The structure of the ͑3 3 1͒ reconstructions of the Si(111) and Ge(111) surfaces induced by adsorption of alkali metals has been determined on the basis of surface x-ray diffraction and low-energy electron diffraction measurements and density functional theory. The ͑3 3 1͒ surface results primarily from the substrate reconstruction and shows a new bonding configuration consisting of consecutive fivefold and sixfold Si (Ge) rings in ͗110͘ projection separated by channels containing the alkali metal atoms. [S0031-9007(98)05973-0] PACS numbers: 61.10.Eq, 61.14. Hg, 68.35.Bs, 71.15.Mb Over the last decade there has been a large effort to understand the structure and properties of reconstructions on elemental semiconductor surfaces. The main driving force behind these reconstructions is the reduction of the number of dangling bonds without introducing too much strain in the surface region. Three structural elements meeting this principle have emerged so far. Particularly important are adatoms which saturate three dangling bonds on (111) surfaces while creating only one unsaturated bond. Adatoms stabilize the clean Ge(111)-c͑2 3 8͒ surface [1] and are also a major stabilizing factor in the dimer-adatom-stacking-fault model of the clean Si͑111͒-͑7 3 7͒ surface [2]. The second structural element that effectively reduces the number of dangling bonds is the dimer frequently found on the (001) surfaces [3]. The third structural element is the p-bonded chain which was first proposed for the clean Si͑111͒-͑2 3 1͒ reconstruction [4]. Yet this simple principle has been of no utility in predicting surface structures as demonstrated by the metal induced ͑3 3 1͒ reconstructions on the (111) surfaces of Si and Ge. Despite the small unit cell, the atomic geometry is still unknown and has been heavily debated over the last ten years [5][6][7][8][9][10][11][12][13]. The mere observation of the symmetry-breaking ͑3 3 1͒ unit cell calls for a unidirectional structural motif and it was, therefore, appealing to introduce p-bonded chains to explain the 3 3 1 periodicity. At present, there are two promising models for the ͑3 3 1͒ reconstruction that have been proposed on the basis of scanning tunneling microscopy (STM) [8], electronic properties [9], and total-energy calculations [10]. The Seiwatz model [see Fig. 1(a)] [7,11,12] consists of parallel p-bonded chains formed by fivefold rings of Si (Ge) atoms in ͗110͘ projection, separated by empty channels, with a top-site adsorbate saturating the surface dangling bonds. The second model is the extended Pandey model [10,13] [see Fig. 1(b)] which consists of a sevenfold ring carrying the p-bonded chain alternating with a five and six-member ring of Si. It is intuitive to describe the ring sequences from these models with the notation 567567 (extended Pandey) and 500500 (Seiwatz model). Unfortunately, neither of these structures is able to explain our surface x-ray diffraction data (SXRD) or low-energy electron diffraction (LEED) data.We determined the structure by using a multipletechnique ...

The ordered overlayer structures formed by Cs adsorbed on a Ru(0001) surface were analyzed by use of low-energy electron diffraction (LEED). The phase diagram reflects the dominance of dipole-dipole repulsions between the adparticles and comprises quasiliquid configurations characterized by diffraction rings up to a coverage 6=0.17, followed by a (2 X 2) structure with maximum intensity of the diffraction spots at 6=0.23. Beyond 6=0.25, a series of structures with rotated unit cells is identified which are followed by a (&3X&3)R30' structure around 6=0.33 ( =completion of the first monolayer). In the 0 (2X2) phase the Cs atoms are located in on-top sites with a Ru-Cs bond length of 3.25+0.08 A, corresponding to a hard-sphere radius of 1.9 A for the Cs atom. In the (&3X &3)R 30' structure, on the other hand, the adatoms occupy threefold hollow hcp sites with Ru-Cs bond lengths of 3.52+0.02 A, corre-0 sponding to a Cs hard-sphere radius of about 2.2 A. The increase in bond length and effective radius of the adparticle is paralleled by the transition of the character of bonding from more "ionic" at 6=0.25 (large dipole moment) to more "metallic" at 6=0.33 (dipole moment reduced by about 30%). The associated change of the type of adsorption site (from on-top to hollow) is qualitatively rationalized by a model according to which inherently less favorable sites may become preferred due to improved effective screening of the dipole-dipole repulsion by the location of substrate atoms in the region between neighboring adatoms.

Oxygen adsorption on to Ru͑10 1 0͒ results in the formation of two ordered overlayers, i.e., a c(2 ϫ4)-2O and a (2ϫ1)p2mg-2O phase, which were analyzed by low-energy electron diffraction ͑LEED͒ and density functional theory ͑DFT͒ calculation. In addition, the vibrational properties of these overlayers were studied by high-resolution electron loss spectroscopy. In both phases, oxygen occupies the threefold coordinated hcp site along the densely packed rows on an otherwise unreconstructed surface. The O atoms are attached to two atoms in the first Ru layer Ru͑1͒ and to one Ru atom in the second layer Ru͑2͒, forming zigzag chains along the troughs. While in the low-coverage c(2ϫ4)-O phase, the bond lengths of O to Ru͑1͒ and Ru͑2͒ are 2.08 and 2.03 Å, respectively, corresponding bond lengths in the high-coverage (2 ϫ1)-p2mg-2O phase are 2.01 and 2.04 Å ͑LEED͒. Although the adsorption energy decreases by 220 meV with O coverage ͑DFT calculations͒, we observe experimentally a shortening of the Ru͑1͒-O bond length with O coverage. The v(Ru-O) stretch mode is found at 67 meV ͓c(2ϫ4)-2O͔ and 64 meV ͓(2 ϫ1)p2mg-2O͔. ͓S0163-1829͑98͒00824-8͔

The local adsorption geometries of the (2 × 2)-N and the ( √ 3 × √ 3)R30 • -N phases on the Ru (0001) surface are determined by analyzing low-energy electron diffraction (LEED) intensity data. For both phases, nitrogen occupies the threefold hcp site. The nitrogen sinks deeply into the top Ru layer resulting in a N-Ru interlayer distance of 1.05Å and 1.10Å in the (2 × 2) and the ( √ 3 × √ 3)R30 • unit cell, respectively. This result is attributed to a strong N binding to the Ru surface (Ru-N bond length = 1.93Å) in both phases as also evidenced by ab-initio calculations which revealed binding energies of 5.82 eV and 5.59 eV, respectively.

No abstract