Vibrational Spectroscopy of Periodic Mesoporous Organosilicas

Mar 24, 2007 - Periodic Mesoporous Organosilica (PMO) Materials with Uniform Spherical Core-Shell Structure. Stefanie Haffer , Michael Tiemann , Micha...
17 downloads 0 Views 487KB Size
5648

J. Phys. Chem. C 2007, 111, 5648-5660

Vibrational Spectroscopy of Periodic Mesoporous Organosilicas (PMOs) and Their Precursors: A Closer Look Frank Hoffmann,† Martin Gu1 ngerich,‡ Peter J. Klar,§ and Michael Fro1 ba*,† Institute of Inorganic and Analytical Chemistry, Justus Liebig UniVersity Giessen, Heinrich-Buff-Ring 58, D-35392 Giessen, Germany, Department of Physics and Material Sciences Center, Philipps UniVersity of Marburg, Renthof 5, D-35032 Marburg, Germany, and I. Institute of Physics, Justus Liebig UniVersity Giessen, Heinrich-Buff-Ring 16, D-35392 Giessen, Germany ReceiVed: October 19, 2006; In Final Form: February 8, 2007

The vibrational (Raman and IR) spectra of four different periodic mesoporous organosilica (PMO) materials and the corresponding bis-silylated precursors were investigated. An assignment of the most prominent bands was possible with the help of supporting and complimentary theoretical DFT calculations. The hitherto widespread belief that vibrational spectroscopy is an appropriate tool to examine the status of the siliconcarbon bond in organosilicon/organosilica compounds is revisited and finally discarded. A thorough analysis of the spectra revealed that (i) ν(Si-C) stretching vibrations are neither eigenmodes of the precursors nor of the PMO materials, (ii) any other vibration involving the Si-C fragment is not necessarily a characteristic group frequency, (iii) there is no simple analogy between different PMO materials with different organic bridges (each material is a “single case”), and (iv) that, as a consequence, vibrational spectroscopy is not a reliable method for monitoring the integrity of the Si-C bonds in PMOs and other related organosilica materials (xerogels and aerogels). Carrying out solid-state NMR spectroscopy is an absolute necessity in order to give valid statements regarding the extent of Si-C bond cleavage, which occurred during the synthesis of organosilica materials.

Introduction In 1999, a new class of mesoporous organic-inorganic hybrid materials called periodic mesoporous organosilicas (PMOs) was independently developed by three research groups.1-3 The synthesis procedure of these materials involves hydrolysis and condensation reactions of bridged organosilsesquioxane precursors [(R′O)3Si-R-Si(OR′)3] in the presence of supramolecular aggregates of long-chain surfactants serving as structuredirecting agents (SDAs). The precursor compounds have been employed in sol-gel chemistry for quite some time. Common organofunctionalized mesoporous pure silica phases (for an overview, see for instance, refs 4-8 and references therein) are usually obtained either via the grafting approach, where the organic functionalities are bound on the silica surface by a postsynthesis treatment, or alternatively by the co-condensation method (so-called “one-pot” synthesis) of tetraalkoxysilanes and terminal organotrialkoxysilanes. In contrast, the organic units in PMOs are two-point attached within the inorganic silica backbone through covalent bonds and are thus completely homogeneously distributed, being a genuine part of the 3D framework structure. So far, quite a number of different PMO materials have been successfully synthesized, among them those with alkyl, aromatic, heteroaromatic, ether, thioether, and dendrimer building (con* To whom correspondence should be addressed. Phone: +49-641-9934100. Fax: +49-641-99-34109. E-mail: [email protected]. uni-giessen.de. † Institute of Inorganic and Analytical Chemistry, Justus Liebig University Giessen. ‡ Department of Physics and Material Sciences Center, Philipps University of Marburg. § I. Institute of Physics, Justus Liebig University Giessen.

taining SiC4 units) organic bridging groups.7,8 As the synthesis of PMO or related organosilica materials is usually conducted in a relatively harsh environment (hydrothermal treatment at very low or very high pH values), the question whether and to what extent the Si-C bonds were cleaved during the synthesis procedure is always a key issue. Solid-state 29Si MAS NMR spectroscopic measurements often are carried out in order to verify and quantify the integrity of the organic bridging groups inside the PMO framework structure. The intensity of Qn signals [Si(OSi)n(OH)4-n, n ) 2-4] of the spectra is proportional to the amount of the respective silicon species and can therefore be used to determine the extent of Si-C bond cleavages, which occurred during the synthesis and/or calcinations/extraction procedure of the PMO/organosilica samples. Sometimes, it is intended to use IR or Raman spectroscopy as alternative or complementary methods to solid-state NMR spectroscopy for the characterization of either real PMOs9-17 or other organofunctionalized mesoporous silica phases.18-20 This technique enables the detection of the presence of possibly remaining surfactants after the calcination/extraction procedure. Furthermore, they allow one to monitor the status of functional groups within the organic moieties inside the walls of PMO materials. Very recently, Morell et al.21 have shown that it is even possible to determine, quantitatively, the proportions of two organic groups of a bifunctional PMO material (containing different amounts of benzene- and thiophene-bridging groups simultaneously) with the help of Raman spectroscopy. However, it is more than questionable whether vibrational spectroscopy is the appropriate tool for yielding definite statements concerning the status of the Si-C bonds, as many authors, including ourselves,15.16 have attempted. The reasons are the following. (i) The correct assignment of bands in the

10.1021/jp0668596 CCC: $37.00 © 2007 American Chemical Society Published on Web 03/24/2007

Vibrational Spectroscopy of PMOs expected wavenumber region of ν(Si-C) modes is a challenging task and often quite ambiguous (fingerprint region). (ii) Due to the solid state and to the altered symmetry of the samples, the mode structure, the degree of Raman and IR activity, and the selection rules are modified compared with the respective molecular species (precursors). (iii) Collective framework (lattice) vibrations are present, which may superimpose bands caused by vibrations of the molecular fragments. (iv) Finally, strong vibrational couplings are to be expected, which give rise to additional wavenumber shifts and intensity changes. The corresponding mode patterns usually involve atoms, which are distributed over the entire molecule. In particular, vibrations of the Si-C bond alone, if they are eigenmodes of the precursor or the PMO products at all, will depend very strongly on the surroundings. These circumstances are somewhat reflected in the inconsistent and nonuniform assignments of modes, which involve Si-C fragments in the existing literature of this field. For instance, Dag et al. 9,10 are confident that one to three Raman bands between 510 and 630 cm-1, which are dependent on the specific organic bridging group, can be assigned to ν(Si-C) stretching vibrations in PMO materials. Temtsin et al.11 assigned two Raman bands around 708 and 1121 cm-1 to silicon-phenyl stretching vibrations in tolyl-bridged PMO materials. Wahab et al.14 argued that the organic moieties in propylamine-bridged PMOs remained intact during the synthesis based on the presence of a band at 1270 cm-1, which they assigned to Si-C vibrations. Rebbin et al.15 assigned two FT-IR bands at 696 and 768 cm-1 to Si-C stretching vibrations of ethane-bridged PMOs, and Morell et al.16 identified two Raman bands at 591 and 753 cm-1 as Si-C stretching modes in large-pore thiophenebridged PMO materials. Tanaka et al.18 recorded FT-IR spectra of samples of mesostructured thin silica films, which were vapor infiltrated with organosiloxanes such as methyltriethoxysilane (MeSi(OEt)3), dimethyldiethoxysilane (Me2Si(OEt)2), or trimethylethoxysilane (Me3Si(OEt)). The spectra of the organofunctionalized films showed two additional bands at 756 and 1258 cm-1, which were attributed to stretching modes of SiCH3 groups. Recently, Dı´az-Morales et al.17 assigned two FTIR bands at 698 and 774 cm-1 to Si-C stretching vibrations in ethane-bridged PMOs. Thus, overall, there are reasonable doubts that conclusions regarding the integrity of the organic components in organic-inorganic hybrid materials can be drawn reliably based on vibrational spectroscopy. The objective of the present work is to unravel the vibrational spectroscopic features of organosilica and PMO materials, in particular, and their respective precursors, in general, with special emphasis on the postulated ν(Si-C) stretching modes. For this purpose, we have chosen a combined experimental and theoretical approach. The vibrational spectra of four different PMO precursors, namely bis(triethoxysilyl)ethane (BTEE), -thiophene (BTET), -benzene (BTEB), and -biphenyl (BTEBP) (see Figure 1), were calculated with the help of density functional theory (DFT). These spectra were compared with (i) the corresponding experimental ones, (ii) the spectra of the respective final PMO materials, as well as (iii) the spectra of the “naked” ethane, thiophene, benzene, and biphenyl compounds of which the PMO precursors can be regarded as derivatives and which are very well documented in the literature. In combination with the displacement vector images of the theoretically calculated vibrational modes, an assignment of the PMO precursor modes is possible. To gain at least an idea of which bands will change in which way (position, intensity, width) upon the transition from the precursors to the final PMO materials, we conducted further theoretical frequency analyses,

J. Phys. Chem. C, Vol. 111, No. 15, 2007 5649

Figure 1. Constitution formulas of the organosilica/PMO precursors of which experimental and theoretical vibrational spectra were analyzed in this study.

namely of the products of the first condensation steps of the precursors, which are, for simplicity, denoted as “dimers”, “trimers”, and “tetramers”, respectively. In order to save CPU time, they were treated in their (already hydrolyzed) -OH form. Although these model compounds might appear as a rather limited approximation regarding real PMO materials, they nevertheless contain at least all atomistic constituents in their right coordination and chemical surroundings; therefore, they should give some clues about the behavior of the entire framework. Experimental Section Raman spectra were acquired using a Raman microscope system (Jobin-Yvon). The green line (514.5 nm) of an Ar-ion laser was used for excitation, and a single grating monochromator with a CCD multichannel detector system was used for recording the Raman spectra. The Rayleigh-scattered laser light was rejected using a holographic notch filter. The excitation power on the sample was less than 20 mW, and the spectral resolution of the detection system was about 1 cm-1. FT infrared spectra were acquired using a Bruker IFS 25 FTIR spectrometer. The precursor samples were measured as a film, and the PMO material samples were measured with the KBr method. The spectral resolution was 2 cm-1. Theoretical DFT calculations were conducted as follows. First, each compound was geometry-optimized to its respective energy minimum at the B3LYP/6-31G(d,p) level of theory using the implementation of GAUSSIAN 03.22 A vibrational analysis followed, employing the same method and basis set. According to the level of theory and basis set, a scaling factor of 0.97 was applied for the whole wavenumber region, irrespective of the kind of vibration (stretch or bending). The vector displacements of the modes were visualized and analyzed with the tool ChemCraft23 for Gaussian output files. Results and Discussion As the C-Halkyl and C-Haryl stretching vibrations are not characteristic of the individual compounds, in all cases, only the wavenumber region between 400 and 1700 cm-1 will be considered, and the modes located there will be discussed. Furthermore, the main focus is placed upon the Raman spectra. The bands of the experimental IR spectra of the solid-state PMO material samples are often very broad, with maxima which are difficult to define. This complicates the comparison with the calculated IR spectra considerably, which are always obtained only as “stick spectra”. In addition, it should be clear that the limited PMO model compounds (dimers, trimers, tetramers) cannot account for the abundance of highly coupled lattice

5650 J. Phys. Chem. C, Vol. 111, No. 15, 2007 vibrations of the quasi-infinite network of a real amorphous solid-state PMO sample. At the level of the precursors, all main features of the Raman as well as IR spectra are reproduced very well (see below and Figure S1-S4 in the Supporting Information), in particular the position of the modes, although some discrepancies regarding the relative intensities and widths of the bands are present. We will show that the applied DFT method is valid and that the achieved accuracy is sufficient in order to extrapolate from the PMO model compounds toward the vibrational spectroscopic features of real PMO samples. Thiophene-Bridged PMOs. In Figure 2, the experimental Raman and FT-IR spectra of thiophene, the PMO precursor BTET, and the final thiophene-bridged PMO material (a-c) as well as the calculated Raman and IR spectrum of the “thiophenetrimer” (d) are shown. In accordance with the literature24 and with the conducted DFT calculations, the main Raman-active characteristic modes of the thiophene compound (point group C2V) can be assigned as follows (the calculated values are given in parentheses): 1404 (1424) and 1356 (1366) cm-1 ν(ring), 1078 (1080) and 1033 (1027) cm-1 δ(C-H)i.p., 832 (813) cm-1 ring breathing, and 604 (594) cm-1 δ(ring)i.p.. The additional characteristic IR-active modes are 1251 (1206) cm-1 δ(CH)i.p., 713 (686) cm-1 δ(CH)o.o.p., and 452 (440) cm-1 δ(ring)o.o.p.. Switching over to the Raman spectrum of BTET, some bands become broader and less intense, some disappear almost completely, some are shifted to lower or higher wavenumbers, and only a few additional bands, caused by the vibrations of the ethoxysilyl groups, are visible. While the symmetry of the thiophene is C2V, the highest possible symmetry of BTET is C2; in fact, due to the conformational degrees of freedom of the ethoxysilyl groups, it will be even lower, that is, C1. A reduction in symmetry should result in an increased number of active normal modes; however, it seems that most of them are only very weakly Raman active, in accordance with the results of the DFT calculations. Furthermore, it is to be expected that all C-H deformation modes of the heterocyclic core will be strongly effected because two of the hydrogens are now being substituted by the triethoxysilyl groups. In detail, the higher wavenumber mode of the two ring stretching vibrations is almost unaffected, now lying at 1413 cm-1 (calculated value, 1425 cm-1). The other ring stretching vibration mode is strongly effected and shifted by about 90 cm-1 to lower values, now lying at 1268 cm-1 (calculated value, 1269 cm-1). The two symmetric in-plane C-H deformation modes formerly entailed two intense bands at 1078 and 1033 cm-1, co-incident to give only one broad and much less intense band at 1089 cm-1 (the DFT calculation gives two very weak Raman-active modes located at 1118 and 1075 cm-1). The ring breathing mode of the bare thiophene (832 cm-1) is not Raman active anymore in BTET; the corresponding band disappears. And finally, the symmetric in-plane ring deformation mode is shifted toward higher wavenumbers from 604 to 647 cm-1 (calculated value, 628 cm-1). For an overview of the assignments of the Raman and IR bands of thiophene and BTET, see Table 1 (a detailed view of the experimental and simulated IR spectrum of BTET can be found in Figure S1). Herewith, the Raman spectrum of BTET is actually completely explained. The only question which remains is whether there are any modes which can be regarded as vibrations, which involve motions of the silicon atom along the Si-C bond, and if these modes are Raman or IR active. Screening all of the 3n - 6 ) 165 vibrational modes obtained by the DFT calculation gave four vibrational modes located at 481, 512, 679, and 689 cm-1, which involve a displacement of the silicon atom along

Hoffmann et al. the Si-C bond direction, but these modes are highly coupled with out-of-plane ring deformation and ethoxysilyl deformation modes; they could also be described as δ(Si-(OEt)3) umbrella vibrations. All of the four modes are almost Raman inactive, but the modes at 512 and 689 cm-1 are highly IR active. The experimental IR spectrum shows a relatively broad band around 531 cm-1, which is assigned to an out-of-plane ring deformation mode. It may be possible that the broadening is partly caused by one of these Si-(OEt)3 umbrella vibrations; however, this band cannot be reliably assigned to a characteristic ν(Si-C) stretching mode of PMO precursors, let alone be regarded as a qualifying band for the proof of intact Si-C bonds. If the precursor does not show any significant or specific Si-C stretching vibration, it is very unlikely that the polycondensation productsthe final PMO materialswill show such modes. The condensation process will lead to a new Si-O-Si bridging motif. Therefore, one would expect that the corresponding Si-O stretching and Si-O-Si deformation modes cause additional bands in the respective IR or Raman spectra. The simulation of these spectra with the help of all-atom DFT calculations is far beyond the current capabilities of computational methods. However, model compounds can be constructed which contain all structural motif elements of the PMO material and thus should contain all of the main features of the vibrational spectra of the PMO. Therefore, we decided to carry out calculations on the product after the first and second condensation step of the thiophene precursor, that is, on the thiophenedimer and thiophene-trimer, respectively. The geometryoptimized structure of the trimer is depicted in Figure 3. Interestingly, while starting with a linear-like conformation, the compound adopts a ring-like geometry in the course of the optimization procedure with a pair of two hydrogen bonds at the end of the chain. The experimental Raman and FT-IR spectra of the thiophene-bridged PMO and the calculated spectra of the trimer are compared in Figure 2c and d. While a fairly good agreement with the experimental Raman spectrum of the final PMO material can already be achieved with the thiophene-dimer, the theoretical IR spectrum deviates considerably, especially in the low wavenumber region (data not shown), from its experimental counterpart. Although the thiophene-trimer model compound is still a very crude approximation of the real PMO material, the calculated Raman spectrum agrees surprisingly well with the experimental Raman spectrum of the final PMO sample. However, the same does not hold to the same degree for the respective IR spectrum. Probably, the trimer model compound suffers from the limited degree of condensation. As a consequence, various Si-O-Si as well as collective framework vibrations (which most of them are expected to show significant IR but not Raman activity) are not present. Comparing the Raman spectra of BTET and the thiophenebridged PMO, the main difference can be identified in the region between 550 and 800 cm-1; the thiophene-bridged PMO material shows three bands with almost the same intensity, two of which are broad with maxima at 591 and 682 cm-1 and a narrower one at 755 cm-1. The result of the theoretical vibrational analysis strongly suggests that these bands are not caused by Si-C modes, which means that the assignment conducted by Dag et al. 9,10 (two to three bands at 510-630 cm-1) and Temtsin et al.11 (two bands at 708 and 1121 cm-1) cannot be confirmed. Instead, the origin of these bands is most likely due to (i) various Si-O-H in-plane and out-of-plane bending modes (altogether more than 20 for the model compound between 400 and 4000 cm-1) and (ii) δ(CH) in-plane

Vibrational Spectroscopy of PMOs

J. Phys. Chem. C, Vol. 111, No. 15, 2007 5651

Figure 2. (a)-(c) Experimental Raman (left side) and FT-IR (right side) spectra of thiophene, the PMO precursor BTET, and the thiophenebridged PMO material. (d) calculated Raman (left side) and IR spectra (right side) of the thiophene-trimer model compound; the envelope curves were generated out of the stick spectrum by Lorentzian broadening with a fwhm value of 20 cm-1.

deformation modes. Most of these bending modes have a low Raman activity but show significant IR intensity. That means, likewise, that the FT-IR band assignment to the ν(Si-C) stretching modes conducted by Tanaka et al.18 (two bands at 756 and 1258 cm-1) seems also to be wrong (see also below).

Benzene-Bridged PMOs. The experimental Raman and FTIR spectra of benzene, the PMO precursor BTEB, the final benzene-bridged PMO material, as well as the calculated spectra of a tetrameric model compound in a cage-like arrangement (see below) are shown in Figure 4.

5652 J. Phys. Chem. C, Vol. 111, No. 15, 2007

Hoffmann et al.

TABLE 1: Assignment of Characteristic IR and Raman Bands of Thiophene and BTETa thiophene

a

mode

Ra/IR

ν1 ν2 ν3 ν4 ν5 ν6 ν7 ν8 ν9 ν10 ν10

b/b b/O O/b b/b b/O b/O b/O O/b b/O O/b O/O

BTET

exp.

calc

Ra/IR

1404/1407 1356 1251 1078/1081 1033 832 748 713 604 452 -

1424 1366 1206 1080 1027 813 725 686 594 440 -

b/O b/b O/O b/O O/b O/O b/b O/O b/O O/b O/b

exp.

calc.

assignment

1413 1268/1272 1089 1079 752 647 531 458

1425 1269 1118 1074 976 726 628 519 481

ν(ring) ν(ring) δ(CH)i.p. δ(CH)i.p. δ(CH)i.p. ring breath. δ(ring)i.p. δ(CH)o.o.p. δ(ring)i.p. δ(ring)o.o.p. δ(Si-(OEt)3)

O/b ) inactive/active; exp./calc. ) experimental/calculated values in cm-1; - ) band not observable or not unambiguously assignable.

Figure 3. Geometry-optimized structure as a ball-and-stick model of the product after the second condensation step of the thiophene-bridged PMO precursor BTET, “thiophene-trimer”. Carbon, gray; hydrogen, white; sulfur, yellow; silicon, orange; oxygen, red.

In accordance with the literature24 and with the conducted DFT calculations, the main Raman-active characteristic modes of the benzene compound (point group D6h) can be assigned as follows (the calculated values are given in parentheses): 1604 (1602) cm-1 ν(ring), 1585 cm-1 [this band is caused by Fermi resonance between the ring stretching mode and a combination mode of the ring breathing (992 cm-1) and ring deformation (606 cm-1) mode], 1174 (1165) cm-1 δ(C-H)i.p. (two-folddegenerated), 991 (988) cm-1 ring breathing, and 606 (602) cm-1 δ(ring)i.p. (two-fold-degenerated). The additional characteristic IR-active modes are 1479 (1478) cm-1 δ(ring)i.p./δ(CH)i.p. (two-fold-degenerated), 1036 (1034) cm-1 δ(CH)i.p., and 673 (650) cm-1 δ(CH)o.o.p. (two-fold-degenerated). The Raman spectrum of the PMO precursor BTEB shows interesting changes with respect to the naked benzene compound. As expected due to the symmetry changes, the Fermi resonance feature at around 1600 cm-1 is no longer visible. The ring stretching mode appears now as a single band with higher intensity at 1593 (1598) cm-1. The two-fold-degenerated inplane C-H deformation mode, located at 1174 cm-1 in the case of the bare benzene, is split into two distinct bands; one of these is shifted moderately and the other strongly toward higher energies, now lying at 1194 (1188) and 1289 (1267) cm-1. The ring breathing mode is strongly effected by the substitution with the ethoxysilyl groups at the 1,4 position and is substantially

shifted toward higher wavenumbers, now lying at 1097 (1086) cm-1, and the formerly two-fold-degenerated in-plane ring deformation mode at 606 cm-1 is again split into two distinct bands, one is located at 671 (662) cm-1 and the other at 634 (629) cm-1. Furthermore, there are three additional bands or groups of bands visible in comparison to the spectrum of the bare benzene. These features can be easily attributed to various typical deformation modes of the methylene and methyl group of the ethoxysilyl moiety. First of all, the δ(CH2) scissoring mode gives rise to the band at 1479 (1491) cm-1, a coupled deformation mode of the methylene (scissoring) and methyl groups (δ(CH3)o.o.p.) leads to the band at 1453 (1468) cm-1, and the band at 1443 (1453) cm-1 can be assigned to an isolated deformation mode of the methyl group (δas(CH3). Second, the band with a maximum at 937 cm-1 (the DFT calculation give six modes with wavenumbers in the range from 955 to 913 cm-1) is caused by a mode which can be described as a combination of the methylene wagging and an asymmetric methyl nick vibration. Third, the broad band around 781 cm-1 (the DFT calculations give eight modes between 796 and 782 cm-1) is again a coupled mode of a methylene rocking and a methyl deformation mode. For an overview of the assignments of the Raman and IR bands of benzene and BTEB, see Table 2; a detailed view of the experimental and simulated IR spectrum of BTEB can be found in Figure S2 As in the case of the thiophene-bridged PMO precursor, the DFT calculation gives hints about modes which resemble something like “pseudo Si-C stretching modes”, involving a movement of the aromatic ring against the rest of the molecule or ring stretching/deformation modes which are linked with motions of the Si atom along the Si-C bond and coupled with δ(Si-(OEt)3) umbrella vibrations. These modes, located at 483, 533, 696, and 760 cm-1 are highly coupled with a very complex pattern of movements of almost all of the atoms of the molecules. The modes at 533 and 696 cm-1 exhibit high IR intensity. A second mode with high IR activity in this region is identified as an out-of-plane ring deformation mode calculated at 524 cm-1. Those two highly active IR modes below 600 cm-1 could account for and correspond well with the two relatively strong bands in the experimental FT-IR spectrum at 546 and 524 cm-1. In the case of the thiophene precursor BTET, the pseudo Si-C vibrations were located at 481 and 512 cm-1, which means that this type of mode also is not characteristic for bis-silylated PMO precursors (due to their varying location) and that it might be impossible to distinguish the Si-R3 umbrella vibrational mode from aromatic out-of-plane deformation modes. In addition, and this is even more important, it is very unlikely that these modes will exist in these forms in the final PMO materials because they exhibit no such isolated or

Vibrational Spectroscopy of PMOs

J. Phys. Chem. C, Vol. 111, No. 15, 2007 5653

Figure 4. (a)-(c) Experimental Raman (left side) and FT-IR (right side) spectra of benzene, the PMO precursor BTEB, and the benzene-bridged PMO material. (d) Calculated Raman (left side) and IR spectra (right side) of the benzene-tetramer model compound; the envelope curves were generated out of the stick spectrum by Lorentzian broadening with a fwhm value of 20 cm-1.

terminal Si-OEt3 motifs anymore; instead, the silicon atoms are integrated into an “infinite” network. Interestingly, the ν4 mode (see Table 2), a pure δ(CH)i.p. mode in the case of the bare benzene, is shifted remarkably toward higher wavenumbers

(now lying at 1133 cm-1) and now also comprises something like a pseudo Si-C vibration, in which the carbon atom connected to silicon makes a compensation movement along the silicon-carbon bond (see also Figure 10). However, it is

5654 J. Phys. Chem. C, Vol. 111, No. 15, 2007

Hoffmann et al.

TABLE 2: Assignment of Characteristic IR and Raman Bands of Benzene and BTEBa benzene mode Ra/IR

exp.

calc.

BTEB Ra/IR

exp.

b/O

1593 1598 1528 1483 1486 1289 1267 1194 1188 - 1133 1097 1086 867 842 671 662 634 629 546 533 524 524

ν1

b/O

1604b 1602

ν2

O/b

1479

1478 (2f) O/b b/O 1165 (2f) b/O 1034 (2f) O/O 988 b/O O/O 693 (2f) O/O b/O 602 (2f) b/O O/b O/b

ν3

b/O

1174

ν4 ν5

O/b b/O

1036 991

ν6

O/b

673

ν7

b/O

606

ν8 ν9

O/O O/O

-

calc.

assignment ν(ring) ν(ring)/δ(CH)i.p. δ(CH)i.p. δ(CH)i.p. ring breath. δ(CH)o.o.p. δ(ring)i.p. δ(Si-(OEt)3) δ(ring)o.o.p.

a O/b ) inactive/active; exp./calc. ) experimental/calculated values in cm-1; - ) band not observable or not unambiguous to assign,; 2f ) two-fold-degenerated. b Fermi resonance leads to a second band at 1585 cm-1.

questionable whether this mode can be used to identify siliconcarbon bonds because the frequency and IR and Raman activity of these modes should be highly dependent on the specific kind of organic bridge (saturated alkyl, unsaturated alkyl, aromatic), if a corresponding band will appear in the spectrum at all. For instance, such a mode is absent in thiophene for symmetry reasons. Accordingly, the mode does not exist in the silylated counterpart BTET, as the two silicon atoms and the organic bridge exhibit no (local) linear geometry. The same holds for the thiophene-trimer. Regarding IR spectroscopy, such modes are not suitable for identification purposes anyway because the corresponding bands fall in a region where already abundant bands are present due to the diverse Si-O vibrations. The main differences between the experimental Raman spectra of the BTEB and that of the benzene-bridged PMO are (i) a new broad band appears at 584 cm-1, (ii) the intensity of the band around 780 cm-1 increases considerably, and (iii) the three δ(CH2) deformation bands of the ethoxy group at 1443, 1453, and 1479 cm-1 disappear, which is outright explainable as due to the absence of ethoxy groups in the final PMO product. The new broad band at 584 cm-1 is most likely again caused by various δ(Si-O-H) deformation modes, whereas the explanation of the relatively strong band at 780 cm-1 is not straightforward. A definite assignment is not possible yet. However, the study of the Raman and FT-IR spectra of BTEB and the benzene-bridged PMO shows again that a typical Si-C group frequency does not exist. We also performed calculations on the benzene-dimer, -trimer, and a special cage-like linked -tetramer whose geometryoptimized structure is shown in Figure 5. The cluster connectivity is inspired from the model of the benzene-bridged PMO pore wall published by Inagaki et al.25 Note, the bridging silicon atoms are saturated with hydrogen, while in the real infinite network of the PMO wall, these would be the connecting points of the next organic bridges. While the simulated spectra of the (linear linked) dimer and trimer resemble, more or less, the calculated spectra of the monomer (BTEB) (spectra not shown), the spectra of the cage-like tetramer show some additional features and resemble, to a larger extent, the experimental spectra. This holds particularly for the Raman spectrum and, to a lesser extent, for the IR spectrum. Especially the widths and intensities of the bands below 600 and around 1000 cm-1 are not very well reproduced in the simulated IR spectrum. Apparently, the tetramer cluster is still a limited model

Figure 5. Geometry-optimized structure (ball-and-stick representation) of the cage-like linked benzene-tetramer model compound. Carbon, gray; hydrogen, white; silicon, orange; oxygen, red.

compound in order to account for all the IR-active nonlocal framework vibrational modes of a real PMO network. The analysis of the calculated Raman spectrum of the tetramer revealed an active mode near 564 cm-1 (unscaled value, 581 cm-1), which can be assigned to δ(Si-O-H) deformations. This mode could eventually explain the experimentally observed band at 584 cm-1. Furthermore, there are two Raman-active δ(ring)i.p. deformation modes of the benzene moieties lying at 749 and 752 cm-1 (unscaled values, 749 and 752 cm-1, respectively), which could probably account for the band at 780 cm-1. A further analysis revealed no additional characteristic features beyond the modes, which were already explained for the monomer. Biphenyl-Bridged PMOs. In Figure 6 the experimental Raman and FT-IR spectra of biphenyl, the PMO precursor BTEBP, and the final biphenyl-bridged PMO material, as well as the calculated spectra of the biphenyl-dimer, are shown. In agreement with the literature24 and with the conducted DFT calculations, the main Raman-active characteristic modes of the biphenyl compound (the point group for the coplanar conformation would be D2; however, the geometry optimization procedure gives the twisted conformation (C1) with a dihedral angle of about 38° around the central C-C bond as the energy minimum) can be assigned as follows (the calculated values are given in parentheses): 1608 (1611) and 1590 (1609) cm-1 ν(ring), 1510 (1504) cm-1 δ(CH)i.p., 1274 (1270) cm-1 ν(C-C)central, 1034 (1028) cm- δ(CH)i.p., 999 (987) cm-1 ring breathing, and 739 (734) and 611 (609) cm-1 δ(ring)i.p.. The additional characteristic IR-active modes are 1480 (1482) and 1429 (1428) cm-1 δ(CH)i.p., 730 (733) and 698 (694) cm-1 δ(CH)o.o.p., and 611 (604) cm-1 δ(ring)i.p.. For an overview of the assignments of the Raman and IR bands of biphenyl and BTEBP, see Table 3; a detailed view of the experimental and simulated IR spectrum of BTEBP can be found in Figure S3.

Vibrational Spectroscopy of PMOs

J. Phys. Chem. C, Vol. 111, No. 15, 2007 5655

Figure 6. (a)-(c) Experimental Raman (left side) and FT-IR (right side) spectra of biphenyl, the PMO precursor BTEBP, and the biphenylbridged PMO material. (d) Calculated Raman (left side) and IR spectra (right side) of the biphenyl-dimer model compound; the envelope curves were generated out of the stick spectrum by Lorentzian broadening with a fwhm value of 20 cm-1.

Looking at the Raman spectrum of the PMO precursor BTEBP, it is interesting to note that there are only three bands with medium to strong intensity left; the intensity of most of

the other bands of the bare biphenyl compound has dropped drastically. The most prominent bands are now located at 1604, 1285, and 1127 cm-1. The first and second one are the known

5656 J. Phys. Chem. C, Vol. 111, No. 15, 2007

Hoffmann et al.

TABLE 3: Assignment of Characteristic IR and Raman Bands of Biphenyl and BTEBPa biphenyl mode Ra/IR exp. ν1 ν2 ν3 ν4 ν5 ν6 ν7 ν8 ν9 ν10 ν11 ν12 ν13 ν13

b/O b/O b/O O/b O/b b/O O/O b/O b/O b/O O/b O/b O/b b/O O/O

1608 1590 1510 1480 1429 1274 1034 999 739 730 698 611 611 -

BTEBP

calc. Ra/IR exp.

calc.

assignment

1611 1609 1504 1482 1428 1270 1039 1028 987 734 733 694 604 609 -

1607 1605 1507 1485 1271 1113 1112 989 732 724 718 637 625 485

δ(CH)i.p. δ(CH)i.p. δ(CH)i.p. ν(C-C)central δ(CH)i.p. δ(CH)i.p. ring breath. δ(ring)i.p. δ(ring)i.p./δ(CH)i.p. δ(CH)o.o.p. δ(ring)i.p. δ(ring)i.p. δ(Si-(OEt)3)

b/O

1604

b/O O/b O/O b/O b/O O/O b/O O/O O/b O/O O/O b/O O/b

1511 1483 1285 1127 994 735 631 503

ν(ring)

O/b ) inactive/active; exp./calc. ) experimental/calculated values in cm-1; - ) band not observable or not unambiguous to assign.

Figure 7. Geometry-optimized structure as ball-and-stick models of the first condensation products of the biphenyl-bridged PMO precursor BTEBP (biphenyl-dimer). Carbon, gray; hydrogen, white; silicon, orange; oxygen, red.

ring stretching and central C-C stretching modes, which are only slightly shifted with respect to the biphenyl molecule. The assignment of the band at 1127 cm-1 is not straightforward. The results of the DFT calculation suggests two possibilities. Most likely, it is the mode with the highest Raman intensity in this wavenumber regime, and this is a δ(ring)i.p./δ(CH)i.p. mode at a calculated value of 1112 cm-1. Finally, the relatively broad and weak bands at 1198 (1190), 783 (the calculation gives eight modes between 797 and 781 cm-1), and 700 (683) cm-1 can be attributed to various δ(CH2)/δ(CH3) modes. To come back to the question of whether there are any Si-C stretching modes, screening all of the 204 vibrational modes revealed four modes at 485, 528, 724, and 742 cm-1 (coupled δ(Si-(OEt)3)/δ(ring)i.p./δ(CH)i.p. modes), which are accompanied by the largest relative movement of Si and the neighboring C atom against each other, but they are, again, highly coupled modes, where almost all other atoms of the molecule also are in motion. The calculated Raman activity of the mode at 724 cm-1 is very low, but the IR intensity is very high, which might account for the relatively strong band in the experimental FT-IR spectrum at 735 cm-1. The pseudo Si-C stretching modes, which are actually δ(CH)i.p.modes (ν6 and ν7 at 1113 and 1112 cm-1, respectively), are similar to the mode described above for the silylated benzene derivative (see also Figure 10);26 both modes show weak IR activity, but ν7 is associated with a moderate Raman intensity. Interestingly, in contrast to the thiophene- and benzenebridged PMO material, the Raman spectrum of the biphenylbridged PMO material shows only one sharp band at 627 cm-1 but no broad band below 600 cm-1. Therefore, we decided to perform a theoretical frequency analysis of the biphenyl-dimer compound too, which is still computationally feasible in a reasonable time frame. The geometry-optimized structure of the biphenyl-dimer is shown in Figure 7. The experimental Raman and FT-IR spectra of the biphenyl-bridged PMO and the calculated spectra of the biphenyl-dimer are compared in Figure 6c and d. It would be desirable to calculate a tetrameric cage-like model compound too, as it is assumed that the biphenyl-bridged PMO shows a similar arrangement of the biphenyl units within the pore walls as that of the benzene units in the respective benzene-bridged PMO, but this was computationally too demanding. A very good agreement of the positions as well as of the intensities of the bands with those of the experimental Raman

spectrum of the final PMO material could already be achieved with the calculated spectrum of the biphenyl-dimer. As in the case of the thiophene-dimer, the theoretical IR spectrum shows some significant deviations, mainly with respect to the widths and intensities of the bands. As stated above, the degree of condensation is not sufficient to cover all of the IR-active nonlocal framework vibrations. Screening the 189 vibrational modes of the biphenyl dimer revealed six modes which are connected with a silicon-carbon stretching motion along their bond. Two are located at 432 and 453 cm-1 and can be again assigned to δ(Si-(OH)3) umbrella vibrations, whereas the mode at 453 cm-1 also involves a δ(Si-O-Si) deformation vibration. Both modes show high IR intensity but no Raman activity, in accordance with the experimental results. The other four modes at 1114, 1119, 1120, and 1121 cm-1 can be described again as δ(CH)i.p. deformation modes, one mode for each phenyl ring,27 in which the carbon atom connected to silicon makes a compensation movement along the silicon-carbon bond (see also Figure 10). These four modes are associated with a high IR and a moderate Raman intensity. Interestingly, in comparison to the “monomer” BTEBP, these modes are shifted again slightly toward higher wavenumbers. Therefore, it might be possible that these modes can account for the relatively broad Raman band found experimentally around 1206 cm-1 for the “polymeric” biphenylbridged PMO material. Ethane-Bridged PMOs. So far, only aromatic organobridged PMOs and precursors were considered. Therefore, we extended our studies to the nonaromatic PMO precursor BTEE, which not only consists of sp3-hybridized carbon atoms (in contrast to the sp2-hybridized in BTET, BTEB, and BTEBP) but also obeys a different symmetry. The Si-CH2-CH2-Si fragment should have, as a first approximation, a C2h symmetry. The altered symmetry should have an impact upon the selections rules and intensities of various bands; the different hybridization status should influence the force constant of the Si-C bond. We did calculations on BTEE, on the ethane-dimer, as well as on the cage-like linked ethane-tetramer model compound. In Figure 8, the experimental Raman and FT-IR spectra of the PMO precursor BTEE (a detailed view of the experimental and simulated IR spectrum of BTEE can be found in Figure S4) and the final ethane-bridged PMO material, as well as the calculated spectra of BTEE and the ethane-tetramer, are shown;

a

Vibrational Spectroscopy of PMOs

J. Phys. Chem. C, Vol. 111, No. 15, 2007 5657

Figure 8. (a) and (c) Experimental Raman (left side) and FT-IR (right side) spectra of the PMO precursor BTEE and the ethane-bridged PMO material. (b) and (d) Calculated Raman (left side) and IR spectra (right side) of BTEE and the ethane-tetramer model compound; the envelope curves were generated out of the stick spectrum by Lorentzian broadening with a fwhm value of 20 cm-1.

the experimental spectra of the bare ethane molecule were not recorded; the spectroscopic data can be found, for instance, in refs 28-30. First of all, the rather weak Raman intensities of BTEE and the ethane-bridged PMO material in the region of 400-

1700 cm-1 are noteworthy. Second, the Raman spectrum of the ethane-bridged PMO material shows interesting differences compared to the other investigated PMO samples below 700 cm-1. While the thiophene- and benzene-bridged PMO samples show two bands in this region (519 and 682 cm-1 for

5658 J. Phys. Chem. C, Vol. 111, No. 15, 2007

Figure 9. Geometry-optimized structure (ball-and-stick representation) of the cage-like linked ethane-tetramer model compound. Carbon, gray; hydrogen, white; silicon, orange; oxygen, red.

thiophene and 584 and 636 cm-1 for benzene) and the biphenylbridged PMO shows one sharp band at 627 cm-1, the ethanebridged PMO sample exhibits only one broad band at 516 cm-1, which is absent in the spectrum of the respective precursor compound BTEE. As a consequence, this means, in turn, that the Raman wavenumber region below 700 cm-1 cannot be used as an (empirical) indicator for intact Si-C bonds in organosilica materials, in general. The calculated Raman spectrum of BTEE and the ethanetetramer (the geometry-optimized structure is shown in Figure 9) are in excellent agreement with the experimental Raman spectrum of BTEE and the one of the ethane-bridged PMO material sample. With the calculated data of the tetramer at hand, it is easy to assign most of the remaining bands of the Raman spectrum of the ethane-bridged PMO sample between 400 and 1700 cm-1 (calculated values of the tetramer are given in parenthesis). The band at 515 cm-1 (535) is most likely caused by a coupled mode which is composed of (a) a tilting vibration of the ethylene unit (as a whole) against the surrounding silica framework (see also Figure 10) and of (nonlocal) Si-O-Si bending vibrations. Further assignments can be conducted as follows: 1407 cm-1 (1424) δ(CH2)wagg, 1267 cm-1 (1259) δ(CH2)twist, and 990 cm-1 (980) ν(C-C). The origin of the very weak and broad band at 797 cm-1 is not clear yet, but there are no indications that it is caused by any kind of ν(Si-C) vibrations. The tetrameric model cluster compound shows two Raman-active modes near 750 cm-1, but these modes are δ(SiH2)wagg vibrations and an artifact resulting from the construction of the finite model in which the bridging silicon atoms are saturated with hydrogen (see above). The FT-IR spectrum of the ethane-bridged PMO sample shows two bands at 699 and 775 cm-1, which were attributed to Si-C stretching vibrations by Dı´az-Morales et al.17, as already mentioned in the Introduction. Studying the output file of the Gaussian03 calculation for the BTEE revealed at least five modes, which involve considerable movements of the carbon

Hoffmann et al. against the neighboring silicon atoms, lying at calculated values of 805, 781, 748, 736, and 619 cm-1. These modes can be described as tilting vibrations (805 and 736 cm-1) and as translational oscillations (781, 748, and 619 cm-1) of the ethylene unit against the rest of the molecule (see Figure 10). As the modes at 781 and 745 cm-1 exhibit relatively strong IR activities (the enwrapping curve gives maxima at 773 and 748 cm-1; see Figure 8), we can tentatively confirm the assignment by Dı´az-Morales et al.17 for the band at 775 cm-1. Interestingly, while the theoretical calculation of the BTEE compound shows no mode between 731 and 647 cm-1, the frequency analysis of the ethane-dimer reveals an additional mode at 684 cm-1 with notable IR intensity, which is again a tilting Si-CH2-CH2-Si vibration. Also the ethane-tetramer shows two IR bands at 777 and 697 cm-1 which can, again, be correlated to the described tilting Si-CH2-CH2-Si vibrations. Thus, it is not unlikely that the assignment for the band 699 cm-1 made by Dı´az-Morales et al.17 could also be right. However, the situation is complicated by the fact that other IR-active modes could fall into the same wavenumber region and could give rise to additional bands, which may mask these “Si-C bands”, for instance, some of the Si-O-H bending modes. Note, a PMO/organosilica particle will always carry considerable amounts of silanol groups at the outer surface. For these reasons, the occurrence of two bands at the aforementioned positions in the FT-IR spectrum can be a hint for intact Si-C bonds but cannot seriously serve as a proof. What is more, a quantification of the ratio of cleaved and intact Si-C bonds is virtually impossible. Furthermore, it should be noted that all other investigated PMO samples in this work do not show these two bands at this position, which means, again, that every PMO/organosilica material should be regarded separately and that a general assignment of bands to Si-C vibrations is not possible. Concluding Remarks The presented analysis has revealed that bis-silylated PMO precursors exhibit no typical ν(Si-C) stretching modes. Two classes of modes of the precursors (or “dimers”) can be interpreted as (pseudo) ν(Si-C) vibrations. The first type are either δ(Si-(OR)3) umbrella vibrations or various kinds of δ(ring)/δ(CH)i.p. modes. The frequencies and IR/Raman activities of the umbrella modes only weakly depend on the specific type of the organic bridge R. However, the final PMO materials should not exhibit these umbrella vibrations at all, as ideally after formation of the PMO network, there should not be any isolated or terminal Si-OR3 groups present. The second class of ν(Si-C) vibrations, in which the silicon atoms perform compensation motions as a result of a collective motion or nodding-type vibration of the organic bridge against the remainder of the molecule/framework, are highly dependent on the local environment and on the symmetry of the Si-R-Si motif. They can only be present if the motif exhibits an inversion center. As a consequence, these modes are absent in BTET. Furthermore, different chemical environments will lead to shifts of the corresponding eigenfrequencies, if those modes will be eigenmodes of the compound/framework at all. Anyhow, these modes are not suitable for identification of the integrity of Si-C bonds because they fall in wavenumber regions (∼ 775 cm-1 for ethane and ∼ 1120 cm-1 for benzene/biphenyl) where already strong absorption due to Si-OH and Si-O-Si fragments, respectively, occurs. All modes of the investigated compounds, which can be interpreted as Si-C vibrations, are synoptically visualized as vector displacement sketches in Figure 10.

Vibrational Spectroscopy of PMOs

J. Phys. Chem. C, Vol. 111, No. 15, 2007 5659

Figure 10. Schematic vector displacement representation and calculated wavenumbers of the vibrational modes, which can be interpreted as Si-C vibrations. (a) Umbrella δ(Si-O3) vibration, (b) δ(CH)i.p. vibrations of BTEB and BTEBP, and (c) various Si-C modes of BTEE with varying Si-C stretching and Si-C bending amounts; for clarity reasons in (b) and (c), only the Si-R-Si fragment is shown. Carbon, gray; hydrogen, white; sulfur, yellow; silicon, orange; oxygen, red.

In summary, the widely held belief that it is possible to identify Si-C bond motifs in PMO materials with the help of vibrational spectroscopy is undermined. Instead, the following conclusions can be drawn from the results of this combined experimental and theoretical study: (i) An unequivocal assignment of IR/Raman bands to discrete ν(Si-C) stretching vibrations is almost impossible; a typical ν(Si-C) stretching mode does not exist and is not a characteristic group frequency. The occurrence and position of such modes/bands is highly dependent on the kind of organic bridge R and the local symmetry of the Si-R-Si motif. (ii) Upon the transition from the precursors to the respective PMO materials, wavenumber shifts and intensity changes of bands occurred, which are almost unpredictable. (iii) There is no simple analogy between various PMO materials with different organic bridges; each material is a “single case”. (iv) Vibrational spectroscopy is not a reliable method for monitoring the integrity of the Si-C bonds in PMO or related organosilica materials; therefore, for valid statements regarding the extent of Si-C bond cleavage, carrying out solid-state NMR spectroscopy is an absolute necessity. Acknowledgment. We thank Ju¨rgen Morell for the preparation of the precursors as well as the PMO materials. We are gratefully acknowledging financial support of the Optodynamic Research Center of the Philipps University Marburg and the Fonds der Chemischen Industrie. Supporting Information Available: Four additional figures: detailed spectral comparison between the experimental and calculated IR spectra of the PMO precursors BTET, BTEB, BTEBP, and BTEE. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Inagaki, S.; Guan, S.; Fukushima, Y.; Ohsuna, T.; Terasaki, O. J. Am. Chem Soc. 1999, 121, 9611. (2) Melde, B. J.; Holland, B. T.; Blanford, C. F.; Stein, A. Chem. Mater. 1999, 11, 3302.

(3) Asefa, T.; MacLachlan, M. J.; Coombs, N.; Ozin, G. A. Nature 1999, 402, 867. (4) Stein, A.; Melde, B. J.; Schroden, R. C. AdV. Mater. 2000, 12, 1403. (5) Sayari, A.; Hamoudi, S. Chem. Mater. 2001, 13, 3151. (6) Stein, A. AdV. Mater. 2003, 15, 763. (7) Hatton, B.; Landskron, K.; Whitnall, W.; Perovic, D.; Ozin, G. A. Acc. Chem. Res. 2005, 38, 305. (8) Hoffmann, F.; Cornelius, M.; Morell, J.; Fro¨ba, M. Angew. Chem., Int. Ed. 2006, 45, 3216. (9) Dag, O ¨ .; Ozin, G. A. AdV. Mater. 2001, 13, 1182. (10) Dag, O ¨ .; Yoshina-Ishii, C.; Asefa, T.; MacLachlan, M. J.; Grondey, H.; Coombs, N.; Ozin, G. A. AdV. Funct. Mater. 2001, 11, 213. (11) Temtsin, G.; Asefa, T.; Bittner, S.; Ozin, G. A. J. Mater. Chem. 2001, 11, 3202. (12) Burleigh, M. C.; Markowitz, M. A.; Spector, M. S.; Gaber, B. P. J. Phys. Chem. B 2001, 105, 9935. (13) Lee, B.; Im, H.-J.; Luo, H.; Hagaman, E. W.; Dai, S. Langmuir 2005, 21, 5372. (14) Wahab, M. A.; Kim, I.; Ha, C.-S. J. Solid State Chem. 2004, 177, 3439. (15) Rebbin, V.; Jakubowski, M.; Po¨tz, S.; Fro¨ba, M. Microporous Mesoporous Mater. 2004, 72, 99. (16) Morell, J.; Wolter, G.; Fro¨ba, M. Chem. Mater. 2005, 17, 804. (17) Dı´az-Morales, U.; Bellussi, G.; Carati, A.; Millini, R., Jr.; O’Neil Parker, W.; Rizzo, C. Microporous Mesoporous Mater. 2006, 87, 185. (18) Tanaka, S.; Kaihara, J.; Nishiyama, N.; Oku, Y.; Egashira, Y.; Ueyama, K. Langmuir 2004, 20, 3780. (19) Wahab, M. A.; Kim, I.; Ha, C.-S. Microporous Mesoporous Mater. 2004, 69, 19. (20) Yang, Q.; Liu, J.; Yang, J.; Zhang, L.; Feng, Z.; Zhang, J.; Li, C. Microporous Mesoporous Mater. 2005, 77, 257. (21) Morell, J.; Gu¨ngerich, M.; Wolter, G.; Jiao, J.; Hunger, M.; Klar, P. J.; Fro¨ba, M. J. Mater. Chem. 2006, 16, 2809. (22) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;

5660 J. Phys. Chem. C, Vol. 111, No. 15, 2007 Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (23) Zhurko, G. A.; Zhurko, D. A. ChemCraft; http://www.chemcraftprog.com/. (24) Dollish, F. R.; Fateley, W. G.; Bentley, F. F. Characteristic Raman Frequencies of Organic Compounds; John Wiley & Sons: New York, 1974. (25) Inagaki, S.; Guan, S.; Ohsuna, T.; Terasaki, T. Nature 2002, 416, 304. (26) Obeying a D2 symmetry, these two modes would fall together to give only one two-fold-degenerated mode.

Hoffmann et al. (27) It is to be expected that, in a totally symmetric “dimer”, only two modes would be present, each of which would be two-folddegenerated. (28) Herzberg, G. Molecular Spectra and Molecular Structure. II. Infrared and Raman Spectra of Polyatomic Molecules; Van Nostrand Reinhold: Princeton, 1945. (29) Van Helvoort, K.; Knippers, W.; Fantoni, R.; Stolte, S. Chem. Phys. 1987, 111, 445. (30) Pinkley, L. W.; Sethna, P. P.; Williams, D. J. Phys. Chem. 1978, 82, 1532.