The structure of the octacyanomolybdate (IV) ion

Syeotrochimica Acta,1965, Vol.21,pp. 1087 to 109%Pergamon PremLtd.Printedin Northern Ireland

The structure of the octacyanomolybdate(I)

ion

S. F. A. KETTLE and R. V. PARISH Department of Chemistry, The University, Sheffield and School of Chemistry, The University, Newcastle-upon-Tyne (Received 28 September 1964) Abstract-The infre-red spectra of K,Mo(CN),.2H,O and of H,Mo(CN)@H,O in solution and of the former also in the solid phase have been measured in the range 1900-2100 cm-l. The spectrum of the solid salt is consistent with the dodecahedral structure reported by HOARD and NORDSIECK. The solution spectra consist of a single, broad band. An explanation for the latter spectrum is proposed, in which the dodecahedron is regarded as a distorted cube, and it is concluded that the vibrational spectra are not diagnostic of the structure. ALTHOUGH the subject has been discussed for many years, it is still uncertain whether the structure of the Mo(CN),*- ion is that of a dodecahedron (D,, symmetry) or of a square antiprism (D&. Energetically, there appears to be little difference between these two [l]. In the former the &,-orbital is not involved in u-bonding, while in the latter all five d-orbitals can participate. However, it is generally accepted that metal-ligand n-bonding is important in cyanide complexes and it is probable that the dodecahedral structure would provide a better opportunity for this. On these grounds the Iigand replacement reactions of the anion have been explained [2]. The electronic spectrum of the ion does not appear to allow a unique diagnosis of the structure and has been interpreted in terms of both models 113-51. The crystal structure of K,Mo(CN),.2H,O has been determined by HOARD and NORDSIECK [6], who found that the anion was approximately dodecahedral, packing and interlocking of the anions resulting in the true symmetry being C,. Several workers have reported and discussed the vibrational spectra of the ion but no definite conclusions emerge from their work. The C-N stretching region (~2100 cm-l) is most readily discussed for here the problem of assignment is minimal; we confine ourselves to this part of the spectrum. The Raman spectrum of an aqueous solution of the potassium salt was found by STAMMREICHand SALA to contain three lines in the C=N stretching region [7]. This observation, coupled with HILDAGO and MATTHIEU’S report [8] of two bands in the infra-red spectrum of the solid, led them to propose that the anion had D,, symmetry in both solid and solution. However, BONINO and FABBRI had already reported three strong bands in the infra-red spectrum of the solid but only one band [l] J. L. HOARD and J. V. SILVERTON,Inorg. Chem. 2, 235 (1963). [2] L. E. OROEL, J. Inorg. NucZear Chem. 14, 136 (1960). [S] [4] [5] [S] [7]

[8]

A. BERTOLUZZA and A-M. MARINANOELI, Rend. Accud. NazZ. XL (4), 10, 82 (1959). J. R. PERDMAREDDI, A. D. LIEHRand A. W. ADAMSON, J. Am. Chem. Sot. 85, 249 (1963). E. K~NIC, Theoret. Chim. Acta 1,23 (1962). J. L. HOARD and H. H. NORDSIECK, J. Am. Chem. Sot. 61,2853 (1939). H. STAFCMREICH and 0. SALA, 2. EZektrochem. 64, 741 (1960). ibid. 65, 149 (1961). A. HILDAGO and J-P. MAF~HIEU, Coopt. Rend. 249, 233 (1959). 1087

S. F. A. KETTLE and R. V. PARISH

1088

in that of an aqueous solution [9]. Neither of these observations appears compatible with either D,, or D,, sy mmetry (see Table 2 for a list of infrared and Raman active bands for a variety of possible symmetries). We have, therefore, re-examined the infra-red spectra of K,Mo(CN),.2H,O both in the solid and in aqueous solution and of H,Mo(CN),.6H,O in aqueous and ethanolic solution. EXPERIMENTAL K,Mo(CN),.2H,O was prepared by the method of FUXMAN and MILLER [lo] and purified by conversion to the cadmium salt and by fractional precipitation. was formed by the action of dilute hydrochloric acid on the H,Mo(CN),.GH,O silver salt and was precipitated by bubbling hydrogen chloride into the solution at 0”. Infra-red spectra of solids were obtained from Nujol mulls or KBr discs, very careful grinding being necessary to achieve maximum resolution. Solution spectra were obtained using either AgCl or As,& windows. Spectra were recorded either on a Unicam SPlOO or a Perkin-Elmer 125 spectrometer, the latter instrument being capable of resolution of O-5 cm-l. RESULTS AND DISCUSSION Solid K,Mo(CN),.2H,O The spectrum of the solid potassium salt agrees well with that reported by BONINOand FABBRI [9] with the exception that we do not find bands at 1113 cm-l and 837 cm-l, although they were occasionally observed for impure samples. The results for the C=N stretching region are shown in Fig. 1 and Table 1. Table 1 also

Fig. 1. Spectrum of K,Mo(CN),.ZH,O (Nujol mull). [g] G. B. BONINO and G. FAEBRI, Rend. Accud. Nuzl. Lincei 20, 566 (1956). H. FURMAN and C. 0. MILLER, Inorg. Synth. 3, 160 (1960).

[IO] N.

The structure of the octacyanomolybdate(IV) Table 1. Infra-rsd spectra in the CzN Llll

[81

Reference

2128 vs Solid K,Mo(CN),.2H,O

2119 8

2103 ~8

2096 s

2080 VW

PI

this work

2129 8 2127 sh

2135 212lV8sh

2136 8 BI 2123 2114 VW 2112 VW. sh ) 2106m ’ 2103 s 1 2086 VW 2081 VW1

v USN

2077 VW 2074 VW1

Y CN15

2062 VW; sh 2060 w 1

v CPN

2103 vs

2112

Raman spectrum 171: 2135 s, pal.; 2.

Assignment

2126 s

2080 w

in solution

Table

stretching region (cm-l)

2080 sh

2062 VW Mo(CN),4-

1089

ion

2111 bv ca 2070 vb, sh

1’ CN

? vCN

vCN v PN,

Y CNL6

2121 8; 2114 b, ah.

Correlation of C=N stretching modes for various configurations Dodeoahedron C*

DZd

A’ T, i--A; 9 r. iA’ F, i,

A”r,i’

I‘---B, r, i Er,iLT1,l

Cube 0,

Square antiprism D,

D 46

A,,r--A,r-A,r A,i-B,i .L Er,i-Elr,i

inA’r,i-AA,r-A -B,r au act A’r,i-BB,r,i E r, 1-lT,,rLi:fizEsr A’ r, i A”r,i

E,r

7

Raman active I.R. aotive Coincidence I =

6 (4) 6 (3) 6 (0)

Raman active;

6 (3) 4 (2) 4 (0)

i = infra-red active;

2 1 0

5 (3) 3 (2) 2 (0)

4 (3) 2 (2) 1 (0)

inact = inactive;

includes the results of previous investigations [8, 9, 111. The activities and symmetries of the CzN stretching modes expected for various possible configurations are shown in Table 2. Four strong bands are seen, together with two moderately strong bands (2106 and 2127 cm-l) and several weak ones. Table 2 shows that D,, or D, symmetry require two or three bands respectively, and these configurations may be ruled out. A satisfactory interpretation may be given in terms of the observed [6] crystal structure. The (idealized) D,, configuration has four infra-red active C%N stretching modes, to which the strongest four bands may be assigned. If the symmetry is lowered to C,, as observed, the two E modes will split to give two bands of comparable, but probably not identical, intensity. The A, modes will become weakly active, so that we anticipate a total of four additional bands, two fairly strong and two rather weak. The two stronger bands are seen at 2106 and 2127 cm-l. The other two may be obscured (e.g. the low frequency tail of the 2103 cm-l band is [ 1 l] E. G. BRAME, F. A. JOHNSON,E. M. LARSENand V. W. MELOCHE,J. Ilzorg. Nuclear C%m. 6, 99 (1955).

8. F. A. KETTLE and R. V. PARISH

1090

broad and unresolved), they may be too weak to be seen or they may possibly be appearing at 2112 and 2114 cm-l. No such correlation of the spectrum can be given in terms of D,, symmetry being lowered to C,. The weak peaks at and below 2086 cm-l are attributable to the presence of isotopes [12]. The Cl3 peaks should have approximately 9% of the intensity of the major bands and the N15 peaks about 3% and should therefore be discernible. The isotope peaks corresponding to the higher frequency major bands, expected at

2050

200

2150

Fig. 2. Spectrum of Mo(C?J)~~- in solution.

about 2095 and 2100 cm-l, are not resolved from the band at 2103 cm-l, although slight irregularities in the band envelope are detectable. Mo(CN)s4- in solution In marked contrast to the well resolved spectrum of the solid, the solution gives only a single broad band, about 40 cm-l wide at half height (Fig. 2). Many attempts were made to resolve this spectrum, including varying the solvent (water, water-ethanol, pure ethanol (for the acid), water-glycerol, saturated aqueous KBr) and freezing or cooling the solutions to -6O”, all without success. The possibility that this spectrum may be the result of superposition of several peaks (from whatever cause) would seem to be ruled out by the simplicity of the Raman spectrum. Although the Raman spectrum is consistent with Dddsymmetry, the visible spectra of the solution and the solid are very similar [5], suggesting that the configuration is the same in both media and presumably DBd(Cs). Neither the Reman nor the infra-red spectrum is compatible with Dzd(Cs) symmetry unless there exists some [lz]

L. H. JONES, J. Phys. Chem. 36, 375, 1209 (1962).

The structure of the octacyanomolybdate (1V)ion

1091

systematic reason for peaks in the infra-red to be weak in the Raman and vice-versa. A possible reason for such behaviour is suggested below, although it should be noted that cyano-complexes [l3] and other complexes involving metal-ligand m-bonding [la] have been reported to give anomalous intensity patterns in Raman spectra. The argument given below is confined to consideration of C%N stretching modes but is perfectly general and would apply equally to the other vibrational modes. (b)

.

Fig. 3. Distortion of a cube to (a) a dodecahedron and (b) a square antiprism.

All the possible structures are distorted versions of the cube (Fig. 3). The correlation of the C=N stretching modes with those of the cube is shown in Table 2. For D,, symmetry, it is seen that B, and E modes derive both from T,, (infra-red active) and T,, (Raman active) modes. If the distortion from the cube is not too great then it seems reasonable to suppose that one set (B, + E) will be predominantly infra-red active and the other set predominantly Raman active. That this will be so may be seen as follows. The dodecahedral arrangement of eight cyanide groups may be split into two groups of four, corresponding to the positive and negative tetrahedra of the cube. Each of these groups transform independently under the operations of the point group, so that the symmetries of the vibrations of one group are duplicated by those of the other group. The vibration of the whole ion may be approximated by taking combinations of the corresponding vibrations of the two groups, i.e. Composite vibration

= sin 8 (first group vibration) & cos 8 (second group vibration),

where 8 = 45’ for a cube and (0 - 45)’ is a measure of the distortion. If the extension of the bonds are labelled as in Fig. 4, the approximate, un-normalized 1131 G. W. CHANTRY and R. A. PLANE, J. Chem. Phys. 85, 1027 (1961). [14]L. A. WOODWARD and M.J. WARE, Spectrochim. Acta 19,775 (1963).

S. F. A. KETTLE

1092

and R. V. PARISH

normal vibrations for 13N 4.P are: A,(+)

= (A + B + C + D) + (P + & + R + 8)

A,(-_)=(A+B+C+D)-((P+Q+R+S)

B,(+)=(A+B-C--_)+(I'+&-E-_-S) B,(-)=(A + B -C - II)-(P +Q -8 E(t)

=

(A - B) + (R - 8)

(C - D) + (P - Q)

=

E(-)

-8)

(A -

B) -

(R 4)

(C -

m-

(P - &)

B,( +) will be strongly infra-red active since the changes in dipole moment for the two groups are in phase, while B,( -) will have only a smal1 resultant change in dipole moment (of magnitude dependent on the value of 0 - 45’) and will be

R Fig. 4

only weakly active. Similarly, E(-) will be strongly infra-red active and E(+) only weakly active. A similar argument may be applied to the Raman activity. The intensity of a Raman band depends on the changes in the components of the derived polarisabilities, which transform as products of the co-ordinate axes. If these components tend to cancel, the intensity of the band will be lowered. The net result is that, to a first approximation, each mode will have the same activity as the corresponding mode of the cube, providing the distortion is not too great. A precisely similar argument may be carried through starting from D, symmetry (the skewed cube), with the same result. The numbers of active bands expected on this basis are shown by the figures in parentheses in Table 2. In every case (except C,) two infra-red and three Raman bands are expected and it will not be possible to deduce the structure from the spectra. Since the infra-red active vibrations are degenerate in the cube, they would be

The structure of the octacyanomolybdate(IV)

ion

1093

expected to be close together for the distorted cube and may not be resolvable (although it is not clear why the bands should be so broad). The Raman active bands arise from two modes of the cube, consistent with the observed spectrum. If the symmetry is further reduced to C,, as has been suggested on the basis of the visible spectrum [4], the TzQ Raman active mode should split into three but it is doubtful whether the corresponding bands would be resolved. (The 2114 cm-l band appears only as a broad shoulder on the 2121 cm-l band). The number of infra-red active bands would rise to three. It would seem, then, that vibrational spectroscopy cannot be used to determine uniquely the structure of the Mo(CN),~- ion in solution. There is, however, nothing to suggest that the structure differs greatly from the dodecahedral arrangement found in the crystal. Acknowledgements-The authors wish to thank Professor N. N. GREENWOODfor his interest and encouragement and the General Electric Company, Schenectady, N.Y., for generous financial assistance (R. V. P.)