Recently I came across this equation:

$$\ce{WO_3 +SnCl_2 +HCl\rightarrow W_3O_8 + H_2SnCl_6 +H_2O}$$

The compound $\ce{W_3O_8}$ that appears in the right hand side of the equation seems to be mixed oxide. Normal calculation of the oxidation state of W in the compound yields $\ce{\frac{16}{3}}$ which is a fraction. A google search on the compound does not give any good results too.

So, is $\ce{W_3O_8}$ a mixed oxide? What are the oxidation states of the different tungsten atoms?

  • 3
    $\begingroup$ This might be helpful. researchgate.net/profile/Jun_Li36/publication/… $\endgroup$ – JM97 Jan 7 '17 at 15:44
  • 1
    $\begingroup$ this dx.doi.org/10.1063/1.3505689 article claims that some 'extra' electrons on W are localized on specific atoms. So it seems that the compound may be classified as mixed oxide. However, it belongs to a weird family of Magneli phases, which are ... somewhat special. They are not non-stechiometric per se, but close to it. I'm not qualified enough to give a full conclusion, though. $\endgroup$ – permeakra Jan 14 '17 at 11:58

TL; DR: If you are forced to list the formal oxidation states, then with a high level of confidence it's tungsten(V,VI) octaoxide $\ce{W^{VI}W^V2O8}$. However, the real product of this reaction would be a number of non-soichimetric tungsten(IV,V,VI) oxides. Formation of molecular oxoclusters or peroxo compounds is very unlikely.

Reduced tungsten oxides are generally non-stoiciometric. Glemser et al. [1] reduced tungsten(VI) oxide hydrates $\ce{WO3 * H2O}$ and $\ce{WO3 * 2 H2O}$ with $\ce{Zn}$ in $\ce{HCl}$ to the blue tungsten oxide hydrates of the general formula $\ce{WO_{3‐x}(OH)_x}$ $(x → 0.5)$ with the lowest reducing degree of $2.65$, which is very close to the one in $\ce{W3O8}$ $(8/3 = 2.6(6))$.

I would expect that $\ce{SnCl2}$ in acidic media would have comparable reducing effect. With the moderately mild reducing agent such as tin(II) chloride there is very little chance of ending up exclusively with $\ce{W3O8}$. Also, a presence of a reducing agent also eliminates a possibility of the formation of the peroxo-compounds, so that with a high level of confidence we are dealing with oxide $\ce{O^2-}$ ligand.

I find it equally improbable that any cluster formation takes place. For the existence of a polyoxotungstate cluster the terminal $\ce{W=O}$ bonds must be "disarmed" in order to prevent further olation/oxolation and aggregation into bulk metal oxide. This is often achieved via the template effect, however even reduced by 1 to 6 electrons forms of 12-heteropolytungstates $\ce{[(P,Si)W12O40]^3-}$, the phosphate- or silicate-templated Keggin structure — one of the most stable — only exist briefly in solution [2, pp. 341–343]. In the presented system there is no suitable anion to preserve the three-membered $\ce{W3O8}$ ring, and it's very likely going to hydrolyze quickly.

To sum these thoughts up, the more appropriate way to write the reaction would be

$$\ce{n WO3 + SnCl2 + 4 HCl → W_nO_{3n-1} + H2SnCl6 + H2O}$$

where $\ce{W_nO_{3n-1}}$ is homologous series analogous to the mixed-valence tungsten blues. On crystallographic level this can be seen as withdrawal of oxygen atoms from $\ce{WO3}$, which is isostructural to $\ce{ReO3}$ (Fig. 1a). New point defects are eliminated by introducing a crystallographic shear (CS, Fig. 1b). As the reduction proceeds, the octahedra $\ce{[WO6]}$ in this structure changing from corner-sharing to edge-sharing motif (Fig. 1c). (Caption and adapted illustration from [3, p. 1086].)

Crystallographic shear in WO3

Figure 1. Crystallographic shear in $\ce{WO3}$ (a) the idealized structure of $\ce{WO3}$, drawn as corner-shared $\ce{WO6}$ octahedra, squares in projection (b) {102} CS planes (arrowed), consisting of blocks of four edge-shared octahedra (c) {103} CS planes (arrowed), consisting of blocks of six edge-shared octahedra.

It's impossible to deduce exact oxidation number for such structure with floating stoichiometry, but sure enough an average oxidation number of tungsten can be found trivially from the $\ce{W_nO_{3n-1}}$ formula:

$$\text{O.N.}(\ce{W})\cdot n - 2(3n - 1) = 0 \qquad\to\qquad \text{O.N.}(\ce{W}) = 6 - 2/n,$$

e.g. for $\ce{W3O8}$ $\text{O.N.}(\ce{W}) = 6 - 2/3 = 5.3(3)$. Boundary case $n = 1$ dictates the lowest $\text{O.N.}(\ce{W}) = 4$, so that $\ce{W_nO_{3n-1}}$ phase consists of a mixed-valence oxide with various ratios of $\ce{W(VI)}:\ce{W(V)}:\ce{W(IV)}$. Exact ratio is to be determined from XPS/XANES analysis for a given specimen; spectrophotometry would be difficult to implement as tungsten oxides have negligible solubility in most solvents unless some peroxide or $\ce{HF}$ is added. Exact location of the reduced tungsten centers can be averaged by finding CS areas with EM/PXRD analysis.

Now, to the $\ce{W3O8}$. In fact, it's already been synthesized by Zakharov et al. as follows [4]:

The sample examined was prepared by partial reduction of $\ce{WO3}$ with carbon at high temperature in combination with high pressure. The reaction occurred between the graphite container material and a pressed tablet of $\ce{WO3}$ in a closed system ($T = \pu{1773 K}$, $P = \pu{50e5 kPa}$). [...] The product obtained was a mass containing minute dark-red crystallites. The colour suggested that considerable reduction had taken place.

and two phases were characterized with EM/PXRD. Authors also suggest that $\ce{W3O8}$ can hardly be obtained under ambient conditions:

The two phases, both of composition $\ce{W3O8}$ $(\ce{WO_{2.667}})$, do not appear in the binary $\ce{W-O}$ system at ambient pressure. $\ce{W3O8(I)}$ has a structure of the $\ce{U3O8}$ type (Andresen, 1958), which is denser than that of $\ce{W3O8(II)}$. The latter has a new type of structure, although some features are shared with other previously known tungsten oxides ({102} CS structures).


Figure 2. Idealized polyhedral representation of $\ce{W3O8(I)}$

enter image description here

Figure 3. Idealized polyhedral representation of $\ce{W3O8(II)}$

So, there are two possible combinations for the formal $\text{O.N.}$s:

  1. $1~\ce{W(VI)} + 2~\ce{W(V)}$;
  2. $2~\ce{W(VI)} + 1~\ce{W(IV)}$.

Unfortunately, oxidation states are not mentioned in the article and I haven't found any citeable sources with the determination of oxidation numbers for $\ce{W3O8}$, but the XPS studies of uranium octaoxide $\ce{U3O8}$ indicated the presence of two oxidation states, $\ce{U(VI)}$ and $\ce{U(V)}$, in a $1:2$ ratio, respectively. Since phase $\ce{W3O8(I)}$ is isostructural, there is a high chance that the scenario 1, $1~\ce{W(VI)} + 2~\ce{W(V)}$, is quite plausible.

Having the CIFs of both structures at hands, I've been trying to perform bond-valence analysis, but with the common tabulated empirical parameters I haven't reached any meaningful results. Mixed-valenced tungsten and molybdenum oxides often contain delocalized $\mathrm{d}$ electrons, making bond-valence analysis alone less meaningful [5].


  1. Glemser, O.; Weidelt, J.; Freund, F. Genotypische Oxidhydrate des Wolframs. Zur Frage der Wolframblauverbindungen. Zeitschrift für anorganische und allgemeine Chemie 1964, 332 (5–6), 299–313. https://doi.org/10.1002/zaac.19643320511.
  2. Advances in Inorganic Chemistry and Radiochemistry; Emeléus, H. J., Sharpe, A. G., Eds.; Advances in inorganic chemistry and radiochemistry; Academic Press: New York, 1967; Vol. 10. ISBN 978-0-08-057859-0.
  3. Encyclopedia of Inorganic Chemistry, 2nd ed.; King, R. B., Ed.; Wiley: Chichester, West Sussex, England ; Hoboken, NJ, 2005. ISBN 978-0-470-86078-6.
  4. Sundberg, M.; Zakharov, N. D.; Zibrov, I. P.; Barabanenkov, Y. A.; Filonenko, V. P.; Werner, P. Two High-Pressure Tungsten Oxide Structures of $\ce{W3O8}$ Stoichiometry Deduced from High-Resolution Electron Microscopy Images. Acta Crystallographica Section B Structural Science 1993, 49 (6), 951–958. https://doi.org/10.1107/S0108768193005701.
  5. Domenge`s, B.; McGuire, N. K.; O’Keeffe, M. Bond Lengths and Valences in Tungsten Oxides. Journal of Solid State Chemistry 1985, 56 (1), 94–100. https://doi.org/10.1016/0022-4596(85)90256-7.
| improve this answer | |

Note that $\ce{W3O8}$ is a non-stoichiometric compound and it is not that simple compound you may think. It has a very complex structure as such its oxidation state is very difficult to calculate. I am not going in details about the structure of the compound and I will only mention it a non-stoichiometric compound actually a oxygen-deficient cluster.

Please give a good read to Xin Huang et.al., J. Phys. Chem. A 2006, 110, 85-92, the link that @JM97 has posted. It says that the $\ce{O-O}$ is much longer than normal oxide bond and so it is assumed to be a peroxo bond $\ce{O2^2-}$. Now, peroxide has oxidation state of -1. So, assuming $\ce{W3O8}$ to be a simple compound (no stoichiometric defect or any kind of defect), the oxidation state is:

$$3x + 8(-1) = 0$$ $$x = +{8/3}$$

| improve this answer | |
  • $\begingroup$ There is no way a molecular cluster of such composition is going to be produced by the OP's reaction, it only exists in the gaseous phase and is very unstable, not to mention that it's a reducing media (no peroxides). The non-stoichimetric oxides are the so-called Wadsley and Magneli phases $\ce{M_nO_{3n-2}}$ and $\ce{M_nO_{3n-1}}$, whereas $\ce{W3O8}$ polymorphic phases preserve its stoichiometry all right. $\endgroup$ – andselisk Feb 5 '19 at 1:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.