Article pubs.acs.org/JPCC
Single versus Dual Attachment in the Adsorption of Diisocyanates at the Ge(100)-2 × 1 Surface Keith T. Wong, Sonali N. Chopra, and Stacey F. Bent* Department of Chemical Engineering, Stanford University, 381 North-South Mall, Stanford, California 94305, United States S Supporting Information *
ABSTRACT: Adsorption of bifunctional molecules on semiconductor surfaces must be understood in order to design systems that expose a free functional group on the surface for applications such as film growth by molecular layer deposition. In this study, we use Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, and density functional theory calculations to investigate the adsorption of three diisocyanate molecules, 1,4-diisocyanatobutane (1,4-DICB), 1,3-phenylene diisocyanate (1,3-PDIC), and 1,4-phenylene diisocyanate (1,4-PDIC), on the Ge(100)-2 × 1 surface. All three molecules undergo preferential [2 + 2] cycloaddition across the CN bond of the isocyanate, but they differ in the fraction that reacts with the surface through one or both isocyanate groups. All three adsorbates also form measurable quantities of side products involving less favorable [2 + 2] cycloaddition across the CO bond of the isocyanate. Our study shows that changing the backbone of a bifunctional molecule can have a large effect on the degree of dual attachment and that designing bifunctional molecules that react with the surface with high selectivity remains a challenge.
I. INTRODUCTION Direct attachment of organic molecules to semiconductor surfaces is of interest for the ability to tune the properties of semiconductor surfaces and inorganic−organic interfaces. Applications of organic functionalization are varied and include organic electronics, chemical and biological sensors, nanoscale patterning, and molecular electronics.1−15 Coupling the established knowledge of the inorganic semiconductor industry with the tailorability of organic molecules will open the door to novel devices and applications. Studies of organic functionalization of semiconductors have focused largely on the (100)-2 × 1 reconstructed surfaces of the group IV elements silicon and germanium.2,16−23 The 2 × 1 reconstruction creates ordered rows of dimersthe fundamental reactive unit of these surfaceswhose properties allow for interesting and useful analogy to organic chemistry.18,20 Uneven distribution of charge within the dimers leads to tilting of the dimer with respect to the plane of the surface. The tilted-up atom is electron rich, while the tilted-down atom is electron deficient, lending some zwitterionic or Lewis acid/base character to the dimers. In addition, a strong σ bond and partial π bond connecting the dimer atoms impart olefinic character. Due to these properties, organic molecules commonly form dative bonds or undergo pericyclic-like cycloaddition reactions with the dimers of the (100)-2 × 1 surfaces of silicon and germanium.2,16,18−25 In the present work, we compare reaction at the Ge(100)2 × 1 surface of three diisocyanate molecules: 1,4-phenylene diisocyanate (1,4-PDIC), 1,3-phenylene diisocyanate (1,3© 2012 American Chemical Society
PDIC), and 1,4-diisocyanatobutane (1,4-DICB). Each contains two isocyanate moieties separated by either a stiff aryl spacer or a flexible alkyl spacer. The germanium surface has seen growing interest in recent years, as there is opportunity to take advantage of some of its favorable properties, such as higher electron and hole carrier mobilities.6,22,26 Diisocyanates and related diisothiocyanates have been successfully implemented in conjunction with diamines for growing organic thin films by molecular layer deposition (MLD).15,27−30 Reaction of bifunctional molecules at the surface is a particularly important topic for MLD since the majority of demonstrated MLD schemes use alternating doses of two homobifunctional precursors to grow a film in a layer-by-layer fashion.31 Ideally, reaction occurs through only one functional group in each step, thus leaving a free functional group to react in the next step. However, use of homobifunctional precursors inherently opens up the possibility of precursors reacting with the surface through both functional groupsa process that leads to adsorption but does not sustain film growth. Thus, understanding how to control single versus dual attachment of bifunctional molecules with the surface is a key step in understanding MLD growth and designing new MLD systems. In this study we show that changing between an aryl and an alkyl spacer or changing positions on an aryl spacer significantly affects the degree of dual attachment of diisocyanate molecules. Received: March 27, 2012 Published: May 31, 2012 12670
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
Figure 1. Potential reactions between an isocyanate functional group and a Ge(100)-2 × 1 surface dimer.
electron diffraction (LEED; Physical Electronics) verified the 2 × 1 reconstruction. FTIR spectra were collected by a BioRad FTS-60A spectrometer in a multiple internal reflection (MIR) geometry with a liquid nitrogen cooled mercury−cadmium− telluride (MCT) detector. The unpolarized IR beam was coupled into and out of the UHV chamber through differentially pumped CaF2 viewports. The beam path outside of the UHV chamber was sealed and purged with air treated by a purge gas generator (Parker Filtration) to eliminate spectral features from H2O and CO2. The spectral range of the IR data is limited by absorption from the CaF2 windows below ∼1050 cm−1. Trapezoidal Ge crystals (19 mm × 14 mm × 1 mm, 45° bevels, Harrick Scientific) were used as the substrate. IR spectra were corrected for baseline fluctuation by manually subtracting spline functions fit to points devoid of spectral features. Physisorbed multilayer spectra were collected by holding the substrate at low temperature to enable condensation of dosed precursor onto the surface; all other spectra were collected at room temperature. Absorbance in multilayer spectra has been scaled to facilitate comparison with chemisorption spectra. X-ray photoelectron spectroscopy (XPS) studies were performed in a separate UHV chamber also described previously.40 The base pressure of this chamber was less than 2 × 10−10 Torr. The Ge substrate (approximately 8 mm × 8 mm × 1 mm, MTI Corp.) was cleaned by a single cycle of argon ion sputtering (1 keV accelerating voltage, 20 mA emission current, 18−20 μA sample current) for 30 min followed by annealing to approximately 800 K. Temperature was monitored by a type K thermocouple attached to the sample holder but not in direct contact with the crystal due to the chamber design. 2 × 1 reconstruction was again verified by LEED (Princeton Research Instruments), and XPS showed that levels of carbon, nitrogen, and oxygen were below the detection limit of our system following the cleaning procedure. C(1s) photoelectron spectra were collected using Al Kα radiation, while O(1s) and N(1s) spectra were collected using Mg Kα radiation (both operating at 300 W; 20 mA emission current, 15 kV anode voltage) due to interference from the Ge Auger series. All spectra were collected with a 25 eV pass energy using a five-channel hemispherical analyzer (SPECS Surface Nano Analysis GmbH). Before fitting photoelectron spectra, background curvature was removed by subtracting a polynomial fit to the spectrum of the clean Ge(100)-2 × 1 surface (for chemisorbed spectra) or subtracting the spectrum of a lower
The cumulated double bond of the isocyanate functional group containing two heteroatoms can potentially undergo a variety of reactions with the Ge(100)-2 × 1 surface. These reactions are summarized in Figure 1. Lone pair electrons on the heteroatoms can be donated to the electron-deficient down dimer atom to form N or O dative-bonded products. Cycloaddition reactions can occur either by a 1,3-dipolar mechanism to form a five-membered ring at the surface or by a [2 + 2] reaction across either the CN or the CO double bonds of the isocyanate. Previous studies of adsorption of isocyanates on Ge(100)-2 × 1 have shown that the CN [2 + 2] cycloaddition product is initially dominant, although in the case of tert-butyl isocyanate there was also a significant amount of dissociated product.27,32,33 The propensity for isocyanates to undergo [2 + 2] cycloaddition forming two directional bonds to the surface per functional group sets them apart from other bifunctional molecules that have been studied,34−38 since the [2 + 2] reaction transfers the surface’s translational and orientational order to the adsorbates before even considering dual attachment. We show that for the three diisocyanates in this study CN [2 + 2] cycloaddition is again the theoretically most favorable and experimentally dominant reaction; however, there are measurable quantities of minor products involving CO [2 + 2] cycloaddition. The occurrence of a less favorable reaction is likely enabled by the adsorbate first being tethered to the surface by CN [2 + 2] reaction. Once adsorbed on the surface, steric hindrance may make CO [2 + 2] cycloaddition of the adsorbate’s second functionality similarly favorable to the CN [2 + 2] reaction. Thus, we find that the nature of the spacer between two functional groups may affect both the extent of dual reaction and the type of reaction that occurs.
II. EXPERIMENTAL SECTION Experiments were performed in two separate ultra-high vacuum (UHV) chambers. Fourier transform infrared (FTIR) spectroscopy studies were performed in a chamber described previously.39 The base pressure of this chamber was less than 2 × 10−10 Torr. Clean, reconstructed Ge(100)-2 × 1 was prepared by two cycles of argon ion sputtering (1 keV accelerating voltage, 20 mA emission current, 9−10 μA sample current) for 20 min followed by annealing to 880 K for 5 min. Temperature was monitored by a type K thermocouple attached directly to the surface of the crystal. Low-energy 12671
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
dose of the precursor (for multilayer spectra). The background subtraction procedure for multilayer spectra removed any contribution from chemisorbed molecules in the first layer at the surface. Spectra were then fit with a Shirley baseline and a chemically realistic number of pure Gaussian components constrained to the same fwhm within a given spectrum. The Ge(3d) photoelectron peak was used as an internal standard for calibrating the energy scale and peak intensity, and relative sensitivity factors determined for our system were used in quantitative comparisons. Intensities in all photoelectron spectra shown have been scaled by the transmission function of our analyzer. 1,4-PDIC (Sigma-Aldrich) is a white solid, 1,3-PDIC (≥95%, Sigma-Aldrich) is a translucent white solid, and 1,4-DICB (≥97%, Acros Organics) is a clear to light-brown liquid at room temperature. Precursors were transferred to sample vials under dry air. 1,4-DICB underwent several freeze−pump−thaw cycles before being dosed into the UHV chamber through a variable leak valve. The solid precursors underwent several sublimation and pumping cycles before being dosed into the UHV chamber. 1,3-PDIC was heated to 40−50 °C and dosed into the chamber through a variable leak valve; the lower vapor pressure of 1,4PDIC required it to be introduced via a directed doser separated from the chamber by a gate valve. Identities of the dosed compounds were verified by an in situ quadrupole mass spectrometer (Vacuum Generators). Exposures are reported in units of Langmuir (L; 1 L = 10−6 Torr s). Quantum chemical calculations were carried out using the Gaussian 03 software suite41 to determine both the energetics of reaction pathways and theoretical IR spectra of potential surface species. We employed the Becke3 Lee−Yang−Parr (B3LYP) three-parameter density functional theory (DFT) method that has been previously shown to provide predictive results for similar systems.23,39,42,43 The Ge(100)-2 × 1 surface was modeled by Ge15H16 and Ge23H24 clusters representing two dimers within a row and two dimers across a trench, respectively. Ge dimer atoms and all adsorbate atoms were modeled using the triple-ζ 6-311++G(d,p) basis set, and subsurface Ge atoms were modeled using the LANL2DZ pseudopotential in order to reduce the computational cost of calculations. Subsurface Ge atoms were terminated with hydrogens, modeled using the 6-31G(d) basis set, in order to fill their valence and approximate neighboring Ge atoms. For optimization calculations, all terminating hydrogens and all cluster Ge atoms except for the top two planes of atoms (including the dimer atoms) were frozen to prevent nonphysical distortion. Local minima and transition states were verified by frequency calculations at the optimized geometries using the same basis sets. All reported energies are zero-point energy corrected. Terminating hydrogens were assigned a mass of 74 amu in frequency calculations to eliminate artificial Ge−H vibrations in the calculated IR spectra, and all calculated frequencies have been scaled by a factor of 0.96.44 Lorentzian lineshapes (4 cm−1 fwhm) with the calculated intensities have been used to represent IR bands in calculated spectra.
Figure 2. Calculated reaction coordinate diagram for reaction of one isocyanate functional group of 1,3-PDIC with the Ge(100)-2 × 1 surface represented by a Ge15H16 two-dimer row cluster. N dative, O dative, 1,3-dipolar cycloaddition, CO [2 + 2] cycloaddition, and CN [2 + 2] cycloaddition products are shown along with transition states connecting these products. All energies are in kcal/mol.
previously been reported for monofunctional isocyanates.32,33 The CN [2 + 2] cycloaddition product is the most thermodynamically and kinetically favored and, thus, should dominate the product distribution, neglecting dual reaction for the time being. The CN [2 + 2] cycloaddition product is formed by passing through an N dative-bonded precursor state (−3.0 kcal/mol) in which a lone pair of electrons on the isocyanate nitrogen are donated to the electron-deficient tilteddown dimer atom. Following N dative bonding, a modest activation barrier (1.9 kcal/mol above the reactants; 4.9 kcal/ mol above the precursor state) in which the cumulated double bond of the isocyanate bends to allow the carbon atom to bond with the second dimer atom must be crossed to reach the C N [2 + 2] cycloaddition product. 1,3-Dipolar cycloaddition is not expected to occur as its formation is endothermic by 4.5 kcal/mol, and the transition state leading to 1,3-dipolar cycloaddition lies well above the energy of the reactants (12.0 kcal/mol). The two dative-bonded products may serve as precursors to further reaction, as described for CN [2 + 2] cycloaddition, but are too weakly bound (−3.0 and −4.4 kcal/ mol for N and O dative products, respectively) to be observed on the experimental time scale at room temperature. Similarly, the CO [2 + 2] cycloaddition product has a low adsorption energy (−2.1 kcal/mol), which is even weaker than its O dative precursor, and is not predicted to be observed at room temperature. Due to the similarity between the reaction pathways for 1,3-PDIC, phenyl isocyanate,33 and tert-butyl isocyanate,32 we expect that these results can be extended to 1,4-PDIC and 1,4-DICB as well. 1,4-PDIC adsorption on Ge(100)-2 × 1 has been briefly studied previously using FTIR spectroscopy and was found to primarily undergo CN [2 + 2] cycloaddition.27 Thus, when considering reaction of only one functional group, all three of the diisocyanates in this study are expected to undergo [2 + 2] cycloaddition across the CN bond of the isocyanate. Figure 3 compares the reaction pathways leading to dual C N [2 + 2] cycloaddition of 1,4-PDIC, 1,3-PDIC, and 1,4-DICB on Ge(100)-2 × 1 calculated using DFT. All three pathways first pass through a single N dative bond, a transition state, and a CN [2 + 2] cycloaddition productthe expected pathway for reaction of a single functional group based on the above
III. RESULTS AND DISCUSSION A. DFT Calculations. Before considering reaction of both isocyanate moieties, we first investigate the reaction pathways for a single isocyanate functional group leading to the expected products in Figure 1. DFT-calculated reaction pathways are shown for 1,3-PDIC/Ge(100)-2 × 1 in Figure 2. The energetics of 1,3-PDIC reaction are very similar to what has 12672
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
Figure 3. Calculated reaction pathways for dual CN [2 + 2] cycloaddition of 1,3-PDIC (blue), 1,4-PDIC (red), and 1,4-DICB (green) on the Ge(100)-2 × 1 surface represented by a Ge23H24 two-dimer trench cluster. All energies are in kcal/mol.
Figure 4. FTIR spectra of (a) physisorbed multilayers of 1,4-DICB on Ge(100)-2 × 1 at low temperature, (b) Ge(100)-2 × 1 saturated with chemisorbed 1,4-DICB at room temperature, (c) the calculated 1,4-DICB single CN [2 + 2] product, and (d) the calculated 1,4-DICB dual CN [2 + 2] product. Calculations were performed using a Ge23H24 two-dimer trench cluster.
analysis for 1,3-PDIC. The slight differences in calculated energy for 1,3-PDIC between those shown in Figures 2 and 3 are attributed to the use of different clusters. While the Ge15H16 row cluster was adequate for calculations involving reaction of one functional group, it was found that reaction across a trench is the most likely geometry when considering reaction of both functional groups; therefore, a Ge23H24 trench cluster was employed for the calculations shown in Figure 3. Interestingly, the N dative bond of 1,4-DICB (−7.1 kcal/mol) is calculated to be several kcal/mol stronger than those of 1,3-PDIC (−1.8 kcal/mol) or 1,4-PDIC (−2.8 kcal/mol). This may be attributed to the butyl group being slightly more electron donating than the phenylene group. However, the differences between the three molecules are negligible after passing through the first transition state (TS1) to form the CN [2 + 2] cycloaddition productthe calculated adsorption energies are all within 0.4 kcal/mol of each other.
Following CN [2 + 2] cycloaddition of one isocyanate, Figure 3 shows the pathways for subsequent CN [2 + 2] cycloaddition of the second isocyanate on the neighboring Ge dimer across the trench. 1,3-PDIC and 1,4-DICB follow N dative bond-mediated pathways similar to reaction of the first isocyanate. For 1,3-PDIC, N dative bonding of the second functional group is exothermic by only 0.5 kcal/molless than one-third the exothermicity of formation of the first dative bond. This difference can be attributed to the strain introduced by first tethering one functional group by CN [2 + 2] cycloaddition. The difference is less for 1,4-DICB, which connects two isocyanate functional groups by a more flexible four-carbon alkyl chain. For 1,4-DICB, N dative bonding of the second functional group is exothermic by 3.0 kcal/mol. While still less than the 7.1 kcal/mol adsorption energy for the first dative bond, the second dative bond is between one-third and one-half as favorable as the first. More notable, however, is that no CN [2 + 2]/N dative state could be found for 1,4-PDIC 12673
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
Figure 5. FTIR spectra of (a) physisorbed multilayers of 1,3-PDIC on Ge(100)-2 × 1 at low temperature, (b) Ge(100)-2 × 1 saturated with chemisorbed 1,3-PDIC at room temperature, (c) the calculated 1,3-PDIC single CN [2 + 2] product, and (d) the calculated 1,3-PDIC dual CN [2 + 2] product. Calculations were performed using a Ge23H24 two-dimer trench cluster.
due to placement of the isocyanate functional groups on opposite sides of the inflexible phenyl ring. Our calculations were successful for 1,4-PDIC attached to the Ge surface by CN [2 + 2] cycloaddition of both isocyanates; however, this state was found to be 3.0 kcal/mol less stable than single CN [2 + 2] cycloaddition of 1,4-PDIC. Thus, even if there is no barrier to cycloaddition of the second isocyanate of 1,4PDIC, the dual CN [2 + 2] state is unlikely to be observed since it is less thermodynamically stable than the preceding single CN [2 + 2] state. For 1,3-PDIC and 1,4-DICB, 4.4 and 7.9 kcal/mol activation barriers, respectively, from the CN [2 + 2]/N dative state lead to the dual CN [2 + 2] product. Such activation barriers may be surmounted at room temperature. Similar to the CN [2 + 2]/N dative state, we observe a small difference in exothermicity of the dual CN [2 + 2] products. The calculations indicate that the dual CN [2 + 2] product of 1,4DICB is 2.3 kcal/mol lower in energy than that of 1,3-PDIC. An alternative way to analyze the data is to consider the ratio of dually to singly tethered adsorption energies. For 1,3-PDIC, dual cycloaddition is 1.69 times as favorable as single cycloaddition, while for 1,4-DICB, it is 1.82 times as favorable. In the ideal case of entirely noninteracting functional groups, this ratio is expected to be 2. 1,4-DICB is closer to achieving the ideal ratio, which we again attribute to the flexibility of the alkyl spacer allowing the second isocyanate to bond to the surface without imparting significant strain on the alkyl chain or surface attachments. We conclude from the DFT calculations that all three molecules will react with the Ge(100)-2 × 1 surface first by undergoing [2 + 2] cycloaddition across the CN bond of one isocyanate group. 1,4-PDIC is not expected to react further since proceeding to dual CN [2 + 2] cycloaddition is endothermic. 1,3-PDIC and 1,4-DICB may continue to react, forming products in which both isocyanates have undergone CN [2 + 2] cycloaddition. Due to surface crowding we expect that both of these molecules will form a mixture of singly and dually tethered products, but 1,4-DICB may form a higher percentage of dually tethered product since dual CN [2 + 2]
cycloaddition is slightly more thermodynamically favorable than for 1,3-PDIC. B. FTIR Spectroscopy. FTIR spectra of 1,4-DICB physisorbed multilayers and 1,4-DICB chemisorbed at room temperature on the Ge(100)-2 × 1 surface are shown in Figure 4 along with calculated IR spectra for the single and dual CN [2 + 2] cycloaddition products. The multilayer spectrum contains the expected features for this molecule: a very strong asymmetric stretching peak associated with the isocyanates at 2264 cm−1, aliphatic C−H stretching modes near 2890−2970 cm−1, a sharp aliphatic C−H bending mode at 1354 cm−1, and other smaller features associated primarily with combinations of aliphatic C−C stretching and C−H bending in the 1100−1500 cm−1 region. Peaks in the multilayer spectra of 1,4-DICB and the other two molecules discussed below were assigned by comparison to DFT-calculated spectra of the unreacted precursors (not shown) and comparison to reference spectra where available. The differences between the spectrum of physisorbed multilayers at low temperature and the chemisorbed spectrum at room temperature give insight into how 1,4-DICB reacts with the Ge(100)-2 × 1 surface. The modes from the multilayer spectrum remain in the chemisorbed spectrum, and in addition, a broad peak around 1653−1672 cm−1 is observed in the chemisorbed spectrum of 1,4-DICB. This is in the region where CO stretching is expected to absorb. CN [2 + 2] cycloaddition leads to formation of a carbonyl, and the calculated positions of CO stretching for single (1670 cm−1) and dual (1657 and 1670 cm−1) CN [2 + 2] cycloaddition products match very well with the experimental chemisorbed spectrum. The broad peak seen experimentally matches particularly well with the calculated spectrum of the dually tethered product, which exhibits two separate CO stretching peaks for the two carbonyls separated by 23 cm−1, although singly tethered 1,4-DICB must also be present since there must be unreacted isocyanate groups to give rise to the NCO stretching peak at 2263 cm−1. There is a small peak in the spectrum of chemisorbed 1,4-DICB at approximately 1609 cm−1 that does not correspond to any peaks in the multilayer spectrum or calculated CN [2 + 2] 12674
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
Figure 6. FTIR spectra of (a) physisorbed multilayers of 1,4-PDIC on Ge(100)-2 × 1 at low temperature, (b) Ge(100)-2 × 1 saturated with chemisorbed 1,4-PDIC at room temperature, (c) the calculated 1,4-PDIC single CN [2 + 2] product, and (d) the calculated 1,4-PDIC dual CN [2 + 2] product. Calculations were performed using a Ge23H24 two-dimer trench cluster.
very good agreement with the calculated position of the carbonyl stretch (1658 cm−1) for the single CN [2 + 2] cycloaddition product of 1,4-PDIC on Ge(100)-2 × 1. DFT calculations show that the carbonyl stretching peak blue shifts significantly to 1704 cm−1 for the dual CN [2 + 2] product. However, the experimental chemisorbed spectrum of 1,4-PDIC does not exhibit significant absorbance near 1704 cm−1, suggesting that 1,4-PDIC does not form a significant amount of the dually tethered productthe same conclusion that was reached based on analysis of the calculated reaction pathways. The chemisorbed spectrum of 1,4-PDIC contains a small peak red shifted ca. 50 cm−1 from the carbonyl stretching peak at 1601 cm−1, similar to 1,4-DICB. This peak is once again assigned to a minor side product. On the basis of DFT calculations, a likely source of a peak near 1610−1630 cm−1 is stretching of the CN bond in an isocyanate that has undergone CO [2 + 2] cycloaddition. In fact, this is the primary identifying feature in the calculated spectra of CO [2 + 2] products (see Figure S1, Supporting Information). While no unassigned peak was noted in this region for 1,3PDIC chemisorbed on Ge(100)-2 × 1, the presence of a CN stretching peak cannot be ruled out since it may overlap with the peak at 1597 cm−1 that corresponds to C−C stretching and C−H bending in the aromatic ring. CO [2 + 2] cycloaddition was deemed unlikely based on DFT-calculated energies; however, the bifunctional nature of the molecules in this study may enable formation of a small amount of less kinetically or thermodynamically favored products. A diisocyanate molecule may first adsorb on the surface to form the C N [2 + 2] cycloaddition product with one isocyanate. Next, the second isocyanate may react with a neighboring dimer, and even if the additional stabilization provided by the second reaction is small, the molecule is already strongly tethered to the surface. DFT calculations on a Ge23H24 trench cluster show that 1,4-DICB, 1,3-PDIC, and 1,4-PDIC bonded to the surface by CN [2 + 2]/CO [2 + 2] cycloaddition are 1.1, 0.2, and 4.5 kcal/mol less stable, respectively, than when bonded to the surface by CN [2 + 2] cycloaddition alone. However, the differences for 1,4-DICB and 1,3-PDIC may fall within the error of the calculations. Also, it is possible that even small
spectra. This peak may result from a small amount of a side product and will be discussed later. Figure 5 shows FTIR spectra of condensed multilayers of 1,3-PDIC on Ge at low temperature, Ge(100)-2 × 1 saturated with 1,3-PDIC at room temperature, and calculated spectra for the single and dual CN [2 + 2] cycloaddition products of 1,3-PDIC on Ge(100)-2 × 1. The multilayer spectrum compares well with the gas-phase spectrum of 1,3-PDIC in the literature,45 indicating that 1,3-PDIC does not decompose while dosing. The multilayer spectrum contains a very intense peak associated with asymmetric stretching of the isocyanates at 2263 cm−1, similar to 1,4-DICB. There are also a number of peaks in the 1500−1650 cm−1 range, as labeled in Figure 5, associated with combinations of C−C stretching in the aromatic ring, C−H bending, and C−N stretching between the aromatic carbons and the isocyanate nitrogens. These features are once again preserved in the spectrum of chemisorbed 1,3-PDIC at room temperature, indicating that the molecule remains largely intact upon chemisorption. Similar to 1,4-DICB, a broad peak that is not present in the multilayer spectrum appears in the chemisorbed spectrum at 1663−1690 cm−1. We attribute this peak to CO stretching resulting from formation of a carbonyl upon [2 + 2] cycloaddition across the CN bond. Calculations indicate that the CO stretching peak appears at 1658 cm−1 for the single CN [2 + 2] product, but this peak blue from shifts 24 cm−1 to 1682 cm−1 for the dual CN [2 + 2] product. Thus, the broad peak seen experimentally again suggests formation of both single and dual CN [2 + 2] products. FTIR spectra of condensed multilayers of 1,4-PDIC at low temperature, Ge(100)-2 × 1 saturated with chemisorbed 1,4PDIC at room temperature, and calculated spectra of single and dual CN [2 + 2] products of 1,4-PDIC are shown in Figure 6. The multilayer spectrum resembles that of 1,3-PDIC and again contains all of the expected features for this molecule. These features are also present in the spectrum of chemisorbed 1,4-PDIC at room temperature, although the C−H stretching modes are too weak to be seen above the noise. Similar to 1,3PDIC and 1,4-DICB, a peak assigned to CO stretching appears in the chemisorbed spectrum at 1655 cm−1. This is in 12675
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
quantification of the amount of singly and dually tethered products, we turn to XPS studies. C. XPS. XP spectra of the Ge(100)-2 × 1 surface exposed to saturation doses of 1,4-DICB, 1,3-PDIC, and 1,4-PDIC are shown in Figure 8. As a reference, multilayer spectra of 1,3PDIC condensed at low temperature are also shown in Figure 8. Multilayer spectra of 1,4-PDIC and 1,4-DICB were also collected but are not shown as they contain essentially identical features as the 1,3-PDIC multilayer spectra (see Figure S2, Supporting Information). The multilayer spectra of 1,3-PDIC are fit to one peak in the O(1s) region at 533.8 eV, one peak in the N(1s) region at 400.5 eV, and three peaks in the C(1s) region at 289.1, 286.3, and 284.5 eV. The highest binding energy carbon (289.1 eV) is attributed to the carbon atoms in the isocyanate functional groups, which are bonded to the more electronegative nitrogen and oxygen. The remaining two C(1s) peaks (284.5 and 286.3 eV) correspond to the carbon atoms in the aromatic ring with the higher binding energy peak attributed to the carbons bonded directly to the isocyanate nitrogens. The overall N:O:C elemental ratio for the multilayer is 2:1.8:7.7 (nitrogen normalized to 2), which is close to the expected ratio of 2:2:8. It is apparent in Figure 8 that the C(1s) peaks in the 1,3-PDIC multilayers are shifted up approximately 0.4 eV in binding energy from the same peaks in the 1,3-PDIC chemisorbed spectrum. We attribute this shift to charging of the multilayer-covered surface, and we expect that the O(1s) and N(1s) multilayer peaks are similarly shifted. The C(1s) spectra of chemisorbed 1,4-DICB, 1,3-PDIC, and 1,4-PDIC each contain three peaks. These can be assigned to the various carbons in each molecule similar to the multilayer, but the O(1s) and N(1s) spectra are much more informative since each molecule contains only two oxygen and two nitrogen atoms. Two peaks are observed in the O(1s) and N(1s) spectra of each molecule when chemisorbed on the Ge(100)-2 × 1 surface. In the O(1s) region, one peak is observed at 532.6− 533.1 eV, which is similar to the binding energy of the multilayer O(1s) peak, after accounting for charging, and can be attributed to unreacted isocyanate oxygen. A second peak is observed 2.1−2.4 eV lower than the first. This lower binding energy peak falls in a range previously observed for carbonyl species;35,40 therefore, we assign it to oxygen in isocyanate functional groups that have undergone CN [2 + 2] cycloaddition resulting in formation of a carbonyl. Similarly, for the N(1s) spectra a higher binding energy peak at 399.8− 400.0 eV matches the N(1s) binding energy of the multilayer and corresponds to nitrogen in unreacted isocyanates. The second N(1s) peak at 398.0−398.3 eV can be attributed to nitrogen in isocyanates that have undergone CN [2 + 2] cycloaddition. Nitrogen covalently bonded directly to the Ge surface, as in the CN [2 + 2] product, is expected to have lower binding energy due to the smaller electronegativity of Ge. FTIR results suggest formation of a small percentage of CO [2 + 2] product. For this product we expect that the N(1s) binding energy would be similar to that of nitrogen in an unreacted isocyanate since the chemical environment of the nitrogen changes little upon CO [2 + 2] reaction. The O(1s) binding energy is expected to shift down to approximately 530.6−531.0 eV34,37,46 due to formation of a covalent bond with germanium and likely overlaps with the O(1s) binding energy of oxygen in CN [2 + 2] products. We also note that dative bonding through either the nitrogen or the oxygen is expected to lead to a measurable shift to higher binding energy, compared to the peak positions in the multilayer, due to loss of
distortions of the cluster compared to the real geometry of the Ge surface may have a particularly large effect on the energetics when dealing with products that interact with multiple dimers. In summary, FTIR spectroscopy of 1,4-DICB, 1,3-PDIC, and 1,4-PDIC corroborates the DFT calculations, indicating that [2 + 2] cycloaddition across the CN bond of the isocyanate is most favorable, although there may also be a small percentage of CO [2 + 2] reaction. FTIR results for 1,4-DICB and 1,3PDICparticularly the CO stretching peakssuggest that these precursors form both single and dual CN [2 + 2] products, whereas the spectrum of chemisorbed 1,4-PDIC matches well with the calculated CO stretching frequency of the single CN [2 + 2] product but not the dual CN [2 + 2] product. Again, these results agree with the calculations, which indicate that dual CN [2 + 2] attachment is likely to occur for 1,4-DICB and 1,3-PDIC but is thermodynamically unfavorable for 1,4-PDIC. Further insight may be gained by directly comparing the intensity of the isocyanate stretching peak in the saturated chemisorbed spectra of the three molecules, as shown in Figure 7. Dual CN [2 + 2] reaction leaves no intact isocyanate functional group and therefore does not contribute to the absorbance in this region; only singly tethered molecules give rise to the isocyanate stretching peak near 2265 cm−1. In Figure 7 it is apparent that the isocyanate stretching peak is strongest for 1,4-PDIC followed by 1,3-PDIC and 1,4-DICB. This
Figure 7. Comparison of the isocyanate stretching peak at 2264−2268 cm−1 in the FTIR spectra of Ge(100)-2 × 1 saturated with (a) 1,4PDIC, (b) 1,3-PDIC, and (c) 1,4-DICB at room temperature.
suggests that 1,4-PDIC forms the highest percentage of singly tethered adsorbates, 1,4-DICB forms the highest percentage of dually tethered adsorbates, and 1,3-PDIC falls in between these two cases. However, the intensity of the isocyanate stretching peak is also affected by the IR absorption cross sections of the particular mode in the three adsorbed molecules and by the total surface coverage at saturation of the three molecules. Assuming that the absorption cross-section of the isocyanate asymmetric stretching mode does not differ significantly among these three molecules, numerical integration of the peaks in Figure 7 indicates that 1,4-DICB and 1,3-PDIC leave 22% and 46%, respectively, the number of free, unreacted isocyanates compared to 1,4-PDIC at the Ge surface. To account for differences in total surface coverage, thus allowing better 12676
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
Figure 8. High-resolution X-ray photoelectron spectra of the O(1s), N(1s), and C(1s) orbitals of (a) condensed multilayers of 1,3-PDIC at low temperature, and saturation exposures at room temperature of (b) 1,4-PDIC, (c) 1,3-PDIC, and (d) 1,4-DICB on Ge(100)-2 × 1.
electron density;34,35,37,38 therefore, our XPS data show no signs of dative-bonded products for any of the molecules. On the basis of these assignments of the peaks in the O(1s) and N(1s) photoelectron spectra, we can quantify the product distribution for each of the three adsorbates, shown in Table 1.
the real Ge surface, but these calculations indicate that dual CO [2 + 2] reaction may be possible for 1,4-PDIC. The presence of some CO [2 + 2] products for all three molecules is supported by our FTIR analysis, but CO [2 + 2] cycloaddition accounts for a minority of the product distribution in all cases. As shown in Table 1, single and dual CN [2 + 2] products account for 86−91% of all three product distributions based on the O(1s) and N(1s) spectra. Ignoring the minor products involving CO [2 + 2] cycloaddition, one-half of 1,4-DICB adsorbates are dually tethered, one-seventh of 1,3-PDIC products are dually tethered, and 1,4-PDIC does not form any dually tethered CN [2 + 2] products. These results corroborate FTIR results showing that a fraction of 1,4-DICB and 1,3-PDIC molecules undergo dual CN [2 + 2] cycloaddition at the Ge(100)-2 × 1 surface and the DFT calculations, indicating that dual CN [2 + 2] cycloaddition is only favorable for 1,4-DICB and 1,3-PDIC. The difference between the degree of dual tethering of 1,4-DICB and 1,3PDIC may be attributed to the flexibility of the alkyl chain of 1,4-DICB compared to the relative stiffness of the phenylene ring of 1,3-PDIC. Relatively free rotation of the alkyl C−C bonds between the two isocyanate groups is expected to minimize strain by allowing bond lengths and bond angles to remain close to their ideal values. Moreover, a flexible spacer may allow dual tethering in more than one geometry at the surface, thus enabling dual tethering at higher coverages when surface crowding may otherwise block reaction. 1,3-PDIC and 1,4-PDIC are restricted to reaction with a second dimer across a trench on the Ge surface since [2 + 2] cycloaddition of the first functional group creates two bonds with the surface, thus aligning the molecule with the dimer. 1,4-DICB, however, is sufficiently flexible that the second functional group may react with a neighboring dimer in the same row. In fact, our DFT calculations show that such intra-row dual CN [2 + 2] cycloaddition of 1,4-DICB, while not as favorable as inter-row reaction, is still thermodynamically downhill from the single CN [2 + 2] product. We can calculate the absolute surface coverage by comparison to a saturation dose of pyridine, which is known to saturate on Ge(100)-2 × 1 at 0.25 monolayer (ML) coverage (1 ML = 1 adsorbate molecule per surface Ge atom).47 On the
Table 1. Product Distribution, Surface Coverage, Fraction of Occupied Dimers, and Relative Number of Free Isocyanate Functional Groups for Ge(100)-2 × 1 Saturated with 1,4DICB, 1,3-PDIC, and 1,4-PDIC at Room Temperaturea
product distribution
CN [2 + 2]
CN [2 + 2]/CN [2 + 2] CN [2 + 2]/CO [2 + 2] CO [2 + 2]/CO [2 + 2] surface coverage (ML) fraction occupied dimers relative amount of unreacted isocyanates
1,4DICB
1,3PDIC
1,4PDIC
43%
78%
90%
43% 14%
13% 9%
0.19 0.58 0.28
0.19 0.48 0.54
10% 0.32 0.69 1.00
a
Product distributions were determined by analysis of O(1s) and N(1s) photoelectron spectra. Surface coverage is shown in ML, where 1 ML is defined as 1 molecule per surface Ge atom. The relative amount of unreacted isocyanates is normalized to 1 for 1,4-PDIC.
For 1,4-DICB and 1,3-PDIC we assume that only single CN [2 + 2], dual CN [2 + 2], and CN [2 + 2]/CO [2 + 2] products form. However, for 1,4-PDIC, these products cannot explain the larger peak at higher binding energy in the N(1s) spectrum; instead, the features of the O(1s) and N(1s) spectra of 1,4-PDIC are best explained by the presence of some dual CO [2 + 2] product. DFT calculations showed that CO [2 + 2] cycloaddition is less kinetically and thermodynamically favorable than CN [2 + 2] cycloaddition, but the dual CO [2 + 2] product may have sufficient adsorption energy to be observed in our experiments. In fact, we calculated adsorption energies ranging from −4.2 to −18.6 kcal/mol for 1,4-PDIC adsorbed by dual CO [2 + 2] cycloaddition. Such wide variations resulted from small changes in the geometry of the cluster, which are particularly important for cases such as this in which a molecule forms a total of four bonds to the surface at two separate dimers. DFT calculations cannot exactly replicate 12677
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
The presence of small but measurable quantities of CO [2 + 2] cycloaddition products for all three adsorbates suggests that achieving high selectivity for bifunctional molecules may present more of a challenge than for their monofunctional analogues. Besides adding the complexity of the possibility of both singly and dually tethered adsorbates, bifunctionality may open up entirely different reaction pathways that would not otherwise be observed for the monofunctional analogues. For a monofunctional molecule, selectivity is sometimes achieved especially on the Ge(100)-2 × 1 surface for which binding energies are often relatively lowvia thermodynamic stability, since adsorption must be sufficiently exothermic to prevent desorption on the experimental time scale. However, once a bifunctional molecule is tethered to the surface by a strong interaction of one functional group, the second functional group may readily react with a neighboring dimer even if the reaction is not highly exothermic. This is demonstrated in this study by the presence of ∼10% CN [2 + 2]/CO [2 + 2] adsorbates for 1,4-DICB and 1,3-PDIC. CO [2 + 2] reaction is expected to have insufficient adsorption energy by itself but may occur for adsorbates already tethered by CN [2 + 2] cycloaddition. Alternatively, we find that 1,4-PDIC forms approximately 10% dual CO [2 + 2] products. While the adsorption energy for single CO [2 + 2] cycloaddition is low, the geometry of 1,4-PDIC allows for dual CO [2 + 2] cycloaddition to occur and the adsorption energy for this product is sufficient for it to be observed in our experiments. Thus, 1,4-DICB, 1,3-PDIC, and 1,4-PDIC provide examples of how selectivity may be reduced for bifunctional molecules by enabling different reaction pathways. This is an important consideration for applications, such as MLD, in which bifunctional molecules are used.
basis of the coverage and product distribution, we can also calculate the fraction of dimers that has reacted with an adsorbate and the relative number of free, unreacted isocyanates at the surface. These data are also shown in Table 1. The surface coverages of 1,4-DICB and 1,3-PDIC are equal and approximately two-thirds that of 1,4-PDIC. Due to the higher percentage of dually tethered adsorbates for 1,4DICB, it actually occupies significantly more dimers at the surface than 1,3-PDIC. 1,4-PDIC, despite forming almost entirely the single CN [2 + 2] product, occupies the greatest fraction of dimers at the surface. From only these three adsorbates it is difficult to identify clear trends in what molecular properties of a bifunctional molecule affect surface coverage or the fraction of reacted dimers. We may speculate that 1,4-PDIC occupies the greatest fraction of surface dimers because calculations indicate that it assumes an orientation largely standing up away from the surface when adsorbed via single CN [2 + 2] cycloaddition. This orientation is unlikely to block a subsequent adsorbate’s access to a neighboring dimer. Conversely, when 1,3-PDIC adsorbs by single CN [2 + 2] cycloaddition, the second isocyanate group inhibits access to a neighboring Ge dimer across the trench regardless of whether it has actually reacted with that dimer, limiting it to 0.25 ML coverage. The relative amount of unreacted isocyanates can be compared to the FTIR spectra in Figure 7, showing the relative intensities of the isocyanate stretching peaks. As discussed above, the isocyanate stretching peaks of 1,4-DICB and 1,3-PDIC are 22% and 46%, respectively, as large as that of 1,4-PDIC. These values agree well with our XPS results, indicating that 1,4-DICB and 1,3-PDIC contain 28% and 54%, respectively, the number of unreacted isocyanate groups as 1,4PDIC. Thus, both the FTIR and XPS data show that 1,4-DICB forms the greatest fraction of dual CN [2 + 2] product, 1,3PDIC forms a smaller fraction of dual CN [2 + 2] product, and 1,4-PDIC forms almost entirely the single CN [2 + 2] product.
■
ASSOCIATED CONTENT
S Supporting Information *
Calculated CO [2 + 2] cycloaddition spectra, multilayer XP spectra for all three adsorbates, and complete ref 41. This material is available free of charge via the Internet at http:// pubs.acs.org.
IV. CONCLUSIONS We investigated the adsorption of three diisocyanates1,4DICB, 1,3-PDIC, and 1,4-PDICat the Ge(100)-2 × 1 surface under UHV using XPS, FTIR spectroscopy, and DFT calculations. DFT predicts that [2 + 2] cycloaddition across the CN bond of the isocyanate will be favored over other reactions and shows that formation of dual CN [2 + 2] products is favorable for 1,4-DICB and 1,3-PDIC but not for 1,4-PDIC. FTIR and XPS data confirm that CN [2 + 2] cycloaddition is the primary reaction pathway but show that a small percentage of CO [2 + 2] cycloaddition products is formed for all three adsorbates. In agreement with the DFT calculations, 1,4-DICB and 1,3-PDIC form significant fractions of dual CN [2 + 2] products, but 1,4-PDIC is limited to only single CN [2 + 2] cycloaddition. While 1,4-DICB forms equal amounts of single and dual CN [2 + 2] products, 1,3PDIC forms one-sixth as much dual CN [2 + 2] product as single CN [2 + 2] product. Thus, we show that the propensity for dual attachment of a diisocyanate at the surface is greatly affected by the spacer between the two functional groups. A flexible alkyl spacer favors dual attachment to the surface, whereas for a stiff aromatic spacer, position of the functional groups on the ring has a large impact on the favorability of attachment through both functional groups. In all cases, however, a mixture of two or more products is observed.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the National Science Foundation (CHE 0910717).
■
REFERENCES
(1) Avasthi, S.; Qi, Y.; Vertelov, G. K.; Schwartz, J.; Kahn, A.; Sturm, J. C. Appl. Phys. Lett. 2010, 96, 222109. (2) Bent, S. F. Surf. Sci. 2002, 500, 879. (3) Cui, Y.; Wei, Q.; Park, H.; Lieber, C. M. Science 2001, 293, 1289. (4) Guisinger, N. P.; Basu, R.; Baluch, A. S.; Hersam, M. C. Ann. N.Y. Acad. Sci. 2003, 1006, 227. (5) Guisinger, N. P.; Greene, M. E.; Basu, R.; Baluch, A. S.; Hersam, M. C. Nano Lett. 2004, 4, 55. (6) Hanrath, T.; Korgel, B. A. J. Am. Chem. Soc. 2004, 126, 15466. (7) Hanson, E. L.; Guo, J.; Koch, N.; Schwartz, J.; Bernasek, S. L. J. Am. Chem. Soc. 2005, 127, 10058.
12678
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
The Journal of Physical Chemistry C
Article
(8) Hurley, P. T.; Ribbe, A. E.; Buriak, J. M. J. Am. Chem. Soc. 2003, 125, 11334. (9) Lin, Z.; Strother, T.; Cai, W.; Cao, X. P.; Smith, L. M.; Hamers, R. J. Langmuir 2002, 18, 788. (10) Lopinski, G. P.; Wayner, D. D. M.; Wolkow, R. A. Nature 2000, 406, 48. (11) Stewart, M. P.; Buriak, J. M. Comments Mod. Chem. A 2002, 23, 179. (12) Strother, T.; Knickerbocker, T.; Russell, J. N.; Butler, J. E.; Smith, L. M.; Hamers, R. J. Langmuir 2002, 18, 968. (13) Wolkow, R. A. Annu. Rev. Phys. Chem. 1999, 50, 413. (14) Yates, J. T. Science 1998, 279, 335. (15) Zhou, H.; Bent, S. F. ACS Appl. Mater. Interfaces 2011, 3, 505. (16) Bent, S. F. Semiconductor surface chemistry. In Chemical Bonding at Surfaces and Interfaces; Nilsson, A., Pettersson, L. G. M., Nørskov, J. K., Eds.; Elsevier: Amsterdam, 2008; p 323. (17) Bent, S. F.; Kachian, J. S.; Rodríguez-Reyes, J. C. F.; Teplyakov, A. V. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 956. (18) Filler, M. A.; Bent, S. F. Prog. Surf. Sci. 2003, 73, 1. (19) Hamers, R. J. Ann. Rev. Anal. Chem. 2008, 1, 707. (20) Hamers, R. J.; Coulter, S. K.; Ellison, M. D.; Hovis, J. S.; Padowitz, D. F.; Schwartz, M. P.; Greenlief, C. M.; Russell, J. N. Acc. Chem. Res. 2000, 33, 617. (21) Kachian, J. S.; Wong, K. T.; Bent, S. F. Acc. Chem. Res. 2010, 43, 346. (22) Loscutoff, P. W.; Bent, S. F. Annu. Rev. Phys. Chem. 2006, 57, 467. (23) Lu, X.; Lin, M. C. Int. Rev. Phys. Chem. 2002, 21, 137. (24) Leftwich, T. R.; Teplyakov, A. V. Surf. Sci. Rep 2008, 63, 1. (25) Tao, F.; Bernasek, S. Functionalization of Semiconductor Surfaces; Wiley-VCH: New York, 2012. (26) Takagi, S.; Tezuka, T.; Irisawa, T.; Nakaharai, S.; Maeda, T.; Numata, T.; Ikeda, K.; Sugiyama, N. Mater. Sci. Eng. B: Solid 2006, 135, 250. (27) Kim, A.; Filler, M. A.; Kim, S.; Bent, S. F. J. Am. Chem. Soc. 2005, 127, 6123. (28) Loscutoff, P. W.; Lee, H.-B.-R.; Bent, S. F. Chem. Mater. 2010, 22, 5563. (29) Loscutoff, P. W.; Zhou, H.; Clendenning, S. B.; Bent, S. F. ACS Nano 2010, 4, 331. (30) Loscutoff, P. W.; Clendenning, S. B.; Bent, S. F. Mater. Res. Soc. Symp. P 2010, 1249, F02. (31) George, S. M.; Yoon, B.; Dameron, A. A. Acc. Chem. Res. 2009, 42, 498. (32) Loscutoff, P. W.; Wong, K. T.; Bent, S. F. Surf. Sci. 2010, 604, 1791. (33) Loscutoff, P. W.; Wong, K. T.; Bent, S. F. J. Phys. Chem. C 2010, 114, 14193. (34) Kachian, J. S.; Bent, S. F. J. Phys. Chem. C 2010, 114, 22230. (35) Kachian, J. S.; Jung, S. J.; Kim, S.; Bent, S. F. Surf. Sci. 2011, 605, 760. (36) Shong, B.; Wong, K. T.; Bent, S. F. J. Phys. Chem. C 2012, 116, 4705. (37) Yang, S.; Kim, Y.; Park, S.; Lim, H.; Lee, H. J. Phys. Chem. C 2011, 115, 9131. (38) Youn, Y.-S.; Kim, K.-j.; Kim, B.; Kim, D. H.; Lee, H.; Kim, S. J. Phys. Chem. C 2010, 115, 710. (39) Mui, C.; Han, J. H.; Wang, G. T.; Musgrave, C. B.; Bent, S. F. J. Am. Chem. Soc. 2002, 124, 4027. (40) Filler, M. A.; Van Deventer, J. A.; Keung, A. J.; Bent, S. F. J. Am. Chem. Soc. 2006, 128, 770. (41) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery Jr., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.et al. Gaussian 03, Rev. D.01 ed.; Gaussian, Inc.: Wallingford, CT, 2004. (42) Konecny, R.; Doren, D. J. Surf. Sci. 1998, 417, 169. (43) Wang, G. T.; Mui, C.; Tannaci, J. F.; Filler, M. A.; Musgrave, C. B.; Bent, S. F. J. Phys. Chem. B 2003, 107, 4982. (44) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.
(45) NIST Mass Spec Data Center, S. E. S., Director. Infrared Spectra. In NIST Chemistry WebBook, NIST Standard Reference Database Number 69; Linstrom, P. J., Mallard, W. G., Eds.; National Institute of Standards and Technology: Gaithersburg, MD. (46) Lee, H.; Youn, Y.-S.; Kim, S. Langmuir 2009, 25, 12574. (47) Cho, Y. E.; Maeng, J. Y.; Kim, S.; Hong, S. J. Am. Chem. Soc. 2003, 125, 7514.
12679
dx.doi.org/10.1021/jp302930g | J. Phys. Chem. C 2012, 116, 12670−12679
