Unusual Two-Stage Kinetics of Ethylene Adsorption on Si(100) Unraveled by Surface Optical Spectroscopy and Monte Carlo Simulation

The adsorption of ethylene on a $\mathrm{Si}(100)\mathrm{\text{−}}2\times 1$ surface in an ultrahigh vacuum has been monitored at room temperature by use of real-time surface differential reflectance spectroscopy, which clearly demonstrated that the adsorption follows a two-stage process. About half a monolayer is obtained for 1 L, while the second stage is much slower, yielding the complete monolayer for an exposure of $\sim 400\mathrm{L}$. The kinetics over the full range has been successfully reproduced by a Monte Carlo calculation. The key point of this two-stage adsorption kinetic lies in the reduced adsorption probability (by a factor of several hundreds) on the Si dimers, neighbors of dimers which have already reacted, with respect to the adsorption probability on isolated dimers. This new kind of adsorption kinetics, due to a repulsion between already adsorbed molecules and additional molecules impinging on the surface, makes it a textbook case for a “cooperative” adsorption process.

Figure 1

(a): SDR spectra for 1 ML of ethylene (empty red squares obtained with 400 L), 0.45 ML corresponding to the scheme (c) (full blue circles, 1 L) and intermediate coverages. (b) SDR spectrum after saturation with atomic hydrogen, leading to a monohydride $\mathrm{Si}(100)\mathrm{\text{−}}2\times 1$: H surface (from Ref. 37). (c) Scheme of a dimer row at 0.45 ML coverage.

Figure 2

Ethylene coverage as a function of exposure for large (a) and small (b) exposures; (c) is a semilogarithmic representation. Empty and filled black squares, experimental points from two sets of experiments. Thick black continuous line, Langmuir kinetics for ${\theta }_{\mathrm{sat}}=0.45\mathrm{ML}$; black dotted and black dot-dashed lines are Langmuir kinetics for ${\theta }_{\mathrm{sat}}=1\mathrm{ML}$; thin blue continuous line is the Kisliuk model for ${\theta }_{\mathrm{sat}}=0.45\mathrm{ML}$ and $K=0.7$; red short-dashed curve is the Kisliuk model for ${\theta }_{\mathrm{sat}}=1\mathrm{ML}$ and $K=95$, red long-dashed curve is the Kisliuk model for ${\theta }_{\mathrm{sat}}=1\mathrm{ML}$ and $K=2.2$. (d) Scheme of precursors, with probabilities of adsorption, diffusion, hopping, and desorption.

Figure 3

Sticking coefficient as a function of coverage in linear (a) and logarithmic (b) representations. Black squares show experimental results from Ref. 14 at 317 K. Black dotted and thin continuous curves [linear in (a)] are Langmuir results for ${\theta }_{\mathrm{sat}}=1$ and 0.45 ML, respectively. Blue dotted curve is the Kisliuk model for ${\theta }_{\mathrm{sat}}=0.45\mathrm{ML}$, with $K=0.7$. Red long-dashed and short-dashed curves show the Kisliuk model for ${\theta }_{\mathrm{sat}}=1\mathrm{ML}$, with $K=2.2$ and 95, respectively. Purple thick continuous curve is the Monte Carlo result.

Figure 4

Experimental coverage vs ethylene dose and Monte Carlo calculations in linear (a) and logarithmic (b) representations. Purple continuous line is best fitting with $p_d=0.3$, $p_{a,2}=0.0022$, and $p_{a,1}=2.p_{a,2}$. Green dotted and dashed lines: $p_d=0.3$, with $p_{a,2}=0.005$ and $p_{a,2}=0.001$, respectively (with $p_{a,1}=2.p_{a,2}$). Inset of (a) shows initial adsorption, brown dotted and dashed lines are $p_d=0.1$ and $p_d=0.8$, respectively, and the values of $p_{a,1}$ and $p_{a,2}\ll 1$ are of little importance.