948
J . Phys. Chem. 1986, 90, 948-952
-
the C M cross sections for heads and tails scattering might be LAB nearly equal, but the Jacobian factor ( u / v ) 2in the C M transformation would favor a higher intensity in the lab for the slower backward scattered flux. Measurements on the product velocity distributions as functions of orientation may help to differentiate among these possibilities.
Acknowledgment. We gratefully acknowledge financial support of these experiments by the National Science Foundation and by
the Robert A. Welch Foundation, and acknowledge a grant from the US-New Zealand Cooperative Science Program of the N S F for travel for P.W.H. P.R.B. thanks the Alexander von Humboldt Foundation for a Senior U S . Scientist Award and Prof. J. P. Toennies and the members of the Max-Planck Institut fur Stromungsforschung in Gottingen for their hospitality during the writing of this manuscript. Registry No. K, 7440-09-7; CF,Br, 75-63-8; KBr, 7758-02-3.
CONDENSED PHASES AND MACROMOLECULES Properties of the Vanadlum Pentoxide Hydrogen Bronzes (H,, V,O,) D. Tinet, M. H. Legay, L. Gatineau, and J. J. Fripiat* Centre National de la Recherche Scientifique. Centre de Recherche sur les Solides d Organisation Cristalline Imparfaite, I B , rue de la Fgrollerie, F-45071 Orleans Cedex 2, France (Received: March 13, 1985)
The electrochemical potential of an electrode containing the bronze H2+V,O5 at equilibrium state has been measured for 0 < x < 1.7 in order to characterize the intermediate phases. 5'V NMR, 'HNMR, and EPR studies have shown that interaction between different cations (V5', V4', V3') plays an important role in the stability of these phases. The V3' ion seems to be stabilized in a 3d14slelectronic configuration,
Introduction A hydrogen bronze is an insertion compound of hydrogen within a host lattice, either an oxide or a chalcogenide, in which no formal chemical bond exists between the inserted proton and the anion. The vanadium pentoxide hydrogen bronze is formed at about 65 OC from a V2O5 microcrystalline powder, coated with small Pt particles, in contact with gaseous molecular hydrogen, through hydrogen spillover. The final hydrogen content x is between 1.5 and 1.9 according to the residual pressure. This hydrogen bronze HhV2OS has been extensively It has been shown that the proton is localized and forms a hydride bond with the vanadium. The electrons are also localized, creating V4' and V3' paramagnetic centers. The exchange interactions between these two paramagnetic species and the small particle size are the origin of a transition between a frustrated superparamagnetic state ( x < 1.65) and an unfrustrated superparamagnetic state (x > 1.65). Because of the amorphous structure of this bronze, the characterization of the final or eventual intermediate phases by X-ray crystallography or neutron diffraction was impossible. This paper aims to characterize intermediate phases in the range 0 C x < 1.5 by electrochemical methods and nuclear magnetic resonance (NMR) and electron paramagnetic resonance (EPR) techniques. At lower hydrogen content x N 0.25 a phase has been recently de~cribed.~ Electrochemical Results The electrochemical potential of the bronze H,V,05 has been measured for 0 < x < 1.7. The electrochemical cell contains a solid electrolyte, which is a Nafion membrane. Nafion stands for perfluorosulfonic acid. It is the only solid proton conductor (1) D. Tinet and J. J. Fripiat, Reu. Chim.Miner., 19, 612 (1982). (2) G. C . Bond, P. A. Sermon,and C. J. Wright, Mater. Res. Bull., 19,
701 (1984). ( 3 ) P. G. Dickens, A. M. Chippindale, S . J. Hibble, and P. Lancaster, Mater. Res. Bull., 19, 319 (1984).
0022-3654/86/2090-0948$0l.50/0
working at room temperature. The electrodes are sheets made of Teflon and graphite mixed with the active materials. The reference electrode is the bronze Hl,6M003,formed in acidic solution by nascent hydrogen, whose potential is about +0.1 V vs. NHE! The counter electrode is the Moo3oxide. The working electrode is the bronze HzxV2O5, formed by hydrogen spillover and transferred into the cell under nitrogen atmosphere. The oxidation of H2*V,O5 is carried out stepwise. Each step begins with a galvanostatic discharge followed by relaxation toward the equilibrium state. At each step the value of x is determined by coulometry. The electrochemical potential of the bronze HhV205 is plotted vs. x in Figure 1. According to the Gibbs phase rule, this diagram can be divided into several domains indicated by vertical dashed lines: Because the potential is invariant with respect to x, the first domain represents the coexistence of two solid phases, HX,oV2O5 and H O , ~ V , O and ~ H2(g). In the second domain, there is a single solid phase H,,V,O5, with 0.25 < x < 0.5, and H2(g). The slight increase of the electrochemical potential may be due to a modification of the lattice potential. In the third, the solid phases H,,,V205 and H1.5V205 coexist with H2(g). The fourth domain is characterized by the existence of a single solid phase H2+V,O5 with 0.65 < x < 1.1 and H2(g), the potential decreasing as x increases. Finally, the fifth domain expresses the coexistence of two solid phases which are H2,2V?O5and H3V205 in equilibrium with H2(g). The reversibility of the system has been tested. The bronze H2*V,O5 with the highest value of x can be formed by hydrogen spillover from H,,,V205, obtained by electrochemical oxidation of H3.4V205. All the hydrogen vanadium bronze phases are better oxidants than the reference H1,6M003electrode. This has also been shown for the C O ~ x i d a t i o n . ~ (4) R. Schollhorn, R. Kuhlmann, and J. 0.Besenhard, Mater. Res. Bull., 11, 83 (1976).
( 5 ) J. P. Marcq, W. Wispenninckx, G.Poncelet, D. Keravis, and J. J. Fripiat, J . Caral., 73, 309 (1982).
0 1986 American Chemical Society
The Journal of Physical Chemistry, Vol. 90, No. 5, 1986 949
Vanadium Pentoxide Hydrogen Bronzes
I
E, Volt
Figure 1. Electrochemical potential of the bronze HZrV205vs. hydrogen content x . The different domains and phases discussed in the text are indicated in this figure.
With the help of N M R and EPR techniques, we shall now describe the properties of the domains delineated electrochemically.
51VNMR Resonance The N M R experiments have been performed using a Bruker pulse spectrometer at 2.1 T. We have used five different samples, HO.sV205, H 1 V d k H1.4V205, Hl,85V205, and H2.3V2059 made by hydrogen spillover using a limited amount of hydrogen in order to get the desired stoichiometry. In spite of its large quadrupole coupling constant (QCC) and thanks to its high relative sensitivity (0.382 with respect to IH), - 1 / 2 central line of the N M R resonance spectrum of the 51V( I = 7/2) is easily observed in V2O5 at 23.66 MHz in a 2.1-T magnetic field. The full width at half-maximum (fwhm) is the order of 0.4-0.5 ppm, in agreement with the value reported by Oldfield6 for a sodium vanadate at 3.52 T. At such low fields the second-order quadrupolar interaction shifts the central line. In HhV2O5, the '/* -1/2 transition is observable in the range 0 < x < 1.5, though, as shown in Figure 2, its intensity decreases as x increases. If each hydrogen atom was transferring one electron to V5+, transforming the V5+ species into paramagnetic V4+ cations, and if V4+ paramagnetic centers were randomly distributed, the 51VN M R spectrum would become unobservable at a much lower value of x . As shown in Figure 3, this is not the case since the reduction of the central line integrated area for x = 1 reaches about 65% the initial integrated intensity observed for x = 0. For x < 0.5 the amount of the 51Vsignal escaping detection (y) increases proportionally with x. Moreover, the line is not displaced or broadened as x goes from 0 to 0.5. For x = 1.5 the 51VN M R signal becomes undetectable. Such a behavior suggests (i) that the 51V nuclei which are observed are located in domains which are not influenced by the paramagnetic species or which experience minimal change in the electric field gradient with addition of H and (ii) that V3+ species are formed before the reduction of V5+ into V4+ is completed. Indeed if it was not the case, the 51Vsignal would disappear for x = 1 whereas it is observable until x < 1.5. In order to explain these observations and in particular to fit the experimental line y =f(x) shown in Figure 3, the following model can be suggested. Let p ( x ) be the electron density per vanadium site, p ( x ) being taken as zero for x = 0, which means that the electron density in V5+ is taken as reference. Additional electron density results from the transfer from the inserted hydrogen moieties to V5+. Thus, we write that the fraction y of 51Vnot observed by N M R as
-
Figure 2. 51V NMR spectra (23.8 MHz) in the bronze HZxVZO5 for different proton contents. Numbers on the right side are normalization
factors.
*Experiment
-
Y=
XP(X)
(1)
(6) F. Oldfield, R. A. Kinsey, B. Montez, T. Ray, and K. A. Smith, J . Chem. SOC.,Chem. Commun., 254 (1982).
e
Theory
e
& e
&
*@
e
*
e
e
e e
Figure 3. Concentration of 51V not observed by NMR vs. hydrogen
content. Then we consider that the V5+ nuclei experience two different states called state I, for x < xo,and state I1 for x > xo,respectively. Thus, p(x) = pa + pb, pa = a, and b
Pb
p(x)
= 1
=a
+ exp[B(x - xo)l
b + 1 + exp[B(x - xo)]
(2)
With a = 0.65, b = 0.40, B = 15, and xo = 0.5,the fit between theoretical and observed y is excellent, as shown in Figure 3. y
The Journal of Physical Chemistry, Vol. 90, No. 5, 1986
950
TABLE I: Full Width at Half-Maximum (AH, C) of the 'H NMR Line at Two Frequencies at Room Temperature X
w,
MHz 40 90
0.5
1
1.4
1.85
2.3
2.7
21.5 17
21.5 21
22.5 21
23.5 25
21.5 23
25 25
reaches 1 for x = 1S6, and the pseudoplateau observed between x = 0.5 and 0.7 is well fitted. This is not surprising since the step observed in that region is imposed by the function (1 exp[B(x - xo)l)-'. In state I (x < xo), p(x) = 1.05 and thus y E x. This means that each hydrogen atom inserted provokes the transformation of one V5+ into one species which is not observable by NMR. In state I1 (x > x0), p(x) = 0.65 and thus y = 0 . 6 5 ~ .This means that the insertion of three hydrogen atoms is responsible for the transformation of about two V5+ into species not observable by NMR. Thus p ( x ) , which has the physical meaning of a density of electrons transferred from the inserted hydrogen to vanadium, is predictable by the stoichiometry for x < xo whereas for x > xo it shows that V3+is formed before x 3 1. The good fit between y observed and y calculated according to eq 1 and 2 must be considered only as supporting the fact that it is only for x 3 1.5 that the V5+signal vanished. In addition, states I and I1 would correspond to two structures where V5+ layers (or domains) are mixed either with V4+layers (or domains) or with V4++ V3+layers (or domains). States I and I1 exist for x < 0.5 and x > 0.7, respectively. The transition occurring in this small range of x is indeed observed in the variation of the electrochemical potential shown in Figure 1. This will be discussed later, but the nice agreement between these two independent observations and the physical meaning of p1 and pII suggest that eq 1 and 2 represent a real aspect of the physical processes occurring when V20s is loaded with hydrogen in the domain 0 < x < 1.5.
Tinet et al.
Inu
1.5
A 90 MHz 0 40 M H Z
+
Proton Resonance Since electrons are transferred from the inserted hydrogen to vanadium, it is most probable that the protons are close to the paramagnetic centers V4+ and/or V3+, V4+ and that the main relaxation mechanism will be through spin diffusion. The ' H signal is expected to be strongly broadened by the electronic-nuclear spin interaction. The magnetic field due to the paramagnetic centers contributes strongly to the local magnetic field, the homodipolar interaction being much weaker. The net result of the two contributions may displace the resonance frequency of some of the 'H nuclei to such an extent that they cannot be observed in the rather narrow frequency window revealed by the Fourier transform of the decay following the 7r/2 pulse. The magnitude of the local magnetic field corresponding to this frequency window is about 23 G (or 98 kHz) f 10% as obtained from the fwhm of the broad and structureless 'H spectra observed at 40 and 90 MHz. Indeed the fwhm of these spectra appears to be independent of x and of the resonance frequency as shown in Table I. However, the integrated intensity of the ' H N M R signal is a function of resonance frequency and of x . Figure 4 shows that the amount of protons detected is higher the lower the field and that its variation with respect to x may be fitted nicely at vo = 90 MHz by eq 3 and at u = 40 M H z by eq 4 0.20 1 + exp[l5(x - l.OS)]
)
(3)
X
0.5
1
1.5
I)
2
Figure 4. Concentration of proton not observed by 'H NMR vs. proton content at two frequencies (40 and 90 MHz).
magnetic center vector. With respect to the resonance field (Ho) and frequency (a),the shift is
where the magnetic moment of the paramagnetic center is
(ji) = P 2 a / 3 k T In HZxV2O5 and in the domain where Vs+ has been transformed into V4+and V3+,any proton is submitted to a local magnetic field which is the sum of the contributions of the surrounding paramagnetic centers, the homodipolar interaction being neglected. Consequently, r-3 must be replaced by Ci#jri;3. Assuming in the first approximation that the shortest value of r is ro N 1.8 A,' that the V-V distance remains of the order of 3 A, and that 7 is perpendicular to the V-V chain, the summation
where hugis the Bohr magneton and nB the effective magneton number, which is theoretically 1.75 and 2.85 for V4+ and V3+, respectively. In a 2.1 -T field (uo = 90 MHz) (CY)= 12 G for V4+ species and 30.8 G for V3+ species, whereas in a 0.93-T field (vo = 40 MHz) these values become 5.3 and 13.6 G, respectively. Equation 7 has been obtained by carrying out the summation till the fourth V neighbor. For a powdered material without any preferential orientation of the crystallites in the magnetic field, the line shape could be described by a function'
where u = 3 cos2 6 , and the number of protons seen in a window should be n, = f(AH) dAH =
CY
dH
(9)
where
X = AH/(a) 0.40
I
where n, is the concentration of protons not observed by 'H NMR. In a paramagnetic material, the angular dependence of the paramagnetic shift of the resonance line of a proton linked to a paramagnetic center is a function oC3 cos26 - 1, where 6 is the angle between the magnetic field H arid 7 is the proton para-
If we express eq 9 with respect to this reduced parameter, the limits between which X has a physical meaning are -2 and + 1. It can be checked easily that the normalized values of n, given by integrating equation between these limits are (7) A. Abragam, 'The Principle of Nuclear Magnetism", Clarendon Press, Oxford, 196 1.
Vanadium Pentoxide Hydrogen Bronzes
However, if because of instrumental constraints the integration limits are -Xeff/2 and +Xeff/2, whereX,ff = AHen/(a)and AHe* is the experimental ‘H line width (-23 G), then
The Journal of Physical Chemistry, Vol. 90, No. 5, 1986 951 TABLE II: Curie Constant (C, Arbitrary Units) and Weiss Temperature (0) Determined from the Slope of the Plot of Integrated Intensity of the EPR Signal vs. T-I, and from the Intercept, Respectively broad line X
0.6 0.8, 1 1, 2
narrow line
C 5.6 3.6
4K
C
0, K
0 0
2.1 1
-100
2.4
0
5
-20
-30
In the intermediate situation where 2 < xepp < 4, the integration limits become -Xeff/2 and 1. According to Figure 4 it is clear that for x < 1.05 and yo = 40 M H z it can be assumed in first approximation that n,* = n, or that the average fraction of protons which are not seen (nu = n, - ns*) is negligible. Under these conditions -Xcff/2 -2 or +Xe, 4 and
( a ) = AHeff/4 = 5.75 G This result is in good agreement with that obtained for ( a ) assuming that the protons are submitted to the influence of V4+ species only. This conclusion is apparently in contradiction with that reported earlier on the composition of the transformed vanadium domains for composition H2xV205 with x = 1. For x > 1.OS,and from Figure 4, at 40 MHz, nu = 0.46 or n, = 0.54. Then Xenobtained from eq 10 is 1.65, which means that ( a ) = 13.9 G. This result is again in agreement with the value of ( a ) calculated for the magnetic field produced at the proton site by V3+ species. This means that, for x > 1.05, the magnetic field limits within which the protons are observed are imposed by the paramagnetic centers with an effective magneton number similar to that assigned to V3+ species. Thus, in a field of 0.93 T (corresponding to vo = H0/(2ry) = 40 MHz) the calculated width of the N M R spectra coincides with the physical limits corresponding to the local field imposed either by V4+ species (x < 1.05) or by V3+ species (x > 1.05). It may be anticipated that in a stronger field, for instance 2.1 T (90-MHz ‘H resonance frequency), ns* (and nu) will be different. Indeed for x < 1.05, the value of ( a ) compatible with AHeff = 23 G is 21 G whereas it is 5 1.4 G when x > 1.OS. Since the natural line width AH is 3 times a, nu becomes an important fraction of x as shown in Figure 4. Indeed for x < 1.05, nu N 0.67 whereas for x 7 1.05, nu N 0.87. In conclusion, from the ‘H N M R line width a_nd from the variation of nu with x and with the applied field IM, three important facts are observed. First, because of the instrumental limitation of the exploration field (or frequency) of the resonance line, the 40-MHz frequency is the upper limit which allows the proton, considered as a spy of the paramagnetic environment, to detect the physical nature of the paramagnetic centers responsible for the highest local magnetic field, namely V4+for x < 1 .OS and V3+ for x > 1.05. Second, since the average local field values ( a )obtained from AHeffat 90 and 40 MHz are not within the ratio of the magnetic fields (2.25) but within ratios 51.4113.9 = 3.7 (for x > 1.05) or 21.015.75 = 3.6 (for x < 1.05), there must be an additional effect which increases the experimental nu observed at 40 M H z with respect to that expected at 90 MHz. This is perhaps attributable to a preferential orientation of the vanadium bronze particles with respect to the field. Indeed such a preferential orientation has been reported for H3,,V205.’ This preferential orientation would split the IH single line into two broad components on each side of the theoretical resonance frequency. This would deplete the intensity of the resonance line arbitrarily centered on this frequency. Third, at 40 MHz, there is a good agreement between the calculated effective magneton number of the paramagnetic centers and the theoretical number for V4+ (x < 1.05) or V3+ (x > 1.05), respectively. This seems to disagree with the conclusion of the 51V N M R study which suggests a contribution of V3+ species becoming important as soon as x > 0.5.
Figure 5. EPR signal of HzxV205 for x = 0.6 and 1.4 a t room temperature.
EPR Data The region 0.5 < x < 1.05 coincides approximately with the domain 0.65 < x < 1.15 where the electrochemical measurements indicate the existence of a defined phase whereas on both sides of the limits, namely for x < 0.5 and x > 1.OS,the electrochemical behavior indicates the presence of mixtures of phases. It may appear paradoxical that it is precisely in the domain of existence of a defined phase that the interpretations of the ‘H and j1V magnetic resonance studies are in conflict. This conflict deals directly with the identification of the paramagnetic centers and indirectly with the extent of the electron transfer from the inserted hydrogen to vanadium. The EPR spectra recorded at room temperature contain a broad (AHpp 500 G) and a narrow line (AHpp 90 G) centered at g = 1.96. This value of g is that characteristic of V4f.8 The Curie constant and the Weiss temperature associated with these lines are shown in Table 11. The integrated intensities of both lines increase rapidly with increasing x till x N 0.5. They decrease in the domain 0.5 < x < 1 and they fall sharply for 1 < x < 1.5. Under the experimental conditions of this work (room temperature) the sole detectable paramagnetic species is V4+,the V3+ signal being not observable.* The broad line may be assigned to V4+ in dipolar interaction with other V4+. Indeed for two V4+ which are 3 A apart, the calculated line width is about 450 G. The narrow line could indicate an exchange (or superexchange) between V4+ centers. It is present in the all domains 0.6 < x < 1, but its Curie constant increases by a factor of -50 in going from x = 1 to 1.2, though its intensity relative to the broad component decreases (Figure 5).
-
-
Model and Discussion The electrochemical data have shown the existence of two defined phases (see Figure 1): the first monophasic domain (phase I) is observed between 0.65 < x < 1.15, and the second one is observed for x 2 1.5 (phase 11). For x < 0.65, the sample is made from a mixture of V205and phase I. For 1.15 < x < 1.5, the sample is composed from a mixture of phases I and 11. (8) A Abragam and B. Bleaney, “Electron Paramagnetic Resonance of Transition Ions”; Oxford University Press, London, 1970. (9) J. Wong, F. W. Lytle, R. P. Messmer, and D. H. Maylotte, Phys. Rev. B. Condens. Matter, 30, 5596 (1984).
952
The Journal of Physical Chemistry, Vol. 90, No. 5, 1986
Tinet et al.
A .U
3d0 3d,, 3d,4s1
3d, ,3d,4s, H2V205
3do
Figure 6. Suggested structure of the bronze at the lower and the upper limit of phase I domain. 3d0 corresponds to a layer where all the vanadium ions are Vs+, and 3d1,3d14slto a layer where all the vanadium ions are transformed either in V4+ (3dl) or in V3+ (3d14s,). B
A
3d
4s
3d
eV i2
Figure 7. Schematic bonding between two V3+ (A and B) with the 3d14s,
configuration: top part, before bonding; below, after bonding. The electronic 4s magnetic moment is canceled, and the 3d magnetic moment fluctuates and averages the local field. The growth of phase I occurs through the formation of domains in which all preexisting V5+ are transformed into V(5-6)+,6 being >1 as suggested by the variation of the 5'V N M R signal with respect to x . The term domain may be understood as a region of superimposed layers within the same particle or as entire layers within one particle. For instance, it is conceivable that layers with V(5-6)+cations are sandwiched between layers with V5+ (Figure 6). From the IH NMR resonance in phase I, the local field observed by the proton may be accounted for by vanadium cation with a 3dl electronic structure. Indeed the effective number of magnetons of the paramagnetic center is close to 1.75, namely that of V4+. The main question then concerns the contribution of the excess of electron transferred from the proton to the vanadium since 6 > 1. The EPR narrow band has been interpreted by assuming an electron exchange between two paramagnetic sites. It may be suggested that the excess electron (6 - l ) , instead of forming vanadium with a 3d2 configuration (namely V3+), occupies the 4s orbital and that two neighboring vanadiums with configuration 3d,4sl form a pair with configuration 3d,4s2, 3dl. This pairing of the 4s orbitals cancels their magnetic moment in such a way that those two V3+ have the same magnetic moment as two 3dl ions, namely two V4+ions. The overlapping of the 4s orbitals is at the origin of an exchange between two vanadium cations and of the narrow EPR signal (Figure 7 ) . In phase I, the characteristic electronic configuration would be such as the number of V(3dl) is equal to the number of V(3d14s,) (see relationship 2 ) . As x increases, the number of transformed layers increases until x = 1, where about one out of every three layers is composed from V5+(see Figure 6). Beyond that limit the system becomes unstable and it demixes into phases I and 11. In phase I1 (x > 1S ) , the V5+ domains have disappeared and they are replaced by a mixture in unequal proportions of V(3d,) and V(3d14sl). Pairing the V(3d14sl)centers is no longer achieved. From the magnetic point of view the local magnetic field of
5460
5468
54irS
5484
b
5492
Figure 8. XANES results for different vanadium oxides and a bronze H,,V,OS with x = 1.6.
V(3d14sl)at the proton site is analogous to that of V(3d2) (V3+) because of the blocking of the orbital moment (Hund's rule for cations with an incomplete shell). In that phase, the reason why the V(3d14sl)species is favored with respect to V(3d2) comes for the observation of the location of the K edge in XANES experiments carried out with the synchrotron radiation at LURE. Indeed as shown in Figure 8 the K edge for H,,V,O5 (x = 1.6) is observed at 2.6 eV with respect to the edge of V203. This is within 0.1 eV that observed for V204 whereas the K edge of V2O5 is at 4.2 eV. Thus in phase 11, the transition of the 1s electron toward 4p levels would not be perturbed by the presence of one electron on the 4s orbital. The preedge peak gives a different picture. Indeed the 1s-3d transition in H,,2V,O5 has its maximum near the position of the shoulder observed for V203. The preedge peaks in V,04 and V205 are displaced toward higher energy. The XANES study should and will be extended. In the present state the only safe conclusion is that for x > 1.05 in H2,V,O5, the 3d2level of the V cation does not behave as would be expected.8 Thus, the hypothesis on the electronic nature of the so-called V3+ centers, namely that the electronic distribution is 3d14s,, is not contradicted by the XANES results. Conclusion
The HzXV2OS(0 < x < 1.9) bronze shows two monophasic domains, 0.5 < x < 1 and x > 1.5. The first phase is a mixture of the original V5+domain and of a domain characterized by the presence of two vanadium species: a V4+ion with its normal 3d, electronic configuration and a pair of V3+ ions with the configuration 3d,4sl. To our knowledge, this last species has never been previously observed. In the second phase the V5+ species have disappeared and are replaced by a mixture of V4+ ions (in decreasing concentration with increasing x) and V3+ ions. This last species could still be 3d,4s, according to a preliminary XANES study.
Acknowledgment. We thank A. Le Mehaute for preparing the electrodes, P. Levitz and J. Petiau for helpful discussion, and the LURE staff. Registry No. 51V,7440-62-2.
