Roger Kelly

Criteria for bombardment-induced structural changes in non-metallic solids

Abstract Structural transformations in non-metallic solids induced by energetic heavy ions at intermediate (∼1013−1016 ions/cm2) and high (∼ 1017 ions/cm2) doses normally take the form of amorphization, crystallization, or stoichiometry changes. Using the results of 72 substances both a temperature-ratio and a bond-type criterion are shown to be successful in predicting the occurrence of amorphization or crystallization. The former, based on a physical model involving thermal spikes, states that amorphization should occur whenever the ratio (crystallization temperature)/(melting point) exceeds 0.30. The latter, of more or less empirical origin, states that amorphization should occur whenever the ionicity is ≤0.47. Stoichiometry changes are basically different from amorphization and crystallization and this is reflected in the criterion. It is shown that whether or not a substance (such as AgBr or TiO2) loses a component (such as Br or O) during bombardment correlates with the heat of atomization showing a...

The sputtering of oxides part i: a survey of the experimental results

Abstract The sputtering behavior of oxides is now fairly well understood in 12 cases. Three categories can be recognized, with some oxides belonging to more than one category. The eight oxides Al2O3, MgO, Nb2O5, SiO2, Ta2O5, TiO2, UO2, and ZrO2 have sputtering coefficients as expected from Sigmunds theory of collisional sputtering, at least provided the surface binding energy can be identified with the heat of atomization. The four oxides MoO3, SnO2, V2O5, and WO3 show high and (except possibly with SnO2) temperature-dependent sputtering coefficients such that the inferred surface binding energics are significantly less than the heats of atomization. For example, V2O5 has S = 12.7 atoms/ion for 10-keV Kr, so that Eb is apparently 1.3 eV/atom as compared with 5.7 eV/atom for atomization. This is taken as evidence for thermal sputtering. The five oxides MoO3, Nb2O5, TiO2, V2O5, and WO3 show preferential oxygen sputtering, such that they first amorphize, then lose oxygen, and finally crystallize as the phas...

An attempt to understand preferential sputtering

Abstract Four aspects of the sputtering phenomenon are considered with reference to binary targets. (1) Direct energy transfer from the incident particles to surface atoms would be expected to cause an enrichment in the heavier component. (2) Energy transfer from the substrate to surface atoms would be expected to cause an enrichment in the less abundant component. (3) Effects relating to surface binding energies are most easily analyzed for alloys, the prediction being here that a target is enriched in the less volatile component. Finally, (4) effects leading to vapor pressures are considered. The actual responses of a rather large number of binary oxide and alloy targets are most simply summarized by noting the following: (a) CdO preferentially loses. O whereas ZnO does not; (b) both Al 0.33 Au 0.67 Au 0.33 preferentially lose Al. As such we conclude that preferential sputtering with oxides is understandable in terms of vapor-pressure effects, though surface binding energies in principle also play a role. By contrast, preferential effects with alloys are understandable in terms both of direct energy transfer and of surface binding energies.

Thermal effects in sputtering

Abstract Of the various sputtering phenomena taking place when solids are bombarded, two have been attributed to thermal effects. The first is prompt thermal sputtering , which can be understood in terms of vaporization from the impact region in accordance with a briefly lived temperature excursion. Doubt is cast on some of the experiments with elemental targets originally intended to demonstrate prompt thermal sputtering in that they involved sufficiently high temperatures as to permit normal vaporization. Work involving polyatomic ions incident on elements is also problematical but recent studies by the groups of De Vries and Husinsky in which d N /d E was measured with Na, Na 2 O, and various halides appear to confirm the existence of prompt thermal sputtering. The second type of thermal effect, slow thermal sputtering , can be understood as due to vaporization from the impact region without the aid of a temperature excursion. Such vaporization is well documented with halides, where halogen is lost by electron sputtering and, for high enough target temperatures, the accumulating metal then gives rise to characteristic signals in the d N /d E spectra which we term “slow thermal sputtering”. For low enough target temperatures, halides tend to develop metallized surfaces. Oxides also show electron sputtering, and, as would be expected, there is a concurrent loss of metal at higher temperatures but metallization at lower temperatures. The yields with oxides are 10 6 –10 8 lower than with halides.

Theory of thermal sputtering

Abstract An energetic ion which is incident on a solid target causes a momentary temperature increase in the impact region, i.e., a so-called thermal spike occurs. Such spikes are capable of causing (or supplementing) disordering, precipitation, crystallization, electronic excitation, stoichiometry change, desorption, and sputtering, it being the contribution to sputtering that we consider here. The approach used is compatible with modern damage-distribution theory. Thus the temperature profile left by the incident ion is taken as a three-dimensional Gaussian with parameters appropriate to power-law scattering, and is used as the initial condition for solving the heat-conduction equation. Let us write this solution as T = T(t, y), where t is time and y is a dimension parallel to the target surface. The vaporization flux from a solid surface is taken as pn ½ (2π kT)−½ where p, the equilibrium pressure of a vapor species containing n atoms, can be written as poexp (−L/T), po and L are constants largely i...

Bombardment-induced photon emission from Al and Al2O3 targets

Abstract Ion impact on Al has been described by several authors as leading to the ejection of excited neutral Al atoms which in turn decay with the emission of photons. We have undertaken a study of these photons as arising from 10 keV Ne bombardment of Al and Al 2 O 3 . One group of experiments involved measuring photon intensities under conditions of equilibrium. By increasing either the current or the oxygen pressure, the intensities approached a well-defined limit such as would be expected if the intensities were proportional to the steady-state oxygen coverage. Remarkable here is that a good description of the experimental results was obtained using a model in which photons are assumed to arise only in the presence of oxygen. A second group of experiments involved the measurement of photon transients. Beam interruption led, when bombardment was resumed, to a transient with a half-time of ~ 3 sec and which was attributed to the amount of oxygen adsorbed on the surface having increased. Changing the current or pressure caused a transient which was about 8 times longer and which was attributed to the amount of recoil-implanted oxygen having changed. There is a close parallel, in the case of bombarded Al, between photon and ion yields; thus, the yields of Al + and Al ++ but not Al +++ are known to depend on the presence of oxygen. The role played by adsorbed or recoil-implanted oxygen is shown to be consistent with the band structures of Al and Al 2 O 3 provided the electron affinity of Al 2 O 3 is ⩽ 0.8 eV.

Ion-impact chemistry in the system titanium-oxygen (studies on bombardment-enhanced conductivity—III)☆

The bombardment of TiO2, whether poly- or single crystalline, with Kr ions leads to an altered surface layer having the following characteristics. It exhibits a high electrical conductivity, has the diffraction pattern of finely polycrystalline Ti2O3, is on the average 110 ± 20 A thick (for 30 keV Kr), and is indefinitely stable in air at room temperature. The formation of the layer is favored by increasing the target temperature. Formation is half complete at (6 ± 2) × 1016 ions/cm2, hence at a dose substantially greater than that for the half completion of sputter equilibrium ([7 ± 2] × 1015 ions/cm2). One model which could lead to Ti2O3 can be excluded fairly readily: this is thermal-spike stimulated vaporization, as the relevant vapor pressures are too low. More satisfactory is a model in which, due to either preferential oxygen sputtering or internal precipitation of oxygen, Ti2O3 nuclei are formed and grow. The reason that the stoichiometry is precisely Ti2O3 can be rationalized by an argument based on surface binding energies (Eb), in the sense that Eb for TiO2 to sputter congruently is 6.4 eV, to yield nuclei of Ti3O5 is 5.7, to yield nuclei of Ti2O3 is 5.1, and to yield TiO is 6.4. A similar rationalization holds also for impact-induced chemical changes observed or inferred with AgBr, CuO, Fe2O3, MoO3, U3O8 and V2O5, except that here thermal-spike stimulated vaporization cannot be excluded.

A gas-release study of the annealing of bombardment-induced disorder∗ (studies on bombardment-induced disorder—I)☆

Abstract Many oxides and diamond-type materials become disordered, e.g. amorphous, when bombarded with heavy ions at doses greater than 10 14 –10 16 ions/cm 2 . Though the most direct evidence for such disorder is probably electron diffraction, much information can also be gained by studying the motion of the ions used to produce the disorder. We have in this way established the following temperatures for disorder annealing (all in °C and for a time scale of 25°C/min): 730 with α-Al 2 O 3 , 445 with Cr 2 O 3 , 535 with α-Fe 2 O 3 , 480 with rutile, 325 with MgO, 285 with NiO, 470 with Ge and 720 with Si. A significant point is that the ratios of the disorder-annealing temperatures to those for atomic-scale cation self-diffusion are similar for most of the materials studied. The temperature widths of the annealing were also measured, and it could be shown that they were, except for MgO and NiO, compatible with single-jump kinetics, namely 45–70°C. It is concluded that the annealing process probably involves the growth of crystalline material towards the bombarded surface, such that the bombarding ions are swept out as the disorder-crystal interface reaches the surface. This model is of interest in implying that the activation enthalpies of the annealing, e.g. 52 ± 3 kcal/mole with rutile, are similar to those for self-diffusion in the disordered phase.

Recoil implantation from a thin source: I. Underlying theory and numerical results

Recoil implantation from a monolayer source, for example an adsorbed layer or a layer of altered stoichiometry, is considered. We introduce the incident ion current I, the fractional surface coverage characterizing the recoil source θ, the differential scattering cross-section dσ, and the integral distribution function F(x, ψ) for recoil-source atoms entering the target at angle ψ and stopping beyond x. The number of atoms implanted beyond x follows as H(x) = IθλNƒ σ F(x,ψ). The receding of the target surface at velocity υ due to sputtering can be allowed for by integrating over x: H(x, t) = (1υ) ∫ H(x′)dx′. For very high doses the result is a steady-state situation in which the number of implanted atoms is given by H(x, ∞). Numerical results of four kinds are presented: H(x, H(x, ∞), the half-time for the build-up of the steady-state situation, and the half-depth of the implanted atoms, in all cases for oxygen on the surface of Be, Al, Mo and W.

On the problem of whether excited states amongst sputtered particles are of thermal origin

Abstract Two models of a thermal nature are currently advocated to explain the presence of ions amongst sputtered particles. In so far as these models are valid, one must consider the possibility that excited neutrals amongst sputtered particles also have a thermal origin, whence that the yields of photons emitted by these excited neutrals will be given by Yield α (ω i /Q) exp(-e i /kT). Here ω i is the degeneracy of a state which lies e i above the ground state and Q is the internal partition function. We have tested the above equation experimentally and can summarize the results as follows: (a) The equation was satisfactorily obeyed by Be, BeO, Al, Al2 O3, Sc, GaAs, Y, In, and Tl, the inferred temperatures lying in the interval 3600–5900 K. (b) The range of temperatures agrees with four independent estimates from other work, (c) The temperatures are, on the other hand, insensitive to the variation of target mass from 9 to 204 u, ion mass from 20 to 84 u, ion energy from 4 to 16 keV, and target crystalli...