Optical lithography at a 193-nm exposure wavelength has been under development at MIT Lincoln Laboratory for several years, supported by ARPA's Advanced Lithography Program. As part of this program, a prototype 193-nm full-field step-and-scan lithographic exposure system was built and installed in the clean-room facilities of MIT Lincoln Laboratory. This exposure system has now been in use for one year, supporting a program of photoresist and lithographic process development at 193 nm. This paper describes the characteristics of the exposure system and some of the advances in 193-nm lithography that have been achieved with the system.

This paper reviews the status of step-and-scan technology. It reviews the main features of the technology and highlights its advantages over the conventional step-and-repeat technology. Data is presented to show the superior field distortion, linewidth control, and overlay performance offered by step-and-scan. The future direction of the technology is discussed in the context of the Semiconductor Industry Association (SIA) roadmap. It is indicated that the step-and-scan technique will allow optical lithography to extend below 180 nm resolution.

New I-line resists are claimed to be usable at 0.35 micrometers design rules. We have examined the suitability of several such materials (JSR IX750, Sumitomo PFi-38a, OCG RX64I) for this purpose and compared them with our production 0.5 micrometers resist, JSR IX700. A variety of criteria have been used, including the measured focus exposure windows at e-min and e-max, DOF vs. CD for grouped and isolated lines as well as contacts, linearity, and proximity response as a function of pitch. A limited study has been done on the impact of embedded phase shift reticles on printing small geometry contacts. We report upon the process improvements observed with two different reticle transmissions, their impact on isofocal bias, as well as the issue of sidelobe formation. Proponents of DUV claim that modern DUV materials exhibit significant advantages in terms of process window and more over are applicable to smaller geometries without the need for supplementary techniques such as phase shifting or modified illumination. In this study, we have examined the performance of a number of DUV materials (BASF ST3.5, OCG ARCH, JSR KRFL2 and an as yet experimental JSR resists) on ASM-L and Nikon excimer laser steppers. Limited results were also obtained using Shipley 2408 and dyed XP-9444 (0.8) on the SVG Micrascan II. Our studies conclude with a comparison of the CD swing observed over a variety of chemically mechanically planarized steps. This has been done for selected I-line and DUV resists with the aid of a TAR or BARC or as in the case of the broad band SVG system either a BARC or a dyed resist.

This paper discusses two lithography strategies for printing 0.35 micrometers features: broad band deep ultraviolet (DUV) lithography using a chemically amplified resist dyed to an optical density of 0.40 per micrometers and the use of high numerical aperture (NA) i-line lithography with an advanced i-line resist. Lithographic results such as linearity, exposure and focus latitudes, reflective notching control, and post exposure bake (PEB) delay time are compared. Results on product wafers are also illustrated. Finally, experimental data is compared with PROLITH/2 simulations.

The optical lithography is extending its life by combining high numerical aperture (NA) optics and shorter wavelength. The shorter wavelength lithography has required the new developments of related technologies. In particular, DUV resists require an entirely different resist chemistry. Much progress has been demonstrated in the field of transparent chemically amplified resists with high sensitivity. However, this DUV lithography ((lambda) equals 248 nm) has been delayed for mass production due to their limitations, such as (i) delay time effects, (ii) high cost ownership due to expensive resist materials and laser maintenance, and (iii) critical dimension (CD) variation over topography caused by multireflection of topographic features. On the other hand, i- line lithography ((lambda) equals 365 nm) has apparently been applied to 64M DRAM of 0.35 micrometers design rule, and attempted to 0.30 micrometers technology which corresponds to 2nd generation 64M DRAM or 1st generation 256 M DRAM. It might be achieved by combination of off-axis illumination (OAI), phase shift mask (PMS) and advanced resist process technique of i-line lithography. Therefore, i-line lithography can be more practical method rather than DUV lithography for the mass production. In this paper, we have optimized the i-line lithographic techniques for the various pattern shape and density for 0.30 micrometers design rule. Optimum duty ratio was tried to find for line and space, contact hole patterns. The basic rule is to keep the minimum Cr width over 0.30 micrometers mask. OAI have been applied to get higher contrast of line and space, and even contact hole patterns, and achieve good pattern fidelities of island patterns. By the implementation of OAI, process latitudes were greatly improved compared to that of conventional techniques. In order to optimize the process over the actual topography, optimum numerical aperture (NA) and aperture of the OAI were selected. In conclusion, 0.30 micrometers design rule device was successfully fabricated by optimizing the advanced i-line lithographic techniques.

A KrF excimer laser using an all solid state Pulse Power Modulator (PPM) has been studied. This PPM configuration replaces the commonly used thyratron switch with a Silicon Controlled Rectifier (SCR) switch combined with a pulse compression-voltage multiplication circuit. Use of this PPM has extended the useful chamber life of a line-narrowed KrF excimer laser from 1.5 billion to 2.5 billion pulses. Broadband KrF laser performance, optimized for mirror based scanner systems, has also been investigated. A minimum broadband chamber life of 5 billion pulses has been demonstrated with this solid state PPM. While a thyratron-based PPM exhibits an expected lifetime of 3 billion pulses, the solid state PPM used in these experiments has been operated for greater than 6 billion pulses without any decrease in performance. Since 72% of the replacement parts cost for the ELS-4000D line-narrowed excimer laser is due to periodic chamber and PPM replacement, significant cost of ownership reduction is realized by extending the lifetime of the chamber and the PPM.

Advanced DUV lithography tools are adapting scanning methods in order to match the increasing demands on throughput, field size, and resolution. The demands on the laser source are changing for both the reflective and the refractive Step&Scan tools. The cost-effective operation of such lithography tool requires 248 nm excimer laser with high power and repetition rates of >= 1 kHz. Precise dose control of the exposure demands not only high repetition rates but also excellent stability of the laser output energy. The combination of the new NovaTube technology laser tube design with the LITHO line narrowing optics design allows 1 kHz operation of the excimer laser with high stability.

A feasibility study of ArF excimer laser as a light source for subquarter micron design rule lithography is presented. Partially narrowed KLES-G1A ArF laser produces 57-pm (FWHM) bandwidth pulses with an average power of 5.1 W. Narrow-band operated KLES-G1A generated 7.2- pm (FWHM) pulses with 1.4 W. Further high output power with narrow-band spectrum was achieved by using injection-locking technique. The injection-locked ArF laser system generated 0.8-pm (FWHM) bandwidth pulses with an average power of 250 W. We also point out some problems appeared in the laser operation.

Accelerated damage experiments were conducted to better predict the lifetime of fused silica optics under Deep-UV lithographic conditions. Real-time monitoring of UV-induced absorption showed that 193-nm radiation produced color centers four to five times faster that 213-nm radiation. Since the two-photon absorption coefficient at 193-nm is also about five times larger than the 213-nm value, a two-photon initiated damage process was supported. Refractive index changes in irradiated samples extrapolated from stress-induced birefringence distributions and a compaction model correlated with the refractive index changes calculated from interferometry measurements. The rates of UV-induced refractive index changes varied by over a factor of two among five different fused silica types, indicating that fused silica chemistry and processing history effect compaction susceptibility. The compaction rates, when extrapolated down to lithographic intensities, predict that the lifetimes of fused silica optical systems may be limited by UV- damage.

A new lithography technique using continuous wave (CW) 266 nm radiation from an all solid state frequency quadrupled Nd:YAG laser is described and demonstrated. This laser has proved to be a highly efficient and promising deep UV light source in fabrication of 0.25 micron design rule device. Furthermore, we obtained 0.2 micron L/S pattern with phase shift mask. Speckle free images are obtained with rotating diffuser. The performance and potential of this new laser as a light source of microlithography are discussed and compared with KrF excimer laser theoretically and experimentally.

The use of intense, broadband, ultraviolet (UV) radiation in the stabilization of photoresist requires stable UV sources and sensors, as well as suitable calibration procedures. We have constructed and measured the stability of a new UV transfer source standard with a spectrum that is identical to that of a commercially available photostabilizer. This device is particularly useful in calibrating the broadband UV detectors that monitor photostabilizer irradiance. The irradiance of this new source in a narrow band around 310 nm is monitored and found to be stable to better than 1.5% over 12 days. Spectral measurements of the source are also presented. The stability of two other commercially available UV sources is investigated and the behavior of the broadband UV sensors is discussed.

With the aim of improving output parameters and operating characteristics of ArF excimer laser intended for submicrolithography intracavity Fabry-Perot etalons able to effectively narrow and tune laser radiation were developed and tested. Used in an YYE 1702 compact excimer laser two innovative etalons permitted to achieve maximum output energy of 5 mJ at 193,3 nm, bandwidth of 20 pm and 0.5 level for 8 ns pulse duration, 20 X 5 mm output beam aperture, 100 Hz pulse repetition rate. The dispersive resonator shows high energy efficiency.

The effectiveness of two methods of optical proximity correction based on feature biasing and subresolution assisting features is compared by simulation and experiments. Parameters examined are overlapping focus- exposure windows for dense lines, semi-isolated and isolated lines, and line-end shortening. Binary and phase-shifting masks containing test and real IC design features are proximity corrected either by commercial software (in the case of feature biasing) or by manual correction using optimized size and placement of assisting features. The results indicate that, while both methods are effective in reducing optical proximity effects, the feature-assisted method is more advantageous in many cases.

One of the most dramatic effects that one encounters when attempting the optical imaging of 0.5 k ASIC gate levels is the truncation or shortening of transistor geometries. This reduces the wafer process latitude and in some cases even eliminates the level-to-level overlay margin. We investigate a number of techniques, including various complexities of mask compensation and modified illumination to mitigate this phenomenon in manners sufficiently general to accommodate ASIC layouts.

In an earlier work we demonstrated the feasibility of automated software correction for imaging and process proximity effects by precompensating mask layout data. As correction techniques and algorithms continue to be refined, it is important to deal with integrating the correction computation into the manufacturing data flow. This paper addresses issues such as computational speed, data volumes, management of the data hierarchy, and the need for geometric operations between various CAD layers within the context of a data hierarchy manager.

A mask correction algorithm (OPTIMASK) has been designed and implemented. Its main ingredients are optical proximity correction (OPC) and optical design rule checker (ODRC). The algorithm is based on the lithographic notion that a mask has to print throughout its defocus budget, taking into account multiple defocus planes. In each defocus plane the aerial image is computed using FAIM, and the design failures are reported via ODRC. The mask correction is subjected to physical restrictions that do not allow any feature couplings to occur. The union of the failures at all defocus values determines the first step taken in correcting the mask. Then a (constrained) Newton optimization scheme is applied to optimize line shrinkage, linewidth control, and corner rounding errors. All the tools needed to optimize a specific layer within a particular cell and return the optimized layer to the original mask file have been implemented. Several examples will be shown.

Practical evaluation method for optical proximity effect correction (OPC) feasibility is newly proposed using EDM (Exposure-Defocus and Mask fabrication latitude) methodology. In this method, printed image on a wafer is characterized by process latitude functions derived from EDM methodology in terms of exposure latitude, depth of focus, and mask line width latitude. In this paper, effectiveness of the process latitude functions will be verified by applying them to intraproximity effect evaluation for isolated lines and angled lines, and also to interproximity effect evaluation for T-shaped patterns and lines & spaces (L/S) patterns in i-line lithography.

The global proximity effects of densed line, semi-isolated line are studied for conventional illumination, off-axis illumination, and finally off-axis illumination in combination with attenuated phase shift masks which have transmittance of 4% and 8%, respectively, by experiments and simulations. To analyze the behavior of proximity effects, the lithographic performances of the super resolution technique are investigated comparing the cross-sectional view of resist pattern profile, useful depth of focus, and the curves of linewidth vs. defocus for 0.30 micrometers , 0.35 micrometers , and 0.40 micrometers pattern size, respectively. The global proximity effect is quantitatively analyzed by fitting the curve for densed line and isolated line to 2nd order polynomials. Off- axis illumination with attenuated phase shift mask is very effective to minimize the proximity effects for the pattern size less than 0.40 micrometers , and have useful depth of focus of 1.0 micrometers for 0.30 micrometers patterns.

A simulated annealing, one of the optimization algorithms, has been applied to the optical proximity correction. Using attenuating phase shifting mask with annular illumination, the stepper condition as well as mask design has been optimized to correct line shortening for typical cell array pattern of dynamic random access memory. A cost function has been designed to reflect the desired dose-window and depth of focus obtained by ED-tree analysis. The performances of escape from a local minimum and the convergency have been demonstrated. Then all parameters that dominated the optical proximity effect have been optimized to minimize the cost function under practical constraints.

Proximity effect correction presumes that a significant proportion of CD variations observed in wafer lithography and processes can be systematically predicted from calculations made on a pattern layout prior to fabrication. Total shape variations include a portion that repeats from chip-to-chip and wafer-to-wafer, and a randomly varying portion. Repeating effects can be compensated by modifying feature shapes on the mask pattern. The project described in this report is a study to characterize the systematic shape distortions in an experimental 0.25 micrometers process. Analyses of variation are made to quantify dependencies on specific variation sources. Two-dimensional 'behavior models,' derived from characterization data measured from processed wafers, can be used to compute shape corrections for arbitrary pattern layouts. This project was undertaken in collaboration with the Hewlett Packard Research Laboratories, Palo Alto, CA.

In automating optical proximity correction, calculation speed becomes important. In this paper we present a novel method for calculating proximity corrected features analytically. The calculation will take only several times the amount it takes to calculate intensity of one point on wafer. Therefore, the calculation will become extremely faster than conventional repetitive aerial image calculations. This method is applied to a simple periodic pattern. The simulated results show great improvement on linearity after correction and have proved the effectiveness of this analytical method.

In this paper, wafer level experimental results on the characteristics of a low-k₁ submicron lithography processes using binary mask and deep ultraviolet (DUV) wavelength (248 nm) printing tool will be discussed. In the study, we investigated the nonlinear response of wafer resist critical dimension (CD) at different pitches to the mask CD. The CD nonlinearity is due to the combination of relatively poor optical aerial image quality at low-k₁ imaging and resist processes. Operating a lithographic process at this nonlinear region presents an issue since only limited process windows are available. In addition, it requires both proximity correction for larger features on the mask and tighter mask CD control. The later requirement, in practice, is rather difficult to achieve with current mask technology. In order to avoid operating low-k₁ lithography processes at CD nonlinear region, improvement of aerial image quality is obviously needed. Although aerial image quality improvement can be achieved with different advanced lithographic enhancement techniques such as phase shift masks and off- axis illumination, there are different process issues related to each of the techniques as well. With conventional illumination source and binary mask, one of the option left to enhance aerial image quality is mask biasing. Our experimental results show that by positively biasing the mask, i.e., mask chrome CD line dimension is larger than targeted resist CD, CD nonlinearity for low-k₁ processes can be improved. For low- k₁ processes, the isofocal bias of different feature sizes (different k₁ factor) was also found to be different. A strong dependence of isofocal bias on pitch was observed. Again, mask biasing can help to shift the isofocal CD of critical dimensions to the target CD. However, in our study, we found that for certain resist, opposite mask biases are required to achieve CD linear response and zero isofocal bias. In this case, tradeoffs between mask CD control, proximity effect, and process window need to be considered.

An attenuated phase shift mask (PSM) is the most promising candidate for the high volume production lithography process among the various PSM types. It has been shown that attenuated PSM improves the lithographic performance such as depth of focus, especially in contact window by its edge enhancement. In this paper, the side lobe effect that restricts the lithographic performance of attenuated PSM and the light intensity distribution have been examined on changing the pattern density and the transmittance by experimental and simulation. The side lobe effect caused by proximity effect is very severe when pitch sizes are in the range of 0.7-0.9 micrometers for 0.35-0.45 micrometers contact hole on mask and it is enlarged by defocus exposure condition. The side lobe effect in this range of pitch size may forms the additional pattern in wafer, which restricts the application of attenuated PSM. The side lobe effect can be removed by additional pattern positioning at the center of four contact hole patterns, but simulation result of Exposure-Defocus tree (E-D tree) shows that lithographic performance of attenuated PSM is decreased by an auxiliary pattern. In the application of attenuated PSM in dense pattern, the relation between performance and side lobe effect is mutually contradictory.

In our earlier work, we have proposed an efficient algorithm for phase- shifting mask design of an arbitrary IC pattern. Experimental results have shown that the algorithm works well for patterns such as gratings, contact-holes, U-patterns, etc.. In this paper, we use the concept of moment expansion to give a rigorous proof of the well-known fact that two-phase masks such as alternating phase-shifting masks or attenuated phase-shifting masks possess a better depth of focus; thus, our proposed double-exposure masks have depth-of-focus enhancement property. In addition, we also investigate the quantization problems related to our algorithm and develop a new quantization scheme based on the so-called halftoning techniques, which is commonly used in the printing industry to render the illusion of the continuous-tone images on binary output devices such as laser printers. Simulation results have shown that the new quantization scheme performs very well even for much larger and complicated patterns. Currently, empirical verification is underway in cooperation with HP ULSI Research Laboratory, Palo Alto, CA.

The virtues of mask-plane assist features for improving imaging performance of generic ASIC layouts in the 0.5k₁ realm has been previously proclaimed. In this report we provide experimental verification and introduce a methodology to automatically deploy these features on ASIC layouts.

In previous work, Cobb and Zakhor developed an automated mask design algorithm using optimization to produce masks which can print at smaller feature sizes. In this work, we build upon our previous approach with special regard to computational efficiency and mask manufacturability to produce an Optical Proximity Correction (OPC) algorithm which operates orders of magnitude faster and produces simpler optimized masks. The algorithm can be used for OPC of Manhattan geometry masks for which phase assignment has been previously completed. Therefore, the OPC problem is divorced from phase-mask design and the two tasks are performed independently. Our algorithm decomposes the mask features into edges and corners which can be moved from their original placements to improve the image characteristics. The resulting optimization algorithm inherently requires computation of O((rho) _o³) where (rho) _o is the density of edges and corners on the mask. A major feature of the algorithm is a new fast intensity computation technique which uses lookup tables to achieve single point intensity computation on the order of O(N_a (DOT) M_r) where N_a is the order of approximation to the optical system and M_r is the number of rectangles in the mask region description. The single point intensity computation time is typically around 300 microsecond(s) ec for fairly complicated masks on a HP 700 series workstation. The resulting algorithm optimized a 36 X 36 micrometers ² test mask in 6 iterations at 11 seconds per iteration pass on the same machine. The new techniques make the described algorithm viable for a production environment with k₁ as low as k₁ equals 0.4.

The application of attenuated PSM to dense contact array is restricted by intense optical interference of sidelobe peaks around hole pattern which is called proximity effect. In order to reduce the effect of proximity, it is necessary to optimize the transmittance of mask field and the exposing energy which cause mask bias. The vast of data are required for doing above job. Using simulation, however, by sidelobe, E- D curves can not be used directly to find the depth of focus. The criterion for sidelobe was defined to find process window from E-D curve. To reduce sidelobe in dense contact hole array, the auxiliary patterns were placed between holes. By simulations and experiments, attenuated PSM with auxiliary pattern could be applied effectively to the dense contact hole of 64MDRAM.

The theoretical foundation for high NA imaging, which has been previously presented is used to examine the effects of the photoresist film on the optical performance of an advanced lithographic lens. We show that the high NA image formed by the lens within the photoresist interacts with the development process; hence, the image behavior is highly correlated to process parameters such as development and photoresist thickness. The effects of aberrations with the photoresist are explored by comparing aberrated and ideal systems. Changes in field related focal deviations and depth of focus (DOF) are shown. A general rule-of-thumb equation is formulated that relates best focus, i.e., focal plane deviation, derived from either simulations or experiments to the amount of wavefront aberrations. The relationship is extended for an approximate calculation of DOF change with aberration. Since the measurement of various amounts of aberration is dependent on the photoresist processing, definitions of best focus using testing procedures such as focus/exposure matrices, are influenced by the ability to measure optical performance.

Strategies for modifying both mask fabrication processes and design data for alternating phase-shifting masks to account for mask scattering phenomena are explored. Results were derived from the rigorous solution of Maxwell's equations using the EMFlex and TEMPEST programs for an etched-quartz fabrications process. By importing the resulting diffracted orders into VCIMAGE, full vector calculation of the aerial image from mask to wafer was obtained. From the rigorous mask simulations, the 0^th and 1^st diffracted orders were translated into an effective transmission and phase based on a thin-mask approximation. With this analysis technique, a 0.25 micrometers line-space grating for the baseline etched-quartz process (4X magnification) showed a transmission error of 7.2% and a phase error of 1.6 degree(s). In order to compensate for these errors, etch-back fabrication techniques, in which the quartz was recessed beneath the chrome, were evaluated to determine the extent to which the transmission and phase errors could be reduced. For the dual etch-back process typically in use today, a residual transmission error of approximately 0.5% could not be completely removed, even for etch-back depths greater than 200 nm. Correction of the phase errors was achieved by reducing the reactive-ion etch depth by 2-3 nm. Design manipulation, in which the 180 degree(s) opening was increased in size, required feature-dependent phase errors as large as 1 degree(s) were present.

The effects of the active area topography on the critical dimension of a polysilicon gate are examined using the rigorous electromagnetic simulation program TEMPEST and SEM pictures of developed photoresist lines. The recessed nature of the active area contributes uneven exposure and hence a nonuniform critical dimension of the polysilicon gate as it reaches the end of the active area and as it traverses over the field oxide. This gate resist formation process over the active area was simulated and patterned in experiments of 0.35 micrometers features using an i-line system with an NA of 0.5 and a sigma of 0.6. The dominate effect of linewidth narrowing near the end of the active region is shown by simulation to track the three dimensional geometrical reflections of light illuminating the bowl shaped region near the end of the active area where the polysilicon covers the field oxide transition. The addition of a TiN anti-reflective coating (ARC) layer and the addition of dye to the photoresist were investigated by studying the energy deposited into the photoresist using TEMPEST as well as SEM pictures of photoresist lines. In both experiment and TEMPEST simulation, a TiN (ARC) layer was found to alleviate the problem while dyed photoresist showed little improvement.

Partial coherent imaging in a high NA stepper is treated with the source integration method. Image formation in 3D is accomplished by the propagation and interference of plane waves. This approach allows the extensive use of FFT and leads to efficient computation of the latent image. In order to further reduce the computation time, we propose a sufficient condition for the grid density in an image plane based on the sampling theorem. Finally, we present a semi-analytical method for the modeling of post exposure bake process in 3D. With these enhancements in the algorithm, a typical 3D latent image problem can be solved in a few second on a workstation.

A comparison is made between several scalar models for the formation of an image in a photoresist film with differing approximations and a vector model. It was found that up to a numerical apertures of 0.7, a full scalar model showed no appreciable deviation from the vector mode. However, when several common assumptions were made to simplify the scalar model, significant errors resulted at a higher numerical apertures.

This paper discusses a new formulation of the Fast Multipole Method of 3D topography scattering in the physical-optics approximation. The novelty of this formulation lies in the use of generalized Sommerfeld integrals to represent the vector multipole waves. This technique allows 3D reflective-notching simulation to be done on an ordinary workstation in O(N^3/2) time, where N is the number of nodes on the surface of the topography. Simulation results for 3D topography with different slope angles are presented. The simulated reflective-notching effects were found to be consistent with a ray-optics model.

Advanced lithography techniques for the sub-half-micron range need a 'balanced' set of parameters to get the necessary enhanced resolution and process latitudes. The influence of every parameter ((lambda) , NA, (sigma) , mask type, illumination geometry) has been studied by extensive use of simulation techniques. The results are visualized in a comparison matrix. So trends can be observed and using this matrix it will be possible to decide whether or not one can use extremely high NA values in order to get high resolution and which additional methods will be necessary for compensation of the reduced focus latitude by NA increase.

Rigorous simulation has been shown to be useful for predicting complicated effects in the optical lithography process. Rigorous simulation models are necessary for accurately predicting behavior from non planar mask and substrate structures and for non-vertical light propagation. As previously shown[l], the benefits of 2D rigorous lithography simulation are many, however, these models are inherently limited. 2D models are not general enough to model contact openings, corners, 3D defect structures or off-axis illumina tion systems. These are important issues in lithography today and require true 3D simulation to help analyze them.

A method is presented for predicting the CD limited yield of a photolithographic process using well established lithography modeling tools. the lithography simulator PROLITH/2 is used to generate several multivariable process response spaces (for example, final resist critical dimension (CD) versus focus, exposure, resist thickness, etc.). Error distributions are determined for each input variable. By correlating the input error distribution with the process response space, a final CD distribution is generated. Analysis of the output distribution produces a predicted parametric CD yield. By analyzing predicted CD yield for multiple geometries, a yield curve (yield vs. feature size) is determined for an i-line process. By comparing CD yield for different nominal exposure doses, a yield-exposure curve is calculated for an i-line process.

This paper reports the simulation study of a new phase-shifting mask (PSM), quartz-etch halftone-rim, applied on one and two dimensional patterns, such as contact hole, line/space and polylayer using positive- tone resist (optical parameters focused mainly on (lambda) equals 365 nm, NA equals 0.5, sigma equals 0.6, halftone chrome T equals 4%). When apply this halftone-rim PSM on 0.35 micrometers dense contact holes, the simulations indicated that optimized rim width is 0.15 (lambda) /NA (0.11 micrometers ). Optimized single side width for biased contact is 0.1 micrometers (0.9 X optimized rim width). Optimized biased contact is 0.55 micrometers (0.35 + 0.1 + 0.1 micrometers ) which is 1.57 times larger of 0.35 micrometers design rule. Exposure latitude is 9.9%. Total depth of focus (DOF) is about 1.2 micrometers . Compared with sized-rim PSM (chrome T equals 0%), this halftone-rim PSM has lower operation exposure dose; higher exposure latitude and larger DOF. When apply it on the 0.35 micrometers dense and isolated line/space, optimized rim width is 0.12 (lambda) /NA (0.09 micrometers ). Optimized single side width for biased space is 0.08 micrometers (0.9 X optimized rim width). The total DOF is about 1.0 micrometers for line/space. By using combination of halftone-rim PSM and off-axis illumination (OAI), the total DOF will reach 1.2-1.5 micrometers for 0.35 micrometers dense line/space. However, the improvement for line/space is insignificant if compared with sized-rim PSM. Halftone-rim PSM has advantages on patterning of contact holes but not on line/space in this study.

Common uses of the term 'depth of focus' (DOF) are explored as it relates to semiconductor lithography. A definition of DOF is given which is most appropriate to photolithography for IC manufacturing and this definition is compared to other DOF metrics. In particular, simple methods for determining DOF (either experimentally or through simulation) lead to DOF-like metrics. These metrics are compared to the definition of DOF and their accuracy are evaluated. Examples of the use of the definition for DOF for studying trends in lithography are given.

Focus shift and process latitude of contact features on both dark-field and light-field attenuated phase-shifting masks and binary intensity masks were examined using experimentally measured aerial images from the Ziess MSM-100 with IBM AIMS software, the scalar and thin-mask approximation in SPLAT, and the rigorous electromagnetic simulator TEMPEST. The dark-field attenuated phase-shifting mask (aPSM) contact holes show the most severe amount of focus shift, although the shift is not much different from that of space openings. Exposure latitude of dark field aPSM features shows a 12% improvement (from 33% to 45%) over conventional binary intensity mask. Depth-of-focus is also improved. under biasing of the mask features can also improve the process latitude of dark-field masks, whereas the contrary is true for light-field mask features. In general, the process latitude of light-field contact features is worse than that of dark-field features, indicating the need for positive deep-UV photoresist technology.

When small feature delineation is considered using existing exposure tools, special techniques might be needed such as phase shift mask, oblique illumination, top surface imaging, etc. When different types of patterns exist simultaneously or island patterns exist predominantly, optical proximity effect will become more important to be controlled. In this study, six different mask types were prepared and evaluated in view of a pattern fidelity and process latitude for 256 mega bit DRAM's storage node patterns. The masks used for this experiment were conventional transmission mask, serif patterned mask, square patterned Transmittance Controlled Mask (TCM), horizontally rectangular TCM, vertically rectangular TCM, and cross patterned TCM. The cross patterned TCM had three different transmittance on it and was evaluated also. In view of both pattern fidelity and process latitude, cross-TCM showed the best result. The vert-TCM also showed fairly good result. But the worst results always came from the conventional mask. From plane surface area point of view, once serif mask or TCMs are used, the areas always improved ranging from 120% to 145% at the best focus condition compared to the convention mask. There was not so much difference among three different transmittance in view of pattern fidelity and process latitude. As one of candidates for optical proximity correction, since small serif delineation on mask level is not easy for devices with small features such as 1 giga bit DRAM or beyond, TCM is more promising which has much bigger and easily writable gray area.

We report on a new system for submicron lithography by fast laser direct writing, where a programmable phase modulating spatial light modulator (SLM) is imaged onto the wafer using flash on the fly exposure with an excimer laser light source. This new principle for maskless optical lithography has been investigated for the first time. A SLM with 512 X 464 pixels has been developed and fabricated using a CMOS active matrix and a reflective, deformable viscoelastic layer on top. Using this light modulator for image generation a demonstrator exposure tool for 0.6 micrometers minimum feature size has been designed and set up including all the components necessary for the exposure of a complete lithographic layer from CAD layout data. The demonstrator is shown to give good quality photoresist pattern on the wafer at a throughput of roughly one 4"-wafer per hour. Based on our experimental results we propose a tool with a throughput of nine 6"-wafers per hour and conclude that the new principle of operation has the potential of high performance optical direct writing lithography.

In order to check the usefulness of an alternating Phase Shift Mask (PSM), we fabricated one and applied to 1.0 micrometers pitch gate-poly pattern of a mask-ROM device. Etched quartz type was used and multiphase shifter method was adopted using a g-line positive tone photoresist. Practically, defect-free PSM could be obtained through 4 times voting method. PSM was written with CORE-2564-PSM reticle writer which showed fast writing speed and simple second layer fabrication due to free of charge dissipation layer, making voting method easily possible. We used the g-line stepper having numerical aperture (NA) of 0.54 and coherence factor ((sigma) ) of 0.5. From this experiment we obtained depth of focus (DOF) of 1.0 micrometers which was the same as that of i-line case (Na 0.45, (sigma) 0.5) and was twice wider than that of conventional g-line case. And there was no loss of critical dimension uniformity for PSM. Conventional i-line reticle and g-line alternating PSM were applied to real device fabrication. Final electrical yield analysis revealed no difference between the two cases, indicating that application of alternating PSM to device fabrication is very promising. Device characteristics of two cases were almost same in speed and low Vcc margin, too.

The application of optical enhancement techniques on high numerical aperture i-line steppers along with attenuated phase shift reticles have enabled contact processing to 0.35 micrometers and below. Previous work has shown that a high numerical aperture and low partial coherence improve contact printing capabilities. This paper focuses on a detailed evaluation of the effect of partial coherence on subhalf micron contact performance using both attenuated phase shift and binary reticles. The initial characterization of contact performance for contact sizes ranging from 0.3 micrometers to 0.45 micrometers confirms previous results that a 10% attenuated phase-shift provides the largest focus latitude when compared with a 6% attenuated reticle and a binary reticle. In extending previous work to 0.3 micrometers , it was confirmed that a larger focus latitude is achieved for a reduced partial coherence and high numerical aperture. Cross section micrographs presented confirm the improvement in profile and focus latitude by reducing the partial coherence. Cross section micrographs also demonstrate the capability with attenuated phase-shift to print 0.3 micrometers contacts at a pitch of 2.5 times the contact width. The pitch does not effect the contact focus latitude until it falls below 2.5 times the contact width. Uniformity (3(sigma) ) across the exposure field is reduced from 0.035 micrometers to 0.021 micrometers as the partial coherence is reduced from 0.6 to 0.3. The uniformity (3(sigma) ) across the wafer is reduced from 0.056 micrometers to 0.023 micrometers for the same reduction in partial coherence. Application of the best case process conditions also produced a focus latitude of 0.75 micrometers for 0.25 micrometers contacts. The results of this study show that i-line processing can be extended to 0.3 micrometers contact processing on random logic devices.

The roles of projection printer illumination and resist/substrate layers in the printability of defects are explored in a physical based approach. In this approach the mutual coherence is first calculated, and then the impact of a defect on a feature is calculated using the image perturbation approach. The concerns are that the use of modified illumination and high NA divergence and reflection of the fields in a thin-film might affect the tendency of a defect to print. Results of this extension for various illuminations, thick resists and antireflection coatings are compared with rigorous thin-film SPLAT simulations. The mutual coherence function (MCF) was found to be strongly dependent of the type of illumination and on the mask geometry; however, the MCF was found to be weakly dependent on the presence of substrate material. This indicates that certain types of illumination sources are more defect tolerant and that the substrate is insignificant in preventing defect printing.

The reduction of image intensity in a rim phase shift mask (PSM) for contact holes has been investigated by computer simulation and experimental lithography. The reduction occurs at contact hole size below 1.0(lambda) /NA. The smaller the size the severer the reduction. A new design of rim PSM called partial rim PSM, is proposed to overcome the problem. Computer simulation of aerial images has shown that the partial rim PSM can enhance image intensity. The enhancement is more significant for smaller size of contact holes while the degradation of image quality is much less than a biased rim PSM proposed previously. The partial rim PSM has been fabricated using the same self-aligned process as for conventional rim PSM fabrication. The contact holes with partial rims and with conventional rims have been printed on a wafer by a g-line stepper. The experimental result has confirmed the improvement in image intensity achieved by the partial rim PSM.

Various techniques for estimating the performance of laser reticle writing systems are examined and compared. The criterion used to measure performance is based on finding the effective maximum error E_m, which is defined as the value for which 99.73% of the error population falls within the limits +/- E_m. The commonly used technique of using 3(sigma) , where (sigma) is the standard deviation, is shown to be a poor estimate when statistical analyses based on the normal distribution do not describe the data well. Improved accuracy is obtained when corrections based on the skewness (gamma) ₁ and the coefficient of excess flatness (gamma) ₂ are included. For system qualification, direct observation of the maximum error among a set of approximately 256 samples of the error is the preferred technique. When the sample size differs significantly from this value, the use of corrections based on (gamma) ₁ and (gamma) ₂ to determine the maximum error is preferred over alternative methods due to its emphasis on the behavior in the tails of the distribution.

A process using a bottom-side antireflective coating, AZ BARLi, has been studied for 0.50 micrometers and sub-0.5 micrometers features using I-line photolithography. Significant improvements were demonstrated for such process parameters as CD swing curve ratio, exposure latitude, and reflective notching of the photoresist. Extensive characterization was done on defects observed between the BARLi and photoresist coatings, and a process developed for their elimination. Factors which had significant effects on the observed number of defects, and their distribution, were the type of photoresist coat program used, solvent treatment of the BARLi surface, and a high temperature bake after photoresist coat. Data is presented for a complete process, which includes plasma etching the BARLi antireflective coating.

The use of organic antireflection layers (ARL) for KrF laser-based DUV lithography at 0.25 micrometers groundrules has been studied. Critical performance considerations include swing curve suppression, coating conformity over topography, defectivity, and etch process control. The use of currently available ARL materials over a range of thickness values reported, and it is shown that substantial swing effects still exist at 0.25 micrometers dimensions, in agreement with theoretical prediction. ARL/resist interfacial defect are seen to be minimized with a new ARL formulation. Atomic Force Microscopy (AFM) data is presented which indicate significant ARL and resist thickness variations over product topography. Preliminary results on an improved ARL etch process and a removable ARL process based upon a uv-harden treatment are also reported. Significant improvements in materials and processes are required for a 0.25 micrometers manufacturing capable system.

We examine the effects of optical thin films on CD control in DUV lithography. A number of different substrates typical of device lithography levels were fabricated. Film stack configurations were chosen for their optical properties as well as process integration compatibility. All wafers were then exposed with an open field to measure swing amplitude, followed by a resolution test pattern to measure proximity effect and dose latitude. The resolution test pattern consisted of 250 nm coded features varying in pitch from 500 to 1750 nm. Results will be presented for optimum CD control over four types of film stacks as a function of resist thickness. In general, it is always best to operate at an extremum of the swing curve in resist thickness. However, the underlying film stack effects the CD control differently at E_min versus E_max. We present the optimum exposure conditions for a number of film stacks, resist thickness, and types of illumination (conventional and annular). These results were corroborated by optical imaging software enhanced to incorporate resist thickness and thin film effects. A synopsis of the role of substrate optical properties on CD control will be presented.

One of the biggest challenges for printing 0.6 micrometers metal lines (1.3 micrometers pitch) over 1.1 micrometers of topography is the shallow depth of focus. Total random errors, or, Built In Focus Errors (BIFE) for the class of steppers used in this study was determined to be 0.48 micrometers . When topography, and field curvature/astigmatism, are factored in, the required depth of focus to print the metal lines is approximately 2.18 micrometers . However, the 'Available' DOF (as determined from PROLITH simulations) is only 1.8 micrometers . The deficit between available DOF and required DOF grows bigger when there are multiple steppers. To overcome this deficiency, various techniques have been studied. Topography reduction by 0.4 micrometers was achieved by the use of Reverse Well technology. Effect of different resist chemistries, off-axis illumination, variation in numerical aperture and partial coherence, on the focus latitude are shown.

The manufacturing of high performance integrated circuits requires tight control of critical dimensions (CD) at poly pattern. Achieving these CD requirements forces process engineers to eliminate linewidth variation caused by reflective notching and thin film effects. Implementation of a bottom antireflective coating (BARC) can reduce or eliminate linewidth variations caused by reflective notching and thin film effects. With the additional process complexity, a number of issues must be resolved to obtain the true benefits of using a BARC material. The following paper discusses the characterization of a commercially available BARC material for 0.35 micrometers i-line poly patterning. Initial characterization of gratings and isolated lines on flat poly test wafers shows the resist process with BARC to have equal and in some cases better process capability compared to a conventional resist process. An 800 angstrom BARC film thickness on flat silicon wafers reduces the amplitude of the resist swing curve by 75%. On topography wafers an 800 angstrom BARC thickness eliminates reflective notching although for some small feature sizes, thin film effects still cause linewidth nonuniformity. Evaluations of BARC thicknesses ranging from 800 angstrom to 2,400 angstrom demonstrate the capability of the BARC to eliminate the swing effect in the resist by optimization of the BARC thickness. Cross sections of the resist profiles before and after etch demonstrate the ability for a repeatable pattern transfer. Wafer-to-wafer data and lot- to-lot data show the successful development of an etch process with a 0.03 micrometers bias. The data presented demonstrates the implementation of an optimized BARC process enhances the CD control capability at poly patterning for 0.35 micrometers i-line processing.

In this work we have investigated the photoresist thin film interference effect on different reflective substrates. Using computer simulated and experimental swing curves, the fitting parameters of an empirical formula based on the two mirror Fabry-Perot etalon are correlated to the reflectivity and resist characteristics. In particular, the phase of the swing curve which corresponds to the phase shift of the light wave at the resist/substrate interface was studied in detail. Therefore, the swing behavior on different reflective substrates (Si, Al, etc.) as well as on oxidized wafers was measured. It is postulated that this difference in phase shift is due to the different nature of the substrates, i.e. the complex refractive index in addition to the thickness. Simulations of the latent image of the photoresist predict that the axial shift of the latent image is also connected to the phase shift of the light wave at the resist/substrate interface and therefore related to the swing curve phase. To illustrate the axial shift of the latent image of the photoresist an improved cross section development technique based on a top coating before cleaving and development is proposed. Possible effects of this phase on the standing waves and on the profile at the resist/substrate interface are also examined with this improved method. Furthermore, Al substrates with a smaller grain size were used in order to explore this influence of diffuse reflectivity on the swing effect.

In the fabrication of integrated circuits, device pattern size determines electrical performance characteristics. Variations in the size of lines, trenches, contacts, straps, etc. result in electrical device performance differences. When the variations occur from chip to chip or wafer to wafer, we find that entire chips function differently (e.g. impacts clock speeds). This paper discusses within-chip variations that can produce undesirable effects which, when severe enough, can kill an entire chip because portions of it have incompatible electrical performance. The modeling and characterization of image-size variations caused by lens aberrations of 5X photolithography steppers with lens field sizes from 15 mm² to 22 mm² are also discussed.

DUV stepper technology has progressed to the point of practical implementation in the deep submicron lithography regime. A comparison of two state-of-the-art DUV lenses, a GCA Thistle and an ASML 5500/90, will be presented. Lens performance parameters of interest to lithographers, e.g., focus and exposure latitude, astigmatism, coma, focal plane deviation, dense to isolated bias, distortion and linewidth uniformity across lens, are reported. These lenses demonstrate the process latitude and resolution capability required for a production-worthy 0.35 micrometers process. In order to derive a meaningful assessment of lens performance, a rigorous lens analysis procedure is required. Using an automated metrology SEM, the Metrologix Metrostep 2002, and a statistical analysis of software package, SAS, the components of variability in lens performance resulting from the specification, measurement and data analysis methods, along with the physical lens design and manufacture, are examined. For example, wafer nonflatness, can have a major impact on focal plane deviation and useable depth-of-field (UDOF). Lens specification can also greatly influence the perceived lens performance. Differences in the description of an acceptable resist profile contributed to a difference of 0.6 micrometers in UDOF values. Several external factors must be carefully considered when performing lens evaluations. The components of variability in lens performance have been identified and methods to account for these components are described.

The phase shift focus monitor technique is based on the fact that phase shift structures which utilize a non-180 degree(s) shift exhibit asymmetric imaging which is proportional to defocus. This asymmetry results in a lateral displacement of the printed features, which can be measured quickly at a number of locations across the field and across the wafer with a registration measurement tool. The monitor provides a means of evaluating focus effects such as lens tilt, chip leveling, astigmatism, and field and wafer flatness. This paper presents a detailed investigation of the focus monitor. Experiments were performed to evaluate the effects of linewidth and exposure dose on the sensitivity of the monitor. An alternative measurement structure was also evaluated. Monitor repeatability was assessed, indicating that wafer-to-wafer variation is the largest source of repeatability error. Unexpected variation in sensitivity and calibration curve offset across the field was observed. These effects make interpretation of the focus data problematic. Potential factors which might contribute to the variability of the monitor were analyzed. Partial coherence uniformity appears to have a significant impact on the monitor, but does not entirely explain the observed effects. Potential use of the monitor as a coherence mapping tool is proposed.

A new latent image metrology technique is discussed that determines best focus with a precision of (sigma) equals 20 nm. This technique uses the existing alignment system of an ASM-L wafer stepper and requires no hardware or software modifications. The user just needs a standard chrome reticle. It can operate for machine setup at the factory, but also in-process for fully automatic self calibration of focus and tilt. A typical measurement takes a few minutes.

An innovative method of determining best focus with an optical exposure tool has been developed. The method uses wedge shaped marks in photoresist that can be measured automatically. The results show that best focus can be measured with a repeatability of 20 nm. The automatic focus measurement system can be used to characterize lens astigmatism, field curvature and tilt. Data shows good correlation with conventional methods using SEM linewidth measurement.

In order to meet industrial requirements for 0.35 micrometers lithography, we have developed a new i-line lens with a high numerical aperture (0.63 NA) and wide image field (31.1 mm diameter). This paper described the image performance of this new i-line lens and the possibility of achieving 0.35 micrometers and 0.3 micrometers lithography with i-line. It is extremely important that the performance of a new generation i-line lens for 0.35 micrometers lithography be confirmed, not only with the conventional illumination technique but with other illumination such as modified illumination or with the phase shift reticle. This new i-line lens has been developed to show the smallest aberration for both its design and manufacturing. As a result of this effort, a total deviation of the field curvature is 0.15 micrometers or less and distortion deviation is 0.03 micrometers or less when the illumination condition is changed. In addition, another experiment for optimizing both NA and sigma was done with 0.35 micrometers critical dimension (CD). It has been confirmed that the optimum NA is 0.55 in order to maximize depth of focus (DOF) for 0.35 micrometers and the optimum Sigma is 0.65 from the view point of the smallest CD bias between isolated and dense features.

An in-situ process has been developed to optimize lens element alignment on a wafer stepper. This technique evaluates and analyzed the optical wavefront using the photoresist line and space profile. The advantage of this technique is that the photoresist sensitivity to the optical wavefront of the lens is automatically considered in the optimization process. The technique uses theoretical prediction and empirical critical dimension (CD) data to adjust the projection lens for optimum performance on the stepper. Successful implementation in production has been achieved. While the procedure has been developed on a high efficiency 1X projection lens, it can also be implemented on a reduction lens.

A custom-built phase shifting interferometer is described for measuring the wavefront quality of Ultratech Stepper 1X projection lenses over a field of 25 mm X 50 mm using wavelengths of 363.8 nm (for i-line systems) and 442.0 nm (for gh-systems). Over an N.A. of 0.42 the wavefront is determined at 240 X 240 points with a calibrated accuracy of (lambda) /20 peak-to-valley (at (lambda) equals 363.8 nm). The interferometer is an unequal path Twyman-Green interferometer with an Argon-laser and a HeCd-laser as light sources. The interferometer test system is integrated on a 10 ft X 4 ft vibration isolation table together with an xy-stage to position the projection lens at different field points. The safe handling of the projection lens is supported by a load/unload mechanism on the table. The measurement wavefront maps at various field points of the tested projection lens are used to optimize the state of alignment of the projection lens.

Practical resolution, which is defined as the minimum geometry for a 1.0 micrometers depth of focus, in conventional krypton fluoride (KrF) excimer laser (248 nm) lithography is 0.30 micrometers . A new illumination technique, which uses a weak quadruple effect to enhance the depth of focus and to solve the current problems in the off-axis illumination techniques, has been developed. This new illumination technique is suitable for use with the attenuated phase shifting mask. With this combination technique, a 1.8 micrometers depth of focus using a 0.45 NA KrF excimer laser stepper can be achieved without a secondary peak in the distribution of light intensity for the various duties 0.30 micrometers space patterns. Even for hole patterns, a 2.0 micrometers common depth of focus can be achieved. These results indicate that KrF excimer laser lithography is a powerful candidate for beyond 0.25 micrometers -rule devices. It is also confirmed that i-line lithography is an expectable candidate for the second generation of 0.35 micrometers -rule devices.

Since island patterns are subject to optical proximity effect, it is not easy to maintain island patterns as good image as same sized periodic lines with a conventional lithographic method such as a chrome transmission mask and on-axis illumination. Five different approaches, including a conventional mask with a conventional illumination, an alternating PSM, a half-tone PSM, a quadruple illumination, and a quadruple illumination with a half-tone PSM, were evaluated to compare pattern fidelity, meaning overlay tolerance, as well as process latitude. Conventional critical dimension (CD) measurement proved not to be a proper method as the criteria of the pattern fidelity. It is found that measurement of the shortest length of the island pattern, regardless of measurement orientation, is more meaningful. Under the condition that the dot size should be as small as possible without generating of any bridging between dots, the alternating PSM showed the widest process latitude. None of the methods except for the alternating PSM showed better result than the conventional method. From this study it is found that only the alternating PSM can substitute the conventional technique.

Subhalf micron lithography suffers severely from optical proximity correction problem. Although the feature sizes are the same on the mask, as the pitch changes (different packing density), the critical dimension (CD) on wafers will change as well. Therefore, different packing densities of line and space will require different exposure energies to resolve individual patterns to their nominal width. This optical proximity effect is difficult to be corrected effectively and drastically reduces the focus process latitudes. Strong phase shifter mask includes sized rim and chromeless shifters and it can compensate for optical proximity correction without sacrificing defocus process latitude. Strong phase shifter mask, applied on DUV lithography (NA equals 0.53) to pattern 0.25 micrometers line widths with various pitch sizes, can deliver a depth of focus of 1.50 micrometers . If a binary mask is used to print 0.25 micrometers of equal line and space pattern, the depth of focus is reduced to 0.80 micrometers . The focus latitude can thus be improved by almost 100% with strong phase shifter mask. Similarly, strong phase shifter mask, applied on I-line lithography (NA equals 0.54) to pattern 0.30 micrometers of line and space with various pitch sizes, can produce 1.20 micrometers of depth of focus. Using the same I-line lithography process with a binary mask, 0.45 micrometers of equal line and space patterns can be printed with 1.10 micrometers depth of focus with no 0.30 micrometers features resolved. The experimental results of strong phase shifter mask applicable to both 0.25 micrometers and 0.35 micrometers technology will be described in detail.

In attenuated phase-shift mask technique for window pattern formation, the amount of mask bias has to be optimized, not only to obtain wider focus margin but also to avoid resist film thickness loss at side-lobe position. In general, larger mask bias is necessary to preclude side- lobe printing, but depth of focus decreases with increasing mask bias. In this paper, surface insoluble layer was applied to prevent this side- lobe printing, instead of larger mask bias addition. At first, the pattern formation capability of 0.35 micrometers window was investigated with an attenuated phase-shift mask of 8% transmittance in i-line lithography (NA equals 0.6, (sigma) equals 0.3, i-line stepper; 1 micrometers thick novolac type resist). Surface insoluble layer was formed by alkaline developer (2.38 wt% tetramethylammonium hydroxide TMAH) treatment followed by water rinse before i-line exposure. After more than 1 min treatment, deep resist film dimple due to side lobe was suppressed almost perfectly, even in small mask bias (<EQ 0.05 micrometers ) condition. As a result, the depth of focus of 0.35 micrometers window increased to 2.0 micrometers (mask bias, 0.05 micrometers ) from 1.4 micrometers (mask bias, 0.1 micrometers ). Next, this technique was also applied to KrF excimer laser lithography to improve the process margin of 0.25 micrometers window formation. A 5% transmittance attenuated phase-shift mask and the combination of KrF excimer laser stepper with 0.5 NA and 0.3 (sigma) and 0.7 micrometers thick chemically amplified positive resist were used. It was found that surface insoluble layer can be formed by the same TMAH alkaline treatment in KrF chemical amplified resist. As a result, the focus margin off 0.25 micrometers window was improved to 2.0 approximately equals 2.5 micrometers (mask bias, 0.0 micrometers ) from 1.5 approximately equals 2.0 micrometers (mask bias, 0.03 micrometers ). In conclusion, we can select more suitable mask bias by means of this surface insoluble layer formation technique, indicting that wider (approximately equals 2.0 micrometers ) depth of focus of window pattern is achieved in both i-line and KrF excimer laser lithography.

A new method for fabricating fine periodic patterns with a large depth of focus is proposed. It is based on flattening the 1st-order intensity distribution and extracting the 2nd-order intensity distribution which has a frequency twice that of the mask pattern. The resolution limit is improved up to the cutoff frequency of the projection system and the image can be created independently of the defocus and the pattern direction. This method is achieved by superposing the images at two focal planes under coherent illumination and it was demonstrated by KrF lithography with a conventional binary mask. A 0.14 micrometers lines-and- spaces pattern was fabricated with a large depth of focus of 2.0 micrometers . A contrast enhancement technique and pattern edge correction method needed for actual application of this method in device pattern fabrication are also discussed.

This paper reports recent progress in achieving subhalf-micron feature sizes with UV laser illumination based on a novel interferometric phase shifting (IPS) technique. In the IPS arrangement, the intensity and amount of phase shift of the shifted beam can be controlled continuously and independently using the same mask. Consequently the method can be considered as a convenient general testbed for practical phase shifting concepts such as strong, weak and attenuated phase shifting. Recent measurements of the lithographic performance of a new concept are reported where phase shifting is combined with off-axis illumination. Experimental as well as simulation data are used to demonstrate this new method. A lithography simulator, Depict from Technology Modeling Associates, Inc. and a related Integrated CAD Framework which is being developed at Rice University was used to simulate and evaluate the performance of the IPS scheme.

We have tried to find optimum combinations of source, mask and pupil filter suitable for all kinds of mask pattern with critical dimension (CD) of 0.18 micrometers by aerial image simulation. The annular aperture radii of deep UV source (wavelength 193 nm) are controlled to give optimum annular illumination for equal line/space pattern. The optimized illumination aperture has inner radius of (sigma) _in equals 0.525 and outer radius of (sigma) _out equals 0.600. Under this illumination, various transmittances of halftone mask are investigated as mask variation and phase distribution of pupil filter is modulated annularly. The aerial image characteristics are compared among eight combinations of illumination, mask, and filter in order to find the most effective combination for each typical mask pattern such as equal line/space, isolated line, isolated space and contact hole. The effective method is annular illumination for line/space, pupil filter for the other patterns. The combination of annular illumination + binary intensity mask + pupil filter is recommended for simultaneous printing of all kinds of pattern.

An optical lithography technique for the formation of T-shaped gate metal with narrow (sub 0.25 micrometers ) channel length, was studied. A special photomask known as the dummy diffraction mask was used to form a T-shaped resist profile by single exposure and development process steps. The sub 0.25 micrometers T-shaped gate metal could be easily obtained with i-line stepper (N.A. 0.4) under ordinary lithographic process conditions and the process margins were found to be enough for the application of device fabrication.

Three types of dies on the wafer including dies on wafer edge, dies adjacent to testkey patterns, and dies with asymmetric chip layout have been identified for mistilting on the lithography exposure processes. There are several approaches i.e. dummy die exposed on the wafer edge, various focus offset, and pattern layout modification etc. can alleviate this problem. However, to complete solve these issues rely on the fundamental and methodical improvements on the stepper leveling system and control software.

We have studied systematic line width (CD) errors as functions of field coordinates for three late-model i-line steppers from different manufacturers by measuring electrical resistance of lines patterned in poly-silicon. The combination of reticle errors with non-linear imaging accounts for a significant fraction of the total line width errors. After removing the effects of reticle errors, CD contour maps are consistent with aberration patterns in which the Strehl intensity is highest at the center of the field.

Defect printability was investigated to assess the reticle defect size tolerance for 0.50 micrometers wafer lithography through the study of programmed reticle defect within line/space pairs. Improvement in the recent generation of steppers with higher numerical aperture and the drive toward design shrinks and 0.35 micrometers wafer technology has made the printability of smaller defects more important. We have studied the printability of opaque and thin flat transparent defects with reticle fabricated with both types of defects ranging in size from 0.3 to 3.0 micrometers . These defects are located on edge and between 2.5 micrometers line/space pairs. Using a 0.60 NA i-line stepper, wafers were printed with four illumination methods: standard with sigma of 0.60, quadrapole, annular, and higher coherence with sigma of 0.30 and the results were observed on a SEM. In addition, the reticle defects were characterized and the measured sizes simulated for further validate the observed results. Transparent defects larger than 0.5 micrometers showed greater printability than their opaque counterparts but had similar printability at the smaller sizes for all illumination methods. Opaque center defects had more influence on CD (critical dimension) loss than those on edge for standard and quadrapole illuminations but were comparable for annular and higher coherence illuminations. With a 10% loss of CD as the tolerance, defect sizes of 0.4 micrometers for opaque and 0.35 micrometers for transparent were determined to be the maximum acceptable for printability of 0.5 micrometers line/space at the wafer plane.

SPARTA has developed a novel dual wavelength interferometer which is directly aimed at stepper stage control and general IC metrology. It utilizes an existing HeNe based dual frequency interferometric system and couples it directly to a second system which provides the real time air turbulence compensation. The optical design provides for colinear optical beam paths for the two systems over the measurement path. Compensated position measurements are provided at a rate of 30 Hz which is sufficient to permit high throughput stage positioning for all modern steppers. Stage position accuracy is determined to 4 nm 3(sigma) and stage precision (which is a two pass operation) can be 5.6 nm 3(sigma) . Future improvements in performance can be expected since the present design is not near fundamental limits. The interferometric system design has a form factor compatible with existing stepper systems. Testing of this system has been carried out in a laboratory environment under a variety of conditions, including those which would simulate a clean room environment. Test results are used in detail and conclusions will be presented which define the impact this system can have on stepper overlay performance and IC metrology. Because the interferometer is the basic ruler upon which much of the stepper metrology, setup, and operation is based, an improvement in performance of this system provides numerous benefits in the areas of stage precision, alignment, lens metrology, and reticle qualification and fabrication. Furthermore, the elimination of air turbulence as a stepper design concern can permit improvements in stepper performance and throughput with fewer engineering compromises.

We found total overlay with respect to optical lithography using an approach similar to quality control technique employed at a semiconductor factory. This approach involves an aligner performance, process quality, reticle error and overlay measurement. This paper further describes new ides for the number of machines to be used for matching and data collection period. Lastly, improvement on total overlay and a prospective view for a future aligner and its usage are also described.

Mix-and-match lithography continues to gain acceptance as a valuable strategy for reducing capital costs and increasing throughput productivity in semiconductor manufacturing. The successful implementation of mix-and-match lithography requires consideration of the unique characteristics of all systems being matched. Among these issues are alignment target placement and alignment strategy. The alignment system for each stepper manufacturer uses specially designed targets for wafer alignment. However, the wafer area available for dedicated alignment targets is typically restricted to maximize the quantity of production die per wafer. One approach to remove the target area limitation is to implement an alignment system based on pattern recognition techniques.

In this paper, an alignment method with oblique illumination and oblique detection named ZBAS (Zero Base-line Alignment System) is proposed and its characteristics and applications are analyzed. It detects the position error from alignment mark at the exposure position even though it is a non-TTL method. The alignment mark, a diffraction pattern with a 4 micrometers width (8 micrometers pitch), is illuminated with a He-Ne laser beam at an incident angle of 75 degree(s) and the first order diffracted beam at this is detected as an alignment signal at a diffraction angle of 60 degree(s). The peak value of the alignment signal with appropriate electrical system is about 20 mV and the signal sensitivity to the wafer position is 0.5 mV/0.1 micrometers . The theoretical analysis of ZBAS was carried out with respect to the effects of pattern depth and resist thickness. The result shows the stable detection is realized by a 2- wavelength (670,780 nm) illumination, in which the variation of the calculated signal is 27% in the range of 0.6 approximately equals 1.2 micrometers of pattern depth, 37% in the range of 0.6 approximately equals 1.2 micrometers of resist thickness. In conclusion, the ZBAS was verified as a potential alignment system in excimer laser stepper, which has a smaller base-line error than the conventional non-TTL method.

This paper presents an automatic, accurate mask aligner based on modified moire technique. In this technique the alignment marks are in the form of gratings. The high slope region of moire signal is used to obtain higher sensitivity and better position control accuracy. Automatic alignment is achieved by using difference of moire signal and its inverted signal obtained by computer. Accuracy for alignment, under the present experimental conditions, is of the order of +/- 0.06 micrometers . Our alignment technique by virtue of its higher alignment accuracy is suitable to X-ray lithography. However, due to nonavailability of a X- ray source the lithography was performed by UV source. Nevertheless, the effort is worth it so as to prove the workability of this technique. The overlay accuracy is estimated to be 0.8 micrometers which is limited by the diffraction effects of the exposure optics.

We present a methodology for predicting photoresist linewidth distributions, and for quantifying the contributions of the key sources of variation. We apply it to evaluate the manufacturability of photolithographic processes under development. The methodology has the advantage of predicting distributions across the entire reticle field. When technology requirements are too stringent, this approach quantifies which elements of the process need to be more tightly controlled. We apply these methods to a 0.5 micron technology and to two 0.35 micron technologies; the latter requires a novel calibration strategy. For production i-line lithography of 0.35 micron features, our results predict that improved exposure dose control is needed.

The focus monitor technique has been shown to be a unique and promising method of characterizing lithography tools. The focus monitor test mask employs phase shifting to translate focus error into a measurable overlay error, independent of exposure. We have also employed an exposure monitor structure to measure dose, independent of focus. The combination of focus and exposure monitors is highly sensitive to factors which affect critical dimensions. Using this test mask we have systematically analyzed the performance of the photo processing tools on our line. This paper will review the data and discuss process capability improvements using both methodologies.

A study of linewidth variations across commercial stepper exposure fields suggests that variations in local value of partial coherence across the exposure field might be responsible for the excessive character of linewidth variance. A methodology to estimate the local value of partial coherence (LVPC) at a random point in the stepper exposure field is proposed and discussed. Estimation of the local value of the partial coherence summarized across the tested exposure field is represented as a 'Coherence map' of the stepper field. For a stepper under test, such a map revealed significant deviation in the local values of the partial coherence from the designed value. Estimated LVPC variation fully accounted for 'excessive' across field linewidth variation as well as for its orientational dependency. Several means of improving line width control in the presence of variance in LVPC are suggested. The impact of the presence of localized partial coherence in the exposure systems used with PSM and OPC technologies is analyzed briefly. Sensitivity to LVPC variations with the suggested monitor design and ways to improve t are also discussed.

The definition of"Moore's Law" has come to refer to almost anything related to the semiconductor industry that when plotted on semi-log paper approximates a straight line. I hesitate to review it's origins and by doing so restrict it's definition. However, today I will review the history and past performance relative to predictions and show where the advances have come from. I will leave the ftiture performance up to you. Certainly continuing on the same slope doesn't get any easier. It presents a difficult challenge to the industry. The original paper that postulated the first version ofthe "law" was an article I wrote for the 35th anniversary issue ofElectronics Magazine in 1965. My assignment was to predict what was going to happen in the semiconductor components industry over the next ten years --to 1975 . In 1965 the integrated circuit was only a few years old and in many cases was not very well accepted. There was still a large contingent in the user community who wanted to design their own circuits and who considered the job of the semiconductor industry to be to supply them with transistors and diodes so they could get on with theirjobs. I was trying to emphasize the fact that integrated circuits really did have an important role to play.

ULSI has heavily depended on develops in optical lithography. However, optical lithography is now facing a major obstacle due to exposure wavelength limitations. To overcome this obstacle, not only the use of shorter wavelengths, but also such new technologies as super-resolution techniques, electron beams, and X-ray lithography are being intensively investigated. This paper reviews recent developments in these technologies and discusses the major issues. The difference in lithographic activities between Japan and the U.S. is also discussed. Finally, recent developments in lithography for experimental 1-Gb DRAMs are presented.

In 1994, The National Technology Roadmap for Semiconductors was revised, updated to reflect the latest information, and published in December 1994. Lithography continues to be a cornerstone for semiconductor process and productivity. The roadmap activity is sponsored by the SIA. The Technical Working Group (TWG), which does the actual work of updating the roadmap, is comprised of members from SEMATECH, SEMATECH FTABs, SRC, industry, universities and government. One of SEMATECH's key goals is to insure that the activities we pursue are well-aligned to the roadmap. The roadmap is focused on requirements and needs of the industry. SEMATECH is focused on solutions. The Lithography roadmap makes clear the desire of semiconductor manufacturers to stay with optics as long as possible since shifts to nonoptical technology are viewed as much higher investment and risk. It is perceived that optics can be a viable manufacturing solution through at least the 0.18 micrometers generation. The TWG and SEMATECH FTAB meetings highlighted the need for greatly increased emphasis and resources on 193 nm technology to bring it to pilot production in 1998 for the 0.18 micrometers generation. SEMATECH has since worked with the industry to define more detailed needs and requirements for a successful 193 program and is working with industry and the government to insure that the technology is available when needed in manufacturing. The needs are clear...cost-effective lithography for each generation. The challenges are...how to get there from here.