Double exposure / Double pattering methodologies are being adopted to extend 193nm optical lithography until the next
generation lithography, most likely the EUV, is solidified. The Double exposure / Double patterning methodologies
require tighter image-placement accuracy and Critical Dimension (CD) controls on a mask than the conventional single
exposure technique. NuFlare Technology's mask writer, EBM-6000 (1), is capable of achieving the required CD control
and high patterning resolution as fine as 35 nm, that are required for the hp 45nm lithography with Double exposure /
Double patterning methodologies, when newly developed resist (i.e. "low-sensitivity" resist) is used, as shown at several
occasions to date. Further, image-placement control with EBM-6000 has been improved based on extensive error
budget analysis to comply with the tight image-placement specifications required by the Double exposure / Double
Patterning lithography. This paper will show the results of the analysis and improvement of the image-placement
accuracy of EBM-6000 series mask writers.
LEEPL mask specifications for LEEPL volume production tool "EBPrinter LEEPL-3000" have been fixed and 1X LEEPL stencil masks for 65nm node device fabrication have been developed and evaluated. "EBPrinter LEEPL-3000" handles a 6025 type mask which is compatible with a 6025 photo mask and a 200 mm wafer type mask. Both of masks have silicon based membranes with a thickness of 0.5 μm - 2.0 μm. Exposure field size of "EBPrinter LEEPL-3000" is 46mmX46mm, and pattern area of the LEEPL mask is the same which covers four fields of an optical stepper and a filed of an optical scanner. Obtaining small patterns, high CD accuracy and high image placement (IP) accuracy are very important for a 1X mask. Less than 70 nm patterns with a CD accuracy of 6.9 nm were obtained. As "EBPrinter LEEPL-3000" has a mask global IP error correction function using a sub-deflector, specification of IP error is not so critical. Spedification of global IP error is less than 30 nm, and local IP error is less than 10 nm. Those are easily obtained with membrane stress control and a multiple exposure method of electron beam mask exposure systems.
Masks for low energy electron proximity projection lithography (LEEPL) require thin membranes, which in turn make the development of low-distortion masks a critical issue for this technology. By using an evaporated resist, a flip side fabrication process is presented here in which mask patterning is carried out with the mask in the same orientation that it will have in the stepper. This new process reduces distortions of a typical LEEPL mask which usually requires patterning on the opposite side of the membrane causing a gravitational sag effects. In addition, an evaporated resist has significant advantages for mask fabrication as membrane distortion is reduced due to the absence of centrifugal force during the resist deposition process. Uniform heat distribution across the membrane during the etching process is also expected since the membrane can now be placed in direct contact with a cooled metal electrode, thereby improving the etch rate uniformity. Also, for large scale production, several mask replicates from the original mask must be made because they have limited lifetime when used in a stepper. Image placement distortion can be minimized and the yield can be improved in mask replication by using an evaporated resist.
The two LEEPL beta-tools were completed in earlier 2002 and have been evaluated for the performance. 50nm CH patterns and 70nm L/S patterns are attained and the CD uniformity of 70nm L/S patterns with 37 shots on a 200mm wafer is under 4nm with the LEEPL beta-tools. In addition, it is proven that the fine tuning deflector can correct a mask and a wafer distortion by giving a minute angle to the electron beam. The mask distortion with respect to orthogonality and magnification is decreased on a wafer by 1/5. By means of this fine tuning deflector, Mix & Match accuracy with any other lithography tools will be better and difficulty of 1X stencil mask fabrication wil be easier. Referring to the data which has been obtained with the evaluation of the LEEPL beta-tools, the first LEEPL production tool dubbed "LEEPL-3000" is under construction to realize the satisfactory ability for 65nm node device fabrication. The shipping of the first LEEPL-3000 is scheduled in earlier 2003 and it is earlier than any other Next Generation Lithography technologies.
Scniiconductor industry currently faces double-hold pains. One is the record-breaking semiconductor depression brought by the IT bubble collapse in the United States, and another is a brick wall called "Lithography Crisis" which stands in front of the technological progress. The former is that no growth of semiconductor market can be anticipated except household electric appliances because of the surplus of the optical communication infrastructure in the United States and sales reduction ofpersonai computer and cellular phone. The latter is that no cost effective solution can be found for next generation lithography, which has been searched for in a long time. Even in the semiconductor market of household electricity appliance, small or middle quantity production in multiple kinds is required to be extremely low cost. it is the new lithography called "LEEPL" (Low Energy Electron Beam Proximity Projection Lithography) that challenges to solve these cost issues. Thanks for simple tool configuration and small niask pattern area, LEEPL promises both tool and mask to be low cost.
We have studied the optical proximity effect, the depth of focus (DOF) and the mask-error-enhancement factor (MEF) of ArF attenuated phase shift masks (att. PSMs) for the application to 100-nm logic gate patterns. In the previous work, we have compared the performances of the binary mask with those of the 6% transmittance ArF att. PSM for printing 100-nm lines, and we obtained the better performances on the att. PSM with the annular illumination. In this paper, we reported the evaluation results of the higher transmittance att. PSMs. We fabricated them with 6-20% transmittance (quartz reference). Those masks have TaSiO bi-layer shifter with two tones. We did not use assist bars. An ArF scanner with NA 0.60 and a positive chemically amplified ArF resist with the thickness of 0.4 um are used. 100 nm semi-dense lines with a pitch of 250 nm are exposed without bias and 100 nm isolated lines are exposed with biases of 5 -15 nm (on wafer). Our results show that the higher transmittance att. PSMs can improve the DOF and the MEF on both semi-dense lines and isolated lines. In our experiments, the highest transmittance of 20% gives the broadest DOF of 0.4 um and the smallest MEF of 1.78 and 1.11 for semi-dense lines and isolated lines respectively. However, some higher transmittance PSMs may create the side lobes on the large patterns. From these results, we demonstrated that that 10% transmittance is suitable for the ArF att. PSMs for the fabrication of 100 nm line patterns.
Electron beam (EB) lithography has often been used for fabricating advanced ULSIs. Recently, to increase the writing throughput, electron beam projection lithography (EBPL) technology has been proposed (100kV acceleration voltage and 20-30(mu) A total currents). When we introduce the EBPL to practical use, it is important to develop a projection mask and a mask data conversion system, because there are many differences between the EBPL mask data and the conventional EB data. In EBPL mask conversion system, it is necessary to divide a full chip data into several 1mm X 1mm (250micrometers X 250micrometers on the wafer) sub-fields, which size is as same as one EBPL shot with format conversion. In this paper, we show the data conversion system that converts pattern data (GDS+U) to EBPL mask data. This system can maintain the hierarchy data structure in the dividing process. The patterns that located on the boundary between neighboring fields were treated as belonging in either field for preventing critical division. As a result, a data conversion system for EBPL mask, that can divide the device data with high speed and high quality, was successfully obtained.
We have developed a stitching accuracy measurement system for electron beam (EB) direct writing and electron beam projection lithography (EPL). This system calculates the amount of a stitching error between two EB shots from SEM images. It extracts a representative edge line of each pattern from the graphical format files (BMP, JPEG etc.) of SEM images and calculates a distance between each edge line as a stitching error. For obtaining a higher stitching accuracy of EB direct writing or EPL machines, it can analyze the relation of amounts and direction of a stitching error with a field size or a field position of these machines. We could successfully measure about 2.0 nm as a stitching error value in 0.1 micrometers L/S resist patterns on a bare-Si substrate and obtain 1.2 nm (3(sigma) ) as the measurement repeatability. It took 2.5 sec. for this system to measure one stitching region.
KEYWORDS: Quantum efficiency, Projection lithography, Lithography, Electron beam direct write lithography, Electron beam lithography, Electron beams, Photomasks, Scanning electron microscopy, Molecules, Chemical analysis
Electron beam (EB) lithography has often been used for fabricating advanced ULSIs. Recently, to increase the throughput, EB projection lithography (EPL) has been proposed. If 100 kV acceleration voltage and 20 to 30 (mu) A beam current are to be adopted in this technology, a high sensitivity resist will have to be developed to achieve a throughput of more than 30 wafers/hour (8'(phi) ). In this paper, we show the photoacid generator (PAG) optimization of a polyhydroxysterene (PHS)-based chemically amplified negative resist for EPL. To evaluate the resist sensitivity and the resolution, we prepared the PHS-based negative resists with PAGs of various quantum yields of acid generation, which were the onium-salt- type PAG, the imide-type PAG, and the alkylbenzene-type PAG. The cross-linker was the melamine-type one. Two simultaneously obtain a high sensitivity of less than 10.0 (mu) C/cm2 and a high resolution of less than 0.10 micrometer, a PHS-based negative resist with the imide-type PAG was most preferable. With this resist, we successfully obtained 0.08-micrometer gate line patterns (128 K sub-array of DRAM), exposed by one 250 X 250 micrometer2 EB shot using a 100-kV EB projection experimental column. In addition, the throughput was estimated to be 30 wafers/hour (8' (phi) ) or more.
In electron-beam cell projection lithography, it is important to achieve high resolution and high throughput enough for use in mass production. Mask bias method has been demonstrated to be very effective in improving both performance. However, factors of mask bias effects have not been clarified. In this paper, we have analyzed the factors of the mask bias effects and have discussed a method for estimating the optimum bias. We have found that the decrease of beam blur due to the Coulomb interaction and expansion the space of streams to decrease the beam overlapping neighboring each other are the main factors and that the back-scattering effect can be negligible. Finally, we have derived and proposed a method for obtaining the optimum mask bias that produces high resolution and high throughput.
KEYWORDS: Signal detection, Error analysis, Electron beam lithography, Lithography, Electron beams, Integrated circuits, Semiconducting wafers, Photomasks, Calibration, Signal to noise ratio
For development and practical fabrication of advanced ultra- large-scale integrated circuits (ULSIs), cell projection (CP) electron beam (EB) lithography has the advantage of high writing throughput, compared with conventional variably shaped (VS) EB lithography. However, when the CP method is used for fabricating advanced ULSIs, the shot stitching accuracy for the minimum feature size between CP and VS EB shots becomes a serious problem. To obtain highly reliable shot stitching accuracy between CP and VS EB shots, we developed a highly accurate stitching method using a cross- correlation method for measurement of the CP EB shot center position. By using the cross-correlation method between ideal and practical detected signals, an estimation stability error of less than 10 nm for the position of the CP EB shot center was achieved. We applied this proposed method in Gbit DRAM pattern fabrication. The mean value for the EB shot stitching accuracy was always suppressed to less than 10 nm. As a result, we obtained a shot stitching accuracy of less than 25 nm between CP and VS EB shots.
We newly developed the mask bias method in electron beam (EB) cell projection lithography to improve the resolution and to increase the throughput. In this method, the open slits of EB mask are shrunk to reduce the ratio of open area in EB mask. This shrinkage decreases the Coulomb interaction effect and the proximity effect. This results that 0.14 micrometer L/S pattern can be resolved even at the maximum shot size (25 micrometer2) and so the writing time of 0.14 micrometer L/S pattern can be reduced by half.
An improved proximity effect correction method including Coulomb interaction effect correction for Gbit DRAMs is presented. When cell projection (CP) electron beam (EB) direct writing is applied to both the cell array region and peripheral region of DRAMs, cell projection shot (CPS) size must be optimized to prevent both the proximity effect and Coulomb interaction effect. Moreover, based on the results of Monte Carlo simulation, it has been shown that optimum doses of each shot must be calculated taking into account the beam blur. These optimum doses are dependent on line width and different even among patterns which have the same pattern density. We obtained 0.15-micrometer lines and spaces (L/S) patterns using a cell projection EB direct writing with the improved correction method.
Electron beam (EB) direct writing has often been used for advanced device development because the dimensions of the pattern required for advanced device fabrication are below the maximum resolution of applicable optical lithography. Recently, EB direct writing has been put to practical use in advanced device production using the cell projection (CP) method, which drastically increases the writing throughput of an EB direct writing system. When CP EB direct writing is used to produce advanced DRAM, however the handling of a large volume of data becomes an important issue. A data preparation system with simple proximity effect correction, which is suitable for the CP method, is therefore required for the reduction of the data preparation time (conversion time). We therefore propose the improved proximity effect correction, which is based on the self-consistent method, using a one dimensional calculation method for CP direct writing. This method can drastically reduce the exposure intensity calculation time because the calculation of a linear integral can be substituted for that of a surface integral. The conversion time to fabricate a 0.15 micrometer Gbit DRAM (10.3 multiplied by 109 rectangle patterns) with a CD accuracy within plus or minus 5%, was reduced to be only 10% of original elapsed time (in conventional 2-D correction method).
In electron-beam (EB) cell projection lithography, the defect in resist pattern caused from the defect on aperture mask (EB mask) must be diminished. We have experimentally evaluated the printability of the defect and confirmed EB mask inspection standard. Line-and-space (L&S) patterns of 0.20 μm width (on the wafer substrate) have been fabricated on the EB mask, which has bridge defects with width of 0.20 μm and height of 0.04-0.20 μm in the line patterns, and exposed with changing the exposure dose. The programmed bridge defect of 0.04 μm height on the wafer (1.0 μm on the EB mask) is not printable at the optimum exposure dose. However, at 90% of the optimum exposure dose, it forms a 0.08 μm disconnection defect. The defect on the EB mask is printable especially in lower exposure dose range. The defect printability is higher than the resolution of L&S pattern. The defect larger than 10 % of line pattern width on the EB mask are not allowed in order to form high quality resist patterns.
NA and (sigma) will be optimized to establish 0.35 micrometers i-line single layer resist process without use of super resolution techniques. Resolution, depth of focus (DOF), and proximity effect are evaluated using a variable NA and (sigma) stepper. NA is varied by an aperture stop in a projection lens. (sigma) is varied by not only an aperture stop (mechanical (sigma) ) in an illumination optics but also intensity distribution of illumination at the aperture stop (effective (sigma) ). Optimized NA and (sigma) are applied to a newly developed high resolution resist. Obtained results show that high NA and high (sigma) stepper has a great availability for 0.35 micrometers device fabrication.
The deviation of phase shift angle from 180° seriously deteriorates the focus latitude. In order to obtain the expected performance of phase shift mask, a Chromium(Cr)/Phase-Shifter/Quartz(Qz) structure is investigated. In this phase shift mask structure, the shifter thickness i.e., phase shift angle, can be precisely controlled, compared with a conventional Shifter/Cr/Qz structure. Spin-on-grass(SOG) is used as the phase shifter material because of its excellent thickness uniformity. Alternating phase shift mask that has the Cr/SOG/Qz structure was fabricated using Ar-laser writing method, and evaluated using a NA=0.45, 6=0.3-0.5,I-line stepper. Obtained results show that this phase shift mask structure is very promising for the subhalfmicron pattern formation.
In order to improve resolution and depth of focus (DOF) in reduction projection aligner, we investigate annular illumination method, in which the center portion of light source is screened.
This paper describes the relationship between resist pattern formation capability and the screened ratio of condenser aperture. And also, the pattern deformation induced by annular illumination is investigated.
First, on the basis of simulation analysis of this illumination method, the optimum optical parameters were selected to obtain high resolution and enough DOF. The effect of annular illumination was confirmed experimentally in i-line exposure. Secondly, to explore optical contrast dependence, we compared the resolution capabilities of both positive and negative resists, having different dissolution characteristics each other. It was found that annular illumination is more effective in low contrast region of light intensity. Thirdly, the dependence on pattern periodicity (L/S and isolated line), and pattern deformation were also examined. This method is effective in periodic pattern but not in isolated pattern, and induces a little pattern deformation in the edge region.
Annular illumination that is a simple method compared with phase shift mask, is a promising for expanding the process margin of 0.35 ?m resist patterning. Furthermore, this method becomes more effective, if adapting thin resist or new high resolution resist, which can be resolved even in low contrast light condition.
The simulation analysis of standing wave effects in deep-UV lithography, which is the most serious problem in 64 MDRAM fabrication, is studied. One lithographic problem for device fabrication is periodic linewidth variation due to multireflection light in the resist film. It was observed that the amplitude of linewidth variation was more than 0.1 micrometers for a 3- component chemically amplified negative resist using a KrF excimer laser stepper (NAequals0.42). So, we simulated the standing wave effect as a function of resist material, exposure tool, and process conditions in order to minimize periodic linewidth variation. The SAMPLE simulator was improved to calculate a series of pattern profiles automatically, to change simulation parameters, and to create a periodic linewidth curve from the simulated pattern profiles. The following discoveries were made: Increasing optical resistance slightly decreases linewidth variation, although the resist pattern profiles degrade significantly. High NA lenses can also decrease linewidth variation under best focus conditions, but, linewidth variation increases significantly under slight defocus. Also, an optimum lens NA value, that minimizes periodic linewidth error, was found. Finally the Anti-Reflecting-Coating (ARC) technique, which can diminish standing wave effects, also degrades pattern profiles. To summarize, in order to minimize linewidth variation due to standing wave effects, it is important to minimize reflectivity, to optimize lens NA and coherency factor by simulation analysis.
Dissolution kinetics, as well as the formation mechanism of a surface insoluble layer produced by dipping into TMAH (tetramethylammonium hydroxide) developer, have been investigated. In the previous paper, we mentioned that dissolution rate characteristics of high resolution novolac resist are clearly divided into three regions. To investigate this dissolution mechanism, we evaluated the temperature dependence of R (Dissolution rate) by changing the exposure dose, PAC (photoactive compound equals dissolution inhibitor) and the TMAH concentration. From Arrhenius Plots of these resist systems, it is considered that R is determined by two competitive reactions in the presence of TMAH; i.e., (a) the complex formation between PAC and novolac resin which produces dissolution inhibition effect, (b) TMAH induced deprotonation of phenolic hydroxy groups in novolac resin which accelerates the dissolution of the resist. Furthermore, we also describe the formation mechanism of a surface insoluble layer produced by dipping into a TMAH developer followed by water rinse, on the basis of the dissolution time of this layer (ts). The resist surface was also analyzed by using FT-IR (Fourier transform infrared spectroscopic measurement) and XPS (X-ray photoelectron spectroscopy). As a result, it was found that (a) water rinse is essential for the surface insoluble layer formation, (b) the ts value is not directly correlated with PAC accumulation in the resist surface and (c) the ts value becomes longer when the amount of penetrated TMAH into the resist increases. These results suggest that the surface insoluble layer is produced via water rinse of PAC-novolac complex described above.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.