Antoinette F. Molless

Optimum Mask and Source Patterns to Print a Given Shape

New degrees of freedom can be optimized in mask shapes when the source is also adjustable, because required image symmetries can be provided by the source rather than the collected wavefront. The optimized mask will often consist of novel sets of shapes that are quite different in layout from the target IC patterns. This implies that the optimization algorithm should have good global convergence properties, since the target patterns may not be a suitable starting solution. We have developed an algorithm that can optimize mask and source without using a starting design. Examples are shown where the process window obtained is between 2 and 6 times larger than that achieved with standard RET methods. The optimized masks require phase shift, but no trim mask is used. Thus far we have only optimized 2D patterns over small fields (periodicities of approximately 1 micrometer or less). We also discuss mask optimization with fixed source, source optimization with fixed mask, and the re-targeting of designs in different mask regions to provide a common exposure level.

Lithographic comparison of assist feature design strategies

Subresolution assist features, when used in conjunction with off-axis illumination, have shown great promise for reducing proximity effects while improving lithographic process window. However, these patterns result in an increased emphasis on the mask manufacturing process, primarily in the areas of mask writing and inspection. In choosing a design strategy, one must be careful to account for the mask making capabilities, such as write tool grid size and linearity, along with the lithographic effect of errors in the mask making process. In addition to mask errors, stepper lens aberrations and expected process variations can also have a large influence on design rules. Generally, design tradeoffs must be made to balance the impact of these for the best overall lithographic performance.

Level-specific lithography optimization for 1-Gb DRAM

A general level-specific lithography optimization methodology is applied to the critical levels of a 1-Gb DRAM design at 175- and 150-nm ground rules. This three-step methodology-ruling out inapplicable approaches by physical principles, selecting promising techniques by simulation, and determining actual process window by experimentation-is based on process latitude quantification using the total window metric. The optimal lithography strategy is pattern specific, depending on the illumination configuration, pattern shape and size, mask technology, mask tone, and photoresist characteristics. These large numbers of lithography possibilities are efficiently evaluated by an accurate photoresist development bias model. Resolution enhancement techniques such as phase-shifting masks, annular illumination and optical proximity correction are essential in enlarging the inadequate process latitude of conventional lithography.

Optimum mask and source patterns to print a given shape

New degrees of freedom can be optimized in mask shapes when the source is also adjustable, because required image symmetries can be provided by the source rather than the collected wave front. The optimized mask will often consist of novel sets of shapes that are quite different in layout from the target integrated circuit patterns. This implies that the optimization algorithm should have good global convergence properties, since the target patterns may not be a suitable starting solution. We have developed an algorithm that can optimize mask and source without using a starting design. Examples are shown where the process window obtained is between two and six times larger than that achieved with standard reticle enhancement techniques (RET). The optimized masks require phase shift, but no trim mask is used. Thus far we can only optimize two-dimensional patterns over small fields (periodicities of ;1 mm or less), though patterns in two separate fields can be jointly optimized for maximum common window under a single source. We also discuss mask optimization with fixed source, source optimization with fixed mask, and the retargeting of designs in different mask regions to provide a common exposure level.

Lithographic effects of mask critical dimension error

Magnification of mask dimensional error is examined and quantified in terms of the mask error factor (MEF) for line and hole patterns on three types of masks: chrome-on-glass (COG), attenuated phase-shifting mask (PSM) and alternating PSM. The MEF is unity for large features, but increases rapidly when the critical dimension (CD) is less than 0.5 (lambda) /NA for line-space patterns and 0.75 (lambda) /NA for contacts. In general dark-field spaces exhibit higher sensitivity to mask dimensional error than light-field lines. Sensitivity of attenuated PSMs is similar to COG masks, even for applications in which attenuated PSMs provide benefits in process latitude. Alternating PSMs have the lowest MEF values. Although the MEF has only a slight dependence on feature nesting for contacts, dense lines and spaces exhibit markedly higher MEF values than isolated features. The MEF of a 0.35 (lambda) /NA isolated line is 1.6 whereas that of a dense line of the same dimension is 4.3 illumination is effective in reducing the mask error sensitivity of dense lines. Dose variation causes changes in the MEF of contacts but has little effect on line-space features; focus error degrades (increases the value of) the MEF of both pattern types. A high diffusion and low contrast photoresist process also worsens the MEF. Consequences of mask CD error amplification include tightening of mask specification, design grid reduction, shift in optimal mask bias and enhanced defect printability.

Application of the aerial image measurement system (AIMS)TM to the analysis of binary mask imaging and resolution enhancement techniques

The newly developed Aerial Image Measurement System (AIMSTM*) was used to quantify the lithographic benefits of several resolution enhancement techniques as compared to standard binary mask imaging. This system, a microscope based stepper emulator, permits rapid characterization of mask images from both binary and phase shifted mask (PSM) patterns at multiple focal planes. The resultant images are captured digitally with a CCD camera and analyzed using an exposure-defocus tree technique to quantify the depth-of-focus as a function of exposure latitude. The AIMS is used to extract both phase and transmission errors from captured aerial images of all the masks evaluated. AIMS results are compared to wafer electrical linewidth data. A 0.5 numerical aperture (NA) DUV stepper was used with a partial coherence of 0.6 combined with IBM APEX-E resist process. Collected data were analyzed using techniques identical to the AIMS analysis, allowing for a high level of consistency. Comparative data focused on binary mask imaging for the verification of the AIMS results. Trends associated with feature sizes and types are discussed.

Alternating phase-shifting mask with reduced aberration sensitivity: Lithography considerations

Aberration sensitivity of alternating phase-shifting masks (PSMs) can be reduced by taking advantage of the trim exposure. Rather than a single phase region bordering each edge of a line, the enhanced alternating PSM technique uses multiple phase regions. The number of phase regions and their widths can be optimized for overall process tolerance including aberration sensitivity and exposure latitude. For exposure with a wavelength of 248 nm and a numerical aperture of 0.68, the optimal number of phase regions is two, with widths between 100 nm and 200 nm. These auxiliary phase regions do not affect the final pattern if a light-field trim mask is used. No extra processing step is necessary. With the enhanced alternating PSM technique, isolated lines of average dimension as small as 36 nm can be delineated using 248 nm lithography with a 3(sigma) linewidth control of 13.4 nm. The mean critical dimension of 36 nm corresponds to k1 equals 0.1.

Linewidth variation characterization by spatial decomposition

Characterization of linewidth variation by a three-step methodology is presented. Via electrical linewidth measurement, sources of linewidth variation with distinct spatial signatures are first isolated by spatial analysis. Causes with similar spatial signatures are then separated by contributor-specific measurements. Unanticipated components are lastly identified by examination of the residuals from spatial analysis. Significant sources include photomask error, flare, aberrations, development nonuniformity, and scan direction asymmetry. These components are then synthesized to quantify the contributions from the three modules of the patterning process: photomask, exposure system, and postexposure processing. Although these modules are independent of one another, their effects on linewidth variation may be correlated. Moreover, the contributions of the modules are found to vary with exposure tool, development track, and lithography strategy. The most effective means to reducing the overall linewidth variation depends on the relative importance between these components. Optical proximity correction is efficacious only for a well-controlled process where proximity effect is the predominant cause of linewidth variation.

Simulation and experimental evaluation of double-exposure techniques

The process windows and capabilities of double exposure techniques with binary and attenuated masks are explored using simulation and experiment, including the effects of resist properties, illumination conditions and overlay error. Here it is shown that by using a low partial coherence factor (sigma) for the two exposures, the total window is considerably improved over that obtained using higher partial coherence illumination. We call this process ORAMEX, which stands for Ordinary Resist And Multiple EXposure. It was found that the process window for nested lines and spaces using ORAMEX is considerably better than that for conventional illumination. This is shown for aerial images and for aerial images plus a resist model with contrast and diffusion length similar to that of state of the art Deep UV resists. In fact, the total process windows found for ORAMEX show good process latitudes for both dense and isolated features, with ORAMEX usually enhancing dose latitude more than single exposure off axis illumination does. Overlay errors are found not to affect the process window for individual features. However, they do affect the common window for every other line (in positive resist) but not spaces. It was also found that using attenuated masks instead of binary masks further improves the process window and resolution of ORAMEX. Experimental results agree with simulation and show a process window for 150 nm lines and spaces with over 0.4 micrometer depth of focus and 15% dose latitude in 0.6 micrometer of resist using ORAMEX and chrome on glass masks. Using attenuated masks and ORAMEX a similar process window (0.4 micrometer DOF and 16% dose latitude) was obtained for 125 nm lines and spaces. Both results were obtained on a 0.6 NA Deep UV stepper using commercial positive resist.

Understanding across chip line width variation: the first step toward optical proximity correction

The prerequisite to successful optical proximity correction is an in depth understanding of the relevant parameters leading to patterning inaccuracies. The work presented in this paper is based on a test chip specifically designed to investigate sources of 1D line width errors, deemed the most critical for optical proximity correction. Data are presented relating line width errors to pitch and pattern density and highlighting the complex interactions between these two main contributors of line width error. The goal of this paper is not to derive a mathematical model for line width variation in etched polysilicon line structures but merely to qualitatively bound the complex interaction of pitch and pattern density based line width variation. The results of this investigation indicate that pattern density has a very significant effect on line width and that lithography, not just reactive ion etch, is significantly impacted by pattern density.