CHAPTER: SIMS RSF SYSTEMATICS USING OUR "SOLIDSTATE" EAs

We in the SIMS community are often concerned with the values of positive and negative secondary ion yields or 'useful yields', and their practical implementation via the SIMS relative sensitivity factor, RSF (inverse ion yields). In the past, some of us have made measurements and plotted values of yield or RSF for positive atomic secondary ions vs ionization potential (IP), and for negative atomic secondary ions vs electron affinity (EA), and discussed the agreement or disagreement of the dependence of yield on exp IP or exp EA. For some elements, the dependence seems good and for others, it seems poor. Various explanations have been put forth (offered) for the apparent agreement or disagreement. The elements for which the agreement is good have low values of IP or high values of EA, that is, high (efficient) positive or negative atomic secondary ion yields. These elements exhibit primarily atomic secondary ions, with various molecular secondary ions having relatively lower intensities.

One issue in these considerations has been the affinity of the element for oxygen, frequently used as the primary bombarding species for positive secondary ion analysis because of the enhancement of positive secondary ions by the presence of oxygen atoms in the surface dipole layer from which the secondary ions are sputtered and to which electrons are easily lost as the secondary atoms leave the composite surface.

Another issue is ESD or electron stimulated desorption, a 'physical' effect and not an 'electronic' effect that can substantially enhance the positive ion yield of highly electronegative elements like F and Cl.

Another issue of study for some of us has been the relative yields of molecular secondary ions, especially as an approach to achieve enhanced detection of the elements of interest when the yield of a certain molecular ion is greater than that of the corresponding atomic ion, which is the case for the elements with higher values of IP and/or lower values of EA. The realization (or observation) that some molecular species have higher yields than the atomic ions could have a significant impact on RSF systematics and the comparisons of yield vs exp IP or EA.

A fixed density of atoms of an impurity introduced into a material during growth or via ion implantation is often used to create SIMS reference standards. If there is a fixed number of impurity atoms per unit volume or in a depth profile in the material being analyzed, and some are sputtered from the surface as atomic ions and some are sputtered from the surface in combination with either matrix atoms or with primary beam atoms equilibrated in the surface, or both, then the total number of impurity atoms sputtered from the surface is the total number available, or the total initial density (implanted fluence). The sum of the intensities of the atomic ion plus all possible molecular ions that contain atoms of the impurity must equal the original number (density) of the atoms of that impurity. If only a fraction of the impurity atoms are contained in the atomic ion and the total fluence is used to calculate an RSF from the atomic ion intensity, an RFS related only to that fraction of the impurity element atoms present in the matrix and not the total number in the matrix (implanted fluence) results.

Let us consider the concept that the fraction of the implanted fluence that is sputtered into each ion component that contains atoms of the implanted impurity fluence should be determined and used to calculate RSFs for each ion component, if comparisons are to be made with the total impurity atom density (implanted fluence) and the IPs or EAs of the element.

In the past, plots of the atomic RSFs or atomic useful ion yields have been generated. The values do not all lie on single lines vs IP or EA because not all secondary ions are atomic. Many are molecular [combined with atoms of the various matrix elements or the primary beam atoms that lie near the surface (dipole layer)].

For elements like Na, with low values of IP, nearly all implanted atoms are sputtered as the atomic ion and the RSF calculated for those elements fall on the predicted or expected line of ion yield or RSF proportional to exp IP. For elements for which only a portion of the implanted atoms are sputtered as the atomic ion, an RSF calculated based on just the sputtered atomic ions should be too high (by that fraction) (the ion yield will be too low). If the distribution of the total implanted atoms into all (or nearly all) molecular ion components in addition to the atomic ion is measured, and the fraction of each is used to calculate RSFs for each, more correct values of RSF may result. These adjusted values of RSF may then more likely fall on the predicted or expected line of dependence on exp IP. The same argument holds for negative ions and values of EA.

In the past, plots of log RSF (inverse ion yield) vs EA have not yielded a single line, but rather a scattering of points through which a single line could not be drawn. The pattern of these data points has been the same for many materials, indicating that the distribution of the impurity atoms between the atomic ion and the sum of all molecular ion components is approximately the same. This distribution or partitioning among atomic and molecular secondary ions may also be nearly the same for most or all materials.

Another concept is that a composite RSF that is characteristic of the total fluence (all secondary positive ions that contain the impurity atom) could be plotted vs EA or IP, using the combining expression (RSF)-1 = (RSFi )-1.

We tested these hypotheses by measuring the relative intensities of the major positive and negative molecular ions in addition to the positive and negative atomic ions, and calculating RSFs for the various components using only the fraction of the implanted fluence that corresponds to the fraction of the total sputtered fluence that is in each ion component, and also by calculating a single RSF for each implanted element using the combining expression given above. These values of RSF are given in the data of this work as RSF.

We then plotted the values of log RSF (adjusted) for the combined RSF determined as described above. Some results are plotted for Si in Fig. 1 for positive ions, and in Fig. 2 and for negative ions. These plots contain more data than previous plots that we have prepared; they contain data for nearly all of the stable elements, for example the lanthanide rare earths.

The data for positive secondary ions (O2 bombardment) show little change from previous plots - because the atomic positive secondary ions predominate over the molecular ions for most elements. Notable exceptions are the high IP elements C, N, F, Cl, and Br. These are also the elements for which ESD (electron stimulated desorption) is proposed to account for the observed enhanced positive secondary ion yields (as discussed elsewhere in this work). The RSFs for these elements are substantially decreased (because the secondary ion yields for the molecular positive ions are substantial). In particular, the RSF for N is moved close to the branch for the halide elements (F and Cl), which is itself lowered. The RSF for Br is lowered enough to fall below this branch, which is not satisfactory. The data for the lanthanide elements fall in a relatively small area on the plot - because their IPs are all close in magnitude. In summary, while a few encouraging changes occur, the overall result is not very helpful.

The situation for the negative secondary ions is different. Many of these data are significantly changed - because the molecular secondary ion intensities are often greater than for the atomic ions. The data (curve) plotted in Fig. 2 appear to fit better a power law dependence than the traditional exponential dependence. An expression that fits the data (curve) fairly well is RFS(-) = 5x1022/EA2.5 cm-3, where the EAs here are those determined in this work, the solid state EAs, which are somewhat different from some of the published values of vacuum EA. We note that there are published experimental vacuum values for only a few elements, and that the values for most elements are calculated or estimated, often with large errors possible. Our work includes values of solid state EAs for many elements, although the errors could be significant. The scatter for the many elements of Fig. 2 is considerably decreased, in addition to the indication of a power law dependence rather than an exponential one. The scatter increases at low values of EA, but is far less than it was when published (largely calculated or estimated) values of vacuum EA were used.

The break in the curve of log RSF for negative ions vs EA still occurs at the value of EA for which 100% of the atoms sputtered from the composite surface of the host matrix are ionized, which is for the elements with EA greater than about 2 eV, which are the elements of column 17 (F, Cl, Br, I). The elements of column 16 (S, Se, Te), plus Pt and Au, elements with values of EA near 2 eV, often fall near the break in the curve.

The breaks in the curve for positive secondary ions are discussed in WSM, and in this work where ESD is noted to account possibly for the enhanced positive secondary ion yield for the elements with high values of electron affinity (e.g F and Cl).