The results show that 113In is preferentially sputtered over 115In in the near-normal direction by about 8.7 ± 2.7%o compared to the oblique direction. This includes a fairly large class of experimental situations from the erosion of filaments and other exposed surfaces in gaseous discharge tubes (eg the original observation by Grove) to the sputtering of solar system material exposed to the solar wind [2]. The sputtering we will describe in this work will exclusively involve the bombardment of multicomponent and possibly inhomogeneous metal surfaces with ions in the ke V energy range.
In the low current and energy range ke V used in these investigations this ratio is constant for any given target composition independent of the ion current.1·2. In the following discussion, the quantities of interest are the relative scattering efficiency of one component of the solid compared to some other component, measured as a function of the surface normal and the dose of the primary ion beam. We will see that the partial scattering efficiency for species in the solid is not simply proportional to their concentrations in the solid.
Clearly, if 6ij > 0, then component i is enriched with respect to component j in the sputtered flux compared to their respective concentrations in the bulk sample. It is therefore clear that if accurate measurements of the enrichment effect are to be made they must be made in the low ft:u.ence limit. In the experiments described in this work, the 'components' whose sputtering yields are compared are all isotopes of the same element.
The nature of the isotopic systems studied varies considerably both in terms of the composition of the target materials and the identity of the type of bombardment. l.
Chapter 2
The large mass difference of 80% was intended to accentuate any mass effect present when sputtering the foils. The analysis of the foils was performed by Caltech PANURGE ion microprobe (a modified Cameca IMS-3f)[15]. At each peak, the count rate is measured in the center and sides of the peak.
The isotope ratio, measured directly here, does not reflect the true isotope ratio of the material sprayed on the surface of the collector. While conducting the SIMS analysis, some concerns arose regarding the accuracy of the results. These tails may be a result of the relatively 'rough' nature of the processed carbon paper surfaces.
There is also evidence that the surface of the Mo samples was at least partially oxidized prior to bombardment. Rutherford Backscattering analysis of the collector sheets has shown that the sputtering yield of Mo isotopes increased by about a factor of 2 during the bombardment.
Chapter 3
Note that except in the 10.0 ke V Xe+ case, the relative composition of the scattered material goes to ,.._ 1.0 in the 'infinite' fluence limit. This is an indication that the high-dose, near-normal steady-state composition estimate of Rf{D is close to the true steady-state value. Furthermore, because we measure relative to the near-normal value, we also underestimate the absolute effect of fractionation compared to the mass ratio, because at steady state the integrated diffuse composition equals the mass ratio, but not the composition in the normal direction.
Finally, the 'fluence range', over which the fractionation changes significantly, is also virtually the same for both energies of a given ion species. It is clear that these systems are all qualitatively similar, at least in terms of dose dependence. At low doses, the lighter isotope is preferably sputtered at a fractionation of 30 to 50% until, after about 2 x 10 16 ions/cm2, the total fractionation goes essentially to zero.
Figures 3.2 to 3.8 show the angular dependence of the relative isotope ratios of 92Mo compared to 100Mo relative to the standard Rf{D, for a range of doses and ion/energy combinations. The general trend in the data in the 5.0 keV Xe+ case is that the fractiona-. In fact, it appears that additional ion beam dose has the only effect of sliding the entire curve down by about 50%0 in the infinite flow limit.
This may be due to changes in surface topology due to ion beam damage. There is evidence, namely Heavy Ion Rutherford Backscattering data, indicating that the surface of the samples suffered from some change in the high flow runs [20]. At any given level there is ,.., 25%0 difference between the isotopic ratios measured near the normal and far from the normal direction.
Other ion/energy combinations produce results that are not entirely different from the 5 kev Xe+ case. In general, the fractionation decreases with dose and, except for the low 10 keV Ar+ dose, the fractionation decreases somewhat with angle from the normal direction. It will be seen in Chapter 5 that these results stand in rather sharp contrast to the predictions of most analytical theories and are in at least partial agreement with other computer simulation efforts [21-23].
Chapter 4
Angular Dependence of the Isotopic Composition
The experimental conditions for the initial sputtering experiment of the liquid In:Ga system were very similar to those for the Mo system, except that the ion beam energies were generally slightly higher (25.0 keV Ar+) and the physical size of the carbon collectors was larger (collector radius of the cylinder ,.., 3.64 em), which allows near-normal measurements to converge within ,.., 5° of the normal direction. The dependence is similar to that of Mo, except that the effect is much smaller. Unlike the fractionation of Mo and In, Ga appears to prefer the heavy isotope at higher angles.
One could speculate that since Ga must cross the In layer to diffuse (probably the cause of the strong decrease in Ga at large angles) and the actual thickness of this layer grows as ,.., s(B), that the lighter mass of se is due to the smaller differential section transfers more efficiently through the layer than a heavier one.
Chapter 5
Predictions of Analytical T heory
One objection to applying this theory directly to the systems under consideration may be that if there is significant oxygen contamination at the surface of the Mo sample, the average mass will be significantly lower than the average mass of the Mo isotopes. In any case, the theory predicts no angular dependence of the isotopic fractionation and in most of the systems studied here there was clear angular dependence, even in the 'low fluence' limit. The results of the experiments performed here compare favorably with the results of MD simulations, indicating both larger fractionations than predicted by equation 5.1 and angular dependence of the isotopic fractionation[23) which is completely ruled out by analytical theory [27].
From a circular area A = 1rb~ax in this plane, the coordinates of the collision partner were selected from a uniform distribution. The identity of the partner was determined by the local isotopic and chemical composition of the target, which was allowed to vary with depth below the surface. The result of the calculation gives, as a 'bonus', the distance of closest approach between the two atoms involved in the collision.
This could optionally be used in the calculation as part of the electronic stopping of repulsive atoms according to the formalism of Oen and Robinson [32]. Even before the collision, one could optionally add electronic loss proportional to the path length and VB. The scattering directions were first determined in the local basis of the input atom and then converted to the target coordinate system.
The composition of the target can also be adjusted and 'relaxed' in response to bombardment by keeping track of the number of atoms born and stopped in discrete layers of the target. As the target composition profile changes, the spray yield naturally responds, eventually leading to a steady-state profile. It seems that the failure of the analytical theory cannot be explained simply by saying that it treats the surface or time evolution of the cascade too naively.
Significant progress has been made in the experimental determination of the angular dependence and the overall magnitude of isotopic fractionation due to sputtering. While the fractionation of the In isotopes falls with angle, in a manner similar to the Mo isotopes of the fixed target experiments, the Ga fractionation increases with angle indicating a different mechanism most likely with the In-rich layer on the surface is associated. Frozen In:Ga eutectic measurements will be difficult to interpret due to the presence of the In-enriched surface layer.
Appendix A
The target chamber pressure invariably increased with the addition of jet, but this additional pressure was probably dominated by the ion source gas and therefore made little contribution to the effective f All parts were rinsed in an ultrasonic bath with reagent grade isopropyl alcohol and dried with warm air before being placed in the UHV environment. After the fully assembled apparatus was loaded into the target chamber, it was rated at 10-6 Torr and then the entire chamber was baked at 300 co. At this base pressure, with the beam current density used in these studies, relation A.2 could be easily satisfied. Average enhancement over the indicated fluence range, calculated with respect to the stable isotope ratio at angles close to the target normal. They do include the uncertainty in the normalization factor and an additional effective uncertainty of 2.0% has been added to account for overall reproducibility [20].Appendix B
Angle 8 from Target normal Figure 3 .5
