X-ray Stepper Research Articles

Development, installation, and performance of the x-ray stepper JSAL 5C

We show the first results from a newly-operation advanced x-ray stepper, and describe an improved type of x-ray mask for use on it. The Mod 5C is the newest x-ray stepper developed by JMAR (JSAL Nanolithography). It is configured for installation at the Center for NanoTechnology (CNTech) on a synchrotron radiation port of the Aladdin storage ring. The alignment system is based on a new IBBI detection system with demonstrated subnanometer resolution. For this system, three microscopes are used for detecting the misalignment between wafer and mask marks with the error signal fed back in the mask positioning stage with the mask locked to the wafer stage through the use of a high-resolution laser interferometer. Both field-by-field and global alignment techniques are possible, and multiple wafer sizes can be used with a NIST format mask held in a kinematic mount, with a gap setting capable of adjustment between 2 and 50microns. A novel beam scanning system was installed to deliver a uniform exposure over a field of 35×35mm2 field. The high power from the storage ring allows exposure times in chemically amplified resists of 5–10s for the CNTech-designed beamline; shorter exposure times could be achieved with a refocusing beamline, and the stepper can be configured for an x-ray point source. To achieve sub-35nm resolution requires very small gaps. A new type of mask had to be developed, on the basis of the mesa mask originally developed at MIT. A very high flatness is required to maintain a parallel mask-wafer exposure environment. Currently, for low cost and easy manufacturing, we employ gold-plated absorber masks with a silicon nitride membrane. From the initial exposures, the resolution achieved is below 50nm with exposure conditions described. The issues of the resolution of the exposure tool are addressed and compared to the outcomes from detailed modeling.

High-power collimated laser-plasma source for proximity x-ray nanolithography

A compact laser-produced plasma x-ray source radiates 1 nm x rays with an average power of 24 W in 2π steradians. The x-ray conversion efficiency is 9% of the laser power delivered on target. The 300 W laser power is generated by a compact diode-pumped, solid-state Nd:YAG laser system. The point source x-ray radiation is collimated with a polycapillary optic to a parallel x-ray beam. The collimated plasma source (CPS) is used to demonstrate proximity x-ray lithography of 100 nm lines with a 16 μm gap between the mask and wafer. The CPS is optimized for integration with an x-ray stepper to provide a complete collimated plasma lithography exposure tool for the manufacture of high-speed GaAs devices.

High-Power Compact Laser-Plasma Source for X-ray Lithography

A compact laser-produced plasma X-ray source that radiates 1 nm X-rays with an average power of 24 W in 2π steradians is presented. The X-ray conversion efficiency is 9% of the laser power delivered on target. The 300 W laser power is generated by a compact diode-pumped, solid-state Nd:YAG laser system. The X-ray source was used to demonstrate X-ray lithography of 75 nm lines. The X-ray source is optimized for integration with an X-ray stepper to provide a complete X-ray lithography exposure tool for the manufacture of high-speed GaAs devices.

Evaluation of new x-ray stepper, the XRA

The exposure performance has been evaluated for the new x-ray stepper, the XRA, which is equipped with global alignment and magnification correction systems. Dose uniformity in the exposure field of 3.9%, stage accuracy of less than 20 nm, and good linearity and stability between the magnification change and applied force were obtained. For the 100 nm node, both critical dimension (CD) control of 10 nm and overlay accuracy of less than 30 nm were obtained using two 4 Gbit masks of the gate and contact hole by the double exposure method. Possible improvements of the dose uniformity and mask pattern CD will result in more accurate CD control. By compensating for the alignment offset, overlay accuracy of 23 nm for the 70 nm node is also expected.

Recent progress in 1×x-ray mask technology: Feasibility study using ASET-NIST format TaXN x-ray masks with 100 nm rule 4 Gbit dynamic random access memory test patterns

This article discusses recent progress in 1×x-ray mask technology in Japan. Proximity x-ray lithography (PXL) using synchrotron radiation light with a wavelength of 0.7–1.0 nm can in principle provide as high a throughput as optical lithography, because PXL and optical lithography both employ photon beams and masks. A high-performance electron-beam mask writer called the EB-X3, suitable for PXL, has been developed by NTT under the Association of Super-Advanced Electronics Technologies (ASET) program. It has a reproducible image placement (IP) accuracy of less than 15 nm. We used the shaped-beam EB-X3 to fabricate 100 nm rule x-ray masks for the gate and contact/hole (C/H) layers of 4 Gbit dynamic random access memory (DRAM) test patterns. The ASET-NIST (National Institute of Standards and Technology) type masks consist of a 350-nm-thick TaBN absorber, a 3-μm-thick SiC membrane (27-mm square membrane area), a 1-mm-thick Si substrate, and a 6.63-mm-thick Pyrex glass frame. We have achieved an IP accuracy for 1×x-ray masks of less than 15 nm at both the gate and C/H levels. The mask-to-mask overlay accuracy between the gate and C/H masks is less than 20 nm; and after magnification correction, it is less than 10 nm. The critical dimension (CD) variations for 100 nm features measured in a 24-mm-square area are less than 100±6.0 nm at both levels. An overlay accuracy for PXL of better than 30 nm (|mean|+3σ) was obtained by the double-exposure method using the XRA volume-production x-ray stepper (Canon, Inc.) and 18 nm overlay ASET-NIST type x-ray masks (gate to contact). An IP reproducibility of less than 7 nm has been achieved for the EB-X3 itself (best data). The resist CD accuracy on the membrane for 4 Gbit DRAM patterns is better than 7 nm (Δ|mean|+3σ=6.6 nm). These data indicate that PXL has the potential for patterning 50 nm node devices with current tools.

Evaluation of a new X-ray stepper XRA

A new X-ray stepper XRA, which is equivalent to β-machine for proximity X-ray lithography (PXL), was installed at ASET Amagasaki-branch, and we have started the evaluation of its performance. The present alignment accuracy using global alignment method is 5.3±18.2 nm ( x) and 3.8±30.5 nm ( y). We evaluate the magnification correction function installed on XRA, and obtain a good linearity between the magnification and the applied force. By considering the dependence of the beam position and its size to the stored current, the critical dimension (CD) repeatability is improved from 15.5 to 5.5 nm (3 σ).

Light Source for X-ray Lithography.

Synchrotron-based proximity x-ray lithography (PXL) has been developed for more than 20 years. The lightsource for high-throughput PXL manufacturing is a synchrotron radiation (SR) ring with multiple x-ray steppers. From the late 1980 's to early 1990 's, many compact SR rings for PXL adopted superconducting magnetsfor achieving their compactness were built in the USA and Japan. The Association of Super-Advanced Electronics Technologies (ASET) was established in Japan in 1996 and ASET had several programs to promote PXL in NTT at Atsugi-shi. An e-beam X-ray mask writer “EB-X3” was developed by NTT and ASET fabricated PXL masks for 4Gbit DRAMs by the writer with 100nm design rules. Sumitomo Heavy Industries (SHI) has developed a new normal-conducting compact SR ring “AURORA-2S” with a beam current 500mA usinga novel idea to increase the field strength of normal-conducting magnets to 2.7 Tesla. In this paper, SR rings for PXL are introduced.

Overlay compatibility between two synchroton radiation steppers

A total overlay accuracy of less than 35 nm was obtained with two synchrotron radiation (SR) x-ray steppers developed by NTT and ASET. This value includes the pattern placement errors of two masks, errors due to the wafer etching process, and errors of two steppers that arise in exposure. The error factors affecting the overlay compatibility among SR x-ray steppers were analyzed using double-exposure experiments. The results show that the overlay error induced by the difference in runout between the two steppers was negligibly small and that the magnification change from shot to shot was caused by gap setting variations between the two steppers. These results demonstrate that a SR x-ray stepper can provide sufficient overlay compatibility for the 100 nm technical node and beyond.

Technology and performance of the Canon XRA-1000 production x-ray stepper

In order to introduce next generation lithography (NGL) into practical use, developing a production-worthy exposure tool is as important as developing a proof-of-concept system and mask technology. The progress of the exposure tool can be used as a measure of the status of each NGL choice. Canon has been focusing on the development of a proximity x-ray lithography (PXL) system for volume production use since the start of development. This has included development of magnification correction, development of a high-seed wafer stage, improvement of illumination intensity, extension of process latitude in the alignment system design, and development of an accurate environment control system. For example, we introduced a new method, using mechanical deformation of the mask, for magnification correction. We also developed a new wafer stage based on a field proven-high-speed wafer stage from our current optical lithography systems. A stepping time (including settling time) of less than 400 ms for a 50 mm step in a low-pressure helium environment was achieved. These achievements have been integrated into our volume production tool: the XRA-1000 x-ray stepper. The XRA-1000 has been installed in ASET and it will be used as the means for other process module development. Its results will demonstrate that PXL technology is the most practical NGL technology for volume production.

Overlay evaluation of proximity x-ray lithography in 100 nm device fabrication

In the semiconductor manufacturing of below 100 nm devices, overlay accuracy is a critical issue for high resolution on proximity x-ray lithography (PXL). We applied PXL for five layers (mark, isolation, gate, contact, and wiring) to fabricate a 100 nm n-type metal–oxide–semiconductor field effect transistor and evaluated the accuracy. Alignment exposure was done with optical heterodyne alignment and magnification correction. The overall overlay accuracy (mean±3σ) was 25–45 nm, and the best was about 25 nm for the contact layer. The overlay error was classified into several components to estimate the error budget. The analysis revealed that the alignment error of the x-ray stepper is less than 20 nm under various conditions in device fabrication. Individual x and y scaling errors caused by mask image placement and the difference of wafer-to-mask gap was from 18–35 nm: the error of common in-plane deformation was about 20 nm. The present accuracy may make it possible to fabricate devices with a 100 nm ground rule.

Evaluation of Overlay Accuracy for 100-nm Ground Rule in Proximity X-Ray Lithography

An evaluation was made of the overall overlay accuracy of proximity X-ray lithography (PXL), which includes mask placement error, the alignment error of the stepper, and deformation arising from wafer processing. The relative placement accuracy of X-ray masks between first and second layers was below 22 nm (3σ) for both the x- and y-directions. Exposure was carried out using an optical heterodyne alignment system and magnification correction, and an overlay accuracy below 30 nm (mean ±3σ) was obtained. The overlay error was broken down into various components to estimate the error budget. The analysis revealed the alignment error of the X-ray stepper to be about 25 nm, and the common in-plane deformation to be about 11 nm. The common in-plane deformation was almost the same as the relative mask placement accuracy without scale error. The accuracy obtained in this study clearly demonstrates the suitability of PXL for the fabrication of 100-nm ground rule devices. Moreover, the analysis indicates that it is also applicable to the 70-nm technology node.

Overlay performance of SR lithography in 64M DRAM layers

Synchrotron radiation lithography is applied to the real dynamic random access memory process. Two full chip x-ray masks whose overlay accuracy is 19.2nm (x) and 26.4nm(y) are prepared. By using Canon x-ray stepper, total overlay accuracy less than 45nm is obtained.

Advances in X-ray Lithography at ASET.

In ASET, in cooperation with NTT, we have developed key systems for X-ray lithography (XRL) such as X-ray steppers and a 100-kV electron-beam (EB) lithography system for x-ray mask fabrication. Mask-to-mask overlay accuracy within 20nm (3σ) between the gate- and contact-hole-layers for 1-Gb DRAM has been achieved. Critical dimension (CD) control of ±14nm for the gate and ±10nm for the contact-hole-patterns have also been realized. The advantages of XRL over advanced optical lithography, in terms of both resolution and process latitude, were demonstrated through computer simulation. The outcomes of several experiments have convinced us that XRL is a particularly promising technology for mass-producing 0.1-μm or smaller devices.

Performance of X-Ray Stepper for Next-Generation Lithography

X-ray steppers have been developed and improved to the point where they are capable of replicating 100-nm patterns. The alignment system, which employs an optical heterodyne system with two wavelengths, was improved by analyzing and reducing the crosstalk and asymmetry in the alignment optics. The overlay accuracy of the steppers with respect to themselves is better than 20 nm, regardless of the wafer mark depth. A magnification correction method using wafer temperature control and chucking force is implemented on the steppers. We have achieved an overall overlay accuracy of less than 35 nm, which includes the pattern placement error for two masks, etching process distortion, and alignment error. This demonstrates the capability of proximity X-ray lithography to meet the requirements of 100-nm lithography.

Magnification correction by changing wafer temperature in proximity x-ray lithography

Magnification errors arising from wafer processing cause significant overlay errors in lithography. We have devised a new, simple method that involves changing the wafer temperature to correct magnification errors on an x-ray stepper. A wafer is first heated in the wafer orientation unit, and then it is allowed to cool as it is transferred to the wafer stage due to the temperature difference between the wafer and the air in the chamber. For a given cooling time, the reproducibility of the wafer temperature is less than 0.3 °C; and the maximum available temperature change is 1.5 °C. The key point is that the placement of patterns on a wafer immediately after the chucking remains fixed regardless of further changes in the wafer temperature, provided that the force holding the wafer to the stage is strong enough. The maximum magnification error that can be corrected by this method is about 4 ppm. This is determined by the maximum available temperature change (1.5 °C) and the coefficient of linear expansion for Si. The results of overlay tests revealed the correction controllability to be below 0.2 ppm. When this correction method was used, the total overlay accuracy was found to be 38 nm (3σ), which includes mask placement error, etching process error, and so on.

Design of a beamline for soft and deep lithography on third generation synchrotron radiation source

ELETTRA is a third generation synchrotron radiation source. The energy spectrum allows the design of beamlines suitable for x-ray lithography from soft to hard x-ray wavelengths. An appropriate lithographic window for micro- and nanofabrication can be obtained by a combination of selected filters and mirrors. As the beamline is interfaced to a vertical x-ray stepper, a uniformity in the beam intensity better than 3% (3σ) in the horizontal direction has to be reached. The present beamline is designed by taking into account the main factors which can affect the beam quality, namely, thermal loading on mirrors and filters, slope errors, and surface roughness of the mirrors. The resulting lithographic resolution at soft x-ray wavelengths is better than 100 nm.

Mask Contamination Induced by X-Ray Exposure

To investigate the growth of mask contamination induced by X-ray exposure, masks are exposed to X-rays using an X-ray stepper. Independent of the exposure conditions, and in the presence or absence of resist coating, the main component of the contaminant is the salt (NH4)2SO4. On some resist-coated wafers, the contaminant growth is governed by the volume of sulfuric materials which evaporate during exposure. Contamination growth can be suppressed by applying a top coat to the resist, as using a resist that does not evaporate as a sulfuric out-gas. Gases were irradiated by X-rays in order to examine the change in environmental pollutants. In humid air, NO, NO2 and NH3 form rapidly up on X-ray irradiation. SO2 also increases up on X-ray irradiation, and changes into H2SO4. H2SO4 easily combines with NH3 which is produced by X-ray irradiation in humid air, and turns into the salt (NH4)2SO4. Mask coating was introduced to suppress the contaminant growth. The mask coated with TiO2 layer does not show contaminant growth, showing a drastic curtailment of contamination by the photocatalysis effect.

Mask membrane deflection caused by mask tilt during the wafer stepping motion in x-ray steppers

A new analysis method for deflection of the mask membrane during the wafer stepping motion in x-ray steppers has been developed. The deflection is caused by the wedge action of the gas film between the mask and the wafer if the mask tilts slightly relative to the plane on which the wafer is traveling. The deflection is not only detrimental to the throughput of x-ray steppers, it may also damage the mask membrane. Lees’ difference approximation method is applied to Reynolds’ equation for the gas film and the equation of motion for the membrane. The results calculated show good agreement with the experimental results (Ref. 1) obtained by Canon Inc. (1) The larger the tilt angle of the mask and the narrower the mask-to-wafer gap, the larger the deflection of the mask membrane. (2) In order to maintain the deflection of the membrane within 1 μm in the calculated mask, the tilt angle of the mask must be within 20 μrad under conditions of a 15 μm mask-to-wafer gap, a 30 mm stepping distance, and a wafer velocity of 100 mm/s. If the stepping distance becomes longer, the tilt angle must be more strictly controlled.

Characterization of pattern geometrical effect on line end shortening in x-ray lithography

Line end shortening in x-ray proximity lithography was characterized from the viewpoint of the pattern geometry. Mask patterns from 0.1 to 0.3 μm linewidth with different pattern density were fabricated by e-beam lithography and subsequent gold electroplating. The mask patterns were printed in commonly well known chemically amplified resists using the Satellite-800 x-ray stepper developed by Sumitomo Heavy Industry with the synchrotron storage ring at the Pohang Accelerator Laboratory at Postech. A former study [J. R. Maldonado et al., J. Vac. Sci. Technol. B 13, 3094 (1995)] had shown the effect of the absorber thickness and the gap between mask and wafer on line end shortening. In this article, the effect on line end shortening of the geometry of a neighboring pattern, such as pattern shape, pattern density and position of the neighboring pattern, was considered. Several resists with different postexposure bake (PEB) temperatures and different gaps between mask and wafer were examined to characterize their effect on line end shortening. The closer neighboring pattern induced larger line end shortening and also higher PEB temperature and larger gap caused larger line end shortening. A perpendicular neighbor pattern to the test pattern induced different line end shortening than a parallel neighbor pattern. In addition, two dimensional aerial image simulations were performed and line end shortening was extracted from the calculated aerial image. Simulation results showed good agreement with the experimental results.

130 nm and 150 nm line-and-space critical-dimension control evaluation using XS-1 x-ray stepper

Critical-dimension (CD) control for 130 and 150 nm line-and-space (L/S) patterns printed with the XS-1 x-ray stepper was evaluated using two kinds of resists: SAL606 and TDUR-N908. The largest factor in the CD variation was the nonuniformity of the x-ray dose, which was ±4.4% in a 20 mm×20 mm field. In replicated resist patterns, the CD variation due to mask-CD variation dropped to less than half the mask-CD variation because of Fresnel diffraction. For 130 nm L/S patterns, the CD variation for an 8-in.-diam wafer was 7.9 nm (3σ) for SAL606, and 12.0 nm (3σ) for TDUR-N908. For 150 nm L/S patterns, the values were 10.5 nm (3σ) and 14.6 nm (3σ), respectively. Although the major factors causing CD variation are dose nonuniformity and mask-CD variation, different resist materials reflect the effects of these factors to different degrees. The CD variation of 150 nm L/S patterns among seven wafers exposed on the same day was ±2.8 nm for SAL606 and ±3.7 nm for TDUR-N908, which is most likely due to fluctuations in the exposure dose of ±2.7%. The total CD variation for SAL606 was roughly estimated to be 9.0 nm (3σ) for 130 nm L/S patterns and 11.9 nm (3σ) for 150 nm L/S patterns. These values are within ±10% of the design rule.