Hydrolysis and Complexation of N,N-Dimethylformamide in New

Jun 6, 2011 - Degradation of Highly Alloyed Metal Halide Perovskite Precursor Inks: Mechanism and Storage Solutions. Benjia DouLance M. WheelerJeffrey...
1 downloads 3 Views 1MB Size
ARTICLE pubs.acs.org/JPCC

Hydrolysis and Complexation of N,N-Dimethylformamide in New Nanostructurated Titanium Oxide Hybrid OrganicInorganic Sols and Gel Thomas Cottineau,†,‡ Mireille Richard-Plouet,† Jean-Yves Mevellec,† and Luc Brohan*,† † ‡

Institut des Materiaux Jean Rouxel, Universite de Nantes, CNRS, 2, rue de la Houssiniere, 44322, Nantes, France Laboratoire des Materiaux Surfaces et Procedes pour la Catalyse, Universite de Strasbourg, CNRS, 25 rue Becquerel, 67087, Strasbourg, France

bS Supporting Information ABSTRACT: We recently synthesized new hybrid organic inorganic sols and gel obtained by the slow hydrolysis of N,Ndimethylformamide (DMF) by TiOCl2 acidic solution. The comprehension of the synthesis mechanism of these materials, made up of TiO framework surrounded by organic molecules, is important to understand their original photochemical properties. The use of Raman spectroscopy provides information on the role of DMF for the complexation of Ti4þ and its subsequent hydrolysis in dimethylammonium chloride (DMACl) and methanoate species during the nanostructuration. The study of the reaction by spectroscopies gives key information on the solsgel formation. Part of DMF is bound to Ti4þ since the beginning of the reaction inducing an oriented growth of the inorganic network. Furthermore, the selective hydrolysis of DMF linked to titanium allows the formation of methanoate species located in the Ti4þ coordination sphere. These surface-attached molecules appear to be of great importance to form the structure of these materials. Moreover, they are involved in the original photoelectrochemical properties of these sols and gel, which can usefully be employed for solar energy conversion and storage.

1. INTRODUCTION The use of TiO2 in various photochemical applications, such as photocatalysis,1 photoelectrochemical solar cells (for the production of hydrogen2 and electricity3), appears attractive for green technologies.4 However, because of a bandgap of 3.2 eV, a maximum of only 5% of the solar energy can be absorbed. Recent developments in nanotechnology offer key ingredients for new energy conversion and storage solutions with a higher efficiency. The enhanced surface area to volume ratio of nanostructured materials leads to singular photoelectrochemical properties that are drastically different from their bulk counterparts. One major change is the predominance of surface states, involved in redox reactions, leading to enhanced photochemical effects.5 Most chemical routes to TiO2 nanomaterials are based on transformations in solution such as solgel processing, hydro- or solvothermal synthesis or metal organic decomposition. Recently, hybrid organicinorganic titanium oxide-based sols and gels with outstanding quantum efficiency of charge separation equal to 46% where developed but, if a general formula (TiOa(OH)b(OR)42ab) is proposed, the crystallographic arrangement of the inorganic network and the interface between the organic and inorganic parts remains misunderstood.5,7 Despite the main advantages of solgel techniques for the preparation of nanomaterials (low temperature processing, homogeneity, versatility, etc.),811 the hydrolytic condensation reactions leading to the 3D framework r 2011 American Chemical Society

are difficult to characterize due to their kinetic control. In conventional solgel methods, the nucleation and growth processes are never separated into two steps, due to very fast hydrolysis and condensation of titanium precursors. Because these sol or gel materials consists of two phases, a liquid phase in which inorganic nanoparticles are dispersed, several analytical methods have to be specifically carried out for their characterization. Starting from the TiOCl2 precursor in an acid aqueous solution, we have shown that the control of polycondensation at the various stages of the synthesis allows us to obtain sols, gel, or solids for which the TiO networks dimensionality is welldefined as 0D, 1D, 2D, and 3D.12 In particular, by controlling the condensation of titanium species in N,N-dimethylformamide (noted DMF), we synthesized novel hybrid organicinorganic sols and gel based on titanium oxide (hereafter noted TiDMF) having original photochemical properties due to an excellent charge separation allowing the reduction of Ti4þ in Ti3þ in large amounts (approximately 10%) under irradiation.13,14 We assume this effect is not only due to the high specific area of the particles but also to the presence of organic groups strongly interacting with the inorganic surface. The comprehension of the hydrolysis Received: February 25, 2011 Revised: April 14, 2011 Published: June 06, 2011 12269

dx.doi.org/10.1021/jp201864g | J. Phys. Chem. C 2011, 115, 12269–12274

The Journal of Physical Chemistry C mechanisms can be useful to define the chemical nature of the interface between the nanoparticles and molecular adsorbates from the liquid phase. This point is important because it plays a key role for the enhancement of the photoinduced charge separation. In previous works, the inorganic TiO nanoparticles of these TiDMF sols and gel were structurally characterized by XAS, XRD, and TEM measurements. A chemical composition was proposed based on vibrational spectroscopies, 1H, and 13C NMR experiments for the organic part.15 The aim of the current study is to take a closer look at the hydrolysis phenomenon considering the DMF δ(OdCN) vibration to probe the interactions between DMF and the inorganic network during the solgel formation. To this purpose, Raman spectroscopy appears as a convenient and easy-handling technique.

ARTICLE

Scheme 1. Reactions of DMF Hydrolysis: a) with Concentrated HCl; b) with Titanium Oxychloride during TiDMF Sols and Gel Formation, the Proportion of Hydrolyzed DMF Can Be Obtained from y/x

2. EXPERIMENTAL DETAILS 2.1. TiDMF Sols and Gel Synthesis. TiDMF sols are synthesized using anhydrous N,N-dimethylformamide ((CH3)2NCHO, 99.8%, Aldrich) as solvent. Titanium oxychloride solution stabilized in an acidic medium (TiOCl2 3 1.4HCl 3 7H2O, 4.85 mol L1 in titanium, Millennium Chemicals). Sols are prepared by adding droplets of DMF to an ice-cooled TiOCl2 solution under continuous stirring. The mixture, which slowly becomes viscous, is stored in a closed container at room temperature. After aging for 8 days, the TiDMF sol turns into syruplike liquid. The sol to gel transition takes place after 7 months at room temperature forming a transparent and waxy solid. The gelation processes can be accelerated by heating at 70 °C, and in this case the sol to gel transition takes only 24 h. TiDMF sols and gel with a titanium concentration of 1.42 mol L1 were selected for comparison with our previous works. They are typically obtained by adding 12 mL of DMF to 5 mL of TiOCl2 precursor leading to a molar ratio nDMF/nTi = 6.4. 2.2. Raman Measurements and Analysis. Raman spectra were recorded at 20 °C on a Bruker RFS 100 FT Raman spectrometer with a laser wavelength of 1064 nm. It allows focusing the beam on a sample area of less than 1 mm2. The spectral resolution is set to 2 cm1. All samples were analyzed in quartz Suprasil cells. The Raman spectra were decomposed to extract the different vibrational bands using OriginLab 7.5 after background removal. A single Raman band is assumed to be represented as a pseudo-Voigt function, as previously described.16 The Gaussian part was fixed to a suitable value (50%) and not refined. The others parameters (areas, positions, and fwhm) of the different peaks where fitted with no constraint using LevenbergMarquardt algorithm. 2.3. NMR Measurements and Analysis. 1H and 13C liquid NMR spectra were acquired at room temperature on an ARX 400 spectrometer using a BBI probe. The recycle times used are 2 s for NMR 1H and 50 s for NMR 13C. The last one has been fixed to obtain the good proportions between each carbon type in DMF. All solutions were maintained in darkness at room temperature in tubes of 5 mm in diameter. Locking was performed with D2O set in an inner tube in such way that no perturbation in the system is induced by this extra solvent. In the chosen conditions, only the species in the liquid part are detected. Fits of peaks were carried out with exclusively Lorentzian curves using dmfit 2002 program.17

3. RESULTS AND DISCUSSION Strong mineral acids are efficient reactants for the hydrolysis of amides at about 100 °C. The products are usually a mixture of the amine and carboxylic acid. When catalyzed by polyoxometalate,

Figure 1. Raman spectra (1064 nm, 20 °C) of a) pure DMF, b) 1.42 mol L1 TiDMF freshly prepared sol (0% hydrolysis), c) 1.42 mol L1 TiDMF sol (32% hydrolysis), d) 1.42 mol L1 TiDMF gel (52% hydrolysis), and e) ZrOCl2/DMF solution (DMF/Zr = 6.4).

this reaction occurs at lower temperature.18 In our experiment, due to the presence of concentrated HCl to stabilize the titanium oxichloride solution, formation of dimethylammonium chloride (DMACl) and methanoic acid (MA) is expected according to the mechanism presented in part a of Scheme 1.19,20 This was confirmed by 1H and 13C liquid NMR, IR, and Raman spectroscopies leading to an equation described in part b of Scheme 1 with a titanium precursor.15 3.1. Evaluation of the DMF Hydrolysis. Raman spectroscopy was used to determine the kinetic of DMF hydrolysis in the TiDMF sols and gel. To this purpose, the DMACl νs(CN) and DMF νs(C0 2N) vibrations' bands (respectively 891.5 ( 0.5 and 866.0 ( 0.2 cm1) are used. The Raman spectra of 1.42 mol L1 TiDMF sols and gel in the 5001000 cm1 area, at different stages of the reaction, are presented in parts bd of Figure 1. These two bands were chosen because they are intense, not overlaid with peaks of other elements, and their relative surface areas are proportional to the DMACl/DMF molar ratio. This assertion was confirmed by recording reference spectra of DMACl mixed with DMF in a different ratio and keeping an overall concentration of CDMACl þ CDMF equal to the initial DMF concentration C0DMF, that is 9.13 mol L1 for TiDMF with CTi = 1.42 mol L1 (Supporting Information for details). According to part b of Scheme 1, the obtained hydrolysis ratio, defined as y/x, can be 12270

dx.doi.org/10.1021/jp201864g |J. Phys. Chem. C 2011, 115, 12269–12274

The Journal of Physical Chemistry C

ARTICLE

estimated from: y CDMACl I DMACl =J DMACl ¼ ¼ x CDMACl þ CDMF I DMACl =J DMACl þ I DMF =J DMF

ð1Þ

with IDMACl(DMF) the intensity of the peak at 891(866) cm1 and JDMACl(DMF) the scattering coefficient of the corresponding DMACl(DMF) peaks, the latter being defined as: I x ¼ J x Cx

ð2Þ

with x = DMACl or DMF. As the scattering coefficient determined from the reference samples are really close (0.117 ( 0.001 and 0.118 ( 0.003 L mol1 for the DMACl and DMF peak, respectively), we consider the denominator as constant and equal to C0DMF. When adding TiOCl2 to DMF, the scattering coefficient of DMF stays in the same range (0.116 ( 0.004 L mol1) at the beginning of the hydrolysis (when DMACl is not formed), furthermore, the widths and the positions of the peaks remain constant ((0.5 cm1) during the whole solgel transition. These observations suggest that no interactions between organic and inorganic precursors affect the scattering coefficient of these two peaks during the reaction. Kinetic evolution extracted from Raman spectra can be used to calculate the amount of the different compounds at each step of the reaction according to the equation of part b of Scheme 1. At the end of the reaction, when the gelation occurs, the DMF hydrolysis ratio (= y/x) reaches 52 ( 1% for 1.42 mol L1 TiDMF. This value is close to the theoretical maximum, obtained after consumption of all chloride ions originating from the TiOCl2 solution (53%; calculated with x = 6.4, y = 3.4, z e 2.6 for 1.42 mol L1 sols). The hydrolysis curve (Figure S2 of the Supporting Information) follows two different kinetic evolutions and a clear transition between them takes place after 8 days when 2 Cl per Ti4þ have been consumed, that is for 32% ( 1 of hydrolyzed DMF for 1.42 mol L1 TiDMF sols. The same trend is observed for less concentrated sol (0.8 mol L1 TiDMF sol) with a transition occurring after 8 days for 13% ( 2 of DMF hydrolyzed also corresponding to 2 Cl per Ti4þ consumed. Our recent investigation revealed that the kinetic evolution of the double hydrolysis reaction changes under the influence of a controlling physical variable, such as temperature or water content for instance. 3.2. Titanium Coordination during the Hydrolysis. Raman diffusion also provides information concerning the coordination of DMF around titanium cations. DMF is a strong Lewis base known to form bond with metallic cation via its oxygen atom.21 The formation of this bond induces a shift of the δ(OdCN) vibration toward higher wavenumbers. Besides the DMF νs(C0 2N) and DMACl νs(CN) peaks, used to calculate the amount of hydrolyzed DMF (Figure 1), other intense contributions appear at lower wavenumbers. In case of pure DMF, a single peak assigned to δ(OdCN) vibration is observed (part a of Figure 1, 663 cm1), when adding TiOCl2 to DMF, a peak characteristic of DMF bound to titanium cations appears (parts bd of Figure 1, ∼690 cm1).20 Meanwhile, the intensity of the peak attributed to free DMF (at 663 cm1) decreases. A similar result is obtained when mixing ZrOCl2 with DMF in the same proportions (part e of Figure 1). A further look at the spectra in Figure 1 indicates that several peaks contribute to the signal of titanium bound to DMF. In Figure 2 is represented a typical decomposition of the Raman spectra in the 500800 cm1 area. The two peaks corresponding to bound titanium located at 693 and 709 cm1 suggest that two different sites are available in the

Figure 2. Decomposition of the different peaks of the Raman signal in the 650 cm1 region, experimental data (black dots), fitted contributions (green curves), and envelope of fitted signal (red).

Ti4þ coordination. The peak frequency of bound solvent relative to the frequency of the free solvent, Δν = νbound  νfree, strongly depends on the ionic radius of the coordinated cations.22 The presence of two peaks corresponding to different lengths of TiO(DMF) bonds suggests a distortion of the titanium cation environment. This observation is in agreement with the EXAFS measurements done on the same samples where two TiO distances were observed for this TiDMF sol and gel (Supporting Information).15 In the case of freshly prepared TiDMF sol (part b of Figure 1), a broad peak appears at 630 cm1. This peak is attributed to the inorganic TiOTi precursor because its fwhm is large (>100 cm1) with respect to the width of the different organic compounds (∼20 cm1). To understand the synthesis mechanism of the TiDMF sols and gel, we aimed at determining the complexation number, n, of DMF bound to titanium cation, which is defined as: n¼

Cb CTi

ð3Þ

where Cb and CTi stand for the concentration in bound DMF and titanium cation, respectively. To this purpose, we modified the approach proposed by Irish.23 The previously described strategy cannot be applied to obtain Cb/CTi because it is not possible to prepare reference solutions including a known amount of DMF coordinated to Ti4þ to check the proportionality between Raman intensities and concentrations for bound DMF. Therefore, the issue is to calculate the concentration in bound DMF by subtracting the concentration in free DMF, Cf, from the total DMF concentration, Ct, considering: Ct ¼ Cf þ Cb

ð4Þ

For this analysis, the intensity of the peaks where normalized by the sum of the intensities of peaks of DMF and DMACl used for the calculation of hydrolysis rate (866 and 891 cm1). As mentioned above, this sum represents a constant (the initial DMF concentration) and this normalization avoids intensity variations due to shifts in incident power or changes of the sol optical density. First of all, the scattering coefficient Jf, correlating the intensity of the δ(OdCN) Raman peak at 663 cm1, If, with the concentration in free DMF has to be estimated. Jf factor is determined by plotting the intensity evolution of the δ(OdCN) vibration as a function of DMF concentration for 12271

dx.doi.org/10.1021/jp201864g |J. Phys. Chem. C 2011, 115, 12269–12274

The Journal of Physical Chemistry C

ARTICLE

reference solutions without TiOCl2. A linear fit with a very good correlation factor (r2 = 0.99799) allows us to experimentally determine that Jf = 0.092 ( 0.002 L mol1. The concentration in free DMF can thus be monitored depending on time and/or hydrolysis ratio of TiDMF sol. The total DMF concentration, Ct, can be expressed as a function of the initial DMF concentration, C0DMF, and the hydrolysis ratio, y/x, according to:   y Ct ¼ C0DMF 1  ð5Þ x Finally, by combination of the different equations, the number of bound DMF, n, can be calculated at each step of the reaction, as follows: n¼

C0DMF ð1  y=xÞJ f  I f CTi J f

ð6Þ

In Figure 3 is summarized the time evolution of the Raman bands in the 680 cm1 area, during TiDMF hydrolysis (part a of Figure 3) and the number of both kinds of DMF per Ti4þ (free and coordinated to Ti4þ) (left axis of part b of Figure 3) as a function of the consumed Cl (bottom axis) or of hydrolyzed DMF (top axis). The uncertainty on free and coordination DMF amount calculated by this method is (0.15. A few minutes after mixing DMF with TiOCl2, half of DMF (n = 3.2) is coordinated to titanium, occupying more than half of the available positions of TiO6 octahedra. This value decreases to 2.1 but the ratio between free and coordinated DMF remains close to one. A clear transition takes place when 2 Cl (by Ti) are consumed; afterward, the amount of coordinated DMF is decreasing faster until gelation occurs. Meanwhile, the amount of free DMF increases, suggesting that only bound DMF molecules are hydrolyzed to MA and DMACl. The evolution of the intensity of TiOCl2 peak (630 cm1) is also plotted (right axis of part b of Figure 3). A decrease of this peak intensity is observed until the band disappears when 2 Cl are consumed. It should be noticed that no displacement of the free DMF peak is observed (663.1 ( 0.4 cm1) and its fwhm remains equal to 14.5 ( 0.3 cm1. Information can also be obtained from the peaks attributed to the bound DMF even if errors on the parameters fitted for these contributions are higher (because of their low intensity and the overlap of the bands). The peaks initially located at 710.1 and 692.3 cm1 are shifted to 700.5 and 691.0 cm1 at the end of the hydrolysis. These displacements accounts for an extension of the TiO(DMF) bond lengths. They can be correlated to the displacement of the TiO bonds already deduced from EXAFS analysis (Supporting Information).15 Furthermore, the decrease in the ratio between the peaks intensity, I690/I710, suggests that, upon sol aging, DMF is mainly coordinated with the shortest TiO distance. This assumption was corroborated by the evolution of distances and coordination number in second neighbors obtained by EXAFS. Similar results were obtained for 1.42 mol L1 TiDMF sols heated at 70 °C. If the kinetic of DMF hydrolysis is more complicated when compared to the hydrolysis at room temperature (Figure S2 of the Supporting Information), the results on the evolution of the number of free and coordinated DMF are really similar, with a clear transition when two Cl/Ti are consumed. A similar analysis done for 0.8 mol L1 TiDMF sols also indicates the same trends, with a change in kinetic when two Cl are consumed. Nevertheless, the results are less precise because the lower amount of bound DMF induces high uncertainty on the

Figure 3. a) Time evolution of Raman spectra during 1.42 mol L1 TiDMF hydrolysis reaction; b) evolution of the intensity of the Raman peak attributed to TiOCl2 (2), number of coordinated (O) or free DMF per Ti4þ (b) as function of chlorine ions consumed/Ti atom or hydrolyzed DMF (%); c) number of mobile (plain) or in interaction with the solid (empty) DMACl (black square) and MA (blue diamond). The red line represents the overall number of MA and DMACl calculated from Raman spectroscopy data.

peaks decomposition of the signal, when compared to the spectra of 1.42 mol L1 TiDMF. 12272

dx.doi.org/10.1021/jp201864g |J. Phys. Chem. C 2011, 115, 12269–12274

The Journal of Physical Chemistry C 3.3. Liquid-State NMR. These new results offer a possibility to get more insight from the NMR results we obtained for the 1.42 mol L1 TiDMF sols. Analytical liquid-state NMR experiments allowed us to probe only mobile species. 13C NMR peaks attributable to methyl groups are detected at 36.1 and 36.2 ppm (doublet) for DMACl and DMF, respectively. Methanoic acid and the aldehyde group arise at 165.2 and 165.9 ppm, respectively. When the viscosity of the sol increases, no signal could be detected indicating that the sample does no more contain any liquid part (in the NMR sense) but colloids in strong interaction with the liquid. This phenomenon takes places when 2.2 Cl and H2O (per Ti4þ) have reacted to hydrolyze 2.2 DMF. Therefore, starting from the DMF and DMACl ratio deduced from NMR study and considering only liquid DMF is observed by this method, we use the number of free DMF obtained from Raman diffusion to extract the number of species resulting from the hydrolysis (DMACl and MA) that are mobile in solution, that is, measured by NMR, and then we deduce the amount in interaction with the solid (part c of Figure 3). The latter are obtained by difference between the values extracted from Raman diffusion (calculated according to part b of Scheme 1 and the hydrolysis ratio) and NMR. Because mobile DMACl and HCOOH species are mainly detected before 0.5 Cl are consumed, we can conclude that the hydrolysis mainly proceeds in solution from free DMF. However, its decrease alone is not sufficient to explain the amount of DMACl and MA formed. Actually, in the meanwhile, a decrease in bound DMF is also observed, indicating that part of it is released in the liquid phase for hydrolysis. In the following step (0.51), mobile DMACl is mainly detected by NMR, whereas mobile MA amount is not sufficient to keep the expected DMACl/ MA ratio equal to 1, as expected from Scheme 1, indicating that some hydrolysis of bound DMF also occurs, leaving methanoate species connected to Ti4þ cations on the nanoparticles dispersed in sols. The presence of methanoate in coordination of titanium was confirmed by EPR and FTIR analysis.13,15 Correlatively, some decrease in bound DMF content is actually observed. In the next step (1 to 2 Cl), DMACl and MA in interaction with the solid are mainly produced by hydrolysis. When the hydrolysis is going on (above 2 Cl, beginning of the sol to gel transition), DMACl and MA are mainly detected in solution by NMR. The amount of DMACl and MA interacting with the solid is almost constant and close to 1. In the final step, free DMF is also detected as the major contribution. Its quantity is increasing by desorption of coordinated DMF. 3.4. TiDMF Sols and Gel Formation Mechanism. The acidcatalyzed DMF hydrolysis can be described as a four step reaction according to the mechanism of part a of Scheme 2.19,24 It is now well admitted that the hydrolysis begins with the activation of the amide by proton addition.25 In the next step, a tetrahedral intermediate is formed through the reaction with a water molecule. Its deprotonation leads the hydroxylation of the complex; this is generally the limiting step. In a third step, a proton is captured by the nitrogen atom and finally the CN bond is broken to produce methanoic acid and ammonium chloride.23 In the case of the TiDMF sols and gel, as observed through Raman experiments, prior to being hydrolyzed, part of DMF complexes Ti4þ cation by entering its coordination sphere leading to an equivalent of the protonated DMF (iminium salt). Hydrolysis of DMF occurs on free and coordinated species according to the mechanisms described in parts a and b of Scheme 2, respectively. The NMR and Raman results give indications on the reaction mechanism involved in sols and gel formation. This mechanism

ARTICLE

Scheme 2. Protonation and Hydrolysis of DMF: a) in Acidic Conditions, b) with Titanium Precursor; (1) Proton Addition or Complexation of Carbonyl Group, (2) Hydroxylation, (3) Nitrogen Protonation, (4) Break of CN Bond (Terminal Carbons Are Not Shown)

arises from the influence of three chemical phenomenons: (i) the competition between hydrolysis on both kinds of DMF, (ii) the condensation of titanium precursor and the extent of the inorganic titanium oxide structure, (iii) the coordination of Ti4þ by organic species, that is DMF then carbonyl groups. The two firsts points are directly controlled by the sol pH. The sol formation mechanism could be detailed as follows: from the beginning of the reaction, half of the available DMF molecules (n = 3.2) enter into the octahedral Ti4þ coordination. Until 0.5 Cl anion per Ti4þ cation has reacted, the pH is low enough to enable the free DMF hydrolysis. The decrease in bound DMF may be attributed to the expansion of the TiOTi framework because the acidity of the sols is not high enough to maintain the species as oligomers or small clusters. From the structural point of view, before the transition of 2 DMF hydrolyzed, the TiO framework is expanding with a 2D KFeF4 type structure, built up of corner-sharing octahedra as determined by EXAFS experiments.15 Because DMF are occupying the octahedral corner, they are transferred to the solution, where they are hydrolyzed. According to part a of Scheme 1, the hydrolysis of DMF consumes HCl leading to an increase in the pH value as already demonstrated in the case of the TiDMF sols by the proton shift in 1H NMR.15 Rapidly, the acidic conditions required for the protonation of free DMF are no more satisfied, slowing down its hydrolysis in solution. In the same time, the protonated equivalent for DMF coordinated to titanium is preferentially hydrolyzed with Ti4þ acting as a Lewis acid. This reaction occurs, on bound DMF, at the surface of the nanoparticles under construction. 12273

dx.doi.org/10.1021/jp201864g |J. Phys. Chem. C 2011, 115, 12269–12274

The Journal of Physical Chemistry C Then, after 0.5 Cl ion is consumed, the hydrolysis of free and bound DMF are in competition. The last one is promoted as HCl is consumed by the hydrolysis leading to the pH increase. In the meanwhile, MA and DMACl interacting with the solid are mainly produced and we infer that the methanoate species are directly linked to the Ti cations, according to the mechanism suggested in part b of Scheme 2, as already confirmed by EPR and IR analysis. DMACl is incorporated in between the layers of the expanding TiO network as suggested by our DRX measurement.15 The increased value of pH also directly influences the polycondensation of TiO nanoparticles: the stability of TiOCl2 is altered and an inorganic condensation occurs. The disappearance of the broad peak attributed to TiOCl2 (630 cm1) can be due to this structural change. As observed by XAS experiments, before 32% of DMF is hydrolyzed, the KFeF4 titanium oxide structure is expending with octahedra-sharing corners, along two dimensions. The 3D growth perpendicularly to the layers is prevented by the presence of coordinated DMF and MA. When the transition at 32% of hydrolyzed DMF occurs, the pH increases at values that do not allow anymore the protonation of free DMF, stopping its hydrolysis. After this point, only DMF coordinated to titanium is hydrolyzed because Ti4þ still acts as a Lewis acid. When all of the Cl have reacted with DMF, less than 0.4 DMF are bound to the Ti4þ and MA is also present in the coordination sphere, whereas 2.6 DMF remains in the liquid part with DMACl and free MA formed during the first step of hydrolysis. After the transition, a structural change leads to octahedrasharing edges and corners structure, based on a FeUS3 model, and to an extension of the inorganic network in two dimensions. This condensation could explain the increase of DMF, DMACl, and MA in the liquid phase, visible after the transition because less titanium oxide corners are available for coordination by organic molecules. Finally, methanoate species are coordinating the cations and DMACl molecules are inserted between the inorganic plans as suggested by IR, EXAFS, and DRX experiments.15

4. CONCLUSIONS In summary, Raman spectroscopy appears as a powerful tool to get insight in the comprehension of DMF hydrolysis and then on DMF complexation in the coordination sphere of Ti4þ during the formation of this kind of material. This detailed comprehension is of great importance as bound DMF and methanoate species formed during the hydrolysis prevent the polycondensation of titanium oxide along some of the growth directions leading to 2D nanostructured sols and gel. Combined with NMR measurements, these results give an indication of the reaction mechanism. The hydrolysis of free and Ti4þ-coordinated DMF are in competition and the second process is promoted, by HCl consumption and decrease in the acidity, allowing the formation of bound methanoate species. The nanometric size of the particles in the TiDMF sols and gel, as well as the presence of organic groups on the surface of the inorganic plans, are responsible for their original optical properties under UV illumination. Next, work will focus on the study of these properties in view to use these sols and gel in photovoltaic and photoelectrochemical applications. ’ ASSOCIATED CONTENT

bS

Supporting Information. Raman calibration curves on reference samples. DMF hydrolysis calculated by Raman experiment for 1.42 and 0.8 mol L1 TiDMF sols. Correlation of the

ARTICLE

TiO measured distances between Raman Spectroscopy and EXAFS experiments. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]; tel: þ33 2 40 37 39 35; fax: þ33 2 40 37 39 95.

’ ACKNOWLEDGMENT T.C. is indebted to the Region Pays de la Loire and CNRS for his grant. This work was also funded by the French Ministere de la recherche through the ACI Energie, ACI Nanosciences, and ANR-PV. ’ REFERENCES (1) Heller, A. Acc. Chem. Res. 1995, 28, 503. (2) Fujhisima, A.; Honda, K. Nature 1972, 238, 37. (3) O’Regan, B.; Gra€etzel, M. Nature 1991, 353, 737. (4) Thompson, T. L.; Yates, J. T., Jr. Chem. Rev. 2006, 106, 4428. (5) Fujishima, A.; Zhang, X.; Tryk, D. A. Surf. Sci. Rep. 2008, 63, 515. (6) Bityurin, N.; Znaidi, L.; Marteau, P.; Kanaev, A. Chem. Phys. Lett. 2003, 367, 690. (7) Fadeeva, E.; Koch, J.; Chichkov, B.; Kuznetsov, A.; Kameneva, O.; Bityurin, N.; Sanchez, C.; Kanaev, A. Appl. Phys. A: Mater. Sci. Process 2006, 84, 27. (8) Livage, J.; Henry, M.; Sanchez, C. Prog. Solid State Chem. 1988, 18, 259. (9) Testino, A.; Buscaglia, M. T.; Buscaglia, V.; Viviani, M.; Bottino, C.; Nanni, P. Chem. Mater. 2004, 16, 1536. (10) Soloviev, A.; Tufeu, R.; Sanchez, C.; Kanaev, A. V. J. Phys. Chem. B 2001, 105, 4175. (11) Mockel, H.; Giersig, M.; Wilig, F. J. Mater. Chem. 1999, 9, 3051. (12) Brohan, L.; Sutrisno, H.; Puzenat, E.; Rouet, A.; Terrisse, H. Titanium Oxide-based Gel Polymer, Int. Pat.WO101436 A2, 2004. Eu Pat., 04 742 604.4, 2005. Jpn Pat., 530327, delivered July 7 2009; US Pat., N°121406760 delivered January 14, 2010. (13) Cottineau, T.; Brohan, L.; Pregelj, M.; Cevc, P.; Richard-Plouet, M.; Arcon, D. Adv. Funct. Mat. 2008, 18, 2602. (14) Pattier, B.; Henderson, M.; P€oppl, A.; Kassiba, A.; Gibaud, A. J. Phys. Chem. B 2010, 114, 4424. (15) Cottineau, T.; Richard-Plouet, M.; Rouet, A.; Puzenat, E.; Sutrisno, H.; Piffard, Y.; Petit, P.-E.; Brohan, L. Chem. Mater. 2008, 20 (4), 1421. (16) Caminiti, R.; Musinu, A.; Paschuna, G.; Piccaluga, G.; Pinna, G. Z. Naturforsch. 1981, 36a, 831. (17) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve, S.; Alonso, B.; Durand, J.-O.; Bujoli, B.; Gan, Z.; Hoatson, G. Magn. Reson. Chem. 2002, 40, 70. (18) Liu, S.; Wang, C.; Zhai, H.; Li, D. J. Mol. Struct. 2003, 654, 215. (19) Brown, R. S.; Bennet, A. J.; Slebocka-Tilk, H. Acc. Chem. Res. 1992, 25 (11), 481. (20) Combelas, P.; Costes, M.; Garrigou-Lagrange, C. Can. J. Chem. 1975, 53 (3), 442. (21) Asada, M.; Fujimori, T.; Fujii, K.; Kanzaki, R.; Umebayashi, Y.; Ishiguro, S. J. Raman Spectrosc. 2007, 38, 417. (22) Umebayashi, Y.; Matsumoto, K.; Watanabe, M.; Ishiguro, S. Phys. Chem. Chem. Phys. 2001, 3, 5475. (23) Deng, Z.; Irish, D. E. Can. J. Chem. 1991, 69, 1766. (24) Zahn, D. J. Phys. Chem. B 2003, 107 (44), 12303. (25) Kresge, A. J.; Fitzgerald, P. H.; Chiang, Y. J. J. Am. Chem. Soc. 1974, 96 (14), 4698.

12274

dx.doi.org/10.1021/jp201864g |J. Phys. Chem. C 2011, 115, 12269–12274