J. MATER. CHEM., 1994, 4( 11), 1749-1 753 1749 170Nuclear Magnetic Resonance Spectroscopy of the Structural Evolution of Vanadium Pentaoxide Gels G. A. Pozarnsky and A. V. McCormick* Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, MN 55455, USA Recent studies of the synthesis of V205gels by acidifying metavanadate salt solutions suggest that VO,' polymerizes into chains of vanadate octahedra. In this study we characterize the growth of vanadate polymers by both solution and magic angle spinning 170nuclear magnetic resonance (NMR) spectroscopy. The spectra are consistent with the formation of a chain polymer with a repeat unit of V02(0H)(OH,),. The 170 NMR spectra also suggest that the chains might connect to each other by hydrogen bonding.Although the chemistry of V205 gels has been of recent interest,'-" the reason that acidification of vanadate solutions produces a characteristic two-dimensional (20) ribbon micro- str~cture'.~,~,~remains unclear. Previous workers have pro- posed that a 2D fragment of V205 crystal is formed,*g9 but recent work using electron paramagnetic resonance (EPR) spectroscopy and 51V nuclear magnetic resonance (NMR) spectroscopy has instead suggested chain polymerization of V02+ .7 In this work, we investigated whether the 170NMR spectra are consistent with the chain growth mechanism by labelling the vanadate solution species and the intermediate polymers with 170 early in the synthesis. By using 170NMR chemical shift assignments from several studies of vanadate structures in solution,'2p20 we were also able to speculate as to whether hydrogen bonding occurs among the polymer chains in a way that might serve to build 2D structures recently observed by cryogenic transmission electron microscopy (cryo-TEM).""' Experimental Samples were prepared by dissolving sodium metavanadate in 1 g of 10 atom% I7O enriched water (Aldrich).A further 1 g of unenriched water was then added to yield a stable 1.0mol dmP3 sodium metavanadate solution (pH 8). By adding the 170enriched water first the extent of 170 enrichment into the vanadate species over the time period studied was increased. A column of Dowex 50W X-2 50-100 mesh ion-exchange beads was charged with hydrochloric acid, then washed with deionized, distilled water and finally used to acidify the solutions.NMR spectra were acquired using a GE 500 MHz NMR spectrometer. 51V NMR solution spectra were acquired at 131.487 MHz using a 90" pulse width of 12 ps, a spectral width of 60 kHz, a relaxation delay of 0.5 s and with 256 transients. 170NMR solution spectra were acquired at 67.8087 MHz using a 90" pulse width of 61 ps, a spectral width of 111 kHz, a relaxation delay of 0.1-0.2 s and with 10000 transients. 51V and I7O chemical shifts were calculated with reference to external samples of VOCl, and water, respectively. Magic angle spinning (MAS) spectra were acquired using a 5 mm Doty probe at a spinning speed of 10 kHz with Si,N, sample rotors.Prior to spectral acquisition, excess solution was removed from the gelled sample by filtration in order to maximize the proportion of 170enriched gel in the solid and -f Present address: Centre for Advanced Materials Processing (CAMP), Clarkson University, Potsdam NY, USA. to remove any dissolved species still present. Spect rometer parameters were the same as for solution spectra except that 1000 and 50000 transients were used for 51Vand I7O MAS NMR, respectively. Results and Discussion The 51V NMR spectrum of the sol immediately after ion exchange is shown in Fig. 1. Detailed discussion of the assign- ments can be found el~ewhere.~ The peaks at -420, -513 and -532 pprn correspond to the fast proton exchange limit of the di- and tri-protonated forms of the decavanadat e anion, and the peak at -545 ppm corresponds to V02 -.7 The remaining peaks at -523 and -537 ppm were assigned to the triprotonated form of the decavanadate anion.7,21,22 The progression of the 51V NMR spectra as gelation occurs has been presented and discussed el~ewhere.~ I7O NMR is used here only to identify the intermediate and final structures of the V205 gel.Owing to the low enrichment and unknown isotope exchange rate used, J kinetic analysis is not expected, and such an analysis has already been performed using 51V NMR and EPR.7 The 170 NMR spectrum immediately after ion exchange is shown in Fig. 2 and the progression of the "0 NMR spectra as gelation occurs is shown in Fig.3. The peaks at 80 ppm (V60); 348 ppm (V30); 705, 843 and 931 ppm (V,O); and 1199 and 1210 ppm (VO) have all been previously iissigned to the oxygen sites indicated in the decavanadate species at I I -280 -360 -440 -520 -600 -680 6 Fig. 1 'lV solution NMR spectrum of reacting solution, t =0 J. MATER. CHEM., 1994, VOL. 4 containing a high concentration of V02+, as demonstrated by the 51V spectrum (Fig. 4). The 170 NMR spectrum of the V60(decavanadate) V -0-V(polymer) \VO(decavanadat e) I I 1400 1000 600 200 6 Fig. 2 170 solution NMR spectrum of reacting solution, t =0 A V-0-V(polymer) 1400 1000 600 200 6 Fig. 3 "0 NMR spectrum of reacting solution; (a) t= 2 h, (b)t=4 h, (c) t=S h pH 2.12-" Each of these peaks represents both protonated and deprotonated forms of the decavanadate anion under- going rapid exchange.This fast exchange limit can depend on the solvent and solution c~ncentration.~~ Apparently at this high concentration some decavanadate anion can be formed that does not exchange quickly enough to cause peak coalesc- en~e.~Hence, the 170 NMR peak at 453 ppm matches a peak observed for an H3V100283- oxygen site undergoing slow exchange.14 It was assigned to the triprotonated decavanadate anion since this peak was present throughout gelation in a fashion mimicking the corresponding 51V NMR peaks at -523 and -537 ppm.7,21,22 The appearance of the triproton- ated form of the decavanadate anion suggested that the decomposition of the decavanadate anion into V02+ might involve protonation of the diprotonated decavanadate anion (decavanadic e.g., H+ 3-13Hf H,V,002,4--H,V,o028 -1OVO2++8H20 (1) In this study, the triprotonated decavanadate anion was present for somewhat longer periods of time (Fig. 3) than with the lower concentration sols studied previ~usly.~ Having accounted for all possible 170decavanadate peaks, we then associated the small 170peaks observed at 220, 580, 1050 and 1380 ppm with non-decavanadate species. The peak at 220 ppm was present throughout gelation.To assign this peak, we compared it with the spectrum of a vanadate solution V02+ solution in Fig. 4 showed the same peak at 220 ppm. Since, according to Howarth and coworkers, the V=O bond of V02+ is not observable in aqueous we assigned this peak to water molecules that are coordinated to the vanadium.This is a triply coordinated oxygen site (V-OH,), so it is reasonable that it should be near the peaks of the V,O groups of the decavanadate anion.12-15 This assignment is also consistent with the chemical shift of the coordinated water in V'"O( H2OI5,+,which appears at ca. 180 ppm in aqueous vanadyl sulfate solution^.'^ The 170NMR peaks at 580, 1050 and 1380ppm clearly disappeared from the solution spectra after the initial polymer growth OCCU~S.~ Previously reported 51V NMR solution spec- tra7 and the 170NMR spectra (Figs. 2 and 3) showed that no new mobile vanadate species were formed in solution during gelation; so it was deduced that these three 170 peaks were associated with the growing vanadate polymer whereas the corresponding 51Vpolymer peaks were not well re~olved.~ The three 170peaks were well resolved, perhaps because the anisotropy of the 170sites in the polymer was less severe than of the 5'V sites.The 580 ppm peak was near to the V,O peaks from decavanadates,12-15 and chemical shift correlations of V,O bond angles in aqueous salt solutions containing dimers and cyclic species show that it might correspond to a H3V10°28* I -400 -440 -480 -520 -560 V20(decavanadate) V30(decavanadate) V-OHOH, 700 600 500 400 300 201 6 Fig. 4 (a) 51VNMR and (b)170 NMR spectra of solution containingvo2 (H20)4+ J. MATER. CHEM., 1994, VOL. 4 V20 bridge at ca.160-180°.1s~'6~'8 I7O NMR studies of vanadate peroxy compounds and of the decavanadate anion suggested that the peak at 1050ppm corresponded to a terminal V-0 group.12p16 This peak probably shows the exchange-averaged shift for a V-OH site on the polymer undergoing the fast reaction;" I/ I/ fVIOH t fVIO-+H+ (2)/I /I The 1380 ppm peak may be associated with the V=O group on the vanadate polymer. Although the V=O site can not be observed for V02+,17,18 I7O NMR studies on VO(NO,), and VOCl, showed that short V=O bonds have a chemical shift of ca. 1400 ppm. The suggestion from previous 51V NMR kinetic studies7 that the polymer is built by chain polymerization of V02+,26 was supported by the appearance of the linear oxygen bridge at 580 ppm in the 170NMR solution spectra.Further evidence was obtained by examining the gel with MAS NMR. The 51V MAS NMR spectrum of the undried gel (Fig. 5) showed only one peak at -547 ppm. This was consistent with spectra reported earlier and corresponds to an octahedral environ- ment.7,27,28 All peaks observed in the corresponding I7OMAS NMR spectrum (Fig. 6) should be associated with this sole 51V environment. The I7OMAS NMR of the wet gel showed -400 -480 . -560 -640 -720' 6 1751 peaks at 223, 688, 1150, 1380, 1396, 1407 and 1430 ppm. The peak at 223 pprn was associated with coordinated water molecules as above. The peak at 688 ppm was near the chemical shift range of V20 environments for decavanad- ates.12-15 The peak at 1150 ppm was near that for terminal VO sites in decavanadates.12-16 The peaks at 1380, 1396, 1407 and 1430ppm were all consistent with V=O bonds17; the formation of multiple peaks (cf.the single solution I7C) NMR peak observed at 1380 ppm in Fig. 2) may result from different degrees of hydrogen bonding in the same way that 170 chemical shifts were affected by hydrogen bonding between water and C =0 gro~ps.~~-~' These peak assignments suggest the identity of the repeat unit of the polymer. Since microscopic observations showed linear polymer growth, the two bridging oxygens should be on opposite sides of the vanadium centre."," Moreovtbr, EPR infrared (IR) and Raman studies showed that one of rhe two water molecules was opposite the V=O bond.3,32-34 A possible structure is shown in Fig.7. There are no reports of optical isomerism, so the equatorially coordinated water and hydroxy groups can presumably switch positions on the vanadium centre. However, it was noted that while the expected ratio of these oxygen sites from Fig. 7 was Obr:0,:OH :OH, = 1: 1 : 1 :2, the ratio of the 170peak intensities was approxi- mately 1 :1:0.5 :1. This discrepancy might be due to the exchange of I7O isotope between the enriched -0112 sites and the less enriched bulk water. This process has been confirmed and studied in vanadyl sulfate solutions.25 Although there is a possibility of short-lived and or low concentration intermediates that are undetectable bj NMR occurring in the polymerization process, the similarity of the repeat unit in Fig.7 to the structure of V02+ suggests that the polymerization process might require no other intermedi- ates. The hydroxy group on the repeat unit could be the result of hydrolysis of a coordinated water on the just-polymerized VO2 .'+ Hydrogen bonding between these chains3' was suggested by the downfield shift of the V20 and VOH peaks of the polymer compared with the signals of these sites in stdution. Since no dissociation of the polymer formed was observed by either I7O or 51V NMR during gelation,' the mobt likely explanation for the disappearance of the solution "0 NMR peaks at 580 and 1050ppm was the broadening caused by continued polymer growth and hydrogen bonding.This might also account for the reappearance of these peaks (shifted to 688 and 1150 ppm) in the I7O MAS NMR. A hypothetical Fig. 5 51VMAS NMR spectrum of V,O, gel structure is shown schematically in Fig. 8 and is consistent with the electron diffraction pattern from cryo-TEM studies on the wet Note that since the 170 peak of the coordinated water shows no downfield shift, it might not interact with any hydroxy groups. Previous cryo-TEM observations have shown that the gel bu'kHvI ribbons are ca. 25 nm in width, so the transverse assembly of V-0-V v--2 1 linear polymers shown in Fig. 8 must cease at some point or \ else sheets would be produced. This cessation might be caused by the presence of negatively charged ligands in the equatorial positions of the repeat unit shown in Fig.7. The reaction mixture remains at a constant pH of 2, and since neither mobile vanadate anions nor counter-ions (e.g. Cl-) are present in the sols at this concentration, only the 1400 1200 1000 800 600 400 200 6 Fig. 6 I7O MAS NMR spectrum of V,O, gel Fig. 7 Repeat unit in vanadia polymer J. MATER. CHEM., 1994, VOL. 4 top view of ribbon planar view of ribbon 0 OH2 0 III I OH2 II HO-V--H-O-V-O-HH-O-V--H-0-V-OH I II It I OH2 0 0 OH2 1 2 3 4 Fig. 8 Wet ribbon structure: proposed assembly of hydrolysed linear polymers into ribbon structure. Individual chains are numbered for identification. linear vanadate polymers can provide such buffering capacity. These negative charges could only be the result of further deprotonation of the hydroxy group on the octahedral repeat unit. VO,(OH,),(OH) + VO,(OH,),O-+Hf (3) We might not expect to see the 0-site in the 170 NMR spectra because of its low concentration and fast exchange.At the concentration where the ribbons were observed with cryo-TEM (0.5 mol dmP3 vanadate), one of every 50 vanadia repeat units should be negatively charged to maintain the pH. If this change were responsible for halting the ribbon's trans- verse growth (by interfering with the hydrogen-bonding mech- anism), then using literature-cited bond and the hypothetical configuration shown in Fig. 8 we might expect the width of a 50 chain ribbon to be in good agreement with that observed with cryo-TEM (ca.25 nm)." The ribbon structure shown in Fig. 8 should be stable upon drying at room temperature since X-ray diffraction, IR and Raman studies have indicated that the short V=O bond and the coordinated water opposite it are retained even after drying.1,32-34,37 Th e same characteristic ribbon structure and electron diffraction pattern are also retained.5,6 1R and NMR studies have shown that the V-OH groups are largely retained upon drying, so apparently there are no condensation reactions between V-OH gro~ps.~~,~'-~'If the linear bridges between vanadia octahedra remain ~nchanged,~'-~~ the only change in structure we might expect with drying would be the loss of the equatorial coordinated water to allow associ- 320°C1-3H20 v2°5 Fig.9 V,O, unit in dried gel ation with the oxygen of the V-0-V bridge of an adjacent chain. This would suggest the structure shown schemati- cally in Fig. 9 with the formula, V,05 -3H,O. Although we note that this indicates somewhat more water than the usual hydrated gel formula of V,05*1.6H,0obtained by thermogravimetry.1,33 Conclusions The I7O NMR spectra obtained was consistent with two earlier points suggested by 51VNMR: (i) that VO,' forms a chain polymer; and (ii) that this polymer can undergo hydroly- sis to form V-OH sites.7 Evidence from "0 NMR also suggested that the hydrolysed, linear polymers may assemble into a ribbon structure by aligning hydroxy groups, bridging oxygens and equatorially coordinated waters to achieve hydrogen bonding.This work was supported by the Office of Naval Research and by a fellowship for G.A.P. from the University of Minnesota Center for Interfacial Engineering, an NSF Engineering Research Center. The authors are grateful for helpful discussions with Professor Martha McCartney (UC Irvine) and Drs. Eric Morrison and Joseph Bailey (3M). References 1 J. Livage, Chem. Mater., 1991,3, 578. 2 N. Gharbi, C. Sanchez, J. Livage, J. Lemerle and J. Lefebvre, fnorg. Chern. 1982 21,2758. 3 J. Lemerle, L. Nejem and J. Lefebvre, J. Inorg. Vucl. Chern., 1980, 42, 17. 4 J. Lemerle, L. Nejem and J. Lefebvre, J. Inorg. ,Vucl. Chem., 1981, 43,2683. 5 J. Livage and J. J Legendre, J. Coll. fnt. Sci., 19h2,94, 75. 6 J. Legendre, P.Aldebert, N. Baffier and J. Livage, J. Coll. Int., 1982,94,84. 7 G. A. Pozarnsky and A. V. McCormick, Chrm. Muter., 1994, 6, 380. 8 J. Livage, M. Henry and C. Sanchez, Prog. Solid Stute Chem., 1988, 18,259. 9 M. Henry, J. P. Jolivet and J. Livage, Structure Bonding, 1990, 77, 154. 10 J. K. Bailey, T. N. Nagase, G. A. Pozarnsky and M. L. Mecartney, in Better Ceramics through Chemistry IV, MRS Proceedings 180, eds. C. J. Brinker, D. E. Clark and D. L. Ulrich, Materials Research Society, Pittsburgh, 1990, p. 759. 11 J. K. Bailey, G. A. Pozarnsky and M. L. Mecartney, J. Muter. Sci., 1992,7, 2530. 12 W. G. Klemperer and W. Shum, J. Am. Chem. Soc., 1977,99,3544. 13 C. J. Besecker, W. G. Klemperer, D. J. Maltbie and D. A. Wright, Inorg.Chem., 1985,24, 1027. J. MATER. CHEM., 1994, VOL. 4 1753 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 V. W. Day, W. G. Klemperer and D. J. Maltbie, J. Am. Chem. SOC.,1987, 109,2991. A. T. Harrison and 0.W. Howarth, J. Chem. SOC., Dalton Trans., 1985,1953. A. T. Harrison and 0.W. Howarth, J. Chem. SOC., Dalton Trans., 1985,1173. R. C. Hibbert, N. Logan and 0. W. Howarth, J. Chem. Soc., Dalton Trans., 1986,369. E. Heath and 0.W. Howarth, J. Chem. SOC., Dalton Trans., 1981, 1105. W. G. Klemperer, in The Multinuclear Approach to NMR Spectroscopy, eds. J. B. Lambert and F. G. Riddell, Reidel Press, Boston, 1983, 671. W. G.Klemperer, Angew. Chem., Int. Ed. Engl., 1978, 17,246. L. Pettersson, B. Hedman, I. Anderson and N. Ingri, Chem. Scr., 1983,22, 254.L. Pettersson, B. Hedman and I. Anderson, Chem. Scr., 1985 25, 309. R. E. Mesmer and C. F. Baes, The Hydrolysis of Cations, John Wiley, New York, 1976,p. 195. C. Detellier, in Modern NMR Techniques and Their Application in Chemistry, eds. A. I. Popov and K. Hallenga, Marcel Dekker, New York, 1991, p. 521. J. Reuben and D. Fiat, Inorg. Chem., 1967,6, 579. C. Madic, G. M. Begun, R. L. Hahn, J. P. Launay and W. E. Thiessen, Inorg. Chem. 1984,23,469. D. Rehder, Bull. Magn. Reson., 1982,4, 82. 0.W. Howarth, Prog. NMR Spectrosc., 1990,22,453. J. Reuben, J. Am. Chem. SOC., 1969,91,5725. 31 32 33 34 35 36 37 38 39 40 41 42 43 44 H. M. Schwartz, M. MacCoss and S. Danyluk, Tetrahedron Lett., 1980,21,3837. C. Sanchez, J. Livage and G. Lucazeau, J. Raman Spectrosc., 1982, 12, 68.P. Aldebert, N. Baffier, N. Gharbi and J. Livage, Muter. Hes. Bull., 1981, 16, 669. L. Abello and G. Lucazeau, J. Chim. Phys., 1984,81,539. G. Pimental and G. L. McClellan, The Hydrogen Bond, lreeman, San Francisco, CA, 1969. H. T. Evans, Inorg. Chem., 1966,5,967. L. Abello, E. Husson, Y. Repelin and G. Lucazeau, J. Solid State Chem., 1985,56,379. R. C. T. Slade, J. Barker and T. K. Halstead, Solid Sta e lonics, 1987,27,221. M. T. Vandenborre, R.Prost, E. Huard and J. Livage, Mtiter. Res. Bull., 1983, 18, 1133. J. Livage, P. Barboux, J. C. Badot and N. Baffier, Better Ceramics Through Chemistry 111, MRS Proceedings 121, eds. C. J. Brinker, D. E. Clark and D. L. Ulrich, Materials Research Society, Pittsburgh, 1988, p. 167. M. Nabavi, F. Taulelle, C. Sanchez and M. Verdaguer, J. Phys. Chem. Solids, 1990,51, 1375. S. Stizza, M. Benfatto, A. Bianconi, J. Garcia, G. Mancini and C. R. Natoli, J. Phys. (Paris), 1986,47-C8,691. S. Stizza, I. Davoli and M. Benfatto, J. Non-Cryst. Solids, 1987, 96, 327. S. Stizza, G. Mancini, M. Benfatto, C. R. Natoli, J. Garcia and A. Bianconi, Phys. Rev. B, 1989,4229. 30 M. Burgar, T. E. St. Amour and D. Fiat, J. Phys. Chem., 1981, 85, 502. Paper 4/02409E; Received 25th April, 1994