Crystal structures of deuterated sodium molybdate dihydrate and sodium tungstate dihydrate from time-of-flight neutron powder diffraction

High-precision structural parameters for Na2MoO4·2D2O and Na2WO4·2D2O are reported based on refinement of high-resolution time-of-flight neutron powder diffraction data. Complementary Raman spectra are also provided.


Chemical context
Na 2 MoO 4 and Na 2 WO 4 are unusual amongst the alkali metal mono-molybdates and mono-tungstates in being highly soluble in water and forming polyhydrated crystals. Additionally, sodium apparently plays a significant role in the solvation of other alkali metal ions to form a range of double molybdate and tungstate hydrates (Klevtsova et al., 1990;Klevtsov et al., 1997;Mirzoev et al., 2010), for example, Na 3 K(MoO 4 ) 2 Á9H 2 O. Both dihydrate and decahydrate varieties of the two title compounds are known, their solubilities as a function of temperature being well characterised (Funk, 1900;Zhilova et al., 2008). The structures of the decahydrates have not yet been reported, although I have established that they are not isotypic with the sodium sulfate analogue, Na 2 SO 4 Á10H 2 O, as had hitherto been thought.
The dihydrates have been the subject of extensive crystallographic studies, from descriptions of their density, habit and measurements of interfacial angles (Svanberg & Struve, 1848;Zenker, 1853;Rammelsberg, 1855;Marignac, 1863;Delafontaine, 1865;Ullik, 1867;Clarke, 1877;Zambonini, 1923), through to determination of absolute unit-cell parameters (Pistorius & Sharp, 1961), and subsequent solution and refinement of their structures (Mitra & Verma, 1969;Okada et al., 1974;Matsumoto et al., 1975;Atovmyan & D'yachenko, 1969;Capitelli et al., 2006;Farrugia, 2007). However, the presence of heavy atoms in these materials makes it impossible to achieve a uniform precision on all structural parameters using X-rays, and even with single-crystal methods that purport to identify hydrogen positions there may be significant inaccuracies. Such problems are minimised using a neutron radiation probe since the coherent neutron scattering lengths of the constituent elements differ by less than a factor of two, being 6.715 fm for Mo, 4.86 fm for W, 3.63 fm for Na, 5.803 fm for O, and 6.67 fm for 2 D (Sears, 1992). Thus one can locate accurately all of the light atoms and obtain a uniform level of precision on their coordinates and displacement parameters. Since the incoherent neutron scattering cross section of 1 H is large (80.3 barns) it is usual to prepare perdeuterated specimens whenever possible (the incoherent cross section of 2 D being only 2.1 barns) as this optimises the coherent Bragg scattering signal above the background, reducing the counting times required for a high-precision structure refinement from many days to a matter of hours on the instrument used for these measurements. These data were therefore measured using Na 2 MoO 4 Á2D 2 O and Na 2 WO 4 Á2D 2 O samples.

Structural commentary
general positions (Wyckoff sites 8c). Note that the atom labelling scheme and space-group setting used here follows Farrugia (2007); consequently there are some differences with respect to other literature sources, although equivalent contacts are referred to in Table 1 and Table 2. The X 6+ ions (X = Mo, W) are tetrahedrally coordinated by O 2À , the Mo-O and W-O bond lengths varying slightly according to the type of coordination adopted by a particular apex: O1 and O4 are each coordinated to Na + and each also accepts two hydrogen bonds; O2 is coordinated to three Na + ions and O3 is coordinated to two Na + ions ( Fig. 1). In both title compounds, X-O1 and X-O4 are the longest contacts and X-O3 is the shortest contact in the tetrahedral oxyanion. The mean Mo-O and W-O bond lengths are in good agreement with those found in the anhydrous crystals (Fortes, 2015). Furthermore, each of the absolute Mo-O bond lengths are identical (within error) to those found by Capitelli et al. (2006); the agreement in W-O bond lengths with Farrugia (2007) is marginally poorer. The Na + ions occupy two inequivalent sites: in one, Na + is six-fold coordinated by two water molecules and four XO 4 2À oxygen atoms, yielding an octahedral arrangement; in the second, Na + is five-fold coordinated by two water molecules and three XO 4 2À oxygen atoms, yielding a square-pyramidal arrangement. These two polyhedra share a common edge (O2-O5) and are connected, moreover, with their inversioncentre-related neighbours along three other shared edges to form a cluster (Fig. 2a). The clusters corner-share via O6 to create a 'slab' parallel to (010) (Fig. 2b). The mean Na-O bond lengths are statistically identical in Na 2 MoO 4 Á2D 2 O and Na 2 WO 4 Á2D 2 O being $1.6% longer in the NaO 6 octahedra and $2.3% shorter in the NaO 5 polyhedra than Na-O bonds in the anhydrous crystals (Fortes, 2015). The agreement in Na-O bond lengths with the X-ray single crystal studies of Capitelli et al. (2006) and Farrugia (2007) is very good. Overall, the agreement in bond lengths and angles for the two independently refined data sets is excellent (Tables 1 and 2).
Although it is more usual to find Na + in octahedral coordination, there are abundant examples of Na + in five-fold coordination, including instances where the NaO 5 polyhedron adopts a square-pyramidal arrangement (Beurskens & Jeffrey, 1961;Císařová ;Sharma et al., 2005;Smith & Wermuth, 2014;Aksenov et al., 2014) or the alternative trigonal-bipyramidal arrangement (Mereiter, 2013;Smith, 2013). A similar combination of NaO 6 and NaO 5 polyhedra to that found in the title compounds occurs in the closely-related hydrates Na 2 CrO 4 Á1.5H 2 O and Na 2 SeO 4 Á1.5H 2 O (Kahlenberg, 2012;Weil & Bonneau, 2014). The two water molecules form hydrogen-bonded chains between the O1 and O4 atoms of the tetrahedral oxyanions; O5-related chains extend along [001] and O6-related chains crosslink them in a staggered fashion (a) Arrangement of NaO x polyhedra into edge-sharing clusters comprised of two Na1O 6 octahedra and two Na2O 5 square pyramids; (b) Arrangement of the clusters shown in (a) by corner sharing to form 'slabs' parallel (010). Ellipsoids are drawn at the 50% probability level.
There are no significant differences in the hydrogen bond geometries of the molybdate or tungstate crystals. The most recent X-ray single-crystal diffraction study of Na 2 WO 4 Á2H 2 O (Farrugia, 2007) implied that one of the water molecules (O5) was involved in a weaker three-centred interaction, although a similarly recent measurement of Na 2 MoO 4 Á2H 2 O (Capitelli et al., 2006) identified a 'normal' linear two-centred interaction for this bond. This work, using neutrons, has been able to accurately and precisely characterise the hydrogen bond geometry, showing that the latter is true for both structures; there is no bifurcated bond and all hydrogen-bonded interactions are of the linear two-centred variety. Presumably the error in Farrugia's analysis arose due to the substantial absorption correction required ( = 18.7 mm À1 ) for an accurate structure refinement from X-ray single-crystal data.
Raman spectra of Na 2 MoO 4 Á2H 2 O and Na 2 MoO 4 Á2D 2 O were first reported by Mahadevan Pillai et al. (1997); subsequently, Luz-Lima et al. (2010) and Saraiva et al. (2013) published the Raman spectra of Na 2 MoO 4 Á2H 2 O and Na 2 WO 4 Á2H 2 O as a function of temperature (13-300 K) and as a function of hydrostatic pressure (to 5 GPa). Both compounds exhibit evidence of a 'conformational change' on cooling through 120 K: the molybdate appears to undergo two high-pressure phase transitions, one at 3 GPa and the second at 4 GPa; the tungstate apparently undergoes a high-pressure phase transition at 3.9 GPa. The Raman spectra reported here (Figs. 4 and 5 and Supporting information) agree well with data in the literature (Table 3). The large blue-shifts in the internal vibrational frequencies of the deuterated water molecule are similar to the square root of the D:H mass ratio; the small blue-shifts of most of the internal modes of the tetrahedral oxyanions are consistent with stronger hydrogen bonding in the deuterated species, as expected (cf. Scheiner & Č uma, 1996;Soper & Benmore, 2008).

Synthesis and crystallization
Coarse polycrystalline powders of Na 2 MoO 4 Á2H 2 O (Sigma-Aldrich M1003 > 99.5%) and Na 2 WO 4 Á2H 2 O (Sigma-Aldrich 14304 > 99%) were dehydrated by drying at 673 K in air. The resulting anhydrous materials were characterised by Raman spectroscopy, X-ray and neutron powder diffraction (Fortes, 2015). This material was dissolved in D  [Symmetry codes: (i) 1 À x, 1 À y, 1 À z; (ii) 1 À x, 1 2 + y, 3 2 À z; (iii) 1 2 + x, 3 2 À y, 1 À z; (iv) 1 2 + x, y, 3 2 À z; (v) x, 3 2 À y, À 1 2 + z; (vi) x, 3 2 À y, 1 2 + z.] 99.9 atom% D) and twice recrystallized by gentle evaporation at 323 K. The molybdate crystallised with a coarse platy habit whereas the tungstate was deposited as a finer-grained material. Once the supernatant liquid was decanted, the residue was air dried on filter paper and then ground to a fine powder with an agate pestle and mortar. The powders were loaded into standard vanadium sample-holder tubes of internal diameter 11 mm to a depth not less than 20 mm (this being the vertical neutron beam dimension at the sample position). Accurate volumes and masses were determined after the diffraction measurements were complete and used to correct the data for self-shielding. The level of deuteration was determined by Raman spectroscopy (see below) to be $91% for both compounds. Raman spectra of Na 2 MoO 4 Á2H 2 O and Na 2 MoO 4 Á2D 2 O in the range 200-3900 cm À1 . Band positions and vibrational assignments are indicated (see also  Table 3). Vertical scales show intensities relative to 1 (XO 4 2À ).

Figure 5
Raman spectra of Na 2 WO 4 Á2H 2 O and Na 2 WO 4 Á2D 2 O in the range 200-3900 cm À1 . Band positions and vibrational assignments are indicated (see also Raman spectra were acquired with a B&WTek i-Raman plus portable spectrometer; this device uses a 532 nm laser (37 mW power at the fiber-optic probe tip) to stimulate Raman scattering, which is measured in the range 170-4000 cm À1 with a spectral resolution of 3 cm À1 . Data were collected for 600 sec at 17 mW for Na 2 MoO 4 Á2H 2 O (as Neutron powder diffraction data for Na 2 MoO 4 Á2D 2 O; red points are the observations, the green line is the calculated profile and the pink line beneath the diffraction pattern represents ObsÀCalc. Vertical black tick marks report the expected positions of the Bragg peaks. The inset shows the data measured at short flight times (i.e. small d-spacings).

Figure 7
Neutron powder diffraction data for Na 2 WO 4 Á2D 2 O; red points are the observations, the green line is the calculated profile and the pink line beneath the diffraction pattern represents ObsÀCalc. Vertical black tick marks report the expected positions of the Bragg peaks. The inset shows the data measured at short flight times (i.e. small d-spacings). bought), 180 sec at 37 mW for Na 2 MoO 4 Á2D 2 O, 300 sec at 17 mW for Na 2 WO 4 Á2H 2 O (as bought) and 220 sec at 37 mW for Na 2 WO 4 Á2D 2 O; after summation, the background was removed and peaks fitted using Pseudo-Voigt functions in OriginPro (OriginLab, Northampton MA). These data are provided as an electronic supplement in the form of an ASCII file. Small quantities of ordinary hydrogen were found to be present in both specimens, the proportion being determined by the ratio of the areas under the 1 / 3 (H 2 O) bands after normalisation relative to the height of the strong 1 (XO 4 2À ) band. The molar abundance of 1 H was used to correct the diffraction data for absorption (see below) and to ensure accurate refinement of the structure (see Refinement).
Time-of-flight neutron diffraction patterns were collected at 295 K using the High Resolution Powder Diffractometer, HRPD (Ibberson, 2009), at the ISIS spallation neutron source, Harwell Campus, Oxfordshire, UK. Data were acquired in the range of neutron flight times from 30-130 msec (equivalent to neutron wavelengths of 1.24-5.36 Å ) for 15.17 hr from the molybdate and 14.40 hr from the tungstate, equivalent to 615 and 590 mAhr of integrated proton beam current, respectively. These data sets were normalized to the incident spectrum and corrected for detector efficiency by reference to a V:Nb nullscattering standard and then subsequently corrected for the sample-specific and wavelength-dependent self-shielding using Mantid (Arnold et al., 2014: Mantid, 2013. In the case of the molybdate, the number density of the specimen was determined to be 3.28 mol nm À3 , with a scattering cross section, allowing for the water being 9.1 mol % 1 H, scatt = 93.81 b and an absorption cross section, abs = 3.66 b; for the tungstate, the number density was 3.01 mol nm À3 , the scattering cross section, allowing for the water being 8.6 mol % 1 H, scatt = 94.19 b and abs = 19.48 b. Diffraction data were exported in GSAS format and analysed with the GSAS/Expgui Rietveld package (Larsen & Von Dreele, 2000: Toby, 2001. The fitted diffraction data are shown in Figs. 6 and 7.