Tribology Letters Vol. 10, No. 1-2, 2001 51 Atomic friction studies on well-defined surfaces R. Bennewitz, E. Gnecco * , T. Gyalog and E. Meyer Institute of Physics, University of Basel, Klingelbergstrasse 82, CH-4056 Basel, Switzerland Atomic friction studies have been performed by means of a friction force microscope (FFM) in ultrahigh vacuum, where well-defined surfaces can be prepared. A home-built FFM allows us to study lateral forces as low as 0.05 nN using rectangular silicon cantilevers. Furthermore, comparison with dissipation measurements performed in non-contact mode are possible. Recent experimental results are presented and discussed in the framework of a one-dimensional Tomlinson model which includes thermal activation. Atomic-scale stick– slip processes on a metallic surface could be repeatedly measured on Cu(111), while the Cu(100) surface was distorted by the tip during the scanning process. A logarithmic velocity dependence of atomic friction has been measured on Cu(111) and NaCl(100) for low scanning velocities. The dissipation found in stick–slip measurements is compared to the power loss detected in dynamic non-contact measurement. KEY WORDS: friction force microscopy; atomic friction; velocity dependence; dissipation; copper; NaCl 1. Introduction Atomic friction processes are the microscopic origin of macroscopic friction, a field of clear technological impor- tance. However, it is difficult to draw conclusions from macroscopic experiments to underlying atomic processes due to the statistic contribution of multiple contacts between macroscopic bodies. Friction force microscopy (FFM) al- lows one to study the friction of a single contact sliding over a surface, where the contact can be as small as just a few atoms. Mate et al. demonstrated in 1987 for the first time that FFM is even able to detect atomic friction by measuring atomic-scale stick–slip processes [1]. The typical result of atomic-scale stick–slip processes in FFM experiments is shown in figure 1. The friction force F L acting on a tip which is scanned with the velocity v parallel to the surface is measured by means of the torsion of a can- tilever bearing the tip. With the periodicity of the atomic sur- face structure the tip sticks to a position until the lateral force built up by the cantilever is high enough to initiate a slip to the next atomic position. The overall appearance of such re- sults has been explained based on the ideas of Tomlinson, who suggested modeling the friction process by a combina- tion of a lateral surface potential and a spring-type potential which move relative to each other [2–5]. In the last years, several well-defined surfaces have been studied by FFM in ultrahigh vacuum with respect to atomic- scale stick–slip processes. It was demonstrated that these phenomena can be observed also on non-layered materials, and that the resolution limits were determined by studying atomic friction close to monatomic steps on the surface [6]. Even reactive surfaces like the Si(111)7 × 7 reconstructed surface have been successfully studied after special tip treat- ment [7,8]. In this paper, we report recent results in the field of atomic friction. In a first section we discuss details of the one- * Present address: Department of Physics, University of Genova, Italy. dimensional Tomlinson model which are important for the interpretation of experimental data, in particular the depen- dence on the scanning velocity. In the experimental part we present atomic friction studies on copper surfaces, the veloc- ity dependence of stick–slip processes, and compare energy dissipation of the stick–slip process with damping measure- ments in the dynamic non-contact mode [9]. Finally, we con- sider future experiments in the field of atomic friction. 2. Modeling In this section, extensions of the one-dimensional Tomlinson model are discussed in order to compare exper- imental results to theoretical considerations on the atomic friction process. We start by defining the simplest potential V(x,t) for the tip being at the position x which is constituted Figure 1. Typical result of an atomic-scale stick–slip experiment using a FFM. The lateral force is recorded while scanning the tip forward (solid) and backward (dashed) in (100) direction over a NaCl(001) surface. The mean lateral force is calculated by dividing the area of hysteresis (filled with grey color) by two times the scanned distance. The thick bar denotes the slope ∂F L /∂(vt) which is taken as an estimate for the effective spring constant k eff [11]. 1023-8883/01/0300-0051$19.50/0  2001 Plenum Publishing Corporation