P. M. Marcus

Determination of the structure of ordered adsorbed layers by analysis of LEED spectra

Abstract Recent work on the structure of ordered overlayers of chalcogens on nickel is reviewed, and discussed as a convincing example of the use of LEED intensity spectra for surface structure determination. Guidelines are suggested for successful application of LEED to surface structure, and the special suitability of the muffin-tin potential for the LEED calculation is brought out. The model of the complex crystal potential used for the calculation, and the systematic determination of the four parameters of the model from the data for clean Ni are described. The application to chalcogen overlayers is illustrated by the case of c(2 × 2)S and c(2 × 2)O on Ni(001) surfaces, and the correspondence between experiment and theory is exhibited for the intensity versus energy spectra of 1 2 1 2 beams. The correspondence is close and detailed; the complex shapes of broad features are reproduced and the average of the magnitude of the deviation between peaks in measured and calculated spectra is less than 2 eV, whereas the accidental correspondence of two spectra is shown to fluctuate around 4 eV. The observed values of the bond lengths of O, S, Se, Te on Ni are shown to be plausible on the basis of the covalent radii and the expected deviations from them. Results for other surfaces and other structures are tabulated. In all cases but one [O/Ni(110)] the chalcogen occupies sites that the next layer of Ni atoms would use.

Determination of the c(2 × 2) structure of sulfur on Fe{001} by low-energy electron diffraction

In the early stages of reaction with sulfur, a clean Fe{001} surface develops a c(2 × 2) superstructure. A low-energy electron diffraction analysis of this structure leads to a model in which the S atoms lie in the four-fold symmetrical sites on the Fe{001} surface, the S-Fe interplanar spacing being 1.09 ± 0.05 A and corresponding to an effective radius of 1.06 A for the chemisorbed S atoms. In contrast to Fe{001} 1 × 1-O, the first interlayer spacing of the substrate here is not significantly expanded.

Atomic and electronic structure of a surface alloy - comparison with the bulk alloy

Abstract A low-energy electron diffraction intensity analysis of the c(2 × 2) structure formed by 1 2 monolayer of Au on a clean Cu {0 0 1} surface reveals a single mixed-layer structure which can be properly called an ordered surface alloy. The layer is buckled, with the Au atoms located 0.1 A outwards from the Cu atoms and the distance between the latter and the second (all Cu) layer expanded by 4.2% with respect to the bulk. This structure of the surface alloy is the same, except for the smaller lattice constant, as that of the first two layers of a {0 0 1} surface of Cu3Au. The third atomic layer is shown to have no ordered Au content. Photoemission experiments with synchrotron light show that the d band of the surface alloy is noticeably narrower than the d band of the bulk-plus-surface Cu3Au alloy.

Atomic underlayer formation during the reaction of Ti{0001} with nitrogen

Abstract Changes in the {0001} surface of Ti during exposure to nitrogen gas are monitored by low-energy electron diffraction (LEED) and Auger electron spectroscopy (AES). Evidence is provided for the formation of a Ti{0001}1 × 1-N phase followed by a Ti {001} 3 × 3−30°-N structure. A LEED intensity analysis of the Ti{0001}1 × 1-N phase reveals that the N atoms penetrate into the octahedral interstitial holes underneath the first layer of Ti atoms. This is the first ordered monoatomic underlayer found in the earliest stages of any solid-gas interaction. The surface structure bears close resemblance to that of {111} planes in bulk TiN. We find a Ti-N bond length of 2.095 A to be compared with the value of 2.120 A in bulk TiN. The analysis of the Ti{0001}1 × 1-N structure indicates that Ti {0001} 3 × 3−30°-N is not a low-coverage phase. The importance of recognizing the existence of 1 × 1 phases prior to the formation of superstructures is emphasized, and some procedures for extracting the information from experimental observations are discussed.

Trends in metal surface relaxation

Abstract New LEED (low-energy electron diffraction) results are reported for Fe{310} and Fe{210} surfaces, including multilayer relaxation of atomic planes both perpendicular and parallel to the surface. The structures of six different surfaces of iron are now known. A comparison of the results yields relaxation trends: top-layer relaxation is found to increase monotonically as the surfaces become more open; for the higher-index surfaces {211}, {310} and {210} “decay curves” of relaxation as functions of depth into the surface show a surface-independent decay length (the depth at which the crystal returns to a bulk-like arrangement) of approximately 2 A.

Structural properties of epitaxial films of Fe on Cu and Cu-based surface and bulk alloys

Thin films of Fe were grown epitaxially at room temperature on Cu{001}, on the ordered surface alloys Cu{001}c(2×2)-Au and Cu{001}c(2×2)-Pd, and on the bulk alloy Cu3Au{001}. The maximum thicknesses attained in well-crystallized Fe films were 18 layers on the first three surfaces but only 3 layers on Cu3Au {00}. Low-energy electron diffraction (LEED) intensity data from 1-, 2- and 3-layer films on Cu{001&} could not be fitted by models with complete layers of Fe, suggesting that the growth of these ultrathin films was not layer-by-layer. The 12-layer films were found to have a tetragonally distorted face-centered-cubic (fcc) structure with a= 3.61 A, c= 3.54 A and 4%-expanded first interlayer spacing. Analysis of the elastic strain gave an equilibrium lattice constant of 3.59 A for fcc Fe at room temperature. Comparison with lattice constants from total-energy band calculations shows that the Fe cannot be in the nonmagnetic phase, but could be in the ferromagnetic phase, or possibly in an antiferromagnetic phase with the same lattice constant. It is suggested that the first interlayer spacing is enlarged owing to the larger magnetic moment of the first layer.

Epitaxial growth of body-centered-cubic nickel on iron

Abstract Studies by low-energy electron diffraction (LEED) and Auger electron spectroscopy of nickel films grown in ultra-high vacuum onto a clean Fe{001} surface show that the films have the body-centered cubic structure with the same lattice constant and the same multilayer relaxation as the clean substrate, as long as they are thinner than about 6 layers. LEED intensity analyses show that the multilayer relaxation of both clean Fe{001} and 3-layer thick Ni films involves 5% contraction of the first and 5% expansion of the second interlayer spacing. These new values of the multilayer relaxation of clean Fe{001} represent an improvement over previous determinations. Thicker Ni films, up to 100 layers, have a complicated structure that is neither b.c.c. nor f.c.c. Short anneals at temperatures between 200 and 650°C cause rapid diffusion of Ni into the Fe substrate with little evidence for formation of the stable f.c.c. phase of Ni.

Relaxation at clean metal surfaces

Abstract An electrostatic model of the energy of the surface layers of a metal is shown to describe in detail complex phenomena of surface relaxation in clean metals. The model accounts for relaxation effects that go many layers deep, that have both parallel and perpendicular components and that show large variations from surface to surface of the same metal. The model adds a new physically plausible assumption to the simple electrostatic model previously proposed by Finnis and Heine, which increases the force binding each ion to its bulk position by an amount fixed empirically for each metal. The equilibrium configuration of surface layers is found by minimizing the energy with respect to rigid translations of ion nets in a fixed electronic background density. The many surface structure parameters thus determined fit low-energy electron diffraction data on six surfaces of bcc Fe and six of fcc Al well in almost all cases.

Magnetism of metastable phases: Band theory and epitaxy (invited)

Total‐energy band calculations are used to analyze the magnetic phases of metallic elements as functions of volume. The calculations utilize a fixed‐spin‐moment procedure, which is described and justified as a natural generalization of density‐functional theory. This procedure finds the ground‐state energies of electronic systems under two constraints, and hence determines the system energy as a function of two variables—volume and magnetic moment. The energy function is used to find the ferromagnetic phases and their ground‐state properties, including bulk moduli and magnetic susceptibilities. The systems studied are fcc Fe, fcc Co, bcc Ni, fcc Pd, and bcc Mn, each of which undergoes a phase transition for small changes of the lattice constant from equilibrium (zero‐pressure) values.

Atomic structure of Fe{111}

Abstract A clean Fe {111} surface was prepared and studied with LEED (low-energy electron diffraction) and AES (Auger electron spectroscopy). A LEED intensity analysis was carried out with a new computational scheme (THIN) specially designed for short interlayer spacings. The results are, for the fust interlayer spacing, d 12 = 0.70 ± 0.03 A and for the inner potential V 0 = 11.1 ± 1.1 eV, the confidence intervals referring to 95% confidence level. Thus, the Fe {111} surface is contracted 15.4% with respect to the bulk (0.827 A).