Contribution of vaporization and boiling to thermal-spike sputtering by ions or laser pulses Roger Kelly* and Antonio Miotello Dipartimento di Fisica dell’Universita ` di Trento and Istituto Nazionale per la Fisica della Materia, I-38050 Povo, Trento, Italy Received 18 March 1999 Here we consider what, in our terminology, we designate as normal vaporization, normal boiling, and phase explosion. In the case of vaporization, one is dealing with the emission of particles atoms or moleculesfrom the extreme outer surface of either a solid or liquid for any temperature exceeding 0 K. In the case of boiling, one is at least ideallydealing with heterogeneously nucleated bubbles which diffuse to the outer surface of a liquid or solid and then escape, the latter being possible for temperatures equal to or exceeding the boiling temperature ( T b ). In the case of phase explosion one is dealing with the consequences of what happens when a liquid approaches the thermodynamic critical temperature ( T tc or T c ), and massive homogeneous nucleation takes place. Although these three mechanisms have been reviewed in reasonable detail in recent work, we will here present evidence, apparently not previously considered, that boiling, whether the distance scale is atomi- cally small 5–15 nm, as for laser-pulse impact on a metal in the absence of thermal diffusionor much larger, has a prohibitive kinetic obstacle because it requires bubble diffusion if the bubbles are formed other than at the outer surface. That is to say, boiling will never be a significant process whether with ion or laser-pulse impact. This leaves vaporization and phase explosion as the only possible thermal-spike processes capable of expelling material from an ion- or laser-pulse bombarded surface in a significant quantity. But even with vaporization it can be shown that a kinetic obstacle, although not as severe as for boiling, will enter. The final result is that only phase explosion will normally be relevant for sufficiently short time scales. S1063-651X9906209-1 PACS numbers: 79.20.Ds, 64.70.Fx, 64.90.+b, 66.90.+r I. INTRODUCTION We have already adequately reviewed the ensemble of mechanisms which arise when ion beams, electron beams, or laser pulses interact with solid or liquid surfaces. Each type of incident particle or radiation leads to a variety of observ- able phenomena. For example, with ions or electrons, one can expect sputtering, mixing, or composition change, each of which will have varieties such as ballistic, thermal-spike, residual-defect-induced, electronic, or exfoliational. See Table 1 of Ref. 1. We here use the term ‘‘thermal spike,’’ rather than simply ‘‘thermal,’’ to emphasize that we are deal- ing with very short-lived effects. The situation with laser-pulse impact is slightly different. Considering just the primary interactions, one has the same family of observable phenomena Table I of Ref. 1. The emphasis is not the same, however: there is a marked impor- tance of thermal-spike processes with laser pulses and a minimal importance with ions. We associate this difference both with the volume of disturbance very small for ions and with the time scale always short for ions and sometimes short for laser pulses. Both ions and the very shortest laser pulses 100 psdo not provide enough time for either boil- ing or vaporization, whereas the converse is partly true for laser pulses having durations of 10–100 ns. See, for ex- ample, Table 4 of Ref. 1. Also, one must not forget that even with longer laser pulses 10–100 nsit is unclear whether boiling is ever significant. This is perhaps the most important point that will be made in the work presented here Sec. II and Table I in particular. What we seek to do here is settle, hopefully in a definitive way, the relevance of normalvaporization and normal boiling. Why this is a problem arises, in our opinion, from a widespread misreading of thermodynamic tables combined with a lack of understanding of bubble diffusion. One would be surprised at how frequent is the use of the term ‘‘vapor- ization’’ to include all possible processes leading to a liquid- to-vapor transition, as well as the assertion that this transition begins abruptly and rapidlyat T b , the boiling temperature. II. GENERAL COMMENTS ON MATERIAL REMOVAL BY VAPOR FORMATION We will show in Tables II, III, IV, and V why there might be confusion in quantities relevant to vapor formation. Table II serves to emphasize that T m , the melting temperature, is well defined. By contrast, T b , the boiling temperature, varies widely depending on the ambient gas pressure. Table III serves to emphasize that normal vaporization is sometimes able to account for significant material loss. At least this is so if the temperature is high enough and the time scale is long enough. For example, 100 ns is sometimes sufficient, but 1 ns is nearly always too little. Older data suggest, if the time scale is moderately short 1 ns, that even phase explosion may possibly be unim- portant 2–5. If this were so, then a liquid-to-vapor transi- tion would be in general excluded, and the only possibilities left would be electronic processes. There is some indication, however, that the criterion ‘‘1 ns’’ is not realistic Sec. II C and Tables IV and V. A. Normal boiling Here we deal with a process, namely normal boiling, that is already adequately discussed Sec. 5.3.2 of Refs. 6,7 *Author to whom correspondence should be addressed. Tel.: +390461881629. FAX: +390461881696. Electronic ad- dress: kelly@science.unitn.it PHYSICAL REVIEW E SEPTEMBER 1999 VOLUME 60, NUMBER 3 PRE 60 1063-651X/99/603/261610/$15.00 2616 © 1999 The American Physical Society