Elucidating the physical properties of the molybdenum oxide Mo4O11 and its tantalum substituted variant Mo2Ta2O11

Abstract Although γ/η-Mo4O11 and Mo2Ta2O11 are used in a variety of industrial applications and can easily be synthesized in a chemical vapour transport (CVT) process or reactions in silica ampoules, respectively, only few data are available concerning their physical properties. In this paper, we further explore the properties of the three compounds with respect to their thermal and magnetic behavior, surface composition, and Raman spectroscopic properties.


Introduction
Elemental molybdenum comprises a tremendous diversity in its range of applications. Utilization covers i.e. the heavy industry, where molybdenum is used in a mass range of 0.25-8% to drastically increase toughness and significantly decrease brittleness of steel. Another important application of molybdenum is its use as a trace component in plant fertilizers (in form of molybdates) and as a trace element in all living organisms in general. Additionally, elemental molybdenum plays an important part in catalysts used in petrochemical processes and as an anode material in X-ray generators [1][2][3][4].
The availability of molybdenum is not that scarce, as large deposits of molybdenum containing ore are known to be situated in China, the USA, Chile, and Canada. However, molybdenum is not available in its genuine form and has to be isolated from naturally occurring molybdenum minerals like Sidwillite, Umohoite, Iseite, Molybdenite, or Ilsemannite. Today, molybdenum is produced by extraction from these minerals as well as by extraction from the copper production process, where it is interspersed in the Chalcopyrite ore and is extracted as a side product. In both processes the ultimately used mineral is Molybdenite (molybdenum disulfide), which is detached from other ores by froth flotation. Subsequently, the molybdenum disulfide is oxidized by roasting to form molybdenum trioxide, which in turn can be reduced with hydrogen via molybdenum dioxide to yield the pure molybdenum. To produce an even purer form of elemental molybdenum, a cleaning step can be introduced, where the molybdenum trioxide is transferred into the ammonium heptamolybdate by reaction with ammonia, before the reduction with hydrogen is proceeded. The reduction process with hydrogen does not work in a simple one step reaction as during the process numerous molybdenum oxides like Mo 4 [1,[5][6][7][8].
Especially, the binary oxides γ/η-Mo 4 O 11 , with γ referring to the phase synthesized at higher temperatures and η to the phase synthesized at lower temperatures, are of particular interest, as they form in large quantities during the reduction process. The first step of the reduction to elemental molybdenum starts with the molybdenum trioxide (MoO 3 ), which is reduced to Mo 4 O 11 , and ends with molybdenum dioxide (MoO 2 ). As of today it is indicated, that the morphology of this first intermediate Mo 4 O 11 is of crucial importance to the morphology of the molybdenum dioxide product and therefore to the elemental molybdenum. The morphology of the final molybdenum product in turn has vital influence on the sinter capabilities of the elemental powder, which is especially important in the general production of molybdenum-based materials [9][10][11].
Disregarding the structural characterizations via X-ray diffraction, hardly any investigations into the physical properties of the individual molybdenum oxides were made. The electrical conductivities of both Mo 4 O 11 phases were investigated by Gruber and Krautz [24]. The specific electrical resistivity was given as 0.53 Ωm (metallic) for η-Mo 4 O 11 and 26 Ωm (semiconducting) for γ-Mo 4 O 11 .
In this work, we want to focus on the characterization of the physical properties of the binary oxides γ/η-Mo 4 O 11 and on one of its tantalum substituted forms, the phase Mo 2 Ta 2 O 11 [25].

Experimental section Synthesis
The pure molybdenum oxides were synthesized by the chemical vapour transport (CVT) method. To prevent contamination with oxygen and water vapour, the initial preparations were made in a glovebox. The reactions were performed in evacuated, torch sealed fused silica ampoules (l = 250 mm, Ø = 25 mm, thickness of the tube walls = 1.8 mm). A mixture of 1636.5 mg MoO 3 (11.4 mmol), 363.7 mg MoO 2 (2.8 mmol) and 176.6 mg TeCl 4 (0.7 mmol) yielding a final Mo:O ratio of 1:2.80 was thoroughly ground together in an agate mortar and transferred into a silica ampoule. To yield the phase η-Mo 4 O 11 , the oven was kept at 798-748 K (educt zone -product zone) for 5 days. Afterwards, the oven was subsequently cooled to room temperature. For the phase γ-Mo 4 O 11 , the oven was kept at 923-873 K for 5 days. Due to the inherent nature of the CVT process, the final product forms as a crystalline compound on the ampoule walls as shown in Figure 2. η-Mo 4 O 11 forms thin layered crystals (up to 3 cm in size). The phase γ-Mo 4 O 11 on the other hand, yields smaller and thicker crystals as shown in Figure 3. The deep wine-red colour and bronze reflections are typical for both phases of Mo 4 O 11 .
To obtain the tantalum substituted phase Mo 2 Ta 2 O 11 , the reaction was performed in an evacuated, torch sealed silica glass ampoule (l = 80 mm, Ø = 25 mm, thickness of the tube wall = 1.8 mm). A mixture of 443.3 mg MoO 3 (3.1 mmol) and 556.7 mg Ta 2 O 5 (1.3 mmol) was ground together in a agate mortar and heated up to a temperature of 1123 K and kept there for 15 days. Afterwards, the ampoule was cooled with a rate of 1 K min −1 to room temperature. The product was obtained as a compressed grey-purple powder ( Figure 4). In comparison to the synthesis described in [25], the temperature is higher (about 100 K) 2 2   and the duration of the heating process longer (compared to 3 h). However, it seems that in contrast to an open air reaction, the silica ampoules enable a reaction without much MoO 3 loss (only 20% MoO 3 excess used), while still yielding the phase pure product Mo 2 Ta 2 O 11 . The synthesis was therefore very similar to [26], where the phase Mo 2 Ta 2 O 11 was falsely described as Ta 2 O 5 · 3MoO 3 (Mo 3 Ta 2 O 14 ).
Relevant details of the data collections and the refinements using the Rietveld method are listed in Table 1.

Thermal investigations
Heat capacity: The thermal behaviour in the temperature range from 303 to 1103 K was studied under inert conditions with a concurrent thermal analysis apparatus (NETZSCH STA F2 Jupiter ® ; NETZSCH-Gerätebau GmbH, Selb, Germany). The NETZSCH Applikationslabor (Selb, Germany) conducted the measurements.
In addition, a further measurement was performed using differential scanning calorimetry (DSC) between 298 and 573 K applying a Netzsch DSC 204 F1 Phoenix equipped with a τ-sensor. The sample with a weight of 17.14(2) mg was placed in an Al-crucible. For this measurement, the heating and cooling rate was set to 6 K min −1 and helium was used as a protective gas.

Thermal expansion High-temperature dilatometry of γ-Mo 4 O 11
The temperature-induced strain was studied between 298 and 720 K on an as grown plane-parallel plate of γ-Mo 4 O 11 using a commercial inductive gauge dilatometer (DIL 402C from Netzsch) equipped with a sample holder made of α-Al 2 O 3 , type S thermocouples and a high-temperature furnace (type 6.219.1-26 from Netzsch). The normal of the plane-parallel plate runs parallel to [100] with a = 2444.9 pm as determined by X-ray diffraction. The thickness of the sample along [100], which is the direction of the temperature-induced strain measurements, was 1.548 mm and the deviation from plane-parallelism of opposing faces was smaller than 10 μm. The γ-Mo 4 O 11 -sample was measured twice in air with heating/cooling rates of 0.6 and 1 K min −1 , respectively. The dilatometer was calibrated with standard samples made of α-Al 2 O 3 .

High-temperature X-ray (HT-XRD) powder diffraction of η-Mo 4 O 11 and Mo 2 Ta 2 O 11
The HT-XRD experiments were conducted using a Rigaku Ultima IV diffractometer with a thermal attachment (Cu-K α , 40 kV and 35 mA, reflection geometry, D/teX Ultra high-speed detector, air atmosphere, 2θ = 10-80°, temperature range 293-1173 K, step size 10 K). Before the HT-XRD experiments, a Si external standard was measured in the  temperature range 293-1273 K in order to control the thermal expansion coefficients. The temperatures of the phase transitions were checked using SiO 2 and K 2 SO 4 . The error in the determination of the temperature did not exceed ±10 K. Experimental data processing by the Rietveld refinement, approximation of temperature dependencies of lattice parameters, and drawing of the α figures were performed using RietToTensor [36].

Magnetic investigations
The single crystals of γ-Mo 4 O 11 , η-Mo 4 O 11 , and Mo 2 Ta 2 O 11 obtained via chemical vapour transport were ground into fine powders in an agate mortar and subsequently packed into polyethylene (PE) capsules and attached to the sample holder rod of a Vibrating Sample Magnetometer (VSM) for measuring the magnetization M(T,H) in a Quantum Design Physical Property Measurement System (PPMS). The samples were investigated in the temperature range of 3-300 K with an applied external magnetic field of 10 kOe. The recorded susceptibilities were corrected by the diamagnetic contributions caused by the PE capsules.

XPS measurements
X-ray photoelectron spectroscopic (XPS) measurements were carried out using a Thermo Scientific MultiLab 2000 spectrometer with a base pressure in the low 10 −10 mbar range. The instrument is equipped with a monochromated Al-K α X-ray source, an Alpha 110 hemispherical sector analyzer as well as a flood gun for charge compensation, providing electrons with a kinetic energy of 6 eV. Wherever possible, the C 1s peak (set to 284.8 eV) was used to calibrate the energy axis shift. Survey spectra were collected to determine the surface purity of the samples, whereas high-resolution spectra of the O 1s, C 1s, Ta 4d, and Mo 3d regions were used to derive information of the relevant oxidation states of the respective elements. Deconvolution was carried out by fitting mixtures of Gaussian and Lorentzian functions (30% Lorentzian character) as peak shapes for each component, limited by several constraints in the fitting process: for each component, the peak splitting was set to a fixed value, depending on the transition and the element. The peak width for the respective spin-orbit components were restricted to the same value. The respective peak area ratios were set to 3:2 for the d 5/2 :d 3/2 component and 4:3 for the f 7/2 :f 5/2 component. A Shirley-type background was used for baseline correction. For calculation of the surface ion concentration, relative sensitivity factor (RSF) and electron mean free path corrections have been applied. The mean free path correction was based on values from the NIST database.

Raman spectroscopic measurements
The Raman spectroscopic measurements were done on a LabRam HR 800 spectrometer equipped with a 1024 × 256 CCD detector ( Peltier-cooled) combined with an Olympus BX41 microscope. All measurements were carried out using a laser wavelength of 532 nm and a total laser power of approximately 12 mW. The laser wavelength was chosen to avoid resonant effects on Mo 4 O 11 , as observed in [37]. A 300 L mm −1 grating (spectral resolution approximately 3.5 cm −1 ) was used and all spectra were baseline corrected using a second order polynomial and normalized by the unit vector method (both LabSpec 6).
The measurements of Mo 2 Ta 2 O 11 were carried out on a powder sample using an Olympus ×50 LNPlanFLN objective (NA = 0.5). The measurements of both the monoclinic and orthorhombic η/γ-Mo 4 O 11 modification were done on a platelet sample using an Olympus ×10 MPlanN objective (NA = 0.25). The two different orientations of the sample relative to the laser polarisation were achieved by turning the sample by 90°. For the measurement of the third orientation of monoclinic η-Mo 4 O 11 , the sample was embedded upright in an epoxy resin and polished to give a smooth surface.

Elemental analysis
Elemental analyses were conducted for all three compounds to determine the composition with regard to molybdenum and oxygen for the binary compounds and with regard to molybdenum, tantalum, and oxygen for Mo 2 Ta 2 O 11 . The analyses were executed via the Mikroanalytisches Labor Pascher (Remagen, Germany).

Results and discussion
The compounds discussed in this paper are very similar with respect to their crystal structures. Both phases, γand η-Mo 4 O 11 , exhibit condensed MoO 6 octahedra forming bands along the c-axis, which are connected through MoO 4 tetrahedra. In the case of η-Mo 4 O 11 , the orientation of the octahedra stays the same throughout the structure, whereas in the case of γ-Mo 4 O 11 the orientation changes from band to band (see Figure 5; top and middle).
As investigated in [28], the phase η-Mo 4 O 11 is prone to twinning along [001]. In the twinning intergrowth area, building failures can occur, resulting in the orientation change of the octahedra, as observed in γ-Mo 4 O 11 . Additionally, the molybdenum atoms are not situated in the centre of the octahedra, but rather moved towards one of the octahedra side planes, therefore yielding three longer (γ: Ø = 206.7 pm; η: Ø = 206.6 pm) and three shorter (γ: Ø = 181.9 pm; η: Ø = 181.9 pm) Mo-O distances (overall average: γ: 194.3 pm; η: 194.2 pm) [15,28]. This effect is most prevalent in those octahedra connected to the tetrahedral MoO 4 entities, whereby the shortest Mo-O distances (γ: Ø = 175.8 pm; η: Ø = 176.3 pm) are found in the tetrahedra [15,28]. Concluding from the Mo-O distances, the molybdenum in the tetrahedra is a Mo(VI) atom and the octahedra are statistically occupied by Mo(V) and Mo(VI) atoms in a ratio of 2:1.
In comparison to the Mo-O distances given above, the Mo-O distances in the binary molybdenum oxide MoO 3 , consisting exclusively of MoO 6 octahedra, range from 167.8  [40]. Therefore, the conclusions drawn above agree well with the coordination dependent Mo-O distances found in the literature.
The phase Mo 2 Ta 2 O 11 was first described by Berendts et al. [25]. The fundamental building block is derived from the ReO 3 type and is built up of Ta V O 6 octahedra (3 × 197(2) and 3 × 189.9(3) pm) and Mo VI O 4 tetrahedra (178(3) and 3 × 182(2) pm). The compound forms a layer like structure stacking along the c-axis (see Figure 5; bottom).
For all three compounds, elemental analysis were performed as depicted in Table 2. The experimentally determined values agree well with the theoretical ones, at least within the accuracy of the methods utilized. The phase Mo 2 Ta 2 O 11 has already been examined before, yielding

Powder diffraction data
The Rietveld refinements of the experimental products are shown in Figure 6. The experimental data is shown in black, the best fit profiles in red, and the difference curves in blue. The experimental data agrees well with the single crystal data from the ICSD, a short comparison of these phases is given in Table 3.

Thermal investigations Heat capacity
In the initial heating process for the phase η-Mo 4 O 11 , a mass loss of 0.9% was observed. The change was attributed to volatile substances adhered to the surface of the sample. After confirming mass consistency by conducting a second heating run, the heat capacity was measured. The heat capacity for η-Mo 4 O 11 was determined to be 0.507 J g −1 K −1 at 373 K. The heat capacity of the phase γ-Mo 4 O 11 showed a non-linear trend with a pronounced peak at 443 K stemming from a phase transition to a yet unidentified phase. This phase transition is fully reversible exhibiting no significant hysteresis and can be classified to be of weakly first order. The heat capacity of γ-Mo 4 O 11 could therefore only be measured up to 443 K and was determined to be 0.584 J g −1 K −1 at 373 K. For the tantalum substituted phase Mo 2 Ta 2 O 11 , the heat capacity was determined to be 0.459 J g −1 K −1 at 373 K. The temperature dependence of the heat capacity for all three substances is depicted in Figure 7. The differential scanning calorimetry (DSC) depicted in Figure 8 indicates the reversibility of the phase transition occurring at 443.4 K, as the peak in the DSC-signal can be observed in the heating curve (red) and the cooling curve (blue).

High-temperature dilatometry of γ-Mo 4 O 11
The temperature-induced strain ε 11 of γ-Mo 4 O 11 was directly determined from the longitudinal strain along [100]. Heating and cooling runs exhibit reversible discontinuities at ca. 440 K, which are hints to a phase transition of weakly first order (Figure 9). The thermal expansion of γ-Mo 4 O 11 is non-linear ( Figure 9); thus, second-order polynomials of the type ε 11 (T) = α 11 ΔT + β 11 (ΔT) 2 with ΔT = T-T 0 are required for a proper approximation of the strain ε 11 over a temperature range between 330 and 436 K. T 0 = 400 K denotes the reference temperature and α 11 = 7.2(4) · 10 −6 K −1 and β 11 = 3.4(1) · 10 −8 K −1 are the corresponding coefficients of linear and squared thermal expansion referring to 400 K. The reproducibility of the linear thermal expansion coefficient α 11 is of the order of 5%.

High-temperature X-ray powder diffraction of η-Mo 4 O 11 and Mo 2 Ta 2 O 11
Both oxides undergo an assumable solid-phase decomposition ( Figure 10).  Figure 11. The unit cell parameters for both compounds were approximated using squared polynomials in the temperature ranges from 293 to 673 K (η-Mo 4 O 11 ) and from 293 to 1023 K (Mo 2 Ta 2 O 11 ). The calculated thermal expansion coefficients at some specific temperatures are given in Table 4.   The monoclinic oxide η-Mo 4 O 11 expands anisotropically and the degree of the anisotropy increases as the temperature increases (Figure 12a). A mixed-anion framework is composed of the MoO 4 tetrahedra and MoO 6 octahedra, which are connected to each other through common vertices. The maximum thermal expansion α 11 is along the direction that is close to the short diagonal of the ac parallelogram (Figure 12a; top), which is consistent with the theory of hinges deformation in monoclinic and triclinic crystals [51,52].
The layered trigonal Mo 2 Ta 2 O 11 expands highly anisotropically and the negative expansion (contraction) is observed within the layers. The Mo atoms are in the tetrahedral coordination of the oxygen atoms and the tantalum atoms are in an octahedral coordination environment. The polyhedra are connected to each other through common vertices forming the thick layers. The maximum expansion is along the c axis (α c = 12.44(6) · 10 −6 K −1 at 293 K) (Figure 12b). The expansion can also be described in terms of hinges deformation. The hinge cell is Ta1-O3-Mo1-O3-Ta1-O2-Ta1. The O2 atoms are on special positions and the Ta1-O2-Ta1 angle equals 180° (Figure 12c), whilst the Ta1-O3-Mo1 angle is 166.4° at room temperature [25] and could increase up to 180° under heating. If the Ta1-O3-Mo1 angle increases, the Mo1 atom could shift in the direction of the O1 atom while two Ta1 atoms come closer to each other. The angles between the polyhedra can vary more than the interior polyhedral angles. Thus, the layer expands along the c-axis and contracts along the a-axis.  (2) 2.13(4) α 33 2.12(3) 1.66 (7) 1.14(2) α a 4.27 (7) 4.51 (2)

Magnetic investigations
For all compounds, the magnetic susceptibility was investigated in zero-field-cooled mode (ZFC) with an applied external field of 10 kOe in the temperature range of 3-300 K ( Figure 13). All curves show almost temperature independent behaviour with small upturns below 25 K, which can be linked to traces of paramagnetic impurities. The binary molybdenum oxides in contrast, exhibit a very different magnetic behaviour. A weak positive but also temperature independent susceptibility is observed, contradicting a potentially expected paramagnetism caused by the Mo 5+ (4d 1 ) cations according to (Mo 5+ ) 2 (Mo 6+ ) 2 (O 2− ) 11 . The present magnetism is more in line with band-paramagnetism, caused by a metallic-like behaviour of the material. This, however, implies that no ordering of the Mo 5+ /Mo 6+ cations but rather a delocalization is present. Studies of the electrical resistivity and the band structures of γand η-Mo 4 O 11 confirm this assumption. The Mo 4d states are filled and therefore lower in energy, the Fermi level therefore is found in the conduction band leading to the metallic character [54][55][56]. The absence of charge ordering has been already predicted in structural studies [28].

XPS measurements
The relevant Mo 3d spectra of the phases ηand γ-Mo 4 O 11 are shown in the panels a and b of Figure 14   The sum formula for Mo 4 O 11 allows for different oxidation states of molybdenum in the structure. With the mean oxidation state being +5.5, Mo 4 O 11 is at first expected to incorporate an equal amount of Mo(V) and Mo(VI), as indicated by Inzani et al. [54]. Taking into account the crystallographic data provided by Knorr and Müller [28], Mo 4 O 11 seems to consist of Mo(VI) inside the tetrahedra and Mo(V) and Mo(VI) in a ratio of 2:1 inside the octahedra, resulting in an overall ratio of 2:2. XPS measurements have previously been undertaken, indicating the presence of Mo(IV), Mo(V), and Mo(VI). However, as described in [59,60], the distribution of Mo(VI), Mo(V), and Mo(IV) depends on the specific temperature, at which the samples were synthesized.
For Mo 2 Ta 2 O 11 (Panels c, d and f of Figure 14), deconvolution of the Mo 3d region is not possible due to strong overlap with the Ta 4d region. The qualitative peak shape (intensity profiles, as well as the low binding energy shoulder) is, by direct comparison to Panels a and b, very similar, inferring an at least qualitatively similar distribution of Mo species. The Ta 4f region shows the presence of three Ta oxidation states between Ta(V) and Ta(III). Ta(V) and Ta(IV) are present in an almost 1:1 ratio, whereas Ta(III) is clearly the minority component (approximately half of the at.-% amount compared to Ta(V) or Ta(IV)). The Ta 4f 7/2 components of Ta(V), Ta(IV), and Ta(III) were measured at binding energies of 29.5, 28.8, and 25.2 eV. Due to charging issues on that particular sample, the Ta 4f binding energies are shifted to too high binding energies with respect to literature data [61]. A protonated OH species is also present at 531.5 eV.

Raman spectroscopic measurements
The Raman spectrum of Mo 2 Ta 2 O 11 with the band positions marked is shown in Figure 15. To the best of our knowledge, no Raman spectrum of Mo 2 Ta 2 O 11 is available in the literature. Therefore, this spectrum was mainly recorded to serve as a future reference for the identification of Mo 2 Ta 2 O 11 .
In contrast, Raman spectra of both η-Mo 4 O 11 [5,62,63] and γ-Mo 4 O 11 [37,63] have been reported in the literature. All of these reports have found little to no difference between the band position, but vast differences in relative band intensities even between the same phase of η/γ-Mo 4 O 11 [5,62,63]. This can in part be explained due to the use of different laser excitations. Also little to no difference was found between the Raman spectra of η-Mo 4 O 11 and γ-Mo 4 O 11 , especially by Olson [63], which is the only report we are aware of that has measured both phases thus far. The Raman spectra measured of η/γ-Mo 4 O 11 in this study are shown in Figure 16. The band positions are consistent with the literature and there is little difference between η-Mo 4 O 11 and γ-Mo 4 O 11 . However, interestingly a strong dependence of the relative band intensities on the orientation of the sample relative to the laser polarization was found. The relative band intensities for the orientation referred to as "pol x" are consistent with what was found by Blume [5] and Borovška et al. [62], whereas the orientation "pol z" is consistent with the relative intensities in [37] and [63]. Thus, orientation effects seem to be a dominated factor in the measurement of Mo 4 O 11 Raman spectra and need to be considered.

Conclusion
Stability up to high temperatures was observed for η-Mo 4 O 11 and Mo 2 Ta 2 O 11 . The phase γ-Mo 4 O 11 on the other hand is only stable to around 443 K, where a phase transition is indicated by DSC and dilatometry measurements. Heat capacity was measured for all three compounds, with the highest value observed in γ-Mo 4 O 11 . As expected, Mo 2 Ta 2 O 11 displays diamagnetism, whereas the binary oxides show a very different magnetic behaviour that is more in line with band-paramagnetism. The vast difference in band intensities found in Raman spectra of η/γ-Mo 4 O 11 are attributed to orientation effects dominating the Raman measurements of η/γ-Mo 4 O 11 . In contrast, no orientation effects are found with the Raman measurements of Mo 2 Ta 2 O 11 .