Sputtering equilibrium

The bombardment of a sample with a dose of high energetic primary ions (1 to 20 keV) results in the destruction of the initial surface and near-surface regions (Sect. 3.1.1). If the primary ion dose is higher than 10 ions mm the assumption of an initial, intact surface is no longer true. A sputter equilibrium is reached at a depth greater than the implantation depth of the primary ions. The permanent bombardment of the sample with primary ions leads to several sputter effects more or less present on any sputtered surface, irrespective of the instrumental method (AES, SIMS, GDOES. ..). 

Compared with XPS and AES sputter depth profiling After achieving sputter equilibrium, and until a layer with different sputtering behavior is reached [3.59], the SN flux represents stoichiometry and not altered layer concentrations evolving because of preferential sputtering effects. 

Equation (3.19) is valid for any species X. If, however, a multi-component material emits only atomic SN after attaining sputter equilibrium, X stands for elements and atoms only, and the total sputter yield Yean be written as ... 

Taking atomic sputtering into account the proportion of the particles emitted as molecules is negligible and the partial sputtering yield for element A in sputter equilibrium can be determined by use of ... 

Fig. 4.49. Comparison of relative intensities measured with IBSCA and SNMS. Conditions sputter equilibrium after bombardment with 5 keV Ar" the samples were oxidic glasses with different content of Na (0.1-7.4 at%) and Pb (4.4-22.1 at%).

Sputtering After ignition of the discharge a bum-in time is required to achieve a sputtering equilibrium. The surface layers have to be removed first, but once equilibrium is reached the discharge penetrates with constant velocity into the sample. Preburning times are usually up to 30 s for metals (at 90 W in argon, zinc 6 s brass ... 

Local Thermodynamic Equilibrium (LTE). This LTE model is of historical importance only. The idea was that under ion bombardment a near-surface plasma is generated, in which the sputtered atoms are ionized [3.48]. The plasma should be under local equilibrium, so that the Saha-Eggert equation for determination of the ionization probability can be used. The important condition was the plasma temperature, and this could be determined from a knowledge of the concentration of one of the elements present. The theoretical background of the model is not applicable. The reason why it gives semi-quantitative results is that the exponential term of the Saha-Eggert equation also fits quantum-mechanical expressions. 

The talk will briefly review some of these developments ranging from high temperature equilibrium plasmas to cool plasmas, PECVD, ion implantation, ion beam mixing and ion assisted etching and deposition. Brief consideration will also be given to sputtering and ionised cluster beam deposition techniques in inorganic synthesis. 

Molecular dynamics simulations yield an essentially exact (within the confines of classical mechanics) method for observing the dynamics of atoms and molecules during complex chemical reactions. Because the assumption of equilibrium is not necessary, this technique can be used to study a wide range of dynamical events which are associated with surfaces. For example, the atomic motions which lead to the ejection of surface species during keV particle bombardment (sputtering) have been identified using molecular dynamics, and these results have been directly correlated with various experimental observations. Such simulations often provide the only direct link between macroscopic experimental observations and microscopic chemical dynamics. 

Synthesis of carbide and nitride films by reactive sputtering. Table 14.3 presents a summary of the experimental conditions for the formation of carbides and nitrides. The phases formed are different from those in Table 14.2. As sputtering is a non-equilibrium technique, it was possible to synthesize the 6-MoC and the fi-WCi x carbides and the... 

Finally, the carbide phases obtained by reactive sputtering are not the same as those obtained by reaction of the metal film with the gas. In this last case it is indeed possible to be closer to the equilibrium thermodynamic conditions than in the sputtering method. 

To examine the solid as it approaches equilibrium (atom energies of 0.025 eV) requires molecular dynamic simulations. Molecular dynamic (MD) simulations follow the spatial and temporal evolution of atoms in a cascade as the atoms regain thermal equilibrium in about 10 ps. By use of MD, one can follow the physical and chemical effects that influence the final cascade state. Molecular dynamics have been used to study a variety of cascade phenomena. These include defect evolution, recombination dynamics, liquid-like core effects, and final defect states. MD programs have also been used to model sputtering processes. 

Such approaches rely on so-called relative sensitivity coefficients (RSFs), ratios of the difference between the sensitivity of various elements, and these cannot be considered as fundamental constants. In fact, they provide no more than a quantitative measurement of the deviation of the method s result from the amount of substance, as issued from primary methods (if available). Other near-equilibrium plasma methods for the analysis of solids (glow discharge, sputtered neutrals secondary ion mass spectrometry) produce quite acceptable results for analytical practice. 

Possibility for the preparation of multi-component alloys with concentrations of elements much higher than their equilibrium values. This is due to the fact that the film composition is mainly determined by the relative contents of components in the stream of sputtered atoms and by the adhesion factor, but is not defined by thermodynamic equilibrium. 

Many attempts have been made to quantify SIMS data by using theoretical models of the ionization process. One of the early ones was the local thermal equilibrium model of Andersen and Hinthome [36-38] mentioned in the Introduction. The hypothesis for this model states that the majority of sputtered ions, atoms, molecules, and electrons are in thermal equilibrium with each other and that these equilibrium concentrations can be calculated by using the proper Saha equations. Andersen and Hinthome developed a computer model, C ARISMA, to quantify SIMS data, using these assumptions and the Saha-Eggert ionization equation [39-41]. They reported results within 10% error for most elements with the use of oxygen bombardment on mineralogical samples. Some elements such as zirconium, niobium, and molybdenum, however, were underestimated by factors of 2 to 6. With two internal standards, CARISMA calculated a plasma temperature and electron density to be used in the ionization equation. For similar matrices, temperature and pressure could be entered and the ion intensities quantified without standards. Subsequent research has shown that the temperature and electron densities derived by this method were not realistic and the establishment of a true thermal equilibrium is unlikely under SIMS ion bombardment. With too many failures in other matrices, the method has fallen into disuse. 

Just as the previously cited work illustrated the role of matrix selection in the quality of the mass spectra, Mei and Harrison [40] studied the underlying equilibrium involved in the formation and removal of metal oxides from the spectra of compacted oxide samples. The analyte for these studies was La203, chosen because of the high affinity of lanthanum for atomic oxygen, a product of the dissociation of both residual air and water vapor in compacted samples. Because the La-O bond strength is very high (8.30 eV), this oxide represents a case in which the dissociation of the original analyte would be difficult as well. On the basis of previous studies [41], it was assumed that approximately 50% of the lanthanum species sputtered from the surface would be free La atoms, and the other half would exist as oxides of one form or another. 

---