We have used dynamical low-energy electron diffraction (LEED) to determine the adsorption site and the geometry of the surface region for the p(2x2) overlayer of potassium adsorbed on Ni(l I I). The structure consists of the potassium atoms adsorbed on top of the Ni atoms with vertical reconstructions of Ni atoms in the first and second substrate layers combined with a slight horizontal reconstruction of the first substrate layer. The potassium-nickel bond length is found to be 2.82+ 0.04 A corresponding to a rather short effective potassium "radius" of about I.S7 A. PACS numbers: 61.14.Hg, 68.55.Eg, S2.65.My The nature of the chemical bond between adsorbed alkali-metal atoms and metal substrates has been a matter of some controversy for the past few years. While some experimental and theoretical results have been interpreted as supporting a picture of ionic bonding at low coverages [1-4],others have been interpreted as supporting a picture of covalent bonding at all coverages [5,6]. Although alkali-metal adsorption systems are among the simplest of chemisorption systems and they have been studied extensively [7], there have been only a few complete structural determinations of them [8-13],as shown in Table I. All of these except one, a LEED study of Cs/Cu (111) [9], indicate that alkali-metal atoms occupy high-coordination sites. No explanation has been proposed for the low-coordination site of Cs/Cu(l 1 1), possibly due to low confidence in the result since it is both unexpected and uncorroborated. Therefore it has been common to assume that the site of adsorption for alkali metals on low-index, atomically flat metal surfaces is the high-coordination site as a consequence of the nondirectional bonding expected of the alkali s orbital [14][15][16][17].The results presented in this paper show that this is not a good assumption in all cases and that our current understanding of the alkali-inetal chemisorption bond is not complete. This experiment was carried out in an ultrahigh vacuum system which was Mumetal shielded and had a base pressure of 6 x 10 " mbar. The data were obtained using a standard Varian four-grid optics in constant-beamcurrent mode and a video data acquisition system [18,19].The crystal was cut to within 0. 25' of the (111) surface and was subsequently mechanically polished and chemically etched. It had been used in adsorption studies for at least two years prior to this experiment and so had been through many cycles of 0.5-keV Ar+ ion bombardment and annealing to 1200 K. After additional cleaning cycles and before running each of these experiments, the crystal was heated to 1200 K and slowly cooled at a rate of about 1 K/s to 120 K in order to minimize the surface defect density [20].The phase diagram and the procedure for producing the p(2x2) structure in the potassium overlayer have been determined in previous LEED experiments [21].The surface was routinely checked for impurities using Auger electron spectroscopy (AES) both before and after adsorption experiments, and after experiments there ...