On the Mechanism of Silicon Activation by Halogen Atoms

ARTICLE pubs.acs.org/Langmuir

On the Mechanism of Silicon Activation by Halogen Atoms Federico A. Soria,† Eduardo M. Patrito,† and Patricia Paredes-Olivera*,‡ †

Departamento de Fisicoquímica and ‡Departamento de Matematica y Física, Facultad de Ciencias Químicas and Instituto de Investigaciones en Fisicoquímica de Cordoba (INFIQC), Universidad Nacional de Cordoba, Cordoba, Argentina

bS Supporting Information ABSTRACT: Despite the widespread use of chlorinated silicon as the starting point for further functionalization reactions, the high reactivity of this surface toward a simple polar molecule such as ammonia still remains unclear. We therefore undertook a comprehensive investigation of the factors that govern the reactivity of halogenated silicon surfaces. The reaction of NH 3 was investigated comparatively on the Cl-Si(100)-2 1, Br-Si(100)-2 1, H-Si(100)-2 1, and Si(100)-2 1 surfaces using density functional theory. The halogenated surfaces show considerable activation with respect to the hydrogenated surface. The reaction on the halogenated surfaces proceeds via the formation of a stable datively bonded complex in which a silicon atom is pentacoordinated. The activation of the halogenated Si(100)-2 1 surfaces toward ammonia arises from the large redistribution of charge in the transition state that precedes the breakage of the Si-X bond and the formation of the Si-NH2 bond. This transition state has an ionic nature of the form Si-NH3þX-. Steric effects also play an important role in surface reactivity, making brominated surfaces less reactive than chlorinated surfaces. The overall activation-energy barriers on the ClSi(100)-2 1 and Br-Si(100)-2 1 surfaces are 12.3 and 19.9 kcal/mol, respectively, whereas on the hydrogenated Si(100)-2 1 surface the energy barrier is 38.3 kcal/mol. The reaction of ammonia on the chlorinated surface is even more activated than on the bare Si(100)-2 1 surface, for which the activation barrier is 21.3 kcal/mol. Coadsorption effects in partially aminated surfaces and in the presence of reaction products increase activation-energy barriers and have a blocking effect for further reactions of NH3.

’ INTRODUCTION Organic-semiconductor interfaces are playing increasingly important roles in fields ranging from electronics1 to nanotechnology2,3 to biosensing.4,5 Numerous protocols are now available for the preparation of Si-C, Si-O, and Si-N bound layers as described in several reviews.6-10 Most efforts to make high-quality chemically modified surfaces begin with an atomically flat hydrogen-terminated surface that can be produced via etching in ammonium fluoride11 When H-terminated Si surfaces are exposed to atomic Cl, the Cl atoms abstract H atoms from the surface, producing HCl and leaving behind Si “dangling bonds” that can then react with additional Cl radicals to produce Cl-terminated surfaces.12,13 Halogen-terminated surfaces are more reactive alternatives to the hydrogenated silicon surface for functionalization reactions. Alkyl Grignards react readily with chlorinated Si(111) to create alkyl-terminated silicon surfaces.14 The efficient assembly of organic molecules on silicon using the Si-N linkage was achieved by reacting amines with chlorinated silicon surfaces.15,16 The halogenation of semiconductor surfaces allows the tuning of their reactivity. For example, the reaction of 1-octadecanethiol is more favorable on the Cl-terminated Ge surface than on the r 2011 American Chemical Society

Br-terminated surface.17 In the case of silicon, the chlorination of the Si(100) surface was found to lower the processing temperature for depositing stable surface NH2 groups using gas-phase NH3.18 The interaction of ammonia with silicon surfaces is of fundamental importance in numerous applications. The silicon surfaces precovered with the NHx functionality are emerging as candidates for the production of sharp interfaces and for the molecular templating of patterns on single-crystalline surfaces.19,20 The amine-terminated Si(100) surface is also the starting point for biological functionalization with amino acids by the formation of amide bonds.4,5 From early studies, it was recognized that ammonia adsorbs molecularly on the Si(100)-2 1 surface in a strongly sitespecific way (with a sticking probability of around 0.9 at room temperature21) and then dissociates into NH2 and H with a saturation coverage of one NH3 per silicon dimer.22,23 Alongside the experimental studies, a large number of theoretical investigations have found that the NH3 molecule is able to adsorb intact in Received: November 25, 2010 Revised: January 24, 2011 Published: February 21, 2011 2613

dx.doi.org/10.1021/la104701g | Langmuir 2011, 27, 2613–2624

Langmuir a metastable molecular state and that there is a barrier to dissociation. The NH3 molecule forms a dative bond with the empty dangling bond of a down atom of a silicon dimer.24 The Si(100) and Si(111) surfaces act as electron acceptors when bonding with amines.25 The dative Si-N bond formed by NH3 is strong: periodic DFT slab calculations give values of between 29.3 and 31.4 kcal/mol.26-28 In turn, competing coadsorption interactions play an important role in the reaction of NH3 on the partially aminated Si(100) surface.29 Surprisingly, Cl-terminated and H-terminated Si(100) surfaces show very different reactivities toward NH3. Ammonia reacts with the Cl-terminated surface to produce surface-bound Si-NH2 species, and H-terminated Si(100) is essentially unreactive toward NH3 at room temperature.18 The same behavior is observed for the reaction of ammonia with the H-terminated Si(111) surface.20 The surface is unreactive up to 300 °C, and starting at 350 °C, Si-NH2 species appear.20 Unlike the reaction of ammonia with the clean Si(100) surface discussed above, the mechanism by which chlorine atoms activate the Si(100) surface for amine formation is not yet fully understood. It was suggested30 that the reactivity of the chlorinated surface could arise from the ability of silicon to redistribute electron density25 that favors the formation of datively bonded complexes with amines.31 Finstad et al.18 suggested that NH3 could form datively bonded adducts because the chlorine-terminated surface polarizes the silicon atoms, inducing a slight positive charge. However, in a recent ab initio investigation of the adsorption of ammonia on Cl-Si(100)-2 1, no datively bonded complexes were found and the high reactivity of the chlorinated surface toward ammonia could not be explained because a high energy barrier (around 36.9 kcal/mol) was obtained for the SiCl þ NH3 f SiNH2 þ HCl reaction32 (SiCl stands for a surface SiCl group). The authors suggested that other factors such as an enhanced microscopic roughness induced by silicon etching upon Cl exposure might be responsible for the surprising reactivity of Cl-terminated Si(100) toward ammonia. We therefore decided to investigate the mechanism by which halogen atoms activate the Si(100)-2 1 surface in the reaction with a polar molecule such as ammonia. As a reference, the reactivity of the halogenated surfaces was compared to the reactivity of hydrogenated and bare Si(100)-2 1 surfaces. A general mechanism was found in which the balance between charge redistribution and steric effects explains the different reactivities of the halogenated surfaces. The interaction of the ammonia lone pair electrons with the positively charged silicon atoms of the surface SiX groups gives rise to a stable datively bonded complex. The transition state that precedes the breakage of Si-X bonds and the formation of the Si-NH2 bond is ionic in nature and is of the form Si-NH3þX-. The influence of steric effects on the surface reactivity is evidenced when surface chlorine atoms are replaced by bromine atoms. The voluminous bromine atoms prevent the nucleophilic attack of ammonia on the silicon atom of a SiBr group, which leads to an increase in the activation-energy barriers. In the last two sections, we discuss the electronic and steric factors that make the chlorinated surface more reactive than the hydrogenated surface. The different contributions to the stability of transition states are also discussed.

’ THEORETICAL METHODS AND SURFACE MODELING First principles atomistic calculations were performed using state of the art plane wave periodic DFT as implemented in the

ARTICLE

Quantum Espresso code.33 Norm-conserving ultrasoft pseudopotentials34 were used for the atomic species. Gradient corrections were included in the exchange correlation functional using the PBE formulation.35 The electron wave functions were expanded in a plane-wave basis set up to a kinetic energy cutoff of 30 Ry (180 Ry for the density). Brillouin zone integration was performed using a (4 4 1) Monkorst-Pack mesh.36 A vacuum thickness of 10 Å was introduced between the slabs to avoid spurious interactions between neighboring replicas. We performed both spin-polarized and unpolarized calculations. However, no spin polarization was observed. A silicon slab with six layers was used to model the 2 1 reconstruction of the (100) face of silicon. The dangling bonds on the bottom surface were saturated with hydrogen atoms. We calculated the equilibrium bulk structure of silicon to ensure that there were no residual forces in the unit cell. We used a0 = 5.48 Å for the lattice constant as determined in previous work.37 The positions of all of the adsorbate atoms as well as those of the four topmost Si layers were fully optimized. The silicon atoms in the lower bilayer were kept fixed in a bulk configuration. All calculations were performed using a (2 2) unit cell. Reaction pathways and energy barriers were calculated using the climbing image nudged elastic band (CI-NEB) method38 as implemented in the Quantum Espresso code.33 The method works by optimizing a number of intermediate images along the reaction path. Each image finds the lowest energy possible while maintaining equal spacing to neighboring images. This constrained optimization is done by adding spring forces along the band between images and by projecting the component of the force due to the potential perpendicular to the band.38 Several steps were followed to identify and verify transition states and intermediates. We first performed a CI-NEB calculation with 13 images between the reactants and the assumed products. This procedure typically yielded several intermediates and transition states. We next launched geometry optimizations using the images on either side of each transition state in order to verify that they effectively converged into the intermediates found in the previous CI-NEB calculation. We finally launched a CI-NEB calculation with seven images for every elementary step. Transition states were further verified by performing a phonon calculation at the gamma point. The dynamic matrix was computed and diagonalized and the normal modes were visualized using the XCrySDen program.39 We thus verified that each transition state corresponded to a saddle point with one imaginary frequency.

’ RESULTS AND DISCUSSION We will first present the energy changes ΔE for the reactions of ammonia with the hydrogenated and chlorinated surfaces to show that ΔE values do not explain the different reactivities on both surfaces. In the following sections, extensive reaction path calculations were performed to identify elementary reaction steps, intermediates, and transition states for hydrogenated and halogenated surfaces. The activation-energy barriers obtained on the chlorinated surface were much lower than on the hydrogenated surface, thus explaining the experimentally observed high reactivity toward ammonia.18 The reactivity of partially chlorinated surfaces (containing SiCl and SiH groups) and partially aminated surfaces (containing SiNH2 and SiCl groups) was also investigated. This allowed us to 2614

dx.doi.org/10.1021/la104701g |Langmuir 2011, 27, 2613–2624

Langmuir

ARTICLE

Table 1. Energy Change ΔE (kcal/mol) for the Reaction of NH3 on the (a) Cl-Si(100)-2 1 and (b) H-Si(100)-2 1 Surfacesa (a) Cl-Si(100)-2 1

ΔE

4SiCl þ NH3(g) f (SiCl)3Si-NH2 þ HCl(g)

8.9

4SiCl þ NH3(g) f (SiCl)3Si-NH2 þ HCl(ads)

5.1

(SiCl)3Si-NH2 þ NH3(g) f (SiCl)2(Si-NH2)2 þ HCl(g)

11.5

(SiCl)3Si-NH2 þ HCl(ads) þ NH3(g) f (SiCl)2(Si-NH3þCl-)2

-1.4

(b) H-Si(100)-2 1

ΔE

4SiH þ NH3(g) f (SiH)3Si-NH2 þ H2(g)

-0.3

4SiH þ NH3(g) f (SiH)3Si-NH2 þ H2(ads)

-0.2

(SiH)3Si-NH2 þ NH3(g) f (SiH)2(Si-NH2)2 þ H2(g) (SiH)3Si-NH2 þ H2(ads) þ NH3(g) f (SiH)2(Si-NH2)2 þ 2H2(ads)

0.04 -2.7

The calculations were performed using a (2 2) unit cell that contains four SiCl or SiH groups. A chlorinated surface with 25% coverage of amine surface groups is denoted as (SiCl)3Si-NH2. a

Figure 1. Reaction products. (a) The reaction of ammonia with the chlorinated surface yields a HCl molecule hydrogen bonded to a surface amine group. (b) The reaction of two ammonia molecules with the chlorinated surfaces yields ammonium chloride moieties. The chloride anion is located within the gulley region.

evaluate the influence of the local environment and coadsorption effects on the surface reactivity. As the surface coverage of amine groups increases, the adsorbed reaction products become ammonium chloride moieties rather than HCl molecules hydrogen bonded to SiNH2

groups. We therefore investigated the reaction of ammonia on a surface containing an ammonium chloride moiety. The high activation barrier obtained proves the blocking effect of reaction products. Energetics of Reactions of NH3 with H-Si(100)-2 1 and Cl-Si(100)-2 1 Surfaces. Table 1 shows the energetics of chlorinated and hydrogenated silicon surfaces after reactions with one and two NH3 molecules per unit cell. The energy changes were calculated for the HCl and H2 products in the gas phase as well as adsorbed on the surface. The reaction of ammonia on the chlorinated surface is endothermic by 8.9 kcal/ mol (Table 1a) whereas it is slightly exothermic on the hydrogenated surface (-0.3 kcal/mol, Table 1b). The energy difference between the hydrogenated and chlorinated surfaces reveals the fact that the Si-Cl bonds are stronger (96.9 kcal/mol) than the Si-H bonds (78.4 kcal/mol), as we showed in previous work.37 The reaction on the chlorinated surface is less endothermic when the HCl product remains adsorbed on the surface (ΔE = 5.1 kcal/mol). This is due to the hydrogen bond interaction of HCl with the surface amine group, as shown in Figure 1a. On the hydrogenated surface, the H2 product interacts weakly with the surface and the energy change of the reaction is nearly the same as when H2 desorbs into the gas phase (ΔE = -0.2 kcal/mol). Table 1a shows that the reaction of ammonia on a chlorinated surface that is partially aminated (25%) is more endothermic (11.5 kcal/mol) than that on the fully chlorinated surface (8.9 kcal/mol) when the HCl product desorbs into the gas phase. However, when the product remains on the surface, an ammonium chloride moiety is formed as shown in Figure 1b. This structure has also been reported in previous work.32 In this case, the reaction is slightly exothermic (-1.4 kcal/mol). Therefore, the formation of the ammonium chloride moiety produces a stabilization of 12.8 kcal/mol as compared to the desorption of HCl in the gas phase. Table 1b shows that on the hydrogenated surface the reaction of a second NH3 molecule on the partially aminated surface has only a minor effect on the energetics of the reaction when the H2 product desorbs into the gas phase (ΔE = 0.04 kcal/mol) and also when it remains adsorbed (ΔE = -2.7 kcal/mol). In conclution, the thermodynamics shows that the reaction of ammonia with the chlorinated surfaces is endothermic for the desorption of the HCl product into the gas phase. Only the formation of a surface chloride moiety makes the reaction slightly exothermic (-1.4 kcal/mol) for a chlorinated surface with a 25% 2615


Langmuir

Figure 2. Reaction of NH3 through the gulley region of Cl-Si(100). (a) Energy profile along the reaction coordinate. The arrows in the upper part of the plot indicate the four elementary reaction steps found. (b) Structure of reactants (panel I), intermediates (panels III, V, and VII), transition states (panels II, IV, VI, and VIII), and products (panel IX).

ARTICLE

coverage of amine groups. In the case of the hydrogenated surface, a slightly exothermic energy change is also obtained (-2.7 kcal/mol) when the surface is partially aminated. These results do not explain the fact that ammonia reacts with the chlorinated surface at a low temperature of 75 °C18 whereas on the hydrogenated surface the reaction occurs at 350 °C.20 Therefore, reaction pathways and activation-energy barriers are required to explain the different reactivities of hydrogenated and chlorinated surfaces. In the next sections, we report an extensive investigation of the topology of the potential energy hypersurface in order to find the corresponding reaction pathways. NH3 Reaction through the Interdimers Row of Cl-Si(100)2 1 and H-Si(100)-2 1 Surfaces. We consider in this section the reaction pathway of NH3 through the interdimers row (the so-called “gulley” region). These results will then be compared with the reaction pathway of NH3 through the dimers row of the next section. Figure 2a shows the energy profile along the reaction pathway. It starts with an absorbed NH3 molecule (inset) and finishes with an HCl molecule hydrogen bonded to a surface amine group (Figure 2b, panel IX). The energy profile shows that the overall reaction has four elementary steps. Figure 2b shows the corresponding intermediates and transition states. The first elementary step corresponds to the diffusion of adsorbed ammonia within the gulley region (Figure 2b, panel III). This process is endothermic with ΔE = 2.7 kcal/mol and has an activation-energy barrier of Ea = 10.7 kcal/ mol. In the transition state (Figure 2b, panel II), the ammonia molecule lies in the plane of the chlorine atoms. In the next step, the ammonia molecule within the gulley region (Figure 2b, panel III) reacts with a silicon atom to produce an intermediate in which the silicon atom is pentacoordinated to three silicon atoms, to a chlorine atom, and to the ammonia molecule (Figure 2b, panel V). This step is endothermic (6.4 kcal/mol) and has an activation-energy of Ea = 11.2 kcal/mol. The pentacoordinated silicon structure shown in panel V of Figure 2b is a datively bonded complex formed by the interaction of the positively charged silicon atom (þ0.5 net charge obtained from a L€owdin population analysis) and the lone pair of ammonia. Its electronic structure will be considered in detail in the Results and Discussion section. The next elementary step corresponds to the rotation of ammonia around the Si-N bond with Ea = 1.30 kcal/mol and ΔE = -0.9 kcal/mol. In the new configuration (Figure 2b, panel VII), a hydrogen atom of ammonia points toward the chlorine atom. In the last step, this hydrogen atom is abstracted by the chlorine atom, producing a HCl molecule hydrogen bonded to the surface amine group (Figure 2b, panel IX). This step has a small activation-energy barrier of 3.2 kcal/mol and is exothermic with ΔE = -2.4 kcal/mol. In the datively bonded complex (Figure 2b, panel V), the SiN bond length is 1.93 Å and the Si-Cl bond length has increased from 2.11 Å (chlorinated surface) to 2.39 Å. When the chlorine atom is eliminated (Figure 2b, panel IX), the Si-N bond length decreases to 1.77 Å for the surface Si-NH2 group. In summary, the steps with the highest activation-energy barriers correspond to the diffusion of ammonia into the gulley region and to the formation of the datively bonded complex. The overall activation barrier is 13.9 kcal/mol (transition-state structure in panel IV), taking as a reference the energy of adsorbed ammonia (inset of Figure 2a). Figure 3 shows the energy profile and the reaction pathway of ammonia with the H-Si(100)-2 1 surface. The reaction has 2616


Langmuir

ARTICLE

Figure 3. Reaction of NH3 through the gulley region of H-Si(100). (a) Energy profile along the reaction coordinate and (b) structure of reactants (panel I), intermediate (panel III), transition states (panels II and IV), and products (panel V).

only two steps. The first step involves the formation of a pentacoordinate silicon atom (Figure 3b, panel III) with Ea = 9.5 kcal/mol and ΔE = 9.2 kcal/mol. The intermediate in panel III of Figure 3b is a datively bonded complex. It has a Si-N bond length of 2.23 Å, which is longer than the Si-N bond length of 1.93 Å for the datively bonded complex on the chlorinated surface (Figure 2b, panel V). This reveals the fact that the silicon atom in the SiH group is less positive than that in the SiCl group; therefore, the bonding with ammonia is weaker. The next step involves the formation of the Si-N bond and the release of a hydrogen molecule with ΔE = -8.5 kcal/mol and a high-activation-energy barrier of Ea = 29.2 kcal/mol. The overall energy barrier for this reaction is 38.3 kcal/mol, which is much higher than the overall activation barrier of 13.9 kcal/mol on the fully chlorinated surface. These results therefore explain the high reactivity of the chlorinated surface. In the Results and Discussion section, we analyze the factors that lead to a lowering of the activation-energy barrier on the chlorinated surface. The reaction of ammonia through the interdimer region of the Cl-Si(100)-2 1 surface was investigated recently by DFT calculations.32 A barrier of 36.9 kcal/mol was reported in this work, which is much higher than the barrier of 13.9 kcal/mol

Figure 4. Reaction of NH3 through the dimers row of Cl-Si(100). (a) Energy profile along the reaction coordinate. The arrows in the upper part of the plot indicate the three elementary reaction steps found. (b) Structure of reactants (panel I), intermediates (panels III and V), transition states (panels II, IV, and VI), and products (panel VII).

found in our work. The difference arises because the minimumenergy path between reactants and products was not investigated in that work.32 Instead, the authors performed constrained geometry optimizations. They first gradually decreased the vertical position of the nitrogen atom of ammonia (while relaxing the other degrees of freedom) until it penetrated the gulley region. Then, they gradually left one chlorine atom on the surface while relaxing the positions of the other atoms. This procedure, however, does not guarantee that the minimum-energy path can be found. Reaction Pathways of NH3 through the Dimers Row of the Cl-Si(100)-2 1 and H-Si(100)-2 1 Surfaces. Previous 2617


Langmuir

ARTICLE

Figure 5. Reaction of NH3 through the dimers row of H-Si(100). (a) Energy profile along the reaction coordinate and (b) structure of the reactants (panel I), transition state (panel II), and products (panel III).

studies of the reactivity of the Cl-Si(100)-2 1 surface toward H2O40 and NH332 have considered only the reaction through the gulley region. However, the reaction through the dimers row considered in this section has not been investigated. The energy profile in Figure 4a shows that there are three elementary steps when ammonia reacts through the dimers row: (a) the formation of a datively bonded complex (Ea = 6.4 kcal/mol, ΔE = 6.0 kcal/ mol), (b) the formation of a surface ammonium chloride moiety (Ea = 1.8 kcal/mol and ΔE = -4.5 kcal/mol), and (c) the release of a HCl molecule (Ea = 10.8 kcal/mol and ΔE = 4.7 kcal/mol) Panel III in Figure 4b shows the structure of the datively bonded complex that has Si-Cl and Si-N bond lengths of 2.15 and 2.14 Å, respectively. It can be appreciated that the chlorine atom is tilted toward the gulley region. In the second step, the chlorine atom is released into the gulley region and the Si-N bond length decreases to 1.91 Å (panel V of Figure 4b). The energy profile in Figure 4a shows that the ammonium chloride moiety (intermediate V) is nearly as stable as the adsorbed ammonia molecule. In the final step, the chloride ion is released outside the gulley region, and in this process, it abstracts a hydrogen atom from the Si-NH3 moiety, giving rise to a HCl molecule hydrogen bonded to the amine groups. The overall activation barrier for the reaction is 12.4 kcal/mol, which is comparable to the value of 13.9 kcal/mol corresponding to the reaction of NH3 through the gulley region discussed in the previous section. On the hydrogenated surface, the reaction of ammonia through the dimers row involves only one step as shown in Figure 5. The activation-energy barrier is 38.3 kcal/mol. This

Figure 6. Reaction of NH3 with the Si(100)-2 1 surface. (a) Energy profiles along the reaction coordinate for the reaction of the first and second NH3 molecules. (b) Structure of the reactants (panel I), transition state (panel II), and products (panel III) for the reaction of the first NH3 molecule. (c) Idem for the reaction of the second NH3 molecule.

value is the same as when ammonia reacts through the gulley region (Figure 3). Reactivity of NH3 with the Bare Si(100)-2 1 Surface. We consider in this section the successive reaction of one and two NH3 molecules through adjacent dimers of the bare Si(100)-2 1 surface. The reaction of NH3 with this surface has been reported in many DFT studies.29 The purpose of this section is to compare the reactivity of the well-known Si(100)-2 1 surface with that of the chlorinated surface. The energy profile for the reaction of the first NH3 molecule (Figure 6a) with the down atom of the dimer (panel I in Figure 6b) shows that the activation-energy barrier is 21.3 kcal/mol and the reaction is exothermic with ΔE = -15.0 kcal/mol. Figure 6b shows the corresponding structures for the reactants, transition state, and products. The energy profiles in Figure 6a show that the reaction of the second NH3 molecule with the adjacent dimer (panel I in Figure 6c) has a lower energy barrier (16.8 kcal/mol) and is more exothermic (ΔE = -25.3 kcal/mol). This indicates that coadsorption effects favor the reaction of successive NH3 molecules on this surface. 2618


Langmuir

ARTICLE

Figure 7. Reaction of NH3 with a partially chlorinated surface. (a) Diffusion of ammonia into the gulley region. The panels show the structures of the adsorbed molecule (panel I), the transition state (panel II), and the intermediate within the gulley region (panel III). (b) Energy profile for the reaction of NH3 with a SiCl surface group. The panels show the structure of the transition state (panel IV) prior to the formation of the datively bonded complex intermediate (panel V) and the transition state (panel VI) prior to the formation of the HCl product hydrogen bonded to a surface amine group (panel VII). (c) Energy profile for the reaction of NH3 with the SiH group. The diffusion within the gulley region has the same energy profile as for the reaction with adjacent SiCl, and structures I-III are the same as in plot a. Note that the energy scales in plots b and c are different. The panels show the structure of the transition state (panel IV) prior to the formation of the datively bonded complex intermediate (panel V) and the transition state (panel VI) prior to the formation of the H2 molecule (panel VII).

We can now compare the reactivities of the bare, hydrogenated, and chlorinated Si(100)-2 1 surfaces toward ammonia. The chlorinated surface is the most reactive with activationenergy barriers of 12.4 kcal/mol (reaction through the dimers row) and 13.9 kcal/mol (reaction through the gulley region). The reactivity of the Si(100)-2 1 surface is intermediate with a barrier of 21.3 kcal/mol. Finally, the hydrogenated surface has a high-activation-energy barrier of 38.2 kcal/mol. The large difference between the barriers of the hydrogenated and chlorinated surfaces therefore explains the different reactivities observed experimentally.18,20 Reactivity of the Partially Chlorinated Surface. To evaluate the influence of the environment on the magnitude of activation-energy barriers, we investigate in this section the reaction of ammonia through the gulley region with a surface that has a 75% coverage of Cl atoms and a 25% coverage of H atoms. Figure 7 shows the energy profiles when the ammonia molecule reacts with adjacent SiCl and SiH surface groups. The first step involves the diffusion of ammonia into the gulley region (Figure 7a, panels I-III). The NH3 molecule within the gulley region (panel III of Figure 7a) may then react with a SiCl

group (Figure 7b) or with a SiH group (Figure 7c). The energy profile in Figure 7b corresponds to a reaction mechanism with four elementary steps: (a) the diffusion of ammonia into the gulley region, (b) the formation of the datively bonded complex (panel VI), (c) the rotation of ammonia around the Si-N bond, and (d) proton transfer to yield a HCl molecule and a Si-NH2 surface group (panel VII). Panels IV and VI indicate the structure of the transition states preceding the formation of the datively bonded complex and the HCl product, respectively. The thin solid line in Figure 7b corresponds to the energy profile on the fully chlorinated surface shown in the previous section. From the comparison of both profiles, it can be observed that the activation barrier for the diffusion of NH3 into the gulley region shows an important decrease on the partially chlorinated surface. The barrier decreases from 10.7 kcal/mol on the fully chlorinated surface to 2.6 kcal/mol on the partially chlorinated surface. This decrease is due to a lowering of steric interactions. We will see in a later section that this barrier greatly increases for the brominated surface. In general, the energy barriers of the others steps decrease only by a few kcal/mol. The activation barrier for the rotation of NH3 along the Si-N bond disappears on the partially chlorinated surface. The overall activation barrier on the partially 2619


Langmuir chlorinated surface is 11.4 kcal/mol, which is a few kcal/mol lower than the value of 13.9 kcal/mol obtained on the fully chlorinated surface. The reaction of ammonia with the SiH group (Figure 7c) has three steps: (a) diffusion into the gulley region (intermediate III), (b) the formation of the datively bonded complex (panel V), and (c) the formation of the Si-NH2 surface group with the release of a H2 molecule (panel VII). The energetics of the diffusion into the gulley region are the same as in Figure 7b, and the structures of the points labeled I-III are shown in Figure 7a. The thin solid line in Figure 7c corresponds to the reaction of ammonia on the fully hydrogenated surface presented in the previous section. Because of the presence of the surrounding chlorine atoms, the diffusion of NH3 into the gulley region now has a small barrier of 2.6 kcal/mol (point labeled II). The other steps are qualitatively similar to those for the fully hydrogenated surface. The overall activation barrier is 31.0 kcal/mol, which is to be compared to the barrier of 38.3 kcal/mol on the fully hydrogenated surface. The lowering of the barrier by 7.4 kcal/ mol is attributed to the stabilization of the transition state by the hydrogen bond interaction of NH3 with the surrounding chlorine atoms: two hydrogen atoms of NH3 point toward chlorine atoms with H 3 3 3 Cl distances of 2.26 and 2.31 Å (panel VI of Figure 7c). This gives an average of 3.7 kcal/mol, which is typical for hydrogen bond interactions. In summary, the partially chlorinated surface has groups with very different reactivities: the reaction of ammonia with SiCl has an overall barrier of 11.4 kcal/mol whereas the reaction with SiH has a barrier of 31.0 kcal/mol. The energy barriers decrease with respect to the fully hydrogenated and fully chlorinated surfaces. This is due to a lowering of repulsive interactions with the chlorine atoms (reaction with the SiCl group) and the stabilization of the transition state by hydrogen bond interactions (reaction with the SiH group). Ideally, the goal of the surface chlorination process is to replace the hydrogen monolayer of the hydrogenated surface fully with a chlorine monolayer. However, during the chlorination process, the surface is partially covered with chlorine and hydrogen. The partial coverage can be detected by IR spectroscopy41 because the local environment affects the SiH and SiCl stretching frequencies, as we showed in previous work.42 Taking into account that SiH and SiCl surface groups have such different reactivities, the partial chlorination of the surface may be viewed as a tool to functionalize the surface selectively at the highly reactive SiCl surface sites. Effect of Reaction Products on the Surface Reactivity. The maximum Si-NH2 coverage obtained experimentally in the reaction of ammonia with the chlorinated surface was 0.33 ML.18 It was suggested that intermolecular interactions could be responsible for the incomplete reaction of the Cl layer. To elucidate this issue, we studied two coadsorption systems: the reaction of ammonia with partially aminated surfaces and the reaction of ammonia on a surface with an ammonium chloride moiety that results from a previous amination step. We calculated the energy profiles for the reaction of NH3 with chlorinated surfaces with 25, 50, and 75% coverage of Si-NH2 groups (Figure S1 in Supporting Information). The overall activation-energy barriers are 14.2, 17.2, and 17.8 kcal/mol, respectively. Although these values are higher than on the fully chlorinated surface (12.4 kcal/mol, reaction through the dimers row), they are not high enough to block further reactions of ammonia. Therefore, we conclude that the increasing coverage of

ARTICLE

Figure 8. (a) Energy profile for the reaction of NH3 on a surface having an ammonium chloride moiety. Panel I in the inset shows the structure of the reactants. (b) Structure of the transition state (panel II) and the reaction products (panel III).

amine groups does not explain the incomplete reaction of the Cl layer observed experimentally.18 However, when NH3 reacts with surfaces that have Si-NH2 groups, the reaction product is an ammonium chloride moiety (Figure 1b and Figure S1 in Supporting Information) rather than an HCl molecule hydrogen bonded to a surface Si-NH2 group (as is the case on the fully chlorinated surface; see panel IX in Figure 2). We therefore investigated the reaction of ammonia with a surface that has an ammonium chloride moiety resulting from a previous amination step. Figure 8a shows the corresponding energy profile. The inset in Figure 8a shows the equilibrium structure of adsorbed ammonia. There is a strong hydrogen bond interaction of 10.7 kcal/mol between the surface ammonium ion and the NH3 molecule with a short hydrogen bond length of 1.76 Å. The energy profile in Figure 8 shows that the reaction occurs in one elementary step, producing another ammonium chloride moiety with the chloride ion within the gulley region (panel III). No datively bonded intermediate is observed in this case. This reaction has a high-activation-energy barrier of 26.5 kcal/mol. Panel II in Figure 8b shows the structure of the transition state. We attribute the high-energy barrier to the electrostatic repulsion between the chloride anions in the gulley region and to the repulsion of the adjacent ammonium ions. This barrier is much higher than the overall barrier of 12.4 kcal/mol (Figure 4a) obtained on the fully chlorinated surface and is also higher than the barriers in the range of 14.2-17.8 kcal/mol on the partially aminated surfaces (Figure S1 in Supporting Information). 2620


Langmuir

ARTICLE

Figure 9. Isocontour (0.06) of electron density during the reaction of NH3 through the dimers row of the chlorinated surface. (a) Datively bonded complex, (b) transition state, (c) ammonium chloride intermediate, and (d) HCl product.

Therefore, the blocking of further reactions with ammonia is due to the presence of reaction products on the surface. The elimination of reaction products is required to reduce the activation-energy barrier of the reaction. This effect has been reported in the reaction between H2O and surface Si-Cl species in the atomic layer growth of SiO2.43 It has been observed that the use of an organic base catalyst such as pyridine greatly accelerates this reaction. In this case, the base helps to eliminate the HCl product from the surface.43 Electronic Factors Contributing to the Stability of Intermediates and Transition States on the Chlorinated Surface. In this section, we investigate the factors that determine the high reactivity of the chlorinated surface as compared to the less-reactive hydrogenated surface. To gain deeper insight into the electronic structure as the reaction of NH3 with the chlorinated surface proceeds, we have plotted in Figure 9 the 0.06 isocontour of electron density during the reaction through the dimers row. For the sake of clarity, Figure 9 shows only one dimer separated by the gulley region. Figure 9a corresponds to the datively bonded complex. The isocontour of electron density clearly shows the Si-Cl and Si-Si

Figure 10. (a) Electron density difference plot ((0.001 isocontour) for the datively bonded complex in the reaction of NH3 through the dimers row of the chlorinated surface. The positive isocontour in red indicates electron density accumulation, and the negative isocontour in blue indicates electron density depletion with respect to the bare NH3 molecule and the bare chlorinated surface with same structure as in the transition state. Isocontour ((0.006) for (b) the datively bonded complex and (c) the transition state.

covalent bonds. In the transition state (Figure 9b), the electron density becomes spherically symmetrical around the chlorine atom, indicating that there is no covalent bonding with the silicon atom. Figure 9c corresponds to the intermediate in which the chlorine atom is within the gulley region (panel V in Figure 4). The spherical distribution of electron density around chlorine shows its anionic nature, as in the transition state (Figure 9b). Figure 9d shows the electron density of the HCl product 2621


Langmuir

ARTICLE

Figure 11. (a) Isocontour ((0.06) of electron density for the transition state in the reaction of NH3 through the dimers row of the hydrogenated surface. (b) Electron density difference plot ((0.001 isocontour) for the same transition state. The positive isocontour in red indicates electron density accumulation, and the negative isocontour in blue indicates electron density depletion with respect to the bare NH3 molecule and the bare hydrogenated surface with same structure as in the transition state.

hydrogen bonded to the surface amine group. Figure 9c,d shows an increasing accumulation of electron density around the Si-N bond indicating the strengthening of this bond as the reaction proceeds. The charge flow that occurs during the formation of the datively bonded complex and the transition state for the reaction of NH3 through the dimers row is shown in the electron density difference plots of Figure 10. We subtracted the density of the bare NH3 and the bare surface from the density of the surfaceNH3 system. Figure 10a shows the (0.001 isocontour of the density difference for the datively bonded complex. There is a region of depletion of electron density around NH3 and a region of charge accumulation around the chlorine atom. This shows the onset of the formation of the ammonium chloride moiety. In turn, the positive charge around NH3 polarizes the electron density of the remaining three SiCl groups, as can be seen in the red lobes of charge accumulation pointing toward NH3. In Figure 10b,c, we compare the (0.005 isocontour for the datively bonded complex and for the transition state. This contour allows one to appreciate a lobe of charge accumulation around the Si-N bond, indicating the formation of this bond. Figure 10c shows that the accumulation of charge around the chlorine atom released into the gulley region is more pronounced than for the chlorine atom of the datively bonded complex (Figure 10b). This clearly shows the anionic nature of the chlorine atom within the gulley region. The higher charge accumulation around the chlorine atom in the gulley region in Figure 10c correlates with a more pronounced charge-depletion region around NH3. The region of

Figure 12. Reaction of NH3 with the brominated surface through (a) the gulley region and (b) the dimers row. The insets show the structure of reactants and products. The thin solid line corresponds to the reaction of NH3 with the chlorinated surface.

charge accumulation around the chlorine atom has the shape of a p orbital, indicating that, as expected, the charge is transferred to the empty p orbital. The L€owdin charge of the chlorine atom in the gulley region is -0.5. The electronic structure of the transition state in the reaction of NH3 with the hydrogenated Si(100) surface through the dimers row (panel II in Figure 5) is quite different from that on the chlorinated surface. Figure 11a shows the 0.06 electron density contour. Only one dimer is shown for the sake of clarity. In the transition state, the Si-H distance has enlarged to 1.936 Å from the equilibrium value of 1.503 Å and the electron density between both atoms is very low, indicating that the bond is broken. On the contrary, there is a high accumulation of electron density between this H atom and the H atom from NH3, indicating the onset of the formation of the covalent bond that will give rise to the H2 molecule. The charge flow in the transition state is shown in the electron density difference plot of Figure 11b. The difference was calculated by subtracting the density of bare NH3 and the hydrogenated surface (both with same geometry as in the transition state) from the density of the transition state. Figure 11b shows a region of charge accumulation between the hydrogen atoms that will give rise to the H2 molecule and a region of charge depletion in the enlarged N-H bond. There is also a region of charge accumulation along the Si-N bond. 2622


Langmuir Therefore, the species involved in the transition state are covalently bonded, and the charge flow shows the breakage of the N-H bond and the formation of H-H and Si-N bonds. No evidence of an ammonium cation is observed as on the chlorinated surface. These results explain the high stability of the transition state on the chlorinated surface. The enlargement of the Si-Cl bond on this surface produces a chloride anion that is stabilized by the electrostatic interaction with the ammonium ion. NH3 is virtually not deformed in the transition state, having a pyramidal structure with the three N-H bonds with nearly the same length. On the contrary, the transition state on the hydrogenated surface involves the breakage and formation of covalent bonds, and the NH3 molecule is very deformed with respect to the gas-phase structure. This is energetically more costly and produces a destabilization of the transition state on this surface. On the chlorinated surface, the deformation energy of NH3 (calculated as the difference between the energy of NH3 with the same geometry as in the transition state and the equilibrium energy of a free NH3 molecule) is only 0.6 kcal/mol, but on the hydrogenated surface, the deformation energy is 20.6 kcal/mol (reaction through the dimers row). Effect of Steric Factors on Surface Reactivity: The BrSi(100)-2 1 Surface. We investigated this surface in order to evaluate the influence of steric effects on the reaction mechanism. Figure 12 shows the energy profiles for the reaction of ammonia through the gulley region (Figure 12a) and through the dimers row (Figure 12b). The thin solid line corresponds to the reaction on the chlorinated surface and is included for comparison. Figure 12 shows that the energy profiles on the Br-Si(100)-2 1 and Cl-Si(100)-2 1 surfaces are qualitatively similar. Effectively, the same number of elementary steps was found on both surfaces. The energetics of the surface reaction of ammonia with a SiX (X = Cl, Br) group (SiX þ NH3(g) f Si-NH2 þ HX(ads)) is similar on both surfaces: ΔE = 6.2 kcal/mol on the brominated surface vs ΔE = 5.8 kcal/mol on the chlorinated surface (reaction through the interdimers row, Figure 12a). The same is observed in Figure 12b for the reaction through the dimer row: the energy change is 7.5 kcal/mol on the brominated surface and 6.3 kcal/mol on the chlorinated surface. However, the activation-energy barriers on the brominated surface are higher. In particular, the major differences are observed in the first reaction step. Figure 12a shows that the diffusion of ammonia into the gulley region has an activationenergy barrier of 22.7 kcal/mol whereas on the chlorinated surface this value is only 10.7 kcal/mol. Figure 12b shows that in the reaction through the dimers row the energy barrier for the formation of the datively bonded complex is 14.3 kcal/mol on the brominated surface and 6.4 kcal/mol on the chlorinated surface. The overall barrier when NH3 reacts through the gulley region is 22.7 kcal/mol (Figure 12a) versus 13.9 kcal/mol on the chlorinated surface. The reaction through the dimers row has an overall barrier of 19.9 kcal/mol (Figure 12b) versus 12.3 kcal/ mol on the chlorinated surface. Therefore, the difference in the activation barriers between both surfaces shows the contribution of steric interactions to the reaction mechanism.

’ CONCLUSIONS We investigated the reactivity of NH3 with the Cl-Si(100)-2 1, Br-Si(100)-2 1, H-Si(100)-2 1, and Si(100)-2 1 surfaces. The factors determining the different reactivities of

ARTICLE

these surfaces toward ammonia have been unveiled in this work. We found that the chlorinated surface is the most reactive, even more so than the bare Si(100)-2 1 surface. The balance between electronic and steric factors explains the different reactivities of halogenated surfaces. The interaction of the ammonia lone pair electrons with the positively charged silicon atoms of the surface SiX groups gives rise to a stable datively bonded complex intermediate. In the transition state preceding the breakage of a Si-X bond and the formation of a Si-NH2 bond, the charge redistribution is more pronounced than in the datively bonded intermediate, giving rise to an ionic transition state of the form Si-NH3þX-. Steric effects also play an important role in the surface reactivity, making brominated surfaces less reactive than chlorinated surfaces. The voluminous bromine atoms prevent the nucleophilic attack of ammonia on the silicon atom of a SiBr group, which leads to an increase in the activation-energy barrier. Therefore, the surface reactivity can be tailored by either changing the nature of the halogen atom or changing its surface coverage. These results also allow us to predict that activationenergy barriers are expected to be higher on the chlorinated Si(111) surface than on the chlorinated Si(100) surface because the 111 surface has a higher density of chlorine atoms than does the 100 surface.44 Coadsorption effects play an important role in the surface reactivity. The reaction of ammonia on a surface with ammonium chloride moieties (the reaction products) has a much higher energy barrier than that on the fully chlorinated surface. This implies that further reactions with ammonia are blocked, and this explains the experimental observation that only a maximum coverage of 0.33 ML of SiNH2 groups is observed.18

’ ASSOCIATED CONTENT

bS

Supporting Information. The energy profiles and the structure of reactants and products for the reaction of NH3 with surfaces with 25, 50, and 75% coverage of Si-NH2 groups. Coordinates and energies of all structures shown in Figures 1-8 and 12. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: 54-351-4344972. E-mail: [email protected].

’ ACKNOWLEDGMENT Financial support from FONCyT (PICT 2005-32893), CONICET (PIP 5903), and SECYT-UNC is gratefully acknowledged. F.A.S. thanks CONICET for a doctoral fellowship. ’ REFERENCES (1) Molecular Electronics: Science and Technology; Aviram, A., Ratner, M., Eds.; Annals of the New York Academy of Sciences; New York Academy of Sciences: New York, 1998; Vol. 852. (2) Aureau, D.; Varin, Y.; Roodenko, K.; Seitz, O.; Pluchery, O.; Chabal, Y. J. J. Phys. Chem. C 2010, 114, 14180. (3) Le Saux, G.; Ciampi, S.; Gaus, K.; Gooding, J. J. ACS Appl. Mater. Interfaces 2009, 1, 2477. (4) Lin, Z.; Strother, T.; Cai, W.; Cao, X. P.; Smith, L. M.; Hamers, R. J. Langmuir 2002, 18, 788. 2623


Langmuir (5) Kim, M. K.; Baik, J.; Jeon, C.; Song, I.; Nam, J. H.; Hwang, H. -N.; Hwang, C. C.; Woo, S. H.; Park, C.-Y.; Ahn, J. R. Surf. Sci. 2010, 604, 1598. (6) Wolkow, R. A. Annu. Rev. Phys. Chem. 1999, 50, 413. (7) Wayner, D. D. M.; Wolkow, R. A. J. Chem. Soc., Perkin Trans. 2002, 2, 23. (8) Buriak, J. M. Chem. Rev. 2002, 102, 1271. (9) Hamers, R. J. Annu. Rev. Anal. Chem. 2008, 1, 707. (10) Ciampi, S.; Harper, J. B.; Gooding, J. J. Chem. Soc. Rev. 2010, 39, 2158. (11) Higashi, G. S.; Chabal, Y. J.; Trucks, G. W.; Ragavachari, K. Appl. Phys. Lett. 1990, 12, 656. (12) Terry, J.; Linford, M. R.; Wigren, C.; Cao, R.; Pianetta, P.; Chidsey, C. E. D. Appl. Phys. Lett. 1997, 71, 1056. (13) Webb, L. J.; Lewis, N. S. J. Phys. Chem. B 2003, 107, 5404. (14) Bansal, A.; Li, X. L.; Lauermann, I.; Lewis, N. S.; Yi, S. I.; Weinberg, W. H. J. Am. Chem. Soc. 1996, 118, 7225. (15) Bergerson, W. F.; Mulder, J. A.; Hsung, R. P.; Zhu, X.-Y. J. Am. Chem. Soc. 1999, 121, 454. (16) Zhu, X.-Y.; Mulder, J. A.; Bergerson, W. F. Langmuir 1999, 15, 8147. (17) Ardalan, P.; Sun, Y.; Pianetta, P.; Musgrave, C. B.; Bent, S. F. Langmuir 2010, 26, 8419. (18) Finstad, C. C.; Thorsness, A. G.; Muscat, A. J. Surf. Sci. 2006, 600, 3363. (19) Perrine, K. A.; Teplyakov, A. V. Chem. Soc. Rev. 2010, 39, 3256. (20) Dai, M.; Wang, Y.; Kwon, J.; Halls, M. D.; Chabal, Y. J. Nat. Mater. 2009, 8, 825. (21) Takaoka, T.; Kusunoki, I. Surf. Sci. 1998, 412, 30. (22) Bozso, F.; Avouris, Ph. Phys. Rev. B 1988, 38, 3937. (23) Dresser, M. J.; Taylor, P. A.; Wallace, R. M.; Choyke, W. J.; Yates, J. T., Jr. Surf. Sci. 1989, 218, 75–107. (24) Widjaja, Y.; Mysinger, M. M.; Musgrave, C. B. J. Phys. Chem. B 2000, 104, 2527. (25) Cao, X.; Hamers, R. J. J. Am. Chem. Soc. 2001, 123, 10988. (26) Kim, H. J.; Cho, J. H. Phys. Rev. B 2004, 69, 233402. (27) Chung, O. N.; Kim, H.; Chung, S.; Koo, J. Y. Phys. Rev. B 2006, 73, 033303. (28) Bowler, D. R.; Owen, J. H. G. Phys. Rev. B 2007, 75, 155310. (29) Owen, J. H. G. J. Phys.: Condens. Matter 2009, 21, 443001 and references therein.. (30) Hamers, R. J. Surf. Sci. 2006, 600, 3361. (31) Cao, X.; Hamers, R. J. J. Phys. Chem. B 2002, 106, 1840. (32) Lange, B.; Schmidt, W. G. Surf. Sci. 2008, 602, 1207. (33) Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I.; Dal Corso, A.; de Gironcoli, S.; Fabris, S.; Fratesi, G.; Gebauer, R.; Gerstmann, U.; Gougoussis, C.; Kokalj, A.; Lazzeri, M.; Martin-Samos, L.; Marzari, N.; Mauri, F.; Mazzarello, R.; Paolini, S.; Pasquarello, A.; Paulatto, L.; Sbraccia, C.; Scandolo, S.; Sclauzero, G.; Seitsonen, A. P.; Smogunov, A.; Umari, P.; Wentzcovitch, R. M. J. Phys.: Condens. Matter 2009, 21, 395502http://www.quantum-espresso.org. (34) Vanderbilt, D. Phys. Rev. B 1990, 41, 7892. (35) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (36) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188. (37) Juarez, M. F.; Soria, F. A.; Patrito, E. M.; Paredes-Olivera, P. J. Phys. Chem. C 2008, 112, 14867. (38) (a) Henkelman, G.; Uberuaga, B. P.; Jonsson, H. J. Chem. Phys. 2000, 113, 9901. (b) Henkelman, G.; Jonsson, H. J. Chem. Phys. 2000, 113, 9978. (39) Kokalj, A. Comput. Mater. Sci. 2003, 28, 155www.xcrysden.org. (40) Lee, H. S.; An, K. -S.; Kim, Y.; Choi, C. H. J. Phys. Chem. B 2005, 109, 10909. (41) Ferguson, G. A.; Rivillon, S.; Chabal, Y. J.; Raghavachari, K. J. Phys. Chem. C 2009, 113, 21713. (42) Juarez, M. F.; Patrito, E. M.; Paredes-Olivera, P. J. Phys. Chem. C 2009, 113, 681.

ARTICLE

(43) Klaus, J. W.; Sneh, O.; George, S. M. Science 1997, 278, 1934. (44) This was confirmed by performing a calculation on a (2 2) unit cell of the Cl-Si(111) surface. The reaction of ammonia on this surface has an activation-energy barrier of 34.4 kcal/mol which is much larger than the barrier of 12.4 kcal/mol found on the Cl-Si(100) surface (reaction through the dimers row).

2624
