A non-topcoat (non-TC) resist is a photoresist that contains a hydrophobic additive, which segregates to the surface and
forms a layer to minimize surface free energy. The improvement of surface hydrophobicity and the suppression of resist
component leaching were confirmed by using this segregation layer. Compared to conventional topcoat process, it is
speculated that the use of non-TC resist will reduce the cost of lithographic materials, improve throughput, and will be
compatible for the scanning speed improvement of immersion scanners. One issue for the non-TC resist is the possibility
of increased defect generation compared to processes using topcoats. It is assumed that the high resist surface
hydrophobicity and the developer insolubility of the hydrophobic additive are main factors causing the increase in defect.
Therefore, it is important to work out solutions for reducing these defects to realize the non-TC resists. A process of
selectively removing the hydrophobic additive between exposure and development process for the purpose of defective
reduction of non-TC resist was investigated. Specifically, wet processing was performed to the wafer after exposure
using an organic solvent to dissolve the hydrophobic additive. As a result, defect count was reduced to less than 1/1000
with the effective removal of the segregation layer without affecting pattern size. These results prove the effectiveness of
the proposed process named 'selective segregation removal (SSR)' treatment in reducing defects for non-TC resists.
In previous studies, we identified filter properties that have a strong effect on microbubble formation on the downstream side of the filter membrane. A new Highly Asymmetric Polyarylsulfone (HAPAS) filter was developed based on the findings.
In the current study, we evaluated newly-developed HAPAS filter in environmentally preferred non-PFOS TARC in a laboratory setting. Test results confirmed that microbubble counts downstream of the filter were lower than those of a conventional HDPE filter. Further testing in a manufacturing environment confirmed that HAPAS filtration of TARC at point of use was able to reduce defectivity caused by microbubbles on both unpatterned and patterned wafers, compared with a HDPE filter.
In this study, we focus on the controllability of a wafer bevel from adhesion and hydrophobicity viewpoints in order to
solve the problems of film peeling and microdroplet formation around wafer bevels, which result in pattern defects.
Hexamethyldisilazane (HMDS) treatment is a common solution to these problems. We examine a novel wafer bevel
treatment utilizing silane coupling agents (SCAs) for obtaining high adhesion and hydrophobicity. SCAs comprise
trimethoxysilanol and organic functional groups. These groups react with inorganic substrates and films just over the
surface subjected to a novel chemical treatment (NCT), respectively. Several organic functional groups both with and
without fluorine are examined. The hydrophobicity is estimated from the static and receding contact angles of water.
The adhesion strength is measured from the stress required for pulling the topcoat film away from the substrate subjected
to the NCT. The coating performance of chemicals on the surface by the NCT and the aging stability of the formulated
solution of the SCAs are examined for optimizing the composition of the NCT solution. Further, we verify the film
peeling behavior and water leakage in wafers having a topcoat, ArF resist, and bottom antireflective coating (BARC)
using a quasi-immersion exposure stage.
A new technology called the double patterning (DP) process with ArF immersion lithography is one of the candidate
fabrication technologies for 32 nm-node devices. Over the past few years, many studies have been conducted on
techniques for the DP process. Among these technologies, we thought that the double Si hard mask (HM) process is the
most applicable technology from the viewpoint of high technical applicability to 32 nm-node device fabrication.
However, this process has a disadvantage in the cost performance compared with other DP technologies since these HMs
are formed by the chemical vacuum deposition (CVD) method.
In this paper, we studied the DP process using a dual spin-on Si containing layer without using the CVD method to
improve process cost and process applicability. Perhydropolysilazane (PSZ) was used as one of the middle layers (MLs).
PSZ changes to SiO<sub>2</sub> through the reaction with water by the catalytic action of amine in the baking step. Using PSZ and
Si-BARC as MLs, we succeeded in making a fine pattern by this novel DP technique. In this paper, the issues and
countermeasures of the double HM technique using spin-on Si containing layers will be reported.
A dynamic receding contact angle (RCA) is a well-known guideline to estimate the degree of watermark (WM)
defects, which shapes circle and bridges inside of the defect and reduces with enlarging the RCA of topcoat (TC).
However, our recent investigation revealed the occurrence of the circular shape defects in spite of using the TC with a
large RCA, bringing about a change of line and space pattern pitch. In this paper, we clarify the origin of these defects
and propose a new key factor of the dynamic surface properties of immersion-specific defects. It was found that the
pitch-change defect is caused by the lens effect of the air bubbles embedded between advancing water meniscus and the
TC. To well understand generation of the bubble defects, we defined the "effective" hysteresis (EH) as the hysteresis of
dynamic contact angle taken the effects of water-absorption into account. An analysis with the EH indicates that the
bubble defect arises from not only to the large ACA but also small amount of water uptake and the amount of
water-absorption could be substituted by the dissolution rate of TC. It was demonstrated that the EH proposed is a new
key factor for estimating the number of bubble defects. The EH is very useful for analyzing the bubble defects in
immersion lithography. The characteristics of the bubble defect are also discussed with a focus on the structure of the
polymer attached to water.
In the manufacture of devices beyond the 45 nm node, it is important to employ a high-performance multi-layer resist (MLR) process that uses silicon containing ARC (Si-ARC) and spin on carbon (SOC). We examined an additional hardening process of SOC by H<sub>2</sub> plasma treatment in order to improve the etching durability of the MLR. The dry etching durability of H<sub>2</sub>-plasma-hardened SOC film showed a drastic improvement, while the wiggling features of the MLR without H<sub>2</sub> treatment observed after SiO<sub>2</sub> etching disappeared completely. The hardening mechanism of SOC was analyzed by Fourier transform infrared spectroscopy (FTIR) with gradient shaving preparation (GSP) and Raman spectrometry. The formation of diamond-like amorphous carbon at a depth of approximately 50 nm was observed and was attributed to the improvement in the dry etching durability. In addition, the MLR stack with hardening has good reflectivity characteristics. The simulated reflectivity at the interface between the bottom of the resist and top surface of the MLR stack with hardening below 0.6% was attained over a wide range of Si-ARC thicknesses and hyper NA (~1.3) regions. The measured refractive indices of the hardened SOC film at 193 nm had a high value at the surface; however, they gradually decreased toward the inner region and finally became the same as those of untreated SOC. This might be the origin of the estimated excellent reflectivity characteristics.
Top coat process is required for immersion lithography in order to prevent both the chemical contamination of scanner optics with eluted chemicals from resist material and the formation of residual droplet under the immersion exposure with high scanning speed. However, defect density of ArF immersion lithography with alkaline developer soluble type top coat material is much higher than that of ArF dry lithography. Mimic immersion experiments comprised of soaking of exposed conventional dry ArF resist with purified water followed by drying step were performed in order to study the immersion specific defects. It was suggested that the origin of immersion specific defects with alkaline developer soluble type top coat was the remaining water on and in the permeable top coat layer that might interfere the desired deprotection reaction of resist during post exposure bake (PEB). Therefore, application of post exposure rinse process that can eliminate the impact of the residual micro water droplets before PEB is indispensable for defect reduction. Post exposure rinse with optimized purified water dispense sequence was noticed to be valid for defect reduction in mimic immersion lithography, probably in actual immersion lithography.
We have developed a new ArF-RELACS (Resolution Enhancement Lithography Assisted by Chemical Shrink) material called AZ-LExp.R720. The principle and process procedure of LExp.R720 are almost identical to those previously developed with KrF lithography. The extent of crosslinking reactions and the mobility balance of chemical components at the boundary between resist and the RELACS film is adjusted to ArF resist chemistry. LExp.R720 can vary shrinkage from 10 to 40nm by controlling the process conditions, mainly the mixing bake temperature. The amount of shrinkage is independent of pattern pitch and focus. We confirmed that pattern profile, lithography margin, CD uniformity, etching resistance, and pattern defects were not deteriorated by the RELACS process with deionized water development. L.ExpR720 was able to get an amount of shrinkage with several of ArF resists, which has commercial applications. In conclusion, we believe that LExp.R720 is extremely useful for 65 nm node and next generation devices.
In the past several years, ArF immersion lithography has been developed rapidly for practical applications. One of the most important topics is the elucidation of a mechanism and its solution of immersion specific defects. In this paper, we report several analytical results of immersion specific defects. First, we classify several possible origins of specific defects that are proposed based on our experiment on the actual immersion process and previous literature. We focused on a droplet of immersion water that was the origin of circular and deformed circular-type defects. Further, a watermark (WM) was created on some types of film stacks with or without the topcoat (TC) on the resist. We observed that all samples exhibited the trace of the WM. From chemical surface analyses, we obtained different types of components in the residue of the WM, which dried spontaneously. These components depended on the tested film stack. Some types were not always derived from leaching materials in the resist. Some components in the residue appeared to be airborne contaminants that were unregulated in machines used in the photolithography process. Based on the results of these tests, we discussed some methods for avoiding defects according to the droplet WM.
193 nm lithography is one of the most promising technologies for next-generation lithography and is being actively evaluated for making it practicable <sup>(1,2)</sup>. First, we evaluated an immersion lithography tool (engineering evaluation tool (EET)) <sup>(3)</sup> and a dry lithography tool (S307E) with the same numerical aperture (NA = 0.85), manufactured by Nikon Corporation. As a result, an increase in the depth of focus (DOF) of the EET to 200 nm in comparison with the DOF (110 nm) of the dry exposure tool was confirmed in a 90 nm isolated space pattern. Next, the optical proximity effect (OPE) in this pattern was evaluated. Generally, when an immersion lithography tool is compared with a dry one with the same NA or both the tools, only an increase in the DOF is found. However, we confirmed that the OPE (The OPE of the 90 nm isolated space pattern is defined as the difference in the space width between a dense space and an isolated space.) of the dry exposure tool for the 90 nm isolated space pattern reduced from 33.1 nm to 14.1 nm by immersion lithography. As the effect of the reduction of 19 nm, the OPE reduced to 15.2 nm by the effect of the top coatings (TCs) and to 3.8 nm by the optical characteristics. An impact of about 5 nm on the OPE was confirmed by the process parameters-film thickness and the pre-bake temperature of the TC. In the case that the solvent was replaced with a high boiling point solvent, the impact changed from 5 to 20 nm further, the replacement of the solvent had a considerable impact on the OPE. However, this influence differs considerably according to the kind of resists; further, it was shown that the addition of acid materials and a change in the polymer base resulted in a high impact on the OPE for a certain resist. Thus, it was demonstrated that the selection of TC is very important for the OPE in immersion lithography.
A chemical shrink technology, RELACS (Resolution Enhancement Lithography Assisted by Chemical Shrink), utilizes the cross linking reaction catalyzed by the acid component existing in a predefined resist pattern. This “RELACS” process is a hole shrinking procedure that includes simple coating, baking, and rinsing applied after conventional photolithography. Our target is realize of sub-70nm hole pattern formation by using new RELACS for ArF resist. At present, RELACS process is introduced to mass production of KrF lithography by using AZ R200 (Product name of Clariant) mainly. Then first of all we reported process performance of conventional RELACS material, AZ R200 with ArF resist. However AZ R200 does not show satisfactory shrinkage on ArF resist. Thereupon, we started on the development of new RELACS corresponding to ArF resist. As the result, we developed new RELACS material including Cross Linking Accelerator (CLA). It was found that CLA is able to improve reactivity of RELACS with ArF-resist. By using this new RELACS, It is Realized sub-70nm hole pattern formation with ArF-Ex lithography and It is able to Control of hole size by mixing bake (MB) temperature and additive ratio of CLA. Moreover this process was realized that thickness of shrunk hole is increased.
Mitsubishi Electric Corporation (MELCO) has developed an advanced microlithographic process for producing 0.1 micrometer contact holes (CH). A chemical shrink technology, RELACS<SUP>TM</SUP> (Resolution Enhancement Lithography Assisted by Chemical Shrink), utilizes the crosslinking reaction catalyzed by the acid component existing in a predefined resist pattern. This 'RELACS<SUP>TM</SUP>' process is a hole shrinking procedure that includes simple coating, baking, and rinse steps applied after conventional photolithography. This paper examines the process parameters affecting shrinkage of CH size. We subsequently evaluated the dependency of CH shrinkage on resist formulation. We conducted investigations of shrink magnitude dependency on each process parameter. (1) Photoresist lithography process: CH size, exposure dose, post development bake temperature. (2) AZ<SUP>R</SUP> R200 [a product of Clariant (Japan) K.K.] RELACS<SUP>TM</SUP> process: Soft bake temperature, film thickness, mixing bake temperature (diffusion bake temperature), etc. We found that the mixing bake condition (diffusion bake temperature) is one of most critical parameters to affect the amount of CH shrink. Additionally, the structural influence of photoacid generators on shrinkage performance was also investigated in both high and low activation energy resist systems. The shrinkage behavior by the photoacid generator of the resist is considered in terms of the structure (molecular volume) of the photogenerated acid and its acidity (pKa). The results of these studies are discussed in terms of base polymer influence on shrinkage performance and tendency. Process impact of the structure and acidity of the photogenerated acid is explored. Though the experimental acetal type KrF positive resist (low activation energy system) can achieve around 0.1 micrometer CH after RELACS<SUP>TM</SUP> processing under the optimized condition, the experimental acrylate type positive resist (high activation energy system) showed less shrinkage under the same process condition. The shrinkage performance of RELACS<SUP>TM</SUP> process largely depends on the resist chemistry used as the underlying layer. Further, shrinkage degree can be controlled by process optimization even for the high activation energy type photoresist.