Substantially low desorption barriers in recombinative desorption of deuterium from a Si(1 0 0) surface

Abstract

We have studied desorption kinetics of deuterium molecules from a Si(1 0 0) surface by means of temperature-programmed desorption (TPD) spectra and isothermal desorptions.Three desorption components, denoted as β 1 ,A , β 1 ,B , and C, can be distinguished in semi-logarithmic plots of the TPD spectra.Their peak positions and intensities are strongly affected by the surface preparation methods employed, either with or without annealing to control the initial D coverage θ D 0 .Peak C appears at the leading edge of the TPD peak.It accounts for only about 5% of the TPD peak at θ D 0 = 1 ML and it diminishes rapidly with decreasing θ D 0 , vanishing at θ D 0 = 0.5 ML .In contrast, together the β 1 ,A and β 1 ,B peaks account for the whole TPD peak at any θ D 0 less than 1.0 ML. The maximum of the β 1 ,A peak is nearly constant at around the maximum temperature of the TPD peak.On the other hand, the β 1 ,B peak appears on the high-temperature side of the TPD peak and it systematically shifts to higher temperatures with decreasing θ D 0 .These results imply that first- and second-order kinetics are operating for the β 1 ,A and β 1 ,B desorptions, respectively.Isothermal desorption experiments confirm the above predicted kinetics for a limited region, namely θ D 0 < 0.5 ML .From the results for the rate curve analysis, the desorption barriers are evaluated to be 1.6 ± 0.1 eV and 1.8 ± 0.1 eV for the β 1 ,A and β 1 ,B desorptions, respectively.These values are substantially lower than the widely accepted value of ∼2.5 eV. To reproduce the measured TPD spectra we take the Arrhenius-type rate equation containing the first- and second-order rate terms for the β 1 ,A and β 1 ,B desorptions.The TPD spectra measured for θ D 0 < 0.4 ML can be reasonably fit with the proposed rate equation when the values given above for E d ,A and E d ,B are used. For θ D 0 > 0.4 ML , however, the TPD curves are not fit with the same values; rather, the best-fit curves require values for E d ,A and E d ,B larger than those given above. Combining the present kinetics results with those obtained by STM along with the studies, the β 1 ,A and β 1 ,B peaks may be attributed to desorption along the 2H path, while peak C may be attributed to desorption along the 4H path. The atomistic desorption mechanism as well as the energy relationship between the desorption barrier and isosteric heat of adsorption are discussed.