The energy and angular distributions of Ni and Al atoms sputtered from Ni 3 Al(100) by 8 keV Ar have been measured using multiphoton resonance ionization detection. The Al atoms, which originate entirely from the top atomic layer of Ni 3 Al(100), are predominantly sputtered along close packed <110> directions. On the other hand the sputtered Ni flux, which arises from both the first and second layers, has a large component normal to the surface. Molecular dynamics computer simulations agree well with experimental results and suggest that the normal emission of Ni atoms arises primarily from second-layer atoms. # 1998 John Wiley & Sons, Ltd. Received 24 April 1998; Revised 22 July 1998; Accepted 24 July 1998 Energy transfer processes that lead to the desorption of atoms and molecules from ion bombarded surfaces involve a complex sequence of nuclear collisions. There have been many spirited debates through the years related to questions about the detailed nature of these collisions. Various strategies ranging from energy transport theories 1 to Monte Carlo simulations 2 have been exploited to gain a certain level of understanding. Of special interest is the case of single crystal metal targets where atoms are observed to eject along specific crystallographic directions. For these systems, molecular dynamics (MD) computer simulations have been successful at reproducing experimental measurements of the energy and angle distributions of neutral atoms.3 This agreement is encouraging since comparisons at this level of detail confirm a high degree of reliability for the MD model.One aspect of the desorption process has involved the depth of origin of the ejected atom.1-4 Computer simulations have shown that for copper single crystals b90% of the particles are emitted from the top layer.5 More recent models suggest that for fcc single crystals, some 2 nd layer ejection can occur through a ring of top-layer atoms and is observed primarily at an angle normal to the surface plane.6-8 Experimental tests of this prediction have been difficult since there is no way to distinguish between layers in a monoelemental target. One particularly elegant experiment involved the measurement of the yield of Ru from Ru{0001} as it is systematically covered with a monolayer of Cu. 9 The results show that the particles do arise almost exclusively from the top layer, although no information regarding the take-off angle of the few atoms emitted from the second layer were reported. Moreover, the overlayer of Cu had an undefined relationship to the Ru substrate.Single crystal metallic alloys represent an interesting model system for these types of studies. It is technically feasible to prepare compounds that have well-defined numbers of atoms of a specific type in a specific layer. Moreover, the atomic configurations can be precisely determined using a variety of spectroscopic approaches. [10][11][12] These attributes are critical to designing experiments that can be compared with theory since important input parameters for molecular dynamics...