J. Phys. Chem. B 1998, 102, 1665-1671
1665
ARTICLES Reduction Kinetics of Zeolite-Hosted Mono- and Polynuclear Titanium Oxide Species Followed by UV/Vis Diffuse Reflectance Spectroscopy: Influence of Location and Coordination Gerd Grubert, Michael Wark,* Nils I. Jaeger, and Gu1 nter Schulz-Ekloff Institut fu¨ r Angewandte und Physikalische Chemie, UniVersita¨ t Bremen, Postfach 330 440, D-28334 Bremen, Germany
Olga P. Tkachenko N.D. Zelinsky Institute of Organic Chemistry, Moscow, Russia ReceiVed: September 11, 1997; In Final Form: NoVember 19, 1997
The kinetics of the reversible reduction of faujasite-hosted mononuclear titanium oxide species prepared by reaction of dehydrated zeolites with TiCl4 at different temperatures are (i) evaluated recording the intensity of an absorption band of Ti(III) species at 16 200 cm-1 by UV/vis diffuse reflection spectroscopy and (ii) compared to those of faujasite-hosted polynuclear titanium oxide species, titanosilicates, and bulk anatase. The reduction kinetics of mononuclear Ti(IV)Ox species could be fitted assuming the existence of three types of Ti sites, one being located at the surface of the zeolite NaY crystals and two inside the void structure. The distributions of Ti, i.e., the ratios Tisurface/Tibulk, resulting from this fit agree with those obtained from photoelectron spectra. The reduction rates for different Ti(IV)Ox species increase with increasing excitation energy of the O f Ti charge transfer, i.e., increasing deviation from octahedral symmetry, and decrease with increasing nuclearity of titanium oxide clusters. The response time of the mononuclear Ti(IV)Ox species toward reduction to Ti(III) is shortened by a factor of 60 in comparison to bulk TiO2. Titanium tetrahedrally coordinated in framework positions in TS-1 cannot be reduced.
1. Introduction The reduction of the dimensions of bulk TiO2 into the nanometer regime yields advanced materials for a variety of applications. TiO2 films with a thickness of less than 50 nm show size quantization effects in one dimension1 and have been proposed for applications such as (i) antireflecting coatings,2 (ii) dielectric layers within metal-insulator-semiconductor devices,3 (iii) electrochromic windows,4 and (iv) oxygen sensors.5-10 Size quantization effects in three-dimensional nanometer-sized titanium oxide particles alter strongly the optoelectronic properties of these materials11-13 and render them prospective for an application in optical filters.14,15 Highly dispersed titanium oxide on different supports has been intensively studied in catalysis and photocatalysis.16-21 Titanium-modified zeolites, especially TS-1, have been used to catalyze the mild oxidation of a variety of organic substrates by H2O2, e.g. olefins, alcohols, ketones, alkanes, or aromatics.22-29 Although the sensing properties of bulk TiO2 and TiO2 thin films toward reducing and oxidizing gas atmospheres are wellknown,5-10 only little effort has been made to investigate the redox properties of titanium in zeolite-stabilized titanium oxide clusters. First results indicated that the strongly distorted coordination sphere of mononuclear titanium oxide species anchored in the * Corresponding author. E-mail:
[email protected].
pores of zeolites30 provides activation sites for reducing molecules such as H2, resulting in improved sensing activity.31 In the following the preparation of highly dispersed titanium oxide species (Ti(IV)Ox) (x stands for the oxygen atoms in the first coordination sphere of Ti comprising OH groups as well as oxygen atoms of the zeolite framework) in the pores of faujasites by loading the host with volatile chlorides from the gas phase and subsequent hydrolysis is reported. It will be demonstrated that the reduction of titanium in zeolites can be followed by optical spectroscopy (DRS-UV/vis) and that the location of the titanium oxide species with respect to the zeolite matrix can be quantitatively deduced from the reduction kinetics and X-ray photoelectron spectroscopy (XPS). 2. Experimental Section 2.1. Materials. Ti(IV) oxide species stabilized in a zeolite matrix were prepared applying the method of chemical vapor deposition (CVD).30,32 Zeolite NaY (Si/Al ) 2.7), synthesized using standard recipes,33 was used as a support. The Na/Al ratio was determined to nearly 1. Thus the present OH groups are only due to SiOH groups from structural defects, e.g., hydroxyl nests,34 or by hydrolysis of siloxane bonds.30 The zeolite1g was put in a flow reactor and dehydrated in a stream of dry nitrogen at 673 K for 12 h. Subsequently, the dry zeolite was loaded at 373 and 673 K, respectively, for 30 min in a N2 stream saturated with TiCl4 at ambient conditions.
S1089-5647(97)02973-8 CCC: $15.00 © 1998 American Chemical Society Published on Web 02/14/1998
1666 J. Phys. Chem. B, Vol. 102, No. 10, 1998
Grubert et al.
TABLE 1: Ti(IV)Ox Species in NaY: Sample Code, Temperature (K), and Number of Cycles for Loading with TiCl4, Ti Content (wt %), and BET Value (m2/G) sample
loading temperature/K
number of cycles
Ti content/wt %
BET value/m2/g
TiNaY-1 TiNaY-2 TiNaY-3 TiNaY-4
373 373 373 673
1 2 3 1
1.9 3.9 8.1 1.3
750 580 430 701
To remove excess TiCl4 physisorbed on the zeolite, the reactor was flushed with dry N2 at 373 K for 2 h. Afterward, the TiCl4 chemically bound to OH groups of the zeolite was hydrolyzed at 373 K in a N2 stream saturated with water. Finally, the sample was calcined in a dry O2 stream at 673 K for 12 h. In some cases the procedure was repeated three times. The Ti content was determined by AAS analysis after dissolution of the Ti zeolites in an autoclave in HF:HNO3 ) 1:4 (440 K, 12 h). As reference materials TS-1, prepared according to the wetness impregnation method,35 and anatase (HOM UV100, Sachtleben Co., particle diameter 4-10 nm) were used. The designations and Ti contents of the samples are listed in Table 1. 2.2. Methods. Diffuse reflectance spectra (UV/vis DRS) were recorded from 200 to 800 nm on a Varian Cary 4 spectrometer equipped with a Praying Mantis and a reaction chamber (Harrick) for in situ measurements. All reflectance spectra were converted to the Kubelka-Munk function (F(R∞))36 using LOT 75% (Oriel Co.) as reference sample. As long as the F(R) values are smaller than 3 the F(R) function is proportional to the absorption of the sample. Samples with F(R) values higher than 3 were diluted with parent zeolite (NaY). Measurements concerning the redox kinetics were carried out by in situ UV/vis DRS in the reaction chamber positioned in the Praying Mantis. The samples were pressed into 1-1.5mm-thick wafers, which could be regarded as infinitely thick with respect to the Kubelka-Munk theory. The gas flows (80 mL/min) through the wafer from top, where the absorption is detected by diffuse reflectance spectroscopy, to bottom. Mixtures of 25 vol % H2 or O2, respectively, in argon were used for reduction and reoxidation. The temperature was varied from 573 to 773 K. At 773 K, i.e., the highest temperature to which the chamber can be heated in the H2 mixture, the most precise kinetic evaluations have been achieved. During reduction and oxidation cycles the reflectance of the samples was continuously recorded at 16 200 cm-1 (datapoint distance 0.5 s.), i.e., the wavenumber at which the greatest changes in the reflectance occur. The samples were repeatedly reduced and reoxidized for many cycles and several days in order to guarantee the reversibility of the reflectivity changes. To obtain similar starting conditions for all samples, they were pretreated by dehydration in a vacuum at 673 K (heating rate 2 K/min) for 12 h followed by an oxidation in a stream of 25 vol % O2 in Ar (80 mL/min) at 823 K for 30 min. In some cases, hydrogen was activated on Pt/NaX zeolite positioned as a thin layer below the sample wafer. The formed activated hydrogen diffuses via spillover to the titanium oxide species. To control the crystallinity of the Ti-zeolites, (i) X-ray diffractograms (XRD) were recorded in a Seiffert diffractometer with Bragg-Brentano geometry using Cu KR radiation under ambient conditions and (ii) adsorption measurements were carried out with a Grimm BET automatic surface analyzer by physisorption of N2 at 77 K.
Room-temperature Raman spectra were taken for undiluted samples in a fine capillary of a Jobin Yvon (Division Instruments) spectrometer using a Stabilite 2016 (Spectra Physics) laser at 514 nm for detection. The characterization by photoelectron spectroscopy (XPS) was performed in a Kratos XSAM 800 spectrometer using Mg or Al KR radiation according to a procedure described elsewhere.37 To avoid possible X-ray-induced changes in the materials, the X-ray power was restricted to 10 kV and 100 mA (100 W). The vacuum applied during the measurements was better than 7 × 10-7 Pa. The data were accumulated in separate regions for normally 1 h per region. To consider a possible charging of the insulating zeolite samples, the measured binding energies were corrected by calibration to the carbon signal of hydrocarbon contaminations from the pump oil deposited on the surface of the zeolite crystals, i.e., C 1s ) 284.5 eV. Additionally, the Si 2p line (103.0 eV) of Y-zeolites was used as a second standard. Elemental ratios have been calculated from the detected signal intensities under consideration of the photoionization cross sections, given by Scofield.38 Titanium concentrations as low as 1 at. % could be reliably detected. 3. Results 3.1. Zeolite Crystallinity. The samples treated only once with TiCl4 (TiNaY-1 and TiNaY-4, Table 1) exhibit only a slight (10-15%) decrease of the BET surfaces compared to the parent NaY zeolite (820 cm2/g), indicating a negligible influence on the crystallinity of the zeolite. In the X-ray diffractograms the absolute and relative intensities of the zeolite Y reflections remain constant, indicating a random distribution of the Ti(IV)Ox species. However for samples where a higher loading is achieved in repeated treatment cycles, the BET surface decreases to 580 m2/g (TiNaY-2) after the second cycle of gas-phase loading and hydrolysis and to 430 m2/g (TiNaY-3) after the third cycle. In the X-ray diffractograms the intensities of Y-zeolite reflections slightly decreased, but no reflections of titanium oxide appear. The drop in the crystallinity can be attributed to the influence of HCl formed during the hydrolysis step, resulting in a break of Al-O-Si bonds in the zeolite framework, the formation of new hydroxyl groups or hydroxyl nests, respectively, and, finally, a local fragmentation of the zeolite framework and the appearance of mesopores.34,39,40 3.2. Nature of Ti(IV)Ox Species. In none of the samples were reflections of anatase or rutile detected by XRD. Furthermore, the absence of anatase is confirmed by the absence of the 144-cm-1 band in the Raman spectra. Figure 1 represents normalized UV/vis DRS spectra after rehydration and calcination of (i) TiNaY-1 and TiNaY-4, loaded in one cycle at different temperatures, (ii) TiNaY-2 and TiNaY-3 loaded at 373 K repeatedly in two or three cycles, respectively, and (iii) TS-1 and bulk anatase for comparison purposes. For all the samples the absorption is significantly blue-shifted compared to that of bulk anatase. The points of intersection between the tangents through the point of inflection in the absorption edges and the abscissa were found around 29 850 cm-1 (TiNaY-1) and 31 500 cm-1 (TiNaY-4) in comparison to about 27 500 cm-1 for bulk TiO2 (Table 2). The TS-1 starts to absorb at around 33 000 cm-1 and has a maximum at 49 500 cm-1, in agreement with previous reports41-43 and calculations for titanium in a tetrahedral environment.44 For the TiNaY samples two main regions of absorption were found with maxima around 34 000 and 45 000 cm-1. Klaas et al. assigned these bands to mononuclear octahedral Ti(IV)Ox species either
Kinetics of Zeolite-Hosted Titanium Oxide Species
J. Phys. Chem. B, Vol. 102, No. 10, 1998 1667
Figure 1. Normalized UV/vis DRS spectra recorded at room temperature of singly loaded TiNaY-1 (373 K, A) and TiNaY-4 (673 K, B), doubly loaded TiNaY-2 (373 K, C), triply loaded TiNaY-3 (373 K, D), bulk anatase (E), and TS-1 (F).
Figure 2. UV/vis DRS spectra of TiNaY-1 (A) and bulk anatase (B) after reduction (Red) and reoxidation (Ox) in the absorption region of the Ti(III) species.
TABLE 2: Points of Intersections between the Tangents through the Points of Inflection in the Absorption Edges and the Abscissa, Binding Energies (BE), and Ti/Si Ratios on the Surface Determined from XPS and in the Bulk from AAS of TiNaY and Reference Samples sample TS-1 TiNaY-1 TiNaY-2 TiNaY-3 TiNaY-4 bulk anatase
point of intersection/cm-1 BE Ti 2p3/2/eV (Ti/Si)surface (Ti/Si)bulk 42 250 29 850 29 400 29 150 31 500 27 450
459.9 459.1 459.0 459.2 459.2 458.5
0.028 0.534 0.214 0.224 0.221
0.032 0.050 0.108 0.215 0.045
monofunctionally bound to isolated OH groups in the zeolite (band around 36 000 cm-1) or bifunctionally attached to vicinal silanol groups (45 000 cm-1).30 The relative intensity of the band at 45 000 cm-1 increases with higher loading temperature. Multiple-loaded samples (TiNaY-2 and TiNaY-3) exhibit steeper absorption edges compared to the 1-fold-loaded TiNaY-1 (Figure 1). The position of the points of intersection of the tangents with the abscissa was found to be red-shifted compared to that of TiNaY-1. However, they remained blue-shifted relative to bulk anatase, indicating a growth of the initially monofunctionally bound Ti(IV)Ox species to clusters with sizes in the nanometer regime, which still show size quantization effects (Table 2). The quantification of the intensities of the XPS signals of Ti 2p3/2 electrons in the Ti(IV)Ox species and Si 2p electrons of the zeolite shows that the observed Ti/Si in the rim of the zeolite crystals, which is sensitive to XPS ((Ti/Si)surface), is for all the Y-zeolite samples increased in comparison to the average value determined by AAS analysis after dissolution of the crystals in HF/HNO3 ((Ti/Si)bulk) (Table 2). In contrast to TS-1, where both ratios are nearly equal, the titanium species are enriched at the outer surfaces of the NaY zeolite crystals by the CVD loading. 3.3. Reduction and Reoxidation. UV/vis DRS spectra recorded for TiNaY samples and reference materials after reduction with 25 vol % H2 in Ar for 2 h and subsequent reoxidation with 25 vol % O2 in Ar for 0.5 h are presented in Figure 2. As observable from the inset, the reduction of the TiNaY samples leads to the occurrence of a shoulder in the absorption ranging from about 18 200 cm-1 to about 16 100 cm-1. For bulk anatase a continuous increase of the F(R) values at wavenumbers lower than about 18 000 cm-1 was found. TS-
Figure 3. Absorption in Kubelka-Munk units at 16 200 cm-1 with time in dependence on the presence of reductive or oxidative atmospheres at 773 K: TiNaY-1 (A), TS-1 (B), and bulk anatase (C).
1, however, shows no significant differences in the KubelkaMunk spectra between the reduced and the oxidized material. The evolution of the absorption intensity at a fixed wavenumber (16 200 cm-1) with time during reduction and reoxidation is depicted in Figure 3. For TiNaY-1 the absorption increases monotonically during reduction with H2 and drops quickly after switching to O2. The changes in F(R) are completely reversible following a complete cycle and are proportional to the concentration of reduced Ti(III) in the sample. Changes in the absorption intensity during reduction can be clearly distinguished from the noise signal after 5-10 s, whereas signal alterations appear within 1 s after exposure to oxygen. In the case of TS-1 the absorption intensity remains almost constant during reduction and oxidation, indicating that tetrahedrally coordinated titanium atoms in the framework are not reducible. For bulk anatase an induction period was found for the reduction. After exposure to hydrogen the absorption increases much slower compared to TiNaY-1, and even after 10 min, the absorption increased to only about 10% of the value observed for TiNaY-1. Compared to bulk anatase the reduction rate of zeolite-hosted Ti(IV)Ox species is strongly enhanced and a much higher percentage of the present Ti(IV) atoms are reduced. By comparison of the samples TiNaY-1, TiNaY-3, and bulk anatase (Figure 4), it is demonstrated that the length of the induction period increases with the number of loadings
1668 J. Phys. Chem. B, Vol. 102, No. 10, 1998
Figure 4. Absorption in Kubelka-Munk units at 16 200 cm-1 with time during reduction at 773 K for samples containing titanium oxide of different type: TiNaY-1 (Ti(IV)Ox, A), TiNaY-3 (Ti(IV)xOy (x > 1), B), and bulk anatase (C).
Figure 5. Absorption in Kubelka-Munk units at 16 200 cm-1 with time during reduction at 773 K: TiNaY-1 without (A) and with spillover hydrogen (A + Pt/NaX) and TiNaY-4 without (B) and with spillover hydrogen (B + Pt/NaX).
with TiCl4 or with the red-shift of the absorption edges (Figure 2), respectively. In Figure 5 the reduction behavior of TiNaY-1 is compared with that of TiNaY-4, which was loaded with TiCl4 at higher temperature. Since sample TiNaY-1 contains more reducible species (Table 1), its absorption intensity at the end of the reduction period, representing the formed Ti(III), is higher compared to that of TiNaY-4. However, the reaction rate is higher in the latter sample, where almost all reducible titanium has reacted within less than 3 min, indicating that loading at higher temperatures generates species that can be reduced faster. For this sample (Figure 5B) the reduction rate could not be enhanced using spillover hydrogen, whereas in TiNaY-1 the fraction of Ti(IV)Ox species, rapidly reduced within the first 3 min, was almost doubled using the activated hydrogen. 4. Discussion 4.1. Nature of Ti(IV)Ox species. During the first loading TiCl4 binds under evolution of HCl to one or more OH groups of the NaY zeolite. After hydrolysis the resulting isolated Ti(IV)Ox species are mononuclear; that is, no Ti-O-Ti bonds are formed. The absorption in the UV region originates from electron charge-transfer transitions from oxygen 2p levels to titanium 3d levels and exhibits band positions depending on
Grubert et al. the oxygen coordination sphere around the titanium.45 For Ti(IV)Ox species bound monofunctionally to one zeolitic OH group (Ti(IV)1) the transition energy is smaller than for bi- or multifunctionally bound ones (Ti(IV)2), where two or more of the oxygen atoms in the coordination sphere are part of the zeolitic framework.46,47 For the latter the deviation from the octahedral coordination sphere is more pronounced, and consequently the excitation energy is higher. Since the intensity ratio for the absorption at 45 000 cm-1 (absorption of monoand bifunctionally bound TiOx species) and 34 000 cm-1 (only absorption of monofunctionally bound species) is higher in sample Ti-NaY-4 in comparison to Ti-NaY-1 (Figure 1), it can be concluded that at higher loading temperatures the tendency of TiCl4 to bind bifunctionally to vicinal OH groups is enhanced. As reported in a previous paper, this tendency is independent of the Ti content.30 At higher temperatures the preferred binding of TiCl4 at vicinal OH groups in the zeolite pores is presumably facilitated due to the higher flexibility of the zeolite framework. The application of molecular dynamics simulations has demonstrated that the distribution of the instantaneous window diameters in zeolite NaY is widened with higher temperatures.48 Taking XPS data (Table 2), which give the strongest enrichment for Ti at the surface for TiNaY-1, also into account, it can be assumed that the monofunctionally bound species (Ti(IV)1) are formed in the pores (Ti(IV)1p) as well as at the outer surface of the zeolite crystals (Ti(IV)1s). At the surface the preferred monofunctional binding of TiCl4 at isolated surface OH groups will occur with high probability, whereas a bifunctional binding at the surface seems to be relatively unlikely since for geometrical reasons the distance between adjacent OH groups should be larger at the outer surface than in the pores. Comparing the (Ti/Si)surface ratios for the samples loaded several times at 373 K, surprisingly a depletion of titanium from the outer surface was found after the second cycle. During the second and third loading the TiCl4 reacts preferentially with OH groups of the already zeolite-bound mononuclear Ti(IV)Ox species under formation of Tix(IV)Oy clusters with x > 1. It seems that these clusters migrate deeper into the zeolite possibly because the interaction between the clusters and the zeolite is stronger and energetically preferred in the pores due to multifunctional anchoring. For none of the samples could a band at 144 cm-1, which is very sensitive in the detection of even traces of anatase,49 be observed in the Raman spectra. From this absence of centrosymmetry it might be deduced that the numbers of x and y in the formed TixOy clusters are very small. The existence of a centrosymmetry is required for the occurrence of Raman-active vibrations. It can, hence, be assumed that the clusters possess titanium atoms with strongly distorted octahedral or even lower coordination due to strong interactions with the zeolite matrix.30 The absence of tetrahedrally coordinated titanium in the TiNaY samples was checked by IR spectroscopy. A band at 960 cm-1, typical for framework titanium in tetrahedral coordination in titanium silicates,50 was only found in sample TS-1 but in none of the TiCl4-loaded Y-zeolites. An influence of the coordination of the titanium or at least of the interaction with the zeolite matrix is also discernible from the binding energies of the Ti 2p3/2 electrons measured by XPS (Table 2). In accordance with literature data51,52 for the tetrahedrally coordinated titanium in sample TS-1 a binding energy of 459.9 eV was observed, which is 1.4 eV higher than for titanium in octahedral coordination in bulk anatase (458.5 eV). For all Y-zeolites modified with TiCl4 the binding energies lie between these values (459.0-459.2 eV), indicating a strongly
Kinetics of Zeolite-Hosted Titanium Oxide Species
J. Phys. Chem. B, Vol. 102, No. 10, 1998 1669
distorted octahedral oxygen environment around the titanium. Similar shifts have been reported for titania-doped glasses in which the coordination number of titanium with oxygen was reduced.53 4.2. Reduction and Reoxidation. The occurrence of an absorption band between 16 000 cm-1 and 18 200 cm-1 for the reduced samples TiNaY-1 and TiNaY-4 can be attributed to the formation of Ti(III) species in accordance with literature data on molecular Ti(III) species in Ti(III)-hexaquo complexes in aqueous solutions or Ti(III)-loaded phosphate glasses, where absorption bands for d-d-electron transitions (2t2g f 2eg) around 17 700 cm-1 have been reported.54 The reduction of TixOy clusters with x > 1 or bulk anatase leads to the appearance of oxygen vacancies of which the concentration depends on the oxygen partial pressure over the sample. In bulk anatase the energy band of the oxygen vacancies lies about 1 eV under the conduction band, resulting in an absorption band at 8300 cm-1 for the excitation of the electron transfer from vacancies to the conduction band.55 The increase of this broad absorption band corresponds to a decrease of the electrical resistance of bulk TiO2.56 The observed increase in the F(R) value starting at 18 000 cm-1 (Figure 2, inset) represents the left flank of this absorption band. Kinetic Model for the Reduction of Mononuclear Ti(IV)Ox in Zeolite Y. For the samples showing no induction period for the reduction, the kinetics can be identified applying a simple model, which is presented in the following. From Figure 3 it is obvious that the oxidation of the Ti(III) occurs faster than the reduction of the Ti(IV) species. Since the collisional cross section of O2 (L ) 0.4 nm2) is larger than that of H2 (L ) 0.27 nm2),57 a limitation of the kinetics by diffusion of the gases through the zeolitic pore system can be ruled out. According to the discussion of the UV/vis DRS and the XPS data, at least three chemically different mononuclear Ti(IV)Ox species can be distinguished, i.e., (i) monofunctionally bound on the external surface of the zeolite (Ti(IV)1s), (ii) monofunctionally bound in the zeolite pores (Ti(IV)1p), and (iii) bifunctionally bound in the pores (Ti(IV)2p). Correspondingly, the semilogarithmic plot, i.e., ln{[Ti(IV)]0/ [Ti(IV)]0 - [Ti(III)])} versus time, where the concentration of Ti(III) is obtained from the absorption at 16 200 cm-1, with progressing reduction could be divided into three linear parts (Figure 6). Considering that the hydrogen is present in large surplus, the reduction can be described as a sum of three independent reactions with pseudo-first-order kinetics. Since it is well-known that the dissociative adsorption of hydrogen is the rate-limiting step for the reduction of many metal oxides,58 and under the assumption that the OH groups in the coordination sphere of Ti are involved in the reaction, it can be assumed that for each type of Ti(IV)Ox species the following reaction proceeds:
Ti(IV)(OH)x + H2 f Ti(III)(OH)x-1 + H2O + H• rate-limiting step Ti(IV)(OH)x + H• f Ti(III)(OH)x-1 + H2O
fast
In summary,
2Ti(IV)(OH)x + H2 f 2Ti(III)(OH)x-1 + 2H2O For the temporal development of the overall concentration of Ti(III) the following mass balance equation holds:
Figure 6. Semilogarithmic scale for the development of the F(R) value with time during reduction of TiNaY-1 (A) and TiNaY-4 (B).
[Ti(III)] ) [Ti(IV)1s]0 + [Ti(IV)1p]0 + [Ti(IV)2p]0 [Ti(IV)1s]0 exp(-k1t) - [Ti(IV)1p]0 exp(-k2t) [Ti(IV)2p]0 exp(-k3t) (1) The index 0 denotes the initial concentration of the reducible Ti(IV)Ox species on the different positions. Figure 7 shows the nonlinear least-squares fits according to a Levenberg-Marquardt algorithm59 resulting from eq 1 for the reduction of TiNaY-1 and TiNaY-4. The fitted kinetic parameters and the resulting initial concentrations for the different Ti(IV)Ox species Ti(IV)1s, Ti(IV)1p, and Ti(IV)2 expressed in wt % are listed in Table 3. The assignment of the species Ti(IV)1s to a location at the outer surface of the zeolite crystals seems to be justified since their relative concentrations determined from the reduction kinetics agrees well with that estimated by XPS. The obtained percentages of surface Ti(IV)Ox species with respect to the total titanium content are listed in Table 4. For the calculation of the value from XPS the detection depth for the Ti 2p and Si 2p photoelectrons due to their limited probability of escaping the zeolite has been estimated to be about 10 nm. This depth might appear too high compared to values given in the literature.60 However, one has to consider the low density of highly porous zeolites, thus enlarging the free path of the electrons. From HREM micrographs it was obvious that the zeolite Y octahedra were partially grown into each other, resulting in particles having a mean diameter of about 1.5 µm and being of more spherical morphology. For the calculation of the volumes of the zeolite Y particles accessible for XPS a spherical shape was assumed for simplification. From this calculation it results that the Ti/Si ratio of the surface region found by XPS is valid for less than 4% of the total volumes of the crystals. From this rough estimate follows that the absolute values of the Tisurface/Titotal ratios are arbitrary but that the comparison between the samples TiNaY-1 and TiNaY-4 are reflected by the XPS measurements and the evaluation of the kinetics as well. The calculated ratio of concentrations for the species (Ti(IV)1p) in the samples TiNaY-1 and TiNaY-4 (Table 3) agrees with the ratio of absorption intensities at 34 000 cm-1 for TiNaY-1 and TiNaY-4 (Figure 1). The order of the reduction rates k2p . k1p > k1s evaluated with highest accuracy at 773 K (Table 3) was found to be valid over the whole measured temperature range (573-773 K). Therefore it can be assumed that the stronger symmetry distortion, imposed on the bifunctionally bound species (Ti(IV)2p), has a larger impact on the
1670 J. Phys. Chem. B, Vol. 102, No. 10, 1998
Grubert et al.
Figure 7. Kinetic analysis of the reduction of TiNaY-1 (A) and TiNaY-4 (B). The dotted line gives the curve fit.
TABLE 3: Parameters of Reduction Kinetics for Mononuclear Ti(IV)Ox Evaluated at 773 K: Initial Concentrations of Species Monofunctionally Bound at the Surface (Ti(IV)1s) or in the Pores ((Ti(IV)1p) or Bifunctionally Bound in the Pores (Ti(IV)2p) of NaY and Corresponding Rate Constants sample
[Ti(IV)1s]0/ [Ti(IV)1p]0/ [Ti(IV)2p]0/ wt % wt % wt % k1s/s-1 k1p/s-1 k2/s-1
TiNaY-1 TiNaY-4
0.62 0.20
1.19 0.53
0.09 0.57
0.13 0.13
0.03 0.03
1.60 1.60
TABLE 4: Fraction of Titanium in Ti(IV)Ox Species at the Surface of NaY in Relation to the Total Titanium Content Deduced from XPS and Reduction Kinetics Tisurface/Titotal sample
from XPS
from reduction kinetics
TiNaY-1 TiNaY-4
0.42 0.18
0.37 0.15
reducibility of the mononuclear titanium oxide species than the different location of the monofunctionally bound species, i.e., in the pores (Ti(IV)1p) or at the external surface (Ti(IV)1s). Obviously, increasing distortion of the symmetry of Ti(IV)Ox facilitates the accessibility of the coordination sphere of Ti(IV) for H2 and, thus, increases the reduction probability. Although little is known about the nature of spillover hydrogen, it can obviously increase the rate of reduction of those species exhibiting a lesser degree of symmetry distortion but not that of the strongly distorted ones (Figure 5). Presumably, the polar nature of the spillover hydrogen is able to induce symmetry distortions, facilitating the entrance into the coordination sphere and, thus, the reduction. The kinetics of reduction with spillover hydrogen are assumed to be more complex and, thus, cannot be determined with the simple model. Induction Period for the Reduction of Polynuclear TixOy Clusters and Bulk Anatase. The simple model cannot be applied for the determination of the reduction kinetics in the case of Tix(IV)Oy clusters (x > 1), TiO2 nanoparticles, and bulk anatase, since additional terms considering gas adsorption and surface to bulk vacancy diffusion must be taken into account. The length of the induction period for the occurrence of Ti(III) in the UV spectra increases going from TiNaY-1, with only mononuclear Ti(IV)Ox species, to TiNaY-3, with Tix(IV)Oy clusters (x > 1), to anatase (Figure 4). It is possible to explain qualitatively the occurrence of this induction period. During the reduction of larger TiO2 particles,
oxygen is removed from the surface and oxygen vacancies remain. Because of the measured enhancement of the reduction rate with spillover hydrogen, it can be stated that the reduction rate must be to a large extent determined by the activation of H2 on the surface. This activation of molecular hydrogen, however, occurs at surface vacancies.61,62 At the beginning of the reduction the concentration of surface vacancies is small and consequently the reduction rate is low. With progressing reduction an increasing number of oxygen vacancies are generated, acting as adsorption sites for H2 and, thus, increasing the reduction rate in an autocatalytic process. The larger the cluster the longer the time necessary to achieve a critical concentration of oxygen vacancies, i.e., the longer the induction period. Above the critical concentration the reduction rate is only determined by vacancy diffusion into subsurface layers, i.e., the surface/volume ratio of the particles.63 The high concentration of energetically less favored vacancies in the surface and subsurface at the end of the reduction leads for all samples to the observed extremely high reaction rate during the following oxidation with oxygen. 5. Conclusions The CVD loading of dehydrated faujasites with TiCl4 and subsequent hydrolysis and calcination leads to the formation of mononuclear Ti(IV)Ox species bound monofunctionally to isolated or bifunctionally to vicinal silanol groups of the zeolite. Increasing loading temperature increases the probability of the bifunctional bonding. During consecutive loading cycles the mononuclear species grow to Q-size Ti(IV)xOy (x > 1) clusters. Whereas tetrahedrally coordinated Ti as a constituent of the zeolite framework is unreducible, the mononuclear and clustered titanium oxide species in the pores of zeolite Y could be reversibly reduced with H2 and reoxidized with oxygen. Hereby, the reoxidation of the formed Ti(III) proceeds about 10 times faster than the most rapid reduction of the initial Ti(IV). For polynuclear species an induction period for the reduction occurs which increases with increasing particle size of the titanium oxide cluster since the number of oxygen vacancies decreases. Analysis of the reduction rate of the mononuclear Ti(IV)Ox species with a simple model assuming first-order kinetics reveals that it accelerates with increasing distortion of the octahedral coordination or with increasing multiplicity of the bonding to the zeolite framework. The kinetic data enable a quantification of the enrichment of Ti at the outer surface of the zeolite qualitatively found by XPS. The found properties of the zeolite-supported Ti(IV)Ox species in H2 and O2 atmosphere at 773 K, like the stability demonstrated by the complete reversibility of the extinctions for a large number of cycles, the short response times, and the, presumably, easier feasible miniaturization compared to ZrO2-based oxygen sensors, make it an interesting material for alternative oxygen sensing using optical detection. Acknowledgment. Financial support by the Deutsche Forschungsgemeinschaft (SCHU 426/9-1) and the European Comission (INTAS 94-1402) is gratefully acknowledged. We thank Dr. M. Froeba (University of Hamburg, Germany) for recording the Raman spectra. References and Notes (1) Shin, H.; Collins, R. J.; de Guire, M. R.; Heuer, A. H.; Sukenik, C. N. J. Mater. Res. 1995, 3, 692. (2) Yoldas, B. E.; O’Keefe, T. W. Appl. Opt. 1979, 18, 333. (3) Burns, G. P. J. Appl. Phys. 1989, 65, 2095.
Kinetics of Zeolite-Hosted Titanium Oxide Species (4) Nabavi, M.; Doeuff, S.; Sanchez, C.; Livage, J. J. Mater. Sci. Eng. 1989, B3, 280. (5) Tien, T. Y.; Stadler, G. E. F.; Zacmanides, P. J. J. Am. Ceram. Soc. 1975, 54, 280. (6) Ketron, L. Bull. Am. Ceram. Soc. 1989, 68, 860. (7) Go¨pel, W.; Hesse, J.; Zemel, J. N. Sensors; VCH: Weinheim, 1989. (8) Kirner, U. K.; Schierbaum, K. D.; Go¨pel, W.; Leibold, B.; Nicoloso, N.; Weppner, W.; Fischer, D.; Chu, W. F. Sens. Actuators B 1990, 1, 103. (9) Tang, H.; Prasad, K.; Sanjines, R.; Levy, F. Sens. Actuators B 1995, 26-27, 71. (10) Li, M.; Chen, Y. Ferroelectrics 1997, 195, 149. (11) Kormann, C.; Bahnemann, D. W.; Hoffmann, M. R. J. Phys. Chem. 1988, 92, 5196. (12) Kooyman, P. J.; van der Waal, P.; Verdaasdonk, P. A. J.; Jansen, J. C.; van Bekkum, H. Catal. Lett. 1992, 13, 229. (13) Liu, X.; Iu, K.-K.; Thomas, J. K. J. Chem. Soc., Faraday Trans. 1 1993, 89 (11), 1861. (14) Lottiaux, M.; Boulestrix, C.; Nihoul, G.; Varnier, F.; Flory, F.; Galindo, R.; Pelletier, E. Thin Solid Films 1989, 170, 107. (15) Rahmlow, T. D. Eur. Pat. Appl. EP 583, 048, 2, 1994. (16) Gra¨tzel, M. Energy Resources through Photochemistry and Catalysis; Academic Press: New York, 1983. (17) Anpo, M. Res. Chem. Intermed. 1989, 11, 67. (18) Willner, I.; Eichen, Y.; Frank, A. J. J. Am. Chem. Soc. 1989, 111, 1884. (19) Das, S.; Muneer, M.; Gopidas, K. R. J. Photochem. Photobiol. A 1994, 77, 83. (20) Liu, Z.; Davis, R. J. J. Phys. Chem. 1994, 98, 1253. (21) Yamashita, H.; Ichihashi, Y.; Anpo, M.; Hashimoto, M.; Louis, C.; Che, M. J. Phys. Chem. 1996, 100, 16041. (22) Huybrechts, D. R. C.; De Bruycker, L.; Jacobs, P. A. Nature 1990, 345, 240. (23) Clerici, M. G.; Bellussi, G.; Romano, U. J. Catal. 1991, 129, 159. (24) Thangaraj, A.; Kumar, R.; Ratnasamy, P. J. Catal. 1991, 130, 294. (25) Camblor, M. A.; Corma, A.; Perez-Pariente, J. Zeolites 1993, 13, 82. (26) Tuel, A.; Ben Taarit, Y. J. Chem. Soc., Chem. Commun. 1994, 1667. (27) Reddy, K. M.; Kaliaguine, S.; Sayari, A.; Ramaswamy, A. V.; Reddy, V. S.; Bonneviot, L. Catal. Lett. 1994, 23, 175. (28) Khouw, C. B.; Dartt, C. B.; Lablinger, J. A.; Davis, M. E. J. Catal. 1994, 149, 195. (29) Selvam, T.; Ramaswamy, A. V. Catal. Lett. 1995, 31, 103. (30) Klaas, J.; Schulz-Ekloff, G.; Jaeger, N. I. J. Phys. Chem. B 1997, 101, 1305. (31) Grubert, G.; Wark, M.; Koch, M.; Schulz-Ekloff, G. Stud. Surf. Sci. Catal. 1996, 105, 1077. (32) Anpo, M.; Aikawa, N.; Kubokawa, Y.; Che, M.; Louis, C.; Giamello, E. J. Phys. Chem. 1985, 89, 5689. (33) Kacirek, H.; Lechert, H. J. Phys. Chem. 1975, 79, 1589.
J. Phys. Chem. B, Vol. 102, No. 10, 1998 1671 (34) Kraushaar, B.; van Hooff, J. H. C. Catal. Lett. 1988, 1, 81. (35) Uguina, M. A.; Serano, D. P.; Ovejero, G.; Van Grieken, R.; Camacho, M. Appl. Catal. A 1995, 124, 391. (36) Kortu¨m, G. Reflectionsspektroskopie; Springer-Verlag: Berlin, 1969. (37) Tkachenko, O. P.; Shpiro, E. S.; Wark, M.; Schulz-Ekloff, G.; Jaeger, N. I. J. Chem. Soc., Faraday Trans. 1993, 89, 3987. (38) Scofield, J. H. J. Electron. Spectrosc. 1976, 8, 129. (39) Kerr, G. T. J. Catal. 1969, 15, 200. (40) McNicol, B. D.; Pott, G. T. J. Catal. 1972, 15, 200. (41) Zecchina, A.; Spoto, G.; Bordiga, S.; Ferrero, A.; Leofanti, G.; Petrini, G.; Padovan, M. Stud. Surf. Sci. Catal. 1991, 69, 251. (42) Thangaraj, A.; Kumar, R.; Mirajkar, S. P.; Ratnasamy, P. J. Catal. 1991, 130, 1. (43) Trong On, D.; Le Noc, L.; Bonneviot, L. J. Chem. Soc., Chem. Commun. 1996, 299. (44) Lever, A. B. P. Inorganic Electronic Spectroscopy; Elsevier: Amsterdam, 1984; Vol. 33. (45) Jørgensen, C. K. Prog. Inorg. Chem. 1970, 12, 101. (46) Ritala, M.; Leskela¨, M.; Nyka¨nen, E.; Soininen, P.; Niinisto¨, L. Thin Solid Films 1993, 225, 288. (47) Morrow, B. A.; McFarlan, A. J. J. Non-Cryst. Solids 1990, 120. (48) Schrimpf, G.; Schlenkrich, M.; Brickmann, J.; Bopp, P. J. Phys. Chem. 1992, 96, 7404. (49) Ohsaka, T.; Izumi, F.; Fujiki, Y. J. Raman Spectrosc. 1978, 7, 321. (50) Smirnov, K. S.; van de Graaf, B. Microporous Mater. 1996, 7, 133. (51) Vetter, S.; Schulz-Ekloff, G.; Kulawik, K.; Jaeger, N. I. Chem. Eng. Technol. 1994, 17, 348. (52) Grohmann, I.; Pilz, W.; Walther, G.; Kosslick, H.; Tuan, V. A. Surf. Interface Anal. 1994, 22, 403. (53) Mukhopadhyhay, S. M.; Garofalini, S. H. J. Non-Cryst. Solids 1990, 126, 202. (54) Bausa, L. E.; Gracia Sole, J.; Duran, A.; Fernandez Navarro, J. M. J. Non-Cryst. Solids 1991, 127, 267. (55) Carnahan, R. D.; Brittain, J. O. J. Am. Cer. Soc. 1965, 48, 365. (56) Cronemeyer, D. C. Phys. ReV. 1959, 113, 1222. (57) Atkins, P. W. Physical Chemistry; Oxford University Press: Oxford, 1978. (58) Hurst, N. W.; Gentry, S. J.; Jones, A.; McNicol, B. D. Catal. ReV.Sci. Eng. 1982, 24, 233. (59) Bevington, P. R. Data Reduction and Error Analysis for the Physical Science; McGraw-Hill: New York, 1989. (60) Somorjai, G. R. Chemistry in Two Dimensions, Surfaces; Cornell University Press: Ithaca, 1981. (61) Go¨pel, W.; Rocker, G.; Feierabend, R. Phys. ReV. 1983, B28, 3427. (62) Zhong, Q.; Vohs, J. M.; Bonnell, D. A. J. Am. Ceram. Soc. 1993, 76, 1137. (63) Kofstad, P. Nonstoichiometrie, Diffusion and Electrical ConductiVity in Binary Metal Oxides; John Wiley & Sons: Canada, 1972.
