In using pulsed laser excitation of surfaces to induce desorption or reaction of adsorbed molecules, it has generally been assumed that the absorbed energy is rapidly randomized, and a thermal model can be used to calculate the surface-temperature change. In this work, the transient temperature jump on a Ag(llO) surface induced by an 8-nsec laser pulse is directly monitored with a psec probe pulse. The probe is based on a temperature-dependent second-harmonic-generation effect. The experiment provides the first direct evidence that the heat-diffusion model can correctly predict the magnitude and the time evolution of the temperature on the surface. PACS numbers: 79.20.Ds, 42.65.Ky, 68.35.Md The use of lasers to induce and probe chemical and physical processes on surfaces is a subject of much recent interest in surface science. x There are a few unique advantages in using lasers to excite an adsorbate/substrate system to induce surface reactions. The initial excitation is mode specific and may lead to the possibility of nonstatistical reaction channels. In many cases, however, thermalization of the initial excitation may be fast compared to the reaction rate, and all reactions would then proceed under thermal conditions. Even for this latter case of equilibration of the initially absorbed energy, the rapid (>10 10 deg/sec) temperature rise caused by pulsed laser excitation can change the relative yields of products coming from competing chemical pathways. The latter advantage has been demonstrated in studies of surface reactions. 2 In order to correctly interpret the events following pulsed laser excitation of surfaces, it is essential to know the time scale of thermalization from the initially excited mode, and the resultant surface-temperature increase. An accurate account of the temperature evolution on the surface is essential to properly determine the kinetic parameters of the surface process.Many surface studies have used powerful nanosecond laser pulses to excite metal substrates in order to induce surface processes. 2 " 9 In all cases, the transient temperature jump at the surface is not experimentally determined, but is calculated by a classical heat-diffusion model. 10 ' 11 This model assumes that the energy absorbed into a specific mode thermalizes instantaneously and that the transfer of heat from the surface to the bulk can be treated by use of bulk heat-diffusion constants. Furthermore, several working assumptions have been widely used to reduce the complexity of the calculation, such as that higher-order terms in the heat-transfer equation can be neglected.For excitation of metals using light from the near ir to the uv, intraband or interband electronic transitions are involved. Several femtosecond experiments have been performed recently for a variety of metals, where the rates of electron relaxation through electron-phonon collisions where shown to be of the order of several picoseconds. 12 " 15 Therefore, it is reasonable to assume that, on a nanosecond time scale, thermal equilibrium is est...