ELSEVIER Microeleetronic Engineering 35 (1997) 557-560 MICROEL£CTRONIC ENGINEERING Accuracy of structure transfer in deep X-ray lithography G. Feiertag, W. Ehrfeld, H. Lehr, A. Schmidt, M. Schmidt Institute of Microtechnology Mainz GmbH, Carl-Zeiss-Stral3e 18-20, 55129 Mainz, Germany Deep X-ray lithography with synchrotron radiation (DXRL) constitutes the key microfabrication process step in the LIGA technology. Microcomponents with a height of some gm up to several mm can be manufactured with sub-gm precision. The pattern transfer accuracy is governed by technological constraints like thermal mask de- formation as well as by various physical effects, e. g. Fresnel diffraction, emission of photo- and Auger elec- trons, fluorescence radiation, radiation scattering and divergence of the synchrotron radiation beam. A computer program has been developed to investigate the significance of these effects to the dose distribution in the resist material, which in turn determines the lateral structure resolution. The paper gives a brief introduction to the calculation procedure and outlines the weight of the different contributions with respect to transfer accuracy. It is shown that beam divergence and diffraction are much less important than the image blur caused by photoelectrons. Fluorescence radiation emitted from the mask mem- brane or the substrate contributes to the dose deposition in the resist if mask membrane or substrate consist of high atomic number material. Radiation scattering is negligible for resist layers which are less than some mm thick. A good agreement is found between calculated dose distributions and measured resist profiles. This allows a partial compensation of the above mentioned accuracy limiting effects in the mask design. I. INTRODUCTION 2. CALCULATION PROCEDURE Microstructures produced by means of the LIGA process are extremely precise and show smooth sidewalls when DXRL is applied. LIGA is therefore the appropriate technology to fabricate components for microoptical, microfluidic or micromechanical applications [ 1 ]. Best results are obtained in a simple shadow printing process, applying synchrotron radiation to project the high precision absorber patterns from a DXRL mask into a thick radiation sensitive polymer layer. Photoelectrons and Fresnel diffraction are the dominant effects which limit the structure transfer accuracy to about 0.2 gm for a 500 gm thick resist layer [1,2]. The thermoelastic deformation of a DXRL mask contributes to deviations from the ideal shadow print with less than 0.2 gm for a 500 gm thick resist layer if Beryllium or Diamond are used as mask membrane materials [3]. This paper presents a detailed analysis of the different contributions which limit the structure transfer accuracy. Model calculations are compared with experimental results which have been obtained by measuring the lateral distance of resist edges after DXRL and resist development using an SEM. A computer code was developed to simulate the contribution of the • Fresnel diffraction at the absorber edges, • divergence of the synchrotron radiation, • photo- and Auger electrons, • fluorescence radiation from mask or substrate, • scattering of radiation. The significance of these effects with respect to structure transfer accuracy is weighted for the above mentioned processes performing model calculations to obtain the dose distribution in the resist. The ra- diation parameters taken into account correspond to the spectral distribution of the synchrotron radiation source DCI, Orsay, France, with a dose of 5 kJ/cm 3 at the bottom of the resist, a vacuum window and a mask membrane made from Beryllium with a total thickness of 1000 gm and a Gold absorber height of 15 gm. Calculated results will be discussed for the irradiation of a 500 gm thick PMMA resist layer. 2.1 Diffraction Diffraction at a semitransparent absorber edge was calculated in the Fresnel approximation [4]. The cal- culation of the dose distribution in the resist con- siders the intensity distribution at an absorber edge 0167-9317(97)/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII: S0167-9317(96)00158-X