Adsorption and Photoinduced Decomposition of ... - ACS Publications

Aug 4, 2010 - TiO2 Nanorod-Derived Synthesis of Upstanding Hexagonal Kassite Nanosheet Arrays: An Intermediate Route to Novel Nanoporous TiO2 Nanoshee...
0 downloads 8 Views 3MB Size
J. Phys. Chem. C 2010, 114, 14121–14132

14121

Adsorption and Photoinduced Decomposition of Acetone and Acetic Acid on Anatase, Brookite, and Rutile TiO2 Nanoparticles ¨ sterlund*,†,‡ A. Mattsson†,‡ and L. O FOI, CementVa¨gen 20, S-901 82 Umeå, Sweden, and Department Engineering Sciences, The Ångstro¨m Laboratory, Uppsala UniVersity, Box 534, S-751 21 Uppsala, Sweden ReceiVed: April 12, 2010; ReVised Manuscript ReceiVed: June 24, 2010

A comparative study of the adsorption and photoinduced degradation (PID) of acetone and acetic acid on thin films of anatase, brookite, and rutile TiO2 nanoparticles is presented. The materials were thoroughly characterized by a wide range of methods, including X-ray diffraction, transmission electron microscopy, and Raman and UV-vis spectroscopies. In situ FTIR transmission spectroscopy was used to follow adsorption and PID reactions. Molecular adsorption of acetone and acetic acid is observed on anatase and brookite, whereas significant dissociation occurs on rutile. It is inferred that adsorbate-surface interaction increases in the order anatase < brookite < rutile, favoring formation of bridge-bonded species on rutile (acetate and formate). Illumination with simulated solar light readily dissociates acetic acid and acetone on all TiO2 samples and produces polymorph-specific intermediate surface species, including acetate, formate, carbonate, and water. PID of surface coordinated acetate is rate determining for complete mineralization of acetic acid and prevents further photooxidation on rutile. On anatase and brookite surfaces, acetate formation is suppressed upon photooxidation of acetone, whereas on rutile, acetate readily forms. On anatase, intermediate species form, which are not observed on either brookite or rutile, suggesting different reaction pathways for the different TiO2 polymorphs. Accurate quantum yield measurements were performed. The quantum yield for PID of acetone is larger for brookite than for anatase and much larger than for rutile. In contrast, the quantum yield for PID of acetate is lower for brookite than for anatase, whereas PID of acetate does not occur on rutile under our experimental conditions. The results are discussed in terms of a balance of strong adsorbate-surface interactions, moderate bonding of intermediate PID surface species, and efficient surface-adsorbate charge transfer of photogenerated electrons and holes. 1. Introduction 1

Ever since the early work by Fujishima and Honda in the 1970s, there has been an increasing interest in photocatalysissin particular, TiO2 photocatalysisswith applications ranging from solar hydrogen production, air and water purification, wet solar cells, and biomedical applications.2,3 The structure of anatase, rutile, and brookite is composed of TiO26- octahedral building blocks, distinguished by the distortion and arrangement of the octrahedrals. Hitherto, most physical and chemical studies have been done on the anatase and rutile phases. In particular, the anatase phase has been extensively studied in photochemical applications and is commonly known to be the most photocatalytic active phase.4-6 However, during the past years, the interest in brookite has increased and a few studies on the synthesis,7-9 optical,10 photochemical,11 and photocatalytic properties of both pure brookite12,13 and TiO2 mixtures with brookite14,15 have been reported. Addamo et al12 reported that brookite exhibits a high photocatalytic activity for 2-propanol and a promising longterm stability after repeated usage. Recent density functional theory calculations also show that the commonly exposed brookite (210) surface is more reactive than the ubiquitous anatase (101) surface and favors formic acid dissociation.16 Hitherto, no comparative experimental photocatalysis studies have been reported for the anatase, rutile, and brookite poly* To whom correspondence should be addressed. E-mail: lars.osterlund@ foi.se. † FOI. ‡ Uppsala University.

morphs. In this work, we report on acetone and acetic acid adsorption and photoinduced decomposition (PID) on wellcharacterized anatase, rutile, and brookite TiO2 nanoparticles prepared by solution-based methods. We employ in situ vibration spectroscopy to assess adsorbate structures and reaction intermediates during gas-phase adsorption and photodegradation.17 We determine quantum yield to facilitate intercomparisons of intrinsic material properties. Such studies are important and can give insights into mechanistic aspects that distinguish the different phases and provide means for rational TiO2 photocatalyst development. 2. Experimental Section The synthesis of anatase and rutile TiO2 nanoparticles was made by solution-based methods as described elsewhere.18 The brookite samples were prepared with slight modifications of the procedure described by Pottier et al19 using alkoxide precursors and will be reported elsewhere. The TiO2 nanoparticles were further prepared as films on Si(111) wafers or quartz substrates by the carbowax method to facilitate quantitative Fourier transform infrared (FTIR) transmission spectroscopy and UV-vis spectroscopy measurements. The thicknesses of the TiO2 films were determined from profilometric measurements (Dektak3, Veeco). Simultaneously, the interference patterns measured with FTIR spectroscopy were used to determine the refractive index according to

10.1021/jp103263n  2010 American Chemical Society Published on Web 08/04/2010

14122

¨ sterlund Mattsson and O

J. Phys. Chem. C, Vol. 114, No. 33, 2010

d ) N/2n(V¯ 1 - V¯ 2)

(1)

where N ) number of interference maxima (or minima) between the wavenumbers νj1and νj2 (in units of cm-1), n ) refractive index, and d ) thickness of the film. The structural, optical, and photochemical properties of the synthesized materials were characterized by a range of different techniques. Transmission electron microscopy (TEM) was done with both a JEOL 2000 FXII equipped with an EDS module (Link AN 1000) and a JEOL 2100F instrument. Grazing incidence and powder X-ray diffractograms (XRD) were obtained for samples heat treated at 723K with a Siemens D-5000 instrument. Quantitative XRD analysis was done on powder samples for brookite and rutile due to low signals from the corresponding thin films. In the UV-vis measurements, total reflectance Rt and transmittance Tt, as well as diffuse reflectance Rd and transmittance Td, were measured for wavelengths between 300 < λ < 800 nm using a double-beam PerkinElmer Lambda 900 UV-vis spectrophotometer equipped with a Spectralon coated integrating sphere. All samples were deposited on quartz windows in these measurements. Raman spectroscopy of TiO2 films deposited on Si substrates was made at ambient temperatures with a LabRam 800HR microscope using a 514 nm Ar ion laser employing 1800 lines mm-1 grating and a 100× objective. Optical damping filters, OD ) 0.6 and 1, were employed to avoid laser-induced heating of the samples. Each spectrum consisted of an average of five consecutive scans, each consisting of a 60 s long sampling time. The spectra were frequency calibrated against the intense Raman active mode of silicon at 520.7 cm-1 of the underlying substrate. The reported spectra were smoothed using a Savitzky-Golay algorithm. Annealing experiments were done in a dedicated reaction cell (Linkam TS 1500) employing a 785 nm diode laser and a 50× objective. The samples were heated at elevated temperatures for 2 h at indicted temperatures and then cooled to 298 K prior to Raman measurements. The reported temperatures are slightly overestimated due to a temperature gradient between the thermocouple mount on the sample cup and the sample position. However, this does not change the qualitative conclusions about the phase transitions reported here. The in situ Fourier transform infrared (FTIR) molecular spectroscopy experimental setup has been described elsewhere.18 Briefly, the FTIR spectrometer is equipped with a custom modified transmission cell, which allows for simultaneous FTIR spectroscopy, illumination of the sample, and gas dosing in a controlled atmosphere. For illumination, simulated solar light was employed (Xe lamp, operated at 200 or 300 W) employing AM1.5 filters together with a water filter to remove the infrared part of the radiation. The light was directed through an optical fiber bundle. The output from the fiber bundle between 200 < λ < 800 nm was 102 and 165 mW cm-2, as measured with a calibrated thermopile detector at the sample position. This corresponds to 11.9 and 19.2 mW cm-2 between 200 < λ < 400 nm, respectively, for the two operating lamp powers, roughly corresponding to the photon power above the band gap of TiO2 (see below). Acetone dosing (analytical grade, Scharlau) was done using a home-built gas generator based on a diffusion tube design, which facilitates calibrated injection of vapor from a liquid reservoir held at a constant temperature into a synthetic air carrier gas (20% O2 and 80% N2).18 A carrier gas flow of 100 mL min-1 was used and controlled via a set of mass flow controllers. A 0.5 mm inner diameter diffusion tube and a

reservoir temperature of 298 ( 0.2 K yielded a steady-state concentration of 320 ppm acetone in the gas feed. Acetic acid (pro analysi, Merck) dosing was done in a similar manner, but due to the lower vapor pressure, a larger diffusion tube (5.5 mm inner diameter) and a higher reservoir temperature (T ) 308 ( 0.2 K) was employed, which resulted in a steady-state concentration of 115 ppm in the gas feed. Formic acid (reagent grade, Scharlau) dosing utilizing a 5.5 mm inner diameter diffusion tube and a 303 ( 0.2 K reservoir temperature gave a steady-state concentration of 420 ppm in the gas feed. Prior to the measurements, the samples were annealed to 673 K in 100 mL min-1 synthetic air and subsequently cooled to 299 K in the same gas feed. This procedure results in samples free of organic contaminants, as determined directly by in situ FTIR spectroscopy. The TiO2 samples were exposed to the reaction gas for ca. 15 min, followed by a 15 min long dwell time with only synthetic air in the gas feed before the irradiation commenced. All spectra was obtained with a 4 cm-1 resolution and a sampling time of 60 s, consisting of 137 consecutive spectra, followed by a 60 s dwell time (exception for the acetone measurements on brookite where 30 s scan and dwell times were used). Post-treatment of spectra included smoothing with a Savitzky-Golay algorithm and baseline corrections. Samples were, in addition, cleaned after repeated usage to facilitate reliable duplicate measurements in a similar manner as described elsewhere.20 Otherwise, carbon deposits generated from the photodegradation experiments may gradually accumulate on the samples, which is particularly evident for rutile (see below). Briefly, the samples were immersed in 1 mM NaOH for 15 min, followed by washing in Milli-Q water, 1 mM HNO3 soaking (15 min), and a new wash in Milli-Q water. Finally, the samples were subjected to mild calcination at 673 K in air for 60 min. This gentle procedure, in combination with the in situ pretreatment annealing procedure described above, yielded contaminant-free samples and reproducible results. Univariate calibration of the absorbance in the FTIR measurements was performed using a liquid sample cell (Omni cell, Specac) employing an analyte concentration covering the absorbance region of interest. Acetone calibration was obtained by dilution with carbon disulfide to appropriate concentrations (Merck, purity > 99.9%). Least-squares fitting of the absorbance versus concentration data using the ν(C-C) band at ca. 1220 cm-1 showed that 1 integrated AU corresponds to 1.88 × 1017 molecules per cm2. Similarly, acetate calibration was done by mixing 98 mg of sodium acetate (puriss, Kebo AB) with 10 mL of methanol (HPLC gradient grade, J. T. Baker) and further dilution in carbon tetrachloride (analyze grade, Merck) and showed that 1 AU for the acetate νa(COO) band at 1580 cm-1 corresponds to 1.76 × 1018 molecules per cm2. 3. Results and Discussion 3.1. Materials. XRD data show only diffraction peaks due to the pure anatase, brookite, and rutile phases. Using the Scherrer formula on the Bragg reflections from the (101), (121), and (101) planes of anatase, brookite, and rutile, respectively, gives particle diameters of 25 and 40 nm for the anatase particles (denoted as A25 and A40, respectively), 12 nm for brookite (denoted as B12), and 20 nm for rutile (denoted as R20); see Table 1. Note that, because the rutile nanoparticles typically are elongated along the 〈001〉 directions (see TEM data below), the XRD diameter using the 〈101〉 reflection does not capture the asymmetric particle morphology and merely quantifies the average particle size. Typical TEM micrographs of the four particle types used in the present study are shown in Figure 1.

Adsorption and PID of Acetone and Acetic Acid on TiO2

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14123

TABLE 1: Physical Properties of the Anatase, Brookite, and Rutile TiO2 Nanoparticles Used in the Present Study particle size (nm) sample

TEM

XRD

band gap from UV-vis (ev)

film thickness, Si substrate (µm)

film thickness, quartz substrate (µm)

anatase A25 anatase A40 brookite B12 rutile R20

23 40 11 9 × 26

25 40 12 20

3.24 n.a. 3.27 3.03

3.7 2.1 0.6 2.5

1.5 n.a. 1.4 1.2

The particle dimensions observed in TEM are in good agreement with XRD data (Table 1). A detailed TEM analysis of the particles21 shows that the anatase particles possess the typical truncated bitetrahedral morphology exposing mainly {101} planes. The rutile particles appear rectangular-shaped in TEM with dimensions of 9 × 26 nm as viewed along the 〈110〉 directions and expose mainly {110} surfaces. The morphology of the brookite particles is more complex, and several particles appear doughnut-shaped and may indicate the coalescence of two or several nuclei.22 In Figure 2 is shown Raman spectra of the anatase, brookite, and rutile samples. Similar to XRD, only Raman bands due to pure anatase,23 brookite,24 and rutile25,26 phases are evident on each sample. Because Raman spectroscopy is known to be sensitive to the absolute phase composition of TiO2,27,28 we again conclude that the particles are phase pure. The bands at 303

and 521 cm-1 originate from the underlying silicon substrate. Observed Raman bands are compiled in Table 2 and are in good agreement with reported literature data. In Figure 2c is also shown Raman spectra obtained on a brookite film after photodegradation experiments and subsequent pretreatment methods (cf. the Experimental Section). It is known that the position of the Eg modes in TiO2 is sensitive to the particle diameter for d < 20 nm due to quantum confinement effects.29 Examination of the spectra in Figure 2c reveals that the position of the brookite Raman peaks does not shift upon pretreatments and photoexperiments, indicating that particle growth is negligible. In fact, brookite TiO2 nanoparticles with a diameter of 11 < d < 35 nm have been predicted to be the most thermodynamically stable form of TiO2.30 The observed stability of the brookite particles can be a possible explanation for the lack of deactivation after repeated use reported by Addamo et

Figure 1. TEM micrographs of (a) anatase (A25), (b) anatase (A40), (c) brookite (B12), and (d) rutile (R20) TiO2 used in the present study.

14124

¨ sterlund Mattsson and O

J. Phys. Chem. C, Vol. 114, No. 33, 2010

Figure 2. Raman spectra of (a) anatase (A25), (b) anatase (A40), (c) brookite (B12), and (d) rutile (R20) TiO2 films on Si wafers. The peak at 521 cm-1 is due to Si from the underlying substrate. For brookite, a Raman spectrum after repeated photocatalytic experiments is also included (dotted line).

TABLE 2: Vibration Frequencies and Mode Assignments of Observed Raman Bands (cm-1) for Anatase, Brookite, and Rutile TiO223-26 anatase

brookite

rutile

position assignment position assignment position

assignment

126 154 196 213 247 323 366 413 454 462 545 585 635

B1g combination Eg A1g

145 197 397 639

a

Eg Eg B1g Eg

Ag Ag Ag B1g Ag B1g B2g Ag, B1g B3g B2g B3ga B2g Ag

154 239 446 609

Tentative assignment.

al12 as well as the reproducible behavior upon various pretreatments reported here. Thermal stability measurements (successive annealing for 2 h at 100 K intervals) of brookite show that phase transformation of brookite to anatase occurs at ∼1100 K and further transition to rutile occurs at 1473 K. Taking into account the temperature gradient present in our experimental setup and the shorter annealing times compared with previous studies on both synthesized13,27,30 and naturally occurring brookite,31 we find good agreement between our measured values of the temperature range at which the brookite-to-anatase phase transition occurs (773-1073 K). For the A25 anatase sample, the phase transition to rutile occurs at 1373 K upon a similar annealing procedure, in good agreement with reported data taking into account the experimental differences.32 The slightly lower transition temperature for the A25 sample compared with

the brookite sample suggests that the anatase particles formed during the brookite-to-anatase conversion have a smaller size than the A25 anatase particles at the same temperature. The observation of a phase transformation from brookite f anatase f rutile is in agreement with the results of Ye et al33 of mixed TiO2 nanocrystals (40.8% brookite, 15.5 nm; 32.7% anatase, 22.7 nm; 26.5% rutile, 27.2 nm). It is, however, at variance with previous results on pure brookite where only a direct transition from brookite to rutile was reported.13,31 However, Zhang and Banfield30 have pointed out that the relative thermodynamic stability of brookite and anatase depends critically on the particle size. Thus, with the particle sizes used here as well as in the work of Ye et al,33 brookite is expected to transform to anatase and finally rutile based solely on thermodynamic arguments. From HR-TEM analyses of anatase and brookite specimens, Penn and Banfield identified intertwined anatase-brookite interfaces and inferred that the oriented attachment and growth of brookite can lead to formation of structures that are different from the bulk; in particular, anatase and brookite interconversion can readily be realized.22 The absorption coefficients and optical band gaps of the TiO2 samples were determined by UV-vis spectroscopy measurements of TiO2 films with known thicknesses deposited on quartz substrates (Table 1). The absorption coefficient was calculated according to R(λ) ) d-1 ln((1 - Rcorr(λ))/Tcorr(λ)), where Rcorr and Tcorr are the corrected values of the total reflectance and transmittance.34 All materials exhibited wavelength-dependent absorption around the optical band-gap characteristic for indirect interband transitions, that is, exhibiting a linear dependence of (Rhν)1/2 versus hν around the band gap.35 The optical band gaps, Eg ) hν, shown in Table 1 were determined by extrapolation of similar plots; see Figure 3b. In particular, Eg was determined to be Eg ) 3.24 eV and Eg ) 3.03 eV for anatase and rutile,

Adsorption and PID of Acetone and Acetic Acid on TiO2

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14125

Figure 3. (a) Absorption coefficient, R, and (b) plot of (Rhν)1/2 vs hν of anatase, brookite, and rutile thin films.

respectively, in very good agreement with tabulated data. The UV-vis spectrum for brookite is very similar to anatase. The similarity of the anatase and brookite UV-vis data is reasonable considering the similarity of their crystal structures. However, the shoulder apparent at ca. 350 nm (3.55 eV) makes the bandgap determination of brookite somewhat ambiguous. We note that reflectance spectra obtained from mixed brookite-rutile powders exhibiting similar structures in the absorption spectra near Eg previously have been observed.36 Indeed, many reported synthesis methods obtain brookite with small fractions of either rutile or anatase.37,38 However, as reported above, we find no evidence for phase impurities in our brookite sample in either Raman or XRD data and disregard this effect here and instead attribute it to intrinsic material properties associated with our brookite specimens, for example, contributions from acoustic and optical modes. Using a narrow region near Eg away from the shoulder in the absorption spectra, we extrapolate Eg ) 3.27 eV for brookite, that is, very close to the Eg for anatase. Using the interference patterns in transmission FTIR on thin TiO2 films with known thicknesses (Table 1), we determined the refractive index for the different films using eq 1 (Table 1). For a porous material, the refractive index is given by n2 ) (ns2 - 1)(1 - P) + 1, where n and ns are the refractive indices of the porous and bulk materials, respectively, and P is the porosity of the material.39 The refractive index for the bulk TiO2 polymorphs is ns ) 2.52 for anatase, ns ) 2.63 for brookite, and ns ) 2.72 for rutile,40 yielding a porosity of 66, 65, 58, and 69% for the A25, A50, B11, and R20 samples, respectively. 3.2. Acetic Acid Adsorption. In Figure 4 is shown FTIR transmission spectra of acetic acid adsorbed on anatase, brookite, and rutile TiO2. The spectra show absorbance bands due to both molecularly adsorbed acetic acid and surface-coordinated acetate species in agreement with other studies.20,41,42 Typical acetic acid peaks are the ν(CdO), δ(CH3), and ν(C-O) bands at ∼1711, ∼1675, ∼1380, and ∼1300 cm-1, respectively.41,42 Acetate is associated with the νa(COO) band observed in the 1516-1544 cm-1 region (surface-coordinated acetate) and in the 1550-1560 cm-1 region (aqueous acetate ions) and the νs(COO) bands in the 1447-1452 cm-1 region (surface-coordinated acetate) and the 1418-1423 cm-1 region (aqueous acetate ions). Ionic acetate is preferentially found on the anatase samples, whereas surfacecoordinated acetate dominates on rutile and brookite. Bands due to δs(CH3), ν(C-C), and F(CH3) at ∼1340, ∼1050, and ∼1025 cm-1 in acetate are clearly resolved on brookite and rutile.20,41 Vibration frequencies and mode assignments are compiled in Table 3. In agreement with previous findings by Rotzinger et al,20 we find that, for acetate, the intensity of the νs(COO) vibration is much larger than the δs(CH3) vibration and occurs at a higher frequency. The positions of the adsorbed acetic acid

Figure 4. In situ FTIR transmission spectra of acetic acid adsorbed on rutile (R20), brookite (B12), and anatase (A40) and (A25) TiO2 nanoparticle films.

and acetate vibrations are shifted compared with condensed acetic acids and matrix data for inorganic acetate complexes in varying amounts for the different TiO2 samples.20,43,44 This is most clearly seen by examining the ν(COO) bands. For rutile, a large red shift of the νa(COO) peak by ca. 25 cm-1 is observed compared with anatase, whereas a minor blue shift is seen for the νs(COO) peak. Examining the splitting of the asymmetric and symmetric ν(COO) modes in surface-coordinated acetate, we can deduce that the small particles (A25) exhibit the largest splitting (∆ν(COO) ) 97 cm-1), and with increasing particle size (A40), it decreases slightly (∆ν(COO) ) 92 cm-1), whereas on brookite, ∆ν(COO) ) 86 cm-1, and on rutile, it is ∆ν(COO) ) 65 cm-1. Following the analysis by Rotzinger et al,20 we conclude that the amount of acetate with ionic character increases in the order anatase < brookite < rutile. Moreover, the preferred coordination of surface acetate is bridging bidentate (µ-coordinated) on rutile and probably also on brookite, which is also consistent with the ∆ν(COO) rules of Deacon et al.45 The results indicate that water displaces adsorbed acetate on anatase and forms aqueous clusters. This is further supported by the observation that molecular water forms on anatase (shoulder at 1630 cm-1); this is not evident on brookite or rutile. Interestingly, the trend of the ν(COO) band splitting, correlates with the nearest-neighbor (nn) Ti-Ti distance of the commonly exposed (101), (210), and (110) surfaces on anatase, brookite, and rutile particles, respectively, which previously have been reported to be important for the formic acid/formate bonding.16 The rutile (110) surface (with the shortest nn distance) is found to favor formic acid dissociation and formation of µ-formate,

14126

¨ sterlund Mattsson and O

J. Phys. Chem. C, Vol. 114, No. 33, 2010

TABLE 3: Compilation of Observed Vibration Frequencies and Proposed Mode Assignments of Acetone, Acetic Acid, and Stable Decomposition Products Adsorbed on Anatase, Rutile, and Brookite TiO217,20,41,48,54 wavenumber (cm-1) adsorbed species acetone

acetic acid

acetate

Ti-O-CdO formate

a

mode assignment

anatase A25

anatase A40

brookite B12

rutile R20

ν(C-C) δs(CH3) δa(CH3) ν(CdO) multilayer ν(C-O) δ(CH3) ν(CdO)

1241 1369 1423 1693 1704 1299 1387 1675 1710 1027 1049 1339 1423 1447 1544 1589 1722 1362 1374 1590

1242 1370 1425 1692 1703 1301 1385 1674 1708 1028 1049 1339 1425 1448 1540 1584 1722 1362 1376 1592a

1243 1371 1424 1689 1714 1308 1377 1677 1714 1028 1049 1339 1424 1452 1538 1574

1241 1369 1423 1687 1298 1375 1673 1699 1024 1050 1343 1451 1423 1516 1552

1362 1381 1562

1354 1379 1542

F(CH3) ν(C-C) δs(CH3) νs(COO)aq νs(COO)ads νa(COO)ads νa(COO)aq ν(CdO) νs(COO) δ(CH) νa(COO)

Absorbance band due to the ν(COO) band in HCOO- ions.

whereas the anatase (101) surface (with the longest nn distance) does not promote dissociation and favors monodentate (or η1coordinated) species coordinated to basic O atoms. Indeed, the TiO2 samples display differences in the ratio of the amount of adsorbed acetic acid and acetate species. The ratio of acetate to acetic acid molecules increases in the following order: large anatase (A40) < small anatase (A25) < brookite < rutile, showing that acetic acid dissociates more readily on rutile than the other polymorphs. Bearing in mind that the nanoparticles expose different facets, including, for example, the more reactive (001) anatase surface, we thus find spectral support for a trend of increasing adsorbate-surface bonding strength (which weakens the intramolecular bonds and hence promotes dissociation) going from anatase < brookite < rutile. However, as pointed out by Li et al,16 the basicity of neighboring surface O atoms must also be considered on a case-by-case basis in more detailed studies because they are involved in deprotonation of the parent molecule as well as available sites for carboxylate coordination.42 We note that, in the case of brookite, this scheme would imply significant spectral contributions from acetic acid, which dissociates in junctions along the [001] direction, separating closely packed building blocks in (210) planes, or other type of lattice defects, because the (210) planes chemically appear similar to anatase and are, therefore, not expected to favor dissociation µ-coordination.16 The thermal stability of acetic acid and acetate on the different TiO2 phases was determined by monitoring the temporal evolution of the adsorbed surface species with the TiO2 samples kept in the dark after acetic acid exposure. These measurements show a slight increase in the absorbance bands associated with acetate and a concomitant decrease in the absorbance bands associated with acetic acid, demonstrating a slow dissociation path of acetic acid at room temperature. The rate of acetic acidto-acetate interconversion was determined from the amplitude of the acetate νs(COO) and the acetic acid νa(CdO) bands, respectively, during a time-period of 30 min in an inert atmosphere. Correcting for acetic acid desorption, which is

inferred from a decreased acetic acid absorbance not accompanied by acetate formation and is known to occur slightly above room temperature for multilayers,46 our results show a notably higher acetic acid dissociation rate, about 3 times, on rutile compared with anatase and brookite, which further supports the arguments above. 3.3. Photoinduced Decomposition of Acetic Acid. In Figure 5 is shown in situ FTIR spectra obtained during photodegradation of acetic acid on anatase, brookite, and rutile TiO2. On anatase and brookite, absorption bands associated with acetic acid rapidly decrease within 5 min. At the same time, absorption bands at ∼1550 and ∼1350 cm-1 due to the typical νa(COO) and νs(COO) vibrations in coordinated formate appear (see Table 3), which eventually disappear after extended illumination times.17,20 On anatase, additional νa(COO) bands at ∼1590 cm-1 due to formate ions17,20 appear at later stages of photodecomposition, signaling formation of aqueous-like clusters. Again, this suggests weaker adsorbate-surface interactions on anatase compared with the other polymorphs and displacement by water formed in the course of the photoreaction, in agreement with previous studies.42 In contrast, on rutile, where most acetic acid dissociates upon adsorption (Figure 4), only negligible spectral differences are observed after extended illumination; acetate is photostable on rutile. In Figure 6, a schematic drawing of the reaction pathways for the different TiO2 samples is shown. Formate formation on anatase and brookite is further supported by the appearance of the typical ν(CH) band at ∼2850 cm-1 on samples A25 and A40 and ∼2877 cm-1 on brookite (spectra not shown) as well as the νa(COO) + δ(CH) combination band at ∼2950 cm-1. The low ν(CH) frequency on anatase is attributed to an increasing amount of water on the surface as a consequence of the photoinduced decomposition reactions, which results in formate ions instead of surface-coordinated species, as seen on drier surfaces. This is further proved by illuminating A40 preadsorbed with formic acid (Figure 7). The ν(CH) band for coordinated formate at 2871 cm-1 shifts to 2849 cm-1 as the photoinduced decomposition progresses. These are not methoxy groups because corresponding ν(CO) bands in the 1000-1100 cm-1 region are lacking (this argument is complicated in the case of acetic acid due to overlapping acetate bands (see Table 3), hence necessitating this indirect proof). On brookite, the formate ν(CH) band appears at 2877 cm-1 during photoinduced decomposition of both acetic acid and formic acid also after extended illumination times, indicating that most formate is coordinated to the surface (cf. Figure 4). The maximum intensity of the absorption bands associated with formate occurs after ca. 6 min of illumination for brookite and the large anatase particles (A40) and somewhat later, ∼12 min, for the smaller anatase particles (A25). Anatase displays two additional absorbance bands at ∼1740 and ∼1720 cm-1 following PID of acetic acid. The time evolution of these two bands is different, and they are thus not correlated. The 1740 cm-1 vibration reaches its maximum intensity at the same time as that of the 2740 cm-1 absorption band (cf. Figure 7) and is, therefore, attributed to formaldehyde.18 The origin of the absorbance bands at 1720 cm-1 is not entirely clear. It reaches its maximum somewhat later than the 1740 cm-1 band (10 min for A40 and ca. 30 min for A25) and is correlated with the absorbance band at ∼1220 cm-1. Furthermore, it is also one of the most stable intermediates with large residues after prolonged UV illumination. It has been suggested that these two absorbance bands are attributed to an asymmetric coordinated end-on Ti-O-CdO species rendering the -CdO moiety a carbonyl vibrational character (see ref 17

Adsorption and PID of Acetone and Acetic Acid on TiO2

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14127

Figure 5. In situ FTIR transmission spectra at different times (min) during photodegradation of acetic acid on (a) anatase (A25), (b) anatase (A40), (c) brookite (B12), and (d) R20 rutile. Simulated solar light with Ptot ) 165 mW cm-2 was employed.

Figure 6. Schematic drawing of the reaction pathways for photoinduced decomposition of acetic acid on anatase, brookite, and rutile. Acetic acid and acetate coexist on the surface prior to illumination. The bidentate coordination of formate is tentative and is drawn as a weakly bonded µ-coordinated species (dashed lines). Bold and normal font styles for carbon denote β-carbon (C) and R-carbon (C), respectively.

and references therein). Because neither of the 1740, 1720, and 2740 cm-1 bands appear on brookite and rutile, which both interact stronger with acetic acid, acetate, and formate, it appears that these bands are associated with reaction pathways involving noncoordinated -CdO, and -COOH moieties or ionic species that we have found are prevalent on anatase and not on the other polymorphs. It has previously been reported that cleavage of the R-carbon in acetate directly forms CO2 on commercial TiO2 (P25, containing mostly anatase TiO2 nanoparticles), whereas the β-carbon forms CO2 via intermediate species, such as methoxy, aldehyde, and formate.41,42,47 We17,48 and others49

Figure 7. In situ FTIR transmission spectra at different times during photodegradation of formic acid on anatase (A40). Simulated solar light with Ptot ) 165 mW cm-2 was employed.

have shown that oxidation of β-carbon proceeds through different pathways depending on how the intermediate products are coordinated to the TiO2 surface, which, in turn, depends on the polymorph, particle size, and defect density. We propose that aldehyde does not form on brookite and rutile because most of the exposed surfaces on these particles promote µ-acetate and formate species, hence providing a different oxidation pathway for the β-carbon. Oxidation of bridge-bonded formate is known to occur by either dehydration or dehydrogenation, producing CO and CO2, respectively, depending on reaction conditions.17,49 After 60 min of illumination of the A40 anatase particles, most PID surface products have disappeared. Minor residues are due to formate, (bi)carbonate species,17,48 as well as the conspicuous 1720 cm-1 band tentatively attributed to coordi-

14128

J. Phys. Chem. C, Vol. 114, No. 33, 2010

Figure 8. In situ FTIR transmission spectra of acetone adsorbed on rutile (R20), brookite (B12), and anatase (A40) and (A25) TiO2 nanoparticle films.

nated CO2. On the smaller anatase particles (A25), acetate residues are evident, whose concentration is slightly higher on brookite. On rutile, acetic acid is converted to acetate, as evidenced by the narrowing and increase of the νs(COO) acetate vibration, but further oxidation does not occur over the time scale of the experiments. 3.4. Acetone Adsorption. Adsorption of acetone on brookite, anatase, and rutile TiO2 nanoparticles shows the typical ν(C-C), δs(CH3), δa(CH3), and ν(CdO) infrared absorption bands at ∼1242, ∼1370, ∼1424, and ∼1690 cm-1 (Figure 8), albeit slightly shifted compared with liquid-phase acetone.44 The peak positions agree well with the peak position of acetone measured previously on anatase.17,18 A chemical shift of a few cm-1 between the different samples is evident. In particular, the down shift of the ν(CdO) frequency signals bonding to coordinately unsaturated Ti atoms through the O atom in the carbonyl group in an η1-coordination.18,50 The high-frequency shoulder of the carbonyl bands for the A25 and A40 samples indicates acetone cluster formation (multilayer), as previously observed.18 Similarly, the shoulder at 1714 cm-1 on brookite is attributed to multilayer adsorption. On rutile, no evidence for multilayer adsorption is observed. We use curve fitting employing a mixed Lorenz-Gaussian function to locate the frequency and intensity of the multilayer and monolayer adsorption bands (Table 3). The results reveal that the monolayer ν(CdO) band is shifted from 1693 cm-1 for A25 down to 1687 cm-1 for rutile. A red shift generally indicates a stronger bonding of the carbonyl group with the surface metal atom. The different ν(CdO) band positions thus indicate that the acetone bonding to the surface increases according to anatase < brookite < rutile. It has recently been proposed that weakly bonded water simultaneously is displaced upon acetone adsorption on hydroxylated anatase.51 We observe a negative absorption band at ∼1620 cm-1 on anatase, which concurs with this conclusion. Interestingly, this is not evident on either brookite or rutile (which is inferred to interact stronger with acetone). This suggests a lesser amount of weakly bonded water on rutile and brookite. Weak absorption bands at 1025, 1450, 1520, and 1570 cm-1 not related to acetone are detected on rutile particles. Comparisons with the results presented in sections 3.2 and 3.3 as well as other studies17,20,41,52 show that these absorption bands are due to formate and acetate. We cannot, however, exclude formation of methoxy species formed from dissociated methyl groups and associated ν(C-O) bands that have been reported

¨ sterlund Mattsson and O to occur in the 1000-1100 cm-1 region on rutile53 due to overlap with acetate δ(CH3) and ν(C-C) bands, as discussed in section 3.3. Nevertheless, our results indicate that rutile interacts stronger with acetone than brookite and anatase, which is the same trend as for acetic acid. To further check whether acetone dissociation on rutile is thermally activated, experiments were conducted in the dark for extended periods with the sample held at 299 K. A slow decrease of the acetone concentration is observed on all TiO2 samples over a time period of 1 h. This is attributed to desorption from the surface, in agreement with TPD studies,50 rather than dissociation, because it is not concerted by evolution of absorption bands due to dissociation products. From the absorbance spectra, we estimate the acetone desorption rate to be 0.24 mL/h. This indicates that the weak 1570, 1520, 1450, and 1025 cm-1 absorption bands due to acetone dissociation products (mainly acetate and formate) are due to interactions with small amounts of reactive sites on rutile, which becomes inaccessible to additional acetone molecules after the first reactions. A dissociative adsorption pathway of acetone has previously also been observed on large rutile TiO2 nanoparticles prepared by microemulsion methods.17 3.5. Photoinduced Decomposition of Acetone. In Figure 9 is shown FTIR spectra for brookite, anatase, and rutile after UV illumination. Comparisons of the spectra in Figure 9 reveal several differences. Acetone rapidly decomposes on brookite, as evident by the FTIR spectra in Figure 9c. Within 5 min, the ν(CdO) and ν(C-C) absorbance bands at 1689 and 1243 cm-1 associated with acetone have completely disappeared and new absorbance bands in the 1360-1381 cm-1 region and at 1446 and 1562 cm-1 due to reaction intermediates appear. The intermediate species reaches their maximum during the first 5 min, after which their concentration starts to decrease. Thus, PID of acetone takes place before the PID of intermediate surface species is completed. After 60 min of illumination, only very small traces of residual surface products are observed. The same PID pathways characterized by the same vibrational signatures and time evolution are observed for both photon fluxes employed here (102 and 165 mW cm-2). Comparing the acetic acid adsorption data discussed above and reported data in the literature, we can identify bridging bidentate coordinated acetate and formate species on rutile in Figure 9d (Table 3).17,20,41,48 On rutile, the intensity of the formate bands reaches a maximum after ca. 12 min and the acetate ν(COO) bands increase throughout the illumination time, whereas for anatase, the formate bands reach their maximum within 5-10 min. Guided by the results in sections 3.2 and 3.3, we conclude that no (or very little) acetate forms on anatase and brookite. This is further proved by the following indirect argument: If acetate did form during PID of acetone, a buildup of acetate species is expected because decomposition of acetate is a slow process (see also section 3.6). Thus, because acetate is not detected, PID of acetone must occur without acetate as a major intermediate step. The vibration bands that appear on anatase during PID of acetone are instead associated with aqueous formate ions (mainly A40) and surface-coordinated formate (mainly A25, B12, and R20).17,20,48,54 Similar to PID of acetic acid, peaks at ∼1720 cm-1 and a shoulder at ∼1740 cm-1 appear upon illumination, signaling formaldehyde and coordinated CO2 on anatase (A25 and A40). Accumulation of µ-acetate and µ-formate on rutile suggests parallel PID reaction pathways because µ-acetate is photostable (section 3.3): one pathway occurring via acetate and one where acetone is converted to formate via

Adsorption and PID of Acetone and Acetic Acid on TiO2

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14129

Figure 9. In situ FTIR transmission spectra at different times (min) during photodegradation of acetone on (a) anatase (A25), (b) anatase (A40), (c) brookite (B12), and (d) R20 rutile. Simulated solar light with Ptot ) 165 mW cm-2 was employed.

Figure 10. Schematic drawing of the reaction pathways for photoinduced decomposition of acetone on anatase, brookite, and rutile TiO2. The bidentate coordination of formate is tentative and is drawn as a weakly bonded µ-coordinated species (dashed lines). Bold and normal font styles for carbon denote β-carbon (C) and R-carbon (C), respectively.

abstraction of the methyl groups into the gas-phase, in agreement with recent UHV studies.52 In atmospheric studies, it is expected that abstracted methyl groups can be readsorbed on the TiO2 nanoparticles. This is supported by studies of PID of acetone on TiO2 at atmospheric pressures with the simultaneous monitoring of the gas phase by FTIR spectroscopy, where no traces of methyl groups in the gas phase were detected.51 In Figure 10, a schematic drawing of proposed reaction pathways for PID of acetone on the different TiO2 polymorphs is presented. Finally, we note that, because no methyl groups were observed in the gas phase by Herna´ndez-Alonso et al51 and

during PID of acetone on TiO2 (anatase, P25), it is expected that the dissociated methyl groups must readsorb on TiO2 as methoxy species (CH3O-Ti) and subsequently convert to formate. In the work by Rusu and Yates,53 small amounts of adsorbed methoxy were observed on TiO2 (anatase, P25) with IR absorbance bands at ca. 1058, 1115, and 2830 cm-1. In our experiments, we do not observe any of these adsorption bands. This indicates that the methoxy species, if present, are shortlived and rapidly converted to formate. This is supported by the findings of Muggli and Falconer,55 where they reported on the adsorption and PID of acetic acid on TiO2 (anatase, P25) in the presence and absence of oxygen. They observed that, in the absence of oxygen, the β-carbon forms CH4 and C2H6. In the presence of oxygen, CH4 forms at a substantial lower rate and the O2 oxidizes the β-carbon to form CO2. They concluded that the oxygen quickly reacted with surface-adsorbed methyl groups before it is protonated into CH4. In Table 3, vibration frequencies and mode assignments of reaction intermediates for PID of acetone are compiled. In particular, we find that the splitting of the formate asymmetric and symmetric ν(COO) bands that appear during PID exhibit the same trend as seen for acetate; that is, the splitting is largest for anatase (230 cm-1), decreases to 200 cm-1 for brookite, and is smallest for rutile (188 cm-1). Again, the main intermediate surface PID products are associated with ion-like species on anatase, whereas it has covalent, bidentate bonding character for brookite and rutile, supporting a stronger bonding of the PID surface products in the order anatase (A40) < anatase (A25) < brookite (B12) < rutile (R20).20,43,48 3.6. PID Rates and Quantum Yield. The number of converted (photodecomposed) molecules per absorbed photon, or quantum yield Φ, was determined for the different TiO2 samples using the mode-resolved FTIR spectra as a function of

14130

¨ sterlund Mattsson and O

J. Phys. Chem. C, Vol. 114, No. 33, 2010

TABLE 4: Number of Absorbed Acetate, NAcetate, and Acetone Molecules, NAcetone; Photons Absorbed in the Films; the PID Rate Constant, kdec, for Acetic Acid, ν(C-C) + δ(CH3), and Acetone, ν(C-C); and the Quantum Yield, Φ, on Different Brookite, Anatase, and Rutile TiO2 Samples. The Total Number of Absorbed Photons, ∫Fphabs(λ)dλ, above Eg and over the Entire UV-vis Region (Values in Parentheses) Are Shownd sample anatase (A25) anatase (A40)a brookite (B12) rutile (R20)

Nacetate × 1016 Nacetone × 1015 ∫Fphabs(λ)dλ × 1018 (s-1 cm-2) (cm-2) (cm-2) 2.81 2.96 1.16 9.84

7.61 6.69 4.69 5.97

5.5 (5.5) 4.7 (4.7) 3.3 (4.3) 9.5 (10.1)

PID rate constant kdec (min-1) acetate 0.03 0.08 0.04 0b

acetone 0.24 0.64c 1.02c 0.17

Φ (molecules photon-1) acetate

acetone

-6

2.6 × 10 1.1 × 10-5 (1.1 × 10-5) -6 8.4 × 10 3.0 × 10-5 (3.0 × 10-5) 2.3 × 10-6 (1.8 × 10-6) 4.8 × 10-5 (3.7 × 10-5) b 3.5 × 10-6 (3.3 × 10-6)

a Eg ) 3.2 eV is used (unpublished data). b No measured decrease of acetate during UV illumination. c The first 4 min. d The PID rates were obtained after the first 10 minutes and with Pout ) 165 mW cm-2.

Figure 11. Evolution of the normalized ν(C-C) acetone peak area (a) and the ν(COO) acetate peak amplitude (b).

illumination time.56 The number of absorbed photons per unit abs (λ), was determined according to area and time, Fph -R(λ)ts Fabs ) ph (λ) ) Fph(λ)(1 - e

(2)

where R(λ) is the measured absorption coefficient (Figure 3), ts the sample thickness, Fph(λ) ) Ee(λ)λ2/hc, where Ee(λ) is the irradiance at wavelength λ, h is Planck’s constant, and c is the speed of light. The spectral data of the lamp and filters were used to determine Ee(λ). The experimental data shown in Figure 3 were used for the absorption coefficient and optical band gap, and the film thickness is given in Table 1. The number of adsorbed molecules per unit area, Nx, was determined from the integrated acetone ν(C-C) absorbance band and the amplitude of the acetate νa(COO) using the measured acetone and acetate calibration curves (section 2). The number of adsorbed acetone and acetate molecules per unit area for the different samples is shown in Table 4. The number of converted molecules per time and unit area, Kxdec, is then obtained from the rate constant

Kxdec ) SkxdecNx

(3)

where x is the adsorbed species (acetone or acetate) and S accounts for stoichiometry. S ) 1 for acetic acid, and S ) 2 for acetone (see below). Combing eqs 2 and 3, the quantum yield, Φ, is determined according to

Φ)

SkxdecNx λg

∫ Fabs ph (λ)dλ 0

where λg ) hc/Eg and is shown in Table 4.

(4)

Examining the spectra in Figure 9, it can be seen that absorptions bands due to acetone are completely removed within only a few minutes of illumination on all samples, except on rutile, where remains of acetone can be seen during the first 20 min. This is also evidident in Figure 11, where the acetone ν(C-C) integrated band area versus time is shown. The acetone degradation rate, kdec [min-1], was determined from the integrated absorbance, A(t), of the ν(C-C) vibration at ∼1240 cm-1 versus time, assuming first-order reaction kinetics, viz. A(t) ) A0e-kt (Table 4). Because kdec is determined for the ν(C-C) absorption band and involves two C-C bonds, S ) 2 in eq 4.18 From the rate constants, it can be concluded that brookite exhibits the highest degradation rate, kdec ) 1.0 min-1, almost twice as high as the best anatase sample (which, in turn, is more efficient than P25).18,21 When the different samples are compared, it can be seen that the brookite sample exhibits the largest Φ. Rutile has the lowest Φ, approximately an order of magnitude lower than that of brookite. The measured PID rates, kdec, and calculated Φ using eq 4 are shown in Table 4. For acetic acid, the intensity of the νs(COO) absorption bands at ∼1450 cm-1 in Figure 5 due to coordinated acetate was used to extract PID rates and Φ. The results are shown in Table 4. Φ is lower for brookite than the smaller anatase particles but significantly higher than for rutile, which showed no PID of acetate. Both Φ and, especially, kdec are lower for acetate than acetone, a factor of ∼4 for the anatase samples and a factor of ∼20 for brookite, indicating that brookite has a clearer preference for acetone decomposition. For the two anatase samples (A40 and A25), the ratio of the quantum yields for acetone and acetate, Φ(acetone)/Φ(acetate), is similar, meaning that the different particle sizes do not influence the preference for degradation of the different molecules. Because acetate is rate-determining (and not photodegraded at all on rutile), acetate is not an intermediate specie created during PID of acetone; otherwise, a buildup of acetate

Adsorption and PID of Acetone and Acetic Acid on TiO2 would be seen in the IR spectra due to the lower Φ for this molecule. On the basis of the results in sections 3.2-3.5, we propose that the main reason that Φ(acetate) for brookite is slightly lower compared with anatase (but larger compared with rutile) is because these results are based on the analysis of coordinated acetate only. Brookite contains a larger fraction of bidentate acetate species compared with anatase, presumably due to the special surface structure of the dominant {210} surfaces, as discussed in section 3.2. Rutile has a lower band gap than anatase and brookite (Figure 3b), and this is known to result in less favorable matching of the CB with the O2 affinity level,2-6 which is the primary reduction pathway in our experiments and source of oxygen radicals. This explains qualitatively the generally inferior activity of rutile compared with the other polymorphs. However, the lower radical generation on rutile is augmented by the formation of µ-acetate on the rutile surface, which does not photodegrade at all during our experiments. In contrast, for acetone, Φ(acetone) is higher for brookite compared with anatase and rutile due to a combination of two effects: (i) facile dissociation and photooxidation of the R carbon to CO2, which avoids formation of surface-coordinated acetate (as it does on rutile) (Figure 10), and (ii) the moderate bonding strength of PID intermediates to the brookite surface (between anatase and rutile), which is weaker than that to rutile, thus avoiding µ-acetate and µ-formate formation (slow oxidation steps), and higher than that to anatase, where H2O (PID produced and adsorbed trace water from the gas atmosphere) displaces adsorbed molecules (as evidenced by the formation of ionic species) and results in a less efficient interfacial charge transfer of the photogenerated electrons and holes. We emphasize that the Φ values reported here represent intrinsic materials properties and are independent of, for example, film thickness, porosity, and number of adsorption sites because they rely on absolute measurements of R(λ) and Nx of samples with known thicknesses and calibrated incident photon power. Furthermore, if the integration in eq 4 is carried out over the whole measured wavelength interval (and not only up to the extrapolated Eg), it results in only small changes in Fphabs and, hence, Φ; for anatase, no changes were observed (Table 4). This shows that the reported Φ values are insensitive to assumptions about the particular choice of the excitonic model to interpret optical data (indirect electronic transition in this case). We, therefore, conclude that approximate methods to assess Eg may be applicable to estimate Φ provided that Nx is known. 4. Conclusions The adsorption and PID of acetone and acetic acid were studied on anatase, rutile, and brookite TiO2 nanoparticles prepared by solution-based methods. All samples were phase pure, as concluded by Raman and XRD measurements. Thermal stability measurements show that brookite is thermodynamically stable and transforms into rutile by first transforming to the anatase phase. The UV-vis optical properties of brookite are similar to those of anatase with similar band-gap energies, and the band gap excitation is interpreted as an indirect interband transition. Analysis of chemical shifts in FTIR spectra indicates that acetone and acetic acid bond weaker to brookite than to rutile but stronger than to anatase. PID of acetone and acetic acid demonstrates polymorph-dependent reaction pathways. Acetate is not an intermediate species on anatase and brookite when acetone is photodecomposed, whereas on rutile, coordinated

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14131 µ-acetate is formed. Decomposition of µ-acetate is rate determining for PID of acetic acid on rutile and is not photodegraded even after 60 min of illumination. Formate is a main intermediate surface product on all polymorphs. Coordinated formate is preferred on rutile and brookite, whereas formate ions form on anatase, in particular, the large anatase particles, which is manifested by the typical upshift in the νa(COO) peaks splitting toward higher wavenumbers and larger ∆ν(COO). On anatase, more molecular water evolves on the surface compared with brookite and rutile, which displaces acetic acid and impedes the interfacial charge transfer to the adsorbed organic molecules. In general, the surface FTIR spectra show that brookite exhibits large similarities with anatase compared with rutile but with larger amounts of surface-coordinated species and less molecular water. A higher quantum yield for the PID of acetone is calculated for brookite compared with the other samples, whereas for coordinated acetate, the quantum yield is slightly lower on brookite compared with anatase. In all cases, rutile is inferior to the other polymorphs. Acetate shows a substantially lower degradation rate compared with acetone, with rutile exhibiting no photoactivity for acetate at all. In general, the observed PID rate correlates with the bonding strength and coordination of acetate and formate: strongly bonded µ-formate and µ-acetate species impede the reaction, whereas a too weak bonding results in site competition with water, ion formation, and a slower interfacial transfer rate of photoelectrons to the organic molecules. The results suggest that brookite TiO2 can be an interesting candidate in photocatalytic applications. Acknowledgment. The authors thank G. Westin and M. Leideborg for the sample preparation and fruitful discussions, K. Jansson and S. Bakardjieva for TEM measurements, and C. Lejon for Raman spectroscopy measurements. This work was supported by the Swedish Armed Forces (contract 460-E4031). References and Notes (1) Fujishima, A.; Honda, K. Electrochemical Photolysis of Water at a Semiconductor Electrode. Nature 1972, 238, 37–38. (2) Peral, J.; Domenech, X.; Ollis, D. F. Heterogeneous Photocatalysis for Purification, Decontamination and Deodorization of Air. J. Chem. Technol. Biotechnol. 1997, 70, 117–140. (3) Carp, O.; Huisman, C. L.; Reller, A. Photoinduced Reactivity of Titanium Oxide. Prog. Solid State Chem. 2004, 32, 33–177. (4) Diebold, U. The Surface Science of Titanium Dioxide. Surf. Sci. Rep. 2003, 48, 53–229. (5) Linsebigler, A. L.; Lu, G.; Yates, J. T., Jr. Photocatalysis on TiO2 Surfaces: Principles, Mechanisms, and Selected Results. Chem. ReV. 1995, 95, 735–758. (6) Fujishima, A.; Zhang, X.; Tryk, D. A. TiO2 Photocatalysis and Related Surface Phenomena. Surf. Sci. Rep. 2008, 63, 515–582. (7) Kominami, H.; Kato, J.; Murakami, S.; Ishii, Y.; Kohno, M.; Yabutani, K.; Yamamoto, T.; Kera, Y.; Inoue, M.; Inui, T.; Ohtani, B. Solvothermal Syntheses of Semiconductor Photocatalysts of Ultra-High Activities. Catal. Today 2003, 84, 181–189. (8) Ozawa, T.; Iwasaki, M.; Tada, H.; Akita, T.; Tanaka, K.; Ito, S. Low-Temperature Synthesis of Anatase-Brookite Composite Nanocrystals: The Junction Effect on Photocatalytic Activity. J. Colloid Interface Sci. 2005, 281, 510–513. (9) Di Paola, A.; Addamo, M.; Bellardita, M.; Cazzanelli, E.; Palmisano, L. Preparation of Photocatalytic Brookite Thin Films. Thin Solid Films 2007, 515, 3527–3529. (10) Zallen, R.; Moret, M. P. The Optical Absorption Edge of Brookite. Solid State Commun. 2006, 137, 154–157. (11) Shibata, T.; Irie, H.; Ohmori, M.; Nakajima, A.; Watanabe, T.; Hashimoto, K. Comparison of Photochemical Properties of Brookite and Anatase TiO2 Films. Phys. Chem. Chem. Phys. 2004, 6, 1359–1362. (12) Addamo, M.; Bellardita, M.; Di Paola, A.; Palmisano, L. Preparation and Photoactivity of Nanostructured Anatase, Rutile and Brookite TiO2 Thin Films. Chem. Commun. 2006, 4943–4945. (13) Kominami, H.; Ishii, Y.; Kohno, M.; Konishi, S.; Kera, Y.; Ohtani, B. Nanocrystalline Brookite-Type Titanium(IV) Oxide Photocatalysts

14132

¨ sterlund Mattsson and O

J. Phys. Chem. C, Vol. 114, No. 33, 2010

Prepared by a Solvothermal Method: Correlation between Their Physical Properties and Photocatalytic Activities. Catal. Lett. 2003, 91, 41–47. (14) Tian, G. H.; Fu, H. G.; Jing, L. Q.; Xin, B. F.; Pan, K. Preparation and Characterization of Stable Biphase TiO2 Photocatalyst with High Crystallinity, Large Surface Area, and Enhanced Photoactivity. J. Phys. Chem. C 2008, 112, 3083–3089. (15) Ardizzone, S.; Bianchi, C. L.; Cappelletti, G.; Gialanella, S.; Pirola, C.; Ragaini, V. Tailored Anatase/Brookite Nanocrystalline TiO2. The Optimal Particle Features for Liquid- and Gas-Phase Photocatalytic Reactions. J. Phys. Chem. C 2007, 111, 13222–13231. (16) Li, W. K.; Gong, X. Q.; Lu, G.; Selloni, A. Different Reactivities of TiO2 Polymorphs: Comparative DFT Calculations of Water and Formic Acid Adsorption at Anatase and Brookite TiO2 Surfaces. J. Phys. Chem. C 2008, 112, 6594–6596. ¨ sterlund, L. In Solar Hydrogen and Nanotechnology; Vayssieres, (17) O L., Ed.; Wiley: Singapore, 2009. ¨ sterlund, (18) Mattsson, A.; Leideborg, M.; Larsson, K.; Westin, G.; O L. Adsorption and Solar Light Decomposition of Acetone on Anatase TiO2 and Niobium Doped TiO2 Thin Films. J. Phys. Chem. B 2006, 110, 1210– 1220. (19) Pottier, A.; Chane´ac, C.; Tronc, E.; Mazerolles, L.; Jolivet, J.-P. Synthesis of Brookite TiO2 Nanoparticles by Thermolysis of TiCl4 in Strongly Acidic Aqueous Media. J. Mater. Chem. 2001, 11, 1116–1121. (20) Rotzinger, F. P.; Kesselman-Truttman, J. M.; Hug, S. J.; Shklover, V.; Gra¨tzel, M. Structure and Vibrational Spectrum of Formate and Acetate Adsorbed from Aqueous Solution onto the TiO2 Rutile (110) Surface. J. Phys. Chem. B 2004, 108, 5004–5017. ¨ sterlund, L. In Solid State Chemistry of Titanium Dioxide; (21) O Nowotny, J., Ed.; Trans Tech Publications Inc.: Stafa-Zurich, Switzerland, 2009. (22) Penn, R. L.; Banfield, J. F. Oriented Attachment and Growth, Twinning, Polytypism, and Formation of Metastable Phases: Insights from Nanocrystalline TiO2. Am. Mineral. 1998, 83, 1077–1082. (23) Ohsaka, T.; Izumi, F.; Fujiki, Y. Raman Spectrum of Anatase, TiO2. J. Raman Spectrosc. 1978, 7, 321–324. (24) Tompsett, G. A.; Bowmaker, G. A.; Cooney, R. P.; Metson, J. B.; Rodgers, K. A.; Seakins, J. M. The Raman Spectrum of Brookite, TiO2 (Pbca, Z ) 8). J. Raman Spectrosc. 1995, 26, 57–62. (25) Porto, S. P. S.; Fleury, P. A.; Damen, T. C. Raman Spectra of TiO2, MgF2, ZnF2, FeF2, and MnF2. Phys. ReV. 1967, 154, 522–526. (26) Katiyar, R. S.; Krishnan, R. S. The Vibration Spectrum of Rutile. Phys. Lett. 1967, 25A, 525–526. (27) Li, J.-G.; Ishigaki, T. Brookite f Rutile Phase Transformation of TiO2 Studied with Monodispersed Particles. Acta Mater. 2004, 52, 5143– 5150. (28) Busca, G.; Ramis, G.; Amores, J. M. G.; Escribano, V. S.; Piaggio, P. FT Raman and FTIR Studies of Titanias and Metatitanate Powders. J. Chem. Soc., Faraday Trans. 1994, 90, 3181–3190. (29) Kelly, S.; Pollak, F. H.; Tomkiewicz, M. Raman Spectroscopy as a Morphological Probe for TiO2 Aerogels. J. Phys. Chem. B 1997, 101, 2730–2734. (30) Zhang, H.; Banfield, J. F. Understanding Polymorphic Phase Transformation Behavior During Growth of Nanocrystalline Aggregates: Insights from TiO2. J. Phys. Chem. B 2000, 104, 3481–3487. (31) Huberty, J.; Xu, H. Kinetics Study on Phase Transformation from Titania Polymorph Brookite to Rutile. J. Solid State Chem. 2008, 181, 508– 514. (32) Hirano, M.; Nakahara, C.; Ota, K.; Tanaike, O.; Inagaki, M. Photoactivity and Phase Stability of ZrO2-Doped Anatase-Type TiO2 Directly Formed as Nanometer-Sized Particles by Hydrolysis under Hydrothermal Conditions. J. Solid State Chem. 2003, 170, 39–47. (33) Ye, X.; Sha, J.; Jiao, Z.; Zhang, L. Thermoanalytical Characteristic of Nanocrystalline Brookite-Based Titianium Dioxide. Nanostruct. Mater. 1997, 8, 919–927. (34) Roos, A. Use of an Integrating Sphere in Solar Energy Research. Sol. Energy Mater. Sol. Cells 1993, 30, 77–94. ¨ sterlund, L. (35) Topalian, Z.; Niklasson, G. A.; Granqvist, C. G.; O Photo-Fixation of SO2 in Nanocrystalline TiO2 Films Prepared by Reactive DC Magnetron Sputtering. Thin Solid Films 2009, 518, 1341–1344.

(36) Di Paola, A.; Cufalo, G.; Addamo, M.; Bellardita, M.; Campostrini, R.; Ischia, M.; Ceccatp, R.; Palmisano, L. Photocatalytic Activity of Nanocrystalline TiO2 (Brookite, Rutile and Brookite Based) Powders Prepared by Thermohydrolsysis of TiCl4 in Aqueous Solutions. Colloids Surf., A 2008, 317, 366–376. (37) Cassaignon, S.; Koelsch, M.; Jolivet, J.-P. Selective Synthesis of Brookite, Anatase and Rutile Nanoparticles: Thermolysis of TiCl4 in Aqueous Nitric Acid. J. Mater. Sci. 2007, 42, 6689–6695. (38) Zhang, Y.; Wu, L.; Zeng, Q.; Zhi, J. An Approach for Controllable Synthesis of Different-Phase Titanium Dioxide Nanocomposites with Peroxtitanium Complex as Precursor. J. Phys. Chem. C 2008, 112, 16457– 16462. (39) Yoldas, B. E. Investigations of Porous Oxides as an Antireflection Coating for Glass Surfaces. Appl. Opt. 1980, 19, 1425–1429. (40) Li, J.-G.; Ishigaki, T.; Sun, X. Anatase, Brookite, and Rutile Nanocrystals Via Redox Reactions under Mild Hydrothermal Conditions: Phase-Selective Synthesis and Physicochemical Properties. J. Phys. Chem. C 2007, 111, 4969–4976. (41) Liao, L. F.; Lien, C. F.; Lin, J. L. FTIR Study of Adsorption and Photoreactions of Acetic Acid on TiO2. Phys. Chem. Chem. Phys. 2001, 3, 3831–3837. (42) Backes, M. J.; Lukaski, A. C.; Muggli, D. S. Active Sites and Effects of H2O and Temperature on the Photocatalytic Oxidation of 13CAcetic Acid on TiO2. Appl. Catal., B 2005, 61, 21–35. (43) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds, 5th ed.; John Wiley & Sons: New York, 1997. (44) Stein, S. E. Names, Structures, Mass and IR Spectra Data. In Nist Chemistry Webbook [Online]; Linstrom, P. J., Mallard, W. G., Eds.; Nist Standard Reference Database Number 69; National Institute of Standards and Technology, Washington, DC, March 2003. (45) Deacon, G. B.; Phillips, R. J. Relationships between the CarbonOxygen Stretching Frequencies of Carboxylates and the Type of Carboxylate Coordination. Coord. Chem. ReV. 1980, 33, 227–250. (46) Kim, K. S.; Barteau, M. A. Pathways for Carboxylic Acid Decomposition on TiO2. Langmuir 1988, 4, 945–953. (47) Lee, G. D.; Tuan, V. A.; Falconer, J. L. Photocatalytic Oxidation and Decomposition of Acetic Acid on Titanium Silicalite. EnViron. Sci. Technol. 2001, 35, 1252–1258. ¨ sterlund, L. A Comparative (48) van der Meulen, T.; Mattson, A.; O Study of the Photocatalytic Oxidation of Propane on Anatase, Rutile, and Mixed-Phase Anatase-Rutile TiO2 Nanoparticles: Role of Surface Intermediates. J. Catal. 2007, 251, 131–144. (49) Uemura, Y.; Taniike, T.; Tada, M.; Morikawa, Y.; Iwasawa, Y. Switchover of Reaction Mechanism for the Catalytic Decomposition of HCOOH on a TiO2(110) Surface. J. Phys. Chem. C 2007, 111, 16379– 16386. (50) Henderson, M. A. Acetone Chemistry on Oxidized and Reduced TiO2(110). J. Phys. Chem. B 2004, 108, 18932–18941. (51) Herna´ndez-Alonso, M. D.; Tejedor-Tejedor, I.; Coronado, J. M.; Anderson, M. A.; Soria, J. Operando FTIR Study of the Photocatalytic Oxidation of Acetone in Air over TiO2-ZrO2 Thin Films. Catal. Today 2009, 143, 364–373. (52) Henderson, M. A. Relationship of O2 Photodesorption in Photooxidation of Acetone on TiO2. J. Phys. Chem. C 2008, 112, 11433–11440. (53) Rusu, C.; Yates, J. T., Jr. Adsorption and Decomposition of Dimethyl Methylphosphonate on TiO2. J. Phys. Chem. B 2000, 104, 12292– 12298. (54) Liao, L.-F.; Lien, C.-F.; Shieh, D.-L.; Chen, M.-T.; Lin, J.-L. FTIR Study of Adsorption and Photoassisted Oxygen Isotopic Exchange of Carbon Monoxide, Carbon Dioxide, Carbonate, and Formate on TiO2. J. Phys. Chem. B 2002, 106, 11240–11245. (55) Muggli, D. S.; Falconer, J. L. Parallel Pathways for Photocatalytic Decomposition of Acetic Acid on TiO2. J. Catal. 1999, 187, 230–237. (56) Serpone, N.; Emeline, A. V. Suggested Terms and Definitions in Photocatalysis and Radiocatalysis. Int. J. Photoenergy 2002, 4, 91–131.

JP103263N