Hydrogen chemisorption and the structure of the diamond C(100)-(2 × 1) surface

Abstract

Upon heating to greater than 1300 K, diamond C(100)-(1 × 1) reconstructs, exhibiting a low energy electron diffraction (LEED) pattern with two domains of (2 × 1) symmetry. No evidence for higher order reconstructions (e.g., c(4 × 2)) is observed, ruling out correlated buckled dimers as a possible structure for the reconstruction. Hydrogen is found to play an important role in the transformation of the (1 × 1) surface during reconstruction. The presence of chemisorbed hydrogen on the (1 × 1) surface is monitored by electron-stimulated desorption time-of-flight spectroscopy (ESD-TOF) as a function of anneal temperature. Two distinct H+ ESD velocity distributions, fast and slow, are observed in the TOF spectra on the polished (1 × 1) surface. Upon heating, the fast component disappears with the onset of the (2 × 1) reconstruction. The slower of the two features remains even after annealing at temperatures of up to 1530 K, demonstrating that some hydrogen remains on the reconstructed (2 × 1) surface. Temperature-programmed desorption (TPD) of H2 from the diamond C(100)-(1 × 1) surface also correlates with the surface reconstruction and the loss of the fast proton peak in the ESD spectra. For the “as polished” surface, the TPD results yielded an integrated hydrogen desorption flux of up to ten monolayers (based on two hydrogen atoms per surface carbon atom, e.g. 3.14 × 1015 H cm−2) indicating that the near-surface region for the as-polished diamond C(100)-(1 × 1) contains a large quantity of absorbed hydrogen. At low hydrogen coverage, the desorption process appears to be first order with an activation energy of ∼ 37 kcalmol and a first order pre-exponential of ∼ 3 × 105s−1. At higher coverages the apparent order increases and the activation energy decreases. Ultraviolet photoelectron spectroscopy (UPS) of the diamond C(100)-(1 × 1) does not reveal any occupied states in the band gap from the valence band maximum (VBM) to the Fermi level (EF). In contrast, UPS for diamond C(100)-(2 × 1) exhibits occupied states in the band gap from the VBM to 1.5 eV above the VBM. No empty states could be found from 1.2 to 5.5 eV above EF for this surface using two-photon photoemission. By analogy to the Si and Ge(100) surfaces, the proposed model for the diamond C(100)-(2 × 1) surface is based on the formation of dimer pairs. The sharp two domain (2 × 1) LEED pattern, the presence of hydrogen on the reconstructed surface, and the observation of occupied surface states in the band gap suggest that symmetric dimers with monohydride termination at each carbon atom of the pair is the most likely structure.