Search

We report the structures and energies from first principles density functional calculations of 12 different reconstructed (111) surfaces of silicon, including the 3 × 3 to 9 × 9 dimer−adatom−stacking fault (DAS) structures. These calculations used the Perdew−Burke−Ernzerhof generalized gradient approximation of density functional theory and Gaussian basis functions. We considered fully periodic slabs of various thicknesses. We find that the most stable surface is the DAS 7 × 7 structure, with a surface energy of 1.044 eV/1 × 1 cell (1310 dyn/cm). To analyze the origins of the stability of these systems and to predict energetics for more complex, less-ordered systems, we develop a model in which the surface energy is partitioned into contributions from seven different types of atom environments. This analysis is used to predict the surface energy of larger DAS structures (including their asymptotic behavior for very large unit cells) and to study the energetics of the sequential size change (SSC) model proposed by Shimada and Tochihara for the observed dynamical reconstruction of the Si(111) 1 × 1 structure. We obtain an energy barrier at the 2 × 2 cell size and confirm that the 7 × 7 regular stage of the SSC model (corresponding to the DAS 7 × 7 reconstruction) provides the highest energy reduction per unit cell with respect to the unreconstructed Si(111) 1 × 1 surface.

In-situ attenuated total internal reflection infrared spectroscopy has been used to examine initial adsorption and reaction steps in atomic layer deposition of HfO2 from tetrakis(diethylamino) hafnium (TDEAHf) on SiO2 and hydrogen-terminated Si(100) surfaces. At low deposition temperatures (25−250 °C), TDEAHf directly reacts with the Si−H surface, resulting in partial removal of Si−H bonds and formation of a four-membered Si−O−Hf−Si bonding structure that can rapidly oxidize. The hydrogen removal process is observed to continue through many cycles of TDEAHf/H2O exposure, signifying continued reactivity of the Hf precursor with the silicon surface. Density functional theory calculations have been performed for various reactions between tetrakis(dimethylamino) hafnium and Si−H surfaces, and several possible reaction pathways for hydrogen removal have been identified and analyzed. The calculations suggest that hydrogen removal proceeds by H abstraction by an amine ligand of the Hf precursor and that the abstraction reaction is made more facile by the presence of OH on the otherwise H-terminated Si surface.

We use density functional theory to investigate the surface chemistry of initial oxidation and hydroxylation of the Ge(100)-2×1 surface by water and hydrogen peroxide. Comparison of the reaction of water on the Si(100)-2×1 and Ge(100)-2×1 surfaces shows that the kinetics of oxidation of the Ge(100)-2×1 surface with water is slower. Our calculations also show that oxidation products on the Ge(100)-2×1 surface are less thermodynamically stable than on Si. We also investigate two competing dissociation reactions of H2O2 on the Ge(100)-2×1 surface. We find that dissociative adsorption via cleavage of the OH bond is less exothermic than OO dissociation. Furthermore, interdimer OO dissociation has a lower activation barrier than interdimer or intradimer OH dissociation, although interdimer dissociation products are found to be less stable compared than those formed from intradimer dissociation reactions. Finally, we find that the oxidation products formed from hydrogen peroxide are more stable than those formed from water.

The atomistic mechanism of the initial atomic layer deposition (ALD) reaction of trimethylaluminum [Al(CH3)3, TMA] on octadecyltrichlorosilane (OTS)-based self-assembled monolayers (SAMs) with different functional termination groups was investigated using density functional theory (DFT). The adsorption of TMA on −NH2-terminated OTS involves formation of a dative bond between the amine lone pair and the Al atom of TMA and is exothermic by 0.86 eV. Conversion of the dative bond to a covalent bond through a four-centered H transfer from the NH2 to the methyl group of TMA to form methane involves a barrier that is 1.22 eV above the adsorbed complex and is exothermic by 0.94 eV. Reaction between TMA and the −OH-terminated SAM results in an adsorption energy of 0.60 eV. Although the dative bond is weaker, the barrier to converting the dative bond to a covalent bond to form methane is only 0.65 eV relative to the complex structure and the overall reaction is exothermic by 1.44 eV. The reaction of TMA with the methyl-terminated SAM forms no adsorbed complex and the ligand exchange reaction to form methane requires a barrier of 1.82 eV and is endothermic by 0.09 eV. These results indicate that the end group of the SAM significantly affects the selectivity toward TMA adsorption and subsequent ALD of Al2O3. Although initiation of ALD reactions involving TMA appears to require lone pairs, even strong dative bonds, such as those between amine groups and TMA, lead to ligand exchange reactions that are not kinetically favorable.

Atomic nitrogen on the Si(100)−(2 × 1) surface is investigated using B3LYP density functional theory to study the incorporation of nitrogen atom into the silicon surface during growth of nitride films on silicon. Several possible structures for nitrogen on the Si(100)−(2 × 1) surface are investigated, including N bridge-bonded into the Si−Si dimer, N bridge-bonded into the Si−Si back-bond, and N inserted by forming three bonds with three Si atoms. Furthermore, the energetics and reaction mechanisms leading to these structures are also calculated. We find that the structure with nitrogen atom bridge-bonded into the Si−Si dimer to be the most thermodynamically stable, with an adsorption energy of 105 kcal/mol and an insertion barrier of 29 kcal/mol. Insertion into the Si−Si back-bond, however has the lowest activation barrier of 11 kcal/mol, although its adsorption energy is 13 kcal/mol lower than insertion into the Si−Si dimer. The third configuration investigated with N forming three bonds with Si atoms has a relatively high activation barrier of 35 kcal/mol and an adsorption energy of 95 kcal/mol. In addition, the formation of this structure “consumes” three Si atoms. Hence, at high nitrogen pressure, this structure is expected to be less dominant than those with N inserted into the Si−Si bonds, which only consume two Si atoms.

We present ¹⁷O and ¹¹B NMR and quantum chemical calculations based on Hartree−Fock (HF) and density functional theory to investigate the site connectivity, the degree of chemical short-range order (CSRO), and the topological short-range order in binary borosilicate glasses, which can serve as a simple model system for more complex borosilicates and other covalent oxide glasses and melts. We also calculate the enthalpy of hydrolysis for each oxygen cluster in such glasses. The stability of ^[4]Si−O−^[3]B was evaluated from the relative energy differences among the bridging oxygen clusters (^[4]Si−O−^[4]Si, ^[4]Si−O−^[3]B, and ^[3]B−O−^[3]B) calculated at the B3LYP/6-311+G(2d,p)//HF/3-21G and B3LYP/6-311+G(2d,p) levels of theory as well as data from ¹⁷O 3QMAS NMR. The fraction of ^[4]Si−O−^[3]B at a boron mole fraction of 0.57 is about 37% (±1), suggesting a significant inter-dispersion of ^[3]B and ^[4]Si, where a random distribution of ^[4]Si and ^[3]B and complete phase separation lead to 50 and 0% of ^[4]Si−O−^[3]B, respectively. ¹¹B MAS NMR shows that the fraction of boroxol rings increases with increasing boron content. The bond angle distribution function G(α) for each cluster was obtained from the lattice energy variation with angle. The calculation for ^[4]Si−O−^[4]Si shows close similarity to that obtained by recent high energy photon diffraction experiments. We also predicted the effects of composition, fictive temperature and degree of disorder among framework cations on G(α). The calculated enthalpies of hydrolysis of ^[4]Si−O−^[4]Si, ^[4]Si−O−^[3]B, and ^[3]B−O−^[3]B are −16.98, −3.03, and 16.61 kJ/mol, respectively, implying differential reactivity with water and that ^[4]Si−O−^[3]B clusters should have intermediate kinetic stability. Quantum chemical simulations also suggest a possible mechanism for the formation of the oxygen tricluster, related to the presence of five-membered rings.

The adsorption of acetone on the Ge(100)-2 × 1 surface has been investigated using multiple internal reflection infrared spectroscopy and density functional theory quantum chemistry calculations and was found to be under thermodynamic control at room temperature. Acetone undergoes attachment at the carbonyl oxygen with loss of an α hydrogen at room temperature on the Ge(100)-2 × 1 surface to form an enol-like adduct. This so-called ene reaction has not been seen on the Si(100) surface, on which the reaction of acetone is under kinetic control and instead a less thermodynamically favored [2 + 2] addition across the carbonyl bond is observed. At low temperature, acetone adsorbs via dative-bond formation with the surface, which we find to be too weak to remain at room temperature in contrast to the stronger dative bonds formed by amines on the Si(100) and Ge(100) surfaces. There is evidence that the acetone ene product may undergo oxygen migration and insertion into Ge−Ge bonds, especially at elevated surface temperatures. Whereas the reactions of organics on Si(100)-2 × 1 are typically under kinetic control, thermodynamic control on the Ge(100)-2 × 1 surface is made possible by the reversibility of weak surface adsorption. The strategic use of thermodynamically controlled reactions on the Ge(100)-2 × 1 surface represents a new and potentially powerful approach for the selective formation of organic-functionalized semiconductor surfaces.

We have used density functional theory along with the cluster approximation to investigate the adsorption of chlorinated silanes on the Si(100)-2 × 1 surface. We have calculated the energetics for various adsorption reactions. We found that dissociative adsorption of chlorosilanes via scission of either the Si−H or Si−Cl bond is exothermic by 50−53 kcal/mol. The vibrational frequencies of the adsorption products were also calculated. It was found that the calculated frequencies for the dissociative adsorption of SiH2Cl2, SiHCl3, and SiCl4 via Si−Cl scission are in good agreement with experiments. Transition state barriers were calculated for adsorption via scission of the Si−Cl bond and via scission of the Si−H bond. We find that the barrier for SiH2Cl2 and SiHCl3 adsorption via scission of the Si−Cl bond is 6−9 kcal/mol lower than the barrier for adsorption via scission of the Si−H bond. This indicates that the reaction pathway is kinetically controlled. Furthermore, the adsorption barrier of SiH2Cl2 is 7 kcal/mol lower than the barrier for SiH4 adsorption. This difference in barrier heights partially explains the experimental observation that SiH2Cl2 has a higher initial sticking coefficient than SiH4.

The room-temperature adsorption of pyrrolidine and its methyl-protected analogue, N-methylpyrrolidine, on the Si(100)-2 × 1 surface has been investigated using multiple internal reflection Fourier transform infrared spectroscopy and ab initio quantum chemistry calculations. For both compounds, initial adsorption occurs by barrierless formation of a dative bond between the nitrogen lone pair and the electrophilic atom of the Si dimer. However, while pyrrolidine proceeds to chemisorb dissociatively through N−H bond cleavage, methylpyrrolidine is shown to be trapped in its dative-bonded precursor state at room temperature due to a substantial barrier for N−CH3 cleavage. Additionally, the saturation coverage of methylpyrrolidine on Si is seen to be significantly less than that of pyrrolidine, due likely to both steric factors and charge-transfer effects.

The use of quantum chemical methods to determine rate constants for some H-atom metathesis reactions using transition state theory and tunneling corrections is explored. Comparisons are made among several methods (DFT, MP2, QCISD), all of which yield similar structures and frequencies for the transition states, but quite different barrier heights. Tunneling corrections are made using either the well-known Eckart method or one based on the WKB approach. We find that we can fit the extant data by varying the barrier heights using either tunneling approach, although the WKB method is both more accurate and more labor intensive. Values of the barrier heights obtained this way are not in good agreement with those obtained from any of the quantum methods.

We investigate the mechanism of NH3 adsorption and initial decomposition on the (2×1) reconstructed Si(100) surface using B3LYP density functional theory. The Si(100)-(2×1) surface is described using cluster approximation. Ammonia is found to adsorb on the “down” atom of the buckled silicon dimer with no activation barrier. We also find that only half of the surface silicon atoms are active sites for ammonia adsorption. Ammonia adsorption on the Si(100)-(2×1) is exothermic with an adsorption energy of 29 kcal/mol. Dissociation of the adsorbed ammonia to form NH2(a) and H(a) proceeds with a low activation energy of 5 kcal/mol below the NH3(g) and bare Si(100)-(2×1) energy. Our calculated recombination desorption energy of 51 kcal/mol is found to be in good agreement with the temperature-programmed desorption experimental result of 47 kcal/mol. Additionally, our calculated vibrational spectra of NH2(a) and H(a) agree within 2% of the experimental high-resolution electron energy loss spectra.

We present an ab initio investigation of the structure and thermochemistry of the products of the chemisorption of 1,3-butadiene on the Ge/Si(100)-2 × 1 surface. The surface was modeled using a Si cluster with one dimer and Ge substitution for surface Si to study the effects of Ge on the cycloaddition. Calculations were performed using the Becke3LYP hybrid density functional theory method. Both the [4 + 2] Diels−Alder and the [2 + 2] cycloaddition products are energetically stable on all the Ge/Si(100)-2 × 1 surfaces investigated. The [4 + 2] Diels−Alder products on the Si−Si, Si−Ge, and Ge−Ge dimers are more stable than the corresponding [2 + 2] cycloaddition product due to ring strain. The binding energies of all cycloaddition reaction products decrease with increasing Ge composition on dimers, which can be explained by differences in bond strength. The structures and energetics will be explained and discussed.

The adsorption of cyclopentadiene and its dimer form, dicyclopentadiene, on the Si(100)-2×1 surface has been investigated using polarized multiple internal reflection Fourier transform infrared (MIR-FTIR) spectroscopy together with ab initio quantum chemistry calculations. The results show that cyclopentadiene adsorbs at room temperature to form a [4 + 2] cycloaddition product and at saturation covers less than two-thirds of the surface. Dicyclopentadiene adsorbs intact, most likely by a [2 + 2] cycloaddition reaction, and is estimated to adsorb onto 30−40% more Si&dbd;Si dimer sites than cyclopentadiene despite its larger size. Dicyclopentadiene is not found to react via a surface-catalyzed retro-Diels−Alder pathway whereby it splits into cyclopentadiene while chemisorbing, even though this pathway is theoretically predicted to be much more thermodynamically favored.

The reaction of ammonia (NH3) on the Ge(100)-2 × 1 surface is investigated using density functional theory (DFT). We find that NH3 adsorbs molecularly onto Ge(100)-2 × 1 via the formation of a dative bond. The calculations also show that, unlike Si(100)-2 × 1, the activation barrier for subsequent dissociation of NH3 adsorbed on Ge(100)-2 × 1 is higher than that of reversible desorption, which indicates that NH3 has a low reactive sticking probability on the Ge(100)-2 × 1 surface. We also predict that nitrogen insertion into the Ge−Ge dimer requires NH3 overexposure because the activation barrier for NH2 insertion into the Ge−Ge dimer is significantly above the entrance channel. The nitridation reaction pathway results in the N−H bridge-bonded state, which is found to be 17.4 kcal/mol more stable than the reactants. We find that the reactions of NH3 on the Ge(100)-2 × 1 surface generally involve higher activation barriers and less stable intermediates than the analogous reactions on the Si(100)-2 × 1 surface.