Geraint Owen

Proximity effect correction for electron beam lithography by equalization of background dose

Compensation for the proximity effect in electron lithography can be achieved by equalization of the backscattered dose received by all pattern points. This is accomplished by exposing the reverse tone of the required pattern with a beam diameter dc=2σb ×(1+ηe)−1/4 and dose Qc=Qe ×[ηe/(1+ηe)], where σb is the radius of the Gaussian spatial distribution function of backscattered electrons at normally exposed pixels, ηe is the ratio of backscattered to forwardscattered energy, and Qe is the dose delivered to normally exposed pixels. This correction method has been confirmed to work for 500‐nm features by computer simulation of electron beam exposure and development and by experiment on a raster scan electron beam lithography system.

Methods for proximity effect correction in electron lithography

A convenient and unambiguous way of characterizing the proximity effect is by use of the modulation transfer function. Six types of correction scheme are compared in this way, these being the use of high beam energies (>20 keV), of low beam energies (<20 keV), the use of multilayer resists, exact dose correction, ‘‘self‐consistent’’ dose correction, and the application of correction exposures. The main conclusions drawn are that the use of high beam energies reduces the proximity effect significantly; that exact dose correction, in addition to performing better than the ‘‘self‐consistent’’ technique, is computationally superior; and that correction exposures are effective, particularly in combination with other correction techniques.

Optimization of Pupil Filters for Increased Depth of Focus

The use of pupil filtering in optical lithography raises the question of an appropriate optimization method for such filters. We have suggested a pattern-independent procedure based on optimal energy concentration. Here, we discuss several different methods of implementing pupil filters in lithographic projection systems. We have assembled a simple telecentric test system and fabricated three filters optimized according to the suggested procedure. The results are shown to be in close agreement with predictions. We also present extensions of our calculations to large numerical apertures and non-unity magnification.

Application of the GHOST proximity effect correction scheme to round beam and shaped beam electron lithography systems

The GHOST proximity effect correction scheme has been applied to a round beam lithography system and also to a shaped beam system. It is shown that the image contrast of the GHOST scheme is constant at all spatial frequencies at which forward scattering is insignificant: the flatness of the modulation transfer function corresponds to the removal of the proximity effect. The effects of errors in the correction dose, the correction beam diameter, and the registration between the pattern and correction exposures have been calculated analytically, and it is shown that the allowable tolerances for these parameters are wide.

SUPPRESSION OF INTERFERENCE EFFECTS IN SPECTROSCOPY USING AN INTEGRATING SPHERE

Interference effects occur in the optical spectroscopy of thin film samples, and they can obliterate absorption spectra, especially if these are weak. A solution to this problem, originally proposed in 1976, is to place the sample inside an integrating sphere in the Edwards configuration so as to collect both reflected and transmitted light simultaneously. Surprisingly, the subsequent published literature on this technique is virtually nonexistent, despite its simplicity and effectiveness. This paper describes its use to measure absorption dips as weak as 2×10−4 in thin film samples.

Obtaining a physical two-dimensional Cartesian reference

A two-dimensional self-calibration method obtains Cartesian traceability for high-precision tools. The calibration method incorporates group theory principles yielding mathematical solutions to a physical Cartesian reference. The calibration method was developed by Stanford University, Hewlett Packard, IBM and funded by the Semiconductor Research Corporation. The method was applied to Leica’s LMS2000 and LMS2020 systems.

Performance results of an electron beam lithography machine and process by means of dc electrical test structures

Direct current electrical tests have been used to measure the performance of electron beam resists and an exposure system. Resist characterization is done by using electrical tests to measure the linewidth defined as a function of an incremental dose. The quality of different resists can be compared by means of δ, the slope of the normalized linewidth versus the logarithm of the incident dose curve. Data have been measured for three negative resists (PCMS 30, PCMS 200, and OEBR 100) and PMMA for two development conditions. Resistors fabricated from 60 nm chromium films, 300 nm polysilicon films, or 60 nm titanium films gave results within ±40 nm of one another for 500 nm lines. Without proximity effect correction, linewidth difference from design value is more than ±300 nm and the 0.5P μm lines were undefined for PMMA and the given development conditions, and 0.5P and 1.0P μm lines were undefined for PCMS 30 and the given exposure conditions. With proximity effect correction, all line sizes are within 100...

All‐reflective phase‐shifting masks for Markle–Dyson optics

Previous work using a Markle–Dyson projection system, operating at 248 nm wavelength, demonstrated 0.19 μm resolution with nonphase‐shifting masks. The use of Levenson‐type phase‐shifting masks should in principle, allow sub‐0.1 μm feature resolution using a k1 of 0.25. To investigate this possibility, a novel reflective phase mask was fabricated and used on the Markle–Dyson system to projection print 0.125 μm lines and spaces in photoresist. An analytical study has been carried out to determine the tolerance of alternating‐phase masks to errors in linewidth and phase.

The effect of resist contrast on linewidth error induced by e‐beam proximity exposure

In electron beam lithography, resolution is limited by two important factors: proximity effect and resist contrast. The relation between these two factors is investigated in a chemically amplified negative resist system. It has been discovered that proximity effect induced critical dimensions variation is not monotonically dependent on resist contrast; the worst proximity effects are apparent at medium contrast values. At higher contrast values, increasing the contrast can reduce the proximity effect; at lower values, increasing the contrast worsens the proximity effect. It has also been found that a parabola‐shaped curve is obtained when interproximity effect is plotted as a function of exposure dose. Dissolution rate ratio and overdevelopment effect were used to explore the impact of resist contrast on e‐beam proximity effect.

The Effect of Gaps in Markle-Dyson Optics for Sub-Quarter-Micron Lithography

The feasibility of Markle-Dyson optics has previously been demonstrated experimentally, using a prototype system in which the wafer is in contact with the mask. However, for semiconductor manufacturing, a gapped system is required. The introduction of a gap creates spherical aberration, which must be compensated. Two possible techniques for doing this are described. The first is to choose suitable materials for the Dyson lens and the mask, and this can work well unless the gap is much greater than 25 µm, in which case the residual aberration is unacceptably large. For this reason, the second compensation technique, the use of an aspheric mirror is preferred, since it affords complete correction. A gapped prototype system using an aspheric mirror is under experimental investigation. In this, the gap is controlled by a capacitance gauge, and the gauge is calibrated using an interferometric technique.