21022
J. Phys. Chem. C 2009, 113, 21022–21028
ARTICLES Infrared and Computational Studies on Interactions of Carbon Dioxide and Titania Nanoparticles with Acetate Groups Ruohong Sui,†,‡ John M. H. Lo,‡ and Paul A. Charpentier*,† Department of Chemical and Biochemical Engineering, Faculty of Engineering, UniVersity of Western Ontario, London, Ontario, Canada N6A 5B9, and Department of Chemistry, UniVersity of Calgary, Calgary, Alberta, Canada T2N 1N4 ReceiVed: May 17, 2009; ReVised Manuscript ReceiVed: October 22, 2009
Understanding the nature of the interactions between growing colloidal nanoparticles and CO2 molecules is of importance for designing and synthesizing well-defined oxide nanoarchitectures during sol-gel processing in supercritical CO2 (scCO2). In this research, attenuated total reflective Fourier transform infrared (ATRFTIR) spectrometry was used for studying the interactions between CO2 molecules and metal acetate bidentate groups. Freshly synthesized aerogels formed using Ti and Zr alkoxides were treated with CO2 and subsequently depressurized to remove the strong infrared absorbance of bulk CO2. The IR spectra from the depressurized TiO2 and ZrO2 samples were compared to the IR spectra from the respective calcined samples showing that the ν2 bending peak of the CO2 split due to Lewis acid and base interactions. Further verification of the spectral assignments and the bonding analysis were obtained by theoretical calculations using the density functional theory (DFT) method and TiO2 nanocluster models. This research demonstrates that the metal acetate group on the Ti and Zr polycondensates is CO2-philic, and therefore these colloidal particles can be stabilized in scCO2. Introduction Supercritical CO2 (scCO2) is an attractive alternative to conventional organic solvents due to its unique features of tunable physical properties and environmental benignness.1 As a result, scCO2 has been widely used for chemical processes, e.g., extraction and separation, chemical reactions, and material processing.2–7 Recently, direct sol-gel reactions in scCO2 have attracted much attention for synthesizing oxide nanomaterials. As examples, SiO2 monolithic aerogels and nanoparticles have been synthesized by the reaction of silicon alkoxides with formic/acetic acid;8–10 TiO2 and ZrO2 nanofibers, nanospheres, and mesoporous monoliths have been produced by polycondensation of metal alkoxides with either acetic acid or water droplets with the aid of surfactants.5,11–15 This approach for synthesizing oxide nanomaterials has many advantages over the conventional sol-gel procedure carried out using organic solvents. These advantages include that the resulting materials maintain nano features and a high surface area after CO2 venting,11 the solvent strength of scCO2 is tunable when temperature and/or pressure is changed providing for tunable particle size,1 and CO2 is an inexpensive, flammable, and sustainable reaction media.1,16 It is well-known that metal alkoxides are sensitive to water and readily form precipitates.17 From the successful synthesis of these oxide nanomaterials in scCO2, it can be rationalized that (1) the polycondensation reaction rates are well controlled, * To whom correspondence should be addressed. Phone: (519) 661-3466. Fax: (519) 661-3498. E-mail:
[email protected]. † University of Western Ontario. ‡ University of Calgary.
and (2) the colloidal particles were stabilized in the scCO2 media. In order to control the reaction rates, two approaches have been used for synthesizing TiO2 nanomaterials. In the first method, nonaqueous organic acids instead of water were used as polycondensation agents of titanium alkoxides,18 and it was found that acetic acid was a mild reaction agent due to the formation of Ti-acetate complexes.19 In the second method, the reverse micelles formed in scCO2 effectively decreased the sol-gel reaction rate.13 However, none of these studies have theoretically examined how the colloidal particles are stabilized in scCO2. Understanding the nature of the interactions and controlling the solubility between macromolecules or colloidal particles and CO2 is of importance for designing, synthesizing, or processing materials with well-defined nanoarchitectures. Due to a lack of polarity and a dipole moment, scCO2 is a poor solvent for most polar solutes and macromolecules.20 However, CO2 has a large quadrupole moment and a polar CdO bond, making a variety of materials with carbonyl or fluoride groups soluble in scCO2.21 The interaction between CO2 and small molecules, polymers, and organometallic compounds has been studied by ab initio calculations22,23 and experimental Fourier transform infrared (FTIR),21,24,25 Raman,3,26,27 and NMR spectrometries.28,29 CO2 may behave as both an electron acceptor and an electron donor.23,30 For example, the interaction of CO2 and the carbonyl functional group has been found to consist of a Lewis acid and Lewis base attraction (LA-LB) between the partial positive carbon of CO2 and the lone pairs of the carbonyl oxygen.24 As well, the C-H · · · O hydrogen bond between CO2 and the LB
10.1021/jp904599t CCC: $40.75 2009 American Chemical Society Published on Web 11/19/2009
Interactions between CO2 and Metal Acetate Groups
J. Phys. Chem. C, Vol. 113, No. 50, 2009 21023
Figure 1. Schematic drawing of possible LA-LB interactions between CO2 and the (a) bridging acetate bidentate, (b) chelating acetate bidentate, and (c) acetate monodentate. M ) Si, Ti, or Zr.
site has been attributed to the high solubility of polymers with a carbonyl moiety in CO2.30 In our previous studies on the synthesis of SiO2, TiO2, and ZrO2 nanoarchitectures in scCO2, in situ FTIR spectra were used to study the mechanism and kinetics of the sol-gel processes.10,12,19,31 The in situ FTIR spectra showed the formation of M-acetate bidentates (M ) Si, Ti, and Zr) during the alkoxides reacting with acetic acid to form nanostructures. Similar to the carbonyl group, the acetate bidentate structure may donate electrons to CO2, resulting in LA-LB interactions, as shown in Figure 1. Because of the high intensity of bulk CO2, and interference from the organic species, it was not possible to directly observe LA-LB interactions between CO2 and the M-acetate groups through the in situ IR spectra during the sol-gel process. However, previous work by Johnston et al. has shown that the kinetics of interactions of CO2 with silica could be elucidated after depressurization.32 Hence, our strategy in this work was to use attenuated total reflective Fourier transform infrared (ATR-FTIR) spectrometry to study the CO2-impregnated TiO2 and ZrO2 sol-gel products under a variety of temperatures and pressures, and then depressurize. For comparison purposes, we have also examined the ATR-FTIR spectra of CO2-impregnated aerogels after calcination, in which any acetate functional groups would be removed. To facilitate the spectral assignments and reveal the bonding interactions between the aerogels and CO2 molecules, density functional theory (DFT) calculations were performed. Experimental Methods In Situ ATR-FTIR. In situ FTIR monitoring of the sol-gel reaction in scCO2 was performed using a high-pressure diamond immersion probe (Sentinel-ASI Applied Systems) attached to a stirred 100 mL autoclave (Parr Instruments). The probe is attached to an ATR-FTIR spectrometer (ASI Applied System ReactIR 4000), connected to a computer, supported by ReactIR software (ASI). The setup was described in detail previously.10 Samples of 1.10 M titanium isopropoxide (TIP) or 1.13 M zirconium butoxide (ZBO) and 4.95 or 2.52 M acetic acid, respectively, were quickly placed in the autoclave, followed by pumping in CO2 and heating to 313-333 K at 31.03 MPa under stirring at 600 rpm. The spectra were collected throughout the reaction at specified intervals. Synthesis of TiO2 and ZrO2 Aerogels. The TiO2 and ZrO2 aerogel particles (denoted as a-TiO2 and a-ZrO2, respectively) were synthesized using the methods previously described.11,12,19 a-TiO2-A and a-TiO2-B aerogels were synthesized using 1.10 M TIP and 1.10 M titanium(IV) butoxide (TBO) with 4.95 and 6.05 M acetic acid, respectively, at 333 K and 31.03 MPa of scCO2. a-ZrO2 was synthesized using 1.13 M ZBO reacting with 2.52 M acetic acid at 313 K and 31.03 MPa of scCO2. Since
Figure 2. Schematic drawing of ATR-FTIR analysis. The aerogel powder was pressed against the diamond mirror to obtain higher absorbance peaks.
aerogels are well-known to adsorb significant quantities of water or organic solvents that also could interact with CO2, special attention was made to dry the aerogels before treating with scCO2. scCO2 drying and vacuum drying at 393 K and 0.8 Pa were conducted until the IR spectrum showed no traces of water peaks at 3000-3600 cm-1. Samples of the dried aerogels were calcined at 773 K to obtain oxides. ATR-FTIR Analysis of CO2-Impregnated Aerogels. The dried aerogels before and after calcination were studied using ATR-FTIR spectroscopy. Since ATR-FTIR can only detect materials a few micrometers from the mirror,24 the absorbance peaks of the powder are low. In order to obtain significant peaks, the aerogel powder was pressed against the diamond mirror by using the agitation shaft with a rubber disk in between (Figure 2). The spectra of the pure aerogels were collected before addition of CO2. Then CO2 was added while the temperature was kept at 313 K and pressures kept at 3.45, 6.89, 13.79, and 27.58 MPa, respectively, and the IR spectra were collected under different conditions. Because of the strong absorbance of bulk CO2, which prevents clear observation of CO2 interacting with the aerogel under high concentration of CO2, FTIR spectra were also collected 1 min after venting of CO2 from the autoclave when adsorbed CO2 was still present. FTIR spectra were collected from 600 to 4000 cm-1 with a resolution of 2 cm-1. The subtraction and curve fitting of the IR spectra were obtained using ACD UVIR Processor version 7.0 software (ACD Inc. Toronto, Ontario). The curve fitting was processed by assuming a Gaussian peak profile, and the limit of the half-peak width was set as 80 cm-1 for a better curve fit. Computational Details The molecular modeling of nanofibers is difficult because of the large number of atoms involved. In order to simplify the
21024
J. Phys. Chem. C, Vol. 113, No. 50, 2009
Sui et al.
Figure 3. In situ IR spectra: (a) 1.3 M tetramethyl orthosilicate (TMOS) reacting with 5.3 M HOAc in CO2 at a reaction time of 360 min, at 50 °C and 27.58 MPa; (b) 1.10 M TIP reacting with 4.95 M HOAc in CO2 at a reaction time of 40 min, at 60 °C and 31.03 MPa; and (c) 1.13 M ZBO reacting with 2.52 M HOAc in CO2 at a reaction time of 40 min, at 40 °C and 31.03 MPa. The spectra were offset for better observation.
calculations, smaller TiO2 clusters were employed to mimic the TiO2 aerogel particles synthesized in this work. Two types of model clusters were built based on the previously reported crystal structure of [Ti6(C2H3O2)6(C3H7O)6O6] which was synthesized by the same methods utilized in this work.33 The first model, denoted by TiO2-m, was constructed by replacing the isopropoxyl (i-PrO) groups in [Ti6(C2H3O2)6(C3H7O)6O6] with hydroxyl (OH) groups. Note that the TiO2 monomer structurally resembles TiO2 aerogel clusters, whose TiO2 units are connected by oxygen linkages. There was another model, denoted by TiO2d, developed in which two TiO2 clusters are bridged by Ti-O-Ti bonds and the terminal O’s are capped by hydrogen. The oxygen linkages between TiO2 clusters provide another set of nucleophilic sites on which CO2 can be adsorbed. The geometry optimizations of the model compounds were performed using a plane-wave-based density functional theory method implemented in the Vienna Ab-initio Simulation Package (VASP),34–36 in which the PBE functional37 was employed in conjunction with the projector-augmented wave (PAW) method38–40 that describes the interaction between ions and electrons. The plane-wave (PW) basis sets with the energy cutoff of 400 eV were used to expand the one-electron orbitals. The k-point sample was done only at the Γ-point because of the relatively large system size. The conjugate-gradient (CG) algorithm40 was used in the optimizations of ionic positions, and a smearing function of 0.15 eV was used to facilitate the electronic convergence. A geometry optimization was converged if the energy change was smaller than 1 × 10-4 eV. The molecules were located inside a 20 × 20 × 20 Å3 chamber, and dipole correction was always imposed to minimize the dipole-dipole interactions between image molecules. For comparison purposes, some of the geometry optimizations were also repeated using the quantum chemical package deMon2k utilizing the conventional basis set method in density functional theory framework.41 In these calculations, the Perdew-Burke-Erzenrhof (PBE) exchange-correlation functional in the generalized-gradient approximation was used.37 The Stuttgart-Dresden relativistic effective core potentials (SDRECP) were employed for Ti,42 O,43 and C43 atoms while a double-ζ valence plus polarization (DZVP) basis set was assigned for H atoms.44 The 3p, 3d, and 4s electrons of Ti were treated as valence electrons. The spd-type auxiliary function
sets (Gen-A2) were used to calculate the Coulomb repulsion energy.45 The default convergence criteria for geometry and energy minimizations (1 × 10-4 and 1 × 10-5 au, respectively) were employed. In all frequency calculations, the numerical Hessians were computed by means of double finite difference method with a step size of 0.015 Å. Results and Discussion In Situ FTIR. Figure 3 shows the in situ FTIR spectra during the alkoxides reacting with acetic acid, in which the absorbance peaks in the region of 1400-1600 cm-1 indicate the formation of M-acetate bidentates (M ) Si, Ti, and Zr).46 As previously shown in Figure 1, the M-acetate bidentates have the potential to interact with CO2 molecules during the sol-gel process in scCO2. However, it was found difficult during synthesis when using the in situ IR results to observe the bidentate interacting with CO2, due to the very high absorbance of bulk CO2 in the autoclave, and the possible interference from organic species. FTIR Spectra in Pressurized CO2. In order to examine the interactions between the M-acetates and CO2, the IR spectra of CO2-impregnated aerogels was measured under various temperatures and pressures. Figure 4 shows TiO2-A in pressurized CO2 under pressures of 3.45-27.58 MPa at 40 °C. In the spectra, the peaks in the region of 1400-1600 cm-1 are attributed to M-acetate bidentate,46 that at 1345 cm-1 is attributed to hydrocarbon groups, that at 1119 cm-1 is attributed to M-O-C, that at 1027 cm-1 is attributed to bridging OR groups,47 and those lower than 900 cm-1 are attributed to oxo bonds, except the peak at approximately 659 cm-1, which belongs to CO2. In this figure, the CO2 peak is very high, obscuring the CO2-aerogel interactions. This phenomenon was attributed to the high concentration of free CO2 molecules in the porous aerogels close to the IR probe. In order to solve this problem, Kazarian et al. cast or hot pressed polymers onto the IR mirror.24 Unfortunately, unlike polymer materials, the fresh oxide aerogels cannot be dissolved or melted without destruction of the acetate functional groups. FTIR Spectra after CO2 Venting. In order to remove the interruption of bulk CO2, ATR-FTIR spectra were also collected immediately after CO2 venting when CO2 molecules were still
Interactions between CO2 and Metal Acetate Groups
J. Phys. Chem. C, Vol. 113, No. 50, 2009 21025
Figure 4. ATR-FTIR spectra of TiO2-A in pressurized CO2: (a) 3.45, (b) 6.89, (c) 13.79, and (d) 27.58 MPa, at a temperature of 40 °C. The spectra were offset for better observation.
Figure 5. ATR-FTIR spectra of (a) a-TiO2-A, (b) a-TiO2-B, and (c) a-ZrO2 before addition of CO2 (black) and at 1 min after venting of CO2 (red).
present in the aerogel matrix. Figure 5 shows the spectra of both the pure and CO2-impregnated aerogels of a-TiO2-A, a-TiO2-B, and a-ZrO2 (prefix “a-” denotes aerogel). In the spectra of CO2-impregnated a-TiO2-A, a-TiO2-B, and a-ZrO2, besides the peaks from the pure aerogels in the region of 680-1680 cm-1, there are new peaks at 668, 654, 649, and 648 cm-1. The very sharp peak at 668 cm-1 is due to free CO2, while the relatively wide peaks in the 642-654 cm-1 region are from the overlapping results of ν2 bending vibration of the impregnated CO2 and vibration of the oxo bonds.24 From Figure 5, there is no significant shift of chelating acetate peak in the region 1400-1600 cm-1 with and without CO2, unlike the case of carbonyl group that has a 4 cm-1 shift.24 This is
Figure 6. IR spectra and curve-fitting results of CO2 interacting with (a) a-TiO2-A, residual sum of squares ) 3.86 × 10-3; (b) a-TiO2-B, residual sum of squares ) 4.01 × 10-4; and (c) a-ZrO2, residual sum of squares ) 1.81 × 10-3. Black curves ) experimental spectra; pink curve ) fitting curve; blue curves ) individual Gaussian function curves; red curve ) residual curve.
likely due to the chelating acetate group being more rigid than the carbonyl group during vibration. In order to examine the vibration wavenumbers of the impregnated CO2, subtraction of the corresponding aerogel
21026
J. Phys. Chem. C, Vol. 113, No. 50, 2009
Sui et al. TABLE 1: Calculated Binding Energies (in kcal/mol) and CO2 Bending Modes (in cm-1) of the Model Compounds (M1-M5)a PBE/PAW
PBE/DZVP/SD-RECP
model
BE
frequency
BE
frequency
M1 M2 M3 M4 M5
2.49 1.44 3.88 2.29 1.58
657, 664, 665, 632, 642,
7.42 3.72 6.55 -
580, 566 -
650 655 651 618 616
a
No PBE/DZVP/SD-RECP data are available for M4 and M5 because of the failure of convergence in geometry optimization.
Figure 7. IR spectra of CO2 impregnated into (a) a-TiO2-A, (b) a-TiO2B, and (c) a-ZrO2 in the ν3 stretching mode region, at 1 min after venting of CO2; (d) CO2 impregnated into a-TiO2-A at 5 min after venting of CO2.
spectrum is necessary. Figure 6 presents the spectra of CO2 impregnated in the aerogels after the subtraction and the curvefitting results. The peaks in the region of 630-665 cm-1 were attributed to CO2 ν2 vibration splitting. Understanding the complex peaks requires a careful consideration of the electronic structures of metal acetate and the nature of the CO2 ν2 vibration. Figure 7 shows the IR spectra of CO2 impregnated into a-TiO2-A, a-TiO2-B, and a-ZrO2 in the ν3 stretching mode region. These spectra show a high peak at 2337 cm-1 and an asymmetric wide peak around 2358 cm-1. The peak at 2337 cm-1 is assigned to the free CO2, while the small right shoulder of the peak at 2337 cm-1 can be assigned to the (ν2 + ν3) - ν2
band.24 The peak at 2358 cm-1 is due to the interaction between CO2 and the aerogels that caused the CO2 ν3 stretching mode split.24 When we compare Figure 7a and 7d, the free CO2 peak at 2337 cm-1 decreased faster than the peak at 2358 cm-1, suggesting that the free CO2 still interacts with the aerogel. Molecular Modeling. The fully optimized structures of TiO2-m and TiO2-d are given in Figure 8. As shown, dimerization does not alter significantly the geometry of the TiO2 cores; a noticeable change is the contraction of Ti-Ti bonds by about 0.1 Å. The bridging Ti-O bonds are about 1.84 Å with the Ti-O-Ti angles from 171° to 173°. The short Ti-O bonds compared to those in rutile and anatase (1.93-1.98 Å) suggest that there exists a π-type interaction between the Ti and bridging O atoms. Interestingly, both the Gaussian basis set and plane-wave calculations yielded similar optimized geometries with an average bond length discrepancy of about 0.02 Å. It has been illustrated in Figure 1a that both the π-electrons of the acetate ligands and the lone pair of electrons residing on the acetate O atoms can contribute to the LA-LB interactions with the CO2 molecule. In addition, the CO2 molecule can be bound simultaneously to two adjacent acetate ligands via the π-interaction. To distinguish these possible adsorption modes, three CO2 binding models were constructed: (1) the M1 model consists of the CO2 molecule lying parallel to the plane of an acetate ligand, with the π-electrons of the acetate ligand donated to the antibonding orbital of CO2; (2) the M2 model corresponds to the CO2 molecule bucketed by two acetate ligands; (3) the
Figure 8. Most stable optimized geometries of CO2 on TiO2 monomer and TiO2 dimer models: (a) M3 configuration; (b) M4 configuration.
Interactions between CO2 and Metal Acetate Groups
J. Phys. Chem. C, Vol. 113, No. 50, 2009 21027
Figure 9. Contour plots of the bonding and antibonding orbitals associated with CO2 and the TiO2-m and TiO2-d models: (a) bonding MO of M3; (b) antibonding MO of M3; (c) and (d) bonding MOs of M4; (e) and (f) antibonding MOs of M4.
M3 model consists of the CO2 molecule bound laterally to the lone pair of the acetate O atom, forming a bent T-shape configuration. Apart from the possible adsorption modes associated with the chelating acetate ligands, the CO2 molecule can also bind to the O atoms that link the two TiO2 core units, as these O atoms possess accessible lone pairs of electrons. Depending on the orientations of the CO2 molecule, we developed the M4 (CO2 perpendicular to the Ti-O-Ti linkage) and M5 (CO2 parallel to the Ti-O-Ti linkage) adsorption models, respectively. The calculated binding energies of CO2 in the adsorption modes M1-M5 are summarized in Table 1. These quantities are generally very small and are similar to the SCF and MP2 values reported for the systems of CO2 coordinated with various carbonyl groups.23 The small adsorption energies in conjunction with the long bond distances (∼3.0 Å) imply the weak LA-LB interaction and the large steric repulsion due to the bulky acetate substituents (Figure 9). Interestingly, the PBE/DZVP/SD-RECP calculations consistently yielded higher binding energies than those obtained in the PBE/PAW calculations. A possible reason that accounts for this difference is the presence of the basis set superposition error (BSSE) in the former approach.48 Furthermore, the PBE/DZVP/SD-RECP calculations predicted that the M1 model is more stable than M3, in contrast to the results of the PBE/PAW calculations. Table 1 also shows the CO2 bending (ν2) frequencies for M1-M5. The accuracy of frequency calculations using the two approaches has been tested on an isolated CO2 molecule. While PBE/DZVP/SD-RECP calculations yielded the CdO bond lengths (1.191 Å) close to the experimental value, the predicted frequencies (2246, 606 cm-1) are substantially lower than the observed frequencies and smaller than those predicted using the PBE/PAW method (2358, 624 cm-1). Therefore, the PBE/ DZVP/SD-RECP frequency calculations were performed only for the most stable M3 model. From our IR experimental results, there are two distinct peaks from a-TiO2-A associated CO2 (663 and 655 cm-1). Based on the data shown in Table 1, we can assign these peaks to the nondegenerate bending frequencies of the CO2 molecule in the M1 adsorption mode, which were predicted to be split by 7 cm-1. The doublet may also arise from the CO2 molecule
embraced by two adjacent acetate groups (M2); the estimated band splitting of 9 cm-1 agrees well with the experimental value but this configuration is less energetically favorable than the M1 mode. In the case of a-TiO2-B associated CO2, however, the experimental IR spectrum consists of a wider combined peak which could be deconvoluted into four individual peaks (Figure 6). These bands can be assigned respectively to the CO2 molecule in M1 (641, 650 cm-1), M2 (650, 658 cm-1), and M3 (641, 658 cm-1) adsorption modes, despite the slightly overestimated theoretically predicted frequencies. The PAW/DZVP/ SD-RECP calculations, on the other hand, underestimated the frequencies by about 80 cm-1, yet they correctly predicted the magnitude of splitting of the two CO2 bending modes (14 cm-1). The noticeable difference between the IR spectra of CO2 impregnated into a-TiO2-A and a-TiO2-B can be explained in terms of their nanostructures. SEM studies19 previously revealed that a-TiO2-A consists of curled nanofibers of about 10 nm diameter. This morphology favors the M1 and M2 adsorption modes, where the CO2 molecule is enclosed by the acetate ligands. The M3 mode is not favored since the CO2 molecule in the more spatially bulky T-shape configuration experiences a substantial steric interaction with the neighboring acetate groups on the curled nanofiber. Unlike a-TiO2-A, a-TiO2-B, which has a fiber diameter of about 80-100 nm, consists of straight nanofibers which are more regularly oriented. The organized configuration of nanofibers gives rise to a larger total surface area and affords an effective interstitial adsorption of CO2 on the acetate ligands (M1, M2, and M3) or oxygen linkages (M4 and M5). As a consequence, all three adsorption modes associated with the acetate ligands were observed. The absence of the absorption bands corresponding to M4 and M5 in the experimental IR spectra may be attributed to the bands falling in the region beyond the lower limit of the scan window of the ATR-FTIR spectrometer, thus making the bands not observable in the present work. In addition, the weaker adsorption of M4 and M5 compared to M1 and M3 also suggests that only a very small portion of CO2 is adsorbed onto these sites, and therefore leads to the extremely low intensity IR bands which may have been hindered by the noise.
21028
J. Phys. Chem. C, Vol. 113, No. 50, 2009
Conclusions For the first time, the interaction of metal acetate and CO2 has been studied using ATR-FTIR spectra and DFT calculations. According to the FTIR study, it can be concluded that a LA-LB interaction exists between CO2 and the M-acetate groups. The electronic structure calculations provide supporting evidence for this proposition, and suggest five possible association modes (M1-M5). Among the various adsorption modes, CO2 is found to be preferentially adsorbed in bent T-shape configuration (M3) to the acetate ligand via the LA-LB interaction with the oxygen lone pair; the adsorption enthalpy of 4 kcal/mol is close to the values previously reported for the systems of CO2 and other carbonyls and bases. However, the site preference is apparently subjected to spatial constraints and the three-dimensional morphology of the TiO2 nanofibers; for instance, the parallel configuration of CO2 (M5) in the dimer model is less favorable than the perpendicular configuration (M4). Despite the fact that the exact modes of adsorption could not be discerned through the experimental and theoretical results, the present study is able to facilitate a better understanding of the formation of the nanostructures during the sol-gel process in scCO2. Future work may be carried out to study the stabilization and growth of the colloidal particle size by using in situ synchrotron techniques.49 Acknowledgment. This work was financially supported by the Canadian Natural Science and Engineering Research Council (NSERC), the Materials and Manufacturing Ontario Emerging Materials program (MMO-EMK), the Canadian Foundation for Innovation (CFI), and the UWO Academic Development Fund (ADF). The DFT calculations presented in this work were performed on the Western Canada Research Grid. Professors Dennis Salahub and Tom Ziegler are gratefully acknowledged for providing the authors with the deMon2k and VASP packages. References and Notes (1) Johnston, K. P.; Shah, P. S. Science 2004, 303, 482. (2) Supercritical fluid technology in materials science and engineering: syntheses, properties, and applications; Sun, Y.-P., ed.; Marcel Dekker: New York, 2002; p ix. (3) Kachi, Y.; Tsukahara, T.; Kayaki, Y.; Ikariya, T.; Sato, J.; Ikeda, Y. J. Supercrit. Fluids 2007, 40, 20. (4) Levit, N.; Tepper, G. J. Supercrit. Fluids 2004, 31, 329. (5) Tadros, M. E. A.; Carol, L. J.; Russick, E. M.; Youngman, M. P. J. Supercrit. Fluids 1996, 9, 172. (6) Cooper, A. I. AdV. Mater. 2001, 13, 1111. (7) Wood, J. A.; Bernards, M. A.; Wan, W.; Charpentier, P. A. J. Supercrit. Fluids 2006, 39, 40. (8) Loy, D. A.; Russick, E. M.; Yamanaka, S. A.; Baugher, B. M. Chem. Mater. 1997, 9, 2264. (9) Moner-Girona, M.; Roig, A.; Molins, E. J. Sol-Gel Sci. Technol. 2003, 26, 645. (10) Sui, R.; Rizkalla, A. S.; Charpentier, P. A. J. Phys. Chem. B 2004, 108, 11886. (11) Sui, R.; Rizkalla, A. S.; Charpentier, P. A. Langmuir 2005, 21, 6150.
Sui et al. (12) Sui, R.; Rizkalla, A. S.; Charpentier, P. A. Langmuir 2006, 22, 4390. (13) Lim, K. T.; Hwang, H. S.; Ryoo, W.; Johnston, K. P. Langmuir 2004, 20, 2466. (14) Stallings, W. E.; Lamb, H. H. Langmuir 2003, 19, 2989. (15) Hong, S.-S.; Lee, M. S.; Lee, G.-D.; Lim, K. T.; Ha, B.-J. Mater. Lett. 2003, 57, 2975. (16) Sheldon, R. A. Green Chem. 2005, 7, 267. (17) Sanchez, C.; Livage, J.; Henry, M.; Babonneau, F. J. Non-Cryst. Solids 1988, 100, 65. (18) Sui, R.; Thangadurai, V.; Berlinguette, C. P. Chem. Mater. 2008, 20, 7022. (19) Sui, R.; Rizkalla, A. S.; Charpentier, P. A. J. Phys. Chem. B 2006, 110, 16212. (20) McHugh, M. A.; Krukonis, V. J. Supercritical Fluids. In Encyclopedia of Polymer Science and Engineering, 2nd ed.; Mark, H. F., Bikales, N., Overberger, C. G., Menges, G., Kroschwitz J. I., Eds.; John Wiley & Sons, Inc.: New York, 1989; Vol. 16. (21) Meredith, J. C.; Johnston, K. P.; Seminario, J. M.; Kazarian, S. G.; Eckert, C. A. J. Phys. Chem. 1996, 100, 10837. (22) Blatchford, M. A.; Raveendran, P.; Wallen, S. L. J. Phys. Chem. A 2003, 107, 10311. (23) Nelson, M. R.; Borkman, R. F. J. Phys. Chem. A 1998, 102, 7860. (24) Kazarian, S. G.; Vincent, M. F.; Bright, F. V.; Liotta, C. L.; Eckert, C. A. J. Am. Chem. Soc. 1996, 118, 1729. (25) Dobrowolski, J. C.; Jamroz, M. H. J. Mol. Struct. 1992, 275, 211. (26) Blatchford, M. A.; Raveendran, P.; Wallen, S. L. J. Am. Chem. Soc. 2002, 124, 14818. (27) Kazarian, S. G.; Martirosyan, G. G. Int. J. Pharm. 2002, 232, 81. (28) Kanakubo, M.; Aizawa, T.; Kawakami, T.; Sato, O.; Ikushima, Y.; Hatakeba, K.; Saito, N. J. Phys. Chem. B 2000, 104, 2749. (29) Yonker, C. R.; Palmer, B. J. J. Phys. Chem. A 2001, 105, 308. (30) Raveendran, P.; Wallen, S. L. J. Am. Chem. Soc. 2002, 124, 12590. (31) Sui, R.; Rizkalla, A. S.; Charpentier, P. A. Cryst. Growth Des. 2008, 8, 3024. (32) Dickson, J. L.; Gupta, G.; Horozov, T. S.; Binks, B. P.; Johnston, K. P. Langmuir 2006, 22, 2161. (33) Sui, R.; Charpentier, P. A.; Rizkalla, A. S.; Jennings, M. C. Acta Crystallogr., Sect. E: Struct. Rep. Online 2006, E62, m373. (34) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 48, 13115. (35) Kresse, G.; Furthmuller, J. Comput. Mater. Sci. 1996, 6, 15. (36) Kresse, G.; Furthmuller, J. Phys. ReV. B 1996, 54, 11169. (37) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865. (38) Blochl, P. E. Phys. ReV. B 1994, 50, 17953. (39) Kresse, G.; Joubert, D. Phys. ReV. B 1999, 59, 1758. (40) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes: The Art of Scientific Computing, 3rd ed.; Cambridge University Press: New York, 2007. (41) Koster, A. K.; Flores-Moreno, R.; Geudtner, G.; Goursot, A.; Heine, T.; Reveles, J. U.; Vela, A.; Salahub, D. R. deMon 2003; NRC: Ottawa, Canada. (42) Dolg, M.; Wedig, U.; Stoll, H.; Preuss, H. J. Chem. Phys. 1987, 86, 866. (43) Igel-Mann, G.; Stoll, H.; Preuss, H. Mol. Phys. 1988, 65, 1321. (44) Godbout, N.; Salahub, D. R.; Andzelm, J.; Wimmer, E. Can. J. Phys. 1992, 70, 560. (45) Andzelm, J.; Radzio, E.; Salahub, D. R. J. Comput. Chem. 1985, 6, 520. (46) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds; Wiley-Interscience: New York, 1997; p 59. (47) Perrin, F. X.; Nguyen, V.; Vernet, J. L. J. Sol-Gel Sci. Technol. 2003, 28, 205. (48) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553. (49) Jensen, H.; Bremholm, M.; Nielsen, R. P.; Joensen, K. D.; Pedersen, J. S.; Birkedal, H.; Chen, Y.-S.; Almer, J.; Sogaard, E. G.; Iversen, S. B.; Iversen, B. B. Angew. Chem., Int. Ed. 2007, 46, 1113.
JP904599T
