The Use of Photoemission to Determine the Electronic Structure of Solids

-The use of photoemission to determine the electronic structure of solids is discussed. The three-step model of photoemission is outlined. Results from crystalline and amorphous Ge are used to illustrate a successful use of photoemission to determine electronic structure. Data from Si is used to illustrate the use of photoemission to investigate surface states and surface reactions. Finally, results from Ni and other materials are discussed to illustrate the minority of cases where difficulty remains in interpreting photoemission in terms of existing models. 1 . Introduction. Pioneering work in the use of UV photoemission as a spectroscopic tool was performed by Apker, Taft, Philipp and Dickey [I] in the 1950s. However, it was only in the 1960s that this technique began to be fi~lly developed [2], [33. Essential to this was the simple three-step model of photoemission developed by Spicer [4], and Spicer and Berglund [3]. This model allows the optical excitation, electron transport through the solid, and escape across the surface to be treated separately. Using this approximation, a very complex phenomena is made tractable and it has proven possible for experimentalists to gain electronic structure and other information directly from their experiments. As will be emphasized later, the approximate nature of this model should always be kept in mind. An important improvement in experinnental techniques in the sixties res~llted fro117 the widespread availability of vacuum monochrornators, allowing measurements to be made at photon energies much higher than the photoemission threshold. This is important in order to examine a larger range of band structure and in order to minimize difficulties due to the rapidly changing escape probability which occurs at the threshold ot response. Another key development was the arrival of vastly improved vacuum technology which made vacuum below lo-' torr, and as low as 10-l' or lo-'' torr relatively easily available. The development of metal vacuum stations have also increased the ease with which complex operations, such as cleaving, can be performed. In signal handling an important deve(*) Work supported in part by the US A~.niy Research Office-Durham and by the National Science Foundation. lopment was the application of AC techniques and phase sensitive detection to the photoemission experiment [5], 161. The use of higher derivatives [7] provided a useful extension of this technique. A review has recently been written of experimental techniques for UV photoemission 171. Normally in pllotoen~ission spectroscopy, PS, the objective is to determine the distribution in energy of the electrons excited by monochromatic radiation, t](E, /I\'), and from this to deduce information about the electronic structure of the solid. Since it provides information on the optical transition probability to final states at specific energies, PS differs in a critical way from optical or X-ray techniques which, in essence, measure the total transition probability to all possible final states due to absorption of a photon of energy liv P(hv) = q(E, hv) dE. i Here I I (E , l i ~ ) is more exactly defined as the probability of a transiticn to a final state between E and E + dE due to absorption of a photon of energy 11v. q(E, 1118) is closely related to the external photoemission energy distribution curves (EDCs) determined experimentally. tl(E, 1111) is illustrated by figure 1 . The critical point is that it is the transition probability to a final state at a given energy E ; whereas, P(lrv) is the integral over all of the possible transitions. Clearly, much detailed information is lost in going from q(E, /iv) to P(l11,) [8], [9]. To illustrate this more clearly, let us examine figure 2 which compares a plot of the optical constant Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1973607