Hydrosilylation of Styrene on Water-Saturated Si(001)-2×1 at Room

Jun 21, 2011 - ... Physics: Condensed Matter 2015 27 (5), 054005. 8.25 Hydrometallation Group 4 (Si, Sn, Ge, and Pb). A.P. Dobbs , F.K.I. Chio. 2014,9...
0 downloads 0 Views 1MB Size
ARTICLE pubs.acs.org/JPCC

Hydrosilylation of Styrene on Water-Saturated Si(001)-21 at Room Temperature F. Bournel,*,† J.-J. Gallet,† D. Pierucci,† A. Khaliq,† F. Rochet,† and A. Pietzsch‡ †

Laboratoire de Chimie Physique Matiere et Rayonnement, UMR CNRS 7614, Universite Pierre et Marie Curie, Paris 6, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France ‡ MAX-Lab Synchrotron, Lund University, P.O. Box 118, 22100 Lund, Sweden

bS Supporting Information ABSTRACT: The Si(001)-21 surface saturated by water is characterized by the passivation of nearly all of the dimerized atoms by H/OH terminations, except isolated dangling bonds, whose areal density is in the range of 1.5 ( 0.2  10 2 defects per Si atom. Therefore the water-saturated Si(001)-21 surface presents similarities with the defective H-terminated Si(001)21 surface (presence of monohydrides and isolated dangling bonds), on which alkene molecules are known to be grafted via a radical-based hydrosilylation mechanism initiated at silicon dangling bonds. These common features stimulated the present study devoted to the reactivity of the water-saturated surface (n+-doped substrate) with styrene (H2CRdCβH C6H5) at room temperature. Using synchrotron radiation X-ray photoemission spectroscopy (XPS), our aim was to estimate the extent of styrene growth and to characterize the chemistry of the adsorbed molecule. XPS showed that styrene does react with the surface: after an exposure of 6.7 L (900 s  0.75  10 8 Torr), we estimate that about 0.2 molecules per Si dimer (∼0.1 molecule per Si atom) are grafted on the surface. The C 1s XPS spectrum is consistent with a hydrosilylation product, Si CRH2 CβH2 C6H5. Indeed we found that the C 1s spectral shape of styrene mono-σ bonded to the water-covered surface is markedly different from that of styrene di-σ bonded to the clean Si(001)-21 surface, confirming the specificity of the reaction product formed on the former surface. Mechanistically, a radical-based hydrosilylation reaction is the most plausible, as theoretical works indicate that the activation barrier of the latter mechanism is much lower than that of a direct, concerted mechanism. The C 1s spectral shape also excludes a reaction of the molecule with surface hydroxyls, leading to the formation of monohydroxyls CβOH(H) (radical-based reaction) or of Si O C linkages (Markovnikov or anti-Markovnikov addition).

1. INTRODUCTION Recent experiments revealed that the Si(001)-21 surface modified by exposure to water vapor can be used as a reactive surface for the attachment of organic molecules such as carboxylic acids1 and ethoxysilanes.2 This surface is commonly described as made of intact Si Si dimers whose dangling bonds are decorated by H and OH endings.3 In fact, due to the competition of two dissociation paths (“on-dimer” and “intra-row” fragmentation4,5), the water-saturated surface exhibits defects (silicon dangling bonds, corresponding to H and OH vacancies) and complex H/OH patterns. We give in Figure 1 a schematic view of the waterreacted surface, as revealed by a scanning tunneling microscopy (STM) study we made recently.6 Hydrides and hydroxyls can be distinguished in STM images, due to an apparent height difference. Their ordering is only local. “Linear” H Si Si OH patterns, where all H's sit on the same side of the dimer row, do exist, but only span a few dimers (typically up to four or five). The observed zigzag patterns are rarer than the aligned patterns. One remarkable observation is the presence of HO Si Si OH units, that is, two OH units on the same dimer, likely paired by a r 2011 American Chemical Society

hydrogen bond, amounting to about 20% of the reacted dimers. H Si Si H units are also seen. STM shows that the silicon dangling bonds, whose density is in the range of 1.210 2 to 1.7  10 2 defects/Si atom, are isolated (hence the term isolated dangling bond, IDB), in contrast with the paired dangling bonds constituting the C-defects of Si(001)-21 exposed to low doses of water.7 Both X-ray photoemission spectroscopy (XPS) measurements of the surface band bending and STM images concur to demonstrate that, on highly doped (n+) substrates, the IDBs are negatively charged.6 Water-saturated Si(001)-21 presents similarities with Hterminated Si(001)-21. First, monohydrides are present on both surfaces. Second, the IDBs (also present on H-terminated Si(001)21 due to incomplete passivation or created by H abstraction under a STM tip) possess the same electronic structure.6 Therefore, the surface reactivity of water-saturated Si(001)-21 with Received: March 29, 2011 Revised: June 14, 2011 Published: June 21, 2011 14827

dx.doi.org/10.1021/jp202913y | J. Phys. Chem. C 2011, 115, 14827–14833

The Journal of Physical Chemistry C

Figure 1. Schematic view of the water-reacted Si(001)-21 surface. (a) Isolated dangling bonds remain on the water saturated surface at room temperature due to the intrarow dissociation channel; (b) HO Si Si H patterns: OH are arranged on the same side of the dimer row (linear) or on alternating opposite side (zigzag), the former being the predominant pattern; (c) the intrarow molecular dissociation can also lead to double OH (double H) decorated dimers.

π-bonded molecules could present mechanisms that are also operative on the H-terminated surface, in particular hydrosilylation reactions. IDBs on H-terminated Si surfaces are responsible for the grafting of alkene molecules at room temperature via a radical chain reaction (this was first discovered by Chidsey and co-workers on the defective H-terminated Si(111)-11 more than a decade ago8,9 and then extended to the H-terminated Si(001)-21 surface10). The basic scheme of the Chidsey mechanism is given in Figure 2a in the particular case of an alkene. It involves the formation of radical adduct, followed by the abstraction of a nearby hydrogen to give the product. Takeuchi et al.,11via a periodic density functional theory (DFT) approach, have calculated an H abstraction barrier energy of only 0.59 eV, in the particular case of styrene on H-terminated Si(001)-21. As an alternative to the radical-based hydrosilylation mechanism, a direct hydrosilylation of the CdC unit with a surface hydride has been proposed by Coletti et al. and studied theoretically using a DFT cluster approach.12 The authors consider a direct concerted pathway that passes through a four-center transition state (Figure 2b), that then evolves to the same final product as that of the Chidsey mechanism. The calculated energy barrier for ethylene is very high, ∼60 kcal 3 mol 1 (∼2.6 eV), much higher that the H-abstraction barrier of the radical chain mechanism. Therefore, as indicated by theoretical calculations,

ARTICLE

the radical-based hydrosilylation mechanism is certainly kinetically favored, at least at room temperature. We have already noted6 that the observed grafting of a carboxylic acid to the water-terminated Si(001)-21 surface1 cannot be used as evidence for a radical-based hydrosilylation on this surface. Indeed, while on H-terminated Si(001)-21 surface the Chidsey mechanism is observed at room temperature for π(CdO) functionalities like those in aldehydes,13 in the case of carboxylic acid an ester link could be formed via a water elimination mechanism between the hydroxyl group of the acid and the surface silanol. Therefore, it is preferable to use alkenes as test molecules for this possible chain reaction. In the present work, we have exposed the water-saturated n+ doped Si(001)-21 surface to several langmuirs (L) of styrene at room temperature, under pressures in the 10 8 Torr range. Styrene is the chosen alkene, because it reacts with the defective H-terminated Si(001)-21 surface,10 contrary to ethylene and propylene for which the radical intermediate is not stabilized enough, and desorbs before abstracting a hydrogen.11 However, the presence of SiOH on water-terminated Si(001)21 could provide reaction mechanisms competitive with the hydrosylilation reaction. Besides H abstraction that leads to the formation of a Cβ H new bond (Figure 3a), OH radicals may be also captured by the radical adduct to form a Cβ OH bond (Figure 3b). Monohydroxyl carbon atoms are indeed formed as adventitious products (by reaction with water or surface hydroxyls) during the thermal hydrosilylation of H-terminated Si(111) by phenylacetylene.14 A direct addition reaction of the styrene CdC unit with a surface silanol could be also envisaged, leading to the formation of Si O C units: this reaction is documented for alkynes reacting with hydroxyls on oxidized silicon surfaces via a Markovnikov or an anti-Markovnikov addition.15 We used core-level XPS to follow the extent of the reaction by measuring the intensity of the C 1s core-level and to determine the chemical nature of the linkage between the molecule and the surface. As discussed in the paper, C 1s spectral analysis enables the identification of the possible bonding units of styrene on water-reacted Si(001) (hydrosilylation products, monohydroxyl carbon, Si O C linkage). Additional information on changes in the areal density of (negatively charged) IDB upon exposure to styrene can also be obtained by measuring the surface band bending with surface sensitive Si 2p core-level XPS.6

2. EXPERIMENTAL SECTION The experiments were carried out at beamline I511-1 (surface end-station) of MAX-Lab Synchrotron Facility (Sweden).16 All surface treatments were made in the ultrahigh vacuum preparation chamber connected to the analysis chamber. Highly doped (phosphorus) n+-type Si(001) wafers (resistivity 0.003 Ω 3 cm, ND ∼ 2  1019 cm 3) were cleaned from their native oxide by flash annealing (Joule effect) at 1100 °C after prolonged degassing at 600 °C in ultrahigh vacuum. Water-covered Si(001)21 surfaces were prepared by exposing the clean surface to H2O at room temperature to 1 L (1 langmuir (L) = 10 6 Torr 3 s) of water (under a nominal pressure of 0.75  10 8 Torr). This exposure was sufficient to saturate the surface. Then the watercovered surface was exposed to increasing doses of styrene at room temperature under a pressure of 0.75  10 8 Torr. The styrene molecule (C6H5 CHdCH2), liquid at room temperature, was purchased from Aldrich (99.99%) and purified by several freeze pump thaw cycles before dosing. The same 14828

dx.doi.org/10.1021/jp202913y |J. Phys. Chem. C 2011, 115, 14827–14833

The Journal of Physical Chemistry C

ARTICLE

Figure 2. Hydrosilylation of alkenes. (a) The radical-based mechanism (Chidsey) on a defective, H-terminated silicon surface. The dot (on top of silicon and on the β-carbon) designates an unpaired electron. (b) The direct, concerted reaction mechanism.

procedure is used to degas ultra pure water (18 MΩ 3 cm). The molecules (H2O and styrene) were introduced in the preparation chamber and onto the silicon surface through a UHV compatible fine leak-valve under constant pressure (background dosing). The XPS measurements were made using a Scienta R4000 electron analyzer. The photoelectrons are detected at 45° from the sample surface normal and at 45° from the polarization vector E contained in the sample surface plan. The C 1s core-level spectra were recorded at hν = 345 eV (with an overall experimental resolution of 100 meV). The Si 2p levels were measured at hν = 150 eV, with an overall resolution of 70 meV. A Shirley background is subtracted to all core-level spectra presented here. Then the spectra are fitted by sums of Voigt curves, that is, the convolution of a Gaussian (of full-width at half-maximum GW) by a Lorentzian (of full-width at halfmaximum LW). For C 1s and Si 2p the chosen LW are 80 meV17 and 45 meV,18 respectively. The total full-width at half-maximum (fwhm) of the Voigt profile is calculated using the formula: fwhm ≈ 0.5346 LW + (0.1269 LW2 + GW2)1/2.19 For the Si 2p spectra fitting we used the statistical spin orbit ratio of the j = 3/2 and j = 1/2 components, and a spin orbit distance of 0.602 eV.20 The zero binding energy (BE) (i.e., the Fermi level) is taken at the leading edge of a clean metal surface in electrical contact with the silicon crystal. The Si 2p3/2 binding energy is found at 99.41 eV for the water-terminated Si(001)-21 surface and at 99.44 eV for the styrene-reacted/water-terminated Si(001)-21 surface. The Si 2p3/2 binding energy is used to measure the energy distance between the surface Fermi level and the top of the silicon valence band,21 and hence the surface band bending qVBB (for a n+ substrate, the clean and water-covered Si(001)-21 surfaces present an upward band bending due to the accumulation of negative charge at the surface and the formation of a depletion layer on the semiconductor side).6 From qVBB, the net surface charge density ns associated with band bending can be calculated using the relation |ns| = (2εNDqVBB)1/2, where ND is the doping concentration and εs the silicon dielectric constant.22

Here we shall examine variations in band bending associated to the adsorption of styrene on water-saturated Si(001)-21. To quantify the amount of deposited carbon, the C 1s kinetic energy is chosen (hν ∼ 390 eV) close to that of the Si LVV Auger peak (∼90 98 eV range). The C 1s spectra are normalized against an equal Si LVV height. Then equivalent carbon surface quantities are given after comparison with the styrene-saturated Si(001) surface (∼6.7 L) for which an average molecule surface density of one molecule per dimer is reported in ref 23.

3. RESULTS AND DISCUSSION Styrene does react with the water-covered surface. The normalized C 1s intensity of the water-covered surface exposed to increasing styrene doses are given in Table 1. By comparison with the styrene-covered Si(001)-21 surface (one molecule per dimer), the estimated coverage at 6.7 L is 0.18 molecules per dimer (0.09 molecules per surface Si atom). The high resolution C 1s spectrum (hν = 345 eV) of the watercovered surface exposed to 6.7 L of styrene is reported in Figure 4b. The fit parameters are reported in Table 2. The spectrum is fitted with three Voigt components, positioned at 284.15, 284.79, and 285.22 eV, respectively (18, 69, and 13% of the spectral weight, respectively). Indeed, if one assumes that the adduct is the result of a hydrosilylation reaction (e.g., the product depicted in Figure 3a), one can, in a first step, distinguish three chemical environments: the phenyl carbons, the β carbon (aliphatic) and the R carbon bonded to silicon. There is evidence in the experimental literature for a chemical shift difference of 0.22 eV in polystyrene between aromatic carbons (binding energy of 284.77 eV) and aliphatic carbons (284.99 eV).24 This is further supported by a recent “density functional theory Δ self consistent field” (DFT ΔSCF) calculation of the ionization energies (i.e., binding energies referenced to the vacuum level) of ethylbenzene,25 that shows that the average 1s ionization energy of the aliphatic carbons is shifted by +0.50 eV with respect to the 1s average 14829

dx.doi.org/10.1021/jp202913y |J. Phys. Chem. C 2011, 115, 14827–14833

The Journal of Physical Chemistry C

ARTICLE

Figure 3. Schemes of possible reactions on a water-saturated Si(001)-21 surface: (a) radical chain reaction involving the jump of a neighboring H atom (as in defective H-terminated Si(001)-21 surfaces); (b) radical chain reaction involving a neighboring hydroxyl; (c) styrene di-σ bonded to a silicon dimer, as in the case of the clean Si(001)-21 surface (“end-on” intrarow di-σ bonding could also exist).

Table 1. Normalized C 1s Intensities Measured at hν = 390 eV for the Water-Covered Surface Exposed to Increasing Doses of Styrene, under a Pressure of 0.75  10 8 Torra surface water-covered Si(001)-21

clean Si(001)-21

styrene doses (L) normalized C 1s intensity 0.45

0.12

4.9

0.16

6.7

0.18

6.7

1.00

a

A comparison is made with the Si(001)-21 surface directly exposed to styrene, for which a styrene surface density of one molecule per dimer is reported in ref 23 at saturation.

ionization energy of the aromatic ones. Therefore we can attribute the main peak at 284.79 eV and its satellite at 285.22 eV to the aromatic carbons and the β carbon (CβH2 unit), respectively. On the other hand, we attribute the lowest binding energy component

at 284.15 eV to the R carbon, bonded to silicon. In fact, the binding energy of Si CH3 units resulting from the dissociation of methyliodide on Si(001)-21 is found at 284.10 eV.26 As this component has an appreciable weight (18%), this excludes an oligomerization process starting from the radical intermediate (step II in Figure 2). So the C 1s spectrum is consistently interpreted assuming that the β carbon captures one H to give a CβH2 unit. We can also definitely eliminate the possibility of an OH abstraction by the β carbon (Figure 3b) and the formation of a monohydroxyl carbon, that should be strongly shifted to higher binding energy than the “phenyl” main peak (284.79 eV) and its CβH2 satellite (285.22 eV). To give an example, in gas-phase ethanol (CH3 CH2 OH), the binding energy (the asterisk denotes the ionized atom) of CH3 C*H2 OH is shifted up by +1.40 eV with respect to that of C*H3 CH2 OH.27 Monohydroxyl carbons formed during the hydrosilylation process of H-terminated Si(111) (by reaction with water or hydroxyls) 14830

dx.doi.org/10.1021/jp202913y |J. Phys. Chem. C 2011, 115, 14827–14833

The Journal of Physical Chemistry C

ARTICLE

produce a photoemission C 1s component at 286.70 eV.14 This is corroborated by DFT (B3LYP) C 1s ionization energy calculations showing that the energy shift between CH3 C*H3 and CH3 C*H2OH is 1.56 eV (see Table S5 in the Supporting Information of ref 14). Therefore, a peak in the ∼286.2 286.7 eV range is expected for a Cβ(H)OH unit, which is not observed in Figure 4b. The addition of the molecule to a silanol group, leading to a Si O C linkage, is also excluded as we do not see any C 1s component at ∼286.5 eV.15 The stoichiometric atomic distribution CR Si/aromatic C/CβH2 is 12.5:75:12.5, somewhat at variance with that deduced from the decomposition of the C 1s spectrum (18:69:13). This can be due in part to π π* shake ups “stealing” intensity from the aromatic carbon peak at 284.79 eV. As shown in the Supporting Information (Figure S1), the shake ups of styrene di-σ bonded to the clean surface (appearing centered at ∼7 eV to lower kinetic energy than the main peak) amount to ∼8% of the whole C 1s spectrum (i.e., ∼10% of the total aromatic ring intensity). The corrected intensity distribution CR Si/aromatic C/CβH2 then would be 16.8:71:12.2. In Figure 4a we give the C 1s spectrum of styrene-saturated Si(001)-21, for the purpose of emphasizing that styrene bonds

in a different manner on the water-covered and on the clean, reconstructed surface. The C 1s spectrum of the molecule di-σ bonded to a silicon dimer28 (Figure 3c) appears much less structured than the one obtained after adsorption of styrene on the water-covered surface (Figure 3a), despite the overall experimental resolution is the same. In particular the Si C component expected at ∼284.10 eV does not show up. Indeed the experimental curve is satisfactorily fitted with a single broad Voigt function of fwhm = 0.91 positioned at 284.61 eV. This means that the range in energy shift for di-σ styrene is much smaller than the ∼1 eV range of the mono-σ adduct of Figure 3a. Recent DFT C 1s ionization energies for styrene di-σ bonded to silicon, equivalent to the 1,2-bridge structure of phenylacetylene, confirm this point (see table S4 in the Supporting Information of ref 14): the R and β carbon 1s ionization energies are found at 289.48 and 289.70 eV, respectively, while the average aromatic carbon 1s energy is at 289.74 eV (with an energy shift range of

Figure 5. XPS Si 2p spectra measured at hν = 150 eV of the watercovered doped Si(001)-21 surface (1 L of water) before (a) and after (b) exposure to 6.7 L of styrene. The takeoff angle of the electrons, with respect to the surface normal, is 45°. The experimental curves (dots) are fitted with sums of Voigt components (black solid lines). The Si 2p3/2 (Si 2p1/2) fitting components are the red solid (red dotted) lines. The Si 2p3/2 binding energy, the widths, and the spectral weights of the various components are given in Table 3.

Figure 4. XPS C 1s spectra measured at hν = 345 eV. The takeoff angle of the electrons, with respect to the surface normal, is 45°. The experimental curves (dots) are fitted (fits are black solid lines) with one or more Voigt components (red solid lines). (a) Clean Si(001)-21 exposed to 6.7 L of styrene at room temperature. (b) Water-terminated Si(001)-21 surface exposed to 6.7 L of styrene at room temperature. Fit parameters are in Table 2.

Table 2. Binding Energies, Gaussian Widths (GW), fwhm, and Spectral Weights of the Voigt Components Used to Fit the C 1s Experimental Spectra Shown in Figure 4a experiment

C 1s binding energy (eV)

GW (eV)

fwhm (eV)

weight (%)

Si 2p3/2 binding energy (eV)

styrene on water-covered

284.15

0.50

0.54

18

99.44

Si(001) (6.7 L)

284.79

0.60

0.54

69

285.22

0.50

0.54

13

284.61

0.87

0.91

100

styrene on clean Si(001) (6.7 L)

99.42

a

The Lorentzian width (LW) was 0.08 eV for all components. Note that the band bending (monitored by the Si 2p3/2 binding energy) is almost the same on both surfaces (within 20 meV); thus C 1s binding energies can be directly compared. 14831

dx.doi.org/10.1021/jp202913y |J. Phys. Chem. C 2011, 115, 14827–14833

The Journal of Physical Chemistry C

ARTICLE

Table 3. Binding Energies (Si 2p3/2), Surface Core Level Shifts (SCLS) Referenced to the Bulk Component Binding Energy, Gaussian Widths (GW), fwhm, and Spectral Weights of the Voigt Components Used to Fit the Si 2p Experimental Spectra Shown in Figure 5a surface water-covered (1 L)

water-covered + styrene (6.7 L)

a

component Si

1+

Si 2p3/2 binding energy (eV)

SCLS (eV)

GW (eV)

fwhm (eV)

weight (%)

attribution

100.37

+0.96

0.41

0.43

9.5

S2

99.67

+0.26

0.30

0.32

13.9

SiH/2nd plane

SiOH

B

99.41

0

0.30

0.32

40.4

bulk

S1

99.12

0.30

0.30

0.32

2.8

Si1+ S2

100.40 99.71

+0.96 +0.27

0.41 0.30

0.43 0.32

10.8 14.2

SiOH SiH/2 plane/Si C

B

99.44

0

0.30

0.32

38.8

bulk

S1

99.14

0.30

0.30

0.32

2.8

subsurface

subsurface

The Lorentzian width (LW) was 0.045 eV for all components. Note that the band bending variation observed after styrene exposure is only 0.03 eV.

0.27 eV). Therefore, the Si bonded carbons cannot be distinguished anymore from the aromatic carbons. The Si 2p spectra of the water-covered surface (measured in surface sensitive conditions at hν = 150 eV), before and after exposure to 6.7 L of styrene, are given in Figure 5 parts a and b, respectively. Fit parameters are collected in Table 3. In Figure 5a we note the absence of the clean surface “up dimer” state6,20,29 that has a surface core-level shift (SCLS) of 0.57 eV from the Si 2p3/2 bulk component B (see Figure S2a of the Supporting Information). This is simply due to the capping of the silicon dangling bonds of the silicon dimers by H and OH fragments (except naturally the IDBs that are below the detection limit). A small peak S1 (2.8% of the spectral weight) with a SCLS of 0.30 eV from bulk Si 2p3/2 is still present. This could be a continuation to the peak C of the clean surface, attributed to subsurface Si (see ref 6 and references therein). The formation of the hydroxyl species shows up as a component denoted as Si1+, with a SCLS of +0.96 eV, with a spectral weight of 9.5% (the contribution of half a surface plane). The S2 component (at +0.26 eV, 13.9% of the spectral weight) encompasses both the Si H (monohydride) species (shifted by +0.26 eV from B)30 and the second silicon plane component of the clean surface. We note that all elemental Si fitting components (S1, B, S2) have a GW of 0.30 eV (fwhm of 0.32 eV), while the GW (fwhm) of the Si1+ state is somewhat larger, 0.41 eV (0.43 eV). The binding energy position of the bulk peak B is found at 99.41 eV, also in accord with our previous measurements.6 The Si 2p spectrum of the water-covered surface exposed to 6.7 L of styrene, shown in Figure 5b, is practically identical to Figure 5a, apart from a small rigid shift of 0.03 eV toward higher binding energy (the photon energy was not changed between these two measurements, to avoid problems). We do not see the appearance of a distinct line related to the formation of a Si C bond. This is not surprising as the Si C component has a SCLS of +0.26 eV, as shown by the Si 2p spectrum of the clean surface directly exposed to styrene (see Figure S2b of the Supporting Information), and is thus contained in component S2, together with the “second plane” and Si H components. The Si1+ (SiOH) peak weight does not vary appreciably (it remains at ∼10% of the spectral weight), which is consistent with the C 1s spectrum indicating that surface hydroxyls are not abstracted. The main information provided by the Si 2p spectra concerns the surface band bending, directly related to the areal density of surface acceptor states in the case of the water-covered n+-doped Si(001)-21 surface.6 The bands are bent upward by 0.45 eV in regards of the clean surface. The above-mentioned rigid shift in

binding energy indicates that the band bending slightly diminishes (by 0.03 eV) after styrene absorption. This means that the overall negative surface charge, and hence the areal density of negatively charged IDB, is only slightly decreased (by ∼5%) after exposure to 6.7 L styrene. C 1s XPS shows that only surface H atoms (not hydroxyls) bond to Cβ. If all H atoms were consumed by the hydrosilylation reaction, then a saturation coverage of one molecule per dimer should result (0.5 molecule per surface Si atom). In fact, as noticed before, the styrene growth at 6.7 L is limited to about 0.18 molecule per dimer (0.09 molecule per surface Si atom). If one assumes that the hydrosilylation of styrene occurs via the radical-chain reaction, then about ∼5.3 molecules are generated by one IDB, whose surface density is ∼0.017 defects per silicon atom6 (after exposure to ∼1 L of water). As discussed in ref 6, the extension of molecular grafting along a dimer row should be limited by the presence of unfavorable H/OH patterns: HO Si Si OH units, amounting to 20% of the water-reacted dimers should block the growth. In fact due to its mobility on the surface (ref 6 and references therein), the IDB can jump to favorable patterns (e.g., aligned H Si Si OH patterns) to start the growth anew. The other cause of adsorption rate decrease is the quenching of dangling bonds, resulting from “intra-row” or the “ondimer” pairing, followed by di-σ adsorption of styrene as depicted in Figure 3c (“on-dimer” case). On the other hand Si 2p XPS shows that the IDB concentration at the surface after an exposure to 6.7 L of styrene exposure is little affected. Therefore, such events are rare on a time scale of ∼103 s. Further complementary experiments should be carried out, examining separately the influence of pressure and exposure time, as the diffusion of dangling bonds on the surface may play a role in the growth process. Finally, the charge state of dangling bond deserves a comment. Figure 1 assumes that the IDB is neutral. However, in the present case, that is, a highly doped (n+) substrate, the STM images6 show that the IDB is negatively charged. Then in a charge-driven reaction scheme, the (doubly occupied) IDB should repel the electron-rich π functionality of styrene. This may have an impact on the barrier energy leading to the radical intermediate (for undoped crystals, calculations show that the formation of the styrene radical intermediate on the defective H-terminated Si(001)-21 is almost barrierless, see ref 11). The impact on its adsorption energy may be less, due to a delocalization of the charge in the phenyl ring and in the silicon substrate. In fact styrene does react on n+ doped H-terminated Si(001)-21.10 Other examples of charge transfer are known on silicon surfaces: the electron-rich ammonia lone-pair attacks preferentially the 14832

dx.doi.org/10.1021/jp202913y |J. Phys. Chem. C 2011, 115, 14827–14833

The Journal of Physical Chemistry C doubly occupied restatoms of the Si(111)-77 surface, while the adatoms are electron-deficient.31 Nevertheless comparing the reactivity of n- and p-doped water-covered Si(001)-21 surfaces (for which the IDB is likely neutral32) with that of the n+-doped water-covered surface would be useful to complete the picture.

4. CONCLUSION AND PERSPECTIVES Given the similarities between the defective H-terminated Si(001)-21 surface and the water-covered Si(001)-21 surface (presence of SiH and of isolated silicon dangling bonds), we have examined if an alkene can be grafted on the latter surface, via a hydrosilylation reaction, as documented for the former surface. Choosing styrene as a test molecule and core-level photoemission as a spectroscopic tool, we have observed that the molecule does react with the water-covered (n+-doped) Si(001)-21 surface at room temperature. The C 1s spectrum of the resulting product is significantly different from that of the adduct di-σ bonded to a silicon dimer of the clean surface. In particular, it exhibits a clear, low binding energy component, characteristic of a mono-σ Si C bond. The C 1s spectral shape also excludes the formation of a monohydroxyl carbon (due, e.g., to OH abstraction by the radical adduct in a radical-based reaction) or of a Si O C bond (direct Markovnikov or anti-Markovnikov addition to a surface silanol). Due to the intensity of the Si C component in the C 1s spectrum, the oligomerization of the molecule via a chain reaction is also excluded. The hydrosilylation reaction that leads to the formation of the Si CRH2 CβH2 C6H5 product is more likely related to a radical-based mechanism (Chidsey mechanism) than to a direct concerted process, as the calculated activation barrier of the former is much lower than that of the latter. We recall that isolated dangling bonds (on which the reaction chain initiates) are naturally present on the water-covered surface, with an areal density of ∼1.7  10 2 defects/Si atom. After an exposure of 6.7 L to styrene, the molecular areal density is ∼0.18 molecule per Si dimer. Therefore, assuming that the radical-based mechanism is valid, one dangling bond could generate the bonding of ∼5.3 molecules on the average. In addition, band bending measurements show that the (negative) surface charge density only slightly decreases (by ∼5%) after exposure to 6.7 L of styrene, showing that dangling bond quenching mechanisms are not efficient. ’ ASSOCIATED CONTENT

bS

Supporting Information. Supplementary XPS spectra and curve fit parameters useful for the interpretation of the present data (C 1s shake up spectrum and Si 2p spectrum of the clean surface exposed to 6.7 L of styrene at room temperature). This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Tel.: 33 144276621; fax: 33 144276226.

ARTICLE

’ REFERENCES (1) Ihm, K.; Kang, T.-H.; Moon, S.; Hwang, C. C.; Kim, K.-J.; Hwang, H.-N.; Jeon, C.-H.; Kim, H.-D.; Kim, B.; Park, C.-Y. J. Electron Spectrosc. Relat. Phenom. 2005, 144 147, 397–400. (2) Fan, C.; Lopinski, G. P. Surf. Sci. 2010, 604, 996–1001. (3) Henderson, M. Surf. Sci. Rep. 2002, 46, 1. (4) Lee, J.-Y.; Cho, J.-H. J. Phys. Chem. B 2006, 110, 18455–18458. (5) Warschkow, O.; Schofield, S. R.; Marks, N. A.; Radny, M. W.; Smith, P. V.; McKenzie, D. R. Phys. Rev. B 2008, 77, 201305. (6) Gallet, J.-J.; Bournel, F.; Rochet, F.; K€ ohler, U.; Kubsky, S.; Silly, M. G.; Sirotti, F.; Pierucci, D. J. Phys. Chem. C 2011, 115, 7686–7693. (7) Hossain, M. Z.; Yamashita, Y.; Mukai, K.; Yoshinobu, J. Phys. Rev. B 2003, 67, 153307. (8) Linford, M. R.; Fenter, P.; Eisenberger, P. M.; Chidsey, C. E. D. J. Am. Chem. Soc. 1995, 117, 3145. (9) Cicero, R. L.; Linford, M. R.; Chidsey, C. E. D. Langmuir 2000, 16, 5688–5695. (10) Lopinski, G. P.; Wayner, D. D. M.; Wolkow, R. A. Nature 2000, 406, 48–51. (11) Takeuchi, N.; Kanai, Y.; Selloni, A. J. Am. Chem. Soc. 2004, 126, 15890–15896. (12) Coletti, C.; Marrone, A.; Giorgi, G.; Sgamelloti, A.; Cerofolini, G.; Re, N. Langmuir 2006, 22, 9949–9956. (13) Pitters, J. L.; Dogel, I.; DiLabio, G. A.; Wolkow, R. A. J. Phys. Chem. B 2006, 110, 2159–2163. (14) Kondo, M.; Mates, T. E.; Fischer, D. A.; Wudl, F.; Kramer, E. J. Langmuir 2010, 26, 17000–17012. (15) Mischki, T. K.; Donkers, R. L.; Eves, B. J.; Lopinsky, G. P.; Wayner, D. D. M. Langmuir 2006, 22, 8359–8365. (16) http://www.maxlab.lu.se/beamline/max-ii/i511/i511.html (accessed Jul 5, 2011). (17) Prince, K. C.; Vondracek, M.; Karvonen, J.; Coreno, M.; Camilloni, R.; Avaldi, L.; de Simone, M. J. Electron Spectrosc. Relat. Phenom. 1999, 101 103, 141. (18) Bozek, J. D.; Bancroft, G. M.; Cutler, J. N.; Tan, K. H. Phys. Rev. Lett. 1990, 65, 2757. (19) Olivero, J. J.; Longbothum, R. L. J. Quant. Spectrosc. Radiat. Transfer 1977, 17, 233. (20) Landemark, E.; Karlsson, C. J.; Chao, Y.-C.; Uhrberg, R. I. G. Phys. Rev. Lett. 1992, 69, 1588–1591. (21) Himpsel, F. J.; Hollinger, G.; Pollak, R. A. Phys. Rev. B 1983, 28, 7014–7018. (22) Sze, S. M. Physics of Semiconductor Devices; Wiley Interscience: New York, 1981. (23) Li, Q.; Leung, K. T. J. Phys. Chem. B 2005, 109, 1420–1429. (24) Beamson, G.; Clark, D. T.; Kendrick, J.; Briggs, D. J. Electron Spectrosc. Relat. Phenom. 1991, 57, 79–90. (25) Kolczewski, C.; P€uttner, R.; Martins, M.; Schlachter, A. S.; Snell, G.; Sant'Anna, M. M.; Hermann, K. J. Chem. Phys. 2006, 124, 034302. (26) Cao, X.; Hamers, R. J. J. Am. Chem. Soc. 2001, 123, 10988. (27) Jolly, W. L.; Bomben, K. D.; Eyermann, C. J. Atomic Data Nuclear Data Tables 1984, 31, 433–493. (28) Coulter, S. K.; Schwartz, M. P.; Hamers, R. J. J. Phys. Chem. B 2001, 105, 3079–3087. (29) Koh, H.; Kim, J. W.; Choi, W. H.; Yeom, H. W. Phys. Rev. B 2003, 67, 073306. (30) Uhrberg, R. I. G.; Landemark, E.; Chao, Y. -C. J. Electron Spectrosc. Relat. Phenom. 1995, 75, 197–207. (31) Lo, R.-L.; et al. Phys. Rev B 2007, 76, 113305. (32) Liu, L.; Yu, X.; Lyding J. W. Mater. Res. Soc. Symp. Proc. 2002, 705, Y6.6.1.

’ ACKNOWLEDGMENT The research leading to these results has received funding from the European Community's Seventh Framework Program (FP7/2007-2013) under grant agreement No. 226716. 14833

dx.doi.org/10.1021/jp202913y |J. Phys. Chem. C 2011, 115, 14827–14833