Ultrafast Diffusion at the Onset of Growth: $\mathrm{O}/\mathrm{Ru}(0001)$

Nanoscopic clustering in a 2D disordered phase is observed for oxygen on Ru(0001) at low coverages and high temperatures. We study the coexistence of quasistatic clusters (with a characteristic length of $\sim 9\AA$) and highly mobile atomic oxygen which diffuses between the energy-inequivalent, threefold hollow sites of the substrate. We determine a surprisingly low activation energy for diffusion of $385\pm 20\mathrm{meV}$. The minimum of the $\mathrm{O}-\mathrm{O}$ interadsorbate potential appears to be at lower separations than previously reported.

Figure 1

Conventional diffraction scan of 0.09 ML $\mathrm{O}/\mathrm{Ru}(0001)$ at 850 K. The narrow first-order peak is centered at $2.68{\AA }^{-1}$. The peak at $\mathrm{\Delta }K\approx 1.34{\AA }^{-1}$, which is entirely due to the adsorbed oxygen, is significantly broader and indicates that any adsorbate order is short ranged. Inset: primitive unit cell of the Ru(0001) substrate; fcc and hcp adsorption sites are indicated by different-colored circles.

Figure 2

(a) $\alpha (\mathrm{\Delta }\mathbf{K})$ from fitting the $⟨110⟩$ experimental ISFs to $F_1$ and $F_2$. Near the edge of the first Brillouin zone (right vertical dashed line), the slow decay rate, ${\alpha }_1$, is expected to tend towards zero. The anomalous peak in ${\alpha }_1$ at that position can be attributed to a misfit because the fast and slow components cannot be distinguished. (b) Arrhenius plots of ${\alpha }_1$ for two values of $\mathrm{\Delta }K$ over a range of temperatures. The black point is excluded from the calculation of the second gradient as it was recorded at the end of a measurement cycle and may be affected by contaminants. (c) A typical ISF, corresponding to $\mathrm{\Delta }K=1.5{\AA }^{-1}$, alongside fits to $F_1$ and $F_2$.

Figure 3

(a) Relative probability generated by the marginalized Bayesian method, in $(\lambda ,\tau )$ space, for the whole dataset. The sharply peaked distribution indicates that the method is successful. (b) Fitted amplitude $A$, showing a clear peak. (c) Fitted constant $C$; the peak width is an indication of the size of the islands. (d) $R^2$ statistic from fitting $F_H$ to the data. $F_H$ and the global parameters provide a good fit to the experimental ISFs in the vicinity of the half-order position but are unsuitable elsewhere. All points associated with $R^2<0.98$ are plotted as red circles, with error bars omitted for clarity.