Mo-Oxide Nanomaterials: Structure−Property Relationships and

Part of the “Alfons Baiker Festschrift”. , * To whom correspondence should be addressed. Phone: +41 44 63 54691. Fax: +41 44 63 56802. E-mail: ...
0 downloads 0 Views 2MB Size
Download PDF
1134

J. Phys. Chem. C 2011, 115, 1134–1142

W/Mo-Oxide Nanomaterials: Structure-Property Relationships and Ammonia-Sensing Studies† Ying Zhou,‡ Kaibo Zheng,§,⊥ Jan-Dierk Grunwaldt,| Thomas Fox,‡ Leilei Gu,§ Xiaoliang Mo,§ Guorong Chen,§ and Greta R. Patzke*,‡ Institute of Inorganic Chemistry, UniVersity of Zurich, Wintherthurerstrasse 190, CH-8057 Zurich, Switzerland, Department of Materials Science, Fudan UniVersity, 220 Handan Rd., Shanghai 200433, China, and Institute for Chemical Technology and Polymer Chemistry, Karlsruhe Institute of Technology, Kaiserstr. 12, D-76128 Karlsruhe, Germany ReceiVed: July 12, 2010; ReVised Manuscript ReceiVed: October 25, 2010

W/Mo-oxides of the hexagonal tungsten bronze (HTB) type have been investigated by X-ray absorption spectroscopy to obtain detailed insight into the substitution process of W by Mo that leads to mixed HTB frameworks. Both the morphology of the nanostructured W/Mo-HTBs as well as the oxidation state of Mo are significantly influenced through the incorporation of different alkali cations into the hexagonal channels of this open structure. A variety of complementary analytical methods, including TG, in situ and ex situ XRD, SEM, and solid-state NMR analyses, were applied to determine the thermal stability of the obtained W/Mo-HTB materials with respect to their alkali cation and NH4+ contents. A strong correlation between composition and stability was found with the Rb-W/Mo-HTBs exhibiting the highest structural and morphological resistance among the series (up to 580 °C). The NH3-sensing properties of selected W/Mooxides in test atmospheres furthermore point to promising features of the Rb-stabilized hexagonal framework materials. 1. Introduction Nanostructured transition-metal oxides (TMOs) are essential building blocks of modern materials chemistry so that they are crucial for future nanotechnological applications.1 Among the multitude of TMO types, the large family of tungsten bronzes (MxWO3 with M representing a wide range of cations, including protons and alkali or alkaline earth metals) offers a particularly high degree of structural flexibility due to the manifold options for the three-dimensional arrangement of corner-sharing and distorted WO6 octahedra.2 The MxWO3 compounds encompass cubic, tetragonal, monoclinic, hexagonal, or pyrochlore modifications, and their selective formation strongly depends on the synthetic conditions. Consequently, they cover a wide spectrum of applications, such as in electrochromic devices,3 catalysis,4 gas sensing,5 and large-scale static displays,6 and they have been widely investigated for superconductivity.7 Among the manifold MxWO3 types, hexagonal tungstates (HTBs) have attracted considerable interest due to their open structure that gives rise to a vivid intercalation and exchange chemistry within the trigonal and hexagonal channels extending along the c axis. These dynamic processes offer unique opportunities for the material tuning of HTBs in order to extend their application spectrum,8 for example, with respect to the facile removal of 137 Cs and 90Sr from radioactive waste via ion exchange.8c Likewise, the specific capacities and Li+ cycling behavior of Li-HTBs are often superior to that of other tungstate phases.9 †

Part of the “Alfons Baiker Festschrift”. * To whom correspondence should be addressed. Phone: +41 44 63 54691. Fax: +41 44 63 56802. E-mail: [email protected]. ‡ University of Zurich. § Fudan University. | Karlsruhe Institute of Technology. ⊥ Present address: Department of Chemical Physics, Lund University, Getingeva¨gen 60, Lund S-22241, Sweden.

Generally, TMO properties can be enhanced in a “synergistic” fashion through the targeted combination of two or more cations into a complex oxide material. This also applies for the substitution of W by Mo in the walls of the HTB channel framework that has been reported to enhance the cationexchange capacity,10 as well as the gas sensor applications11 and the electrochromic and lithium-ion transport properties.12 However, the transfer of such mixed TMO materials onto the nanoscale, followed by optimization of the particle shape, stability, and size, is required to fully explore their technical potential. We have established a hydrothermal morphology control approach for W/Mo-HTBs in the course of our previous studies: their nanoscale shape can be easily modified through the use of alkali cations as additives. Whereas the addition of LiCl and NaCl favors nanorod growth, their hierarchical organization into microspheres takes over in the presence of KCl and RbCl as auxiliary substances.13 With the help of in situ energy-dispersive X-ray diffraction (EDXRD) techniques, we linked these morphological phenomena to different additivedependent kinetics and mechanisms of hydrothermal HTB formation: the hierarchically structured spherical K-, Rb-, and Cs-W/Mo-HTB architectures are all formed via the same diffusion-controlled pathway, whereas the mechanistic description of Li- and Na-W/Mo-HTB nanorod formation is less straightforward.14 Despite the considerable impact of W/Mo-substitution reactions on the properties of HTB materials, very little is still known about the influence of the alkali cations situated within the channels on the local environments, optical and electronic properties, thermal stability, and oxidation states of the surrounding W and Mo atoms in the channel walls. It is well known that the open channel structure of the HTBs depends on the intercalation of stabilizing neutral or cationic species. Therefore, pure hexagonal W- or Mo-oxides in the strict sense of the word

10.1021/jp106439n  2011 American Chemical Society Published on Web 11/30/2010

Nanostructured W/Mo-Oxides have probably never been accessed because even intercalated trace amounts of stabilizing substances, that can be close to analytical detection limits, are sufficient to maintain their structural motif.8 To understand how the subtle structural chemistry of this versatile channel system affects its application potential, we applied a wide range of analytical techniques and we finally rounded off the study with sensing performance tests. First, we investigated the local structure and oxidation state of W and Mo in W/Mo-HTB frameworks using X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) methods. In the next step, the thermal stability of W/Mo-HTBs was studied with complementary methods, namely, thermal analysis in combination with in situ and ex situ powder X-ray diffraction (XRD). The mobility of Li+ in the channels of Li-W/Mo-HTBs after thermal treatment at 500 °C was monitored in terms of solid-state NMR spectroscopy. Moreover, we have studied the gas-sensing properties of W/Mo-oxides with respect to the detection of low ammonia levels to explore their environmentally relevant properties. Given that tungstates, in general, are promising sensor materials,5 their compositional and morphological tuning is an interesting approach to enhance their performance and, furthermore, to tailor their properties for specific sensing applications. 2. Experimental Section 2.1. Synthesis. Nanostructured W/Mo-oxides were prepared from a hydrothermal approach described in our previous work.13 In a typical procedure, 148 mg of (NH4)6[H2W12O40] · 3H2O (ammonium metatungstate; in the following, AMT), 58 mg of MoO3 · 2H2O, 2 mmol of MCl (M ) Li-Rb), and 2 mL of 25 vol % acetic acid were added to a Teflon-lined stainless steel autoclave with a capacity of 23 mL. The autoclave was sealed, heated for 48 h at 180 °C, and subsequently cooled to room temperature. The precipitate was collected after filtration, washed with distilled water and ethanol, and dried in air. 2.2. Analytical Methods. X-ray diffraction (XRD) analysis was conducted on a STOE STADI P diffractometer in transmission mode (flat sample holders, Ge monochromator, and Cu KR1 radiation) operated at 40 kV and 40 mA. In situ hightemperature X-ray diffraction (HT-XRD) patterns were recorded on a PANalytical X’pert diffractometer coupled with a Paar HKT-1200 high-temperature XRD chamber using Cu KR radiation. For scanning electron microscopy (SEM), performed on a LEO1530 (FEG) microscope, samples were dispersed in ethanol and subsequently deposited on a silicon wafer. The specimen was investigated without a conductive coating at rather low voltage (2 kV) to minimize charging effects. A JEOL-6060 electron microscope with a Bruker energy-dispersive X-ray spectrometer (EDXS) was employed for approximate elemental analyses. N and H analyses were conducted on a Leco CHN (S)-932 instrument. Alkali contents were determined by Mikroanalytisches Labor Pascher, Remagen, Germany. Optical absorption spectra were recorded on a Lambda 650s spectrometer. Conduction mechanisms were examined by impedance spectroscopy in the frequency range from 20 to 1 MHz at a temperature of 245 °C in air. Samples were fabricated via spincoating W/Mo-oxide suspensions in isopropanol onto interdigital Au electrodes. During the measurements, an alternating current (ac) potential amplitude of 10 mV rms was generally added to the direct current (dc) potential of the working electrode. Thermal analysis was performed on a STA 449C apparatus between 50 and 900 °C with a heating rate of 10 °C min-1 in a N2 atmosphere. Platinum crucibles were used to exclude reactions with the container material. NMR spectra were recorded at room temperature on a Bruker DRX-500 spectrom-

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1135 eter, which has been modified for solid-state measurements with MAS sample spinning up to 15 kHz (4 mm rotors) and highpower 1H decoupling. The NMR frequency was 194.41 MHz for 7Li (I ) 3/2, QM ) 0.045). Brunauer-Emmett-Teller (BET) surface area measurements were performed on a Quadrasorb SI in N2-adsorption mode. The samples were degassed at 150 °C for 5 h under vacuum prior to the measurement. 2.3. X-ray Absorption Measurements. X-ray absorption (XAS) measurements were performed at the KMC-2 beamline at BESSY (Berlin, Germany) using the double-crystal SiGe(111) monochromator, operating at 1.7 GeV. The beam intensity was stabilized by MOSTAB electronics with an accuracy of 0.3%. The detector system consists of three ionization chambers, and a reference W foil for energy calibration was used for the incident and outcoming X-ray intensities. Mo K-edge spectra were recorded at the beamline X1 at HASYLAB (DESY, Hamburg, Germany) using the double-crystal Si(311) monochromator. Mo foil was used for energy calibration (70% detuning of the crystal). To guarantee the reproducibility of the experimental data, at least two spectra were recorded for each sample. Data reduction of experimental absorption spectra and EXAFS fitting and simulation were carried out using WinXAS3.215 and following the standard procedures. To keep the systematic error during data analysis at the same level, all data were treated in an analogous manner. The energy scale of each EXAFS scan was calibrated by assigning the first inflection point of the L3 edge of the corresponding W foil and the K edge of the Mo foil. Pre-edge background subtraction and normalization was carried out by fitting a linear polynomial to the pre-edge region and a cubic polynomial to the postedge region of the absorption spectra. A smooth atomic background, µ0(k), was obtained by applying a cubic polynomial. After background removal, the data were transformed into k-space. Fourier transforms of the W L3 edge and Mo K edge χ(k), which was multiplied by k3, were taken over similar photoelectron wavenumber ranges, typically from 2 to 12 Å-1, and a Bessel window was used in the Fourier transform. The FEFF716 program package was used to calculate phase and amplitude factors, followed by least-squares curve fitting of the data. 2.4. Gas-Sensing Measurements. To fabricate thick film gas sensors with a thickness around 10 µm, 0.2 mL of terpineol mixed with 5 wt % ethylcellulose and 0.04 mL acetylacetone was added to 200 mg of the respective W/Mo-HTBs to obtain a paste. The paste was coated onto a quartz slide covered by Ni interdigital electrodes. Each bar of the interdigital electrodes was 200 µm in width, and the spacing between the neighboring bars was 200 µm as well. The film was then sintered in air to remove the organic additives at 150 °C for 10 min, 350 °C for 10 min, and 500 °C for 30 min, respectively, and then cooled to the room temperature naturally.17 Gas-sensing properties were measured using a home-built testing system, which consists of a test chamber, sensor holder with embedded heater, Agilent oscilloscope-54810A, mass flow controllers, and a data acquisition system. Prior to the measurement, the gas sensor was aged for 1 h at the required temperature in an Ar atmosphere. During the sensor measurements, the target gas was mixed with pure Ar using mass flow controllers to adjust the desired concentration with the total flow rate maintained at 200 sccm. The resistance response was defined as RAr/Rammonia, where RAr is the resistance of the sensor in a pure Ar gas atmosphere and Rammonia is the resistance in the ammonia-containing test gas atmosphere.5

1136

J. Phys. Chem. C, Vol. 115, No. 4, 2011

Zhou et al.

TABLE 1: Analytical Characterization of Hydrothermally Synthesized W/Mo-HTB-Oxides sample

composition

typical morphology

BET (m2/g)

particle dimensions 10-40 nm diam, 200 nm-1 µm length 20-100 nm diam, 300 nm-2 µm length 40-100 nm diam, 500 nm-3 µm length spheres (1-2 µm diam), rods (20-70 nm diam) spheres (1.5-3 µm diam), rods (20-100 nm diam)

WMo

(NH4)0.24W0.77Mo0.23O3

nanorods

78.8

WMo_Li

(NH4)0.16Li0.07W0.75Mo0.25O3

nanorods

64.6

WMo_Na

(NH4)0.09Na0.19W0.79Mo0.21O3

nanorods

58.4

WMo_K

(NH4)0.04K0.23W075Mo0.25O3

hierarchical nanostructures

82.4

WMo_Rb

(NH4)0.01Rb0.24W0.79Mo0.21O3

hierarchical nanostructures

56.3

3. Results and Discussion 3.1. Local Structure and Composition of Nanostructured W/Mo-HTBs. All obtained W/Mo-oxides displayed the hexagonal tungstate structure type, in line with our preceding results.13 Table 1 provides a representative summary of the phase, composition, morphology, and BET surface area of the products emerging from the different hydrothermal AMT/ MoO3 · 2H2O/MCl systems. In close analogy to the W-HTB systems,18 the fraction of ammonium cations incorporated into the HTB framework strongly depends on the size of the alkali cation. As the larger alkali cations are less prone to exchange reactions, the ammonia content decreases from the W/Mo-HTBs to the Rb-W/Mo-HTBs. The W/Mo-ratios were determined with EDXS measurements, and the entire series of compounds contained an approximately 3:1 distribution of both elements within the experimental error of the method, thereby exceeding the 2:1 ratio of W/Mo in the starting materials. Furthermore, all products exhibited high BET surface areas (between 55 and 82 m2/g), which are higher than the typical values for alkalifree (NH4)0.26WO3 nanorod samples (37 m2/g).18 The Cs-W/ Mo-HTBs of the series were not included in the present discussion for two reasons: first, the formation of pyrochlore tungstate side products has to be circumvented through special optimization work, which renders them less straightforward for technical applications,13b and their gas-sensing behavior under the given conditions deviated considerably from the trends observed for the other compounds (cf. below). Although the phase of the products could be assigned with laboratory X-ray powder diffraction methods, their accuracy is not sufficient to elucidate the detailed local environments of the different cations. The small particle size together with the low degree of crystallinity of the M-W/Mo-HTBs (M ) Li to Rb) led to very broad fwhm values (full width at half-maximum) that rendered Rietveld refinements impossible. XAS methods were the appropriate choice for the present study because the determination of the coordination and oxidation state of a target element with XANES/EXAFS does not require the presence of a crystalline material.19 This wide applicability of XAS furthermore facilitates the assignment of local environments to dopant elements in solid solutions, such as the Mo atoms incorporated into the HTB framework. Figure 1a,b shows the W L3- and Mo K-edge XANES/ EXAFS spectra recorded on the entire W/Mo-oxide series (cf. Table 1). The XAS spectra at the W L3 edge show a single large peak (white line) above the edge that corresponds to a transition of 2p3/2 electrons toward the 5d state.20 The edge position is usually strongly related to the oxidation state of the absorber ion with lower oxidation states leading to lower energies.19b As the W L3 absorption edge energies for all W/Mooxide samples remained practically constant at 10 204 eV, the presence of alkali cations within the channels does not seem to affect the oxidation state of the tungsten atoms in an obvious

manner. However, the intensity of the white line increased with the size of the alkali cation from 1.76 for the alkali-free W/Mooxide reference sample to 1.96 for Rb-W/Mo-HTB. The increase of the white line intensity was recently linked to a decreasing extent of condensation among the tungsten species, that is, their structural arrangement at medium distances.21 Interestingly, the incorporation of alkali cations with increasing ionic radii reduced the condensation level of the W species within the W/Mo-HTB oxides. On the other hand, all XAS investigations at the Mo K edge (cf. Figure 1b) had a characteristic weak pre-edge feature around 19 995 eV prior to the sharp absorption rise at around 20 000 eV in common that is due to the dipole-forbidden transition from 1s to the nd (n ) 3 or 4) level (by dipole selection rules).22 The intensity of the pre-edge peak is generally

Figure 1. (a) Normalized W L3-edge XAFS spectra of W/Mo-oxides prepared in the presence of different alkali chlorides (inset: W L3-edge XANES spectra) and (b) normalized Mo K-edge XAFS spectra (inset: Mo K-edge XANES spectra).

Nanostructured W/Mo-Oxides

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1137

higher for tetrahedrally coordinated Mo(VI) than the values observed for an octahedral environment. A comparison of a representative W/Mo-HTB oxide sample with NaMoO4 as a reference for tetrahedrally coordinated Mo(VI) clearly showed that the HTB compound has a less intense pre-edge peak and a more pronounced feature on the ascending part of the curve (denoted with A and B in Figure S1, Supporting Information): this indicates that all the Mo atoms in the W/Mo-HTB samples are octahedrally coordinated. In contrast to the constant +VI oxidation state of W observed for all the samples, the position of the Mo K edge shows a clear dependence on the incorporated alkali cation that points to a higher redox flexibility of the Mo centers. Whereas the Mo K-edge positions for the pristine W/Mo-HTB oxide, Li-W/Mo-HTB, and Na-W/Mo-HTB remain constant at 20 010 eV, the values for K-W/Mo-HTB and RbW/Mo-HTB are shifted to lower energies at 20 009 and 20 007 eV, respectively, thereby pointing to a lower oxidation state of Mo in the corresponding samples. These valence changes can be quantified by evaluating the XAS-edge positions using references with different valences V (Mo metal (z(V) ) 0), MoO2 (z(V) ) +4), and MoO3 (z(V) ) +6)), and the linear relationship between the Mo valence and the Mo K-edge energy shift f(E) that is depicted in Figure S2 (Supporting Information) can be expressed as follows:23

z(V) ) -0.000724 + 0.51658 f(E)

(1)

The average valence of Mo in W/Mo-HTB, Li-W/Mo-HTB, and Na-W/Mo-HTB was determined to +5.7 on this basis, and the values for K-W/Mo-HTB and Rb-W/Mo-HTB were calculated to be +5.2 and +4.1, respectively, on the basis of the above linear function. As the hexagonal channels may contain both NH3 and NH4+,8a a more accurate determination of the sample compositions beyond these trends is a demanding task because it would also include a full discussion of possible oxygen vacancies. We have tried to determine the NH3/NH4+ ratio by means of solid-state NMR, but given that these attempts were not successful, we prefer to limit the stoichiometric accuracy to the tentative values given in Table 1. The above valence trends shed an interesting light on the different responses of W and Mo toward the incorporation of alkali cations into the channel structure: whereas the Mo centers display versatile redox properties, the oxidation state of W is not affected by the ion-exchange process. Previous studies have also shown that Mo(VI) is more easily reduced in the HTB framework than W(VI).24 More detailed information about the coordination numbers and bond distances of the W and Mo atoms can, furthermore, be obtained from curve-fitting analyses of the W L3- and Mo K-edge EXAFS spectra, respectively (for typical Fourier transformed spectra (FT, without phase correction), cf. Figure S3, Supporting Information). The close resemblance of the FT spectra recorded at the W L3 and Mo K edges is a good prerequisite for the facile replacement of W with Mo in the channel walls of the HTB structure. The first peak appears at 0.8-2.2 Å and can be assigned to the first coordination shell consisting of six oxygen atoms, whereas the following peak in the region from 2.2 to 3 Å arises from multiple scattering (MS). The W/Mo-W/Mo coordination shell is finally reflected in the third backscattering contribution (from 3 to 4.3 Å).21 Although the key structural motif of hexagonal WO3 has already been determined from powder X-ray diffraction data in 1979,25 the lack of suitable single crystals rendered the complete structural elucidation quite difficult so that the occurrence of two subsets of W-O distances was suggested. This hypothesis

Figure 2. UV/vis spectra of the alkali-W/Mo-HTB oxides.

was revised in a recent attempt to fit the EXAFS spectra recorded at the W L3 edge of hexagonal WO3: as two W-O distances turned out to be insufficient to obtain a good fit, the authors proposed to use three different W-O distances instead.26 These findings also apply for our current measurements that can best be fitted with a set of three W/Mo-O distances. The fit was performed in r-space, and the prefactor S02 was fixed at 0.8 (for typical fit results, cf. Figure S3 as well as Tables S1 and S2 (Supporting Information)). First, the coordination sphere of the Mo atoms bears a close resemblance to that of tungsten as the majority species, and the mean Mo-O distances appear to be slightly shorter than the corresponding W-O values (for details, cf. the Supporting Information). A clear-cut relationship between the individual bond lengths and the type of the incorporated alkali cation could not be established. 3.2. Optical and Electronic Properties. All W/Mo-HTB oxide samples have a blue color in common, and their UV/vis spectra are displayed in Figure 2. The band gaps (Eg) of W/MoHTB, Li-W/Mo-HTB, Na-W/Mo-HTB, K-W/Mo-HTB, and RbW/Mo-HTB were assigned as 2.98, 2.90, 2.97, 3.01, and 2.97 eV, respectively, according to the onset of the absorption edge. The obtained values are higher than the band gap reported for K-W-HTB oxide (2.80 eV).2c Nevertheless, the band gaps are not clearly related to the different alkali cations situated within the channels so that the slight differences between the individual values are due to the different particle sizes and morphologies of the alkali-W/Mo-HTB oxide samples (cf. Table 1 and Figure S4 (Supporting Information)). In addition, the materials were studied with impedance spectroscopy to examine the type of conductivity mechanism. As shown in Figure 3a, all W/Mo-HTB oxides exhibit a single semicircled Nyquist diagram at 245 °C in air, which indicates that electronic conductivity is the predominant mechanism. Note that the M-W/Mo-HTB oxides maintain the hexagonal structure motif after heat treatment at 245 °C (for details, cf. section 3.3). The Nyquist response can be differentiated in terms of R-C parallel circuits, as schematically depicted in Figure 3b, representing the combination of electron transport processes for both bulk and grains.,27a and modeled R-C values are listed in Table S3 (Supporting Information). In summary, the grain resistance (Rgb) values of the M-W/Mo-HTB oxides exceed the bulk resistances by 2-3 orders of magnitude (Table S3, Supporting Information), thereby indicating that the overall resistance can be mainly attributed to the influence of the grain boundaries on the material properties. This is in line with the general understanding that intergrain effects considerably influ-

1138

J. Phys. Chem. C, Vol. 115, No. 4, 2011

Zhou et al.

Figure 3. (a) Nyquist diagram of the M-W/Mo-HTB oxide series (M ) Li-Rb; 245 °C in air) and (b) equivalent circuit model.

Figure 4. Representative TG analyses of the pristine W/Mo-HTB oxide and Rb-W/Mo-HTB oxide.

ence the electronic transport mechanisms in nanostructured metal oxides with small particle sizes and high surface areas.27b 3.3. Thermal Stability of M-WMo-HTBs (M ) Li-Rb). As sensor fabrication often involves continuous heating processes, the sensing material under consideration has to be stable with respect to structure, composition, and, if possible, morphology. This renders thermal stability investigations of the W/Mooxides an indispensable prerequisite for their potential use in gas sensors. The M-W/Mo-HTB series (M ) Li-Rb) was thus investigated with TG methods, and representative results are shown in Figure 4. The alkali-free W/Mo-HTB underwent a mass loss of 3.0% below 300 °C that is probably due to residual amounts of H2O or acetic acid. In the next stage (300-520 °C), further release of NH3 and H2O led to a mass loss of 2.5% (note that the nitrogen content obtained from elemental analyses is 1.5%). No further mass loss upon raising the temperature to 750 °C was observed, and at higher temperatures, the evaporation of MoO3 sets in, as had been previously reported.13a The TG curves of Li-, Na-, and K-W/Mo-HTB basically followed the same trend with different mass losses for H2O and NH3, respectively, that agreed well with the elemental analyses (Table 1). Rb-W/Mo-HTB, however, exhibits a different thermal behavior because an additional mass loss peak (0.44%) was observed at 550 °C with a corresponding thermal effect that remains to be investigated in detail: for the evaporation of Rbcontaining species, the temperature appears to be too low so

that this phenomenon may arise from more firmly bound water molecules in the channels that are only removed upon stronger heat treatment or could be traced back to loss of other species involving oxygen. Moreover, the structural changes accompanying the thermal treatment were monitored with the help of in situ powder XRD. Figure 5 shows the temperature-dependent in situ XRD patterns of W/Mo-HTB and Rb-W/Mo-HTB. At ca. 435 °C, W/MoHTB displayed the onset of a phase transition into a modification of lower symmetry, as can be seen from the splitting of the (111) and (112) reflections together with the appearance of new reflections. The XRD patterns recorded at 480 °C permitted its identification as monoclinic W0.71Mo0.29O3 (JCPDS No. 76-1279, S.G. P21/n, a ) 7.446 Å, b ) 7.424 Å, c ) 7.613 Å, β ) 91.24°). Interestingly, the Rb-containing HTB framework displayed a much higher thermal stability up to around 580 °C, and a further temperature rise led to the formation of a mixed oxide phase with a hitherto unassigned structural motif. Generally, the thermal resistance of the HTB framework is a function of the incorporated cations that are indispensable to maintaining the open structural framework: the hexagonal form of WO3 not only is metastable at room temperature with respect to the monoclinic and orthorhombic modifications but also has not been accessed yet in pure form without any stabilizing ions or molecules.8 Therefore, the collapse of the alkali-free, Li-, Na-, and K-W/Mo-HTB frameworks goes hand in hand with the observed mass loss of the NH4+ ions that are obviously essential to keeping up the structural motif (Figures 4 and 5). However, their role can be taken over by the larger Rb+ cation that exerts a stronger stabilizing effect than its smaller analogues Li+, Na+, and K+ so that the Rb-W/Mo-HTB oxides contain the lowest amounts of NH4+ among the series (Table 1). Concerning the formation of Na-HTBs, a minimum value of 160 ppm Na+ in the precursor has been reported to be essential for the abovementioned structural stabilization effect.28 The remaining amounts of the smaller cations (Li+, Na+, K+) after heat treatment are obviously insufficient to sustain the hexagonal framework, whereas the Rb-W/Mo-HTBs contain enough Rb+ to withstand treatment in air up to 580 °C (cf. Figures 4 and 5). EDXS determinations of the residual alkali amounts after heat treatment at 500 °C for 1 h demonstrated that these values were in the same range of the initial contents prior to heating. In the case of thermally treated Li-W/Mo-HTBs, the effect of the thermally induced phase changes was monitored with solidstate NMR investigations (for details, cf. below and Figure 8). All samples heated to 500 °C were structurally characterized

Nanostructured W/Mo-Oxides

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1139

Figure 5. In situ XRD patterns monitoring the thermal behavior of the (a) pristine W/Mo-HTB oxide and (b) Rb-W/Mo-HTB oxide.

Figure 6. Ex situ XRD patterns of the W/Mo-HTB oxide series after heat treatment at 500 °C for 1 h: (a) pristine W/Mo-oxide, (b) Li-W/ Mo-oxide, (c) Na-W/Mo-oxide, (d) K-W/Mo-oxide, and (e) Rb-W/Mooxide.

afterward in more detail with ex situ powder XRD (cf. Figure 6). The alkali-free W/Mo-HTB as well as the Li- and Nacontaining compounds were almost quantitatively converted into a monoclinic phase with a structural motif corresponding to that of W0.71Mo0.29O3 (JCPDS No. 76-1279; cf. above). The intermediate ionic radius of K+ prevents the complete collapse of the HTB structure so that it is partially retained besides the monoclinic phase, as can be seen from Figure 6. Rb-W/MoHTB, however, remains structurally unaltered at 500 °C, as to be expected from the in situ XRD screening (Figure 5). In addition to the consequences of the thermal treatment for the structure and composition of the HTBs, the morphological changes were also investigated: SEM images of the pristine W/Mo-HTBs as well as of the Li- and Na-containing analogues demonstrated that the nanorods emerging from hydrothermal synthesis13b (cf. Figure S4, Supporting Information) aggregated into larger and less anisotropic particles at 500 °C with individual sizes ranging from several tens of nanometers to several hundred nanometers (Figure 7). As the K-W/Mo oxide product formed after heating is biphasic, its morphology is a mixture of rods and irregularly shaped particles (Figure 7d). In

Figure 7. Representative SEM images of the W/Mo-oxide series after heat treatment at 500 °C for 1 h: (a) pristine W/Mo-oxide, (b) Li-W/ Mo-oxide, (c) Na-W/Mo-oxide, (d) K-W/Mo-oxide, (e) Rb-W/Mooxide, and (f) Rb-W/Mo-oxide with higher magnification.

line with the morphological changes, the BET surface area of W/Mo, Li-W/Mo, Na-W/Mo, and K-W/Mo oxides decreased obviously to 4.2, 4.7, 5.9, and 3.5 m2/g, respectively, after heat treatment at 500 °C. Most importantly, the structural stability of Rb-W/Mo-HTBs goes hand in hand with morphological resistance so that the hierarchical arrangement of nanorods withstands the thermal treatment at 500 °C, as can be seen from Figure 7e,f, with a BET surface area of 10.8 m2/g. In summary, the choice of the stabilizing cation is decisive for the thermal properties of M-W/Mo-HTBs and Rb-W/Mo-HTBs turned out to be the most promising candidates for the application in sensors at high operational temperatures.

1140

J. Phys. Chem. C, Vol. 115, No. 4, 2011

Our preceding solid-state 7Li NMR spectroscopy investigations on the mobility of Li+ within the W/Mo-HTB host framework showed that it depends on the channel typesstrigonal or hexagonalsthat can, in principle, both be occupied depending on the size of the incorporated cations.13b After the heat treatment at 500 °C, the Li+ cations are situated in the structurally different environment of the above-mentioned monoclinic W/Mo-phase (cf. Figure 6). We thus applied 7Li NMR spectroscopy as another complementary technique to differentiate the present W/Mooxide types with respect to their application-oriented properties by monitoring the effect of thermal treatment on the Li+ mobility and environment in the Li-W/Mo-oxide samples. A comparison of the HTB spectra prior to heating and those of the monoclinic Li-containing phase is shown in Figure 8 together with a schematic representation of the different structural features of the host lattice. Whereas the major structural changes only exerted a slight influence on the 7Li solid-state NMR, the 7Li MAS spectrum of the monoclinic mixed oxide still showed an extensive sideband pattern (Figure 8e) that is comparable with the results of the pristine Li-W/Mo-HTB sample (Figure 8b). This implies the presence of quasi spherical symmetry with dominating chemical shift anisotropy (CSA) effects and just small quadrupolar interactions, as had been already observed on the untreated sample.13b However, the sharpening of the static 7 Li powder pattern from 5.8 to 2.0 kHz after the heating process can be interpreted as a first indication of an accompanying increase of the Li+ mobility that was further confirmed through a solid-state 7Li-T1 time measurement. Generally, the obtained T1 data are an additional, sensitive probe for the dynamic behavior of the respective lithium sites. After treatment of the Li-W/Mo-oxide at 500 °C, the T1 time displayed a more than 3-fold decrease from 481 to 144 ms. Furthermore, the presence of only one isotropic shift in the MAS spectra indicated the presence of a single site for Li+ remaining after the thermally induced phase transition: other than the dual channel system of the HTBs (Figure 8c), the monoclinic W/Mo-oxide host framework offers only one tunnel type to be occupied by the Li+ cations (Figure 8f). 3.4. Gas-Sensing Properties. The temperature-dependent resistance response of the W/Mo-oxide series toward NH3 as a test gas was monitored, and Figure 9 displays the results for the different sensor types obtained with 100 ppm NH3 diluted in pure Ar. Here, the resistance response was defined as the relative resistance: S ) RAr/Rammonia. The resistance of all W/Mooxide-based sensors generally decreased in the presence of NH3, which is a characteristic feature of n-type semiconductors.29 As mentioned above, this trend does not apply to the Cs-W/Mooxide member of the series: its different and more complex sensing behavior is still under investigation and will be reported elsewhere. The discontinuous and nonlinear resistance change with temperature among the series is also typical of oxide sensors,30 and it arises from the type of chemisorption on the surface: although lower temperatures do not provide sufficient energy for chemisorption, the desorption rate outweighs the adsorption rate again at considerably higher temperatures. This results in maximum sensitivities at medium temperatures (275 °C) for all oxides under investigation so that they display a lower operating temperature than pure WO3 nanofibers at 350 °C.5e When operated at the optimized temperature of 275 °C, all alkali-containing oxide samples exhibit higher sensitivities than the alkali-free W/Mo-oxide, and irrespective of the actual structure motif, the overall resistance response of the oxide materials increases with the size of the alkali cations, that is, SRb_WMo > SK_WMo > SNa_WMo > SLi_WMo (cf. Table 2).

Zhou et al.

Figure 8. Solid-state 7Li NMR spectra of Li-W/Mo-HTB (a) without MAS and (b) with MAS and (c) schematic structure representation of the HTB framework, compared to the corresponding spectra of monoclinic Li-W/Mo-oxide after heat treatment (d) without MAS and (e) with MAS and (f) schematic representation of the monoclinic W/Mooxide host structure.

Figure 9. Response resistance S vs operating temperature for NH3 detection (100 ppm) with sensors based on the investigated W/Mooxide series.

As structural stability and a high surface area are key requirements for sensor materials, the Rb-W/Mo-HTB is of the highest interest among the investigated compounds, and this corresponds with its maximum sensitivity of 11.1 among the series that exceeds the lowest value observed for the alkalifree W/Mo-oxide (1.2) almost by a factor of 10. The same trends apply for the resistance response as a function of the gas concentration at 275 °C (Figure 10).

Nanostructured W/Mo-Oxides

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1141

TABLE 2: Resistance Response at 275 °C towards 100 ppm NH3 of the Sensors Based on the W/Mo-Oxide Series sample

WMo

WMo_Li

WMo_Na

WMo_K

WMo_Rb

S

1.2

2.1

7.7

8.6

11.1

Although the sensitivity of WO3, in general, has been proven using a variety of gases, such as NO2, O3, H2S, H2, and NH3,5 its ammonia-sensing properties have always been described as rather weak in comparison with superior properties regarding the detection of NO2.5,31 The present results clearly show that the resistance response of W/Mo-oxides toward NH3 can be remarkably enhanced through the structural incorporation of additional cations, thus demonstrating the importance of structure-property relationships among the large family of tungsten oxides (cf. Table 2). However, several parameters, especially structure, composition, and morphological and thermal stability of the product, have to be optimized, and the phase change of the M-W/Mo-HTB oxides (M ) Li-K) during sensor preparation has to be taken into consideration. With respect to these stability criteria, Rb-W/Mo-HTB oxides are the most promising sensor candidate emerging from the present study because all other oxides undergo a complete or partial phase transition into a monoclinic W/Mo-oxide modification during the annealing process at 500 °C that is required to remove the organic additives for the completion of the sensor fabrication. As the gas sensitivity increases with the amount of HTB phases remaining in the oxide components after thermal treatment, it is reasonable to conclude that the W/Mo-HTB oxide type displays a higher resistance response toward NH3 than the monoclinic modification. This hypothesis agrees well with recent reports on WO3 that confirm a better NH3-sensing performance of the binary nanoscale HTBs in comparison with monoclinic WO3 particles.32 Furthermore, Rb-W/Mo-HTB exhibited the highest resistance response, which can be attributed to its unique morphology compared with the remaining members of the W/Mo-oxide series: as shown in Figure 7, Rb-W/Mo-HTB retained its hierarchical arrangement of nanorods with a higher surface area of 10.8 m2/g after heat treatment at 500 °C, whereas the other W/Mo-oxides showed intense agglomeration effects. The high surface area of Rb-W/Mo-HTB enhances the number of absorption sites for NH3 sensing, which, in turn, improves sensing properties. This is in line with a recent study that identified the surface area as the most influential parameter in the comparison of the NO2-sensing properties of different tungsten oxide

Figure 10. Resistance response vs NH3 concentration for the investigated W/Mo-oxide sensor series at the optimized operating temperature of 275 °C.

modifications.33 The present results furthermore illustrate the importance of selecting the appropriate cationic stabilizer for the hexagonal HTB channels to render them sufficiently stable for technical applications so that their further performance optimization according to standard ammonia-sensing oxides34 will be the next step. 4. Conclusions A series of nanostructured M-W/Mo-HTB oxides (M ) Li-Rb) with band gaps in the range between 2.90 and 3.01 eV were hydrothermally synthesized, and the materials were found to exhibit mainly electronic conductivity at 245 °C. As the substitution of tungsten by molybdenum in the framework structure of HTBs (hexagonal tungsten bronzes) can substantially alter their properties and nanoscale morphologies, the local environments and oxidation states of W and Mo centers in nanostructured M-W/Mo-HTBs (M ) Li-Rb; approximately, W/Mo ratio ) 3:1) were investigated with XANES/EXAFS spectroscopy. The molybdenum atoms display a more flexible redox behavior upon the introduction of ammonium and alkali cations into the hexagonal HTB channels than the tungsten host framework so that the Mo oxidation state decreases from +5.7 to +4.1 with increasing size of the alkali guest cations. The choice of the incorporated species influences not only the nanoscale morphology but also especially the thermal stability of the M-W/Mo-HTB phases. Whereas their Li- and Nacontaining representatives undergo a structural transformation into a monoclinic W/Mo-oxide upon heating to 500 °C, the presence of the larger K+ cation leads to a partial persistence of the HTB phase under analogous conditions. This stabilizing influence of the cation is most pronounced in the Rb-W/MoHTBs that display both structural and morphological stability up to 500 °C. Ammonia-sensing tests on the entire mixed oxide series revealed an enhanced resistance response of the thermally treated compounds with respect to binary hexagonal WO3, and nanostructured Rb-W/Mo-HTB exhibited the best performance. The superior properties of nanostructured Rb-W/Mo-HTB over other tungsten oxide sensor materials are the result of strategic cation intercalation that stabilizes the structural framework and the nanoscale morphology upon thermal treatment during sensor fabrication. This renders Rb-W/Mo-HTB a promising candidate for further optimization toward stable and sensitive NH3 sensors. Follow-up investigations on the morphology-sensing relationships of doped nanoscale W/Mo-HTBs are now in progress in order to tap the technological potential of these complex oxides for the fabrication of miniaturized nanosensors, and detailed work on the Rb-W/Mo-HTB performance optimization (e.g., selectivity, stability, influence of the composition) is under way. Acknowledgment. We thank BESSY (Berlin, Germany) and HASYLAB (DESY, Hamburg) for providing beam time at beamline KMC-2 for the W L3-edge and at beamline X1 for Mo K-edge XAS experiments, respectively. We are grateful to the Electron Microscopy ETH Zurich, EMEZ, as well as to the Center for Microscopy and Image Analysis, University of Zurich, for experimental support. The authors thank Dr. D. Koziej and Prof. M. Niederberger (Department of Materials, ETH Zurich) for help with in situ XRD measurements. Financial support by the EU (Contract RII-13-CT-2004-506008), by the Swiss National Science Foundation (SNF Professorship PP002114711/1), by the University of Zurich, and by the Sino Swiss Science and Technology Cooperation (SSSTC, EG05-092008) is gratefully acknowledged.

1142

J. Phys. Chem. C, Vol. 115, No. 4, 2011

Supporting Information Available: Mo K-edge XANES of W/Mo-oxides and Na2MoO4, linear relationship between the average Mo valence and the corresponding Mo K-edge positions, representative first-shell fit of a W/Mo-HTB oxide sample, and structural parameters obtained from the fitting of EXAFS spectra at the W L3 edge and at the Mo K edge of M-W/Mo-HTBs (M ) Li-Rb). This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) (a) Rao, C. N. R.; Deepak, F. L.; Gundiah, G.; Govindaraj, A. Prog. Solid State Chem. 2003, 31, 5–147. (b) Patzke, G. R.; Krumeich, F.; Nesper, R. Angew. Chem., Int. Ed. 2002, 41, 2446–2461. (c) Goesmann, H.; Feldmann, C. Angew. Chem., Int. Ed. 2010, 49, 1362–1395. (d) Pinna, N.; Niederberger, M. Angew. Chem., Int. Ed. 2008, 47, 5292–5304. (2) (a) Zivkovic, O.; Yan, C.; Wagner, M. J. J. Mater. Chem. 2009, 19, 6029–6033. (b) Guo, J. D.; Reis, K. P.; Whittingham, M. S. Solid State Ionics 1992, 305, 53–56. (c) Gu, Z. J.; Ma, Y.; Zhai, T. Y.; Gao, B. F.; Yang, W. S.; Yao, J. N. Chem.sEur. J. 2006, 12, 7717–7723. (3) (a) Shibuya, M.; Miyauchi, M. Chem. Phys. Lett. 2009, 473, 126– 130. (b) Balaji, S.; Djaoued, Y.; Albert, A. S.; Ferguson, R. Z.; Bruning, R. Chem. Mater. 2009, 21, 1381–1389. (c) Han, W.; Hibino, M.; Kudo, T. Solid State Ionics 2000, 128, 25–32. (4) (a) Francke, L.; Durand, E.; Demourgues, A.; Vimont, A.; Daturi, M.; Tressaud, A. J. Mater. Chem. 2003, 13, 2330–2340. (b) Huo, L.; Zhao, H.; Mauvy, F.; Fourcade, S.; Labrugere, C.; Pouchard, M.; Grenier, J.-C. Solid State Sci. 2004, 6, 679–688. (5) (a) Ponzoni, A.; Comini, E.; Sberveglieri, G.; Zhou, J.; Deng, S. Z.; Xu, N. S.; Ding, Y.; Wang, Z. L. Appl. Phys. Lett. 2006, 88, 20310–20312. (b) Polleux, J.; Gurlo, A.; Barsan, N.; Weimar, U.; Antonietti, M.; Niederberger, M. Angew. Chem., Int. Ed. 2006, 45, 261–265. (c) Szilagyi, I. M.; Wang, L. S.; Gouma, P. I.; Balaszsi, C.; Madarasz, J.; Pokol, G. Mater. Res. Bull. 2009, 44, 505–508. (d) Pokhrel, S.; Simion, C. E.; Teodorescu, V. S.; Barsan, N.; Weimar, U. AdV. Funct. Mater. 2009, 19, 1767–1774. (e) Wang, G.; Ji, Y.; Huang, X. R.; Yang, X. Q.; Gouma, P. I.; Dudley, M. J. Phys. Chem. B 2006, 110, 23777–23782. (f) Li, X. L.; Lou, T. J.; Sun, X. M.; Li, Y. D. Inorg. Chem. 2004, 43, 5442–5449. (6) Whittingham, M. S. In Solid State DeVices; Chowdari, B. V. R., Radhakrishnan, S., Eds.; World Scientific: Singapore, 1988. (7) Shengelaya, A.; Reich, S.; Tsabba, Y.; Mu¨ller, K. A. Eur. Phys. J. B 1998, 12, 13–15. (8) (a) Szilagyi, I. M.; Madarasz, J.; Pokol, G.; Kiraly, P.; Tarkanyi, G.; Saukko, S.; Mizsei, J.; Toth, A. L.; Szabo, A.; Varga-Josepovitso, K. Chem. Mater. 2008, 20, 4116–4125. (b) Szilagyi, I. M.; Madarasz, J.; Pokol, G.; Hange, F.; Szalontai, G.; Varga-Josepovitso, K.; Toth, A. L. J. Therm. Anal. Calorim. 2009, 97, 11–18. (c) Griffith, C. S.; Sebesta, F.; Hanna, J. V.; Yee, P.; Drabarek, E.; Smith, M. E.; Luca, V. J. Nucl. Mater. 2006, 358, 151–163. (d) Han, W.; Hibino, M.; Kudo, T. Bull. Chem. Soc. Jpn. 1998, 71, 933–937. (9) Reis, K. P.; Ramanan, A.; Whittingham, M. S. J. Solid State Chem. 1992, 96, 31–47. (10) Griffith, C. S.; Luca, V. Chem. Mater. 2004, 16, 4992–4999.

Zhou et al. (11) Merdrignac-Conanec, O.; Moseley, P. T. J. Mater. Chem. 2002, 12, 1779–1781. (12) Xue, B.; Peng, J.; Xin, Z. F.; Kong, Y. M.; Li, L.; Li, B. J. Mater. Chem. 2005, 15, 4793–4798. (13) (a) Michailovski, A.; Krumeich, F.; Patzke, G. R. Chem. Mater. 2004, 16, 1433–1440. (b) Kiebach, R.; Pienack, N.; Bensch, W.; Grunwaldt, J.-D.; Michailovski, A.; Baiker, A.; Fox, T.; Zhou, Y.; Patzke, G. R. Chem. Mater. 2008, 20, 3022–3033. (14) Zhou, Y.; Pienack, N.; Bensch, W.; Patzke, G. R. Small 2009, 5, 1978–1983. (15) Ressler, T. J. Synchroton Radiat. 1998, 5, 118–122. (16) Newville, M. J. Synchroton Radiat. 2000, 8, 322–324. (17) Zheng, K.; Zhou, Y.; Gu, L. L.; Mo, X.; Patzke, G. R.; Chen, G. R. Sens. Actuators, B 2010, 148, 240–246. (18) Michailovski, A.; Kiebach, R.; Bensch, W.; Grunwaldt, J.-D.; Baiker, A.; Komarneni, S.; Patzke, G. R. Chem. Mater. 2007, 19, 185– 197. (19) (a) Rehr, J. J.; Albers, R. C. ReV. Mod. Phys. 2000, 72, 621–654. (b) Haaβ, F.; Adams, A. H.; Buhrmester, T.; Schimanke, G.; Martin, M.; Fuess, H. Phys. Chem. Chem. Phys. 2003, 5, 4317–4324. (20) Yamazoe, S.; Hitomi, Y.; Shishido, T.; Tanaka, T. J. Phys. Chem. C 2008, 112, 6869–6879. (21) Pauporte, T.; Soldo-Olivier, Y.; Faure, R. J. Phys. Chem. B 2003, 107, 8861–8867. (22) Beale, A. M.; Sankar, G. Chem. Mater. 2003, 15, 146–153. (23) Ressler, T.; Wienold, J.; Jentoft, R. E.; Neisius, T. J. Catal. 2002, 210, 67–83. (24) Genin, C.; Driouiche, A.; Gerand, B.; Figlarz, M. Solid State Ionics 1992, 53-56, 315–323. (25) Gerand, B.; Nowogrocki, G.; Guenot, J.; Figlarz, M. J. Solid State Chem. 1979, 29, 429–434. (26) Moscovici, J.; Rougier, A.; Laruelle, S.; Michalowicz, A. J. Chem. Phys. 2006, 125, 124505. (27) (a) Labidi, A.; Jacolin, C.; Bendahan, M.; Abdelghani, A.; Gue´rin, J.; Aguir, K.; Maaref, M. Sens. Actuators, B 2005, 106, 713–718. (b) Xu, J. Q.; Pa, Q. Y.; Shun, Y. A.; Tian, Z. Z. Sens. Actuators, B 2000, 66, 277–279. (28) Pfeifer, J.; Balazsi, C.; Kiss, B. A.; Pecz, B.; Toth, A. L. J. Mater. Sci. Lett. 1999, 18, 1103–1105. (29) Balazsi, C.; Wang, L. S.; Zayim, E. O.; Szilagyi, I. M.; Sedlackova, K.; Pfeifer, J.; Toth, A. L.; Gouma, P. I. J. Eur. Ceram. Soc. 2008, 28, 913–917. (30) Zheng, K.; Gu, L.; Sun, D.; Mo, X.; Chen, G. Mater. Sci. Eng., B 2010, 166, 104–107. (31) (a) Cao, B. B.; Chen, J. J.; Tang, X. J.; Zhou, W. L. J. Mater. Chem. 2009, 19, 2323–2327. (b) Ashraf, S.; Blackman, C. S.; Palgrave, R. G.; Naisbitt, S. C.; Parkin, I. P. J. Mater. Chem. 2007, 17, 3708–3713. (32) Wang, L.; Pfeifer, J.; Balazsi, C.; Gouma, P. I. Mater. Manuf. Processes 2007, 22, 773–776. (33) Liu, Z.; Miyauchi, M.; Yamazaki, T.; Shen, Y. Sens. Actuators, B 2009, 140, 514–519. (34) (a) Rout, C. S.; Hedge, M.; Govindaraj, A.; Rao, C. N. R. Nanotechnology 2007, 18, 205504–205513. (b) Du, N.; Zhang, H.; Chen, B.; Ma, X.; Liu, Z.; Wu, J.; Yang, D. AdV. Mater. 2007, 19, 1641–1645.

JP106439N