Practical Aspects of 51V and 93Nb Solid-state NMR Spectroscopy

www.elsevier.com/locate/pnmrs

Progress in Nuclear Magnetic Resonance Spectroscopy 53 (2008) 128–191

Practical aspects of 51V and 93Nb solid-state NMR spectroscopyand applications to oxide materials

O.B. Lapina a,*, D.F. Khabibulin a, A.A. Shubin a, V.V. Terskikh b

a Boreskov Institute of Catalysis, Prosp. Lavrentieva 5, Novosibirsk 630090, Russiab Steacie Institute for Molecular Sciences, National Research Council Canada, Ottawa, Ont., Canada K1A 0R6

Received 29 September 2007; accepted 6 December 2007Available online 4 March 2008

Keywords: Solid-state NMR; Modern NMR techniques; NMR in catalysis; Vanadium oxide catalysts; Niobium oxide catalysts; Vanadia; Niobia; NMRof Group VB elements; 51V NMR; 93Nb NMR; 181Ta NMR

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1292. Group VB elements, NMR properties, and solid-state NMR concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

0079-6

doi:10.

Abb

contindoubleroscopGIPAWhigh-fimagic-quantusubstitsubstitfrequeSTMAdouble

* CoE-m

2.1. Definition of NMR parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1312.2. Vanadium-51 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1312.3. Niobium-93 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1332.4. Tantalum-181 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

3. Modern NMR techniques most suitable for studying 51V and 93Nb in solids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
3.1. NMR spectra of stationary samples (quadrupolar echoes, QCPMG and nutations) . . . . . . . . . . . . . . . . . . . . . . . . 1343.2. Magic-angle spinning (MAS) and high-speed magic-angle spinning (HS MAS) . . . . . . . . . . . . . . . . . . . . . . . . . . . 1353.3. Satellite transition spectroscopy (SATRAS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1363.4. Spinning sidebands analysis of selected transitions (SSTMAS). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1393.5. MAS and static spectra analysis (MASSA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1393.6. Multi-quantum MAS (MQMAS) and satellite transition MAS (STMAS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1413.7. Heteronuclear correlation spectroscopy (HETCOR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1443.8. Double resonance experiments (SEDOR, REDOR, TRAPDOR, and REAPDOR) . . . . . . . . . . . . . . . . . . . . . . . . 144
565/$ - see front matter � 2007 Elsevier B.V. All rights reserved.

1016/j.pnmrs.2007.12.001

reviations: CP, cross polarization; CS, chemical shielding; CSA, chemical shift anisotropy; CT, central transition; CW, continuous wave; CW-NMR,uous wave nuclear magnetic resonance; DAS, dynamic-angle spinning; DeNOx, nitrogen oxide abatement; DFT, density functional theory; DOR,

rotation; DQ, double-quantum; DQ STMAS, double-quantum satellite transition magic-angle spinning; EDAX, energy dispersive X-ray spect-y; EFG, electric field gradient; ESR, electron spin resonance; FID, free induction decay; FT-NMR, Fourier transform nuclear magnetic resonance;

, gauge-including projected augmented-wave; HETCOR, heteronuclear correlation spectroscopy; HFI, hyperfine interaction tensor; HFMAS,eld magic-angle spinning; HREM, high resolution electron microscopy; HSMAS, high-speed magic-angle spinning; KTN, KTa(1�x)NbxO3; MAS,angle spinning; MASSA, magic-angle spinning and static spectra analysis; MQMAS, multiple quantum magic-angle spinning; MQ, multiplem; NMR, nuclear magnetic resonance; NQR, nuclear quadrupolar resonance; QCPMG, quadrupolar Carr-Purcell Meiboom-Gill; PBN, Ba-uted Pb(Mg1/3Nb2/3)O3; PMN, Pb(Mg1/3Nb2/3)O3; PMN/PT, (1�x)Pb(Mg1/3Nb2/3)O3/xPbTiO3; PSN, Sc-substituted Pb(Mg1/3Nb2/3)O3; PZN, Zr-uted Pb(Mg1/3Nb2/3)O3; REAPDOR, rotational echo adiabatic passage double resonance; REDOR, rotational echo double resonance; RF, radioncy; SATRAS, satellite transition spectroscopy; SBV, Strongly bound vanadium; SEDOR, spin–echo double resonance; ST, satellite transition;S, satellite transition magic-angle spinning; SSTMAS, spinning sidebands analysis of selected transitions; TRAPDOR, transfer of population inresonance; VOCS, variable offset cumulative spectrum.

rresponding author.ail address: [email protected] (O.B. Lapina).

mailto:[email protected]

O.B. Lapina et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 53 (2008) 128–191 129

3.9. Triple Resonance experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1463.10. Advantages of high magnetic field strengths. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

4. DFT and other quantum chemical computational approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1485. 51V NMR data compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

5.1. Chemical shielding and quadrupolar tensor parameters in individual vanadium compounds. . . . . . . . . . . . . . . . . . 149
5.1.1. Tetrahedral Q0 sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.1.2. Tetrahedral Q1 sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.1.3. Tetrahedral Q2 sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1575.1.4. Associated non-axial VO5 and VO6 sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1575.1.5. Isolated and associated trigonal VO4 pyramids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1575.1.6. Isolated octahedral VO6 and tetragonal VO5 pyramids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1585.1.7. Associated tetragonal pyramids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1605.1.8. Strongly associated octahedral sites in decavanadates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.2. Correlating local environment of vanadium nuclei in VOx species with 51V NMR parameters . . . . . . . . . . . . . . . . 160
6. 93Nb NMR data compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.1. Chemical shielding and quadrupolar tensor parameters in individual niobium compounds . . . . . . . . . . . . . . . . . . . 162
6.1.1. Six-coordinated compounds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1626.1.2. Four-coordinated compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1656.1.3. Five-coordinated compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1666.1.4. Seven- and eight-coordinated compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
6.2. 93Nb NMR chemical shift scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
7. 181Ta NMR data compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1698. Paramagnetic effects in 51V and 93Nb solid-state NMR spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
8.1. Presence of paramagnetic cations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1718.2. Systems with vanadium in mixed oxidation states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

9. Recent applications of solid-state 51V and 93Nb NMR in oxide materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
9.1. Applications of solid-state 51V NMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
9.1.1. Bio-structural chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1759.1.2. Materials chemistry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1759.1.3. Catalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176

9.2. Applications of solid-state 93Nb NMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
9.2.1. Characterization of ferroelectrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1829.2.2. Silicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1839.2.3. Miscellaneous applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
9.3. Multinuclear solid-state NMR in vanadia and niobia catalysts supported on Al2O3 . . . . . . . . . . . . . . . . . . . . . . . 184
9.3.1. Vanadia sites in VOx/Al2O3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1849.3.2. Niobia sites in NbOx/Al2O3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1859.3.3. Niobia–vanadia species in (Nb–V)Ox/Al2O3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
10. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

1. Introduction

There are many similarities between the first three ele-ments of the Group VB. Two of these three elements, vana-dium and niobium, were discovered in 1801, and Tantalumwas discovered shortly after, in 1802. All three elementsbear mythological names: vanadium is named after theScandinavian goddess Vanadis because of its beautifulmulticolored compounds, and niobium is named afterNiobe, the daughter of Tantalus, the namesake ofTantalum.

All three elements find numerous applications in chem-ical industry, electronics, and metallurgy. Vanadium is avery versatile metal: besides its main use in steel manufac-turing, vanadia-based catalysts are frequently used forlarge-scale sulfuric acid production [1], for cleaning flue

gases, for selective oxidation of hydrocarbons [2–4], forreduction of nitrogen oxides with ammonia [5,6], and forproduction of bulk chemicals [4,7–9]. Vanadium haloper-oxidases [10] have potential as catalysts in industrial-scalebio-catalytic conversions. Recent years have also broughtgrowing interest in niobium-based oxide systems oftenshowing improved catalytic properties [11–13]. Even smallamounts of niobium oxide added to a catalytic mixturemay considerably enhance catalytic activity, selectivity,and long-term stability [13]. Niobium oxide itself or mixedwith other oxides (Nb2O5–SiO2, Nb2O5–Al2O3, Nb2O5–TiO2, Nb2O5–V2O5, etc.) is frequently used as a supportfor catalytically active metals or other metal-oxide catalysts[11–13]. Hydrated niobium pentoxide (‘‘niobic acid’’,Nb2O5ÆnH2O) and niobium phosphate have been shownto have unusually high surface acidity, significant catalytic

130 O.B. Lapina et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 53 (2008) 128–191

activity, exceptional selectivity, and high stability in manyacid-catalyzed reactions [13].

From the NMR point of view, all three elements, vana-dium, niobium, and tantalum, are quite similar as well.Each element has only one dominant isotope suitable forNMR spectroscopy, 51V, 93Nb, and 181Ta, each is quadru-polar having a half-integer nuclear spin.

Vanadium-51 (natural abundance 99.76%) has spin 7/2and an electric quadrupolar moment of only 0.05 barn.The relative receptivity of 51V NMR is 0.38 compared to1H NMR. The 93Nb nucleus (natural abundance 100%)has spin 9/2. Niobium-93 is one of the most NMR-recep-tive nuclei with a receptivity of 0.482 relative to 1H and afavorably low quadrupolar moment of 0.32 barn. On theother hand, tantalum-181 (natural abundance 99%, spin7/2) has one of the largest known quadrupolar moments,3.44 barn, which hampers considerably 181Ta NMR exper-iments on solid samples (see below).

The first solid-state 51V, 93Nb, and 181Ta NMR worksappeared in the late 1950s. The first reported 51V NMRspectra were recorded by Knight and Cohen in 1949 [14]on polycrystalline Pb(VO3)3 and V2O5 samples to deter-mine the magnetogyric ratio for the 51V nucleus. Similarresults for 93Nb were published in 1951 [15] and for 181Tain 1959 [16]. In 1961, Ragle and coworkers [17] describedthe 51V magnetic anisotropy in a polycrystalline V2O5

sample.In 1967–1969 several reports were published on single-

crystal 51V and 93Nb NMR. Gornostansky et al. deter-mined parameters of the magnetic shielding anisotropyand the quadrupolar coupling constants for 51V in V2O5

[18] and in KVO3 [19]. Single-crystal 93Nb NMR experi-ments had allowed one to measure the 93Nb quadrupolarcoupling constant in LiNbO3 [20,21].

Continuous wave (CW) NMR was successful in study-ing quadrupolar coupling parameters in the solid state,since such spectra often showed well-defined discontinuitiescorresponding to quadrupolar interactions (of the first-order for a small value and of the second-order for a largervalue of quadrupolar constant). This approach was used todetermine quadrupolar coupling constants in several vana-dates [18,22–26]. Similar 93Nb experiments were performedon LiNbO3 [27–31].

With the development of pulsed FT-NMR instrumentsit became more practical to determine magnetic shieldingparameters of 51V nucleus by NMR spectroscopy [32,33].However, in these first experiments the first-order quadru-polar effects were often very difficult to observe.

Early 51V NMR results obtained with CW-NMR andFT-NMR have been summarized by Pletnev et al. [22].The introduction of sample spinning at the magic angleto the external magnetic field (MAS), with spinning speedsup to 6 kHz, brought considerable improvements in spec-tral resolution allowing the identification of as many astwo or three non-equivalent vanadium sites in some com-pounds, and also allowed determination of the isotropicchemical shift values for each site [34].

Unlike the early success of 51V NMR spectroscopy,93Nb NMR studies were somewhat slow to follow. In manyrespects, this was because of the larger 93Nb quadrupolarmoment. In a solid the electric field gradients arising fromthe electronic cloud at the nucleus can interact with thenuclear quadrupolar giving rise to considerable spectralbroadening. Nevertheless, several Nb-containing systemswere studied in detail using combined static and conven-tional MAS 93Nb NMR [35–43].

Not surprisingly, 51V solid-state NMR has become animportant tool in characterizing the local structure ofvanadium sites in many vanadium-based systems [34,44].Modern NMR techniques such as ultrahigh-speed MASwith spinning speeds in excess of 35 kHz, MQMAS,SATRAS, and others methods, have allowed accurateinformation to be obtained on the local structure of vana-dium sites, i.e. (i) the number of non-equivalent vanadiumsites, (ii) coordination numbers, (iii) the nature of atomsin the first coordination sphere, (iv) distortion of the firstcoordination sphere, (v) association of vanadium–oxygenpolyhedra. In addition, spin–echo mapping spectra orultrahigh-speed MAS experiments have helped to identifyV5+ atoms bound via an oxygen atom to V4+ or anotherparamagnetic species [45]. Defects and distortions in thecrystal structure can be revealed by analysis of distribu-tions of the chemical shielding and quadrupolar tensorparameters [46].

A growing interest in niobium-based catalytic systems,as well as the practical importance of some Nb-containingpiezoelectric and optoelectronic materials, have stimulatedseveral recent 93Nb NMR studies employing modern solid-state NMR techniques [47–52]. Using conventional andultrahigh magnetic field facilities, ultrahigh-speed MAS,DQ-STMAS, solid-echo and computer modeling, chemicalshifts and quadrupolar tensor parameters have beenreported for a considerable number of Nb compounds[53–55]. It has been shown, that the 93Nb isotropic shiftis sensitive to the coordination number of Nb sites. Arecently proposed 93Nb NMR chemical shift approachallows determination of the coordination number in NbOx

polyhedra [55].Because of the relatively large 181Ta quadrupolar

moment and the low resonance frequency, there is only avery limited number of reports on 181Ta NMR in the solidstate [56]. Starting from the first 181Ta NMR work onKTaO3 [16], the total number of 181Ta NMR papers pub-lished so far is less than 10 [16,57–63].

This review on the current state of the solid-state 51Vand 93Nb NMR spectroscopy includes both previouslypublished and original results. Advantages and restrictionsof various solid-state NMR techniques as applied to vana-dium and niobium are discussed with illustrations from avariety of vanadium and niobium containing oxide materi-als, including individual highly crystalline compounds,solid solutions, glasses, and catalysts. The main purposeof this work is to provide readers with the latest compre-hensive compilation of 51V and 93Nb NMR data in oxide


materials, and to demonstrate the great potential of 51Vand 93Nb NMR in solid-state chemistry.

2. Group VB elements, NMR properties, and solid-state

NMR concepts

The Group VB has only three nuclei most commonlyconsidered for NMR spectroscopy, 51V, 93Nb, and 181Ta.The energy level diagrams for these three quadrupolarnuclei are rather similar. An example for a spin-9/2 system,i.e. for 93Nb, is shown in Fig. 1 together with the corre-sponding calculated powder patterns perturbed by thefirst-order and the second-order quadrupolar interactions.NMR properties of 51V, 93Nb, and 181Ta are summarizedin Table 1.

2.1. Definition of NMR parameters

In this review, we will use the following convention forthe quadrupolar coupling constant, CQ (MHz), and theasymmetry parameter, gQ, as:

CQ ¼eQV ZZ

h; gQ ¼

V YY � V XX

V ZZð1Þ

Here, Q is the quadrupolar moment of the nucleus, andVXX, VYY, VZZ are the principal values of the traceless elec-tric field gradient (EFG) tensor Vij ordered in a sequence|VZZ| P |VXX| P jVYY| with the common designationVZZ = eq for its largest principal component, while gQ is

20000 10000 0

A

200 0 -200 -400

C

B

(ppm)

400

Fig. 1. Effects of the quadrupolar interactions on a spin-9/2 system in a solid saat 9.4 T including all transitions. (C) Second-order powder pattern of the cenm0 = 97.9 MHz.

the asymmetry parameter of the EFG (or nuclear quadru-polar) tensor.

For a chemical shift tensor we use the following defini-tion for the isotropic chemical shift, diso, the chemical shiftanisotropy (CSA), dr, and the CSA asymmetry parameter,gr:

diso ¼ 13ðdXX þ dYY þ dZZÞ ð2Þ

dr ¼ dZZ � diso ð3Þ

gr ¼dYY � dXX

dZZ � diso

ð4Þ

Here, dXX, dYY, and dZZ are the principal components ofthe CSA tensor.

Note that the CSA tensor d is related to the chemicalshielding anisotropy tensor r as

d ¼ Iriso � r ð5Þ

Here, I is the unit matrix and riso is the isotropic value ofthe chemical shielding tensor for a selected reference com-pound. Absolute chemical shieldings for some commonlyused reference compounds can be found in [64].

2.2. Vanadium-51

The vanadium-51 nucleus has high NMR receptivityand a convenient resonance frequency, which is very closeto that of the frequency of 13C. The 51V NMR spectra ofsolid samples show not only quadrupolar interactions,but also sizable magnetic shielding effects. When present

-9/2

-7/2

-5/2

-3/2

-1/2

1/2

3/2

5/2

7/2

9/2

6

2

-1

-3

-15/4

15/4

-3

-126

9(16sin4θ + 2sin22θ )

7(0sin4θ + 6sin22θ )

m B0 only Zeeman Interaction

-hν0m

I-order quadrupolar

)1cos3(288

2 −θQhCII-order quadrupolar

096νQhC

5(9sin22θ - 12sin4θ )

3(11sin22θ - 20sin4θ )

(12sin22θ - 24sin4θ )

(24sin4θ - 12sin22θ )

3(20sin4θ - 11sin22θ )

5(12sin4θ - 9sin22θ )

-7(0sin4θ + 6sin22θ )

-9(16sin4θ + 2sin22θ )

-10000 -20000

mple. (A) Energy level diagram. (B) Calculated first-order powder patterntral transition for gQ = 0. Simulation parameters CQ = 20 MHz, gQ = 0,

Table 1NMR properties of the Group VB elements

Isotope Spin Naturalabundance (%)

Quadrupolarmoment (barn)

Sensitivityrelative to 1H

NMR frequency(MHz) at

Chemical shiftrange (ppm)

Referencesample

9.4 T 21.1 T

51V 7/2 99.75 �0.05 0.38 105.2 236.6 2000 VOCl393Nb 9/2 100.0 �0.32 0.48 97.8 220.0 4000 NbCl5/CH3CN181Ta 7/2 99.99 3.44 0.04 47.9 107.7 3450 K[TaCl6]


together, these two contributions can considerably compli-cate NMR spectra and their analysis. When studyingmultiphase amorphous systems, such as most of the van-adia-based heterogeneous catalysts, the 51V NMR spectrabecome extremely complicated, and advanced NMR tech-niques together with spectral simulations are often requiredfor their successful interpretation.

It is common for 51V to exhibit quadrupolar couplingconstants ranging from 2 to 6 MHz, but rarely exceeding10 MHz [34]. At the same time the magnetic shieldinganisotropy is normally below 1000 ppm, and is often foundwithin the 100–500 ppm range, depending on the coordina-tion environment.

Simulated static spectra of all NMR transitions for aspin-7/2 nucleus, i.e. 51V, with a quadrupolar coupling con-stant CQ = 4.5 MHz are shown in Fig. 2A (CQ = 4.5 MHz,gQ = 0–1). The whole spectrum, including the outer±5/2 M ±7/2 transitions, extends over 1.5 MHz, whichrequires a radio-frequency pulse for its homogeneous exci-tation shorter than a few tenths of microsecond.

Modern FT-NMR solid-state spectrometers are well-equipped for obtaining 51V NMR spectra in solid materials

600 400 200 0 -200 -400 -600

0.0

0.2

0.4

0.6

0.8

1.0

1000

A B

, ppm

ηQ

δ

Fig. 2. NMR powder patterns calculated for a spin-7/2 system of a solid polycdifferent values of gQ (CQ = 4.5 MHz). (B) Chemical shielding anisotropy effect

with all the quadrupolar satellite transitions present. Min-imal requirements are reasonably high magnetic fields(P9 T), MAS spinning speeds exceeding 15 kHz, extremelyshort radio frequency pulses, normally shorter than 0.5 ls,and very fast digitizing rates. For example, a 20 MHz(50 ns) digitizing rate is necessary to provide a 10 MHz fullspectral width.

Along with the quadrupolar interactions, as alreadymentioned above, it is also important to consider the mag-netic shielding while analyzing 51V NMR spectra. Themagnetic shielding is commonly described as a chemicalshift anisotropy tensor with three principal tensor compo-nents. The chemical shift anisotropy is most pronouncedin the central transition. This is illustrated in Fig. 2B fora series of spectra calculated with different CSA asymmetryparameters, gr.

For 51V NMR the central transition is very oftenaffected by the first-order quadrupolar interactions(Fig. 3). Both the quadrupolar and the magnetic shieldinginteractions are described by second-rank tensors, whichmake it necessary to define their relative orientation. Thisrelative orientation is given by three Euler angles, a, b,

500 0 -500 -1000

1.0

0.75

0.5

0.25

0.0

ησ

, ppmδ

rystalline sample at 105.2 MHz. (A) The first-order quadrupolar effects ats on the central transition at different values of gr (dr = 500 ppm, CQ = 0).

500 0 -500 -1000 -1500 -2000500 0 -500 -1000 -1500 -2000δ, ppm

A B

1

2

3

4

5

6

78

1

2

3

4

5

Fig. 3. Combined effects of the quadrupolar and magnetic shielding interactions on the NMR spectra for a spin-7/2 system of a solid, polycrystallinesample. (A) Simulated with NMR parameters typical for V2O5, m0 = 105.2 MHz CQ = 0.797 MHz, gQ = 0.08, diso = �609 ppm, dr = 645 ppm, gr = 0.11,a = 42�, b = 126�, c = 5�; (1) full spectrum and transitions (2) +7/2 M +5/2, (3) +5/2 M +3/2, (4) +3/2 M +1/2, (5)+1/2 M �1/2, (6) �1/2 M �3/2, (7)�3/2 M �5/2, (8) �5/2 M �7/2. (B) Effects of the relative orientation of the CS and quadrupolar tensors. Only the Euler angle b is varied as following: (1)b = 160�, (2) b = 140�, (3) b = 126�, (4) b = 110�, (5) b = 90�.


and c. Therefore, in general case, there are eight indepen-dent parameters in the Spin-Hamiltonian describing asolid-state 51V NMR spectrum. These parameters are threeprincipal components of the CSA tensor, any two principalcomponents of traceless quadrupolar tensor, and threeEuler angles describing relative orientation of the CSAand quadrupolar tensors (see also Eqs. (1)–(4) above forthe definition of commonly used corresponding NMR +parameters).

In the first-order of perturbation theory, each NMRtransition has only three singularities (Fig. 3). It is there-fore necessary for spectra of static samples to analyze atleast three separate transitions in order to obtain the fullset of eight independent parameters. This requires observa-tion of not only the central +1/2 M �1/2 transition, butalso two satellite transitions. The central transition line-shape can be used to determine the CSA parameters, whilethe quadrupolar interaction parameters can then be deter-mined from the satellite transitions. We note, however, thatthe magnetic shielding affects not only the central transi-tion but the satellite transitions as well, i.e. the singularitiesare now affected by both interactions (Fig. 3B).

Let us consider the most narrow ±3/2 M ±1/2 satellitetransitions in the magnetic field of 9.4 T and the quadrupo-lar coupling constant of CQ = 4.5 MHz (gQ = 0). The totalwidth of these transitions will be about 4500 ppm due tothe quadrupolar interactions only. These transitions arewider by some 15% if the magnetic shielding, dr = 500 ppmand gr = 0, is also present. In this case if the magneticshielding effects are not taken into consideration, the calcu-

lated quadrupolar coupling constant could be easily over-estimated by as much as 15%.

2.3. Niobium-93

93Nb NMR spectra, even at very high magnetic fields,are often dominated by the quadrupolar interactions. Forthe central transition +1/2 M �1/2, the second-orderquadrupolar perturbation results in a characteristic powderpattern. Non-central, or satellite, transitions are spread farfrom the central Larmor frequency. At the same time, forI = 9/2 nuclei, the satellite transitions are closer to the cen-tral transition compared with those from nuclei with lowerspin quantum numbers (for a given CQ). Thus, for I = 9/2not only a central transition, but also several satellite tran-sitions can routinely be observed using conventional solid-state NMR spectrometers.

At present, there is no universal method for quadrupo-lar nuclei with a half-integer spin that allows one to obtaina complete set of the quadrupolar and chemical shielding(CS) tensor parameters for the wide range of values ofthe quadrupolar constant and the chemical shift anisotropyoften found for 93Nb. Depending on the absolute values ofthese parameters and on their relative magnitude only cer-tain NMR techniques or a combination of several NMRtechniques can be applied successfully. As it has been sug-gested in [47–54,65–67], the most suitable techniques for93Nb are experiments at high magnetic fields applyinghigh-speed MAS, solid-echo, MQMAS, pure-phase nuta-tion, and STMAS techniques [55].


2.4. Tantalum-181

Due to the relatively large 181Ta quadrupolar moment(3.44 barn), 181Ta NMR studies on solid samples are rareand seriously hampered by strong line broadening causedby quadrupolar relaxation. Another complication is thelow resonance frequency and the consequent low receptiv-ity. 181Ta may be considered as a low-gamma nucleus andas such is very difficult to work with, particularly for solidsamples. However, the availability of ultrahigh magneticfields for solid-state NMR research makes 181Ta NMR fea-sible, yet challenging. The NMR spectra of 181Ta for solidsamples are similar to those of 51V and 93Nb and will berepresented by a set of central and satellite transitions,unless the nucleus is in an environment with high symme-try. The large 181Ta quadrupolar moment ensures thatthe NMR spectra will most likely be dominated by quadru-polar interactions, but magnetic shielding effects should notbe ignored.

3. Modern NMR techniques most suitable for studying 51V

and 93Nb in solids

3.1. NMR spectra of stationary samples (quadrupolar

echoes, QCPMG and nutations)

The first solid-state 51V and 93Nb NMR spectra wererecorded on stationary samples [22,24,27–32,68–70]. Eventhough numerous line-narrowing techniques like MASand MQMAS have since been introduced, recordingNMR spectra on static samples remains important andoften indispensable for proper interpretation of the NMRdata either in individual crystalline compounds or in suchcomplex multiphase systems as heterogeneous catalysts.

In most cases, static 51V and 93Nb NMR spectra repre-sent a superposition of the central and satellite transitions.By manipulating with the excitation bandwidth of rf pulses,it is possible to obtain separately the spectrum of the cen-tral transition only, and the spectrum including all or atleast several satellite transitions. However, increasing thesweep width in order to record broader spectra often leadsto amplified acoustic ringing of the probehead. Whenacoustic ringing is present, it causes the loss of the firstfew points in the FID, which results in significantly dis-torted spectra. In the worst cases, the very broad spectracan be lost altogether solely due to the dead time of theprobe/receiver. This problem can be circumvented byapplying techniques involving spin–echo pulse sequences.

There are two main types of echo pulse sequences usedin solid-state NMR of quadrupolar nuclei with a half-inte-ger spin: the Hahn-echo [71] and the solid-echo [65,72]. Inthe solid-echo (or quadrupolar echo) pulse sequence,p1 � s1 � p2 � s2 � AQ, rf pulses p1 and p2 are not neces-sarily multiples of a p/2 pulse, which makes this sequenceuseful for recording very broad spectra. In the Hahn-echopulse sequence the pulse durations are defined asp2 = 2 · p1 and are normally multiples of the p/2 pulse.

For NMR of qudrupolar nuclei, such as 51V and 93Nb,with many satellite transitions present, it is often difficultto define a single p/2 pulse. This can be done, however,for each separate transition, for example, for the centraltransition only. In practice, the Hahn-echo pulse sequenceis often used not only to obtain the spectra of central tran-sitions, but also to carefully calibrate the p/2 pulse, whichthen can be applied in more advanced experiments employ-ing various n-quantum coherence filters.

The line shape in static spectra, either of the centraltransition, or when superimposed with several satellitetransitions, is very sensitive to durations of pulses in theecho sequences and to delays between pulses. Inaccuratechoice of pulses or delays may lead to distorted line shapes,and therefore to an incorrect interpretation of the spectra.The situation becomes even more complicated, when thereis more than one site present in the system. The effects ofquadrupolar echoes for half-integer spins in static solid-state NMR spectra have recently been discussed in[65,73,74].

Improved signal-to-noise ratio in static spin–echo spec-tra can be achieved by recording the whole echo spectra,via a modified solid-echo pulse sequence, p1 � s1 �p2 � AQ, as demonstrated by Massiot et al. [75] and Wuand Dong [76]. The whole-echo technique is beneficialwhen the intrinsic spin–spin relaxation time of the sample,T2, is long sufficiently and the lines are sufficiently broad toavoid truncation. In such cases the sensitivity can beimproved quite easily by a factor of

p2, while still preserv-

ing the correct lineshape. The whole-echo acquisition isoften used to improve signal-to-noise in MQMAS and sim-ilar experiments (see below).

Recently re-introduced to NMR of half-integer quadru-polar nuclei, a Carr-Purcell Meiboom-Gill technique,QCPMG, provides even greater sensitivity enhancementover the conventional spin–echo [77–81]. In QCPMGexperiments, a standard solid-echo pulse sequence asshown above is followed by a series of p pulses with awhole-echo acquisition after each p pulse. The resultingtrain of whole echoes can than be Fourier transformed togive a series of equally spaced sharp spikelets outliningthe static powder pattern, somewhat resembling an MASspectrum. Another approach in processing QCPMG spec-tra involves adding together all echoes in the train and thentreating the resulting sum as a regular whole echospectrum.

An example of a 93Nb QCPMG NMR spectrumrecorded for La3NbO7 is shown in Fig. 4. In this case useof the QCPMG spectrum reproduces very well the staticpowder pattern obtained for the same sample via use of atraditional spin–echo approach. Either of the two spectracan be used to determine the quadrupolar and CSA param-eters for Nb sites in this compound. While the two spectraare of about the same quality, equally suitable for furtheranalysis, the QCPMG spectrum required only 64 scans toacquire, while the solid-echo spectrum was obtained in4096 scans, i.e. took 64 times longer to acquire. Generally

-800 -1000 -1200 -1400δ, ppm

1

2

3

Fig. 4. 93Nb NMR spectra of a stationary La3NbO7 sample of apolycrystalline solid recorded at 21.1 T. (1) Experimental spin–echospectrum. (2) Spectrum (1) simulated with the following parameters:CQ = 49 MHz, gQ = 0.275, diso = �968 ppm, dr = �113 ppm, gr = 0.69,a = 50�, b = 27�, c = 72�. (3) Experimental QCPMG spectrum. Thenumber of accumulated scans was 4096 for (1) and 64 for (3).


speaking, a QCPMG pulse sequence containing n full ech-oes in its acquisition cycle delivers sensitivity enhancementby a factor of n

p2 over a traditional single-echo

experiment.Because of this considerably improved sensitivity, today

the QCPMG technique finds many applications in solid-state NMR of low-gamma and low natural abundancequadrupolar nuclei. Most recently QCPMG, often com-bined with a variable offset cumulative spectrum approach,VOCS, is being used for systems with extremely large quad-rupolar coupling constants [82]. Particularly large quadru-polar coupling constants are also expected for mosttantalum compounds. Naturally, in such systems 181TaQCPMG NMR should become the exploratory techniqueof choice. 51V and 93Nb NMR benefits from QCPMG incases where the vanadium or niobium content is small,for example, in many supported catalysts, or in vanadium-or niobium-doped electronic materials.

While valuable, QCPMG does not always provide veryaccurate line shapes as shown by Ooms et al. [83], and thismay complicate analysis of the spectra, or even lead to awrong interpretation of the experimental data. QCPMGalso requires sufficiently long T2 relaxation times to acquirea train of echoes. Unfortunately in many important sys-tems such as supported heterogeneous catalysts T2 relaxa-tion times are often not long enough.

For stationary samples, Samoson and Lippmaa havedeveloped a nutation technique, which is based on theacute sensitivity of the static line shape to the duration ofan excitation rf pulse [84]. By comparing the nutation fre-quency with the frequency of the excitation rf pulse it ispossible to determine the quadrupolar coupling parame-ters. The nutation technique is not very useful for 51V, sincefor this nucleus it is typical to observe only the first-order

quadrupolar interactions. In contrast, for the 93Nb NMRwith strong second-order quadrupolar effects even in thehighest magnetic fields, the nutation technique has beenproven quite informative, particularly when performed ina two-dimensional fashion [48,50].

One of the advantages of recording static NMR spectraof quadrupolar nuclei, including 51V and 93Nb, is simplicityof execution, and that it does not involve purchasing orbuilding very expensive and maintenance-demandingMAS probes. Also, the static line-shape does not requiresignificant computing power to analyze. On the other hand,static NMR spectra have limited use mostly due to theirbroadness and the resulting insufficient spectral resolution.It is also a very challenging task to obtain correct lineshapes. When there are more than one or two individualsites present, it often becomes impossible to interpret aspectrum. Even the slightest distortion in the local nucleusenvironment may render useless any attempts to obtainmeaningful spectral information.

3.2. Magic-angle spinning (MAS) and high-speed magic-

angle spinning (HS MAS)

The first 51V MAS NMR spectra of vanadium com-pounds [85], were recorded by using MAS to successfullyminimize effects of dipolar interactions, magnetic shielding,and the first-order quadrupolar interactions. At the sametime, the low spinning speeds used, frequently below5 kHz, often resulted in multiple spinning sidebands over-lapping with isotropic 51V lines thus complicating the inter-pretation of the spectra. In these earlier 51V NMR works,MAS sidebands were considered a nuisance, and everyattempt was made to minimize their number in the spectra.Low MAS spinning speeds have little effect in resolving93Nb NMR spectra broadened by strong quadrupolareffects.

Recent advances in the MAS probe technology haveresulted in considerably increased spinning speed ratesand improved spinning stability. All major manufacturersof solid-state NMR equipment are now offering MASprobes capable of spinning speeds ranging from 35 to70 kHz [86–90]. However, the final choice of the MAS spin-ning speed still greatly depends on the system under inves-tigation and on the applied NMR technique. For example,the spinning speed should be as low as possible in order toanalyze the spinning sidebands of the satellite transitions(SATRAS, see below). At the same time, high spinningspeeds or several different spinning speeds are oftenrequired to resolve non-equivalent sites.

The 51V NMR chemical shift range in solid vanadia-based systems has been reported to exceed 1200 ppm, withmost of the shifts falling within ca. 500 ppm. To completelyfree this spectral window from MAS spinning sidebandswhile performing experiments at 9.4 T (51V resonancefrequency of 105.2 MHz), it would be necessary to spinthe sample at speeds exceeding 40 kHz. Lower spinningspeeds may render impossible a correct interpretation of


the spectra in a multi-component system. At higher mag-netic fields, the MAS spinning speed should be even faster,i.e. exceeding 60 kHz at 14 T, and 90 kHz at 21 T.

Spectral resolution in a MAS spectrum is limited by theline width of the isotropic lines. In 51V NMR spectra thefirst-order quadrupolar effects and the magnetic shieldingeffects are effectively averaged by fast MAS. At the sametime, the second-order quadrupolar effects are averagedby MAS only partially, thus leading to a residual broaden-ing of the lines even under infinitely fast MAS. Additionalbroadening in the spectra is introduced by a distribution ofthe NMR parameters due to the nature of the sample, i.e.due to various defects, surface effects, a lack of, or low levelof crystallinity, etc. In such cases a numerical simulation ofthe spectrum is the only suitable approach to achieve a cor-rect interpretation. Only a few of the currently availablecomputational tools for spectra simulations, offer theoption of including the distribution of NMR parametersin their fitting procedures [91].

The distribution of the magnetic shielding parametersand the quadrupolar coupling parameters affect theNMR spectra in a somewhat similar fashion. In additionto overall line broadening, the singularities in the line shapeare less pronounced. It is important to note that MAS spec-tra are more sensitive to the distribution of the magneticshielding parameters (Fig. 5B, spectrum 2), while the distri-bution of the quadrupolar parameters is particularly obvi-ous in the static spectra (Fig. 5A, spectrum 3). Thedistribution in both, the magnetic shielding and the first-order quadrupolar interactions, results in a homogeneousline broadening, while the distribution in the second-orderquadrupolar interactions results in a quasi-homogeneousline broadening.

In the case of 93Nb NMR, when the line width is mostlygoverned by the quadrupolar interaction, only the highestavailable MAS spinning speeds will be effective at lowmagnetic fields. Considerably improved resolution in the93Nb NMR spectra can be achieved at the highest available

2000 1000 0 -1000 -2000

3

2

1

A

, ppm δ

Fig. 5. Effects of distributions of NMR parameters on (A) powder patterns in s7/2 system at 105.2 MHz without distributions with the following parametersdistribution of 20 ppm. (3) As in (1) but with CQ distribution of 400 kHz.

magnetic fields by applying the highest available MASspinning speeds.

Today, the high-speed MAS technique is frequentlyapplied in a combination with advanced line-narrowingpulse sequences, including MQMAS, STMAS, and CP/MAS.

3.3. Satellite transition spectroscopy (SATRAS)

The SATRAS technique is based on a numerical analy-sis of the integral intensities of the MAS spinning side-bands [92–96]. Therefore, both the spectral resolutionand the spinning stability are important factors in theSATRAS spectra analysis. As an example, some simulatedSATRAS spectra are shown in Fig. 6A. It is common toobserve a spectrum being a superposition of the severaltransitions. The resulting spinning sideband patterns canbe quite complicated, which sometime makes it very diffi-cult to interpret them (Fig. 6). For example, in the caseof 51V, every spinning sideband is often a superpositionof up to four individual lines from different satellite transi-tions. Because the spinning sidebands from different satel-lite transitions have different line width and slightlyshifted relative to each other, it is important to know whatcontribution each makes into the integral intensity beingmeasured.

Effects of the line broadening on the MAS spinning side-bands of the transitions ±7/2 M ±5/2, ±5/2 M ±3/2, and±3/2 M ±1/2 are shown in Fig. 6B. Even the slightest addi-tional broadening, here by only 100 Hz, may lead to con-siderable broadening of the spinning sidebands fromsome transitions, i.e. ±7/2 M ±5/2, making them impossi-ble to detect by NMR. As a result, in some cases the inte-gral intensities of the spinning sidebands cannot bemeasured with sufficient precision.

It is known [97], that the integral intensity of a transition±m M ±(m � 1) is proportional to I(I + 1) � m(m � 1), i.e.for a spin-7/2 nucleus this means that the integral intensity

2000 1000 0 -1000 -2000

B

, ppm δ

tationary samples and (B) on 5 kHz MAS spectra. (1) Calculated for a spin-, dr = 200 ppm, gr = 0, CQ = 2 MHz, gQ = 0.2. (2) As in (1) but with diso

8000 4000 0 -4000 -8000

1

2

1800 1500 1200

1

2

3

A

δ, ppm δ, ppm

B

Fig. 6. Simulated MAS NMR spectra for a spin-7/2 system. (A1) As-recorded SATRAS spectrum. (A2) Integral intensity spectrum of the spinningsidebands. (B1 and B2) Effects of the broadening on the MAS spinning sidebands of the transitions ±7/2 M ±5/2, ±5/2 M ±3/2, and ±3/2 M ±1/2. (B1)10 Hz broadening. (B2) 100 Hz broadening. (B3) Intensities of the satellites for all transitions (black stack) and without a ±7/2 M ±5/2 transition (graystack). Spectra were calculated at 105.2 MHz with the following parameters, CQ = 5 MHz, gQ = 0.4, dr = 200 ppm, gr = 0.7, a = 20�, b = 20�, c = 0�,mr = 10 kHz.

Table 2Relative quadrupolar shift d(2)(m) and line broadening Dm of the spinningsidebands of the satellite transitions caused by the second-order quadru-polar interactions for spin-7/2 nuclei

m 1/2 3/2 5/2 7/2

Dm/D(m = 1/2) 1 0.622 �0.511 �2.4d(2)(m)/d(2)(m = 1/2) 1 0.4 �1.4 �4.4


of the spectrum is distributed as 16:15:12:7 from the centraltransition to the satellite transitions. If the ±7/2 M ±5/2transition becomes ‘‘invisible’’, this would result in the lossof up to 14% of the total spectral intensity. However, thecontribution of each satellite transition into the spinningsidebands depends upon the quadrupolar and magneticshielding parameters and the MAS spinning speed, i.e.the resulting distribution of intensities across the spectrummay not even follow the pattern mentioned above. All thismay lead to considerable errors in determining the spectralparameters from the analysis of the SATRAS spectra.

To somewhat minimize possible errors, it is critical tocarefully choose the proper experimental conditions andthe pulse sequence. Selection of adequate rf pulse dura-tions, rf power, pulse phases, relaxation delays, all helpto define which part of the spectrum is being excited, i.e.which satellite transitions are being observed.

According to Samoson [98], the spectral width, d(2)(m),and broadening, D(m), caused by the second-order quadru-polar interactions for a particular satellite transition aregiven by:

dð2ÞðmÞ ¼ 3

40

C2Q

m20

IðI þ 1Þ � 9mðm� 1Þ � 3

I2ð2I � 1Þ21þ

g2Q

3

!ð6Þ

DðmÞ ¼ 3

128

C2Q

m20

6IðI þ 1Þ � 34mðm� 1Þ � 13

I2ð2I � 1Þ21þ

g2Q

3

!ð7Þ

where I is the nuclear spin, quantum number m = 1/2 rep-resents the central transition, while m = 3/2 represents thesatellite transitions between m = ±1/2 and m = ±3/2, etc.Relative spectral widths of the spinning sidebands from dif-ferent transitions for a spin-7/2 nucleus are summarized inTable 2. The broadest are the ±7/2 M ±5/2 transitions,while all other satellite transitions are actually narrowerthan the central transition. It is clear, that the transitions+1/2 M �1/2, ±1/2 M ±3/2, and ±3/2 M ±5/2 are the eas-iest to observe experimentally. At the same time, recording

the ±5/2 M ±7/2 transitions with sufficient signal-to-noiseratio would require much longer accumulation times.

Because the SATRAS technique employs analysis of theintegral intensities of the satellite transitions, it is more con-venient to perform calculations taking into account only+1/2 M �1/2, ±1/2 M ±3/2, and ±3/2 M ±5/2 transi-tions, and to exclude the ±5/2 M ±7/2 transitions. Thisconsiderably simplifies the analysis and normally producesmuch more accurate results. The full MAS spectrum of allfour quadrupolar transitions is shown in Fig. 7. Unaidedanalysis of such spectra is not straightforward, since evenvisible singularities cannot unambiguously be assigned toany particular satellite transition. The following algorithmhas been developed in our group to simplify analysis of theSATRAS spectra of quadrupolar nuclei with half-integerspin greater than 3/2, including 51V and 93Nb.

First, the individual satellite transitions should be iden-tified in the spectrum (Fig. 7). Since the outermost satellitetransitions are the most sensitive to the quadrupolar inter-actions, it is convenient to use these transitions to estimatethe upper limit of the quadrupolar coupling constant.Excluding the ±5/2 M ±7/2 transitions (Fig. 7, spectra 3and 8) as mentioned above, the next outermost satellitetransitions, ±3/2 M ±5/2, should be analyzed (Fig. 7, spec-tra 4 and 7). These transitions have the narrowest spinningsidebands, which makes it easier to identify these particulartransitions. In general, any satellite transition ±(m � 1) M

±m is asymmetric with the line shape depending on the

10000 5000 0 -5000 -10000

x2

x2

x10

x10

1

3

4

5

6

7

8

1000 0 -1000

2

δ, ppm

δ, ppm

Fig. 7. Simulated MAS NMR spectra for a spin-7/2 system. (1) Full spectrum representing a superposition of all transitions. (2) Central +1/2 M �1/2transition. (3–8) Satellite transitions +7/2 M +5/2, +5/2 M +3/2, +3/2 M +1/2, �1/2 M �3/2, �3/2 M �5/2, �5/2 M �7/2, respectively. The verticalintensities of some satellite transitions are scaled as shown. Spectra were calculated at 105.2 MHz with the following parameters, CQ = 5 MHz, gQ = 0.4,dr = 200 ppm, gr = 0.7, a = 20�, b = 20�, c = 0�, mr = 10 kHz.

7500 7000 6500 6000 5500δ, ppm

Fig. 8. Simulated MAS NMR spectrum for a spin-7/2 system showingdetails of the �3/2 M �5/2 transition (narrow lines) and the ±5/2 M ±7/2transition (broad lines). Spectra were calculated at 105.2 MHz with thefollowing parameters, CQ = 5 MHz, gQ = 0.4, dr = 200 ppm, gr = 0.7,a = 20�, b = 20�, c = 0�, mr = 10 kHz.


Euler angles defining the relative orientations of the chem-ical shift and quadrupolar tensors. As a result the followingsimplified analysis may lead to somewhat overestimatedvalues of the quadrupolar coupling constants.

Each satellite transition line shape has three singulari-ties, similar to the three singularities produced by valuesof the chemical shift tensor for the central transition. Forthe transitions ±3/2 M ±5/2, these singularities are themost pronounced and easiest to use in estimates. We notethat in the static spectra with a non-zero asymmetryparameter of the quadrupolar interaction, the singularitiesare practically unobservable. At the same time in the MASspectra these singularities can be seen as a characteristicshape of the spinning sideband envelope. This becomesparticularly important for the ±3/2 M ±5/2 satellite transi-tions, when the intensity of the spinning sidebandsdecreases sharply at the positions of singularities, and isadditionally highlighted by very broad spinning sidebandsfrom the ±5/2 M ±7/2 transitions (Fig. 8).

When the ±3/2 M ±5/2 transitions have been identified,it is then possible to estimate the value of the quadrupolarcoupling constant. All other NMR parameters responsiblefor the line shape of the satellite transitions, including theasymmetry parameter and the Euler angles, cannot be esti-mated in a similar simplified fashion. To obtain theseparameters a complex numerical analysis of the integral


intensities of the spinning sidebands is required, ofteninvolving an extremely time consuming minimization pro-cedure and requires considerable computing resources [99].

To summarize, the SATRAS technique has a certainadvantage of being able to precisely determine all NMRparameters at once from a single one-dimensional NMRspectrum. At the same time this technique is applicableand limited only to highly crystalline samples with smallto moderate quadrupolar coupling constants. Anotherobvious limitation is the large volume of calculationsrequired.

3.4. Spinning sidebands analysis of selected transitions

(SSTMAS)

It is not uncommon while performing SATRAS analysisto obtain not one but several different sets of NMR param-eters fitting the experimental spectrum equally well. In suchcases another analytical approach can be employed to iden-tify the correct set. This approach uses lineshape analysis ofspinning sidebands from selected satellite transitions, thuscalled spinning sidebands analysis of selected transitionsor SSTMAS. This approach has been found particularlyuseful for spin-9/2 systems including 93Nb [54,55].

As a first iteration step, an approximate value of thequadrupolar coupling constant can be estimated from thelow-field static spectra. This approximate CQ value is fur-ther refined via lineshape analysis of spinning sidebandsfrom selected satellite transitions in MAS spectra. In thecase of 93Nb, the analysis is usually performed for the satel-lite ±3/2 M ±5/2 transitions and the central +1/2 M �1/2transition. Relative shifts and broadening of the satellitetransitions with respect to the central transition for I =9/2 were calculated earlier by Du et al. [53,54] using Eqs.(6) and (7) [98] (Table 3).

As shown in Table 3, for I = 9/2 the m = 5/2 satellitetransition has the smallest broadening of 0.055 in respectto the central transition. As for the second-order quadru-polar shift, only for the m = 3/2 satellite transition thequadrupolar shift has the same sign as the central transi-tion. These two effects are important in calculating thequadrupolar coupling constants from experimental 93NbMAS spectra.

The range of quadrupolar coupling constants that canbe estimated like this is 15–50 MHz at 9.4 T and 20–80 MHz at 21.1 T [55]. In experimental spectra the line-shape of spinning sidebands will also depend on variations

Table 3Relative quadrupolar shift d(2)(m) and line broadening Dm of the spinningsidebands of the satellite transitions caused by the second-order quadru-polar interactions for spin-9/2 nuclei [54]

m 1/2 3/2 5/2 7/2 9/2

Dm/D(m = 1/2) 1 0.764 0.055 �1.125 �2.778d(2)(m)/d(2)(m = 1/2) 1 0.625 �0.5 �2.375 �5

in gQ [53,54]. A numerical simulation of the full spectrumand the central transition of MAS spectra allows for accu-rate determination of CQ, gQ, and diso values.

Using the values of CQ, gQ, and diso obtained from theanalysis of satellites, it is then possible to calculate thechemical shift anisotropy parameter dr from analysis ofthe static spectra recorded at different magnetic fields.The static spectra recorded at higher fields are more infor-mative in determining the chemical shift tensor parameters.For example at 21.1 T even the small values of dr

(�100 ppm) can be calculated accurately for compoundswith CQ � 20 MHz, as discussed below.

The SSTMAS technique can be illustrated with anexample of 51V NMR in LaVO4, where the second-orderquadrupolar interaction has to be taken into account.Experimental 51V MAS spectra recorded for LaVO4 areshown in Fig. 9. To perform a SATRAS analysis of thisspectrum it is necessary first, to measure and plot inte-grated intensities from a hundred or so MAS spinning side-bands. Next, computer modeling needs to be performed tofit the experimentally measured intensities. In the SSTMAStechnique, just a single line with a pronounced quadrupolarshape is often sufficient for a complete analysis. As shownin Fig. 9 by a dotted line, the fit of the central line obtainedwith the SSTMAS technique has been obtained by takinginto account not only the quadrupolar coupling andthe magnetic shielding tensors, but also their relativeorientation.

Also shown in Fig. 9 are the calculated spectra of thecentral line with fixed parameters of the quadrupolar andchemical shielding tensors including their relative orienta-tion. Acute sensitivity of the line shape of selected transi-tions in MAS spectra to variations in NMR parametersmakes it possible to apply this technique in cases when fullMAS spectra suitable for SATRAS analysis are not readilyavailable.

3.5. MAS and static spectra analysis (MASSA)

SATRAS or even SSTMAS analysis may be com-pletely impossible for amorphous or disordered systemswhere lines are often considerably broadened and it isimpossible to obtain the full NMR spectrum with allthe satellite transitions present. In such cases only thecentral transition can normally be observed. Facing thisproblem Shubin et al. [100] have developed a techniquebased on simultaneous analysis of several MAS spectrarecorded with different spinning speeds and the staticspectrum. In some experimental situations this approach,called magic-angle spinning and static spectra analysis, orMASSA, allows one to obtain the full set of NMRparameters for quadrupolar nuclei with reasonableaccuracy.

In contrast to analyzing only the intensities of MASspinning sidebands as in SATRAS, the MASSA techniqueemphasizes the importance of the complete lineshape anal-ysis involving both MAS and static spectra. In this case not

, ppm-8000-6000-4000-20000200040006000

=

-640-620-600-580

A

B

-600 -610 -620 -630

C

1

2

3

ppm ppm

δ

Fig. 9. 51V MAS NMR spectra of LaVO4 at 9.4 T, mr = 10 kHz. (A) Full experimental spectrum. (B) The isotropic line (solid) shown with the simulation(dotted) calculated with the following parameters: CQ = 5 MHz, gQ = 0.4, dr = 200 ppm, gr = 0.7, a = 20�, b = 20�, c = 0�. (C) Simulated isotropic line atdifferent values of dr and b, (1) dr = �172 ppm and b = 88�, (2) dr = �272 ppm and b = 88�, (3) dr = �272 ppm, b = 8�.

0 -500 -1000 -1500

1

, ppm

2

3

4

5

6

δ

Fig. 10. Experimental and calculated 51V NMR spectra of a VOx/TiO2

catalyst (20 wt% V2O5) at 9.4 T. (1) Simulated static spectrum. (2)Experimental static spin–echo spectrum. (3) Simulated 12.2 kHz MASspectrum. (4) Experimental 12.2 kHz MAS spectrum. (5) Simulated14.1 kHz MAS spectrum. (6) Experimental 14.1 kHz MAS spectrum.Experimental static and 12.2 kHz MAS spectra, (2 and 4), were simulta-neously used in the optimization procedure. The final simulation of allthree calculated spectra, (1, 3, and 5), was calculated using a single set ofNMR parameters as following: CQ = 14.7 MHz, gQ = 0.59, dr = 650 ppm,gr = 0.02, diso = �611 ppm, d1 = �281 ppm, d2 = �292 ppm, d3 = �1261ppm, a = 20�, b = 62�, c = 42�.


only singularities in the line shape are taken into account,but the complete MAS and static lineshapes are simulta-neously computed. Another advantage of the MASSAtechnique is its applicability in situations when the CSAand second-order quadrupolar effects are comparable.However, the MASSA technique still remains computa-tionally intensive.

An example of MASSA analysis is presented inFig. 10 for a VOx/TiO2 catalyst (20 wt% V2O5) [100].This catalyst was prepared by the spray drying technique[101]. In such catalysts vanadium is present on the sur-face in the form of strongly bound vanadium (SBV), asdiscussed below in Section 9.1.3. 51V MAS spectrarecorded for this sample are extremely complex(Fig. 10) and very difficult to analyze using traditionalapproaches. Since only the central transition wasobserved, neither SATRAS nor SSTMAS techniqueswere applicable. Any attempts to represent experimentalMAS spectra recorded at different spinning speeds by asingle set of several overlapping resonances also failed.This suggested that in this case the complex lineshapeof MAS spinning sidebands was determined to a largeextent by the second-order quadrupolar effects. The mag-nitudes and the relative orientation of 51V quadrupolarcoupling and CSA tensors could be determined only byfitting NMR parameters using simultaneously static andMAS spectra. A good agreement was observed betweenall experimental and simulated spectra fitted with a singleset of parameters as illustrated in Fig. 10. This exampleshows that inclusion of a static spectrum into the optimi-zation procedure is very desirable.


3.6. Multi-quantum MAS (MQMAS) and satellite

transition MAS (STMAS)

The multi-quantum MAS technique is the most suitablefor quadrupolar nuclei with small magnetic shieldinganisotropy and pronounced second-order quadrupolareffects [102,103], as often observed, for example, for 27Al.However, for 51V, and for 93Nb in lesser extent, the mag-netic shielding anisotropy can be very significant. In suchcases the MQMAS technique requires more accurate set-ting up, with carefully chosen pulse lengths and delays.

The most popular 3QMAS pulse sequence is composedof three pulses instead of two pulses in the originally pro-

π2

-1

p1 p3 p4

p

t1 t2z

+3

-3

0

A

π2

-1

p1 p3 p4

p

t1 t2z

-1

0+1

B

ππ 2

p1 p3 p4

t1 t2z1

-1

-10

-2

p2

+1 +2

C

π π π2τ

τ

τ

τ τ

ττ

τ

τ

τ τ τa

21

D

1

m

2 2 a d

2n r

Fig. 11. Typical pulse sequences for high-resolution 2D NMR of quadrupolar nSEDOR (low-resolution technique for static samples), (F) REDOR, (G) Texcitation, p2 CT-selective, p3 z-filter mixing, and p4 detection pulse. Reprodu

posed sequence, where the third p/2 central transition selec-tive pulse is added to the first two pulses. This pulsesequence is shown in Fig. 11A from Ref. [104]. The thirdselective pulse is used to filter out all satellite transitionsthat are being excited by the second pulse, and allowingthrough only the central transition, using this so-calledzero-quantum filter.

For example, the optimal excitation of the triple-quan-tum coherence for spins I = 7/2 and 9/2 nuclei can beachieved by applying 120� and 90� rf pulses [105]. This isillustrated in Fig. 12A, where the theoretical and experi-mental excitation profiles are presented depending on thepulse length for a case of mrf/mQ = 1.25, 2.5, 5. All of them

π

τ

τ

τ

τ τ

τ τ

ππ2E

I

Sn

23

π π2

F

I

Srotor

0 1 2 3 4

π

π π

ψ ψ

2G

I

S

H π2

I

Srotor

0 5 10

x y x y y x x y x y x y y x x y

τ

uclei. (A) MQ(3Q)MAS, (B) STMAS, (C) DQ STMAS, (D) QCPMG, (E)RAPDOR, (H) REAPDOR. Radio frequency pulses are defined as p1

ced with permission from Refs. [81,104,115].

0.0

0.5

1.5

1.0

2.0 7/2

0.0

0.5

1.5

1.0

2.09/2

0 60 120 180 240 300 360

1(o)

3/2

, 3/

2

±

±

0.1

0.0

0.2

0.3 7/2

0.1

0.0

0.2

0.3

0 20 40 60 80 100 120 140

1/2

, 1/

2

±

± 9/2

θ 2(o)θ

ρ ρ

Fig. 12. Optimization of MQMAS experiments for spin-7/2 and spin-9/2 nuclei. (A) Buildup curves for generation of triple-quantum coherence as afunction of the pulse angle h1. (B) Dependence of triple- to single-quantum coherence transfer processes on the rf pulse angle h2. Investigations focused onthe 0 fi +3(t1) fi �1(t2) transfer pathway, and employed initial excitation pulses h1 = 120� for I = 7/2 and h1 = 90� for I = 9/2. Computation assumed thefollowing ratios between mQ and mrf (gQ = 0): mq/mrf = 5 (-ÆÆ-),mQ/mrf = 2.5 (- - -), mQ/mrf = 1.25 (–). Experimental data (s) collected for the latter ratio withK59

3 CoðCNÞ5 (I = 7/2) and Li93NbO3 (I = 9/2) samples at 9.4 T, and subsequently normalized to match the theoretical profiles. Reproduced withpermission from Ref. [105].


are normalized to the intensity of the central +1/2 M � 1/2transition, i.e. the maximum intensity of the +3/2 M �3/2satellite transition is 50% greater that the central transition.

While performing MQMAS experiments, it is importantto achieve the most complete conversion of the triple-quan-tum coherence, evolving during t1 time, into detectedsingle-quantum coherence. The final intensity of the sin-gle-quantum coherence as a function of the conversionpulse is shown in Fig. 12B, with the optimal conversionpulse found to be 45� for I = 7/2 and 35� for I = 9/2.Because the intensity of the observed signal is now threetimes lower than the intensity of the signal from the centraltransition alone, the difference in sensitivity needs to becompensated by increasing the number of accumulatedscans by 32 times. The relaxation time of the triple-quan-tum coherence is proportional to the coherence order, inthis case by a factor of 3. If this is also taken into accountfor each increment in t1 evolution time, then 33 = 27 timesthe number of scans need to be accumulated in order toobtain the same signal-to-noise level as in a simple one-dimensional single-pulse spectrum. This should be furthermultiplied by the number of increments in the t1 direction:at least 32 or more increments are usually required.

It is clear, that two-dimensional MQMAS spectrarequire long recording times. Attempts to cut acquisitiontimes by decreasing the number of scans would usuallyresult in a poor signal-to-noise ratio. A simple estimateshows that when a simple one-dimensional spectrum canbe acquired in as fast as 60 s, MQMAS for the same com-pound may require several hours. This makes MQMASnot very feasible for samples with less than 10 wt% or soof vanadium or niobium content, which is typical for solidcatalysts. All this renders the use of MQMAS very limitedin studying catalytic systems.

However, MQMAS can still be quite useful, particularlyin cases where there are several sites with close isotropicchemical shifts. As an example, a 51V 3QMAS spectrumrecorded for AlVO4 is shown in Fig. 13A. In this com-pound there are three non-equivalent vanadium sites.These three sites are completely resolved in 3QMAS spec-tra, and the quadrupolar coupling parameters can also beobtained for each site. Analysis of the one-dimensionalNMR spectrum for this compound is complicated by thesignals being strongly overlapping. The MASSA techniqueis not very efficient in this case due to relatively weak sec-ond-order quadrupolar effects, and SATRAS is compli-cated by over 120 individual spinning sidebands needingto be resolved, further attributed to each particular transi-tion, and finally integrated.

Recently a two-pulse MQMAS sequence has been sug-gested for measuring CSA parameters when the magneticshielding is small and cannot be determined from one-dimensional spectra with enough precision [106]. Indeed,it is very difficult to accurately measure the 51V chemicalshift anisotropy when it is less than 100 ppm, since at thismagnitude the dipolar interactions often conceal any mag-netic shielding effects. Taken without any additionaltransformations, a 3Q projection of the 3QMAS spectrumis effectively a spectrum recorded at three times the spec-trometers magnetic field. Because the CSA interactionsmeasured in parts per million do not depend on the mag-netic field strength while the dipolar interactions areinversely proportional to the magnetic field on the partsper million scale, the number of spinning sidebands inthe 3Q dimension is increased three times at the samespinning speed. This improves accuracy in determiningthe chemical shift anisotropy as well as the asymmetryparameter.

-800-750-700-650

-800

-750

-700

-650

V1

V2

V3

-900 -1000 -1100 -1200

-400

-800

-1200

-1600

Nb1

A B

Nb1

DCT

ST1

ST2

δid 93Nb, ppmδid

51V, ppm

δ 51V, ppm

δid51V, ppm δid

93Nb, ppm

δ 93Nb, ppm

-580

-600

-620

-640

-540

-560

-590 -600 -610 -620δ 51V, ppm

-950

-1000

-1050

-1150 -1200 -1250 -1300

δ 93Nb, ppm

C

Fig. 13. (A) 51V 3QMAS sheared spectrum of AlVO4 obtained at 9.4 T. (B) 93Nb 3QMAS sheared spectrum of BiNbO4 obtained at 9.4 T. (C) 51V STMASsheared spectrum of LaVO4 showing different correlations. CT is the autocorrelation of the central transition, ST1 correlates the central transition with the±1/2 M ±3/2 satellite transition, and ST2 correlates the central transition with the ±3/2 M ±5/2 satellite transition. The experiment has been recorded at9.4 T using 10 kHz MAS. (D) 93Nb DQ STMAS spectrum of Te3Nb2O11 recorded at 21.1 T using 20 kHz MAS.


93Nb MQMAS spectra are usually not complicated byCSA. At the same time, the second-order quadrupolareffects are very pronounced, thus simplifying the experi-mental setup compared with 51V [50]. Even at lower spin-ning speeds the spectral resolution in MQMAS spectracan sometimes be improved because of using the seconddimension that spreads out overlapping resonances andMAS spinning sidebands. An example of 93Nb 3QMASNMR is shown in Fig. 13B for BiNbO4.

To summarize, the main advantage of the MQMAStechnique, although being inherently of low sensitivityand time-consuming, is increased spectral resolution, andoffering the possibility to determine the isotropic chemicalshifts and the quadrupolar coupling parameters in astraightforward simplified manner.

The satellite transition MAS (STMAS) [107,108] tech-nique has been proposed as a method complementary toMQMAS. STMAS provides better sensitivity and is lessdependent on CSA and the strength of quadrupolar inter-actions. As a result the sites with low or high quadrupolarcoupling constants should be easier to observe in STMASthan in MQMAS. The STMAS has certain similarities withboth MQMAS and SATRAS experiments as it correlatesin a two-dimensional spectrum the central transition with

the satellite transitions (m,m � 1). In fact the STMASexperiment can be described as a two-dimensionalSATRAS experiment. The STMAS experiment requires avery accurate setting of the magic angle which is also sim-ilar to SATRAS [108]. A typical STMAS pulse sequence isshown in Fig. 11B.

The first 51V STMAS experiment of LaVO4 was pub-lished in [45]. Three signals were observed, correspondingto the central transition signal (CT), the ±1/2 M ±3/2satellite transition signal (ST1), and the ±3/2 M ±5/2 satel-lite transition signal (ST2) (Fig. 13C). The composite quad-rupolar coupling constant, k ¼ CQð1þ g2

Q=3Þ1=2, and theisotropic chemical shift, diso, extracted from the positionof the center of gravity of the ST1 line (k = 6.2 MHz,diso = �605 ppm), and those obtained from the frequenciesof the center of gravity of ST2 (k = 6.0 MHz, diso =�600 ppm) agreed well with SATRAS and MQMAS data[45]. However, this multiplicity of each signal may compli-cate interpretation of the spectra, especially when severalsites are involved.

Fig. 11C shows a double-quantum STMAS sequence,DQ STMAS. A simple modification of the standardSTMAS sequence by placing a selective p pulse beforethe t1 period correlates the double-quantum satellite transi-

δ(51V), ppm

-600 - 620 - 640

0

-5

δ(1 H

), p

pm

Fig. 14. 51V–1H HETCOR NMR spectrum of Ba(VO3)2ÆH2O obtainedusing 7 kHz MAS at 9.4 T.


tions with the central transitions, DQ±1 fi CT. This corre-lation also refocuses the anisotropic second-order quadru-polar interactions yielding ridge-shaped resonances in the2D spectra. The double-quantum STMAS spectra are freeof diagonal and outer satellite transition peaks similar as infiltered single-quantum spectra.

The rotor synchronization is important for efficient fil-tering of the diagonal and outer satellite transition peaksin DQ STMAS spectra [66,67,104]. The undesired signalsfrom CT and ST±2 can leak through the double-quantumfilter because the soft p pulse (p2, p4 in Fig. 11C) can inducesome coherence transfer to double-quantum despite theirlow efficiencies. The central transition has a zero first-orderquadrupolar shift and the outer satellite transitions ST±2

have a first-order shift twice that of ST±1. Therefore thesecoherences are not refocused at the mixing pulse if the t1

evolution time is carefully chosen for the rotor-synchroni-zation of the first-satellite transition coherence. The com-bined effects from low coherence transfer efficiencies androtor-de-synchronization of these leak-through coherencesexplain the superb performance on the suppression of theCT and ST±2 peaks in DQ STMAS, as illustrated inFig. 13D for Te3Nb2O11.

3.7. Heteronuclear correlation spectroscopy (HETCOR)

HETCOR spectroscopy is a classic two-dimensionalNMR correlation technique developed to probe short-range correlations and distances between heteronuclei,and is very popular in liquid-state NMR spectroscopy.The recent development of the necessary hardware hasmade HETCOR also useful in solid-state NMR [109–112], including research in vanadia and niobia based cata-lysts. Potentially HETCOR may become an important toolin the solid-state NMR practice.

The most popular HETCOR approach in solid state isbased on cross-polarization (CP). Even though one-dimen-sional CP/MAS is not always successful in measuring corre-lations between protons and quadrupolar nuclei due to veryfast relaxation of the latter, HETCOR allows these correla-tions to be measured, as for example, for 51V–1H pairs.

For spin one-half nuclei CP matching conditions aredescribed by the Hartmann–Hahn equation, cHB1H =cXB1X, [113], where B1H and B1X are the rf fields of thetwo nuclei during the contact time. A more general Hart-mann–Hahn rule applicable to quadrupolar nuclei requiresthe nutation frequency of the quadrupolar nucleus tomatch the effective proton rf field: cHB1H = xnut. Thequadrupolar nutation frequency, xnut, depends on manyparameters, including the quadrupolar coupling constantand the asymmetry of the quadrupolar coupling tensor.All this complicates dramatically the optimization of CPmatching conditions, which becomes even worse whenthere are several non-equivalent quadrupolar sites with dif-ferent nutation frequencies in the system.

Another problem while using HETCOR arises due tovery short relaxation times caused by the quadrupolar

relaxation mechanism, which result in mixing signals fromnuclei connected to protons and nuclei not connected toprotons. This complication is somewhat circumvented byusing complex cycling pulse sequences, which are oftenrotor-synchronized.

An example of a 51V–1H HETCOR experiment is shownin Fig. 14 for Ba(VO3)2ÆH2O. According to the crystalstructure, there are two non-equivalent vanadium sites inthis compound with similar tetragonal pyramid oxygenenvironments. However, vanadium–proton distances forthe two sites are quite different, 2.005 and 2.727 A. Thismay explain why the efficiency of the 1H to 51V magnetiza-tion transfer is also different for two sites. A simpleone-dimensional 51V MAS NMR spectrum of Ba(VO3)2ÆH2O has two lines corresponding to two vanadium sites.These one-dimensional spectra were used to optimize theHETCOR experimental conditions for this compound(Fig. 14).

It is also interesting to note, that although the one-dimensional 1H MAS NMR spectrum has only a singlepeak, the HETCOR spectrum contains a second compo-nent of lower intensity from H–V interactions correspond-ing to the second vanadium site. These signals were simplyoverlapping in the 1H MAS NMR spectrum.

One of the advantages of the 51V–1H HETCOR NMRtechnique is the possibility of studying V–H bonding, aswell as increased spectral resolution due to the seconddimension. As for many other two-dimensional techniques,HETCOR suffers from low sensitivity, thus requiringlonger acquisition times.

3.8. Double resonance experiments (SEDOR, REDOR,

TRAPDOR, and REAPDOR)

Kaplan and Hahn [114] proposed a spin–echo doubleresonance (SEDOR) pulse sequence to probe connectivityand interatomic distances between adjacent nuclei in solids.A fixed-time variation of the SEDOR pulse sequence isshown in Fig. 11E [115]. SEDOR experiments areperformed on stationary samples and as such have only


limited use in 51V NMR spectroscopy of solids due to lackof spectral resolution.

Follstaedt and Slichter [116] applied 51V–63Cu SEDORto identify 51V impurities dissolved in metallic copper.They were able to resolve two 51V resonances in CuV solidsolutions belonging to isolated impurities in copper andundissolved vanadium metal remaining in samples.SEDOR data were also useful for identification host 63Cunuclei adjacent to the vanadium impurity.

The rotational echo double resonance (REDOR) exper-iments are similar to SEDOR, except that the spin I echoesare now synchronized with magic-angle spinning (Fig. 11F)[117]. Brown et al. [118] have used the REDOR techniquefor a 51V/15N pair to study phenylamide groups bonded toVOx species on a silica surface. REDOR has allowed thedetermination of a 51V–15N dipolar coupling constantand hence the bond length. The efficiency of the polariza-tion transfer was increased by additional 15N/1H crosspolarization. These and similar REDOR experimentsrequire triple-channel NMR probes.

Fig. 15. 29Si CP/MAS spectrum of [AsW9O33(t-BuSiO)3(VO)]3� and the corrgreen, Si). Dashed inset shows isotropic lines without {51V} decoupling. Fusidebands are marked with asterisk. Reproduced with permission from Ref. [1

Other examples of REDOR involving 51V have beenreported by Hudalla et al. [119]. They studied 31P–{51V}and 51V–{31P} REDOR spectra of ZrV2�xPx O7 solid solu-tions applying two-pulse sequences with two and four rota-tion cycles. The authors determined the V–P bond length inthis system to be 3.42 A.

TRAPDOR (transfer of populations in double reso-nance), another variety of the double resonance NMRtechnique, has been proposed by Grey et al. [120,121].TRAPDOR involves nuclear polarization transfer similarto CP/MAS, however, the transfer is now performed underadiabatic conditions (Fig. 11G). Kim and Grey [122] used51V–17O TRAPDOR NMR to confirm the assignment ofresonances in the disordered anionic conductors a-Bi4V2O11 and c-Bi4V1.7Ti0.3O10.85. TRAPDOR has alsobeen shown to be sensitive to probe mobility involvingoxide ion hops between equivalent crystallographic sites.

The pulse sequence for a REAPDOR (rotational echoadiabatic passage double resonance) experiment is shownin Fig. 11H [123]. This experiment combines REDOR

esponding ORTEP view (orange, V; gray, As; blue, W; red, O; black, C;ll line inset shows isotropic lines with CW {51V} decoupling. Spinning25].


and TRAPDOR approaches, but is less susceptible toerrors when conditions of a perfect adiabatic passage arenot met. Huang et al. [124] have demonstrated that theREAPDOR spectroscopy is applicable to distancemeasurements between spin-1/2 and -7/2 pairs, i.e. 31P– 51V.Using REAPDOR they measured P–V distances in mono-vanadium substituted K4PVW11O40, 1-K7P2VW17O62, and4-K7P2VW17O62 polyoxoanionic salts. Numerical simula-tions of the experimental NMR data yielded very goodagreement with the averaged P–W/P–V distances deter-mined from the X-ray diffraction measurements in the sameor related compounds.

3.9. Triple Resonance experiments

Only a limited number of triple resonance experimentsinvolving 51V or 93Nb have been reported so far. A seriesof 51V/29Si/1H, 31P/13C/1H, and 29Si/13C/1H experimentswere performed by Bonhomme et al. [125] to study a[AsW9O33(t-BuSiO)3(VO)](n-Bu4N)3 complex. In theirexperiment they detected 29Si magnetization decay, whilethe population transfer was performed from 1H via 51V.This approach resulted in dramatically improved spectralresolution, allowing direct observation of 51V–29Si splittingin the 29Si spectra from a scalar coupling with the valueJ(51V–29Si) = 28 Hz (Fig. 15). We note here that the two-bond heteronuclear J-coupling in 93Nb–29Si was firstobserved by Kao and co-authors [126] using simple one-pulse experiments for 93Nb and 29Si. They reportedJ(93Nb–29Si) = 64 Hz in Rb4(NbO)2(Si8O21).

3.10. Advantages of high magnetic field strengths

The availability of NMR instruments operating at mag-netic fields exceeding 18–20 T is increasingly attractive forsolid-state NMR research. Besides providing improvedsensitivity, there are many other advantages of using ultra-high magnetic fields for studying half-integer-spin quadru-polar nuclei [127,128] such as 51V, 93Nb and 181Ta.

Considering quadrupolar coupling interactions, understationary conditions the full width of the central transition

-400-300-200-100

1

2

kHz

A

10

Fig. 16. Effects of the magnetic field strength on the 93Nb NMR spectra recor(A) Spectra scaled in kilohertz. (B) Spectra scaled in parts per million.

line shape, DmCT, broadened due to second-order quadru-polar interactions is inversely proportional to the reso-nance frequency, m0, as

DmCTðHzÞ / C2Q=m0 ð8Þ

Performing experiments at higher magnetic fields resultsin proportionally narrower static line widths in caseswhere this type of broadening is dominant. For example,the full central transition linewidth in the 93Nb spectrumrecorded for a stationary La3NbO7 sample at 9.4 T(97.8 MHz) is about 225 kHz. When the same spectrumis recorded at 21.1 T (220.0 MHz) this line is now onlyabout 100 kHz wide (Fig. 16A), i.e. the scaling factoris 2.25, very close to what is predicted from the corre-sponding magnetic field values. If the two spectra arecompared on the parts per million scale, the narrowingat higher field becomes even more dramatic, since onthe parts per million scale the second-order quadrupolarlinewidth of the central transition is inversely propor-tional to the square of the resonance frequency, i.e. thescaling factor in this case is 5.1 (Fig. 16B).

This type of narrowing at high magnetic fields simplifiesconsiderably the recording of the static spectra of centraltransitions affected by the second-order perturbations.Narrower static lines at higher magnetic fields offer bettersensitivity and require much less rf power for homogeneousexcitation. Spectra can be obtained much faster than atlower fields, and in fewer steps if acquired using astepped-frequency acquisition technique. Even the largestquadrupolar coupling constants, earlier accessible onlyvia nuclear quadrupolar resonance NQR, can now bedirectly measured with solid-state NMR.

We note, however, that in contrast to the second-orderbroadening, the first-order quadrupolar effects do not scalewith the resonance frequency, i.e. the full span of the satel-lite transitions will still cover the same spectral width mea-sured in Hertz regardless of the magnetic field strength.

As in stationary samples, under MAS conditions thelinewidth of the central transition is also inversely propor-tional to the resonance frequency. This is illustrated inFig. 17 where two 93Nb MAS NMR spectra are compared

-3000-2000-1000000

δ, ppm

1

2

B

ded at 9.4 T (above) and 21.1 T (below) for a stationary La3NbO7 sample.

-300-250-200-150-100-50050

9.4 T

21.1 T

kHz

Fig. 17. Effects of the magnetic field strength on 30 kHz MAS 93Nb NMRspectra of Te3Nb2O11 recorded at 9.4 T (below) and 21.1 T (above). 0 -500 -1000 -1500

1

2

3

4

δ, ppm

5

Fig. 18. 51V NMR spectra of a 2.5% V2O5/ZrO2 catalysts. (1) Staticspectrum at 9.4 T. (2) Simulated spectrum (1). (3) 35 kHz MAS spectrum at9.4 T. (4) 18 kHz MAS spectrum at 21.1 T. (5) Simulated spectrum of (4).

-660-640-620-600-580

-660-640-620-600-580

-1200-1000-800-600-400-2000δ, ppm

1

2

Fig. 19. 51V NMR static spectra of Ba3(VO4)2. (1) Recorded at 9.4 T.Inset shows the central transition. (2) Recorded at 21.1 T. Inset shows thecentral transition (upper spectrum) together with its simulation (lowerspectrum).


for Te3Nb2O11. Even though both spectra were recordedwith the same MAS spinning speed (30 kHz), the spectrumobtained at 21.1 T is considerably simplified comparingwith the spectrum recorded at 9.4 T.

This narrowing at higher magnetic fields makes MASexperiments feasible in systems where the quadrupolar cou-pling constants are quite large. Improved spectral resolu-tion in MAS spectra recorded at high fields becomes ofgreat importance when there are several sites present,which is very typical for complex multi-component sys-tems, including many advanced materials and catalysts,as will be discussed in more details below. Here we illus-trate this with 51V MAS NMR spectra obtained for a2.5% V2O5/ZrO2 sample at two magnetic fields, 9.4 and21.1 T (Fig. 18).

Even though, the spectrum at 9.4 T was recorded whenspinning the sample at 35 kHz the resolution in this spec-trum is much lower than in the 18 kHz MAS spectrumobtained at 21.1 T. In the latter spectrum up to five individ-ual sub-spectra can be resolved from five different types ofVOx species on the ZrO2 surface. It is clear that the NMRspectra recorded for these and similar systems at high andultrahigh magnetic fields contain more useful structuralinformation, which can be used to improve their propertiesand performance.

Important structural information can be obtained fromstudying magnetic shielding tensors as well. This is alsofacilitated at ultrahigh magnetic fields, since the chemicalshielding anisotropy measured in Hertz scales linearly withthe magnetic field strength. Very small chemical shieldinganisotropies become more apparent at higher magneticfields, even if at lower fields they are effectively concealedby other types of line broadening, including quadrupolaror dipolar interactions. For example in a static 51V NMRspectrum recorded for Ba3(VO4)2 at 9.4 T the central tran-sition line is rather broad and featureless mostly due to

quadrupolar broadening (Fig. 19). These broadeningeffects diminish at higher magnetic fields, and the samespectrum recorded at 21.1 T shows now the central transi-tion line typical for an axial anisotropy of the chemicalshielding tensor, which can be simulated, and the anisot-ropy parameters extracted.

The fact that the chemical shielding anisotropy scaleslinearly with the magnetic field is particularly importantfor 93Nb NMR, where even at reasonably high fields of7–14 T the static spectra are often still dominated by the


quadrupolar interactions. Only at ultrahigh magnetic fieldsdo the CSA effects become clearly visible in such spectraand can accurately be measured as will be demonstratedbelow. Ultrahigh field NMR spectroscopy is gaining inpopularity for studying CSA effects in a variety of quadru-polar nuclei [129–131].

On a cautionary note, the very same fact that the chem-ical shielding anisotropy measured in Hertz scales linearlywith the magnetic field may complicate recording MASspectra at very high magnetic fields. This is particularlyimportant to remember while performing 51V MASNMR experiments, where spectra are often dominated bythe magnetic shielding interactions. In such cases to achievethe same resolution on the parts per million scale in MASspectra it is often necessary at higher magnetic fields to spinsamples at much faster MAS spinning speeds. For exam-ple, to just reproduce on the parts per million scale a spin-ning sideband pattern originating from CSA in the 35 kHzMAS spectrum recorded at 9.4 T, it would require spinningthe same sample at 79 kHz at 21.1 T. This is not alwayspossible in practice.

Another complication arises when the line width is dom-inated by a chemical shift distribution. This situation is typ-ical for many glasses and amorphous materials. In such casesemploying ultrahigh magnetic fields does not necessarilyresult in improved spectral resolution. Because the chemicalshift distribution is not affected by the magnetic field strengthor the resonance frequency, the line width measured in partsper million remains the same regardless of the field. More-over, when measured in Hertz, the line dominated by thechemical shift distribution will actually be broader at higherfield. However, the spectral resolution may still improve withincreasing field if T2 values do not vary strongly as a functionof applied magnetic field strength.

Two other types of interactions to be mentioned here,dipolar and scalar J-coupling, are field-independent whenmeasured in Hertz, and in solid-state 51V and 93Nb NMRshould be studied at moderate magnetic fields. At ultrahighfields these two interactions are often concealed by broaderchemical shift distributions, particularly in materials whichare not of extremely high crystallinity.

Paramagnetic effects in 51V and 93Nb NMR spectra willbe discussed in more details below. Here we mention onlythat the frequency contributions to the magnetic shieldingfrom both the Fermi contact and pseudocontact interac-tions are directly proportional to the magnetic fieldstrength. At high magnetic fields this may cause severe linebroadening and even disappearance of some resonancesaffected by paramagnetic interactions. In paramagnetic sys-tems employing ultrahigh magnetic fields does not neces-sarily lead to better or improved spectra quality.

4. DFT and other quantum chemical computational

approaches

Theoretical computations of NMR parameters (CSAand EFG tensors) in transition metal compounds represent

a separate topic of quantum chemical research. We referthe reader to textbooks and introductory articles [132–134] and to a recent comprehensive review [135]. Chemicalshift and EFG computations by the density functional the-ory (DFT) approach are implemented in some popularquantum-chemical packages. DFT methods are now inthe most commonly used for the theoretical prediction ofNMR parameters for nuclei with high atomic number itwas concluded by Autschbach [135] that the Hartree–Fockapproach is, in general, not adequate, except maybe for thespecial case of high metal oxidation states of d10 systems.We also note that relativistic effects can have a seriousinfluence on calculated NMR parameters. Non-relativisticcalculations of the chemical shift in metals usually producereliable results only for 3d and 4d series elements, while for5d and f-block metals a proper treatment of relativisticeffects is mandatory. Even for 4d-block elements relativisticcorrections to the absolute shieldings can be significant, butthey usually have similar value for different computedstructures and, therefore, are almost canceled in relativechemical shift calculation being masked by the broaderrange of the relative chemical shifts. Even though justfew in number, all recent theoretical predictions of 51VNMR properties in solid state have also been performedwith the use of the DFT method.

Using two different basis sets Pooransingh et al.[136,137] and Ooms et al. [138] computed the quadrupolarand chemical shielding anisotropies by a DFT method forfour crystallographically characterized oxovanadium (V)complexes mimicking the active site of vanadium haloper-oxidases and hydroxylamido. The calculated tensors are ingeneral agreement with the experimental solid-state NMRdata. It was concluded that the combination of 51V solid-state NMR and computational methods is beneficial forinvestigating electrostatic and geometric environments indiamagnetic vanadium systems with moderate quadrupolaranisotropies.

Based on isolated and embedded cluster models Geereported [139] DFT calculations of the 51V EFG tensor ele-ments in orthovanadates to be in very good agreement withexperimental results. It was concluded that the local vana-dium–oxygen covalent bond structure had a more pro-nounced effect on the 51V EFG than did the rest of thecrystal. This is in agreement with earlier DFT calculationsbased on the linearized augmented plane-wave method forperiodic solids [140]. In this latter work, DFT calculationsprovided reliable assignment of 27Al and 51V resonances tospecific crystallographic sites in the asymmetric unit ofAlVO4. It was shown that the magnitude and orientationof the EFG tensors are largely determined by thep � p(27Al) and p � p, d � d(51V) orbital contributions tothe valence electrons, while the lattice itself provides onlyminor contributions for both nuclei.

Despite continuous development of quantum chemicalmethods, reliable calculations of shielding parameters fortransition metals have not been possible until recently.The approaches commonly used before were based on


finite, in a number of atoms, models using localizedatomic orbitals. In a recently published study [141] peri-odic DFT calculations of 51V NMR shielding parameterswere reported for the first time employing the gauge-including projected augmented-wave (GIPAW) pseudo-potential approach [142]. Using pseudopotentials specifi-cally constructed for vanadium Truflandier et al. [141]were able to perform accurate periodic calculations ofthe 51V shielding tensor in AlVO4 as well as to performEFG calculations. Analysis of the relative orientation ofthe EFG and shielding tensors has shown that these‘‘first-principles’’ calculations can indeed give access toEuler angles describing relative orientation of these twotensors.

Most recently an ultrasoft pseudopotential (USPP)modification of the GIPAW method has been applied tosystematically calculate 49Ti and 51V shielding parametersin a series of molecular and extended periodic systems[143]. Thirteen different vanadate compounds, with totalof 18 distinct vanadium sites, were computed, and a verygood correlation was found between theoretical and exper-imental values of isotropic chemical shift. Obtained dataallowed straightforward assignment of 51V resonances inAlVO4, a- and b-Mg2V2O7, and Ca2V2O7 compounds.Quite reasonable agreement between experimental and the-oretical CS anisotropy parameters was also obtained forortho- and pyrovanadates.

Table 4Different types of vanadium sites

5. 51V NMR data compilation

5.1. Chemical shielding and quadrupolar tensor parameters

in individual vanadium compounds

The vanadium coordination number in vanadium-con-taining oxide compounds varies from 4 to 6 with vanadiumfound in different VOx environments. For example, VO4

sites (x = 4) may exist either as tetrahedral sites (Td) or tri-gonal pyramids (Pd) (Table 4). Association of VOx polyhe-dra is often described as Qn, where n corresponds to thenumber of shared oxygen atoms. Thus tetrahedral Q0 sites(n = 0) are isolated species, while Q1 are dimers (n = 1),and Q2 (n = 2) correspond to chains of VO4. A similar sit-uation exists for VO5 and VO6 species. However, the levelof association of VO5 and VO6 polyhedra can be higher.For example, VO6 are isolated at n = 0, dimeric at n = 1and 2, form chains at n = 2,3,4, layers at n = 4, and poly-oxoanions at n = 6 (Table 4).

Precisely measured 51V NMR parameters are currentlyavailable for many individual binary vanadium oxide com-pounds of compositions V2O5–MxOy, where M is mono-,di-, tri-, tetra- or pentavalent metal, or V2O5–XzOk, whereX is phosphorus, arsenic or sulfur. Some of the ternaryvanadium compounds, i.e. V2 O5–MxOy–XzOk, and evenseveral V-containing biological systems, have also recentlybeen studied [93,94,144–149]. It is noteworthy, that almost


all types of vanadium sites shown in Table 4 can be foundin these individual compounds.

5.1.1. Tetrahedral Q0 sitesSites Q0 correspond to vanadium in tetrahedral oxygen

coordination where individual VO43� tetrahedra are iso-

lated from each other. This type of vanadium coordinationis typical for almost all known orthovanadates:

M+:

200

Fig. 20. 5

spectrum.CQ = 1.53(3–5) Sim±3/2 M ±

Li3VO4 [150], Na3VO4, K3VO4 [151], Cs3VO4

[152,153]; Tl3VO4.
M2+: Ca3(VO4)2 [154], Sr3(VO4)2 [155], Ba3(VO4)2
[156], Mg3(VO4)2, Zn3(VO4)2, Pb3(VO4)2 [157].M3+: AlVO4 [158], BiVO4, LaVO4 [159,160], LuVO4,

YVO4 [161], ScVO4 [161], CeVO4 [162], RhVO4.M5+: TaVO5, NbVO5 [163], VNb9O25, VTa9O25 [164].

In many of these compounds VO43� tetrahedra are

almost ideally symmetric with V–O distances in the rangeof 1.6–1.8 A, and all these distances usually within less then±0.1 A of the average V–O distance.

As an example, experimental and calculated 51V MASNMR spectra of Li3VO4 are presented in Fig. 20. The

0 1000 0 -1000 -2000 -3000δ, ppm

1

2

3

4

5

≈

≈

1V MAS NMR spectra of Li3VO4 at 9.4 T. (1) Experimental(2) Simulated spectrum with parameters: m0 = 105.2 MHz,

MHz, gQ = 0.05, dr = 12 ppm, gr = 0.8, diso = �544 ppm.ulated satellite transitions ±7/2 M ±5/2, ±5/2 M ±3/2, and1/2 correspondingly.

experimental spectrum obtained for this compound underspinning at 12.5 kHz has very high resolution, sufficientto detect and identify a minor, less than 1%, impurity ofLiVO3. The 51V MAS NMR spectrum of Li3VO4 repre-sents a well-resolved superposition of six individual sub-spectra from six satellite transitions, ±5/2 M ±7/2,±3/2 M ±5/2, and ±1/2 M ±3/2. These spectra can beanalyzed within the SATRAS approach as describedabove. CSA and quadrupolar parameters determined forthis compound from SATRAS analysis are summarizedin Table 5. The value of the quadrupolar coupling constantfound for this compound is rather small, CQ = 1.53 MHz,which explains why the second-order quadrupolar effectsare virtually undetectable in this case.

Similar spectra have been obtained and analyzed for allstudied orthovanadates. Based on observed 51V NMRparameters (Table 5) it is possible to conclude that vana-dium in regular Q0 tetrahedral oxygen environment hasalmost spherically symmetric magnetic shielding tensorwith small CSA anisotropy, dr < 100 ppm. The quadrupo-lar coupling constant, CQ, is found in the range of 1–6 MHz, while the asymmetry parameter of the quadrupolartensor varies from 0 to 1 depending on the compound[93,94,149].

5.1.2. Tetrahedral Q1 sites

Vanadium in tetrahedral oxygen coordination of Q1

type is sharing an oxygen atom with adjacent VO43� and

forming a V2O74� dimer with two VO3 units connected

via a bridging V–Ol2–V link. Sites Q1 are typical in pyro-vanadates of monovalent and divalent metals. Q1 tetrahe-dra in pyrovanadates are usually somewhat distortedcompared with more symmetric Q0 units in orthovana-dates. A slight distortion in Q1 is caused by the bridgingV–Ol2 bond, which is normally longer than the threeremaining V–Ot bonds.

In pyrovanadates of divalent metals two vanadiumatoms in V2O7

4� are usually non-equivalent. Dependingon the V2O7

4� structure two classes of pyrovanadates canbe distinguished. In pyrovanadates of the thortveitite type(structural analogs of Sc2Si2O7) the V–Ol2–V angle is at180�, i.e. the V2O7

4� dimer has D3d symmetry, as for exam-ple, is observed in a-Zn2V2O7 and Cd2V2O7. At the sametime in pyrovanadates of the dichromate type (structuralanalogs of K2Cr2O7) the V–Ol2–V angle is at about 140�and the V2O7

4� dimer has C2v symmetry, i.e. as found ina-Mg2V2O7, b-Mg2V2O7, Ca2V2O7, BaCaV2O7, and a-BaZnV2O7. 51V NMR parameters for many pyrovanadatesof divalent metals have been reported by Nielsen et al.[147].

For two thortveitite compounds, a-Zn2V2O7 andCd2V2O7, the values of CQ were determined as 3.9 and6.0 MHz, and the value of dr were found to be �119 and�173 ppm. For dichromate compounds CQ is found withina wider range of values from 1.5 to 10 MHz, while there aretwo general regions for dr, one is from �262 to �57 ppm,and the second is from 70 to 250 ppm.

Table 551V NMR parameters of individual vanadium compounds

Compound Ref. Site CQ

(MHz)

gQ dr

(ppm)

gr �diso

(ppm)

�d11

(ppm)

�d22

(ppm)

�d33

(ppm)

�d^

(ppm)

a/o b/o c/o

VO4 type Q0

Li3VO4 [34] 1.52 50 544

[22] 1.51

[149] 1.53 0.05 12 0.80 544.3 533 543 556

Na3VO4 [185] 545

[22] 1.05

K3VO4 [34] 100 560

Cs3VO4 [34] 56 0.82 576 520 580 626

[149] 3.50 0.29 30 0.18 572.9 555 561 603

Tl3VO4 [34] 30 480

Mg3(VO4)2 [145] 1.05 0.56 31 0.53 557.3 533 550 588 0 64 85

[34] 40 557

[186] 15 1 557 542 557 572

Ca3(VO4)2 [34] 100 615

[22] 2.05

Sr3(VO4)2 [34] 20 610

[186] �14 0.86 608(618) 631 619 604

[22] 0.53

[149] 0.53 0.01 10 0.50 610.9 603 608 620

Ba3(VO4)2 [34] 20 605

[22] 0.75

[186] 11 0.94 603(604) 593 604 616

[149] 0.76 0.01 5 0.27 602.9 600 601 608

Zn3(VO4)2 [145] 1.04 0.99 41 0.68 522.5 488 516 564 0 61 67

[185] 522

[186] 28 0.93 522(520) 493 519 548

Pb3(VO4)2 [186] 1.01a �32 0.41 486(498) 521 508 467

Mg2Sr(VO4)2 [186] �38 0.66 581(586) 617 592 548

Mg2Ba(VO4)2 [186] �37 0.16 575(584) 606 600 547

Mg2Pb(VO4)2 [186] �51 0.18 549(544) 574 565 493

AlVO4 [34] 1 62 0.16 668 630 640 730

2 55 1.00 747 605 745 800

3 75 0.40 780 710 800 830

[149] 1 3.50 0.70 90 0.30 662.0 604 631 752

2 3.30 0.70 90 0.30 743.0 685 712 833

3 4.50 0.70 90 0.30 776.0 718 745 866

[187] 1 4.05 0.84 87 0.74 660.5 585 649 748 114 29 88

2 2.35 0.93 �120 0.72 743.6 847 760 624 137 27 58

3 3.08 0.62 �82 0.88 775.7 853 781 694 130 25 15

ScVO4 [149] 3.95 0.01 40 0.83 674.9 638 672 715

YVO4 [34] 30 664

[22] 4.75 0 0

[149] 4.85 0.01 40 0.10 665.8 644 648 706

LaVO4 [34] 72 �0.85 609 555 616 657

[22] 5.21 0.69 0.69

[106] 5.20 0.68 50 0.71 604.7 562 597 655 165 77 36

[149] 5.21 0.69 72 0.09 612.1 573 579 684

CeVO4 [188] 5.62 0.21 177 0.01 427 338 339 604

LuVO4 [149] 4.28 0.01 5 0.43 663.4 660 662 668

BiVO4 [34] 80 0.63 420 355 405 500

[145] 4.94 0.36 94 0.32 421.1 359 389 515 2 90 30

[189] 4.82 0.41 138 0.099 416.8 341 354 555

[190] 5.04 0.4 97 0.7 428 345 414 525 0 66 169

RhVO4 [191] 605

LiCdVO4 [192] 3a �125 0 630

NbVO5 [145] 1.2 0.39 70 0.17 791.4 750 762 861 90 25 90

TaVO5 [145] 0.85 0.28 53 0.24 773.2 740 753 826 90 37 90

VNb9O25 [193] 3.95 0.12 100 0.01 602.0 552 553 702

VTa9O25 [193] 4.5 0.1 160 0.06 617 532 542 777

(continued on next page)


Table 5 (continued)


(MHz)

gQ dr

(ppm)

gr �diso

(ppm)

�d11

(ppm)

�d22

(ppm)

�d33

(ppm)

�d^

(ppm)

a/o b/o c/o

VO4 type Q1

Na4V2O7 [34] 1 560

2 575

K4V2O7 [34] 64 0.94 578 500 582 642

Cs4V2O7 [34] 1 543

2 567

Tl4V2O7 [34] 52 0.85 504 443 512 556

a-Mg2V2O7 [34] 1 30 0.50 555 510 570 585

2 83 0.12 617 570 580 700

[147] 1 3.29 0.69 �57 0.91 549.2 604 552 492 80 1 0

2 4.82 0.43 103 0.34 603.5 534 570 707 2 89 41

b-Mg2V2O7 [34] 1 120 0.83 560 460 560 680

2 50 0.80 650 590 660 700

[147] 1 10.10 0.44 �262 0.10 494.4 639 612 232 0 25 0

2 4.80 0.39 �113 0.90 639.3 747 645 526 36 4 17

[186] 1 �36 0.8 551(574) 611 571 542

2 �73 0.68 647(655) 717 667 582

Ca2V2O7 [34] 1 56 0.64 574 528 564 630

2 62 0.55 578 530 564 640

[147] 1 1.58 0.90 71 0.54 574.9 520 559 646 8 90 16

2 7.33 0.43 530 0.50 534.0 137 402 1064 28 90 69

[186] 62 0.68 575 523 565 637

Sr2V2O7 [34] 1 43 0.58 557 523 548 600

2 68 0.44 582 480 620 650

3 64 0.31 588 480 632 652

4 66 0.30 592 480 638 658

[149] 1 2.50 0.85 78 0.40 582.0 527 559 660

2,3,4 3.20 0.90 561.5

[186] 1,2,3,4 83 0.12 561,581 584,593 542 552 672

Ba2V2O7 [34] 1 73 0.34 579 530 555 652

2 27 0.04 588 574 575 615

3 40 0.38 600 535 625 640

[149] 1 2.04 0.95 �51 0.53 596.0 635 608 545

2 2.37 0.85 10 0.95 587.0 577 587 597

3 2.09 0.76 53 0.94 577.0 526 575 630

[186] 1,2,3 579,589,599

a-Zn2V2O7 [185] 120 0.67 620 500 640 720

[145] 3.97 0.54 �119 0.69 615 514 597 734 71 10 77

[147] 3.86 0.56 �119 0.62 616.6 713 639 498 128 0 90

[186] 117 0.65 620(612) 494 632 709

Cd2V2O7 [185] �193 0.00 563 370 660 660

[147] 6.00 0.41 �173 0.27 562.7 673 626 390 131 0 90

Pb2V2O7 [185] 110 0.45 522(510) 430 480 620

[186] 89 0.22 521(516) 462 482 606

BaCaV2O7 [147] 1 2.57 0.32 100 0.65 581.6 499 564 682 82 38 90

2 3.20 0.85 99 0.49 598.6 525 574 698 95 27 90

ZrV2O7 [34] 110 774 710 802 824

[149] 2.443 0.169 75 0.05 775.9 737 740 851

a-BaZnV2O7 [147] 1 3.81 0.58 143 0.10 650.1 571 586 793 2 36 90

2 5.91 0.86 252 0.16 608.4 462 503 860 132 17 50

VO4 type Q2

a-AgVO3 [194] 244 0.34 365 202 285 609

NH4VO3 [34] 2.98 0.28 234 0.64 572 380 530 807

[93] 2.91 0.3 237 0.71 563.7 361 529 801 75 23 34

[23] 2.88 0.3 250 0.28

[24] 2.95 0.19

[25] 2.76 0.37

[144] 2.95 0.3 240 0.7 569.5 366 534 810 78 23 32

LiVO3 [144] 3.18 0.87 221 0.7 573.4 386 540 794 154 9 4

[145] 3.25 0.88 228 0.72 573.2 377 541 801 105 21 72


Table 5 (continued)


(MHz)

gQ dr

(ppm)

gr �diso

(ppm)

�d11

(ppm)

�d22

(ppm)

�d33

(ppm)

�d^

(ppm)

a/o b/o c/o

a-NaVO3 [34] 247 0.65 582(573) 368 530 820

[23] 3.65 0.6 253 0.87

[185] 263 0.65 577 360 530 840

[144] 3.8 0.46 259 0.68 572.7 355 531 832 30 15 15

[145] 3.77 0.49 249 0.67 572.8 365 532 822 156 3 75

[189] 2.3 0.95 208 0.68 569.7 395 536 778

KVO3 [34] 308 0.63 547 294 490 856

[23] 4.36 0.75 227 0.88

[185] 313 0.64 553 300 500 870

[144] 4.2 0.80 290 0.65 552.7 313 502 843 90 0 31

[145] 4.15 0.85 307 0.66 547.6 293 495 855 102 8 25

RbVO3 [34] 302 0.65 570(561) 313 508 863

[26] 4.33 0.72 216 0.14

[145] 4.23 0.64 314 0.69 565.4 300 517 879 100 28 0

CsVO3 [34] 291 0.66 583(571) 330 522 863

[24] 3.92 0.62

[145] 4.1 0.48 314 0.67 577.4 315 526 891 112 21 16

TlVO3 [34] 265 0.72 528 300 490 793

[144] 3.67 0.71 265 0.76 529.1 296 497 794 114 21 4

[145] 3.76 0.71 267 0.76 529.1 294 497 796 106 30 9

VO5,6 axial symmetry

b-NaVO3 [144] 4.2 0.55 512 0.17 510.4 211 298 1022 255 177 48 9

[189] 2.4 0.92 477 0.02 508.4 265 275 985 270

(NH4)2V6O16 [195] 1 2.75 0.98 470 0.01 509.7 272 277 980 275

2 2.325 0.34 401 0.05 546.0 335 356 947 346

K2V6O16 [34] 432 0 503 290 290 935 290

[144] 1 2.45 0.44 405 0 548.1 346 346 953 346 � 65 90

2 3.03 0.89 459 0 510.1 281 281 969 281 � 61 56

Rb2V6O16 [34] 432 0.00 503 290 290 935 290

[149] 1 2.664 0.986 448 0.16 513.8 254 326 962 290

2 2.34 0.30 400 0.05 547.0 337 357 947 347

Cs2V6O16 [34] 445 0.00 508 296 296 953 296

[149] 1 1.25 0.42 460 0.03 513.0 276 290 973 283

2 1.68 0.69 400 0.05 575.0 365 385 975 375

Tl2V6O16 [34] 465 0.00 700 485 485 1165 485

a-Mg(VO3)2 [34] 373 0.43 577 310 470 950 –

[26] 6.79 0.63 220 0.00

[146] 7.50 0.34 310 0.30 533.9 332 425 844 – 0 52 0

Ca(VO3)2 [34] 509 0.15 571 278 355 1080 317

[24] 3.3 0.8

[19] 3.16 0.6

[26] 2.81 0.6

[146] 3.06 0.51 517 0.18 563.0 258 351 1080 305 74 86 35

[189] 2.94 0.67 530 0.13 556.3 257 326 1086 292

Ca(VO3)2Æ4H2O [146] 1 4.18 0.92 297 0.30 580.0 387 476 877 – 150 34 54

2 3.73 0.57 492 0.16 529.5 244 323 1022 284 144 50 88

a-Sr(VO3)2 [146] 1 4.22 0.12 218 0.32 639.1 495 565 857 – 128 86 43

2 5.65 0.31 244 0.61 585.9 389 538 830 – 56 6 54

Ba(VO3)2 [34] 249 0.30 660(701) 540 614 950 –

[146] 1 3.68 0.16 190 0.41 658.5 525 602 849 – 153 70 60

2 5.56 0.04 265 0.59 590.6 380 536 856 – 20 8 90

Zn(VO3)2 [185] 387 0.36 517(533) 270 410 920 –

[146] 6.86 0.40 333 0.02 493.8 324 331 827 328 0 53 2

[189] 4.84 0.27 256 0.49 491.4 301 426 747 –

Cd(VO3)2 [34] 330 0.18 500 305 365 830 335

a-Cd(VO3)2 [146] 1.70 1.00 484 0.15 521.5 243 316 1006 280 0 90 47

b-Cd(VO3)2 [146] 6.46 0.47 311 0.31 468.2 264 361 779 – 22 55 0

Pb(VO3)2 [185] 457 0.02 533(543) 310 320 1000 315

[146] 1 3.75 0.13 465 0.15 529.8 262 332 995 297 90 36 0

2 6.98 0.31 428 0.12 479.7 240 291 908 266 90 50 0

V2O5 [34] 640 0 610 310 310 1270 310

[93] 0.797 0.00 645 0.11 609.0 251 322 1254 287 42 126 –



Table 5 (continued)


(MHz)

gQ dr

(ppm)

gr �diso

(ppm)

�d11

(ppm)

�d22

(ppm)

�d33

(ppm)

�d^

(ppm)

a/o b/o c/o

[19] 0.805 0.04 593 0.20 90 142 180

[196] 0.811 0.04 620 0.15 609 252 340 1229 296 58 128 –

[46] 0.799 0.21 636 0.11 612 259 329 1248 294 140 127 145

V2O5ÆnH2O

xerogel

[148] 1 1.34 0.7 574 0.49 622.5 194 476 1197

1 0 1.33 0.45 611 0.51 618.5 157 469 1230

100 0.83 0.08 613 0.00 619.4 313 313 1233 313

1000 0.86 0.10 640 0.00 619.4 299 299 1259 299

2 1.39 0.60 787 0.62 593.2 �45 444 1380

2 0 1.71 1.00 667 0.30 592.6 159 359 1259

200 1.70 1.00 591 0.15 594.6 255 343 1186 299

2000 1.61 0.98 1056 0.81 597.2 �358 497 1653

3 0.76 0.94 730 0.53 663.6 105 492 1394

4 0.73 0.31 713 0.14 572.3 166 266 1285 216

V2O5Æ1.5H2O

xerogel

[197] 1 1.34 0.70 575 0.50 623 377 541 952

2,4 1.40 0.60 780 0.60 596 249 509 1029

3 1.40 0.94 460 0.53 662 463 601 923

5 0.73 0.30 600 0.14 576 358 412 958

VOAsO4 [34] 758 0.049 617(614) 218 255 1370 237

VOPO4 [34] 841 0.00 734(705) 285 285 1547 285

aI-VOPO4 [180] 1.55 0.55 820 0.00 691.0 281 281 1511 281 54 163 –

b-VOPO4 [180] 1.99 0.59 818 0.00 755.0 346 346 1573 346 6 43 –

[198] 1.45 0.44 818 0.05 735 306 346 1553 326 – 35 15

aII-VOPO4 [180] 0.825 0.52 582 0.67 776.0 287 680 1358 7 0 60

[198] 0.625 0.09 922 0.08 755 256 330 1680 293 0 3 0

c-VOPO4 [198] 1 0.554 0.68 955 0.15 755 206 349 1710 278 – 28 42

2 1.32 0.55 942 0.07 739 235 301 1681 268 – 81 77

K2V8O21 [34] 480 0.00 570 330 330 1050 330

KVO3ÆH2O [34] 604 0.00 606 305 305 1210 305

NaVO3ÆH2O [34] 500 0.00 530 280 280 1030 280

[25] 3.94 0.64

NaVO3Æ1.5H2O [197] 1 3.10 0.90 420 0.00 548 408 408 828 408

2 2.40 0.80 370 0.00 535 412 412 782 412

VO6 isolated

NH4VO(SO4)2 [34] 902 0.04 658 187 227 1560 207

NaVO(SO4)2 [34] 838 0.03 592 160 185 1430 173

KVO(SO4)2 [34] 900 0.04 650 180 220 1550 200

RbVO(SO4)2 [34] 888 0.04 662 198 235 1550 217

CsVO(SO4)2 [34] 859 0.05 671 220 263 1530 242

KVO(SeO4)2 [34] 850 0.05 630 185 226 1480 206

RbVO(SeO4)2 [34] 834 0.04 636 203 235 1470 219

CsVO(SeO4)2 [34] 795 0.04 575 160 195 1370 178

KVOSO4SeO4 [34] 870 0.03 630 180 210 1500 195

RbVOSO4SeO4 [34] 864 0.03 636 190 220 1500 205

VO6 dimer

K4(VO2)2(SO4)2S2O7 [34] 704 0.06 696 325 365 1400 345

Rb4(VO2)2(SO4)2S2O7 [34] 722 0.08 703 315 371 1425 343

Cs4(VO2)2(SO4)2S2O7 [34] 650 0.00 690 365 365 1340 365

[199] 1 2.97 0.74 641.5 0.07 671.5 326 374 1313 350

2 2.97 0.74 641.5 0.07 678.5 333 381 1320 357

VO6 associated

K3VO2(SO4)2 [34] 494 0.22 566 266 374 1060 –

Rb3VO2(SO4)2 [34] 626 0.15 562 201 297 1188 249

Cs3VO2(SO4)2 [34] 621 0.15 559 201 297 1180 249

Na3VO2SO4S2O7 [34] 737 0.03 578 200 220 1315 210

K3VO2SO4S2O7 [34] 660 0.13 600 230 315 1260 273

V2O3(SO4)2 [34] 778 0.11 632 204 287 1410 246


Table 5 (continued)


(MHz)

gQ dr

(ppm)

gr �diso

(ppm)

�d11

(ppm)

�d22

(ppm)

�d33

(ppm)

�d^

(ppm)

a/o b/o c/o

V2O3(SeO4)2 [34] 697 0.11 633 247 322 1330 285

KVO2SO4 [34] 664 0.05 696 350 380 1360 365

RbVO2SO4 [34] 709 0.09 711 325 390 1420 358

CsVO2SO4 [34] 466 0.12 544 284 340 1010 312

KVO2SeO4 [34] 520 0.00 540 280 280 1060 280

RbVO2SeO4 [34] 428 0.00 572 358 358 1000 358

CsVO2SeO4 [34] 465 0.19 535 258 348 1000 303

KVO2SO4Æ3H2O [34] 517 0.21 503 190 300 1020 –

RbVO2SO4Æ3H2O [34] 513 0.23 507 190 310 1020 –

KVO2SeO4Æ3H2O [34] 488 0.27 507 197 330 995 –

RbVO2SeO4Æ3H2O [34] 503 0.28 492 170 310 995 –

CsVO2SeO4Æ3H2O [34] 498 0.27 500 183 318 998 –

NaK2[(VO2)3(SO4)3]Æ5H2O [34] 460 0.00 490 260 260 950 260

NaRb2[(VO2)3(SO4)3]Æ5H2O [34] 446 0.00 494 270 270 940 270

NaCs2[(VO2)3(SO4)3]Æ5H2O [34] 447 0.00 503 280 280 950 280

VO6 isolated

[(n-C4H9)4N]3[VW5O19] [200] 0. 605 0.65 200 0.95 504.8 310 500 705 – 90 30 0

Cs3[VW5O19] [200] 1.30 0.8 466 0.10 519.6 263 310 986 287 70 0 40

K4PVW11O40 [201] 0.94 0.48 537 0.16 561.3 250 336 1097 293 – 64 26

[(n-C4H9)4N]4PVW11O40 [201] 1.13 0.61 521 0.13 543.0 249 317 1064 283 – 52 42

Cs4PVW11O40 [201] 1.25 0.53 589 0.25 563.9 196 343 1153 269 – 55 52

[NaxCs4�x]PVW11O40 [201] 1.15 0.15 590 0 562.4 267 267 1152 267 – 63 50

a1-4-K7P2VW17O62 [124] 1.73 0.37

a2-4-K7P2VW17O62 [124] 0.88 0.75

VO6 dimer

[(n-C4H9)4N]3H[V2W4O19] [200] 1.05 0.95 418 0.10 505.2 275 317 923 296 70 0 50

Na2Cs2[V2W4O19]Æ6H2O [200] 1.56 0.35 456 0.20 526.0 252 344 982 298 80 60 50

a-1,2-H3K2[PV2W10O40]ÆCH3OHÆ14H2O

[201] 1.79 0.81 516 0.07 542. 5 266 303 1059 285 – 62 36

a-1,2-[(n-C4H9)4N]5�PV2W10O4 [201] 1.75 0.77 549 0.01 557.0 280 285 1106 283 – 81 39

a-1,2-Cs5PV2W10O40 [201] 1.51 0.57 479 0.0 533.1 294 294 1012 294 – 63 42

VO6 three and more associated

a-1,2,3-K6PV3W9O40 [201] 3.93 0.87 371 0.0 522.1 337 337 893 337 – 0 38

a-1,2,3-[(n-C4H9)4N]6PV3W9O40

[201] 5.50 0.75 451 0.05 444.4 208 230 895 219 – 45 40

a-1,2,3-Cs6PV3W9O40 [201] 3.5 1.0 428 0.0 519.9 306 306 948 306 – 87 5

Cs4[H2V10O28]Æ4H2O [202] 1 3.0 0.9 400 0.20 518.0 278 358 918 318

2 2.5 0.7 600 0.70 531.0 21 441 1131 –

3 2.5 0.9 450 0.50 540.0 203 428 990 –

4 4.0 0.5 450 0.20 545.0 275 365 995 320

5 6.4 0.4 350 0.20 427.0 217 287 777 252

Yellow foam [202] 1 3.90 0.30 400 0.20 435.0 195 275 835 235

2 3.00 0.90 350 0.20 530.0 320 390 880 355

3 2.00 1.00 140 0.20 564.0 480 508 704 494

White foam [202] 1 2.00 1.00 100 0.20 564.0 504 524 664 514

2 2.50 0.70 200 0.30 578.0 448 508 778 478

Green foam [202] 3.00 0.90 350 0.50 525.0 263 438 875 –

K5NaV10O28Æ10H2O This work 1 3.70 0.90 300 0.10 424.0 259 289 724 274

2 2.20 0.99 345 0.07 488.0 303 328 833 316

K4Na2V10O28Æ10H2O This work 1 421.5

2 3.70 0.90 300 0.10 426.8 262 292 727 277

3 431.0

4 2.20 0.99 345 0.07 489.0 304 329 834 317



Table 5 (continued)


(MHz)

gQ dr

(ppm)

gr �diso

(ppm)

�d11

(ppm)

�d22

(ppm)

�d33

(ppm)

�d^

(ppm)

a/o b/o c/o

5 495.0

6 501.0

7 506.0

8 511.0

Rb4Na2V10O28Æ10H2O This work 1 3.70 0.90 300 0.10 424.0 259 289 724 274

2 430.0

3 435.0

4 2.20 0.99 345 0.07 489.0 304 329 834 317

5 493.0

6 497.0

7 505.0

8 511.0

(NH4)4Na2V10O28Æ10H2O This work 1 3.70 0.90 300 0.10 423.0 258 288 723 273

2 427.0

3 432.0

4 2.20 0.99 345 0.07 490.0 305 330 835 318

5 495.0

6 502.0

7 508.0

8 514.0

Cs2V4O11 [149] 1 1.25 0.42 620 0.06 510 184 222 1133 203 – 90 –

2 1.68 0.69 256 0.13 575;581 430 464 831 447

[34] 1 610 0.05 510 190 220 1120 205

2 253 0 575(572) 445 445 825 445

Bi4V2O11 [189] 1 4.0 0.98 48 0.20 498.6 470 480 546 475

2 4.5 0.96 489 0.08 422.2 158 197 911 178

[203] 510

K3V5O14 [149] 1 1.75 0.40 480 0.08 517.0 258 296 997 277

2 0.22 0.20 80 0.60 624.0 560 608 704 –

[34] 1 486 0.00 500(503) 260 260 990 260

2 620

Rb3V5O14 [34] 1 463 0.00 496(517) 285 285 980 285

2 619

Tl3V5O14 [34] 1 489 0.00 500(504) 260 260 993 260

2 594

Pb5(VO4)3Cl [185] 508

VO[SiO(OtBu)3]3 [149] 2.45 0.1 445 0.05 776 542 564 1221 553

VO[SiOPh3]3 [177] 414 0.04 736 510 525 1150 518

3.19 0.13 420 0.08 731.4 504 538 1151 521

[(c-C6H11)7(Si7O12)VO]2 [177] 3a 401 0.087 714 500 535 1115 518

VO[OAmt]3 [204] 107 0.15 681 619 636 788 628

VO[OSiMe3]3 [204] 204 0.05 714 607 617 918 612

(„SiO)3VO [177] 2.5a 480 0.042 710 450 470 1190 460

(„SiO)3VOÆ2H2O [177] 620 0.129 580 250 330 1200 290

a Values of CQ estimated in this work.


Negative values of dr, detected in some thortveitite com-pounds, as well as for one of the vanadium sites in a-Mg2V2O7, is explained by certain structural features ofthe local vanadium coordination. For example, in b-Mg2V2O7 besides four V–O bonds in VO4 tetrahedra thereis an additional V–O distance at 2.4–2.8 A, which can bedescribed as the fifth V–O bond.

According to Hawthorne and Calvo [165] in barium piro-vanadate Ba2V2O7 there are four structurally non-equivalenttypes of VO4 tetrahedra. Similar findings were also reportedfor isostructural Sr2V2O7 [166]. All four of these VO4 tetrahe-dra are distorted in a somewhat similar fashion, while threeV–Ot distances remain almost equal in length, a bridgingV–Ol2 bond forming a V2O7

4� dimer is usually longer by

about 0.1 A. This type of distortion results in 51V MASNMR spectra illustrated in Fig. 21 for Ba2V2O7. The MASspinning sidebands from the central and satellite transitionsextend over 10,000 ppm (1 MHz at 9.4 T). At the same timethe central transition is sufficiently well resolved to showthree individual lines with relative intensities of 1:2:1. Sinceaccording to the structural data there are four non-equivalentvanadium sites in the crystal structure, it is clear that two ofthese sites overlap in this spectrum. The 51V NMR parame-ters determined for the four sites are very close (Table 5).

Summarizing 51V NMR data available for Q1 vanadiumsites (Table 5), we can conclude that vanadium in slightlydistorted tetrahedral sites sharing one common oxygenatom with the adjacent tetrahedra (Q1 type) has larger

-620-600-580-560

δ, ppm-5000-4000-3000-2000-10000100020003000

1

2

δ, ppm

Fig. 21. 51V MAS NMR spectrum of Ba2V2O7 obtained at 9.4 T using 10 kHz MAS. (1) Full spectrum. (2) Isotropic region of the MAS spectrum showingthree isotropic lines.


CSA anisotropy, 100 < dr < 200 ppm, compared with Q0

sites; the quadrupolar coupling constant varies from 2.5to 10 MHz, while the CSA asymmetry parameter gr

changes from 0.1 to 0.9 [93,94,149].

5.1.3. Tetrahedral Q2 sitesTetrahedral sites of Q2 type are found in metavanadates

of ammonia, alkali metals and some other monovalentmetals, MVO3, (M+ = Li [167], Na [168], K and NH4

[169], Tl [170], Rb, and Cs [171]). In these compounds cor-ner-sharing VO4 tetrahedra form infinite two-dimensionalchains. As a result, the V–Ol2–V bonds are further elon-gated, thus causing additional distortions of VO4 tetrahe-dra. In many MVO3 compounds with Q2 sites 51V NMRparameters have been carefully characterized withSATRAS [144] and were found to fall into quite narrowranges. For example, the quadrupolar coupling constantin MVO3 is often within 2.8–4.3 MHz, while the chemicalshielding anisotropy is within the 217–314 ppm range.

The crystal structure of Ba(VO3)2 is similar to metavana-dates of alkali metals [172], where corner-sharing VO4 tetra-hedra form extended 2D-chains of Q2 units, although inBa(VO3)2 there are two non-equivalent vanadium sites.

Generally speaking, vanadium in strongly distorted Q2

tetrahedral sites with adjacent tetrahedra sharing two com-mon oxygen atoms has large values of anisotropy(200 < dr < 500 ppm); the quadrupolar coupling constantvaries from 2 to 7 MHz; and the CSA asymmetry gr rangesfrom 0.6 to 0.8 [93,94,149].

5.1.4. Associated non-axial VO5 and VO6 sitesAssociated non-axial VO5 and VO6 sites are typical for

some metavanadates of divalent metals M(VO3)2. For

example, distorted octahedral VO6 sites are found in com-pounds of the brannerite structural type (analogs ofThTi2O6) with space group C2/m, i.e. in Mg(VO3)2 [173],Zn(VO3)2 [174] and b-Cd(VO3)2 [175]. The low-tempera-ture modification a-Sr(VO3)2 (Pnma) contains two non-equivalent distorted VO6 sites [176]. 51V NMR parametersfor these compounds were reported in [146] and are sum-marized in Table 5. 51V NMR spectra are characterizedby CSA values similar to those for Q2 sites,(200 < dr < 400 ppm), however, the CSA asymmetryparameter gr is smaller that in Q2, 0.3 < gr < 0.6, but some-what higher than in pyramidal vanadium sites (see below).

5.1.5. Isolated and associated trigonal VO4 pyramids

Trigonal VO4 pyramids are usual in VO[SiO(OtBu)3]3,VO[SiOPh3]3, [(c-C6H11)7(Si7O12)VO]2, („SiO)3VO, andsimilar systems. Vanadium in these compounds is four-coordinated with one short V@O bond. When V–Ol2–Vbonding is absent or very weak, the pyramids are said tobe isolated. Vanadium occupying these sites has an axiallysymmetric chemical shielding tensor with a large anisot-ropy (400 < dr < 500 ppm), the quadrupolar coupling con-stant can vary from 1 to 4 MHz, while the CSA asymmetryparameter gr changes from 0 to 0.2, and d^ � �400 to�600 ppm [144–149,177].

Pyramidal association, i.e. as observed in Cs4V2O11,leads to smaller anisotropy mostly due to changes in dII,while d^ remains practically unchanged (at �450 ppm inCs4V2O11). The crystal structure of Cs2V4O11 has beenreported in [178]. There are three different vanadium sitesin this structure, including two types of trigonal VO4 pyra-mids, and one type of a trigonal VO5 bipyramid having dif-ferent degrees of association (Fig. 22). Corresponding 51V

-2500-1500-5005001500δ, ppm

-600-500

1

2

3

δ, ppm

V1

V2 V3

Fig. 22. 51V MAS NMR spectra of Cs2V4O11 obtained at 9.4 T. (1) Experimental 51V MAS spectrum. (2) Simulated spectrum corresponding to VO4 sites.(3) Difference spectrum corresponding to VO5 sites. Inset shows the isotropic region of the MAS spectrum. The crystal structure of Cs2V4O11 is given onthe left.


MAS NMR spectra for this compound are shown inFig. 22. Three isotropic lines can be easily identified inthese spectra, two narrower lines with close isotropic chem-ical shifts belonging to trigonal VO4 pyramids, and abroader line attributable to vanadium in distorted octahe-dral environment of trigonal VO5 bipyramids. 51V NMRparameters determined for all three sites are given inTable 5.

1000 0 -1δ, ppm

*

Fig. 23. 51V NMR spectra of aI-VOPO4 at 9.4 T. (1) Experimental static spectrucalculated with the following parameters, CQ = 1.55 MHz, gQ = 0.55, dr = 880above. The isotropic line is marked with asterisk.

5.1.6. Isolated octahedral VO6 and tetragonal VO5 pyramids51V NMR parameters for four of five known [179] crys-

talline modifications of VOPO4 are summarized in Table 5.In VOPO4 vanadium occupies distorted octahedral VO6

sites. Each of these octahedrons is composed of four oxy-gen atoms in equatorial positions and two axial oxygenatoms. One of two axial oxygens has a short V@O bond,therefore the vanadium coordination in such sites is

000 -2000 -3000

1

2

3

m. (2) Experimental 10 kHz MAS spectrum. (3) Simulated MAS spectrumppm, gr = 0, diso = �691 ppm. The building unit of the aI-VOPO4 is shown

-1500-1000-5000

A

BC

Fig. 24. 51V NMR spectra of Cs4(VO)2O(SO4)4 measured at 9.4 T. (A) 14 kHz MAS spectrum. (B) Isotropic region of the MAS spectrum. The isotropiclines are marked with asterisks. (C) Static spectrum.


described as octahedral pyramidal. Since there is no bridg-ing via V–Ol2–V corner sharing, individual VO6 octahedraare considered isolated. Coordination of vanadium sites inaI-VOPO4 together with the corresponding 51V NMR spec-tra is shown in Fig. 23 [180].

Similar isolated octahedral VO6 pyramids are also typi-cal for VOAsO4, and some other compounds includingMVO(SO4)2, (M+ = NH4, Na, K, Cs, Rb), MVO(SeO4)2

(M+ = K, Rb, Cs), and MVO(SO4)(SeO4) (M+ = K, Rb).

02000

δ, pp

1

2

A

ppm-600-500-400

**

C

Fig. 25. 51V MAS NMR spectra of Rb2V6O16 recorded at 9.4 T using 14 kHz Mtwo sub-spectra. (B1) First sub-spectrum calculated with CQ = 2.66 MHz,spectrum calculated with CQ = 2.34 MHz, gQ = 0.30, dr = 400 ppm, gr = 0.05,indicating isotropic lines for two vanadium sites.

Vanadium in isolated octahedral pyramids has an axiallysymmetric chemical shielding tensor with a large anisotropy(750 < dr < 900 ppm); at the same time the quadrupolar cou-pling constant is often less than 2 MHz, while gr changesfrom 0 to 0.2, and d^ � �180 to �250 ppm.

Association of octahedral pyramids is usually accompa-nied by some decrease in anisotropy, as was observed, forexample, for dimeric species in Cs4(VO)2O(SO4)4 wheredr is about 700 ppm (Fig. 24).

-4000-2000

m

-4000-200002000δ, ppm

B

1

2

AS. (A1) Experimental spectrum. (A2) Simulated spectrum composed ofgQ = 0.99, dr = 448 ppm, gr = 0.16, diso = �514 ppm. (B2) Second sub-diso = �547 ppm. (C) Isotropic region of the MAS spectrum with asterisks


5.1.7. Associated tetragonal pyramids

Association of tetragonal VO5 pyramids occurs inM2V6O16 (M = NH4, Cs, Rb, Tl). All these compoundsare isostructural with the P21/m space group [181]. Thereare two types of vanadium site in this structure occupyingtwo types of tetragonal pyramids in the ratio of 1:2. Eachpyramid is connected to two adjacent pyramids via edgeor corner sharing.

The 51V MAS NMR spectrum of Rb2V6O16 shown inFig. 25 indicates the presence of two non-equivalent vana-dium sites in the ratio of 1:1.8, in good agreement with thecrystal structure described above. The 51V NMR parame-ters for these two sites do not differ significantly (Table5). Some metavanadates of divalent metals also have asso-ciated tetragonal VO5 pyramids in their structure, theseinclude, for example, Ca(VO3)2 and a-Zr(VO3)2 (Table 5).

5.1.8. Strongly associated octahedral sites in decavanadatesThe molecular structure of a V10O28

4� polyanion wasfirst reported by Evance [182]. The model structure of thispolyanion is presented in Fig. 26A. (adopted from Ref.[183]). This structure consists of 10 VO6 octahedra labeledas V1; V1

�; V2; V2�; V3; V3

�; V4; V4�; V5; and V5

�, whichwithin the polyanion results in only three non-equivalentvanadium sites in the ratio of V5/V3,4/V1,2 = 2:4:4. TwoV5 octahedra are buried entirely inside the polyanion,and their edges are shared with other octahedra in thestructure. Two pairs of V3,4 and V1,2 octahedra are similar

2000 -2000 -40000

3000 2000 1000 -10000

δ, ppm

A

BC1

2

δ, ppm

*

Fig. 26. 51V 10 kHz MAS NMR spectra of decavanadates obtained at 9.4 TK4Na2V10O28Æ10H2O. (D) Rb4Na2V10O28Æ10H2O. The isotropic lines are indicwith the simulated spectra shown below. The calculation parameters used in simgr = 0.1, diso = �424 ppm; (D2) CQ = 2.2 MHz, gQ = 0.99, dr = 345 ppm, gr =

in their structure and form an external part of the polyan-ion. In aqueous solutions, V10O28

6� anions are stable andshow characteristic 51V NMR chemical shifts of�425 ppm for V5, �506 ppm for V3,4, and �524 ppm forV1,2 sites [184].

In the solid state all vanadium sites in the polyanionbecome non-equivalent, as observed, for example, inK5NaV10O28Æ10H2O, K4Na2V10O28Æ10H2O, Rb4Na2V10O28Æ10H2O, and (NH4)4Na2V10O28Æ10H2O. As a result, corre-sponding 51V MAS NMR spectra are usually very compli-cated due to closely overlapping lines from different sites.The isotropic shifts are usually in the range of �425 to�500 ppm. While the magnetic shielding anisotropy israther small for all these sites, it is smallest for internalV5 octahedra with the highest degree of association.

The nature of charge-balancing cations in decavana-dates has only moderate effects on the isotropic shift val-ues. The resonances corresponding to V5 sites experienceshift from �425 to �430 ppm upon changing cations fromK+ to NH4

þ and finally to Rb+. The V3,4 and V1,2 reso-nances at around �500 ppm become more separated forNH4

þ, and somewhat less separated for Rb+.

5.2. Correlating local environment of vanadium nuclei in

VOx species with 51V NMR parameters

Studies of vanadium-51 NMR spectroscopy of solidsamples has a long tradition. There is a large volume of

2000 -2000 -40000

-2000 -3000 -4000 -5000

D

1

2

δ, ppm

*

. (A) Structure of V10O286� from Ref. [183]. (B) K5NaV10O28Æ10H2O. (C)

ated with asterisks. In (C and D) the experimental spectra are comparedulations were as following: (C2) CQ = 3.7 MHz, gQ = 0.9, dr = 300 ppm,0.07, diso = �489 ppm.

0

0.2

0.4

0.6

0.8

1.0

0 200 400 600 800 1000

VO4 Q0

VO4 Q1

VO4 Q2

VO4 axial

VO5,6 non-axialVO6 axial

VO5 axial

η σ

δσ, ppm

Fig. 27. Correlation between 51V asymmetry, gr, and anisotropy, dr,parameters of chemical shielding tensor obtained for various vanadiumcompounds.

0

1

2

3

4

-200 -300 -400 -500 -600

VO4 trigonal pyramids

VO6 strongly associatedVO5,6 associated pyramidsVO5,6 isolated pyramids

δ⊥, ppm

CQ, M

Hz

for δσ>200 ppm, ησ<0.2

Fig. 28. Correlation between the value of perpendicular component of the51V NMR chemical shielding tensor, d^, and the quadrupolar couplingconstant, CQ, for vanadium polyhedra with one short V@O bondcharacterized by dr > 200 ppm and gr < 0.2.


experimental data available for a wide variety of systemsand materials. In Table 5 we summarize all the experimen-tal results currently available in the literature and someresults just recently obtained by us for individual vanadiumoxide compounds with well-characterized vanadium localenvironment. These results have been accumulated overyears by many research groups often working with differentconventions for reporting CSA and quadrupolar parame-ters. For convenience of use, all the data listed in Table 5have been recalculated from those originally reported andare now presented according to notations adopted in thisreview as outlined above.

There have been several reported attempts to correlatethe local structure of vanadium sites with 51V NMRparameters. The most comprehensive correlation reportedearlier by Lapina et al. [34] was between the coordinationof vanadium in VOx polyhedra and the value of the chem-ical shift anisotropy. Using this correlation these authors

Table 6Eight types of VOx sites that can be recognized by 51V NMR

dr (ppm)

Tetrahedral Q0 sites <100Tetrahedral Q1 sites 100–200Tetrahedral Q2 sites 200–500Associated non-axial VO5 and VO6 sites 200–400Isolated octahedral VO6 and tetragonal VO5 pyramids >700Associated pyramids VO5 and VO6 200–400Strongly associated octahedra VO6 300–400Trigonal pyramids VO4 400–500

were able to distinguish four different types of vanadiumenvironments, including tetrahedral sites Q0, Q1, and Q2,as well as vanadium in distorted octahedral environmentof the V2O5 type.

In this report, we are revising these earlier observationsby specifying a much wider range of vanadium environ-ments. Correlating the chemical shielding anisotropy andthe CSA asymmetry parameter as shown in Fig. 27 it isnow possible to clearly differentiate five different types ofvanadium coordination environments. In addition to tetra-hedral sites Q0, Q1, Q2 these now include non-axial VO5

and VO6 sites, and axial VOn (n = 4,5,6) species. The lattergroup contains isolated and associated VO6 octahedra, iso-lated and associated VO5 tetragonal pyramids, and trigonalVO4 pyramids, which are difficult to discriminate usingonly CSA parameters.

In addition to characteristic values of the chemicalshielding anisotropy and the CSA asymmetry parametersfor each site outlined in Table 6, we also note several otheruseful tendencies and correlations.

The first observation is related to the value of the isotro-pic chemical shift, which in Q0 sites was found to dependon the nature of the counter-ion. Thus, the isotropic chem-ical shift will usually increase as following: M+ = Li fi

gr �d^ (ppm) CQ (MHz) Key NMR parameters

0–6 dr, gr

0.1–0.9 2.5–10 dr, gr

0.6–0.8 2–7 dr, gr

0.3–0.6 3.5–6 dr, gr

0–0.1 200–350 0–2 dr, gr, d^, CQ

0–0.2 200–400 2–3 dr, gr, d^, CQ

0–0.2 200–350 3–4 dr, gr, d^, CQ

0–0.2 400–500 1–4 dr, gr, d^, CQ


Na fi K fi Cs; M2+ = Zn fi Mg fi Ba fi Sr; M3+ = Bi fiCe fi La fi Lu fi Y fi Sc.

In a similar fashion, the quadrupolar coupling constant,CQ, in vanadates of M+ is typically ranging from 1 to3.5 MHz, in M2+ vanadates this constant falls in a verynarrow range of 0.5–1 MHz, and in M3+ vanadates CQ

can be from 3.5 to 5.6 MHz. Such a broad distribution inCQ is most likely due to relatively short V–O distances invanadates normally not exceeding 1.7 A. Even small devia-tions of these distances from the average value would resultin considerable electric field gradients at the location ofvanadium nucleus.

The chemical shift anisotropy in Q0 sites does not exceed100 ppm, indicating high symmetry of the vanadium coor-dination in these species.

Unfortunately, the values of the chemical shift anisot-ropy and the CSA asymmetry parameter (Fig. 27) do notallow differentiating between various types of axial vana-dium species, including isolated and associated VO6 octa-hedra, isolated and associated VO5 tetragonal pyramids,and trigonal VO4 pyramids. This can be done, however,by analyzing effective values of d^ and the value of thequadrupolar coupling constant in cases with large CSA,dr > 200 ppm, and gr < 0.2. The effective values of d^ canbe estimated as d^ � 1/2 (d1 + d2), where di are the compo-nents of the CSA tensor. An example of such correlation isshown in Fig. 28. Using this correlation it is now possibleto identify four more types of vanadium coordination withaxial symmetry of CSA (Table 6).

We also note that while isolated pyramids would usuallydemonstrate large magnetic shielding anisotropies, an asso-ciation of pyramids often leads to decreasing values of theanisotropy. The situation is opposite for tetrahedra, wherethe smallest anisotropy is always found in isolated species,while it is gradually increasing with the extent of associa-tion as shown in Fig. 29.

-1000-5000ppm

-1000-5000

isolated

dimer

chains

isolated

2D structure

3D structure11

22

33

⊥ ||δδ

δδ δ

Fig. 29. Correlation between local structure of vanadium sites andassociation of VOx polyhedra with the type and magnitude of CSA. Thevalue of di is determined by a V@O distance: variation of this distancefrom 1.56 to 1.68 A leads to di variation from 1600 to 800 ppm. The valueof d^ corresponds to different types of VOx pyramids: for VO5 and VO6 itlies in the range from �200 to �350 ppm, for VO4 – in the range �450 to�600 ppm.

Using results presented above it is now possible to differ-entiate up to eight different vanadium coordination envi-ronments by their 51V NMR parameters. Thesecorrelations will be useful for studying new vanadiumoxide systems, in particularly those systems and materialswith disordered or poor crystalline structures.

6. 93Nb NMR data compilation

6.1. Chemical shielding and quadrupolar tensor parameters

in individual niobium compounds

The niobium coordination number in niobium-contain-ing oxide compounds varies from 4 to 8.

6.1.1. Six-coordinated compounds

The most typical are compounds with six-coordinatedniobium. In this work the 93Nb NMR parameters are sum-marized for the following six-coordinated niobium com-pounds: Li3NbO4 [205,206], Bi3NbO7 [207], La3NbO7

[208], BiNbO4 [209], LiNbO3 [210], NaNbO3, KNbO3

[211,212], SnNb2O6 [213], Sn2Nb2O7, K8Nb6O19, Te3Nb2O11

[214], NbVO5 [163], VNb9O25 [164], Pb(Mg1/3Nb2/3)3

(cubic, tetragonal, and rhombic symmetry), Pb3Nb4O13,PbNb2O6, Pb2Nb2O7, Pb5Nb4O15, Pb3Nb2O8 [47]. 93NbNMR parameters for these compounds are summarizedin Table 7.

Niobium compounds with cubic symmetry, includingLi3NbO4, Pb(Mg1/3Nb2/3)3, Bi3NbO7, and Pb3Nb4O13 haverelatively small quadrupolar coupling constants usually notexceeding 20 MHz. The 93Nb isotropic chemical shift forthese compounds is found within the �900 to �990 ppmrange. Measurements at high magnetic fields allow determi-nation of the chemical shift anisotropy, which is normallynot exceeding 140 ppm. Because CQ and the chemical shiftanisotropy are not extremely large, it is often possible toaccurately calculate a full set of 93Nb NMR parameterswhich fits all the experimental spectra, including static,MAS, HFMAS, and 3QMAS spectra.

Examples of MAS, HFMAS, and static 93Nb NMRspectra for Li3NbO4 are shown in Fig. 30. The static93Nb NMR spectrum recorded at 97.7 MHz (9.4 T) hassingularities of the first-order quadrupolar perturbations(Fig. 30B). Note, that in addition to the central transitiononly the ±1/2 M ±3/2 transitions can be observed underthese experimental conditions. The value of CQ estimatedfrom the static spectrum is 12 MHz. In the 93Nb MASspectra at 9.4 T the satellite transitions for m = 1/2, 3/2,and 5/2 can also be seen (Fig. 30A). The spinning side-bands analysis described above produced accurate valuesof CQ = 11.5 MHz and gQ = 0.1. The 93Nb MAS NMRspectrum recorded at higher field had revealed a moderatechemical shift anisotropy, dr = 140 ppm (Fig. 30C). Theisotropic chemical shift for Li3NbO4 was found at�950 ppm in all these experiments.

93Nb NMR spectra of six-coordinated compounds withnon-cubic symmetry, i.e. orthorhombic, rhombohedral,

Table 793Nb NMR parameters of individual niobium compounds

Compound Ref. CQ (MHz) gQ diso (ppm) gr dr (ppm) a�, b�, c� Field Method

Four-coordinated

YNbO4 [55] 81 0.41 �800 0 0 0, –, 0 Static[55] 80 0.5 �750 HF MASThis work 82 0.38 �845 0.48 200 165, 16, 100 HF Statica

LaNbO4 [55] 70 0.3 �650 0 0 StaticThis work 89 0.15 �800 0.3 200 145, 9, 103 HF Statica

PrNbO4 [55] 87 0.25 �400 0 0 StaticThis work 87 0.25 �350 – – HF Static

Five-coordinated

Na5NbO5 [55] 11.1 0.01 �903 0 0 MAS[55] 11.1 0.01 �903 0.6 120 0, 31, 90 HF MASa

CaNb2O6 [55] 50 0.8 �990 0.5 �300 Static[55] 50 0.8 �990 0 �100 MAS

Six-coordinated

Pb(Mg1/3Nb2/3)O3 [47] <0.8 �900 NutationBi3NbO7 [55] <20 �900 MASLi3NbO4 [55] 11.5 0.1 �950 MAS

[55] 12.0 0.1 �950 0.35 135 Static[55] 12.0 0.1 �954 0.3 140 HF MAS

a-BiNbO4 [55] 23 0.35 �963 0 0 MAS[55] 23 0.4 �963 0.25 �180 HF MAS[55] 20.7 0.79 �974 3QMASThis work 20.7 0.51 �977 0.56 �150 24, 22, 78 Statica

Pb3Nb4O13 [50] 13.7 �995 3QMASPb(Mg1/3Nb2/3)O3 [47] �17 �954 to �980 NutationPb(Mg1/3Nb2/3)O3 [47] >62 �954 to �980 NutationLa3NbO7 [55] 49 0.21 �980 0 0 Static, MAS

This work 49 0.275 �968 0.69 �113 50, 27, 72 HF Statica

K8Nb6O19 This work 86 �970 HF StaticSnNb2O6 [55] 38.7 0.67 �1010 0 0 Static

[55] 40 0.45 �1010 0 0 MASLiNbO3 [55] 22 0.01 �1004 Static

[55] 22.2 0.2 �996 MAS[35] 22 0.2 �1004 MAS[21] 22.0 0 �1009 Single crystal[50] 22.1 – �1004 3QMAS

Cs4Nb11O30 (3 sites) [222] 16 – �1062 HF 3QMASNaNbO3 [50] 22.7 – �1073 3QMAS

[222] 20 – �1070 HF 3QMASNaBa2Nb5O15 (2 sites) [222] 21 – �940 HF 3QMAS

24 – �1079KNbO3 [55] 21.5 0.6 �1015 0 MAS

[47] 23.1 0.80 �1069 HF MASSn2Nb2O7 [39] <20 �1050 MASPMN pyrochlore [50] 26.8 �1014 MASPbNb2O6 [50] 16.8 �1113 3QMAS

[48] 19 0.5 �1090 0.2 230 45, 20, 30 Single crystalTe3Nb2O11 [55] 22 0.6 �1166 0 �150 MAS

[55] 22 0.6 �1176 0.1 �250 HF MAS[55] �1207 HF STMAS

PNb9O25 (3 sites) [222] 12 – �1177 HF 3QMAS25 – �1202

This work 20 �1167 HF Static and 3QMAS29 �120468 �1250

NbOPO4 [222] 21 – �1316 HF StaticNbVO5 [35] 16.5 0.9 �1338 MASNb3(NbO)2(PO4)7 (3 sites) [222] 30 – �1583 HF MASNa3.04Nb7P4O29 (4 sites) [222] 32 – �1287 HF MAS

27 – �1328Na4Nb8P4O32 (4 sites) [222] 29 – �1285 HF MAS

30 – �1338Pb2Nb2O7 (9 sites) [50] 13.6 �1003 3QMAS

17.0 �978(continued on next page)


Table 7 (continued)

Compound Ref. CQ (MHz) gQ diso (ppm) gr dr (ppm) a�, b�, c� Field Method

Pb5Nb4O15 (10 sites) [50] 16.6 �1013 3QMAS17.9 �975

Pb3Nb2O8 (10 sites) [50] 20.6 �999 3QMAS�951

H-Nb2O5 (15 sites) [222] 24 �1204 HF 3QMAS

Seven-coordinated

LaNb5O14 (3 sites) [55] �1200 (NbO6) HF STMAS�1230 (NbO6)�1267 (NbO7)

K2NbF7 [53] 38.5 0.35 �1600 0 �200 HF MAS

Eight-coordinated

NaNb3O8 (2 sites) [55] �1250 (NbO7) MAS�1500 (NbO8)

Cs3NbO8 This work �1600 HF Static

a Simultaneous simulations of MAS and static spectra recorded at 9.4 and 21 T.

, ppm-5500-4500-3500-2500-1500-500500150025003500

CT

±1/2 ±3/2 ±3/2 ±5/2 ±3/2 ±5/2

-920 -960 -1000 -2800 -30004000 3800

A

2000 0 -2000 -4000

1

2

0 -1000 -2000

B C

, ppm , ppmδ δ

δ

Fig. 30. 93Nb NMR spectra of Li3NbO4 obtained at 9.4 and 21.1 T. Experimental spectra are shown above, calculated spectra are shown below. (A)10 kHz MAS spectrum at 9.4 T. Details of the central transition +1/2 M �1/2 and two satellite transitions, ±3/2 M ±1/2 and ±5/2 M ±3/2 are shown asinsets. (B) Static spectrum obtained with the quadrupolar echo pulse sequence at 9.4 T. (C) 9 kHz MAS spectrum at 21.1 T. The same simulationparameters were used in all three simulations as following: CQ = 11.5 ± 0.5 MHz, gQ = 0.1, diso = �950 ± 5 ppm, gr = 0.30 ± 0.05, dr = 135 ± 5 ppm.The two latter parameters were not used in simulations shown in (A).


monoclinic, rhombic, tetragonal, and trigonal, are oftenbroadened due to strong quadrupolar interactions. Analy-sis of such spectra is complicated and requires a variety of

approaches as illustrated below for LiNbO3. Of all niobiumcompounds, LiNbO3 is the most investigated with solid-state 93Nb NMR [20,21,27–29,31,215,216]. As a matter of


fact, all new NMR techniques applicable to I = 9/2 havefirst been tested on LiNbO3 [50,65,66,217].

The 9.4 T 93Nb MAS NMR spectrum of LiNbO3 spunat 30 kHz, shows a lineshape dominated by the second-order quadrupolar interactions (Fig. 31A). Well-pro-nounced spinning sidebands from the satellite transitions±1/2 M ±3/2, ±3/2 M ±5/2, and the central transition+1/2 M �1/2 allows the determination of CQ, gQ, anddiso values (Fig. 31A, Table 7). While improvement isexpected in 93Nb resolution provided by DAS spectros-copy [50], it is offset by the significant homonuclearNb–Nb dipolar interactions dominating the centerbandin the isotropic dimension of the DAS NMR spectrumof LiNbO3 (Fig. 31B). At the same time, the 93Nb3QMAS NMR spectrum of LiNbO3 shows resolutionof approximately an order of magnitude better than inthe DAS spectrum (Fig. 31C). The full-widths at ahalf-maximum height of the 93Nb resonances of LiNbO3

were found to be 9 kHz in MAS spectra, 5.8 kHz inDAS, and only 0.7 kHz in a 3Q projection of the3QMAS spectrum. Since the 93Nb CQ is large in thiscase, �22 MHz, the two-dimensional nutation spectros-copy can also provide complementary information(Fig. 31D). The 2D nutation spectrum of LiNbO3 showsa single Nb site with its center of gravity at 4mRF corre-sponding to a CQ value of ca. 20 MHz [48].

Regardless of different NMR approaches, either MAS,Static, DAS, 3QMAS, 2D nutation, or SSTMAS at 9.4 T(Fig. 31E), the NMR parameters obtained for LiNbO3

are rather similar (Table 7).Magnetic fields above 18–20 T are most beneficial for

niobium compounds with large quadrupolar coupling con-stants exceeding 20 MHz or even larger, as, for example, inBiNbO4. Static, MAS, 3QMAS at 9.4 T and MAS 93NbNMR spectra recorded for this compound at 21 T areshown in Fig. 32. These spectra are resolved well enoughto accurately measure the magnetic shielding anisotropy,which in this compound was found to be small (only150 ppm). Certainly, such small magnetic shielding anisot-ropy would be very difficult to detect at lower magneticfields (Fig. 32A).

At even larger quadrupolar coupling constants, inexcess of 50 MHz, ultrahigh magnetic fields become anecessity. When static experiments are performed at sev-eral fields, at least one high-field measurement is requiredas shown for La3NbO7 (Fig. 33). In this compound theniobium coordination in zig-zag chains of NbO6 octahe-dra is very close to symmetric. Nevertheless, due to thefact that Nb atoms are off-center, the observed 93NbNMR line is broad (Fig. 33) and the quadrupolar cou-pling constant is almost 50 MHz with the symmetry ofthe quadrupolar tensor close to axial (gQ = 0.2). The iso-tropic shift (�980 ppm) is somewhat smaller than in alkaliniobates.

The following conclusions can be drawn from analysisof the six-coordinated niobium compounds with non-cubicsymmetry.

(i) In such compounds the isotropic chemical shift seemsto depend on the ionic character of the niobium sub-lattice. In niobates of M(+1, +2,+3) elements theniobium sublattice has more anionic character, andthe isotropic shifts are in the range from �1000 to�1100 ppm. When M(+4) or M(+5) is present, theniobium sublattice has more covalent (towards cat-ionic) character and the isotropic chemical shifts areshifted to �1200 to �1300 ppm.

(ii) The quadrupolar coupling constant depends on thesite symmetry and not on the coordination number.CQ can vary in a wide range, from hundreds of kilo-hertz to hundreds of megahertz.

(iii) The chemical shift anisotropy is more pronouncedwhen atoms of different types are present in the firstcoordination sphere, e.g. O and F.

6.1.2. Four-coordinated compounds

Compounds having Nb with coordination four arerather rare. A few of them belong to the MNbO4 type, withM = Y, La, Pr, Nd. The NbO4 tetrahedra in MNbO4 arestrongly distorted with two types of Nb–O distances equalto1.9522 and 1.8359 A [218]. Neighboring NbO4 tetrahedraare placed in close proximity, with distances from niobiumto neighboring oxygen atoms of about 2.4–2.5 A. For thesereasons, NbOx polyhedra in MNbO4 are often consideredas intermediate between isolated NbO4 tetrahedra andedge-sharing chains of NbO6 octahedra with two longNb–O bonds [218–220].

93Nb NMR spectra of MNbO4 differ significantly fromthe spectra of compounds with well-defined NbO6 octahe-dra in their structure. Thus the 93Nb NMR static spectrumof YNbO4 is very broad, and at low magnetic fields it waspractically impossible to record this spectrum without dis-tortions (Fig. 34A) [54]. Computer simulation of this spec-trum has produced a very large quadrupolar couplingconstant of 80 MHz and the isotropic shift unusuallyshifted to �800 ppm. Similar parameters were also foundfor LaNbO4 (CQ = 70 MHz, diso = �650 ppm) (Table 7).

Ultrahigh field static NMR experiments (Fig. 34B) sig-nificantly improve the situation: it was possible to obtaina complete set of the chemical shielding and quadrupolartensors parameters which describe precisely the static spec-tra obtained at two fields (Fig. 34 and Table 7).

Crystalline structures of rare earth niobates MNbO4 arevery similar. Unfortunately, almost all these compoundsare paramagnetic, which complicates their 93Nb NMRstudies. When paramagnetic effects are not as strong, e.g.for Pr and Nd, it was possible to obtain 93Nb NMR spectraand to measure the corresponding NMR parameters(Table 7). The 93Nb isotropic chemical shifts in PrNbO4

and NdNbO4 differ considerably from YNbO4 andLaNbO4, most likely due to paramagnetic effects.

Thus, the 93Nb isotropic chemical shifts for four-coordi-nated Nb sites occur at the lower field range (�400 to�900 ppm) compared with the six-coordinated Nb sites(�900 to �1300 ppm). Due to significant distortions of

10000 0 -10000

-600 -900 -1200 -1500

5000 -5000 -15000

1

2

3 -5500 -6000

4

5

, ppm

A

B C

D E

δ2, ppm δ2, ppm

νrf

δ1, ppm

δ2, ppm

-1000 -1100 -1200

-1000

-1100

-1200

δ1, ppm

δ1, ppm δ1, ppm

δ

≈

Fig. 31. Experimental and calculated 93Nb NMR spectra of LiNbO3 measured at 9.4 T. (A) 30 kHz MAS spectrum. Details of the central transition+1/2 M �1/2 and two satellite transitions, ±3/2 M ± 1/2 and ±5/2 M ±3/2, are shown as insets. Simulated spectra are shown below with parameters:CQ = 22 MHz, gQ = 0.2, diso = �996 ppm, mr = 30 kHz. (B) 93Nb DAS NMR spectrum. (C) 93Nb 3QMAS NMR spectrum. (D) Pure phase 2D nutation93Nb NMR spectrum. (E) 93Nb STMAS spectrum. Spectra (B–D) are reproduced with permission from Ref. [50]. Spectrum (E) is reproduced withpermission from Ref. [66].


the local environment in four-coordinated sites and theirpseudo-octahedral type, the quadrupolar coupling con-stants are usually very large (>70 MHz). The magneticshielding anisotropy is less than 200 ppm.

6.1.3. Five-coordinated compounds

Among the compounds studied so far, only two haveniobium in coordination state five, Na5NbO5 andCaNb2O6.

-800 -1000 -1200δ, ppm

1

2

C

-500 -1000 -1500

δ, ppm

1

2

B

ppm160 80 0 -80 -160

12

8

4

0

-4

D

0 -500 -1000 -1500δ, ppm

1

2

A

Fig. 32. Experimental and calculated 93Nb NMR spectra of BiNbO4. (A) 30 kHz MAS spectrum obtained at 9.4 T. (B) Static spectrum obtained at 21.1 T.(C) 30 kHz MAS spectrum obtained at 21.1 T. Calculated spectra shown below experimental spectra in (A), (B) and (C) were calculated with the same setof NMR parameters, CQ = 23 MHz, gQ = 0.35, diso = �963 ppm, dr = �150 ppm. (D) DQ STMAS spectrum recorded at 21.1 T, mr = 20 kHz.

2000 1000 0 -1000 -2000 -3000 -4000 -5000

O1

O1

O1

O1O3

O3O2

O2

O2

O2

O1

O1

La Nb

δ, ppm

1

2

Fig. 33. 93Nb MAS NMR spectra of La3NbO7 obtained at 9.4 T (35 kHz). (1) Experimental spectrum. (2) Simulated spectrum with the followingparameters: CQ = 49 MHz, gQ = 0.21, diso = �980 ppm, gd = 0, dr = 0 ppm, mr = 35 kHz. The structure of La3NbO7 consists of zig-zag chains of NbO6

octahedra with lanthanum ions occupying two types of polyhedra as illustrated.


The 93Nb NMR spectrum for Na5NbO5 is differentfrom the spectra of similar four-coordinated Nb com-pounds by having a much smaller quadrupolar couplingconstant (11 MHz) and a significant high-field shift of

diso. The complete set of NMR parameters for this com-pound was determined from NMR experiments at differ-ent magnetic fields (Fig. 35 and Table 7). The CS tensorparameters and the tensors orientation were obtained


only from high-field measurements. The small value ofthe quadrupolar coupling constant reflects the symmetricnature of the Nb sites and their isolated character. Thehigh-field shift as compared with four-coordinated com-pounds is due to the higher coordination number.

The isotropic 93Nb NMR shift of ��990 ppm found forCaNb2O6 is typical for five-coordinated sites, and the rela-tively large CQ � 50 MHz is the result of orthorhombicsymmetry.

Therefore, the following 93Nb NMR features have beenrevealed for five-coordinated Nb sites. In comparison withfour- and six-coordinated sites, five-coordinated Nb dem-onstrates an intermediate range of the isotropic chemicalshifts. The quadrupolar coupling constant depends on thelocal symmetry. It is small for symmetric isolatedsites, and large for pseudo-octahedral sites with one longNb–O bond.

2000 1000 0 -1000 -2000 -3000 -40

A

δ, ppm

Fig. 34. 93Nb NMR static spectra of YNbO4 obtained with a quadrupolar solidare show at the top. The spectra at the bottom were calculated with the folldr = 200 ppm, a = 165�, b = 16�, c = 100�.

-700 -800 -900 -1000 -1100

A

1

2

δ, ppm

Fig. 35. 93Nb MAS NMR spectra of Na5NbO5 obtained at (A) 21.1 T and (B) 9calculated with the following parameters: (A2) CQ = 11.1 MHz, gQ = 0.01, diso

gQ = 0.01, diso = �903 ppm, gr = 0, dr = 0 ppm, mr = 30 kHz. Also shown o±3/2 M ±1/2, and the central transition +1/2 M �1/2 together with simulationsquare-pyramids as illustrated on the left side.

6.1.4. Seven- and eight-coordinated compounds

Two of three studied 7- and 8-coordinated Nb com-pounds, LaNb5O14 and NaNb3O8, have several non-equiva-lent Nb sites in their crystal structure. Only in Cs3NbO8 isthere a single eight-coordinated site according to thereported crystal structure.

The structure of LaNb5O14 consists of three types ofNbOx polyhedra, edge-sharing pentagonal NbO7 bipyra-mids forming chains, which are interconnected by corner-sharing NbO6 octahedra. The 93Nb NMR spectrum for thiscompound is very complicated due to presence of three dif-ferent Nb sites with overlapping resonances. To obtain reli-able 93Nb NMR parameters for all three sites the high-fieldDQ STMAS technique was applied. The 93Nb DQ STMASspectrum of LaNb5O14 has isotropic shifts at �1200,�1230, and �1267 ppm, corresponding to two octahedralNbO6 and one NbO7 polyhedron, respectively [55].

00 -5000 0 -1000 -2000

1

2

δ, ppm

B

-echo pulse sequence at (A) 9.4 T and (B) 21.1 T. The experimental spectraowing parameters: CQ = 82 MHz, gQ = 0.38, diso = �845 ppm, gr = 0.48,

B

4000 2000 0 -2000 -4000 -6000

-2700 -3000 -3300

δ, ppm

.4 T. The experimental spectra are show at the top. The spectra below were= �903 ppm, gr = 0.6, dr = �120 ppm, mr = 8 kHz. (B2) CQ = 11.1 MHz,n the right are fine details in the satellite transitions ±3/2 M ±5/2 ands. The structure of Na5NbO5 is composed of isolated sodium and niobium

-500 -1000 -1500 -2000δiso, ppm

-1000-900 -1100 -1200 -1300 -1400

Cubic M(+1,2,3) M(+4) M(+5)

NbO4

NbO5

NbO6

NbO7

NbO8

Fig. 36. 93Nb NMR chemical shift scale for NbOx polyhedra.


NaNb3O8 forms a channel structure comprised of twotypes of chain-forming niobium polyhedra, edge-sharingNbO8 dodecahedra and edge-sharing distorted pentagonalbipyramids NbO7 [221]. However, for this compound,93Nb NMR experiments have been performed so far onlyat 9.4 T, where it was very difficult to determine accurate93Nb NMR parameters for two non-equivalent Nb sites.The 93Nb NMR static spectra were too broad to distin-guish the two individual components with typical second-order features of the central transitions. Nevertheless, itwas clear, that these spectra were indeed a superpositionof at least two sub-spectra, both strongly shifted up-field.From the high-speed 93Nb MAS spectrum the isotropicshifts for both Nb sites were estimated. For seven-coordi-nated niobium sites the isotropic shift was found at ca.�1250 ppm and for eight-coordinated niobium sites at�1500 ppm.

According to the crystal structure, there is only onetype of eight-coordinated Nb site in the Cs3NbO8 com-pound, the only Nb site in the structure. Yet, the 93NbNMR experiments indicate the presence of at least twooverlapping components. Despite fairly symmetric oxy-gen coordination, four of eight O–Nb bonds are at1.99 A, and other four at 2.05 A. The 93Nb NMR spec-tra show quite strong quadrupolar coupling interactions,CQ = 62 MHz, large CSA, and a considerable up-fieldshift (�1600 ppm).

When compared with six-coordinated niobium com-pounds, seven- and eight-coordinated Nb sites usuallydemonstrate high-field shifts, while the magnitude of thequadrupolar coupling constant is mostly determined bythe local symmetry.

CSA and quadrupolar parameters for a large number ofindividual niobium-containing oxide materials are summa-rized in Table 7. Many of these experimental results arepresented here for the first time.

6.2. 93Nb NMR chemical shift scale

By analyzing the data reported in Table 7 it is straight-forward to make several important conclusions regardingthe relationship between the isotropic 93Nb NMR chemicalshifts and the niobium coordination environment.

For four-coordinated Nb sites, the isotropic chemicalshifts corrected for the second-order quadrupolar pertur-bations occur from �650 to �950 ppm, withCQ > 70 MHz.

For five-coordinated Nb sites the isotropic chemicalshifts are observed in the range from �920 to �990 ppm,CQ � 10–50 MHz, dr � 200 ppm.

For six-coordinated Nb sites the isotropic shift diso var-ies from �900 to �1300 ppm, CQ from 1 to 100 MHz, anddr from 0 to 300 ppm. The range of �900 to �1000 ppmfor diso is typical for cubic symmetry. For six-coordinatedNb sites with non-cubic symmetry, the 93Nb isotropicchemical shifts are influenced by the ionic character ofthe niobium sub-lattice. In niobates of M(+1,+2,+3) ele-

ments the niobium sub-lattice has more anionic character,and the isotropic shifts are in the range from �1000 to�1100 ppm. When M(+4) or M(+5) is present, the nio-bium sub-lattice has more cationic character and the iso-tropic chemical shifts are from �1200 to �1300 ppm. Thevalue of the quadrupolar coupling constant for six-coordi-nated Nb sites varies over a wide range and depends on thesite symmetry. The chemical shift anisotropy is large onlywhen atoms other than oxygen are present in the first coor-dination sphere, for example, fluorine.

For seven-coordinated niobium sites the isotropic shiftvaries from �1200 to �1600 ppm while for eight-coordi-nated niobium sites the isotropic shift occurs at fieldshigher than �1500 ppm.

On this basis the 93Nb NMR chemical shift scale hasbeen proposed for niobium compounds (Fig. 36) [55].

7. 181Ta NMR data compilation

Only a very limited number of solid-state 181Ta NMRstudies has been reported to date, including several earlyworks on K[TaO3] [16,57,62,63], MI[TaY4] (MI = Cu, Tl;Y = S, Se, Te) [58], tantalum metal [59], and TaY2

(Y = S, Se) [60,61]. The 181Ta chemical shift range in solu-tion extends over 3450 ppm, and is limited by [TaCl6]� spe-cies at the low-field limit and by [Ta(CO)6]� at the high-field limit [56]. The shielding sensitivity of 181Ta is about1.6 times that of 93Nb and is comparable with that of95Mo. The 181Ta NMR spectrum of cubic KTaO3 consistof a rather narrow (5 kHz) central transition, with satellitetransitions spread out over ±100 kHz around the centralline. Doping KTaO3 with niobium, lithium and sodium[63] leads to considerable line broadening due to distortionof the cubic structure in pure KTaO3.

8. Paramagnetic effects in 51V and 93Nb solid-state NMR

spectra

In general, the effects of paramagnetic metal ions onNMR spectra (chemical shifts, relaxation times, etc.) are


well described for paramagnetic molecules and paramag-netic solids with low and intermediate concentration ofparamagnetic centers [223–225]. Paramagnetic shifts areparticularly useful in discovering medium and long-rangestructural information. We will not discuss here the originof all the different paramagnetic contributions to the chem-ical shift referring our readers to the cited literature. Wenote, however, that all the paramagnetic effects under con-sideration may be effectively incorporated in the second-rank hyperfine interaction (HFI) tensor, which can bedirectly probed by magnetic resonance experiments. TheFermi contact and pseudocontact contributions to thechemical shift are the most important in the context ofthe present discussion.

The contact shift is caused by the contact term in thespin-Hamiltonian as was first derived by Fermi. This shiftis associated with the isotropic contribution to the HFI ten-sor from the electron spin density located directly at thenucleus site. For paramagnetic ions of the VB Group,and especially for vanadium, unpaired electrons are notcompletely concentrated at the ion location, but are alsotransferred via direct delocalization and spin polarizationto remote atoms of the surrounding ligands. The patternof electron spin delocalization is not easily predictable fromfirst principles, especially for solids. This is one of the rea-sons why the information obtained from contact shift val-ues is often limited to the general picture of the electronspin delocalization over the nearest nuclei active in NMR.

In disordered paramagnetic systems with magneticallyanisotropic paramagnetic centers there is an additional ori-entationally averaged contribution to the isotropic value ofthe chemical shift called the pseudocontact, or dipolar,shift. It originates from the anisotropy of the net electronspin moment connected with a susceptibility tensor fixedat the molecular structure. The most common representa-tion for a pseudocontact shift is given by

dpc ¼ 1

12pr3Dvaxð3 cos2 h� 1Þ þ 3

2Dvrh sin2 h cos 2u

� �ð9Þ

with

Dvax ¼ vZZ �vXX þ vYY

2; vrh ¼ vXX � vYY ð10Þ

where Dvax and Dvrh are the axial and rhombic anisotropyparameters of the magnetic susceptibility tensor v of aparamagnetic metal, while two angles h and / define theorientation of an electron–nucleus vector in the frame ofthe magnetic susceptibility tensor. Somewhat differentexpressions for the pseudocontact shift can be found in[225]. Interpretation of the pseudocontact shifts dependsto a lesser extent on the electronic structure theory ofmolecules or solids, comparing with similar interpreta-tions of the Fermi contact shifts. Nevertheless, such inter-pretation often requires comparison to a diamagneticanalog with similar geometry and charge distribution.For example, for lanthanides unpaired electrons aremainly located in inner orbitals with little or no overlap

with orbitals of the surrounding ligands. This makes theFermi contact shifts negligible compared with the pseudo-contact contributions to the paramagnetic shift. In somecases pseudocontact shifts can be experimentally mea-sured even for nuclei located as far as 40 A from the para-magnetic center.

Therefore, by measuring local fields at the location of anucleus, it is possible to obtain information on the spin-density transfer from the paramagnetic ion, and also toestimate the distance between the observed nucleus andthe paramagnetic ion.

This simplified picture of electron-nuclear interactionsbecomes incomplete or even incorrect for solids with a highconcentration of paramagnetic centers and especially formagnetically ordered materials. In these situations otherapproaches originating from solid-state physics need tobe applied [226–230].

While V5+ containing phases are directly characterizedby conventional 51V NMR, this is not the case for mate-rials with vanadium atoms in lower oxidation states. It ispossible, however, though indirectly, to obtain informa-tion concerning the nature, location and the oxidationstate of vanadium centers from NMR spectra of neigh-boring atoms. An example of such an investigation forvanadium–phosphorus systems was reported by Lashieret al. [231] and later by Tuel et al. [232]. They haveapplied a so-called 31P spin–echo mapping technique toprobe vanadium paramagnetic centers by observing the31P chemical shift over a very large spectral region. Incompounds containing only diamagnetic V5+ centers the31P NMR chemical shifts are in the range 20–40 ppm; atthe same time for compounds containing paramagneticV4+ and V3+ the 31P NMR chemical shifts range from1600 to 2600 ppm and at around 4650 ppm, respectively.The 31P NMR line shift is directly proportional to thedensity of unpaired electrons. Thus, the 31P NMR chem-ical shift has provided information about the number ofparamagnetic vanadium species in the first coordinationsphere of phosphorus atoms, as well as on the oxidationstate of such species.

Applying the spin–echo mapping technique to 51V underultra-high speed MAS (and large spectral width) condi-tions, it is possible to detect 51V NMR signals of V5+ atomsin close proximity to paramagnetic centers.

In this work we report our recent 51V NMR data for anumber of rare earth vanadates where paramagnetic effectsare caused by the presence of paramagnetic cations. Wealso summarize results available for vanadium bronzes,where some vanadium is in a paramagnetic V4+ state,and for Cuban-like vanadium compounds, where closelyspaced paramagnetic centers form diamagnetic pairs whenplaced in persistent magnetic fields.

Since all these compounds are true paramagnetics, specialcare should be taken in choosing appropriate experimentalconditions while performing NMR experiments, i.e. theamount of the sample in the MAS rotor should be as smallas possible.


8.1. Presence of paramagnetic cations

Early studies of the effects of the paramagnetic centerson the 51V NMR spectra were usually performed withCW-NMR spectroscopy. The most important results of51V CW-NMR in vanadium oxides, bronzes, and variousvanadates of s-, d-, and f-elements have been summarizedby Pletnev et al. [22], together with some experimentalspectra, and several examples of the cluster MO-calcula-tions of the electric field gradients and the magneticinteractions.

Examples of 51V MAS NMR spectra for several vana-dates of rare earth and 3d-elements are presented inFig. 37, with the corresponding spectral parameters givenin Table 8. The 51V quadrupolar coupling constant foundfor iron orthovanadate FeVO4 is rather small, while forrare earth orthovanadates the quadrupolar coupling con-stant exceeds those found in meta- and orthovanadates ofthe Group I and Group II elements. The asymmetryparameter of the quadrupolar tensor is zero for most ofthese compounds. It was reported earlier by Pletnev et al.

δ, ppm14000160001800020000

4

δ, ppm020004000

*

*

*

Fig. 37. 51V MAS NMR spectra of MVO4 vanadates with paramagnetic cationmarked with asterisks. Note an extremely large shift of +17,000 ppm observed2.5 mm MAS probe with mr = 30 kHz.

[233,234] that the 51V NMR isotropic chemical shifts forCr, Fe, Co, and Ni orthovanadates are shifted to low fieldcompared with that of the reference KVO3 solution. Indeedthis is the case for FeVO4, where the isotropic chemicalshift in the 35 kHz MAS spectrum is at 17,000 ppm, i.e.at much lower field to the reference sample. Because ofthe considerable distance of 3.4 A between vanadiumatoms and the paramagnetic centers, the direct contactinteractions of 51V nuclei with electrons located on the par-tially occupied 3d shells is unlikely. This strong effect onthe isotropic chemical shift can be explained by transferof electron density from the magnetic d-electrons via theatomic orbitals of nonmagnetic ions, vanadium, andoxygen.

The isotropic 51V NMR chemical shifts in rare earthorthovanadates can be formally presented as:

diso ¼ d0 þ dp; ð11Þ

where d0 is the chemical shift originating from the elec-tronic structure of the filled orbitals, and dp is the shift

1

2

3

-4000-2000

*

s: (1) PrVO4, (2) YbVO4, (3) EuVO4, and (4) FeVO4. The isotropic lines arein the Fe-containing sample. The spectra were obtained at 9.4 T using a

Table 851V NMR parameters for several vanadates of rare earth and 3d-elements

Compound CQ (MHz)a gQa diso (ppm)b dr (ppm)b Ion configuration 2S+1LJ base state

FeVO4 1.7 � 17,000 � 3d5 11S5/2 (11)CeVO4 3.5 0.9 �559 200 4f1 2F5/2 (6)PrVO4 5.59 0 �253 �520 4f2 3H4 (9)NdVO4 5.5 0 �242 �550 4f3 4I9/2 (10)EuVO4 4.95 0 �935 – 4f6 7F0 (1)HoVO4 4.55 0 0 – 4f10 5I8 (17)ErVO4 4.51 0 30 – 4f11 4I15/2 (16)TmVO4 4.43 0 90 – 4f12 3H6 (13)YbVO4 4.25 0 100 – 4f13 2F7/2 (8)

a Values of CQ and gQ are from Ref. [22].b Values of diso and dr are obtained in this work.


from interactions of the vanadium atom with unpairedelectrons of the M3+ ions.

When diso = d0, as found in diamagnetic vanadates,the isotropic chemical shifts are in the range typicalfor vanadium in an oxygen environment, i.e. as inCeVO4, LuVO4, ScVO4, YVO4, and LaVO4 (Table 8).This confirms that the d0 values are about the samefor the whole range of the rare earth orthovanadate.Therefore, all the changes in diso observed for rare earthvanadates are due to changes in dp. Also, there was nocorrelation found between the values of the isotropicchemical shifts and the g-factor anisotropy in ESR spec-tra [235,236]. Therefore, the contribution of the pseudo-contact interactions to the chemical shift can be dis-missed as negligible.

Vanadates containing mixed paramagnetic and dia-magnetic cations are of interest for developing advancedlithium ion batteries. One such system based on LiCox-

Ni1�xVO4 compounds with paramagnetic Co and Niwas recently studied by Stallworth et al. [192]. In theirpreparations they varied the Co/Ni ratio, with x assumingvalues from 0 (only Ni present), 0.2, 0.5, 0.8, and 1.0(only Co present). The workers reported observing tworesonances in the 51V NMR spectra. One resonance inthe range 2500 to 4200 ppm having a Gaussian line shapeand strongly affected by paramagnetic interactions wasattributed to tetrahedral vanadium sites. The second reso-nance with the isotropic chemical shift of �630 ppm andexhibiting first-order quadrupolar broadened satellites wasassigned to octahedral sites. This contradicts, however,the correlation presented above in Section 5. Thelatter line at �630 ppm exhibiting the first-order quadru-polar interactions should instead have been attributedto tetrahedral vanadium sites. Moreover, diamagneticorthovanadates LiCdVO4, LiZnVO4, and LiMgVO4

where vanadium has a tetrahedral oxygen environmentshows very similar 51V NMR spectra. It is clear that thesetetrahedral sites are surrounded by diamagnetic Li andnot affected by paramagnetic interactions. At the sametime, the resonance at 2500–4200 should correspond tooctahedral sites and tetrahedral sites with paramagneticneighbors.

8.2. Systems with vanadium in mixed oxidation states

Vanadium bronzes, compounds somewhat similar tometals in their properties, are formed by incorporating Melements into the vacancies of the VOn oxide structure.The charge balance is achieved by transferring two valenceelectrons of M ions to vanadium ions, which are nominallypresent in two oxidation states. To underline the fact thatvanadium bronzes are actually oxides, sometime they arecalled oxide bronzes, or non-stoichiometric compoundsincorporating one-, two-, and three-charged ions. ElementM incorporated into the crystal structure of the oxide nor-mally occupies only a fraction of the available voids, withthe filling factor depending on the ionic radius of Mn+.By varying the concentration of M, it is possible to obtaina number of MxVOn phases with different composition[237]. Thus for V2O5 there are four such phases, a, b, c,and d. When x is not exceeding 0.04, the a-MxV2O5 phaseis formed, representing solid solutions of M1+,2+,3+ ionsincorporated into V2O5 [238]. The structure of this a-phaseis similar to V2O5. The b-phase is formed when0.2 < x < 0.4. In the b-phase there are three non-equivalentvanadium sites, two of them represent deformed octahedralcoordination, and the third site is five-coordinated. Furtherincorporation of Mn+ leads to formation of c and d phasesat x greater than 0.6.

One may expect that different types of phase transitionsin MxV2O5 obtained upon varying x could be followedwith 51V NMR. However, while studying LixV2O5

(0.4 < x < 1.4) Nakamura et al. [239] observed that for dif-ferent values of x the isotropic 51V shifts did not changesignificantly. Based on this observation it was concludedthat the VO5 square pyramidal structure of V2O5 isretained without drastic structural modifications.

As already discussed above, the isotropic 51V shift is nota very informative probe of the vanadium environment, themost informative being the chemical shift anisotropy andthe quadrupolar coupling constant. Indeed close inspectionof the 51V NMR spectra presented in the above mentionedpaper shows that variations in x do lead to changes invanadium environment. At loading x = 0.4 the structureis similar to V2O5, with one of the sites being of the b-phase


or a residual part of the a-phase. The bronze of b or c typeis formed at a value of x of about 0.8, as indicated by abroad anisotropic line with a chemical shift of �797 ppm.The isolated tetrahedral sites are formed at even higher x

loadings, x > 1, which is supported by the small anisotropyobserved and the moderate value of the quadrupolar con-stant. This type of transformation seems to be typical forother MxV2O5 bronzes, where several consecutive phasetransitions can be observed up to x = 0.8: a fi b fi b 0 fi c.Whereas at x > 0.8 the bronze structure is usually breakingapart.

The b(b 0) type vanadium bronzes of Li, Na, K, Ag, Bi,and Sr are structurally similar (Table 9). At normal condi-tions these compounds are paramagnetic metals with well-pronounced conductivity along the b axis. At lower tem-perature these compounds transform into non-metallicphases, and at the liquid He temperature convert eitherinto ordered ferromagnetic phases as in Na0.33V2O5, or intoa diamagnetic phase as in Ca0.3V2O5. The crystal structureof these bronzes can best be described as a strongly dis-torted V2O5 phase with three non-equivalent vanadiumsites, V1, V2, and V3. It is often assumed, although it isnot certain, that the paramagnetic V4+ ions, or unpaired

Table 9Parameters of the 51V NMR spectra in oxide bronzes and compounds contain

Compounds CQ (kHz)

Cuban-like compounds

V2S4(S2COEt)4 3120

V2S4(S2COi-Pr)4 3120

V2S4(S2CNBu2)4 31264,4 0-Bipyridinium cations, (C10H10N2)[(VO2)4(PO4)2]

Metal

metal V –

Oxides

b-VO2 6940a-VO2 1900Bronzes

Ca0.3V2O5 21004000

Bi4V2O10.6

Cs0.35V2O5

Na0.67V2O5 –Li0.3V2O5 –Na0.33V2O5 4000

K0.25V2O5

Ag0.35V2O5

Sr0.17V2O5 –Ba0.17V2O5 –Composites

PANI-V2O5-nanocomposites

electrons, are located either in V1 or V2, or in both, whileV5+ ions occupy only V3. From the data presented in Table9, it is clear that the line at �900 ppm is typical for allbronzes of the b type, and perhaps can be attributed toV5+ located near V4+. If, however, V4+ ions or unpairedelectrons occupy V1 and V2 sites, than V5+ shifted to thehigh field may also occupy V1 and V2 sites. In Ca0.3V2O5

and Cs0.35V2O5 bronzes 51V NMR signals were detectedat 2120 and 1565 ppm respectively, which is very close tosignals from diamagnetic V4+–O–V4+ pairs in b-VO2. Thiscorrelates with the non-magnetic state of Ca0.3V2O5 atliquid He temperature. It is interesting, that in Ca0.3V2O5

the 51V NMR signal is at �4766 ppm, very close to metallica-VO2. In similar metallic bronzes 51V NMR signals can beobserved from �1718 to �739 ppm. The more metallic thebronze becomes, i.e. moving from Na0.33V2O5 toNa0.67V2O5, the more the 51V NMR signal is shiftedtowards the signal found in pure vanadium metal.

Some representative 51V NMR spectra for vanadiumbronzes and VO2 are shown in Figs. 38 and 39. It is impor-tant to recognize, that the interpretation of such spectra isnot always straightforward, which sometimes leads toheated debates and controversy in the literature [240].

ing V4+–V4+

gQ dr (ppm) gr diso (ppm) Ref.

0.4 �960 0.001 70 This work54

0.4 �960 0.001 105 This work8256

0.40 �956 0.001 135.0 This work�580.1 [241]�593.4�599.2

– – – 5200 This work

0.46 1151 0.88 2113 [242]0.4 175 – �4788 [242]

0.9 270 1.0 2120 This work0.9 1000 – �4766

�1447 [189]1000 – 1565 This work800 – �775800 – �810

0.9 370 0.8 �739 This work– 835 0.92 �869 This work– 1600 – �1200 This work

�900�620 This work�900�435 This work�900

– 860 – �1718 This work– 1250 0.01 �1381 This work

�8500 [243]


For example, the high-temperature phase of a-VO2 isclearly metallic with characteristic anisotropy in properties,and at the same time this phase is characterized by theabnormal 51V NMR chemical shift at �4788 ppm, andopposite in sign to the Knight shift in vanadium metal.

4000 2000 0 -2

δ, pp

*

*

Fig. 38. 51V NMR MAS spectra of polycrystalline vanadium bronzes recordedlines are marked with asterisks.

0200040006000

δ, ppm

2

1

*

*

Fig. 39. 51V MAS NMR spectra of two vanadium compounds with V4+–V4+ pmarked with asterisks. The structure of a [V2S4]4+ unit is shown above. Dudiamagnetic as also supported by 51V NMR. The spectra were obtained at 9.4

This behavior can be explained by dominating contribu-tions of the electron polarization transfer between groundelectron orbitals and those with unpaired d-electrons. Ina similar fashion, to describe the non-metallic diamagneticelectronic state of the low-temperature b-VO2 phase, it was

000 -4000 -6000 -8000

1

2

m

at 9.4 T under 30 kHz MAS, (1) Ba0.17V2O5 and (2) Ca0.3V2O5. Isotropic

-4000-2000

airs in the structure. (1) VO2 and (2) V2S4(n-Bu2NCS2)4. Isotropic lines aree to a very short distance between two V4+ ions these compounds areT using a 2.5 mm MAS probe at mr = 30 kHz.

Fig. 40. 51V solid-state NMR spectra of vanadium chloroperoxidaseacquired at 14.1 T at two MAS spinning speeds of 17 kHz (1) and 15 kHz(2). The isotropic line at �520 ppm is marked with an asterisk. The insetshows spinning sidebands from the satellite transitions indicating largeCQ. Each spectrum took about 5 days to acquire. Reproduced withpermission from Ref. [136].


suggested to use either a model involving mixedoxidation states, V3+ and V5+, or a model of the diamag-netic V4+–O–V4+ (S = 0) pairs. The latter model is moreprevalent, and is supported by the isotropic 51V NMR shiftvalues found in four model compounds with Cuban-likestructures containing diamagnetic pairs V4+–S–V4+

(Fig. 39).From the practical point of view 51V CW-NMR spec-

troscopy used in early studies on vanadium bronzes[233,234,237,240] has certain advantages over modern Fou-rier-transform NMR spectroscopy. The most important isa possibility of recording extremely broad spectra withalmost unlimited spectral width. At the same time, pulsedNMR experiments with the same samples would requireextremely broad sweep widths, short high-power RF pulsesand very high spinning speeds if MAS is to be applied. Onthe other hand, modern NMR approaches bring increasedspectral resolution and the ability to resolve non-equivalentvanadium sites, which is impossible by CW-NMR.

9. Recent applications of solid-state 51V and 93Nb NMR in

oxide materials

9.1. Applications of solid-state 51V NMR

The last two decades have seen a dramatic increase inthe number of 51V NMR applications to a variety of solidsranging from materials science and biology to solid-statephysics. In heterogeneous catalysis, solid-state 51V NMRhas become almost a routine characterization technique.The seeming simplicity of 51V NMR has brought about avery wide user base, yet has had a somewhat negativeimpact as well. The main problem remains with inadequateNMR hardware which is being applied to study very chal-lenging and complex systems. Herein, we have selected afew of the most representative examples of successful 51VNMR applications from the recent literature, coveringmajor research areas, including bio-structural chemistry,materials science, and catalysis.

9.1.1. Bio-structural chemistry

Perhaps the most impressive example of the solid-state51V NMR spectroscopy applied directly to probe vanadiumcenters in proteins has been reported very recently by Poo-ransingh-Margolis and coworkers while studying 67.5-kDavanadium chloroperoxidase [136]. Each molecule of thisprotein contains a single vanadium atom, which translatesinto only 1 lmol of vanadium spins in the entire sample.Despite this very low concentration of vanadium sites,the authors were able to detect and analyze the spinningsidebands of the central and satellite transitions (Fig. 40).The quadrupolar and chemical shift anisotropy tensorswere determined by numerical simulations of the spinningsidebands and the line shapes of the individual spinningsidebands of the central transition. For these vanadiumsites the authors reported a value for the quadrupolar cou-pling constant CQ of 10.5 MHz and the chemical shift

anisotropy dr of 520 ppm. DFT calculations of the NMRspectroscopic observables for an extensive series of activesite models indicates that the vanadate cofactor in theseenzymes is most likely anionic with one axial hydroxo-group and an equatorial plane consisting of one hydroxo-and two oxo-groups. This approach has yielded thedetailed coordination environment of the metal centers,which is unavailable from other experimental measure-ments. Solid-state 51V NMR is expected to be generallyapplicable for studies of diamagnetic vanadium sites inother metalloproteins and similar systems.

9.1.2. Materials chemistry

Hybrid organic–inorganic materials based on vanadiumoxide have received significant attention due to their ionicand electronic properties with potential applications in var-ious technologies, such as reversible cathodes in lithiumbatteries, electrochromic devices, or even in heterogeneouscatalysis. Many vanadium compounds exhibit a layeredstructure allowing the reversible intercalation of a widerange of organic molecules to occur.

Durupthy et al. have recently studied vanadium oxidefoams prepared by mixing V2O5 with hydrogen peroxideand solution of 1-hexadecylamine [202]. Despite the disor-dered nature of these foams, using solid-state 51V NMR theauthors have been able to successfully identify a variety ofpolyoxovanadates contained in them. It has been shownthat the local environment of vanadium as probed by 51VMAS NMR consists of octahedral and tetrahedral vana-dium sites corresponding to decavanadate [HxV10O28]

(6�x)�


and [V4O10]4�/[V2O7]4� polyanions. However, the relativeamounts of different polyoxovanadate species could notbe controlled via synthetic means. The layered structureof foams is most likely formed by a self-assembly ofamines, while [HxV10O28](6�x)� and [V4O10]4�/[V2O7]4�

polyanions intercalate into interlayer spaces. In a similarfashion, the structure of the tailor-made macroporousvanadium oxide foams has been examined by 51V NMRin [244] (Table 5). It has been revealed that the local struc-ture of vanadium sites in these foams is similar to thatreported earlier for V2O5 xerogels [148].

Grey et al. [245,246] used variable temperature 51V MASNMR to study the synthesis of Li1+rV3O8 (r = 0.1–0.2)materials via a xerogel precursor route. Li1+rV3O8 materi-als are among several other layered lithium-containingvanadates with attractive electrochemical properties forapplications in rechargeable lithium batteries. These mate-rials generally consist of V3O8 layers also found in hewet-tite, CaV6O16Æ9H2O, with octahedrally and penta-coordinated vanadium atoms. Intercalation can occurbetween the layers. During synthesis the solid componentof the xerogel undergoes a series of phase transitionsinvolving layered phases with decreasing water concentra-tions, while the hewettite framework is largely maintainedthroughout the reaction. In the case of the dried liquidxerogel component the formation of anhydrous Li1+rV3O8

occurs through decomposition of lithium vanadates, whichwas followed first by formation and then by progressivedehydration of the hydrated hewettite structure.

Valence states of vanadium in VO2 have been studiedwith 51V MAS NMR in [242] in order to reveal if V–V pairsconsist of either V5+–V3+ or V4+–V4+. To answer this chal-lenging question, the authors applied the high-speed MAStechnique with spinning speeds up to 35 kHz at two differ-ent magnetic fields, 9.4 and 14.1 T. These spectra unambig-uously show the presence of a single V4+ site in the low-temperature non-metallic b-form of VO2. Variable-temper-ature 51V MAS NMR studies performed in the temperaturerange from 25 to 90 �C indicate that the phase transitionfrom non-metallic b to metallic a- VO2 is also accompanied

Fig. 41. Model of the surface structure in VOx/TiO2 ca

by a dramatic change in the Knight shift value, from 2115to �4788 ppm.

Michalaka and Mohammad were the first to directlydetect 51V NMR spectra from more than two valence statesof vanadium present simultaneously in the same compound[247]. They identified three individual resonances in the 51VNMR spectra of the low-temperature phase of a-NaV2O5.These three resonances were attributed to three differentvanadium valences in the system. NMR data were com-pared with the structural models developed for thiscompound.

9.1.3. Catalysis

Heterogeneous vanadia-based catalysis has been a tradi-tional stronghold for solid-state 51V NMR spectroscopystarting with its earliest applications in late 1950s, almostimmediately after the discovery of the magnetic resonancephenomenon. We have reviewed the subject of 51V NMR incatalysis on several occasions, including our first compre-hensive review published in this very journal in 1992 [34].Another review appeared in the Encyclopedia of NuclearMagnetic Resonance in 1996 [44], and the most recentwas published in 2003 in Catalysis Today [45]. A consider-able number of research papers on this subject has beenpublished since then, many using solid-state 51V NMR astheir primary research tool. In this review, we will discussseveral of the many recent publications, focusing mostlyon those having methodological interest for developmentof the 51V NMR technique. Some examples are selectedto specifically illustrate recent NMR hardware develop-ments, including the availability of extremely high spinningspeeds for MAS in excess of 35 kHz, as well as the wideraccessibility of ultrahigh magnetic fields for solid-stateNMR.

Binary catalysts containing vanadia supported on inertsupports have been studied in detail not only by NMRbut by many other techniques, and most recently by com-puter modeling. The model shown in Fig. 41 is one of sev-eral introduced to describe surface vanadia species andadopted by many research groups.

talysts. Reproduced with permission from Ref. [6].


Although this particular model was developed forVOx/TiO2 catalysts, it has a more general applicability toother similar binary vanadia systems. Depending on thepreparation technique as well as on the surface concentra-tion of vanadia, the support nature, and the treatmentconditions, several distinct types of VOx species can beidentified and characterized by solid-state NMR.

Isolated or associated surface vanadia species are formedat low vanadia loadings, normally at loadings not exceed-ing a theoretical monolayer. The structure of these speciesdepends on the number of chemical bonds connectingvanadium to the support, and also on the treatment condi-tions. This can be the best illustrated by samples preparedvia chemical grafting. The grafting technique suggestsdirect chemical reaction between surface hydroxyl groupsof the support with vanadium compounds being deposited,usually VOCl3 or VO(OR)3:

VOCl3 þ nðHO–Ti–Þ ! VOCl3�nðOTi–Þnhere ‘‘n’’ could be 0, 1, 2, or 3. The resulting surface vana-dium species, their structure and the number of chlorineatoms left in this reaction can be identified in the 51VMAS NMR spectra by their isotropic chemical shifts andthe type and value of the chemical shielding anisotropies[248].

For example, gas-phase grafting of VOCl3 onto an ana-tase surface at room temperature leads to formation of tet-rahedral VOCl2(OTi–) species with each vanadium havingonly a single V–O–Ti bond with the surface and thetwo chlorine atoms in the first coordination sphere(diso = �350 ppm, gr = 0.2, dr = �230 ppm). DepositingVOCl3 at 110 �C results in stronger bonding of vanadiumto the surface as VOCl(OTi–)2 species, now with twoV–O–Ti bonds and only one chlorine atom in the first coor-dination sphere (diso = �433 ppm, gr = 0.4, dr = �220 ppm).Even stronger bonding of vanadium occurs at 250 �C,when VO(OTi–)3 species form (diso = �520 ppm, gr = 0.3,dr = 310 ppm, CQ = 2.2 MHz, gQ = 0.3).

-1800-1000-200

δ,ppm

init

+H2O

A

1

2

Fig. 42. (A) Static and (B) 35 kHz MAS 51V NMR spectra recorded at 9.4 Tcatalysts, A1 – after H2O adsorption, B1 – after catalytic reaction.

The next step of the catalyst preparation involves ahydration–dehydration procedure at 350 �C. Regardlessof the initial grafting temperature, the corresponding 51VMAS NMR spectra always showed three similar groupsof signals (Fig. 42B, spectrum 2):V1 with diso = �537 ppm,gr = 0.1, dr = 250 ppm, CQ = 2.1 MHz, gQ = from 0 to 1;V2 with diso = �570 ppm, gr = 0.9, dr = 230 ppm,CQ = 4.3 MHz, gQ = 0.7; V1 with diso = �645 ppm,gr = 0.3, dr = 600 ppm, CQ = 7 MHz, gQ = 0.4.

Only MAS spinning of these samples at 35 kHz (at 9 T)frees all isotropic lines from overlapping with MAS spin-ning sidebands. From their NMR parameters it is straight-forward to assign V1 to trigonal VO4 pyramids, V2 toassociated tetrahedral sites of Q2 type, and V3 to distortedoctahedral sites (see above). High-field isotropic shiftsfound for V3 sites are most likely due to strong interactionsbetween vanadium and TiO2, for example, via two or threeV–O–Ti bonds. At the same time, V1 sites should corre-spond to vanadium weakly bonded to TiO2, i.e. via onlyone or two V–O–Ti bonds. This assignment correlates wellwith the electronegativity of the atoms in the second coor-dination sphere. Isotropic 51V chemical shifts progresstowards more negative values as the electronegativity of asecond metal atom decreases. Consequently, the higherelectron density around vanadium in V3 species can beattributed to their stronger interaction with TiO2 thanthose of V2 and V1 species [249]. Taking into account com-plementary Raman data [250,251] it is plausible to suggestthat V3 species are monomeric distorted VO6 octahedrawith one short V@O bond which are interacting stronglywith the support.

Each group of signals in Fig. 42B, spectrum 2 in turnconsists of two to three closely overlapping lines. Thiscan be explained by vanadium interacting with OH groupslocated not only on the densest plane (001), but also oncleavage planes other than (00 1). Formation of at leastsix to nine non-/equivalent vanadium sites indicates thatthere should be also association of vanadium species:

δ,ppm

-700-600-500

ial

afterreaction

V1 V2 V3

1

2

B

for VOx/TiO2 catalysts prepared by grafting technique. A2, B2 – initial


VOCl2ðOTi–Þ ��!H2OVOðOHÞ2ðOTi–Þ ��!350 �C

VOðOHÞðOTi–Þ2��!350 �C

VOðOTi–Þ3VOClðOTi–Þ2 ��!H2O

VOðOHÞðOTi–Þ2 ��!350 �CVOðOTi–Þ3

2VOðOHÞðOTi–Þ2 ��!350 �Cð–OTiÞ2–VO–O–VO–ðOTi–Þ2ð–OTiÞ2–VO–O–VO–ðOTi–Þ2 þ VOðOHÞðOTi–Þ2��!350 �C

VOðOTi–Þ3–O–VOð–OTiÞ–O–VOðOHÞðOTi–Þ2

According to complementary 1H MAS NMR data[34,248,252], both bridging and terminal OH groups areinvolved in bonding VOx species, with on average twoV–O–Ti bonds found after the hydration–dehydrationprocedure.

During catalytic reactions both the structure and con-tent of surface sites undergo certain transformations. Thusunder DeNOx conditions the weakly bound V1 sites on theTiO2 surface change significantly, with their relative con-tent increasing, and the sites themselves becoming moresymmetric. At the same time, V2 sites are practically unaf-fected under these conditions. Similarly, only minorchanges can be detected for V3 sites (Fig. 42B, spectrum 1).

Grafting VOCl3 onto a silica surface with subsequenthydration and dehydration at T P 350 �C almost alwaysleads to strongly bound VO(OSi–)3 species having threeV–O–Si bonds with the surface. 51V NMR spectra ofdehydrated VOx/SiO2 samples prepared this way are char-acterized by an axially symmetric anisotropy of the CSAtensor, with parameters typical for trigonal pyramids, i.e.

Fig. 43. Wide-line 51V NMR spectra recorded at 9.4 T for VOx/SiO2 catalystCatalyst after H2O adsorption. Structures of vanadium sites in the dehydrapermission from Ref. [255].

diso = �710 ppm, d^ = �460 ppm, gr = 0.04, dr = 480 ppm,and CQ = 2.5 MHz (Fig. 43).

Upon hydration the surface vanadia species interactreadily with water molecules forming various surface com-plexes, i.e. VO3(OH), (VO3)n, VO2(OH)2, V10O26(OH)2,VO(OH)3, V10O27(OH), (V2O7)4�, or V10O28

6�, dependingon the net surface pH, as was demonstrated using Ramanspectroscopy by Deo and Wachs [253]. This agrees withrecent 51V NMR data, which have also shown that surfacecomplexes remain attached to the surface, though not asstrongly as in the dehydrated state [148,254,255]. Surfacevanadia species in hydrated VOx/SiO2 are similar to thosefound in V2O5ÆnH2O gels.

Based on 51V and 17O NMR data, a total of five differentvanadium species were identified and quantified in vanadiagels at various stages of hydration [148,254,255]. Compar-ative analysis of 17O MAS and 3QMAS NMR, 51V MASNMR, and thermogravimetric data has provided an addi-tional insight into the coordination states of water mole-cules during hydration, and led Fontenot et al. topropose a model for the gel structure. Upon rehydrationof the layered gel there is observed to be a single preferredsite for initial water readsorption. Oxygen atoms of thesereadsorbed water molecules are readily exchangeable withall other types of oxygen sites even at room temperature(Fig. 43).

The surface vanadia species described above have beenfound not only in the catalysts prepared by grafting, butalso in vanadia-based catalysts prepared by other tech-niques: impregnation, mechano-chemical activation, co-

s prepared by grafting technique. (1) Initial catalyst after dehydration. (2)ted catalyst and in the catalyst after H2O adsorption are adopted with


precipitation, or spray-drying. However, in all these cata-lysts other types of vanadia are also present in additionto surface species, including larger polyanions, stronglybound vanadium (SBV), binary phases, and V2O5.

Strongly bound vanadium (SBV) is often found in cata-lysts prepared via spray-drying of the mixture of TiO2

and vanadyl oxalate followed by a thermal treatment.Under these conditions a coherent interfacial boundary isformed between two crystalline phases, TiO2 and V2O5.According to HREM and EDAX data this interfacialboundary consists of closely spaced vanadium and tita-nium atoms in a 1:1 ratio forming a mixed-cation layer.The structural arrangement of this interface does notdepend on vanadia loading and is persistent in sampleseven after extraction of excess V2O5 and soluble compo-nents [101]. It also remains mostly intact during catalyticreactions and under water adsorption.

The structure of vanadium sites in this strongly boundvanadium phase was first studied with 51V NMR by Shubinet al. [100]. They have found that these vanadium speciesare characterized by unusually large 51V quadrupolar cou-pling constants, in the range from 14 to 16 MHz, whereasthe principal components of the CSA tensor are similarto those found in bulk V2O5. Examples of the 51V NMRspectra recorded for the SBV phase formed in VOx/TiO2

catalysts are shown in Fig. 44.It is interesting, that earlier theoretical studies also pre-

dicted a possibility of strong bonding and intergrowthbetween crystalline phases of anatase and vanadium pent-oxide [256]. According to these calculations and 51VNMR data, vanadium sites in the SBV phases are in dis-torted octahedral oxygen coordination similar to V2O5;yet the very large quadrupolar coupling constant, almost

4000 2000 0 -2000 -4000 -6000

initial

afterreaction

1

2

δ, ppm

-700-600-500

Fig. 44. 51V MAS NMR spectra recorded at 9.4 T for strongly boundvanadium sites in VOx/TiO2 catalysts prepared by spray-drying. (1)Experimental spectrum with the full set of spinning sidebands. (2)Simulated spectrum with the following NMR parameters: CQ = 4.1 MHz,gQ = 0.12, dr = 520 ppm, gr = 0.1, diso = �634 ppm, dispersion of distri-bution DgQ = 2 kHz. Inset shows isotropic lines in the spectra of the initialcatalyst and the catalyst tested in the DeNOx catalytic reaction.

an order of magnitude of that found in V2O5, indicatesconsiderable structural distortions.

Strongly bound vanadium has later been found in sam-ples prepared not only by spray-drying, but also in samplesprepared by many other techniques, including grafting,impregnation and mechano-chemical activation. However,51V NMR parameters found for SBV in all these systemsmay vary, most likely due to impurities in anatase, specificsof the preparation procedure, or in some cases due to thepresence of water. The differences are found mostly in thequadrupolar coupling constant [257], indicating that differ-ences among samples exist in the long-range order, whilethe short-range order reflected in the CSA parameters isusually close and is similar to V2O5. The most symmetricSBV species are found in samples prepared by grafting.

In VOx/TiO2 samples prepared via ball milling followedby thermal treatment the parameters of the stacking geom-etry are quite different from all other studied systems. TheSBV phases in these samples also lack the coherent fittingbetween lattices of the different phases with rising of thestrain dislocations at the boundary. 51V NMR experimentsof these samples revealed two different types of octahe-drally coordinated vanadium species [258]. More distortedspecies are formed predominantly during milling, whereasmore symmetric species are found after thermal treatment.In both cases the distortion is less axial than in the bulkV2O5.

The structure of SBV species also depends on the sup-port used. Thus on tetragonal ZrO2 a variety of sites canbe found but all of them have a tetrahedral coordination.51V 3QMAS NMR spectra indicate the presence of at least10 non-equivalent vanadium sites in these samples(Fig. 45).

Strongly bound vanadium species are also formed onmonoclinic ZrO2, and Nb2O5, but not on SiO2.

-1000

-800

-600

-400

-200

0

ppm

-500 -550 -600 -650δ, ppm

δ Id,

ppm

Fig. 45. 51V 3QMAS NMR spectrum of a VOx/ZrO2 catalyst prepared byimpregnation. The spectrum was obtained at 9.4 T.

1000 0 -1000 -2000

δ, ppm

1

2

3

4

5

x2.3

6

Fig. 46. 51V MAS NMR spectra recorded at 9.4 T for VOx/Al2O3

catalysts prepared by impregnation with vanadyl oxalate. (1) Spectrum ofthe initial 25 wt% V2O5/Al2O3 sample. (2) Spectrum of the sample afterextraction of soluble components. Also shown sub-spectra representing (3)V2O5, (4) V10O28

6�, (5) AlVO4, and (6) SBV species.


Large vanadium-containing species like decavanadates,V10O28

6�, and aggregates of V2O5, are formed on the sur-face either when vanadium loading exceeds the monolayercoverage, or when dispersion of vanadia is not sufficientlyhigh. Decavanadates can only be unambiguously detectedin 51V MAS NMR spectra recorded under very fastmagic-angle spinning, in excess of 30–35 kHz. At ambientconditions V10O28

6� species were detected in almost allstudied VOx/Al2O3 catalysts with vanadium contentexceeding the monolayer coverage. One such example of

0 -1000 -2000δ, ppm

-1000 -1500

A

1

2

3

1

2

3

Fig. 47. (A) Static 51V NMR spectra recorded at 9.4 T for (1) (V2O5–WO3)/TiOand (3) V2O5. Inset at the top shows fine details in the satellite transitions. (B) Sconstant DCQ at around 799 kHz according to the Gaussian function gðDQ = 50 kHz, (4) DQ = 0 kHz. Inset at the top shows fine details in the satelli

a 51V MAS NMR spectrum recorded for a VOx/Al2O3 cat-alyst is shown in Fig. 46. Several vanadium sites can beidentified in this spectrum by deconvoluting it into sub-spectra corresponding to surface tetrahedral sites,V10O28

6� species, V2O5 and AlVO4. It is noteworthy, thatafter extraction of soluble components, the AlVO4 phaseis still retained in the sample together with SBV sites(Fig. 46, spectrum 2).

When V2O5 is formed on the surface, its structure isalways distorted due to structural defects in the surface.This can be seen particularly well by observing the broad-ening of the satellite transitions as shown in the static spec-tra in Fig. 47. The parameters of the magnetic shielding arepractically not affected.

Binary phases may form under certain conditions uponinteraction of vanadium with the substrate. For example,when Al2O3, Nb2O5 or ZrO2 are used as supports, the for-mation of AlVO4, NbVO5, and ZrV2O7 can be observed(Figs. 46 and 48) [259]. Titania and silica supports do notnormally form binary compounds with vanadium.

Third component, i.e. an impurity in the starting mate-rials or when a modifier agent is added, can considerablyalter the structure of vanadium species. For example,when phosphorus is present during spray-drying, thestructure of the resulting SBV species is affected. Herephosphorus is often being incorporated into the finalproduct with the molar ratio of V/P/Ti close to 1:1:1.Vanadium in these V–P–Ti species is found in the tetra-hedral coordination typical for isolated or weakly associ-ated tetrahedra (Fig. 49) [180]. New compounds havealso been found in VOx/TiO2 catalysis modified withsodium [260].

Bulk vanadium-containing catalysts based on bismuthvanadate, Bi4V2O11, are promising candidates for applica-tions in the oxidative coupling of methane. Bismuth vana-date itself shows unusually high anionic conductivity

0 -1000 -2000

-1000 -1500

1

32

4

1

32

4

B

δ, ppm

2 catalysts after catalytic reaction, (2) starting (V2O5–WO3)/TiO2 catalyst,imulated static 51V NMR spectra with the distribution of the quadrupolarxÞ ¼ ð1=Dp2pÞ expð�x2=2D2

QÞ. (1) DQ = 100 kHz, (2) DQ = 75 kHz, (3)te transitions.

Fig. 48. 51V MAS NMR spectra recorded at 9.4 T for (A) VOx/Nb2O5 and (B) VOx/ZrO2 catalysts. Vanadium loading was 2 wt% in (1), 4 wt% in (2) and8 wt% in (3). The isotropic lines are marked with asterisks. Reproduced with permission from Ref. [259].

-1000-5000 -1500

1

2

3

δ, ppm

Fig. 49. 51V NMR spectra obtained at 9.4 T for a 5%P2O5/10%V2O5/TiO2

catalyst prepared by spray drying. (1) Stationary sample. (2) 10 kHz MAS.(3) 30 kHz MAS.


sought for advanced composite materials. Not surprisingly,these vanadium systems are currently under scrutiny ofteninvolving solid-state 51V NMR.

According to 51V NMR data reported by Delmaire et al.[189] there are at least three different vanadium sites ina-Bi4V2O11, two of these sites are tetrahedral and one isin bipyramidal trigonal or has a distorted octahedral oxy-gen coordination. The majority of the vanadium (about70%) occupies regular tetrahedral sites and the rest is foundin bipyramidal trigonal or distorted octahedral sites.Reduction of a-Bi4V2O11 under a flow of hydrogen gas at330 �C results in progressive conversion of V5+ into V4+.At first a broad signal due to paramagnetic V4+ ions

appears in the 51V NMR spectra underneath rather narrowresonances from the starting material a-Bi4V2O11. Follow-ing further reduction a new resonance at �1447 ppm withmultiple MAS sidebands becomes apparent. This reso-nance is dominant in Bi6V3O16, the final product of bis-muth vanadate reduction (Fig. 50). In this compound theratio V5+/V4+ = 2/1. The room-temperature crystal struc-ture of Bi6V3O16 can best be described as ribbons ofV3O10 units in which one V4+O6 octahedron shares equato-rial corners with two V5+O4 tetrahedra intercalatedbetween sheets. The paramagnetic shift observed in 51VNMR spectra of Bi6V3O16 was attributed to strong interac-tions between V5+ ions and unpaired electrons of V4+ ions.

Summarizing the results outlined above we would like toonce more accentuate the idea that the 51V NMR techniqueis a unique research tool indispensable for studying van-adia catalysts and related materials. The possibility to clo-sely follow the formation and transformations ofvanadium-containing centers during synthesis and undercatalytic conditions makes it possible to purposefullydesign catalysts with well-defined and, more importantly,desirable properties.

9.2. Applications of solid-state 93Nb NMR

While 51V NMR finds most of its applications in catal-ysis, 93Nb NMR is gaining in popularity in studying mate-rials for advanced electronic devices and semiconductors.In these challenging systems 93Nb solid-state NMR demon-strates a great potential to provide both qualitative andquantitative information about chemical environments,cation ordering, and the motional behavior of cations. Sev-eral 93Nb NMR examples in solids have already been dis-cussed above, and some applications have recently beenreviewed [55].

Fig. 50. 51V MAS NMR spectra recorded at 9.4 T for Bi4V2O11 reduced under a hydrogen gas flow at 330 �C for (1) 10 h and (2) 16 h. The isotropic linesof different vanadium sites are marked as 1, 2, 3. Reproduced with permission from Ref. [189].


9.2.1. Characterization of ferroelectrics

Niobia-based ferroelectric materials are currently thesubject of intensive research, with several groups reportingtheir 93Nb NMR findings, usually complemented by resultsfrom other research techniques, including neutron and X-ray-diffraction experiments.

Thus polycrystalline solid-solution relaxor ferroelectricsPb(Mg1/3Nb2/3)O3 (PMN) and (1�x)Pb(Mg1/3Nb2/3)O3/xPbTiO3 (PMN/PT) have been studied using 93Nb MASand nutation NMR at 9.4 and 14.1 T [47]. The authors

Table 1093Nb resonances found in PMN materials from 93Nb MAS and nutationNMR experiments [47]

Site 1 2 3

dobs (93Nb, MAS),(ppm)

�900(narrow)

�954 to �980 (broad)

CQ (93Nb, nutation),(MHz)

<0.8 �17 >62

B-site symmetry Cubic Tetragonal RhombicAssignment Nb(OMg)6 Nb(ONb)6�x(OMg)x

(x = 1–5)

obtained accurate atomic-level information about the localstructure and chemical disorder of Nb cations occupyingB-sites. In these and similar systems 93Nb MAS and nuta-tion NMR spectra usually show three 93Nb resonanceswith drastically different quadrupolar coupling constants,assigned to Nb sites with cubic, axial and rhombic localsymmetry of the neighboring Mg/Nb configurations. Typ-ical 93Nb NMR parameters for these three sites are summa-rized in Table 10. Three 93Nb resonances in PMN were alsoobserved by Prasad et al. [48], and by Cruz et al. [49].

According to Fitzgerald et al. [47] the Nb sites corre-sponding to sharp 93Nb signals at �902 ppm are mostlikely associated with cubic Nb(OMg)6 configurations inMg-rich ordered-domain regions. Two broader 93NbNMR resonances with intermediate and high CQ valueswere assigned to two types of lower symmetry NbO6 octa-hedra with local configurations of Nb(ONb)6�x(OMg)x

type, where x may vary from 1 to 5. One set of configura-tions has axial local symmetry, and another has lowerrhombic symmetry.

93Nb NMR allows the monitoring of compositionalchanges in PMN/PT solid-solutions. At low PT content,


Ti ions substitute both Mg and Nb in the B-sites of PMN.At increased PT content (>10 mol%) the simultaneouspresence of three cations Mg, Ti, and Nb randomlyoccupying B-sites leads to increased structural disorder ofthe B-sites. This causes the sharp 93Nb signal fromnear-cubic Nb(OMg)6 sites to almost disappear, andfurther changes to occur in the lineshapes from theother two resonances assigned to axial and rhombicNb(ONb)6�x(OMg)x sites in Nb-rich regions of PMN. Tisubstitution for Nb in both Mg-rich and Nb-rich regionsof PMN was found to be nonselective. At the same timethere are distinct differences between Nb B-sites locatedin Mg- and Nb-rich regions in PMN and PMN/PT materi-als at low PT contents [47].

A comparative study of lead-based perovskite relaxorferroelectrics, Pb(Mg1/3Nb2/3)O3 (PMN), substituted byZr, Sc, and Ba (PZN, PSN, PBN correspondingly), haverecently been reported to show differences in the localchemical environments, in the relative degree of B-siteordering, and in the motional behavior of the B-site cat-ions [51]. 93Nb MAS NMR spectra provided conclusiveevidences for significant B-site disorder in PMN, whileB-site ordering was found in scandium-substituted(PSN) materials. In agreement with earlier studies 93NbMAS NMR spectra of PMN indicate the presence of sev-eral resonances with very different CQ values, which wereassigned to different types of Nb B-sites as in pseudo-cubic, axial and rhombic symmetry. At the same time inPSN there was found to be a predominantly single Nb sitewith an intermediate value of CQ, which was assigned toaxially distorted B-sites. Substitution by Ba results inincreased B-site ordering in the Mg-rich and Nb-rich nan-odomains of PMN.

Scandium substituted ferroelectrics (PSN) have alsobeen studied by Blinc et al. [52,261]. 93Nb NMR spectrapresented there suggest a homogeneous structure of thePMN and PSN phases studied. Both structures can be con-sidered within a spherical random bond-random fieldmodel [262] where disorder of the Pb2+ sites is representedby the order parameter described as a continuous vector ofvariable length. While studying similar PSN materials,Laguta et al. [263] reported excellent agreement betweendata obtained from neutron and X-ray-diffraction experi-ments and via multinuclear solid-state NMR. In this work207Pb, 45Sc, and 93Nb NMR spectra of partially orderedPSN were obtained at temperatures between 77 and420 K to reveal a first-order phase transition at 360 K. Thistransition was particularly straightforward to detect insomewhat narrower 45Sc and 207Pb NMR spectra originat-ing from chemically ordered regions of the crystal, but notin the broader 93Nb NMR spectra.

9.2.2. Silicates

Silicates containing niobium are quite often studied with93Nb NMR, usually in combination with other nucleiNMR, for example, 29Si. Thus the structure of nenadkevi-chite minerals, Na0.9K0.06Ca0.03 Nb0.7Ti0.3Si2O6.7(OH)0.3Æ

2H2O, was characterized by multinuclear 23Na, 29Si, and93Nb high-resolution solid-state NMR [41]. In agreementwith the crystal structure reported for this mineral, 93Nbsolid-state NMR suggests that both titanium and niobiumare in a distorted octahedral coordination. Due to strongerdipolar interactions, replacing low-abundant 47Ti and 49Tiwith 93Nb having high natural abundance results in a sig-nificant decrease in the 29Si spin–spin (T2) NMR relaxationtimes.

Static 93Nb NMR spectra have been reported for a seriesof (Nb2O5)x(SiO2)1�x materials prepared via a sol-gel tech-nique [42]. The authors suggested that niobium is incorpo-rated into silica as a four-coordinated species. Thisconclusion is somewhat questionable, however, since notonly are the reported 93Nb NMR spectra very similar tothose normally found for Nb2O5, but also peaks corre-sponding to bridging Nb–O–Nb groups can be clearly seenin the 17O NMR spectra. The reported isotropic 93NbNMR chemical shifts values of about �1200 ppm are alsoclose to those characteristic of six-coordinated niobium. Itappears that niobium was indeed incorporated into SiO2,but not as the isolated tetrahedral Nb species the authorsstipulated.

A combination of 93Nb and 29Si solid-state NMRand X-ray diffraction data has provided a correlationbetween NMR parameters and the local structure inRb4(NbO)2(Si8O21) silicate [126]. For the first time heter-onuclear two-bond J-coupling between a quadrupolarnucleus (93Nb) and a spin-1/2 nucleus (29Si) was observedin the solid state. This work has opened up new opportu-nities for investigating possible relationships betweenJ-coupling and structural parameters in solids.

9.2.3. Miscellaneous applications

Among some other recent applications of 93Nb NMR inthe solid state we would like to bring our readers attentionto research by Du et al. [54] who applied modern NMRtechniques to study local structures and oxygen/fluorineordering in Cdpy4NbOF5 (py = C5H5N) and [pyH]2[Cd-py4(NbOF5)2] compounds. Their 93Nb NMR spectra wereacquired at ultrahigh magnetic field of 19.6 T and MASunder ultrahigh spinning speeds (43 kHz). The 93NbNMR spectra of both compounds are dominated not onlyby quadrupolar but also by the chemical shift interactions,consistent with a highly asymmetric Nb environment(Fig. 51). To extract the nuclear quadrupolar coupling con-stant CQ, the asymmetry parameter gQ, and the isotropicchemical shift diso in the presence of the large CSA, theauthors applied the original methodology as described indetail in Section 3.5. The values of the 93Nb NMR param-eters found for these two compounds are summarized inTable 7.

More recently Lo et al. [264] reported 93Nb NMR datafor a series of half-sandwich niobium metallocenesCp 0Nb(I)(CO)4 and CpNb(V)Cl4. They have found the93Nb chemical shifts for Nb(I) metallocenes ranging from�1900 to �2050 ppm, and for Nb(V) metallocenes from

Fig. 51. (A) 93Nb NMR spectra of Cdpy4NbOF5. (B) 93Nb NMR spectra of [pyH]2[Cdpy4(NbOF5)2]. (1) Stationary samples at 8.5 T. (2) 20 kHz MAS at8.5 T. (3) 43 kHz MAS at 19.6 T. Reproduced with permission from Ref. [54].


�600 to �715 ppm. The observed difference in the chemicalshift values for two Nb oxidation states was explained byshielding effects resulting from the Cp 0-metal p-bondinginteractions. Although metallocenes are not the subject ofthis review, this paper is a good example of a well-executedand comprehensive solid-state 93Nb NMR study.

Another interesting recent case involving paramagneticvanadium species was reported by Ponce et al. [43]. Whilestudying 51V and 93Nb NMR spectra of HxNbVO5

(x = 1.4) bronze, they found no differences in the chemicalshift and the line width of the central transition in this com-pound with respect to those observed in NbVO5. Fromthermodynamic calculations the charge transfer fromhydrogen to the host lattice in HxNbVO5 appears to becomplete. Thus the formal oxidation state of vanadium inthis bronze should be V3.5+ and the abundance of V4+

and V3+ paramagnetic centers should have preventedobservation of 51V and 93Nb resonances. The reportedNMR results are difficult to explain within the thermody-namic model of the complete charge transfer, unless someelectron-pairing occurs. Solving this problem may proveto be a challenging task requiring additional experimentaland computational techniques to be involved.

93Nb NMR spectra of a mixed-cation KTa(1�x)NbxO3

(KTN) single crystal have been recorded by Runkel et al.[265]. The spectra consist of two components, a sharp cen-tral +1/2 M �1/2 transition along with an unresolvedbackground from quadrupolar-induced first-order satel-lites. These spectra were interpreted as being from Nb ionsoccupying off-center positions rather than from those in thehigh symmetry central perovskite site. The angular depen-dence of the second moments of the satellite backgroundfurther showed that these distortions are of rhombohedralsymmetry, i.e. Nb ions are indeed displaced along the [111]body diagonals. Another recent single-crystal 93Nb NMRstudy involved measurements of the 93Nb EFG parametersin Nb-doped single TiO2 crystals [266].

93Nb and 27Al NMR studies of composites between par-tially hydrolyzed aluminum cations and layered calcium

niobate perovskites suggest the presence of structurallyunaltered Ca2Nb3O10 layers in these materials with inter-layer regions occupied by polymeric Al hydroxylcations[36].

9.3. Multinuclear solid-state NMR in vanadia and niobia

catalysts supported on Al2O3

This example of solid-state 51V and 93Nb NMR is per-haps one of the best to fully illustrate the advantages of thistechnique for studying complex multi-component systemshaving tremendous practical importance. The multinuclearNMR approach, involving simultaneous use of both 51Vand 93Nb nuclei, has allowed us to characterize these cata-lysts in great detail, and to provide a basis for improvingtheir performance in real practice. In the next few para-graphs we will summarize results of several years ofresearch and the efforts of many of our colleagues.

9.3.1. Vanadia sites in VOx/Al2O3

As it has been mentioned above, several types of van-adia sites can be identified by 51V NMR in supported van-adia catalysts. In VOx/Al2O3 systems these include weaklyand strongly bound surface species, AlVO4, large V10O28

6�

anions, and distorted V2O5 [267]. The relative amounts ofthese sites on the surface depend on the preparation proce-dure, the total vanadium concentration, humidity, and pHof the surface. What is important in this context is thateach of these sites has its own distinct 51V NMR signature(Fig. 52).

Tetrahedral VO4 species weakly bonded with the surfacevia one or two bonds have similar static and MAS 51VNMR spectra with isotropic shifts found in the range from�540 to �580 ppm. The NMR spectra of stronger boundVO4 species are shifted upfield and usually consist ofbroader isotropic lines at �620 ppm. Large vanadia poly-anions V10O28

6� on alumina surface are characterized bystatic 51V NMR spectra typical for an axially symmetricchemical shielding anisotropy. At high MAS spinning

0 -500 -1000 -1500

-1500-1000-500400 0

δ, ppm

V2O5

AlVO4

V10O286-

OV

OOO

V2O5, 10.8 wt.%

VOx/Al2O3

Fig. 52. (Left) 51V MAS and static NMR spectra at 9.4 T of different vanadium species formed in VOx/Al2O3 catalysts as indicated. (Right) Relativecontent of different vanadium species in a sample containing 10.8 wt% of V2O5. The experimental 51V MAS NMR spectrum of this sample is shown on thetop.


speeds of 35 kHz these spectra can be resolved into threeresonances with isotropic shifts diso of �420, �490, and�510 ppm. Static and MAS 51V NMR spectra of theAlVO4 phase consists of three relatively narrow overlap-ping lines from three non-equivalent vanadium sites in tet-rahedral oxygen coordination with an isotropic shifts diso

of �668, �747, and �780 ppm. The V2O5 phase on the sur-face is identified by an axially symmetric line in the static51V NMR spectra with d^ = �310 ppm, di = �1250 ppm.Under MAS conditions this spectrum transforms into asingle set of narrow spinning sidebands with isotropic shiftof �610 ppm. Some distribution of NMR parametersreflects a defect structure of this phase on the surface.

At low vanadium loading, only tetrahedral species(weakly and strongly bound to the surface) are formedon the alumina surface. With increasing vanadium contentthe relative concentration of these sites decreases, at thesame time decavanadate V10O28

6� species are becomingprevalent. Further increase in the vanadium content resultsin formation of AlVO4 and V2O5 phases. The relative con-tent of each type of vanadia species can be determined byanalyzing the corresponding 51V NMR spectra as shownin Fig. 52 for a sample containing 10.8 wt% of V2O5.

Upon dehydration the surface vanadia species inVOx/Al2O3 undergo certain transformations, which caneasily be monitored with 51V NMR. Thus tetrahedral VO4

species (weakly and strongly bound to the surface) becomemore distorted in the dehydrated state. Decavanadate spe-cies are converting into significantly distorted tetrahedralsites of both types, with the ratio between them dependingon the initial content of V10O28

6�, i.e. the share of stronglybound species increases with increase in V10O28

6�. BothAlVO4 and V2O5 phases adopt a more regular structureupon dehydration. The relative ratio of different vanadium

sites in the dehydrated samples depends on the relative ratioof V sites in the corresponding hydrated samples.

According to 27Al MAS NMR recorded for the samecatalysts there are no strong interactions between V andAl in VOx/Al2O3 samples at different vanadium loadings.Only at the very high vanadium content, when the AlVO4

phase is formed, the new 27Al MAS NMR line appearsat �10 ppm corresponding to AlVO4 species. The lattercan be better seen in corresponding 3Q MAS 27Al NMRspectra.

Interaction of vanadia with alumina in VOx/Al2O3 canalso be monitored by 1H MAS NMR of hydroxyl groups.Vanadia species interact with the surface of Al2O3 viabridged OH groups (d1H 1.6 ppm) and terminal OH groups(d1H �0.2 ppm). At the same time hydrogen-bondedhydroxyl groups (d1H 7.4 ppm) and bulk OH groups(d1H 3.7 ppm) remain largely intact. These data indicatehigh dispersion of vanadia species over the Al2O3 surface.At the same time there is no noticeable change in the bulkAl2O3 structure.

9.3.2. Niobia sites in NbOx/Al2O3

As discussed above and by Lapina et al. [55] the 93NbNMR chemical shift is sensitive to Nb coordination inoxide materials. Representative static 93Nb NMR spectrafor a series of NbOx/Al2O3 samples recorded at 21.1 Tare shown in Fig. 53. This is perhaps the first time suchspectra have been reported for NbOx/Al2O3 catalysts. Allrecorded spectra are rather broad with no narrowingobserved under MAS. The broadening is most likelycaused by both the quadrupolar interaction and by thechemical shift distribution. When recorded at differentmagnetic fields, the line width measured in parts per millionscales inversely with the magnetic field strength, but only

δ, ppm-3000-1000500

3

2

1

Fig. 53. 93Nb NMR spectra of stationary NbOx/Al2O3 samples recordedat 21.1 T. (1) 4% Nb2O5/Al2O3. (2) 8% Nb2O5/Al2O3. (3) 16% Nb2O5/Al2O3.


linearly and not as a square function expected for purequadrupolar broadening (see above). At low niobia loadingthe coordination number of Nb sites in NbOx/Al2O3 isclose to six, i.e. as found in octahedral NbO6 species. Athigher Nb loadings a new 93Nb NMR resonance can beseen in the spectra recorded at 21.1 T. By the value of itschemical shift this resonance can be attributed to seven-or eight-coordinated Nb sites. At lower magnetic fields

OV

OOO

OV

OOO

OV

OOO

NbAl

NbAl

Nb2O

Al2O3

Al2O3

Al2O3

Low niobia loa

High niobia co

Al2O3

Al2O3

Al2O3

OV

OOO 10

V2O5

OV

OOO

V2O5

OV

OOO

AlVO4

V o

r N

b lo

adin

g

Vanadi

Fig. 54. Surface species formed in VOx/Al2O3, N

the spectra are too broad for this new resonance to beresolved.

As was mentioned above, in 27Al MAS NMR spectra ofVOx/Al2O3 catalysts we normally do not see strong interac-tions between vanadia and alumina. However, this is notthe case for niobia catalysts. In similar 27Al MAS NMRspectra recorded for NbOx/Al2O3 samples there wasdetected a very characteristic 27Al resonance attributed tofive-coordinated Al sites [268]. The intensity of this lineincreases with Nb content. It is reasonable to suggest stron-ger interactions of niobia with alumina, with possibleincorporation of Nb into alumina. The Nb coordinationnumber increases with loading, thus indicating interactionsamong neighboring Nb species, and this supports theisland model of niobia deposition on the alumina surface.

Additional support for the island model can be found in1H MAS NMR spectra recorded for the same NbOx/Al2O3

samples. Supported niobia species indeed interact with thesurface hydroxyl groups as vanadia species do. However, atsimilar surface concentrations of niobia and vanadia thecontent of residual OH groups is always considerablyhigher in NbOx/Al2O3 than in VOx/Al2O3 samples, appar-ently due to higher dispersion of vanadia species on the sur-face, while niobia species tend to form aggregates.

9.3.3. Niobia–vanadia species in (Nb–V)Ox/Al2O3

Vanadia and niobia can simultaneously be depositedonto alumina to prepare mixed (Nb–V)Ox/Al2O3 samples.The nature of niobia and vanadia species in such systems is

NbAl

NbAl

5

ding

verage

NbAl NbAl

Al2O3

Low niobia loading

OV

OOO

OV

OOO

OV

OOO

a feels niobia, but niobia is not affected by vanadia

bOx/Al2O3, and (Nb–V)Ox/Al2O3 catalysts.


of great interest, since they often show a synergetic cata-lytic behavior by exceeding the performance of individualsingle-component VOx/Al2O3 and NbOx/Al2O3 catalysts.The following conclusions about (Nb–V)Ox/Al2O3 can bedrawn from multinuclear NMR experiments.

First of all, the new 51V NMR line with the isotropicchemical shift of �650 ppm appears in 51V NMR spectrain the presence of niobia. The relative intensity of this lineincreases with niobia loading, and becomes even more pro-nounced in dehydrated samples. It is reasonable to attri-bute this line to (Nb–V)Ox species, though it is not yetNbVO5. At the same time there are only very minorchanges in the corresponding static 93Nb NMR spectra.27Al MAS NMR spectra recorded for (Nb–V)Ox/Al2O3

are similar to those for NbOx/Al2O3, with no strong inter-actions detected between vanadia species and the aluminasupport even at high Nb and V loadings. An almost lineardecrease in the surface concentration of OH groups with(Nb–V)Ox loading is observed in the 1H MAS NMRspectra.

All these NMR findings can be explained by the islandmodel introduced above for the NbOx/Al2O3 system. First,the niobia precursors interact strongly with the aluminasurface forming islands of mixed (Nb–Al) oxides. The van-adia precursors then interact with both oxides and disperserandomly on the surfaces of Al2O3 and the mixed (Nb–Al)oxide islands. Schematically this is shown in Fig. 54.

Thus, the multinuclear NMR approach has allowed usin this case not only to identify various surface species inthese catalysts, but also to suggest the most likely modelof the surface structure. Testing these catalysts in catalyticreactions would provide further input in understandingstructure–properties relationships in these complexsystems.

10. Conclusions

We have reviewed practical aspects of solid-state NMRspectroscopy of Group VB elements pertaining to studyinga large variety of oxide materials. Some of these materialsalready have tremendous importance in large-scale indus-trial applications, as for example, vanadia-based systemsin heterogeneous catalysis, other materials are currentlybeing investigated for possible future uses in electronicsand advanced composites. Solid-state NMR offers a uniqueand indispensable tool for learning in great detail about thestructure and the underlying properties of these materials.While 51V and 93Nb NMR spectroscopy has already estab-lished itself as a popular research vehicle, there is no doubtthat 181Ta NMR is also about to show its great potential.The availability of advanced NMR hardware, includingultrahigh magnetic fields and ultrahigh-speed MAS probeswill make 181Ta NMR experiments on solids feasible andwill further expand the borders of 51V and 93Nb NMRapplications. We have compiled perhaps the most compre-hensive database to date on 51V and 93Nb NMR parame-ters in oxide materials, yet we understand that this

database is destined to grow as more and more databecome available every day. We hope this work will con-tribute to ever continuing progress in development of novelsolid-state NMR techniques for quadrupolar nuclei. Mod-ern NMR spectroscopy combined with other complemen-tary spectroscopic and computational approaches notonly advances our fundamental knowledge in material sci-ence and related disciplines, but also helps us to developbetter technologies and to discover new ways of improvingeveryday life.

Acknowledgements

Our research in this field has been generously supportedby the Russian Foundation for Basic Research, most re-cently by a RFBR Grant 07-03-00695-a. This review wouldnot be possible without our many collaborators, coauthors,and colleagues to whom we are extremely grateful. We ded-icate this review to the memory of our dear teacher andfriend the late Prof. V.M. Mastikhin. We thank Prof. P.Bodart (Universite des Sciences et Technologies de Lille),Dr. K.V. Romanenko (Boreskov Institute of Catalysis,Novosibirsk), Prof. J.-B. Espinose (ESPCI, Paris), Dr.Z.H. Gan (NHMF Laboratory, Florida), Prof. J.-P.Amoureux (Universite des Sciences et Technologies deLille), and Prof. M. Guelton (Universite des Sciences etTechnologies de Lille) among others for their expertise incarrying out some of the experiments and for helpful dis-cussions. Many individual vanadium and niobium com-pounds have been skillfully synthesized by Prof. V.N.Krasilnikov (Institute of Solid State Chemistry, Ekateribn-urg), Prof. M.G. Zuev (Institute of Solid State Chemistry,Ekaterinburg), Prof. V.L. Volkov (Institute of Solid StateChemistry, Ekaterinburg), and Prof. V.E. Fedorov (Niko-laev Institute of Inorganic Chemistry, Novosibirsk). Re-search on catalytic systems have been done over manyyears in close collaboration with Prof. M. Bonares (Institu-to de Catalisis y Petroleoquimica, Madrid), Prof. B. Grzy-bowska (Institute of Catalysis and Surface Chemistry,Krakow), Prof. I. Wachs (Lehigh University, Bethlehem,USA), Prof. Dr. H. Knozinger (Ludwig-Maximilians-Uni-versitat Munchen), Prof. R. Fehrmann (Technical Univer-sity of Denmark, Lyngby), Prof. Z. Sobalik (HeyrovskyInstitute of Physical Chemistry, Prague), Dr. V.M. Bond-areva (Boreskov Institute of Catalysis, Novosibirsk), Dr.L.G. Pinaeva (Boreskov Institute of Catalysis, Novosi-birsk), Prof. G.A. Zenkovets (Boreskov Institute of Catal-ysis, Novosibirsk), and Dr. L.G. Simonova (BoreskovInstitute of Catalysis, Novosibirsk). Some NMR spectraat 21.1 T were recorded at the Canadian National Ultra-high Field NMR Facility for Solids (Ottawa, Canada), anational research facility funded by the Canada Founda-tion for Innovation, the Natural Sciences and EngineeringResearch Council of Canada, the Ontario InnovationTrust, Recherche Quebec, the National Research CouncilCanada, and Bruker BioSpin, and managed by the Univer-sity of Ottawa (www.nmr900.ca).

http://www.nmr900.ca


References

[1] G.K. Boreskov, Catalysis in the Production of Sulfuric Acid,Goskhimizdat, Moscow, 1954 (in Russian).

[2] G.K. Boreskov, Heterogeneous Catalysis, Nauka, Moscow, 1987 (inRussian).

[3] D.J. Hucknall, Selective Oxidation of Hydrocarbons, AcademicPress, London, 1974.

[4] B.E. Leach, Applied Industrial Catalysts, Academic Press, New-York, 1983.

[5] A. Andersson, J. Catal. 69 (1981) 465.[6] G. Centi, Appl. Catal. A 147 (1996) 267.[7] T. Machej, P. Ruiz, B. Delmon, J. Chem. Soc., Faraday Trans. 1

(1990) 731.[8] F. Trifiro, I. Pasquon, J. Catal. 12 (1968) 412.[9] O.V. Krylov, Catalysis by Nonmetals, Academic Press, New York,

1970.[10] K. Kenneth, C.-P. Joao, C.C. Debbie, Vanadium: The Versatile

Metal, American Chemical Society, Washington, 2007.[11] M.A. Banares, I.E. Wachs, R.M. Martin-Aranda (Eds.), in: Fourth

International Symposium on Group Five Compounds, Toledo,Spain, 9–12 April 2002, UNED, Madrid, 2002.

[12] I.E. Wachs (Ed.), in: Fifth International Symposium on Group FiveCompounds, Hancock, Massachusetts, USA, 18–21 May 2005,Lehigh University, USA, 2005.

[13] K. Tanabe, Catal. Today 78 (2003) 65.[14] W.D. Knight, V.W. Cohen, Phys. Rev. 76 (1949) 1421.[15] R.E. Sheriff, D. Williams, Phys. Rev. 82 (1951) 651.[16] L.H. Bennett, J.I. Budnick, Bull. Am. Phys. Soc. 4 (1959) 417.[17] J.L. Ragle, J. Chem. Phys. 35 (1961) 753.[18] S.D. Gornostansky, G.V. Stager, J. Chem. Phys. 46 (1967) 4959.[19] S.D. Gornostansky, C.V. Stager, J. Chem. Phys. 48 (1968) 1416.[20] G.E. Peterson, J.R.J. Carruthers, J. Solid State Chem. 1 (1969) 98.[21] R. Kind, H. Granicher, Solid State Commun. 6 (1968) 439.[22] R.N. Pletnev, N.A. Gubanov, A.A. Foiev, NMR in Oxide Vana-

dium Compounds, Nauka, Moscow, 1979 (in Russian).[23] J.F. Baugher, P.C. Taylor, T. Oja, P. Bray, J. Chem. Phys. 50 (1969)

4914.[24] S.L. Segel, R.B. Creel, Can. J. Phys. 48 (1970) 2673.[25] R.N. Pletnev, V.A. Gubanov, A.K. Chirkov, Zh. Struk. Khim. 17

(1976) 938 (in Russian).[26] R.N. Pletnev, Zh. Neorg. Khim. 51 (1977) 2359 (in Russian).[27] G.E. Peterson, A. Carnevale, J. Chem. Phys. 56 (1972) 4848.[28] E.N. Ivanova, A.V. Yatseako, N.A. Sergeev, Solid State Nucl.

Magn. Reson. 4 (1995) 381.[29] J. Blumel, E. Born, T. Metzger, J. Phys. Chem. Solids 55 (1994) 589.[30] F. Wolf, D. Kline, H.S. Story, J. Chem. Phys. 53 (1970) 3538.[31] A.F. McDowell, M.S. Conradi, J. Haase, J. Magn. Reson. 119

(1996) 211.[32] V.M. Mastikhin, O.B. Lapina, V.N. Krasilnikov, A.A. Ivankin,

React. Kinet. Catal. Lett. 24 (1984) 119.[33] V.M. Mastikhin, O.B. Lapina, L.G. Simonova, React. Kinet. Catal.

Lett. 24 (1984) 127.[34] O.B. Lapina, V.M. Mastikhin, A.A. Shubin, V.N. Krasilnikov, K.I.

Zamaraev, Prog. Nucl. Magn. Reson. Spectrosc. 24 (1992) 457.[35] J. Davis, D. Tinet, J.J. Fripiat, J.M. Amarillo, B. Casal, E. Ruiz-

Hitzky, J. Mater. Res. 6 (1991) 393.[36] S. Hardin, D. Hay, M. Millikan, J.V. Sanders, T.W. Turney, Chem.

Mater. 3 (1991) 977.[37] K. Kato, C. Zheng, J.M. Finder, S.K. Dey, Y. Torii, J. Am. Ceram.

Soc. 81 (1998) 1869.[38] A.H. Munhoz, S. Rodrigues, T. Pinnavaia, Adv. Sci. Technol. 16

(1999) 521.[39] L.P. Cruz, J.-M. Savariault, J. Rocha, J.-C. Jumas, J.D. Pedrosa de

Jesus, J. Solid State Chem. 156 (2001) 349.[40] H. Yoshida, H. Nishihara, S. Yokota, M. Ohyanagi, T. Nakaoki, Z.

Naturforsch. A: Phys. Sci. 53 (1998) 309.

[41] J. Rocha, P. Brandao, Z. Lin, A.P. Esculcas, A. Ferreira, M.W.Anderson, J. Phys. Chem. 100 (1996) 14978.

[42] K.O. Drake, D. Carta, L.J. Skipper, F.E. Sowrey, R.J. Newport,M.E. Smith, Solid State Nucl. Magn. Reson. 27 (2005) 28.

[43] A.W. Glynn, A.L. Ponce, X. Lin, J.J. Fripiat, Solid State Ionics 84(1996) 213.

[44] V.M. Mastikhin, O.B. Lapina, Vanadium Catalysts: Solid StateNMR, in: D.M. Grant, R.K. Harris (Eds.), Encyclopedia of NuclearMagnetic Resonance, vol. 8, Wiley, New York, 1996, pp. 4892–4904.

[45] O.B. Lapina, A.A. Shubin, D.F. Khabibulin, V.V. Terskikh, P.R.Bodart, J.-P. Amoureux, Catal. Today 78 (2003) 91.

[46] A.A. Shubin, O.B. Lapina, D. Courcot, Catal. Today 56 (2000) 379.[47] J.J. Fitzgerald, S. Prasad, J. Huang, J.S. Shore, J. Am. Chem. Soc.

122 (2000) 2556.[48] S. Prasad, P. Zhao, J. Huang, J.J. Fitzgerald, J.S. Shore, Solid State

Nucl. Magn. Reson. 14 (1999) 231.[49] L.P. Cruz, J. Rocha, J.D.P. de Jesus, J.M. Savariault, J. Galy, Solid

State Nucl. Magn. Reson. 15 (1999) 153.[50] S. Prasad, P. Zhao, J. Huang, J.J. Fitzgerald, J.S. Shore, Solid State

Nucl. Magn. Reson. 19 (2001) 45.[51] J.J. Fitzgerald, J. Huang, H. Lock, Solid-state 25Mg, 45Sc and 93Nb

MAS NMR studies: local B-site chemical environments and order-ing in Pb(Mg1/3Nb2/3)O3 (PMN) and related relaxor ferroelectrics,in: C. Galassi, M. Dinescu, K. Uchino, M. Sayer (Eds.), PiezoelectricMaterials: Advances in Science, Technology and Applications. vol.76, NATO Sci. Ser. Kluwer Academic Publishers, Norwell, 2000. pp.203–218.

[52] R. Blinc, A. Gregorovic, B. Zalar, R. Pirc, V.V. Laguta, M.D.Glinchuk, J. Appl. Phys. 89 (2001) 1349.

[53] L.-S. Du, R.W. Schurko, N. Kim, C.P. Grey, J. Phys. Chem. A 105(2001) 760.

[54] L.-S. Du, R.W. Schurko, N. Kim, C.P. Grey, J. Phys. Chem. A 106(2002) 7876.

[55] O.B. Lapina, D.F. Khabibulin, K.V. Romanenko, Z.H. Gan, M.G.Zuev, V.N. Krasil’nikov, V.E. Fedorov, Solid State Nucl. Magn.Reson. 28 (2005) 204.

[56] D. Rehder, B. Wolf, J. Magn. Reson. 68 (1986) 157.[57] J.J. Van der Klink, F. Borsa, Phys. Rev. B: Condens. Matter 30

(1984) 52.[58] K.D. Becker, U. Berlage, J. Magn. Reson. 54 (1983) 212.[59] J.I. Budnick, H. Lawrence, L.H. Bennett, J. Phys. Chem. Solids 16

(1960) 37.[60] H. Nishihara, G.A. Scholz, M. Naito, R.F. Frindt, S. Tanaka, J.

Magn. Magn. Mater. 31-34 (1983) 717.[61] M. Naito, H. Nishihara, S. Tanaka, J. Chem. Phys. C 16 (1983) 387.[62] E.B. Doering, J.S. Waugh, J. Chem. Phys. 85 (1986) 1753.[63] S. Rod, F. Borsa, J.J. Van der Klink, Phys. Rev. B: Condens. Matter

38 (1988) 2267.[64] J. Mason, Multinuclear NMR, Plenum Press, New York and

London, 1987.[65] P.R. Bodart, J.-P. Amoureux, Y. Dumazy, R. Lefort, Mol. Phys. 98

(2000) 1545.[66] S.E. Ashbrook, S. Wimperis, J. Magn. Reson. 156 (2002) 269.[67] J.P. Amoureux, L. Delevoye, S. Steuernagel, Z.H. Gan, S. Gana-

pathy, L. Montagne, J. Magn. Reson. 172 (2005) 268.[68] K. Paulsen, D. Rehder, Z. Naturforsch. A37 (1982) 139.[69] M.A. Habayeb, O.E. Hileman, Can. J. Chem. 58 (1980) 2115.[70] A. Allerhand, J. Chem. Phys. 52 (1970) 2162.[71] E.L. Hahn, Phys. Rev. 80 (1950) 580.[72] P. Mansfield, Phys. Rev. 137 (1965) A961.[73] P.P. Man, E. Duprey, J. Fraissard, P. Tougne, J.-B. d’Espinose,

Solid State Nucl. Magn. Reson. 5 (1995) 181.[74] P.P. Man, Quadrupole couplings in nuclear magnetic resonance, in:

R.A. Meyers (Ed.), Encyclopedia of Analytical Chemistry, Chich-ester, New York, 2000, pp. 12224–12265.

[75] D. Massiot, I. Farnan, N. Gautier, D. Trumeau, A. Trokiner, J.P.Coutures, Solid State Nucl. Magn. Reson. 4 (1995) 241.


[76] G. Wu, S. Dong, Solid State Nucl. Magn. Reson. 20 (2001) 100.[77] F.H. Larsen, J. Skibsted, H.J. Jakobsen, N.C. Nielsen, J. Am. Chem.

Soc. 122 (2000) 7080.[78] A.S. Lipton, J.A. Sears, P.D. Ellis, J. Magn. Reson. 151 (2001) 48.[79] R. Lefort, J.W. Weinch, M. Pruski, J.-P. Amoureux, J. Chem. Phys.

116 (2002) 2493.[80] G.M. Bowers, A.S. Lipton, K.T. Mueller, Solid State Nucl. Magn.

Reson. 29 (2006) 95.[81] F.H. Larsen, H.J. Jakobsen, P.D. Ellis, N.C. Nielsen, J. Magn.

Reson. 131 (1998) 144.[82] I. Hung, R.W. Schurko, J. Phys. Chem. B 108 (2004) 9060.[83] K.J. Ooms, K.W. Feindel, V.V. Terskikh, R.E. Wasylishen, Inorg.

Chem. 45 (2006) 8492.[84] A. Samoson, E. Lippmaa, Phys. Rev. B: Condens. Matter 28 (1983)

6567.[85] E. Oldfield, R.A. Kinsey, B. Montez, T. Ray, K.A. Smith, J. Chem.

Soc. Chem. Commun. (1982) 254.[86] A. Samoson, Extended magic-angle spinning, in: D.M. Grant, K.

Harris (Eds.), Encyclopedia of Nuclear Magnetic Resonance 9,Wiley, Chichester, 2002, pp. 59–64.

[87] A. Samoson, T. Tuherm, J. Past, J. Magn. Reson. 149 (2001) 264.[88] M. Ernst, A. Samoson, B.H. Meier, Chem. Phys. Lett. 348 (2001)

293.[89] A. Samoson, T. Tuherm, Z.H. Gan, Solid State Nucl. Magn. Reson.

20 (2001) 130.[90] A. Samoson, T. Tuherm, J. Past, A. Reinhold, T. Anupold, I.

Heinmaa, Top. Curr. Chem. 246 (2005) 15.[91] A.A. Shubin, O.B. Lapina, E. Bosch, J. Spengler, H. Knozinger, J.

Phys. Chem. B 103 (1999) 3138.[92] J. Herzfeld, A.E. Berger, J. Chem. Phys. 73 (1980) 6021.[93] J. Skibsted, N.C. Nielsen, H. Bildsøe, H.J. Jakobsen, Chem. Phys.

Lett. 188 (1992) 405.[94] C. Jager, J. Magn. Reson. 99 (1992) 353.[95] J. Skibsted, N.C. Nielsen, H. Bildsøe, H.J. Jakobsen, J. Magn.

Reson. 95 (1991) 88.[96] A. Samoson, E. Kundla, E. Lippmaa, J. Magn. Reson. 49 (1982)

350.[97] M.H. Cohen, F. Reif, Quadrupole Effects in Nuclear Magnetic

Resonance Studies of Solids, Academic Press Inc., New York, 1957.[98] A. Samoson, E. Lippmaa, A. Pines, Mol. Phys. 65 (1988) 1013.[99] A.A. Shubin, O.B. Lapina, G.M. Zhidomirov, in: IXth AMPERE

Summer School, Novosibirsk, 1987, p. 103.[100] A.A. Shubin, O.B. Lapina, V.M. Bondareva, Chem. Phys. Lett. 302

(1999) 341.[101] G.A. Zenkovets, G.N. Kryukova, S.V. Tsybulya, E.M. Al’kaeva,

T.V. Andrushkevich, O.B. Lapina, E.B. Burgina, L.S. Dovlitova,V.V. Malakhov, G.S. Litvak, Kinet. Catal. 41 (2000) 572.

[102] L. Frydman, J.S. Harwood, J. Am. Chem. Soc. 117 (1995) 5367.[103] A. Medek, J.S. Harwood, L. Frydman, J. Am. Chem. Soc. 117

(1995) 12779.[104] H.-T. Kwak, Z.H. Gan, J. Magn. Reson. 164 (2003) 369.[105] J.-P. Amoureux, C. Fernandez, L. Frydman, Chem. Phys. Lett. 259

(1996) 347.[106] U.G. Nielsen, H.J. Jakobsen, J. Skibsted, Solid State Nucl. Magn.

Reson. 23 (2003) 107.[107] Z.H. Gan, J. Am. Chem. Soc. 122 (2000) 3242.[108] Z.H. Gan, J. Chem. Phys. 114 (2001) 10845.[109] P. Caravatti, L. Raunschweiler, R.R. Ernst, Chem. Phys. Lett. 100

(1983) 305.[110] A. Bielecki, D.P. Burum, D.M. Rice, F.E. Karasz, Macromolecules

24 (1991) 4820.[111] D.P. Burum, A. Bielecki, J. Magn. Reson. 94 (1991) 645.[112] S. Steuernagel, Solid State Nucl. Magn. Reson. 11 (1998) 197.[113] S.R. Hartmann, E.L. Hahn, Phys. Rev. B: Condens. Matter 128

(1962) 2042.[114] D.E. Kaplan, E.L. Hahn, J. Phys. Radium 19 (1958) 821.[115] M.J. Duer, Solid-state NMR Spectroscopy: Principles and Applica-

tions, Blackwell Science, Malden, MA, 2002.

[116] D.M. Follstaedt, C.P. Slichter, Phys. Rev. B: Condens. Matter 16(1977) 21.

[117] T. Gullion, J. Schaefer, J. Magn. Reson. 81 (1989) 196.[118] C. Brown, R. Achey, R. Fu, T. Gedris, A.E. Stiegman, J. Am. Chem.

Soc. 127 (2005) 11590.[119] C. Hudalla, H. Eckert, R. Dupree, J. Phys. Chem. 100 (1996)

15986.[120] C.P. Grey, W.S. Veeman, A.J. Vega, J. Chem. Phys. 98 (1993) 7711.[121] C.P. Grey, A.J. Vega, J. Am. Chem. Soc. 117 (1995) 8232.[122] N. Kim, C.P. Grey, Science 297 (2002) 1317.[123] T. Gullion, Chem. Phys. Lett. 246 (1995) 325.[124] W. Huang, A.J. Vega, T. Gullion, T. Polenova, J. Am. Chem. Soc.

129 (2007) 13027.[125] C. Bonhomme, C. Coelho, T. Azais, L. Bonhomme-Coury, F.

Babonneau, J. Maquet, R. Thouvenot, C.R. Chim. 9 (2006) 466.[126] H.M. Kao, K.H. Lii, Inorg. Chem. 41 (2002) 5644.[127] L.B. Alemany, R.L. Callender, A.R. Barron, S. Steuernagel, D.

Luga, A.P.M. Kentgens, J. Phys. Chem. B 104 (2000) 11612.[128] J.F. Stebbins, L.-S. Du, S. Kroeker, P. Neuhoff, D. Rice, J. Frye,

H.J. Jakobsen, Solid State Nucl. Magn. Reson. 21 (2002) 105.[129] K.W. Feindel, K.J. Ooms, R.E. Wasylishen, Phys. Chem. Chem.

Phys. 9 (2007) 1226.[130] K.J. Ooms, V.V. Terskikh, R.E. Wasylishen, J. Am. Chem. Soc. 129

(2007) 6704.[131] D.L. Bryce, E.B. Bultz, Chem. Eur. J. 13 (2007) 4786.[132] F. Jensen, Introduction to Computational Chemistry, John Wiley &

Sons, Chichester, 1999.[133] S. Moon, S. Patchkovskii, First-principles calculations of paramag-

netic NMR shifts, in: M. Kaupp, M. Buhl, V.G. Malkin (Eds.),Calculation of EPR and NMR parameters: Theory and Applica-tions, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2004(Chapter 20).

[134] M. Buhl, NMR of transition metal compounds, in: M. Kaupp, M.Buhl, V.G. Malkin (Eds.), Calculation of EPR and NMRparameters: Theory and Applications, Wiley-VCH Verlag GmbH& Co. KGaA, Weinheim, 2004 (Chapter 20).

[135] J. Autschbach, Struct. Bond. 112 (2004) 1.[136] N. Pooransingh-Margolis, R. Renirie, Z. Hasan, R. Wever, A.J.

Vega, T. Polenova, J. Am. Chem. Soc. 128 (2006) 5190.[137] N. Pooransingh, E. Pomerantseva, M. Ebel, S. Jantzen, D. Rehder,

T. Polenova, Inorg. Chem. 42 (2003) 1256.[138] K.J. Ooms, S.E. Bolte, J.J. Smee, B. Baruah, D.C. Crans, T.

Polenova, Inorg. Chem. 46 (2007) 9285.[139] B.A. Gee, Solid State Nucl. Magn. Reson. 20 (2006) 171.[140] M.R. Hansen, G.K.H. Madsen, H.J. Jakobsen, J. Skibsted, J. Phys.

Chem. B 110 (2006) 5975.[141] L. Truflandier, M. Paris, C. Payen, F. Boucher, J. Phys. Chem. B

110 (2006) 21403.[142] C.J. Pickard, F. Mauri, Phys. Rev. B: Condens. Matter 63 (2001)

245101.[143] L. Truflandier, M. Paris, F. Boucher, Phys. Rev. B: Condens. Matter

76 (2007) 035102.[144] J. Skibsted, N.C. Nielsen, H. Bildsøe, H.J. Jakobsen, J. Am. Chem.

Soc. 115 (1993) 7351.[145] J. Skibsted, C.J.H. Jacobsen, H.J. Jakobsen, Inorg. Chem. 37 (1998)

3083.[146] U.G. Nielsen, H.J. Jakobsen, J. Skibsted, Inorg. Chem. 39 (2000)

2135.[147] U.G. Nielsen, H.J. Jakobsen, J. Skibsted, J. Phys. Chem. B 105

(2001) 420.[148] C.J. Fontenot, J.W. Wiench, M. Pruski, G.L. Schrader, J. Phys.

Chem. B 105 (2001) 10496.[149] D.F. Khabibulin, A.A. Shubin, O.B. Lapina, Correlation of 51V

NMR parameters with local environment of vanadia sites, in: J.Fraissard, O.B. Lapina (Eds.), Magnetic Resonance in Colloid andInterface, vol. 76, NATO ASI, Kluwer, Dordrecht/Boston/London,2002, pp. 537–546.

[150] R.D. Shannon, C. Calvo, J. Solid State Chem. 6 (1973) 538.


[151] R. Olazcuaga, J.M. Reau, G. le Flem, P. Hagenmuller, Z. Anorg.Allg. Chem. 412 (1975) 271.

[152] A.D. Kelmers, J. Inorg. Nucl. Chem. 21 (1961) 45.[153] A.D. Kelmers, J. Inorg. Nucl. Chem. 23 (1961) 279.[154] R. Gopal, C. Calvo, Z. Kristallogr. 173 (1973) 67.[155] W. Carrillo-Cabrera, H.G. von Schnering, Z. Kristallogr. 205 (1993)

271.[156] G. Liu, J.E. Greedan, J. Solid State Chem. 110 (1994) 274.[157] H. Kasatani, T. Umeki, H. Terauchi, J. Phys. Soc. Jpn. 61 (1992)

2309.[158] E. Arisi, S.S.A. Palomares, F. Leccabue, B.E. Watts, G. Bocelli, F.

Calderon, G. Calestani, L. Righi, Mater. Sci. 39 (2004) 2107.[159] Y. Oka, T. Yao, N. Yamamoto, J. Solid State Chem. 152 (2000) 486.[160] C.E. Rice, W.R. Robinson, Acta Crystallogr. Sect. B: Struct. Sci. 32

(1976) 2232.[161] B.C. Chakoumakos, M.M. Abraham, L.A. Boatner, J. Solid State

Chem. 109 (1994) 197.[162] K.-J. Range, H. Meister, U. Klement, Z. Naturforsch. Teil B:

Anorg. Chem. Org. Chem. 45 (1990) 598.[163] H. Chahboun, D. Groult, B. Raveau, Mater. Res. Bull. 23 (1988)

805.[164] M.T. Casais, E. Gutierrez-Puebla, M.A. Monge, I. Rasines, C. Ruiz-

Valero, J. Solid State Chem. 102 (1993) 261.[165] F.C. Hawthorne, C. Calvo, J. Solid State Chem. 26 (1978) 345.[166] J. Huang, A.W. Sleight, Mater. Res. Bull. 27 (1992) 581.[167] R.D. Shannon, C. Calvo, Can. J. Chem. 51 (1973) 265.[168] F. Marumo, M. Isolbe, S. Iwai, Y. Kondo, Acta Crystallogr. Sect. B:

Struct. Sci. 30 (1974) 1628.[169] H.T. Evans, Z. Kristallogr. 114 (1960) 257.[170] M. Ganne, Y. Piffard, M. Tournoux, Can. J. Chem. 52 (1974) 3539.[171] F.C. Hawthorne, C. Calvo, J. Solid State Chem. 22 (1977) 157.[172] T. Yao, Y. Oka, N. Yamamoto, Inorg. Chim. Acta 238 (1995) 165.[173] N.N. Hok, C. Calvo, Can. J. Chem. 50 (1972) 3619.[174] G.D. Andreetti, G. Calestani, A. Montenero, Z. Kristallogr. 168

(1984) 53.[175] K. Mocala, J. Zio’lkowski, J. Solid State Chem. 69 (1987) 299.[176] O.G. Karpov, M.A. Simonov, T.I. Krasnenko, O.A. Zabara,

Kristallografiya 34 (1989) 1392 (in Russian).[177] N. Das, H. Eckert, H. Hu, I.E. Wachs, J.F. Walzer, F.J. Feher, J.

Phys. Chem. 97 (1993) 8240.[178] Y. Oka, S. Fumihiko, Y. Takeshi, Y. Naoichi, J. Solid State Chem.

134 (1997) 52.[179] F.B. Abdelouahab, R. Olier, N. Guilhaume, F. Lefebre, J.C. Volta,

J. Catal. 134 (1992) 151.[180] O.B. Lapina, D.F. Khabibulin, A.A. Shubin, V.M. Bondareva, J.

Mol. Catal. A: Chem. 162 (2000) 381.[181] E.V. Sokolova, M.A. Simonov, U.V. Karyakin, A.V. Zavolskaya,

Kristallografiya 28 (1983) 862 (in Russian).[182] H.T.J. Evance, Inorg. Chem. 5 (1966) 967.[183] Y. Shan, S.D. Huang, J. Chem. Crystallogr. 29 (1999) 93.[184] O.W. Howarth, R.E. Richards, J. Chem. Soc. (1965) 864.[185] H. Eckert, I.E. Wachs, J. Phys. Chem. 93 (1989) 6796.[186] S. Hayakawa, T. Yoko, S. Sakka, Bull. Chem. Soc. Jpn. 66 (1993)

3393.[187] U.G. Nielsen, A. Boisen, M. Brorson, C.J.H. Jacobsen, H.J.

Jakobsen, J. Skibsted, Inorg. Chem. 41 (2002) 6432.[188] R. Cousin, D. Courcot, E. Abi-Aad, S. Capelle, J.P. Amoureux, M.

Dourdin, M. Guelton, A. Aboukais, Colloids Surf. A 158 (1999) 43.[189] F. Delmaire, M. Rigole, E.A. Zhilinskaya, A. Aboukais, R. Hubaut,

G. Mairesse, Phys. Chem. Chem. Phys. 2 (2000) 4477.[190] P.R. Bodart, Y. Dumazy, J.P. Amoureux, C. Fernandez, Magn.

Reson. Chem. 37 (1999) 223.[191] O.B. Lapina, V.M. Mastikhin, A.V. Nosov, T. Beutel, H. Knozin-

ger, Catal. Lett. 13 (1992) 203.[192] P.E. Stallworth, X. Guo, E. Tatham, S.G. Greenbaum, M. Arrabito,

S. Bodoarado, N. Penazzi, Solid State Ionics 170 (2004) 181.[193] D.F. Khabibulin, K.V. Romanenko, Z.H. Gan, E. Arkhipova, M.

Zuev, O.B. Lapina, Magn. Reson. Chem. 45 (2007) 962.

[194] S. Hayakava, Y. Toshinobu, S. Sumio, J. Solid State Chem. 112(1994) 329.

[195] A.A. Shubin, O.B. Lapina, in: V Russian Conference on Structureand Dynamics of Molecular Systems, vol. 1, Mariisky StateUniversity, Ioshkar Ola, 1998, p. 36–41 (in Russian).

[196] C. Fernandez, P. Bodart, J.-P. Amoureux, Solid State Nucl. Magn.Reson. 3 (1994) 79.

[197] O. Durupthy, N. Steunou, T. Coradin, J. Maquet, C. Bonhomme, J.Livage, J. Mater. Chem. 15 (2005) 1090.

[198] R. Siegel, N. Dupre, M. Quarton, J. Hirschinger, Magn. Reson.Chem. 42 (2004) 1022.

[199] O.B. Lapina, V.V. Terskikh, A.A. Shubin, K.M. Eriksen, R.Fehrmann, Colloids Surf. A 158 (1999) 255.

[200] W. Huang, L. Todaro, L.C. Francesconi, T. Polenova, J. Am. Chem.Soc. 125 (2003) 5928.

[201] W. Huang, L. Todaro, G.P.A. Yap, R. Beer, L.C. Francesconi, T.Polenova, J. Am. Chem. Soc. 126 (2004) 11564.

[202] O. Durupthy, M. Jaber, N. Steunou, J. Maquet, G.T. Chandrappa,J. Livage, Chem. Mater. 17 (2005) 6395.

[203] D. Franklin, I.E. Wachs, H. Eckert, D.A. Jefferson, J. Solid StateChem. 90 (1991) 194.

[204] B. Alonso, C. Sanchez, J. Mater. Chem. 10 (2000) 377.[205] K. Ukei, H. Suzuki, T. Shishido, T. Fukuda, Acta Crystallogr. Sect.

C: Cryst. Struct. Commun. 50 (1994) 655.[206] T. Shishido, H. Suzuki, K. Ukei, T. Hibiya, T. Fukuda, J. Alloys

Compd. 234 (1996) 256.[207] A. Castro, E. Aguado, J.M. Rojo, P. Herrero, R. Enjalbert, J. Galy,

Mater. Res. Bull. 33 (1998) 31.[208] A. Kahn-Harari, L. Mazerolles, D. Michel, F. Robert, J. Solid State

Chem. 116 (1995) 103.[209] M.A. Subramanian, J.C. Calabrese, Mater. Res. Bull. 28 (1993)

523.[210] R. Hsu, E.N. Maslen, D.D. Boula, N. Ishizawa, Acta Crystallogr.

Sect. B: Struct. Sci. 53 (1997) 420.[211] A.W. Hewat, J. Phys. C: Sol. St. Phys. 6 (1973) 2559.[212] V.A. Shuvaeva, M.Y. Antipin, Kristallografiya 40 (1995) 511 (in

Russian).[213] T.S. Ercit, P. Cerny, Can. Mineral. 26 (1988) 899.[214] J. Galy, O. Lindqvist, J. Solid State Chem. 27 (1979) 279.[215] K.J.D. MacKenzie, M.E. Smith, Multinuclear Solid State NMR of

Inorganic Materials, Pergamon Elsevier, Oxford, 2002.[216] Y. Watanabe, T. Sota, K. Suzuki, N. Iyii, K. Kitamura, S. Kimura,

J. Phys.: Condens. Matter 7 (1995) 3627.[217] M.D. Meadows, K.A. Smith, R.A. Kinsey, M. Rothgeb, R.P.

Skarjune, E. Oldfield, Proc. Natl. Acad. Sci. USA 79 (1982) 1351.[218] H. Weitzel, H. Schroecke, Z. Kristallogr. 152 (1980) 69.[219] S. Tsunekawa, T. Kamiyama, K. Sasaki, H. Asano, T. Fukuda, Acta

Crystallogr. Sect. A: Found. Crystallogr. 49 (1993) 595.[220] N.A. Godina, T.I. Panova, E.K. Keller, Izvestiya AN SSSR Inorg.

Mater. 5 (1969) 1974 (in Russian).[221] K.-J. Range, M. Wildenauer, A.M. Heyns, Angew. Chem. Int. Ed.

Engl. 27 (1988) 969.[222] A. Flambard, L. Montagne, L. Delevoye, S. Steuernagel, Solid State

Nucl. Mag. Reson. 32 (2007) 34.[223] J.P. Jesson, The paramagnetic shifts, in: G.N. La Mar, W.D.

HorrocksJr., R.H. Holm (Eds.), NMR of Paramagnetic Molecules:Principles and Applications, Academic Press, New York andLondon, 1973, pp. 1–52.

[224] I. Bertini, C. Luchinat, Coord. Chem. Rev. 150 (1996) 1.[225] I. Bertini, C. Luchinat, G. Parigi, Prog. Nucl. Magn. Reson.

Spectrosc. 40 (2002) 249.[226] T. Moriya, Prog. Theor. Phys. 16 (1956) 23.[227] V. Jaccarino, Nuclear resonance in antiferromagnets, in: G.T. Rado,

H. Suhl (Eds.), Magnetism, vol. 2, Academic Press, New York, 1965,p. 307.

[228] M.I. Kurkin, E.A. Turov, NMR in Magnetically Ordered Materialsand its Applications, Nauka, Moscow, 1990 (in Russian).

[229] P. Lemmens, P. Millet, Lect. Notes Phys. 645 (2004) 433.


[230] J. Stohr, H.C. Siegmann, Magnetism from fundamentals to nano-scale dynamics, Springer Series in Solid-State Sciences, vol. 152,Springer-Verlag, Berlin, Heidelberg, 2006.

[231] J. Li, M.E. Lashier, G.L. Schrader, B.C. Gerstein, Appl. Catal. 73(1991) 83.

[232] A. Tuel, L. Canesson, J.C. Volta, Colloids Surf. A 158 (1999) 97.[233] R.N. Pletnev, V.N. Lisson, V.A. Gubanov, Fizika Tverdogo Tela 15

(1973) 558 (in Russian).[234] R.N. Pletnev, A.A. Sidorov, V.N. Lisson, A.K. Chirkov, Dokladi

Akademii Nauk SSSR 236 (1977) 1159 (in Russian).[235] R. Schowalter, Phys. Status Solidi B 48 (1971) 743.[236] U. Ranon, Phys. Lett. A 28 (1968) 228.[237] A.A. Fotiev, V.L. Volkov, V.K. Kapustkin, Bronses of Vanadium

Oxides, Nauka, Moskva, 1978 (in Russian).[238] D. Tinet, J.J. Fripiat, Rev. Chim. Mineral. 19 (1982) 612.[239] K. Nakamura, D. Nishioka, Y. Michihiro, M. Vijayakumar, S.

Selvasekarapandian, T. Kanashiro, Solid State Ionics 177 (2006) 129.[240] N.I. Lazukova, V.A. Gubanov, Zh. Struk. Khim. 18 (1977) 10.[241] F.-N. Shi, F.A.A. Paz, J. Rocha, J. Klinowski, T. Trindade, Eur. J.

Inorg. Chem. 2004 (2004) 3031.[242] U.G. Nielsen, J. Skibsted, H.J. Jakobsen, Chem. Phys. Lett. 356

(2002) 73.[243] I. Karatchevtseva, Z. Zhang, J. Hanna, V. Luca, Chem. Mater. 18

(2006) 4908.[244] F. Carn, N. Steunou, J. Livage, A. Colin, R. Backov, Chem. Mater.

17 (2005) 644.[245] N. Dupre, J. Gaubicher, D. Guyomard, C.P. Grey, Chem. Mater. 16

(2004) 2725.[246] M. Dubarry, J. Gaubicher, D. Guyomard, N. Steunou, J. Livage, N.

Dupre, C.P. Grey, Chem. Mater. 18 (2006) 629.[247] R. Michalaka, H.A.H. Mohammad, Solid State Nucl. Magn. Reson.

26 (2004) 187.[248] L.G. Pinaeva, O.B. Lapina, V.M. Mastikhin, A.V. Nosov, B.S.

Balzhinimaev, J. Mol. Catal. 88 (1994) 311.[249] D. Courcot, B. Grzybowska, Y. Barbaux, M. Rigole, A. Ponchel, M.

Guelton, J. Chem. Soc., Faraday Trans. 92 (1996) 1609.

[250] L. Lietti, P. Forzatti, G. Ramis, G. Busca, F. Bregani, Appl. Catal.B 3 (1993) 13.

[251] G.T. Went, L.J. Leu, S.J. Lombardo, A.T. Bell, J. Phys. Chem. 96(1992) 2235.

[252] V.M. Mastikhin, I.L. Mudrakovsky, A.V. Nosov, Prog. Nucl.Magn. Reson. Spectrosc. 23 (1991) 259.

[253] G. Deo, I.E. Wachs, J. Phys. Chem. 95 (1991) 5889.[254] C.J. Fontenot, J.W. Wiench, M. Pruski, G.L. Schrader, J. Phys.

Chem. B 104 (2000) 11622.[255] C.J. Fontenot, J.W. Wiench, G.L. Schrader, M. Pruski, J. Am.

Chem. Soc. 124 (2002) 8435.[256] D.C. Sayele, C.R.A. Catlow, M.-A. Perrin, P. Nortier, J. Phys.

Chem. 100 (1996) 8940.[257] U.G. Nielsen, N.U. Topsøe, M. Brorson, J. Skibsted, H.J. Jakobsen,

J. Am. Chem. Soc. 126 (2004) 4926.[258] O.B. Lapina, A.A. Shubin, A. Nosov, E. Bosch, J. Spengler, H.

Knozinger, J. Phys. Chem. 103 (1999) 7599.[259] C. Martin, D. Klissurski, J. Rochac, V. Rives, Phys. Chem. Chem.

Phys. 2 (2000) 1543.[260] V.V. Terskikh, O.B. Lapina, V.M. Bondareva, Phys. Chem. Chem.

Phys. 2 (2000) 2441.[261] R. Blinc, B. Zalar, A. Gegorovic, R. Pirc, M.D. Glinchuk,

Ferroelectrics 240 (2000) 1473.[262] R. Pirc, R. Blinc, Phys. Rev. B: Condens. Matter 60 (1999) 13470.[263] V.V. Laguta, M.D. Glinchuk, I.P. Bykov, R. Blinc, B. Zalar, Phys.

Rev. B: Condens. Matter 69 (2004) 54103/1.[264] A.Y.H. Lo, T.E. Bitterwolf, C.L.B. Macdonald, R.W. Schurko, J.

Phys. Chem. A 109 (2005) 7073.[265] S. Rankel, B. Zalar, V.V. Laguta, R. Blinc, J. Toulouse, Ferroelec-

trics 314 (2005) 165.[266] K. Sato, S. Takeda, S. Fukuda, T. Minamisono, M. Tanigaki, T.

Miyake, Y. Maruyama, K. Matsuta, M. Fukuda, Y. Nojiri, Z.Naturforsch. A: Phys. Sci. 53 (1998) 549.

[267] O.B. Lapina, V.M. Mastikhin, L.G. Simonova, Y.O. Bulgakova, J.Mol. Catal. 69 (1991) 61.

[268] J. Klinowski, Prog. Nucl. Magn. Reson. Spectrosc. 16 (1984) 237.

Documents

Practical Aspects of 51V and 93Nb Solid-state NMR Spectroscopy