Nils E. Erickson

ESCA study of fractional monolayer quantities of chemisorbed gases on tungsten

Abstract X-ray photoelectron spectroscopy (ESCA) has been used in a study of CO and O2 chemisorbed on a polycrystalline tungsten sample. Working under ultra-high vacuum conditions, the surface was cleaned and then covered with known monolayer and fractional monolayer quantities of adsorbed CO and O2. The O(ls) and C(ls) spectral features were detected, and the influence of an adsorbed layer on the tungsten spectral features was determined. A chemical shift of 3.4 eV in the O(ls) line from chemisorbed CO is related to the different modes of bonding of CO to tungsten. A model calculation of the photoelectron yields expected from an adsorbed monolayer is in good agreement with the experimental results.

ESCA study of carbon monoxide and oxygen adsorption on tungsten

Abstract The chemisorption of both CO and O 2 on a clean tungsten ribbon has been studied using an ultrahigh vacuum X-ray photoelectron spectrometer. For CO, the energy and intensity of photoemission from O(1s) and C(1s) core levels have been studied for various adsorption temperatures. At adsorption temperatures of ∼100 K., the “virgin”-CO state was the dominant adsorbed species. Conversion of this state to more strongly-bound β-CO is observed upon heating the adsorbed layer to ∼320K. Thermal desorption of CO at 300⩽ T ⩽640 K causes sequential loss of α 1 -CO and α 2 -CO as judged by the disappearance of O(1s) and C(1s) photoelectron peaks characteristic of these states. Oxygen adsorption at 300K gives a single main O(ls) peak at all coverages, although at high oxygen coverages there exist small auxiliary peaks at ∼2eV lower kinetic energy. The photoelectron C(1s) and O(1s) binding energies observed for these adsorbed species are all lower than for gaseous molecules containing C and O atoms. For CO adsorption states there is a systematic decrease in photoelectron binding energy as the strength of adsorption increases. These observations are in general accord with expectations based on electronic relaxation effects in condensed materials.

X-ray photoelectron spectroscopic study of the adsorption of N2 and NO on tungsten

Abstract X-ray photoelectron spectroscopy (ESCA) has been used in a study of N 2 and NO adsorbed on a polycrystalline tungsten ribbon. The sample was flash cleaned under ultrahigh vacuum conditions, and cooled to either 300 K or 100 K for the adsorption studies. Large chemical shifts, as great as 8 eV, were observed between the N (1s) spectra associated with the weakly chemisorbed γ-nitrogen states and the strongly chemisorbed β-nitrogen states. Chemical shifts in both the N(1s) and O(1s) spectra suggest that NO is largely non-dissociatively chemisorbed at 100 K. In general, the binding energies of N(1s) and O(1s) electrons in the adsorbed layers are smaller than the binding energies for the same atoms in small gaseous molecules. In addition, the binding energies associated with the weakly-bound states of NO and N 2 are invariably greater than the binding energies associated with strongly chemisorbed species.

X-ray photoelectron spectroscopic study of the physical adsorption of xenon and the chemisorption of oxygen on tungsten (111)

Abstract X-ray photoelectron spectroscopy (ESCA) has been used to study the physical adsorption of Xe and the chemisorption of oxygen by W (111). An ultrahigh vacuum ESCA spectrometer has been modified such that thermal desorption behavior from the W (111) crystal can be directly compared with ESCA spectra of the adsorbed species. In addition, since the work function of a W (111) crystal covered with one monolayer of Xe is accurately known from previous work, the binding energy of the Xe (3 d 5 2 ) adsorbate level can be accurately compared to the gaseous Xe (3 d 5 2 ) level. When Xe is physisorbed to 1 monolayer the Xe (3 d 5 2 ) level exhibits a binding energy (relative to the vacuum level) which is 2.1 eV below that found for Xe (g). At lower Xe coverages the shift becomes monotonically greater, approaching 2.6 eV at a Xe coverage of 0.05. This 0.5 eV shift downward is accompanied by an increase of only 0.05 eV in adsorption energy as coverage decreases, and may be partially caused by the presence of ~ 10–20 % of extraneous adsorption sites other than W (111) which adsorb Xe with higher adsorption energy. The adsorption energy of Xe may also be increased by coadsorption of oxygen and the Xe (3 d 5 2 ) binding energy exhibits a corresponding shift downward as adsorbed oxygen coverage is increased to θo = 0.5. Electronic relaxation processes affecting the final state are dominant factors in determining the magnitude of the chemical shift upon adsorption, in agreement with the predictions of Shirley. The magnitude of the relaxation effect seems to be very sensitive to small changes in Xe adsorption energy. Similar effects have been seen for chemisorption of CO. The adsorption of O2 at 120 K by W (111) yields a single broad O(1s) peak whose line-width decreases with increasing coverage. The final spectra at θo = 1 monolayer are very similar to those obtained at temperatures of 300 K or above on polycrystalline tungsten.

Results of a joint auger/esca round robin sponsored by astm committee E-42 on surface analysis. Part II. Auger results

Abstract We report the results of a round robin involving kinetic-energy (KE) and relative-intensity measurements on high-purity samples of copper and gold by Auger-electron spectroscopy. These results were obtained using 28 different instruments or analyzers manufactured by four companies. We found that the spread in reported KE values ranged from 7 eV at a KE of 60 eV to 32 eV at a KE of ∼2025 eV. The total spread in reported intensity ratios ranged from a factor of ∼38 for the ∼6O eV and ∼92O eV peaks of Cu to a factor of ~ 120 for the ∼70 eV and ∼2025 eV peaks of Au. We have analyzed the observed trends in some detail. The systematic error of kinetic-energy measurements increases with kinetic energy for many instruments. Even though all instruments were adjusted with the use of 2 keV elastically scattered electrons, the spread in the reported positions of the ∼2025 eV Au peak indicates that the instruments were not adequately calibrated. Examples of erratic response were found in the measurements of relative intensities; it was believed, though not proved, that the more extreme values of intensity ratios were associated with instrumental malfunctions or operator mistakes. As in the similar ESCA round robin (Part I), the spread in reported Auger kinetic energies and relative intensities demonstrates clearly the need for standards (e.g., calibration methods, operating procedures, and data analysis) to ensure that data of known accuracy can be obtained routinely. Until suitable standards are available, interested individuals may find it useful to compare measurements using their own Auger or ESCA instruments with the group results and the trends found in the round-robin results. We have conducted an extensive round robin consisting of AES measurements on high-purity samples of Cu and Au. Participants were asked to measure the kinetic energies and relative intensifies of designated Auger peaks under specified conditions. This round robin was conducted contemporaneously with a similar ESCA round robin, the results of which have already been published [1]. The AES round robin had three principal objectives. First, it was intended to assess the overall accuracy of KE and relative-intensity measurements in a relatively straightforward AES measurement. An earlier round robin [2] with catalyst samples demonstrated substantial spreads of reported data, and it was believed that comparisons of data obtained for cleaned metallic samples should give a more accurate picture of the current state-of-the-art. With a larger number of participants in the present round robin than in the catalyst round robin, we in fact find a comparable spread in the raw data. The spread in the reported KE measurements is a function of kinetic energy, and ranges from 7 eV at a KE of 60 eV to 32 eV at a KE of 2 keV. The imprecision of the KE measurements is typically ∼1–3 eV. The total spread in the reported intensity ratios ranges from a factor of ∼38 for Cu (at an incident energy of 3 keV) to a factor of ∼120 for Au (at the same incident energy). The imprecision of the intensity ratios is typically less than 10%. Second, it was desired to determine the variation of AES intensities as a Cu sample was displaced along the analyzer axis with respect to its optimum position. Many of the participants found maxima in the intensities of the Cu ∼60 eV and ∼920 eV peaks at or very close to that sample position found to be optimum for the 2 keV elastic peak. Other participants found intensity maxima at sample positions up to ∼4 mm away from the 2 keV elastic-peak position; these participants found that when the sample was at the optimum position for the 2 keV elastic peak, the intensities of the ∼60 eV and ∼920 eV Cu peaks could be as low as 50% of the corresponding maximum peak intensities that were found when the sample was displaced. Third, it was intended to measure the Auger KE for the “adventitious” carbon that forms on initially clean samples in the ambient vacuum of each instrument. Few participants made this measurement, although it was observed that the spread of reported energies for the carbon Auger peak was comparable to that found for the low-energy (60–7OeV) peaks in Cu and Au. We have examined the KE and relative-intensity data in some detail. We found it useful to compute deviations of individual KE measurements from the median values of the reported measurements for the selected Auger peaks. These deviations were plotted as a function of kinetic energy and lines were drawn connecting data points obtained using the same instrument. Such plots can be regarded as “error functions” or “calibration curves” based on the use of the group median values as reference data. These curves indicate that for many instruments the error of KE measurements increases approximately linearly with KE (unlike the behavior found in similar plots for binding-energy measurements in the ESCA round robin, in which the error was often nearly constant or slowly varying with binding energy). The large spread (32 eV) in the reported positions of the Au M5N6,7N6,7 peak at a KE of ∼2024 eV was considered particularly significant, since most instruments were adjusted and aligned using elastically scattered electrons at an energy of 2 keV. This observation clearly indicates that the working KE scales of the instruments were not adequately calibrated using the elastic-peak method; this problem is believed to be due to the insufficient accuracy of the 2 kV power supplies or the voltmeters used to display the KE scales rather than to any intrinsic deficiencies in the use of the elastic-peak method. There were several examples, however, in which plots of the intensities of the ∼60 eV and ∼920 eV Cu peaks as a function of sample position had maxima for both peaks at a position different from that found optimum for the 2 keV elastic peak. These observations indicate that sample alignment by the elastic-peak method was not done with sufficient accuracy in some laboratories. Finally, while the imprecision in the locations of the elastic peak and of Auger peaks in the round robin was typically 1–3 eV, the overall inaccuracy of the KE measurement was usually substantially larger. Most participants found that ratios of peak heights for the low-energy and high-energy transitions in Cu and Au decreased slowly as the incident electron energy was increased from 3 keV to 8 keV. Some participants, however, obtained qualitatively different dependences on incident energy; these results were attributed to mistakes, instrument malfunctions, or to inadequate alignment. Our experience in the ESCA round robin indicated that operator mistakes or instrumental problems were responsible for most of the outliers in comparisons of measured intensity ratios. We suspect (although we have not proved) that the more extreme values of peak-height ratios in the AES round robin have a similar origin. The AES intensity data were analyzed to search for mechanisms that could account for the large range of reported intensity ratios. We considered several possible origins for the more extreme data values. First, we examined the reported peak-height ratios for Cu and Au, to search for possible variations of the instrumental transmission functions from their “ideal” values. Second, we considered whether relatively large amounts of residual surface carbon could account for the observed intensity ratios. Third, we tested whether the instruments which exhibited “non-ideal” behavior (probably because of significant stray magnetic fields or inadequate sample alignment) when the samples were translated parallel to the analyzer axis were also the ones which gave the more extreme peak-height ratios. Fourth, we investigated whether probable variations in the amplitude of the modulation voltage applied to the analyzers would modify significantly the ratios of the observed intensities. Fifth, we considered the effects of the differing energy resolutions of the analyzers in the round robin. Finally, we considered effects due to variations in surface roughness caused by the different ion-sputtering conditions used for initial cleaning of the samples. None of these factors alone could account for the more extreme variations of the peak-height ratios, although it is possible that some of these factors could affect certain specific instruments while a different combination of the factors could be important in other cases. Although we were unable to demonstrate conclusively the nature of instrumental artifacts or possible operator mistakes in the various intensity measurements, we believe that the spread in the reported intensity ratios is associated with specific measurement problems in particular individual laboratories. A variety of factors have been identified here to account for the more modest but nevertheless distressing range of intensity ratios (a factor of ∼2 for Cu and of ∼5 for Au) for the majority of the participants. The spreads in the energies and relative intensities of Auger and photoelectron peaks in this AES and the previous ESCA round robin indicate clearly that improved calibration and operating procedures are required for both Auger and ESCA measurements. Published data, for example, are of little value unless credible statements of accuracy can be associated with the numerical results. We hope that the standards needed for improved measurements can be developed by the ASTM Committee E-42 on Surface Analysis together with other interested parties.

X‐ray photoelectron spectroscopy/Ar+ ion profile study of thin oxide layers on InP

The effect of incremental ion bombardment on the surface layers of an aqua regia etched InP sample was studied by monitoring the components of the In 3d5/2 and O 1s x‐ray photoelectron spectroscopy (XPS) lines as the sample was bombarded with low energy (1 keV) Ar+ ions. The changes in the stoichiometry of the surface produced large shifts in the position of the In 3d and O 1s lines that were not paralleled by shifts in the P 2p line. Analysis of these shifts indicated that the surface was covered with a mixture of indium hydroxide and indium phosphate, with the phosphate closer to the InP substrate. It is proposed that this layer structure is due to differences in the dissolution rates of the oxidation products in the acid etch and the effect of the distilled water rinse. It may be possible to alter the composition of such oxides by carefully tailoring the etch conditions to optimize the kinetics for the particular oxide phase required. The analysis of the XPS lines also showed that the InP substrate was...

Oxide formation on ruthenium observed by TDS and ESCA

Oxygen adsorbed at room temperature on a Ru(1010) surface shows a single peak in its thermal desorption curve, the peak maximum shifting to lower temperature with higher initial concentration in accordance with second order kinetics. Additional peaks in the thermal desorption spectrum of oxygen from Ru(1010) were observed after exposure of the ruthenium above 600 K to oxygen. These peaks, desorbing with first order kinetics, are attributed to oxide formation. The change in the size of the peaks with respect to the temperature of the surface during oxygen exposure is correlated with an oxygen sticking coefficient that is constant up to 850 K and decreases thereafter. The ESCA spectrum of a surface with the ’’oxide’’ showed an increase in the height of the 0(1s) peak but no chemical shift. Sulfur adsorbed on Ru(1010) prevents the adsorption of oxygen.

An XPS study of formaldehyde and related molecules adsorbed on the surface of a tungsten (100) single crystal

Abstract X-ray photoelectron spectroscopy has been used to study the adsorption and catalytic decomposition of formaldehyde on a W(100) single crystal. Comparison with the O(1s) spectra of CO(ads), CO2(ads) and O(ads) has been carried out in an attempt to understand the surface complexes formed from H2CO. It has been shown that H2CO dissociates at 100 K upon adsorption up to ca. 1/2 monolayer. Above this coverage, condensation of undissociated H2CO occurs. A surface complex leading to the liberation of CO2 from the formaldehyde layer has been detected by XPS. However, no complex uniquely related to an intermediate which yields a small quantity of CH4 has been detected by XPS.

The Use of X-Ray Photoelectron Spectroscopy (ESCA) for Studying Adsorbed Molecules

An ultrahigh vacuum X-ray photoelectron spectrometer has been used to study a number of cases of adsorption on tungsten single crystals. The choice of adsorbates spans a wide range from dissociative chemisorption to nondissociative chemisorption to physisorption.