developed for electron-beam lithography. Since a large membrane induces large image placement (IP) error, which is pattern dependent, a method of correcting EB data has been studied to compensate the membrane distortion. In this study, firstly, the effect of crystal anisotropy of a Si membrane to the distortion is examined by making masks from blanks with different orientations. The influence of the anisotropy is found to be small and simulation based on isotropic modeling should be applicable. Secondly, a finite element method (FEM) called ANSYS and Pseudo-FEM are used to predict distortions for three masks with 8 mm-, 12 mm-, or 18 mm-square die of an opening ratio of 0.2 on a 24 mm-square membrane. The simulation results are compared with the results obtained in the experiment on anisotropy and a previous experiment. Qualitative agreement is observed between simulation and experiment but quantitative agreement is obtained only after introduction of adjustment factors. A suggestion is made to improve the IP correction scheme for EB data.

Predicting electron projection lithography mask membrane image placement errors

Electron projection lithography (EPL) has been identified as a viable candidate of the next-generation lithography technologies for moderate-volume production of systems-on-a-chip for the sub-65nm nodes. EPL is especially suited for the fabrication of contact structures. The development of a low-distortion mask is essential for meeting the stringent requirements at these lower nodes. This research focuses on quantifying in-plane distortions (IPDs) due to mask fabrication and the effects of different mounting schemes in the e-beam writer and exposure tools for a 200mm EPL stencil mask. In order to quantify the IPD of the freestanding membranes, three-dimensional finite element (FE) models have been developed and benchmarked with experimentally measured data. Metrology measurements in both the membrane-side up and membrane-side down configurations can be used to correct for image placement errors (due to the fabrication and chucking of the mask) by compensating for IPD during the exposure process. To correct for higher order distortions associated with pattern specific errors, FE models were generated to simulate the entire process flow, including chucking. As a sample test case, a stencil pattern was replicated uniformly over the membrane arrays. The actual mask was fabricated and metrology measurements were taken. Selected membranes were experimentally measured to compare with FE results and the data correlated well in both direction and magnitude. Results illustrate that FE modeling and simulation can be used to predict mask distortions due to fabrication. Such techniques can be instrumental in meeting the stringent error budget required at the lower lithographic nodes.

Distortion of chucked extreme ultraviolet reticles from entrapped particles

Successful imaging of patterns with critical dimensions less than 45nm with extreme ultraviolet lithography (EUVL) requires stringent controls on all sources of image placement (IP) errors. Among the potential sources of IP error is the mechanical distortion of the patterned mask when mounted in the exposure tool. An EUVL reticle can exhibit both in-plane distortion and out-of-plane distortion due to the presence of debris lodged between the mask and the electrostatic chuck. Even particles with a compressed height as small as 100nm have the potential to consume a significant portion of the IP error budget. To alleviate this problem, a thorough understanding of the response of the reticle∕particle∕chuck system during electrostatic chucking is essential. This article describes experimental indentation testing to characterize relevant nanoscale material properties and the subsequent use of the data in finite element models that simulate the system response under typical chucking conditions.

Investigation of overlay errors due to the interaction of optical and extreme ultraviolet mask fabrication processes

Overlay tolerances are becoming increasingly severe as lithography technology drives the minimum integrated-circuit feature size below 65nm. Manufacturable solutions at the lower nodes are essentially unknown. The goal of this article is to investigate overlay errors at the 45-nm node, for a mix-and-match of optical and extreme ultraviolet (EUV) lithographies. In particular, image placement errors induced during mask fabrication are predicted for a specified test pattern, which includes the possibility of having individual layers on the device wafer patterned by either optical or EUV lithographic processes. Finite element (FE) models have been developed to simulate the response of both optical and EUV masks during fabrication and chucking. In order to track image placement errors of the actual features within the pattern area, submodeling techniques were developed for the FE simulations. In addition, equivalent material properties were determined for the individual patterns used for the different optical and EUV masks. The FE simulations clearly indicate that the most significant source of overlay error is the fabrication of the optical reticle, primarily due to differences in chucking of the mask during e-beam patterning and exposure. Modeling results are being used by the industry to establish guidelines for reticle chucking and fabrication standards.

Image placement error: closing the gap between overlay and imaging

We perform experimental and theoretical studies of image placement error induced by aberrations of the projection lens. The goal is to experimentally determine the magnitude of the image placement errors, to compare experiments and simulations, and to screen possible correction strategies. The calculations and experiments are done for ArF lithography. Theoretically, we simulate image placement error using the projection lens aberration data and simulators such as Prolith or Solid-C. Features with low and high sensitivities to lens aberrations are identified, together with a reference feature that has low sensitivity for image placement error. Dedicated reticles are fabricated to print various features at different illumination conditions on the same substrate. The resulting patterns could be analyzed using top-down scanning electron microscopy (SEM), but also optically with the standard optical overlay tool KLA5200. For both techniques, the experimentally found image placement errors are in excellent agreement with simulations. In simulations, we calculate the dependency of the image placement error on pattern density, pattern orientation, and illumination conditions. These tendencies are experimentally reproduced. We conclude with a case study that demonstrates a possible correction strategy for image placement error.

Alternating phase-shifting mask design for low aberration sensitivity

The aberration present in the lenses of exposure systems can cause placement errors to the images produced by alternating phase-shifting masks (PSMs). In reality, when the aberration signature varies from one lens to another, the magnitude of placement error also varies. It remains a question of how the alternating PSM should be designed, so that the image placement error, on average, can be minimized. To achieve this goal, we are interested in optimizing the phase width of an alternating PSM with a fixed critical dimension (CD). The constraint of the optimization is the mean of root mean square (rms) aberrations for a set of interest of exposure systems. To begin the analysis, the image placement error is expressed as a function of illumination, mask spectrum, and wave aberration. A Monte Carlo technique is then applied to produce random samples of wave aberration and image placement error. This analysis shows that a global minimum of mean image placement error is likely to occur at phase widths between 0.2[/numerical aperture (NA)] and 0.4(/NA). This is further confirmed by analytically considering the expected value of the square of the image placement error. The methodology of finding the optimal phase width is applicable to the design of all alternating PSMs.