chapter 38, Potential Energy Sputtering of EUVL Materials

Excerpt

38.1 Introduction

Of the many explanations suggested for the erosion of critical EUVL components, potential energy (PE) damage remains relatively uninvestigated. Unlike the familiar kinetic energy sputtering, which is a consequence of the momentum transferred by an ion to atoms in the target, PE sputtering occurs when an ion rapidly collects charge from the target as it neutralizes. Since the neutralization energy of a singly charged ion is typically on the order of 10 eV, PE effects are generally neglected for low-charge-state ions, and hence in the bulk of the sputtering literature. As an ion's charge state is increased, the PE increases rapidly; e.g., PE(Xe¹⁺) = 11 eV, PE(Xe¹⁰⁺) = 810 eV, PE(Xe²⁰⁺) = 4.6 keV. By comparison, the binding energy of a single atom on a surface is typically about 5 eV, so even inefficient energy transfer mechanisms can lead to large quantities of material being removed; e.g., 25% efficiency for Xe¹⁰⁺ corresponds to ≈40 atoms∕ion. By comparison, singly charged Xe ions with ≈20 keV of kinetic energy sputter only about 5 atoms∕ion at normal incidence, and less than 1 atom∕ion at typical EUV source energies (as determined using TRIM and sputter-yield data tables).

EUV light sources are optimized for producing approximately 10¹⁶ Xe ions per shot with an average charge state of q = 10 in the core plasma. At operational rates of ≈10 kHz, the number of ions produced per second becomes a whopping 10²⁰. Even if only one in a billon ions reaches the collector, erosion rates could exceed ≈10¹² atoms per second, severely reducing the collector lifetime (for an average yield of 10 atoms∕ion). In addition, efforts to reduce contamination effects may contribute to reduced neutralization and even larger PE damage rates. In order to provide accurate estimates for collector lifetimes and to develop mitigation schemes, National Institute of Standards and Technology (NIST) is working to understand and quantify potential energy damage mechanisms on materials relevant to EUVL. Accurate PE damage rates can then be used for projecting component lifetimes as source plasma conditions are modified and characterized.

This chapter will serve to provide an introduction and some background to the physics of highly charged ions (HCIs) and some of the relevant experimental work in the literature. It will first provide a brief background and an overview of the interaction of HCIs with solids as it is currently understood. Secondly, it will present current data from screen test measurements performed to isolate and evaluate the effects of PE damage on critical EUVL materials. It will then speculate on the implications of work to date and the outlook for EUVL development, and finally summarize.

38.2 Interactions of HCIs with Solids

When singly charged ions interact with solids, the transfer of the ions' forward momentum is the dominant damage-forming mechanism, creating lattice dislocation and sputtered atoms. Kinetic energy sputtering is a thoroughly studied and well-understood process for most elements and kinetic energies. Extensive experimental work has generated data on sputter rates as a function of kinetic energy for nearly every known combination of elements in the periodic table. The compilation of these data has resulted in the accumulation of accurate parameters for use in analytical fits such as Yamamura's semiempirical model (based on Sigmund's theory of sputtering). With increased interest in technologies that employ ion energies nearer to sputtering thresholds and at nonnormal incidences, the models have been refined to increase their accuracy in the low-energy and light-ion regimes. In addition to semiempirical fits to actual data, the Monte Carlo simulation SRIM (TRIM) has been widely tested and accepted as an accurate benchmark for quantitatively describing ion-solid interactions, particularly stopping ranges and sputter yields. In recent history, SRIM has been used to generate accurate predictions at arbitrary energies and incidence conditions for even further refinement of semi-empirical formulas that more accurately model the low-energy (threshold) regime and the low-mass-ratio regime. This vast compilation of knowledge, fit functions, and simulations makes estimation of kinetic energy damage relatively easy and accurate, but such simulations neglect charge and PE effects entirely.