Search

The authors describe a plasma ashing system where a stationary jet of hot, activated gases removes photoresist from a scanning wafer. The jet is created by a reactant stream flowing through a 2.45 GHz surface wave discharge in a 6 mm quartz tube. For O2/N2 plasmas in the medium pressure range from 20 to 100 Torr, a luminous plasma jet emerges from the end of the discharge tube that transports both heat and reactive species to the wafer. A single scan results in a Gaussian track profile with a standard deviation of 7 mm for the source-to-substrate distance of 9 mm. A simple model of the ashing process, which assumes a thermally activated ash rate and Gaussian distributions for both power density and reactant flux, unifies the dependence of effective ash rate on the substrate temperature and scan speed at a constant power. The best fit activation energy at 2.5 kW is 0.23 eV, about half of the value found in conventional downstream ashing, implying that diffusion plays a significant role in limiting the ash rate. The peak thermal power density in a 2.5 kW jet at 80 Torr is 160 W/cm2, resulting in an effective instantaneous ash rate of 2.5 mm/min for a scan speed of 70 cm/s and 200 °C chuck temperature. This implies that the time to clear a 1.2 μm thick resist coating from a 300 mm wafer is 18 s. [ABSTRACT FROM AUTHOR]

As the quality of EUV-wavelength mask inspection microscopes improves over time, the image properties and intensity profiles of reflected light can be evaluated in ever-greater detail. The SEMATECH Berkeley Actinic Inspection Tool (AIT) is one such microscope, featuring mask resolution values that match or exceed those available through lithographic printing in current photoresists. In order to evaluate the defect detection sensitivity of the AIT for dense line patterns on typical masks, the authors study the linewidth roughness (LWR) on two masks, as measured in the EUV images. They report the through-focus and pitch dependence of contrast, image log slope, linewidth, and LWR. The AIT currently reaches LWR 3σ values close to 9 nm for 175 nm half-pitch lines. This value is below 10% linewidth for nearly all lines routinely measured in the AIT. Evidence suggests that this lower level may arise from the mask’s inherent pattern roughness. While the sensitivity limit of the AIT has not yet been established, it is clear that the AIT has the required sensitivity to detect defects that cause 10% linewidth changes in line sizes of 125 nm and larger. [ABSTRACT FROM AUTHOR]

Microfield exposure tools (METs) have and continue to play a dominant role in the development of extreme ultraviolet resists and masks. One of these tools is the SEMATECH Berkeley 0.3 numerical aperture (NA) MET. Here, the authors investigate the possibilities and limitations of using the 0.3 NA MET for sub-22-nm half-pitch development. They consider mask resolution limitations and present a method unique to the centrally obscured MET, allowing mask patterning resolution limitations to be overcome. The method, however, comes at the cost of increased sensitivity to mask surface roughness. They also explore projection optics resolution limits and describe various illumination schemes allowing resolution enhancement. At 0.3 NA, the 0.5k1 factor resolution limit is 22.5 nm, meaning that conventional illumination is of limited utility for sub-22-nm development. In general, resolution enhancing illumination encompasses increased coherence. They study the effect of this increased coherence on line-edge roughness (LER), which, along with resolution, is another crucial factor in sub-22-nm resist development. Due to coherence induced LER limitations, addressing the development at 16 nm half pitch and beyond will ultimately require higher NA systems. [ABSTRACT FROM AUTHOR]

Periodic structures, such as gratings and grids, are required in a variety of applications including spectroscopy, photonic and phononic devices, and as substrates for basic studies in materials science. Interference lithography readily forms periodic patterns in photoresist, but conventional approaches, using a Lloyd’s mirror or Mach–Zehnder configuration, suffer from a number of shortcomings including difficulty in aligning patterns with respect to pre-existing structures on a substrate and difficulty in precisely repeating a given spatial period. Coherent diffraction lithography (CDL), a mask-based approach, utilizes the well-known Talbot effect to accurately replicate the one- or two-dimentional pattern on a mask by reimaging the mask pattern in photoresist. Moreover, with appropriate alignment marks on the mask, one can align the replicated pattern relative to pre-existing patterns on the substrate. The authors describe the design, construction, and utilization of a dedicated CDL apparatus that permits replication, at a well-defined mask-substrate gap, of the periodic structure of a phase mask. The system also incorporates interferometric-spatial-phase imaging for aligning the replicated pattern relative to fixed fiducials on a substrate. They obtained high quality replications of a mask pattern, consisting of a 600 nm period grating, from the 1st to the 52nd plane of reimaging, i.e., from 1.55 to 40.16 μm. [ABSTRACT FROM AUTHOR]

One method currently being employed to reduce the overall lithography process complexity and cost is the utilization of a topcoatless photoresist. The development of these materials administers an additive to create the same hydrophobic characteristics as those created by advanced topcoats. The main challenge for topcoatless resists is to increase the hydrophobicity without causing too much inhibition at the resist surface which leads to bridging or residue-type defects. The key to such a design lies in creating a balance between leaching control and dissolution characteristics of the resist without degrading lithography performance and increasing defectivity. The addition of these hydrophobic additives into existing ArF photoresist systems has been shown to increase both receding contact angle and advancing contact angle in water-based immersion lithography. In this work, the authors have demonstrated that the defectivity levels of topcoatless resist are equal to or better than the industry standard of topcoat systems. This was achieved by optimizing process conditions. This article will report the influences of the develop process, postcoat apply bake/postexposure bake (PAB/PEB) temperatures, and immersion specific rinses on defect performance of topcoatless resists. It was found that pattern collapse defects along nonexposed regions for some topcoatless material can be drastically reduced by PAB/PEB temperature optimization. Complete elimination of missing pattern defects achieved by use of de-ionized water immersion specific rinses for all materials tested. Finally, the impact of the PAB temperature on surface properties will be investigated. [ABSTRACT FROM AUTHOR]

In modern photolithography, model based optical proximity correction (MBOPC) has evolved from a nice-to-have feature to a must-have feature and has been widely adopted to improve the process throughput. The purpose of MBOPC is to adjust the designed pattern on the photomask to introduce mask perturbations such that the layout printed on the wafer is as close as possible to the drawn layout. In regular MBOPC process, the polygons are dissected into segments and corrections are conducted segment by segment. While it is relatively easy and straightforward to find the optimal segment size for most one-dimensional features, it is hard for many two-dimensional features, especially line-end areas. By solving a few segments around line-end areas together, multisegment solver (MSS) showed a good contour match. However, this method is expensive on model calculation, as there is more than one segment size to be optimized and a large number of searching iterations is needed, which makes its runtime sensitive to the number kernels retained in OPC model. In this study, a novel method of computing the signal change in the MBOPC process with MSS is proposed by analyzing the optical kernel sensitivity to mask perturbation. The experimental results show that while the correction accuracy of the MBOPC process with a simplified model is maintained, the MBOPC runtime is reduced by 37%. [ABSTRACT FROM AUTHOR]

The use of the tetrapropylammonium hydroxide (TPAH) and tetrabutylammonium hydroxide (TBAH) aqueous developer solutions is proposed as an alternative to the tetramethylammonium hydroxide (TMAH) aqueous developer solution (semiconductor industry standard). A polyhydroxystyrene-based extreme ultraviolet (EUV) resist was utilized at a film thickness of 60 nm. To confirm the effectiveness of these alternative developer solutions in improving linewidth roughness (LWR) performance, patterning exposures were carried out using the small field exposure tool with annular (σouter0.7/σinner0.3) illumination conditions. Dissolution contrast curves of EUV resist using the TMAH, TPAH, and TBAH developer solutions have shown similar dissolution characteristics which means that the use of these alternative developer solutions might have minimal effect on the resist resolution limit and sensitivity. Imaging performance analysis results have shown negligible effect on the resolution capability and sensitivity. A 20% LWR improvement was observed with the TBAH developer solution [7.2 nm at 32 nm 1:1 line and space (L/S)] comparing with the TPAH and standard TMAH developer solutions (both at 9.0 nm at 32 nm 1:1 L/S). This improvement was perceived to be the result of the significant increase in dissolution contrast (higher Rmax and larger slope) with the use of the TBAH developer solution. [ABSTRACT FROM AUTHOR]

Sphere-forming polystyrene-block-poly(t-butyl acrylate) (PS-b-PtBA) diblock copolymer with catalytic amounts of photo-acid generator (PAG) formulated a pixelated photoresist. In thin films with single-sphere thickness, hexagonal arrays of spheres (∼20 nm diameter on a 40 nm pitch) of PS within a matrix of PAG segregated in PtBA was obtained through solvent annealing. Upon exposure and post-exposure baking, the soluble PtBA matrix was converted to insoluble poly(acrylic acid), such that a negative pattern could be formed in the chlorobenzene developer. The concept of pixelation was demonstrated by exposing line and space patterns with increasing widths. In contrast to the width of the exposure fields that increased monotonically, the widths of the pixelated resist structures after development were quantized with respect to an integer number of rows of spheres. Furthermore, line edge roughness could be correlated with the size of each pixel (diameter of spherical domain). [ABSTRACT FROM AUTHOR]

Ashing rate and nonuniformity are the usual parameters used for the evaluation of industrial ashers. A detailed study of the wafer temperature and of its effect on the two contributions to the diminution of photoresist thickness, namely, shrinkage and chemical reactions, gives a solid ground for understanding the variations of ashing rate with process conditions. It is in particular shown how variations of shrink with temperature affect the measured ashing rate. This analysis serves as a basis for proposing tests that could be used as “standards” and ease the objective comparison of industrial ashers, based on real performances. They include a determination of ashing rate on hard-baked wafers for characterization of radical density, a demonstration of the ion-free nature of the process, and a determination of wafer heat-up rate. Due to the very reproducible nature of the distribution of ashing rates over a wafer, it is argued that the “range” parameter should be used over standard deviation for the measurement of nonuniformity within a wafer. [ABSTRACT FROM AUTHOR]

The roughening mechanism of ArF photoresist during etching was investigated to find out why CF3I gas reduces the line edge roughness (LER) in the photoresist pattern better than CF4 gas. Since the plasma of reactive ion etching (RIE) consists of ultraviolet (UV) photons, radicals, and ions, the authors used a UV lamp and a neutral beam source for evaluating the effect of different plasma compositions on the photoresist roughness. The roughness was found not to increase only by UV photons or F radicals, but increase under the CF4 RIE plasma which has both UV photons and F radicals. A C–F modified layer was generated on the resist surface because the UV damaged C==O bonds reacted with F radicals and the resist surface became softer and shrank. Since CF3I plasma has a lower UV intensity and fewer F radicals compared with CF4 plasma, the shrinkage on the sidewall of the photoresist was suppressed and resulted in a smaller LER when this plasma was used. [ABSTRACT FROM AUTHOR]

In this paper a continuous approximation to the critical ionization (CI) model has been derived and shown to match the CI model extremely closely. Further, both of these models were shown to match the widely used Mack model of dissolution over the lithographically significant ranges of development rates. The variance of the dissolution rate is shown to arise from the variation in the development path required to bypass randomly insoluble polymer molecules rather than just the variance in the polymer solubility itself. Percolation theory, where the percolation probability is equal to the probability that a polymer molecule will become soluble, has the potential for providing the theoretical framework required to determine this variance in dissolution path. [ABSTRACT FROM AUTHOR]

The authors report a study of the impact of surface chemical factors on etch rates along with surface and line edge roughness introduction for prototypical photoresist (PR) materials and structures during plasma-based pattern transfer employing fluorocarbon (FC) discharges. For selected photoresist materials and model polymers (193 nm PR, 248 nm, PR, and polymethyladamantyl methacrylate), the influence of bulk polymer properties on plasma durability was clarified by comparing etch rates, surface roughness introduction, and profile evolution of nanostructures. They studied the effects of both fluorocarbon fragment deposition and polymer surface fluorination by gas phase fluorine atoms on plasma etching resistance and surface roughness evolution of the organic materials by comparing discharges fed with C4F8/Ar or CF4/Ar/H2 gas mixtures. The spatial frequency distribution of surface roughness was obtained using fast Fourier transformation of atomic force microscopy data. A graphitic layer was formed for Ar containing discharges on the polymer surfaces. Fluorocarbon deposition on the damaged photoresist affected roughening in two opposing ways: ion-induced mixing with the damaged polymer layer increased surface roughening, whereas for simple FC precursor deposition a reduction in surface roughness was seen. The latter effect was especially important during profile evolution of three-dimensional structures. Fluorination of the photoresist surfaces by fluorine-rich plasma increased polymer etching yields, and for highly fluorinated surfaces inhibited the formation of the graphitic surface layer. The destruction of the adamantyl structure is usually found in fluorocarbon/argon discharges and is a major origin of roughness evolution for 193 nm PR materials. Process conditions having high etch yields were found to improve the roughness results of 193 nm PR after etch. The fluorination of the photoresist materials prevented the formation of characteristic small scale roughness features at the cost of large scale roughness introduction. Use of low energy density process conditions prevented the introduction of large scale roughness and can be explained by suppression of surface roughness growth by an ion-induced transfer mechanism. Alternatively, CF4/H2 processing also showed improved roughness results due to a separate layer deposition of the fluorocarbon film on top of the photoresist material. The etch results for all process conditions can be combined in a surface roughening model where the roughening behavior of the film scales linearly with the energy density delivered to the polymer surface during processing. Even for a range of feedgas chemistries, adamantyl-containing polymers show enhanced roughening rates, suggesting that the instability of the adamantyl structure used in 193 nm PR polymers is the performance limiting factor for processing these PR materials. [ABSTRACT FROM AUTHOR]

In this work, a low-cost x-ray micromask is developed by sputtering a lead (Pb) film on a Mylar sheet substrate through microshadow masks and the x-ray mask is experimented for patterning an SU-8 negative photoresist on a stainless steel substrate. In addition, the required Pb layer thickness as a function of SU-8 photoresist thickness is studied and compared to that of conventional gold x-ray mask. The Pb layer and SU-8 photoresist thicknesses are varied from 8 to 19 μm and from 150 to 350 μm, respectively. Sputtering is selected for Pb thick film deposition due to its high sputtering yield. The Pb mask is used for x-ray lithography of SU-8 photoresist with 125 μm wide microchannel patterns, designing for microfluidic chip fabrication. The x-ray source for x-ray lithography is produced by synchrotron radiation at Siam Photon Laboratory, Thailand. For 180 μm thick SU-8 photoresist, Pb film thickness of around 8 μm is required to block x ray at a dose of 4200 mJ/cm3. This is less than twice the thickness required for a gold absorbing layer, which is about 5 μm. A similar relationship is seen for different SU-8 photoresist thicknesses. In addition, a steep sidewall angle of ∼89.5° for SU-8 microchannel pattern is obtained. The results demonstrate that the Pb based x-ray mask offers high lithographic quality at a very low cost. Therefore, it is highly promising for commercial applications. [ABSTRACT FROM AUTHOR]

The authors investigated the etching of grooves in low-k in Cu technology. Correlation between the line edge roughness (LER) and the time-dependent dielectric breakdown (TDDB) reliability for 100 nm pitch Cu interconnects was investigated. They controlled LER by using various gases to etch multilayer photoresist. CF3I gas was found to reduce LER better than conventional gases such as CF4 and CHF3 because CF3I has higher etching selectivity of photoresist against spin-on glass film. The LER did not affect measures of electrical performance such as wiring resistance, capacitance, and leakage current, but did affect TDDB lifetime because, according to their simulation, the electric field was strongly enhanced at curvatures in the interconnects. The maximum electric field (Emax) was also determined to evaluate the effect of LER on TDDB lifetime. All their results show that CF3I etching is promising for creating reliable Cu interconnects with smaller pitches. [ABSTRACT FROM AUTHOR]

A corner rounding metric has been used to determine the deprotection blur of Rohm and Haas XP 5435, XP 5271, and XP 5496 extreme ultraviolet (EUV) photoresists as base wt % is varied, an experimental open platform photoresist (EH27) as base wt % is varied, and TOK EUVR P1123 and FUJI 1195 photoresists as postexposure bake temperature is varied. In the XP 5435, XP 5271, XP 5496, and EH27 resist platforms, a six times increase in base wt % reduces the size of successfully patterned 1:1 lines by over 10 nm and lowers intrinsic line-edge roughness (LER) by over 2.5 nm without changing deprotection blur. In TOK EUVR P1123 photoresist, lowering the PEB temperature from 100 to 80 °C reduces measured deprotection blur (using the corner metric) from 30 to 20 nm and reduces the LER of 50 nm 1:1 lines from 4.8 to 4.3 nm. These data are used to drive a lengthy discussion about the relationships between deprotection blur, LER, and shot noise in EUV photoresists. The authors provide two separate conclusions: (1) shot noise is probably not the dominant mechanism causing the 3–4 nm EUV LER floor that has been observed over the past several years; (2) chemical contrast contributes to LER whenever deprotection blur is large relative to the printed half-pitch. [ABSTRACT FROM AUTHOR]

Because volatile species containing Si and S desorbed from photoresists during exposure to light can irreversibly damage optical elements inside a photolithography tool, accurate quantification of vapor fluxes is essential to evaluate risk. The authors compare measurements using two vapor-collection methods (gas and ion chromatographies) and one solid state analysis method (secondary ion mass spectrometry) to characterize outgassing from two silicon-containing and two all-organic 193 nm photoresists during exposure at atmospheric pressure. Results from all three techniques are in good general agreement with the notable exception of Si-containing products. Investigation of the influence of the gas atmosphere above the resist during exposure reveals that gas-surface interactions involving oxygen and perhaps water are a previously unrecognized mechanism for release of Si during exposure of otherwise inert polymers. This result shows that it is essential to take the gas environment inside the photolithographic tool into account when designing resist polymers for low Si outgassing. [ABSTRACT FROM AUTHOR]

The authors evaluated the performance of the negative-tone resist ma-n 1405 for lift-off and hybrid processes by combining electron-beam and/or laser lithography techniques with metal deposition. Electron lithography and gold deposition allow the fabrication of useful structures such as circles, lines, and coaxes with a resolution of nearly 70 nm. The ability of this resist to withstand a hybrid process was also tested. A constricted line was transferred to the silicon substrate by lift-off, where the large (1 μm linewidth) and the small (200 nm linewidth) features were, respectively, patterned by laser and electron-beam direct writing on a single resist level. [ABSTRACT FROM AUTHOR]

Hydrogen silsesquioxane (HSQ) has interesting applications as an electron-beam resist and hardmask. In this work, HSQ was investigated with regard to the postcoat delay, isofocal dose for an optimum process window, e -beam proximity effect correction, and the molecular structure in order to better understand the processing. Several independent methods were set up and applied to characterize the structuring of HSQ with e-beam including contrast measurements, basedose-over-critical-dimension tests, the isofocal dose method, the doughnut test, and Fourier transform infrared analysis. HSQ was coated on 300 mm bare silicon wafers and exposed with a 50 kV variable shaped e-beam writer in the dynamic random access memory pilot line environment of Qimonda Dresden and Fraunhofer CNT. The postcoat delay showed no significant influence on the exposure results. A dose difference between the basedose and the isofocal dose was observed, which indicates a working point in a suboptimal process window related to the poor contrast. The FTIR analysis showed a molecular structure change up to 600 μC/cm2. Finally, a point-spread-function for HSQ has been generated for the first time. In summary, the results from this detailed characterization show promise for a manageable process. [ABSTRACT FROM AUTHOR]

The deprotection blur of Rohm and Haas XP 5435, XP 5271, and XP 5496 extreme ultraviolet photoresists has been determined as their base weight percent is varied. The authors have also determined the deprotection blur of TOK EUVR P1123 photoresist as the post-exposure bake temperature is varied from 80 to 120 °C. In Rohm and Haas XP 5435 and XP 5271 resists seven times and three times (respective) increases in base weight percent reduce the size of successfully patterned 1:1 line-space features by 16 and 8 nm with corresponding reductions in deprotection blur of 7 and 4 nm. In XP 5496 a seven times increase in base weight percent reduces the size of successfully patterned 1:1 line-space features from 48 to 38 nm without changing deprotection blur. In TOK EUVR P1123 resist, a reduction in post-exposure bake temperature from 100 to 80 °C reduces deprotection blur from 21 to 10 nm and reduces patterned line-edge roughness from 4.8 to 4.1 nm. [ABSTRACT FROM AUTHOR]

One recurring problem in nanoscale processing is roughening of photoresist (PR) materials during plasma etch. We studied the plasma etch behavior of 248 nm PR, 193 nm PR, and poly methyladamantyl methacrylate while changing the source power level (400–1200 W), adjusting the bias power to change the self-bias voltage Vdc (-50 to -150 V), and varying the pressure (10–80 mTorr) and the amount of fluorocarbon gas additive to the Ar discharge (0%–10% c-C4F8 in Ar). The authors found that the PR removal is dominated by the ion energy and fluence. Surface fluorination enhanced the removal rates. Two linked mechanisms for the roughening behavior of the films during processing were identified. Changes of PR top surface roughening behavior in response to variations of bias power and pressure could be interpreted by a model of roughness formation which is dominated by a physical pattern transfer mechanism, i.e., roughness amplification through selective ion-induced transfer. When the plasma source power was varied, they observed that roughness formation was linked to the surface energy density deposited during processing. As the energy required to volatilize a volume element from the surface increased, the surface roughening rates grew proportionally. This conversion of excess energy into roughening was found to depend on the molecular structure of the polymer, with adamantyl polymers having a very high roughening constant. Additional effects on the etch behavior arise from fluorination of the samples, as quantified by x-ray photoelectron spectroscopy. High F 2s/F 1s intensity ratios, which indicate deeper fluorination, were measured for rough surface conditions. Smaller F 2s/F 1s ratios indicate near-surface fluorination and correspond to smoother top surfaces and feature sidewalls. Molecular compounds showed roughening behavior relative to the respective cross-linking behavior even when processed in pure Ar discharges, suggesting that the measured surface fluorination mirrors surface morphology. When plasma etching three-dimensional trenches and contact holes patterned in PR, the authors found that the sidewall roughness changed with process parameters in a fashion similar to that seen for blanket surface roughness introduction using the same etch conditions. A close correlation between the surface and sidewall roughness results was obtained, suggesting that their model of polymer surface roughening also applies to resist sidewall evolution during plasma etch. [ABSTRACT FROM AUTHOR]

Microfield exposure tools (METs) continue to play a dominant role in the development of extreme ultraviolet resists. One of these tools is the 0.3 numerical aperture SEMATECH Berkeley MET operating as a resist and mask test center. Here the authors present an update on this tool, summarizing the latest test and characterization results. They provide an update on the long-term aberration stability of the tool and present line-space imaging in chemically amplified photoresist down to the 20 nm half-pitch level. Although resist development has shown substantial progress in the area of resolution, line edge roughness (LER) remains a significant concern. In this manuscript the authors further present a summary of recent LER performance results and consider the effect of mask contributors to the LER observed from the SEMATECH Berkeley MET. [ABSTRACT FROM AUTHOR]

A combination of atomic force nanolithography and optical lithography, using amorphous arsenic sulfide as photoresist and a protective metallic layer mask, was used to fabricate metallic nanowires. It is shown that the scratch of the metallic mask can be performed without damage on the photoresist layer by a careful combination of the force applied by the microscope tip and a proper choice of the diamond tip scratching direction. The dimensions of the nanowires can be controlled by the depth and width of the scratched lines and the thickness of the deposited material used to grow the nanowires. The process has led to the deposition of continuous metallic nanowires with 200 nm width and 25 μm length. [ABSTRACT FROM AUTHOR]

Mask contributors to line-edge roughness (LER) have recently been shown to be an issue of concern for both the accuracy of current resist evaluation tests as well the ultimate LER requirements for the 22 nm production node. Problems arise from mask absorber LER as well as mask multilayer roughness leading to random phase variations in the reflected beam. Not only do these mask contributors effect the total measured LER in resist but they also have an impact on LER spatial scaling characteristics such as power spectral density and the related descriptors of correlation length and roughness exponent. Understanding how these metrics respond to mask effects may lead to an experimental mechanism for evaluating the importance of mask contributors to observed LER in resist. Here the authors present a detailed study of mask-induced LER spatial characteristics. The authors further describe the influence of illumination conditions and defocus on the metrics and compare the results to those expected from intrinsic resist LER. The results show power spectral density analysis to be a promising technique for distinguishing mask and resist contributors to LER. [ABSTRACT FROM AUTHOR]

The effect of fluorocarbon film and surface pretreatments on roughness formation of 193 nm photoresist (PR) during short time fluorocarbon plasma exposure was investigated. The present work complements two earlier reports by this group on surface modifications of 193 nm PR during plasma etching. The authors employed a shutter approach to minimize initial plasma transient effects on processing of PR surfaces. Surface chemical conditions after plasma etching were observed by x-ray photoelectron spectroscopy. The authors investigated the effect of deposited fluorocarbon film and pretreatments using several gas chemistries on PR roughening. Pretreated samples exhibited smaller roughness after plasma etching as compared to specimens processed without pretreatment. Three main mechanisms were identified for surface roughness reduction after pretreatment: (a) the formation of a fluorinated surface layer—having a large amount of fluorine on the PR surface at the beginning of the etch reduces PR surface roughening, (b) the improvement of durability of the PR under plasma exposure by removal of the ester group, and (c) a rapid fluorination and a reduction in the PR etch rate during the initial etch period due to the formation of a protective film on the top of the PR. The authors conclude from this work that the initial surface chemical state is an important factor that determines the degree of surface roughness formation for 193 nm PR during the initial etch period. [ABSTRACT FROM AUTHOR]

Projection maskless patterning (PMLP) is based on a programmable aperture plate system and on charged particle projection optics with 200× reduction, providing thousands of electron or ion (H+, He+, Ar+, Xe+, C60-) beams working in parallel on the substrate. As part of the European CHARPAN project a PMLP proof-of-concept tool has been realized. Using resolution templates, with 10 keV H+ multibeams a resolution of 16 nm lines and spaces was achieved in HSQ resist across the proof-of-concept tool 25×25 μm2 exposure field at an exposure dose of 25 μC/cm2. Enhancing the dose by 10% there was <1 nm increase in linewidth. With 10 keV Ar+ multibeams resistless nanopatterning of various materials was accomplished. Inserting a wired programmable aperture plate system providing ∼4000 beams, first HSQ resist exposure and patterning results have been accomplished, implementing gray scale exposure techniques. The system is being upgraded to a PMLP engineering tool integrating an aperture plate system with complementary metal oxide semiconductor electronics providing ∼40 000 programmable beams, a precursor gas injection system for in situ ion multibeam induced etching and deposition, and a laser-interferometer controlled high-precision vacuum stage. Industrial PMLP nanotechnology applications are discussed. [ABSTRACT FROM AUTHOR]

The degradation of projected images using TM polarization is not intrinsic because losses in image contrast can be recoverable. By controlling the photoresist/substrate interface reflectivity, high modulation for TM polarization can be maintained for angles approaching 90° in a photoresist. Although there is calculated loss of image contrast with increasing polarization angle using suppressed reflection (i.e., with an antireflection coating), the loss is not nearly as large when imaging over a reflective substrate. These results can potentially impact the design of illumination, possibly away from most recent TE-only schemes for oblique imaging angles and high numerical apertures (NA). Several cases of TM illumination are presented combined with tuned substrate reflectivity for 0.93 NA, 1.20 NA, and 1.35 NA and compared to results using TE illumination. Additionally, a scheme for frequency doubling with a single TM polarized exposure is presented. Using a single exposure and selective TM polarization, the reflective component produced at the photoresist/substrate interface is utilized for a four-beam imaging scenario. In doing so, patterns deposited into a photoresist film with double density are made possible. [ABSTRACT FROM AUTHOR]

With continuous reduction in linewidth of the VLSI devices, the pattern integrity of photomasks becomes considerably more important than ever. Consequently, requirement for the defect repair technology on photomasks is more severe and strict. Focused ion beam (FIB) technology has been widely used for defect repairing in photomask industry. Therefore, the performance of the FIB mask repair tool has to be improved especially in repair accuracy and precision. The FIB repair processes are classified into two kinds; one is additive repair using FIB induced deposition for missing patterns, the other is subtractive repair using gas assisted FIB etching for extra patterns. In both processes, precursor gas is applied onto the processing area through a small nozzle. Thus, the repair processes are controlled by the FIB irradiation and the precursor gas supply. Important characteristics of the repairs, such as size, shape, and placement of the repair area, are defined by the FIB scanning control. As conventional FIB systems used raster type of beam scanning for the repair processes, the size, shape, and placement could be controlled with the unit of pixel size (typically about 6–12 nm). However, in order to satisfy the recent requirement, more precise beam control is needed. The authors have developed a vector scanning system to meet the requirement. The vector scanning system enables us to control ion beams more precisely and more arbitrarily. Furthermore, unicursal beam scanning can be applied to the repair processes, which minimizes beam blanking times. By adopting the vector scanning, repair precision is improved. Additionally, sidewall angle of the repair region is also improved because the repair shape is formed without beam blanking. [ABSTRACT FROM AUTHOR]

Control of linewidth in patterns written with interference lithography (IL) is an important part of process control for many applications, requiring an understanding of how exposure parameters, such as dose and dose modulation, vary across the exposed area, and how a given photoresist will respond to changes in those parameters. This article presents a model of linewidth variation in a common IL system, Lloyd’s mirror, along with some results derived from that model that relate to optimizing the linewidth uniformity that can be achieved for a given application. An extension to a previously reported resist-characterization method that makes it possible to directly measure how changes in dose modulation affect linewidth is also discussed. [ABSTRACT FROM AUTHOR]

Extreme ultraviolet lithography (EUVL) using 13.5 nm wavelength light is the leading candidate to succeed 193 nm immersion lithography, enabling semiconductor chips with features smaller than 22 nm. Several major programs worldwide have developed this technology in recent years [D. A. Tichenor et al., OSA Proceedings on Soft X-Ray Projection Lithography, edited by A. M. Hawryluk and R. H. Stuten (1993), Vol. 18, p. 79; H. Kinachita, OSA Proceedings on Soft X-Ray Projection Lithography, edited by A. M. Hawryluk and R. H Stulen (1993), Vol. 18, p. 74; J. P. H. Benschop, W. M. Kaiser, and D. C. Ockwell, Proc. SPIE 3676, 246 (1999)] and in 2006, ASML shipped the first EUV Alpha Demo tools (NA=0.25 full-field scanners) to IMEC in Belgium [A. M. Goethals et al., Proc. SPIE 6517, 651709 (2007)] and CNSE in Albany [O. Wood et al., Proc. SPIE 6517, 6517–041 (2007)], USA. Currently the development of preproduction tools with targeted shipment of 2009 is well under way. This paper discusses the most critical items for EUVL development, namely, EUV imaging and EUV sources. Furthermore, it elaborates on the necessary development of masks and resists and, for example, quantifies how resist diffusion length can impact imaging capabilities. Results obtained and lessons learned with the Alpha Demo tools are discussed, as well as potential solutions to some of the remaining challenges. Additionally, this paper explains how EUV can realize high productivity (>100 wafers/h) and high resolutions (<22 nm) to continue the cost-effective shrink of semiconductors for several generations. [ABSTRACT FROM AUTHOR]

This study investigated the extreme ultraviolet (EUV) lithographic performance of negative-tone molecular resists based on 2,7-bis[bis(2,3,5-trimethyl-4-hydroxyphenyl)methyl]naphthalene (MGR002) and their negative-tone imaging mechanism. EUV imaging experiments were performed using the high-numerical-aperture (NA=0.3), small-field EUV exposure tool (HINA). Patterning results showed the resolution of one resist to be 29 nm at an EUV exposure dose of 18 mJ/cm2 and the obtainable aspect ratio to be as high as 2. Analyses by matrix-assisted laser desorption/ionization mass spectroscopy, Fourier transform infrared spectroscopy, and UV-visible absorption spectroscopy of EUV-exposed resists revealed that one reason for the good performance is that the negative-tone imaging mechanism involves both cross-linking and a change in polarity. [ABSTRACT FROM AUTHOR]

The dissolution characteristics of the resist film into alkaline developer have been investigated in order to understand inherent extreme ultraviolet (EUV) resist characteristics and improve resist performance. The combined increase in dissolution “slope” and dissolution contrast (Rmax/Rmin) was confirmed to improve the resolution capability of an EUV resist. Also, a higher Rmin value, which means easy and smooth dissolution of the resist into the alkaline developer, will mean a lower line-width roughness (LWR). However, more evidence and further detailed investigation are necessary to prove this relationship between this dissolution characteristic and LWR. It was also found that the slope of dissolution rate curve for the poly(hydroxystyrene) resist increases as protecting ratios is increased. However, it was found that higher protection ratios increase the amount of released resist outgassing upon EUV exposure. The dissolution rate analysis of presently available high performance EUV resist Selete standard resist 2 (SSR2) was also performed. Based on these results, a high slope and dissolution contrast value of the SSR2 correlated with its improved resolution capability at the 26 nm hp L/S. [ABSTRACT FROM AUTHOR]

In order to meet the linewidth roughness (LWR) requirements for the 16 nm node, postprocessing methods need to be investigated to reduce the LWR after the lithography step. We present the results of five different techniques applied to a single extreme ultraviolet photoresist. The results show that rinse has the most promise in achieving the nearly two time LWR improvement needed. However, other techniques such as etch/trim, hardbake, vapor smoothing, and ozonation give at least 10%–20% LWR reduction and could be further optimized. Some of the physical based techniques which melt the photoresist reduce the midspatial frequency (50–10 nm period) roughness, whereas chemical based techniques reduce the low order spatial frequencies (∼500–50 nm period). Hence, a combination of techniques may be the ultimate solution. [ABSTRACT FROM AUTHOR]

Resist critical dimension (CD) control and line edge roughness (LER) reduction has been one of the most challenging issues for sub-100 nm feature patterning in integrated circuit manufacturing. Among those factors dominating CD and LER, photoacid distribution homogeneity and diffusivity are major elements which are correlated to resist material design and have a direct impact on the lithography performance. In this work, a mesoscale stochastic model has been applied to investigate the joint effect of photoacid distribution homogeneity and diffusivity on resist lithography performance. Simulation results suggest that the high photoacid generator (PAG) loading and low photoacid diffusivity provided by polymer bound-PAG resist systems can provide superior lithography performance as compared to traditional blended-PAG resists, which is in good agreement with our previous experimental characterization of polymer bound-PAG resists. The results also suggest that resist image blur is proportional to the square root of the product of photoacid concentration generated at the line edge and the photoacid diffusion coefficient. LER was observed to be proportional to the product of the standard deviation of the extent of deprotection along the nominal line edge and the reciprocal of the gradient of the deprotection profile along the resist line edge. [ABSTRACT FROM AUTHOR]

The authors have extended the atomic force microscopy-based technique to measure intrinsic material roughness after base development to evaluate a number of different polymer types in resist formulations. These polymers include environmentally stable chemical amplified photoresist type copolymers and terpolymers, methacrylate polymers, and fluoropolymers. The surface roughness of resists containing these polymers was measured along with the clearing dose with both extreme ultraviolet (EUV) and deep ultraviolet (DUV) exposures. Selected resists containing a representative sampling of different lithographic polymers were imaged with both EUV and DUV exposures. It has been suggested by many that there is a fundamental inverse relationship between resist sensitivity and LER that leads to a fundamental limit to resist performance. It was found that no simple relationship exists between intrinsic material roughness (IMR) and sensitivity and instead some other, more complex relationship between the material properties of the polymer and resist process that is determining both the IMR and sensitivity. It was also found that no simple relationship existed between resist sensitivity and LER. This suggests that resist performance is not innately limited by any fundamental law but is instead open to further improvements through the use of new materials or material combinations. [ABSTRACT FROM AUTHOR]

A contact-hole deprotection blur metric has been used to monitor the deprotection blur of an experimental open platform resist (EH27) as the wt % of base and photoacid generator (PAG) were varied. A six times increase in base wt % is shown to reduce the size of successfully patterned 1:1 line-space features from 52 to 39 nm without changing deprotection blur. Corresponding isolated line edge roughness is reduced from 6.9 to 4.1 nm. A two times increase in PAG wt % is shown to improve 1:1 line-space patterning from 47 to 40 nm without changing deprotection blur or isolated line edge roughness. A discussion of improved patterning performance as related to shot noise and deprotection blur concludes with a speculation that the spatial distribution of PAG molecules has been playing some role, perhaps a dominant one, in determining the uniformity of photogenerated acids in the resists that have been studied. [ABSTRACT FROM AUTHOR]

By the development of the double exposure technique or the double patterning technique, the pattern placement error of a photomask is interesting because of its impact on the size and position of wafer pattern. Among various sources to induce the pattern placement error, we have focused on the charging effect of the FEP-171 resist and have shown that the resist charging effect generates the pattern position error of a clear pattern and the critical dimension variation of a dark pattern. Based on experiment and simulation, we present quantitatively the dependence of position error on pattern density, pattern shape, and writing order. Furthermore, we have discussed the model to describe the charging effect and its agreement with experiment and the correction method to remove the resist charging effect. [ABSTRACT FROM AUTHOR]

Double patterning is a manufacturing process targeted for the 22 nm half pitch manufacturing node that harbors strong potential for reaching high volume manufacturing. The double patterning process requires twice as many manufacturing steps for a device layer as are required for a traditional single reticle device layer. However, the double patterning process does have a direct interaction between the individually processed layers at the final etch processing step. Current optical proximity correction process modeling capabilities can be designed to account for the interactions between these two processes. However, it is unclear at this time whether or not real process differences are present due to an interaction between the two patterns in double patterning. It is also not certain if proximity effects play a role in the process. This study will use currently available data from the IMEC double patterning process to determine if there is a significant interaction between the photoresist and hardmask patterns during substrate etch, and if there is a significant interaction, a discussion of how a process model could be constructed to capture this effect. [ABSTRACT FROM AUTHOR]

The study investigates the use of ultraviolet nanoimprint lithography (UV-NIL) for patterning three dimensional (3D) structures. Generating the 3D structures is a challenging task especially on an insulating substrate such as quartz. These transparent molds are essential for the UV-NIL process. The 3D profiles were created on the negative tone photoresist, Microresist ma-N2403 using a Raith150 electron beam lithography (EBL) tool in a single step variable dose controlled exposure. The developed 3D resist profiles subsequently were utilized as the 3D masking layer. The 3D patterns were transferred into the quartz mold substrates by a single-step reactive ion etching (RIE). The replication of the 3D mold structure by using the UV-NIL technique requires a two-step imprint process. The master mold profile was replicated onto a Microresist Ormocomp US-S4 resist on the first imprint to become the soft mold. The cured Ormocomp soft mold, subsequently was used as a mold for replicating the 3D pattern structures on the Microresist mr-UVCur06 resist in the second imprint step to create a positive replica of the original mold. A test pattern of a 3D pyramid-shaped array with multilevel features was successfully been replicated using this technique. The replicated 3D pattern on the resist could be utilised as the final 3D masking layer for the pattern transfer onto final substrate using RIE. [ABSTRACT FROM AUTHOR]

The mechanism of modified layer formation for 193 nm photoresist (PR) during a short time (up to ∼10 s) fluorocarbon plasma exposure was investigated. We employed a shutter approach to achieve rapidly steady-state plasma condition when processing PR surfaces. The time evolution of the optical constants and the thickness of the modified layer on the PR surface were obtained using two layer optical modeling of ellipsometric data for the processed PR material. This enabled us to determine the time-resolved etching rate of the PR and the kinetics of modified layer formation. The change in the surface chemical composition of the PR materials was determined by x-ray photoelectron spectroscopy (XPS). A graphitic layer with a higher refractive index as compared to the bulk PR material was formed on the PR surface within a few (∼3 s) seconds of plasma exposure. The XPS data revealed that before a fluorinated surface developed, cleavage at ester groups of the side chain in the polymer and dangling bond formation took place, leading to cross-linking. To investigate the influence of the oxygen content of the polymer on surface roughness formation, we compared the surface evolution of oxygen-rich 193 and 248 nm PRs, which have a smaller oxygen content. Remarkable differences in the etching behavior during the initial plasma interaction period were observed for the two materials. Whereas for 193 nm PR, etching was observed immediately and the material exhibited higher surface roughness, for the 248 nm PR material fluorocarbon film deposition took place initially. Once a fluorinated surface had developed, steady-state etching took place, but the 248 nm PR exhibited lower surface roughness than the 193 nm material. XPS measurements showed that when comparing the fluorine content of the surface layer to the oxygen content, the fluorine content was relatively more important for the 248 nm PR than for the 193 nm PR. For the latter, oxygen in the side groups of the bulk PR enhanced the PR etching rate initially and led to a rapid surface roughness formation. This coincides with the development of a fluorinated surface. [ABSTRACT FROM AUTHOR]

The mechanisms of modified layer formation and the effect of surface condition on the roughness formation of 193 nm photoresist (PR) materials during short time fluorocarbon plasma exposures were investigated. The authors employed a shutter approach to minimize the plasma transient effects on processing of PR surfaces. Photoresist etching behavior and surface roughness formation during the initial etch period depend strongly on ion energy, pressure, and stage temperature, and reflect the degree of surface fluorination for these process conditions. The authors found that a high etch rate for low fluorination conditions leads to rougher surfaces as compared to etching of highly fluorinated PRs. The comparison of x-ray photoelectron spectroscopy and atomic force and scanning electron microscopy data for PR materials processed using different plasma operating conditions showed that smooth surfaces resulted when during the initial period of plasma-polymer interaction the fluorine content of the PR surface was large relative to the oxygen content of the unexposed polymer. One interpretation of this observation is that a larger amount of fluorine on the PR surface relative to the bulk oxygen content of the PR material reduces the microscopic etch nonuniformity, e.g., due to selective etching of ester groups. The grain size distribution of PR roughness was determined through Fourier transformation of the atomic force microscopy data. The results indicate that a process dependent grain size distribution is introduced initially (approximately several seconds) and subsequently only grows in amplitude as the root mean square roughness increases. The authors interpret this observation as surface roughness being formed at the beginning of the etch process and that subsequently the height of the grains grows by the roughness transfer into the underlying polymer material. [ABSTRACT FROM AUTHOR]

With the continuous scaling down of the critical dimensions (CDs) of semiconductor devices, the linewidth roughness (LWR) becomes a non-negligible parameter that needs to be controlled within 1 nm (at 3σ) for the 32 nm node and beyond. In this article, the authors have used a CD-atomic force microscopy to investigate the evolution of the LWR during the subsequent lithography and plasma etching steps involved in the patterning of polysilicon transistor gates. The authors demonstrate that the LWR present on the etching mask [photoresist/bottom antireflective coating (BARC), SiO2 or amorphous carbon hard masks] right before the gate etching is transferred into the polysilicon during the HBr/Cl2/O2 gate etching step. Thus, the final polysilicon LWR directly is strongly dependent on the lithography and plasma etching steps preceding the gate etching step. The authors show that by applying plasma treatment to minimize the resist mask LWR prior to all the other etching steps or by optimizing the BARC opening plasma chemistry, the final polysilicon LWR can be minimized. The authors also demonstrate that the introduction of hard masks (SiO2 or amorphous carbon) helps to reduce further the final polysilicon gate LWR. Finally, the authors discuss the role of the ion bombardment in the smoothening/roughening of the pattern sidewalls during plasma processes. [ABSTRACT FROM AUTHOR]

Poly(methyl methacrylate) (PMMA) is a transparent thermoplastic with important applications as a positive resist for various radiation sources. When used as a photoresist, PMMA is typically used with wavelengths shorter than 240 nm, as that is the commonly accepted upper limit of effectiveness. However, the authors have shown patterning of nonamplified PMMA films at 254 nm, which is significant because 254 nm radiation can be produced using inexpensive low-pressure mercury vapor lamps. Data for the etch depth as a function of exposure dose (0–12 h), developer temperature (20–35 °C), and etch time were collected. Dissolution rates of up to many microns a minute are possible, and the dissolution rate ratio of exposed over unexposed PMMA can reach over 3000. This demonstrates the feasibility of PMMA exposure using deep-UV at 254 nm. [ABSTRACT FROM AUTHOR]

This paper reports a new patterning method, the complementary-structure micropatterning (CSMP) technique, to fabricate the undercut structures for the passive-matrix display of organic light-emitting diodes (OLEDs). First, the polyvinylpyrrolidone (PVP) stripe patterns with a trapeziform cross-section were formed by micromolding in capillaries. Then the photoresist was spin coated on the substrate with the patterned PVP stripes and developed in water. The PVP was dissolved and lifted off from the substrate with the photoresist deposited on it, resulting in the undercut structures of the remaining photoresist. The undercut structures with different configurations were obtained by adjusting the photoresist thickness. The undercut structures were further used as separators in the patterning of the passive-matrix display of OLEDs. No visible performance difference was observed compared with OLEDs patterned by the traditional method. [ABSTRACT FROM AUTHOR]

This article reports a novel hydrostatic pressure mechanism for soft UV imprinting. With this mechanism, the pressure distribution over the whole substrate is uniform. In addition, the nonuniform deformation of the soft mold during the imprinting process can be avoided. The mechanism employs a gasbag inside a closed chamber, which upon inflation compresses the whole polydimethylsiloxane (PDMS) mold/substrate stack not only from the surface but also from all the sides. The microstructures on the PDMS mold are then replicated onto photoresist-coated substrates. The result shows that the soft UV imprinting using a hydrostatic pressure mechanism can provide uniform pressure for imprinting. It also provides the hydrostatic pressure state on the stack so that micropatterns can be replicated with high fidelity from soft mold to photoresist on substrates. The accuracy of replication, even at the edge, has been experimentally verified. [ABSTRACT FROM AUTHOR]

Photoresist characterization is an important part of process control for any lithography technique. This paper describes a means to experimentally determine how the linewidth in interference lithography varies with characteristics of the exposure, using far fewer exposures than are needed using methods previously described in the literature. In this method, a sample is exposed twice, with the sample rotated a few degrees between the two exposures. From micrographs of the developed resist patterns it is possible to determine how linewidth varies with both average dose and dose modulation, and what dose modulation is required to produce an acceptable pattern for that dose and spatial period. The information generated by this method is directly applicable to interference lithography, but can also be used to study the performance of photoresist for use with other lithography systems. [ABSTRACT FROM AUTHOR]

Photoresist defects can occur in large, seemingly noncritical features in high-resolution photolithography although it is capable of resolving features of much smaller dimensions. Both optical illumination and wafer topography may contribute to the formation of resist defects in large features. This letter examines optical simulations to uncover the specific nature of these resist defects and proposes local correction methods to improve the photolithographic process. For example, adding a back-etched phase feature in the center of a large isolated trench feature reduces the illumination-responsible resist scumming effect, and optimizing the multilayered film stack can eliminate the resist defect when printing a large contact feature on the structured wafer substrate. [ABSTRACT FROM AUTHOR]

Self-assembled multilayers, composed of alternating layers of α,ω-mercaptoalkanoic acids and Cu2+ ions (“molecular rulers”), are used as organic sidewall spacers and etch resists for the fabrication of registered microstructures with precisely tailored nanometer-scale spacings. The method outlined here eases the stringency of the lithographic processing for patterning second-generation features. Additionally, a new method to lift off the self-assembled multilayered films for the generation of the tailored nanogaps is demonstrated. The advantages of these techniques are discussed. [ABSTRACT FROM AUTHOR]

In electron-beam (e-beam) grayscale lithography, a three-dimensional (3D) structure is transferred onto the resist or substrate. In either case, accurate control of the remaining resist depth and, accordingly, profile is critical for successful fabrication of the structure. Usually, the remaining resist depth control is guided by the empirically derived dose (or exposure)-depth relationship using a two-dimensional model. However, such an approach may require multiple calibrations and also lead to significant dimensional errors for nanoscale structures due to the depth-dependent variation of exposure and the nonlinearity between exposure and developing rate. In this study, a resist developing model is incorporated into e-beam dose control schemes in order to take the exposure variation and nonlinearity into account. Through computer simulation, it has been demonstrated that significant improvement in dimensional accuracy may be achieved by including the 3D resist developing model in the e-beam dose control for fabricating nanoscale 3D structures. [ABSTRACT FROM AUTHOR]

The authors report on the fabrication of binary spiral zone plates and investigate an approach using gray scale scanning-electron-beam lithography to fabricate spiral-phase plates. The optical performance of the spiral zone plates is characterized via photoresist exposures and compared to theory. [ABSTRACT FROM AUTHOR]

Arrays of nanodots and nanoholes were patterned with a highly coherent tabletop 46.9 nm laser on high resolution hydrogen silsesquioxane photoresist using multiple exposure interferometric lithography. The authors observed for λ=46.9 nm radiation a penetration depth in excess of 150 nm. This laser-based extreme ultraviolet interferometric setup allows printing of 0.5×0.5 mm2 areas with different nanoscale patterns using a compact tabletop system and exposure times of tens of seconds. [ABSTRACT FROM AUTHOR]