Insights into the Exfoliation Process of V2O5·nH2O Nanosheet Formation Using Real-Time 51V NMR.

Nanostructured hydrated vanadium oxides (V2O5·nH2O) are actively being researched for applications in energy storage, catalysis, and gas sensors. Recently, a one-step exfoliation technique for fabricating V2O5·nH2O nanosheets in aqueous media was reported; however, the underlying mechanism of exfoliation has been challenging to study. Herein, we followed the synthesis of V2O5·nH2O nanosheets from the V2O5 and VO2 precursors in real time using solution- and solid-state 51V NMR. Solution-state 51V NMR showed that the aqueous solution contained mostly the decavanadate anion [H2V10O28]4- and the hydrated dioxovanadate cation [VO2·4H2O]+, and during the exfoliation process, decavanadate was formed, while the amount of [VO2·4H2O]+ remained constant. The conversion of the solid precursor V2O5, which was monitored with solid-state 51V NMR, was initiated when VO2 was in its monoclinic forms. The dried V2O5·nH2O nanosheets were weakly paramagnetic because of a minor content of isolated V4+. Its solid-state 51V signal was less than 20% of V2O5 and arose from diamagnetic V4+ or V5+.This study demonstrates the use of real-time NMR techniques as a powerful analysis tool for the exfoliation of bulk materials into nanosheets. A deeper understanding of this process will pave the way to tailor these important materials.

Introduction

In the last few years, the synthesis of two-dimensional (2D) materials based on transition metal chalcogenides and oxides with thicknesses of a few layers has attracted renewed attention because of the different chemical, physical, and semiconducting properties of these materials compared to their bulk (three-dimensional, 3D) counterparts.[1−3] Vanadium oxides are earth-abundant compounds, which have important applications in catalysis,[4] batteries,[5−7] supercapacitors,[8−10] and sensors.[11] Thus, many research groups have focused on the synthesis of 2D vanadium oxides from their bulk precursors.[12−14] Of particular interest among these 2D materials are those based on the hydrated vanadium pentoxides (V2O5·nH2O), which have been shown to exhibit improved electrochemical behavior and semiconducting properties compared to anhydrous V2O5.[15] The improvements are typically ascribed to the presence of H2O or H+ ions between the V2O5 layers in V2O5·nH2O, which can be synthesized in the form of hydrogels,[16] xerogels,[17,18] nanobelts,[19] and nanosheets.[5−7] Nanostructured V2O5·nH2O has attracted research interest as it can be easily fabricated into a freestanding film,[5] which is easier to handle than, and is thus advantageous compared to, an amorphous or crystalline powder or gel.

V2O5·nH2O is commonly synthesized either by an ion-exchange route using sodium metavanadate solution or via a sol–gel route using a mixture of hydrogen peroxide (H2O2) and V2O5.[20,21] In both cases, a dark red compound is formed with a layered structure. Recently, Etman et al. synthetized V2O5·nH2O nanosheets using aqueous exfoliation of a mixture of V2O5 and VO2, resulting in black/green films.[7] The synthesis was monitored by in situ X-ray diffraction (XRD) studies, which revealed that the V2O5·nH2O phase started to form after 90 min of reflux in water at 80 °C. A more detailed understanding of the underlying mechanism of the formation of this V2O5·nH2O product is challenging because vanadium is a transition metal with a very complex chemistry and many different stable oxidation states.[21] As XRD and wide-angle X-ray scattering provide data mainly on the long-range order in compounds, other techniques are important to assess developments in noncrystalline materials or at the solid–liquid interface during the reaction.

Nuclear magnetic resonance (NMR) is a characterization technique complementary to XRD and has been used to study the ion-exchange or sol–gel synthesis routes of V2O5·nH2O.[22−26] It can provide information about the local order, coordination states, and protonated or deprotonated oxygen atoms. Typically, solid compounds are monitored by magic-angle spinning (MAS) solid-state (ss)-NMR, whereas solution-state NMR is used to observe dissolved compounds. However, both solid- and solution-state 51V NMR are complicated by localized unpaired electrons in paramagnetic V4+ ions, which may bleach out the signal arising from the NMR detectable (diamagnetic) 51V5+ ions in various ways.[27] MAS ss-NMR is able to distinguish between delocalized (metallic) and localized (paramagnetic) electronic states via Knight shifts and paramagnetic shifts.[28] Paramagnetic, which refers here to Curie or Curie–Weiss paramagnetism, V4+ ions with localized unpaired electrons cannot be directly studied with NMR, but their presence has the effect of bleaching the 51V signal of nearby V5+ ions, allowing the presence of V4+ to be probed indirectly. However, V4+ ions with localized unpaired electrons can be studied by other techniques, for example, electron spin resonance (ESR). When the V4+ ions are less than 2.7 Å apart, their unpaired electrons may pair and turn the corresponding materials from paramagnetic to diamagnetic, which does give a detectable 51V4+ NMR signal.[27,29,30]

In this paper, we report on real-time solid- and solution-state 51V NMR studies performed during the synthesis of V2O5·nH2O nanosheets from a 1:4 mixture by weight of commercial monoclinic VO2(M) and V2O5. The interpretation of the 51V NMR results was linked with those from ESR and 1H NMR and used to elucidate the mechanism of the aqueous exfoliation process and formation of V2O5·nH2O nanosheets.

Results and Discussion

Morphology and Structure of V2O5·nH2O Nanosheets

The V2O5·nH2O nanosheets were synthesized in water from a 1:4 mixture by weight of monoclinic VO2(M) and commercial V2O5, and the chemical and thermal analyses are described elsewhere.[7] The XRD pattern (Figure a, gray) of the as-prepared V2O5·nH2O nanosheets displayed broad peaks, which were indexed as 00l, reflecting the preferred orientation of the layered structure of the nanosheets. This pattern was recorded in a reflection configuration using an in-house diffractometer (λ = 1.5406 Å). One possible solution to overcome the preferred orientation was to perform XRD in transmission mode with, for example, a high-energy X-ray source (synchrotron radiation, λ = 0.7766 Å). Notably, the XRD pattern (see Figure a, black) recorded in this way was very similar to that collected in the reflection mode using an in-house diffractometer, suggesting a disordered stacking between V2O5·nH2O layers over the a–b plane. Interestingly, the transmission electron microscopy (TEM) images showed that V2O5·nH2O had a typical nanosheet morphology with a different lateral size thickness ranging from 30 to 220 nm (see Figure b). In addition, the selected area electron diffraction (SAED) pattern of V2O5·nH2O had powder rings (see Figure c), which provided additional support for disordered stacking between the layers over the a–b plane.

Figure 1

(a) XRD patterns of V2O5·nH2O nanosheets in transmission mode using synchrotron radiation (black) and low-energy X-ray reflection (gray) mode. The remaining traces of the V2O5 phase are marked by “*”. To the right, the TEM image (b) and SAED (c) of the V2O5·nH2O nanosheets are shown. In (c), the inset shows the crystals from which SAED was obtained.

Local Structure of V2O5·nH2O Nanosheets

MAS ss-NMR can provide fruitful information about the local structure and oxidation states of metal ions. The commercial V2O5 precursor possessed a layered anisotropic structure of distorted VO6 octahedral building units,[13,31] and its 51V isotropic shift was −611 ppm (see Figure , black).[27] The isotropic shift of the largely disordered nanosheets of V2O5·nH2O was slightly lower in magnitude, −596 ppm, and the spinning side band manifold was broader, having nearly double the number of spinning side bands (see Figure , red). In addition, the individual side bands exhibited an increased inhomogeneous line width, which suggested highly distorted geometries of the vanadium sites. The enhanced line width was in agreement with observations from XRD and SAED.

Figure 2

Weight-normalized 51V MAS ss-NMR spectra of V2O5 (black) and V2O5·nH2O nanosheets (red). The signal intensity for V2O5·nH2O is scaled up by a factor of 20. The isotropic shift is marked by “+”.

In previous NMR studies of  V2O5·nH2O gels, synthesized using H2O2-based or ion-exchange methods, up to five different 51V NMR peaks have been observed and attributed to various vanadium sites.[32,33] The corresponding 51V isotropic shifts have been in the interval of −572 to −663 ppm. However, in this study, the intrinsic and symmetric 51V line width of the 51V NMR peaks of the nanosheets of V2O5·nH2O exceeded 80 ppm and thus prevented potential multiple vanadium sites from being resolved. On the basis of previous studies,[32] the 51V isotropic shift of the nanosheets of V2O5·nH2O suggested octahedral vanadium sites with one water molecule bonded to the vanadyl oxygen or vanadium pentoxide with shifted subunits.

Notably, the integral intensity of the total 51V signal including the spinning side band manifold of the V2O5·nH2O nanosheets was less than 20% of the V2O5 precursor. The broad line width of the side bands means that the second-order quadrupolar broadening cannot be measured from the line shape, and the contributions of the first-order quadrupolar interaction and shift anisotropy to the spinning side band manifold cannot be easily separated; consequently, the quadrupolar couplings were not measured.[34] One possible reason for the signal loss relative to the precursor would be a phase transition similar to that of VO2(M)–metallic-VO2(R) because of frictional heating from MAS.[35,36]

The magnetic susceptibility data (see Figure S1) indicated a weak paramagnetic behavior of the nanosheets with no observable magnetic phase transition. A fit of these data returned a Weiss constant of zero, pure Curie paramagnetism which we ascribe to isolated (noninteracting) V4+ ions that were incorporated between the V2O5 nanosheets during the course of the reaction. In turn, we ascribe the large reduction in the observable signal from V2O5 to a paramagnetic bleaching effect, where the nuclear relaxation of V5+ is enhanced by the proximity of the paramagnetic V4+ ions.[27] As the synthesis was performed in an aqueous mixture of VO2(M) and V2O5, the former provided a possible source of paramagnetic V4+, given the 1:4-fraction of VO2(M) and V2O5. V2O5·nH2O has been found to contain about 10 mol % of V4+ according to Etman et al.,[7] which was in agreement with the observed weakly paramagnetic behavior. The presence of V4+ was here confirmed by ESR (see Figure S2). The corresponding spectra each had a broad peak with an isotropic g-value of 1.95 at room temperature. This value matched well with those reported for other V4+-containing materials.[23,29,37]

Water Molecules in V2O5·nH2O Nanosheets

The distribution and location of H2O in V2O5·nH2O are important for the electrochemical behavior and semiconducting properties.[20,21] In relation to the positioning of H2O, Pozarnsky and McCormick suggested a chain model with a H2O molecule and a −OH group in the equatorial plane and an additional H2O pointing downward.[25] Hence, we recorded a static 2H ss-NMR spectrum (see Figure ) on the V2O5·nD2O nanosheets and observed three distinct resonances with decreasing intensities at chemical shifts of 1, 3.3, and 7 ppm. Similar results were obtained from 1H NMR experiments under MAS, but the 1H background of the probe and rotor complicated the interpretation (data not shown). Takeda et al. have observed rotational restricted motion in hydrated V2O5 as evidenced by the features of the 2H powder patterns.[22] In this work, neither bulk H2O nor strongly coordinated H2O could be observed. Instead, the features of the static 2H spectrum suggested that H2O had a high mobility. However, H2O or −OH groups bonded or coordinated to V4+ were not easily detectable under static conditions, as they were expected to be strongly shifted by a Fermi-contact shift to the paramagnetic V4+ ion, exhibit large resonance broadening, and have short relaxation times. The observed 2H chemical shifts at 3.3 and 1 ppm were attributed to D2O and −OD on the surface of the nanosheets, respectively.[38] The chemical shift at 7 ppm may have been due to −OD groups, in which the O atom bridges between two V atoms.

Figure 3

Static 2H NMR of V2O5·nD2O nanosheets. The 2H NMR spectrum was recorded using a quadrupolar echo sequence.

Probing Nanosheet Formation by Real-Time 51V NMR

To elucidate the formation of nanosheets, we applied real-time solid- and solution-state 51V NMR to follow the reaction of the solid phases and the dissolved species separately.

Dissolved Species

VO2 and V2O5 with a mass ratio of 1:4 were blended with 550 μL of H2O and 50 μL of D2O in an NMR tube, and solution-state 51V NMR spectra were recorded in real time during the reaction. The observed vanadium species and their 51V shift are summarized in Table . It was evident that the 51V signals of the decavanadate anion [H2V10O28]4–, resonating at −419, −503, and −522 ppm, exhibited increasing integral intensities for up to 2 h after mixing, whereas the 51V signal of the hydrated dioxovanadate cation [VO2·4H2O]+ at −549 ppm retained a constant integral (see Figure ). The decavanadate anion is believed to be produced from 10 dioxovanadate cations under acidic conditions in aqueous solutions. However, if this reaction had occurred here, there must also have been an additional process where dioxovanadate cations were produced. Furthermore, the rate of its formation has to be equal to the rate of consumption, so as to keep the overall dioxovanadate concentration unaltered throughout the entire reaction. It has furthermore been discussed that the decavanadate anion is not responsible for the formation of the nanosheets because it is highly acidic and hence would prevent further condensation reactions.[39] Notably, a broad feature at a shift of about −297 ppm was observed (see Figure ), which Rehder has suggested to be related to a VO2+ derivate.[40] The broadening of this 51V signal might also be attributed to polymeric vanadium species or species that contain V4+ in close vicinity.

Table 1

Dissolved Vanadium Species Observed with Solution-State NMRa

dissolved vanadium species	51V shift (ppm)
[H2V10O28]4–	–419, −503, −522
[VO2·4H2O]+	–549

The peak at −297 ppm could not be assigned.

Figure 4

Stacked real-time solution-state NMR spectra as a function of time. “x” marks an unassigned peak. The inset shows the normalized 51V signal integral as a function of time for [H2V10O28]4– (black) and [VO2·4H2O]+ (red).

The dioxovanadate cation was exchanged with H2VO4–, which has a pKa of 3.8.[41] It is, hence, highly likely that the observed chemical shift of [VO2·4H2O]+ was due to an average of both the cationic and anionic forms and is highly pH-dependent. The observed chemical shift agreed well with the reported one at a pH of 3.8.[41] The formation of the decavanadate anion produces H+, which lowered the pH to 2.6 at the end of the reaction, which in turn shifted the 51V signal of the dioxovanadate cation from −550 ppm to less negative chemical shift values.

An aqueous suspension of VO2 did not produce any 51V NMR peaks at room temperature or 80 °C. Hence, we assumed that all of the 51V signals (in Figure ) including the broad peak at −297 ppm resulted from V2O5 and its reaction products despite its low solubility (0.7 g/L = 3.8 mmol/L at room temperature).[42] It is worth mentioning that VO2 is, however, slightly water-soluble particularly under acidic conditions, and the following V4+ species, which are NMR silent, might be present: [VO·5H2O]2+, VOOH+, and a dimer VO2(OH2)2+; the latter most likely formed from the coupling of two VOOH+ species.[43] V4+ species in an aqueous VO2/V2O5 mixture at 80 °C was confirmed by the hyperfine coupling between an electron and 51V in the ESR spectrum showing eight peaks (see Figure S2). These features of the ESR spectrum suggested isolated V4+ species in solution,[29] most likely [VO·5H2O]2+. Furthermore, the ESR spectrum had a broad feature, which was attributed to solid VO2, and an aqueous mixture of solely V2O5 did not give any ESR signal.

The same vanadium species have been observed in preparations from other approaches using ion-exchange and sol–gel methods.[23−25,37] Our observations were consistent with findings from other synthetic methods taking into account the acidic conditions with pH = 3.8. Furthermore, the observed 51V NMR signals agreed well with earlier findings on the concentration and pH dependence for vanadium species formed in aqueous solutions.[21]

Etman et al. have reported on an onset of the V2O5·nH2O formation after 90 min using real-time XRD,[7] while the formation of decavanadate leveled out after 2.5 h. By comparing those findings with the ones of this study, the question arose if the dissolved species were responsible for the formation of nanosheets or if the observations of the decavanadate anion and the dioxovanadate cation were solely due to various side reactions of the aqueous vanadium chemistry. Many mechanisms have been proposed for the formation of nanostructured gels,[23−25,37] and in our view, the most relevant are those that have dealt with ion exchange.[23,25] However, notably, all of them have derived these compounds from vanadium-based species in solution, whereas here we instead started from two commercial solid compounds (V2O5 and VO2).

Solid Species

To access information on the solid phases during the reaction, we performed real-time MAS ss-NMR experiments on the reaction mixture (0.3 mg of VO2, 1.2 mg of V2O5, and 20 μL of H2O) at 7 kHz MAS. Analyses of the 51V NMR spectra in Figure a,b showed a reduction of the integral of the whole 51V signal spinning side band manifold due to the V2O5 phase, which has an isotropic chemical shift of −611 ppm. The reduction was observed both for low-flip-angle direct excitation (Figure c, gray) and in a Hahn echo experiment (Figure c, red). The normalized 51V NMR integrals of a repeated Hahn echo experiment (Figure c, black) coincided well with the first reaction. Vanadium-containing compounds have a very large 51V NMR shift range, which in turn required that we moved the observation window by changing the carrier frequency and retuning the probe to observe various species. By comparing weight-normalized 51V MAS ss-NMR spectra of fresh commercial VO2(M) and V2O5, the observed broad 51V signal of VO2(M) at approximately 2100 ppm[30] was consistent with <1% of the V2O5 signal in the solid phase (see Figure S3). This intensity was lower than expected and was ascribed to the broad 51V resonance of VO2 being harder to excite and having a shorter relaxation time than those of V2O5. The other 51V signal resonating at negative shift is most likely the impurity V2O5 (see Figure S3). Consequently, an amount of approximately 0.3 mg of VO2(M) was undetectable in the real-time MAS ss-NMR experiments. The lab-scale synthesis was prepared under similar conditions as in real-time MAS ss-NMR experiment and revealed traces of unexfoliated V2O5 at the end of the synthesis, which agreed with the remaining signal in the 51V NMR spectra detected after 38 h (see Figure c, red and gray), and was therefore attributed to this unexfoliated V2O5 precursor. Despite attempts to observe other 51V signals, for example, from VO2, none were detected. One reason might have been the small rotor volume of 20 μL, which required a VO–H2O ratio that is 17 times larger to assure a good signal-to-noise ratio and to minimize the uncertainty of the weighed amount of solids, as compared to the lab-scale synthesis. In total, approximately 1.5 mg of VO was present.

Figure 5

Low-flip-angle direct excitation (a) and Hahn echo (b) MAS ss-NMR spectra at 70 °C for the VO2/V2O5 mass ratio of 1:4 extracted from real-time experiments at time = 0 (black trace) and time = 38 h (red trace). (c) Normalized 51V integrals vs time for reactions with aged VO2 (blue) or fresh VO2 (Hahn echo red; direct excitation gray) in the reaction mixture. The black curve is a repetition using a Hahn echo.

Interestingly, the synthesis of the V2O5·nH2O nanosheets failed when V2O5 was used solely as the precursor (data not shown), suggesting that VO2(M) or dissolved species formed from VO2(M) initiated the formation of the V2O5·nH2O nanosheets. A requirement of VO2(M) for this synthesis was proposed by Pozarnsky and McCormick who reported on the formation of V4+ species by ion exchange and the consumption of those during the reaction.[23,41] They suggested that [V4+O·5H2O]2+ reacted with [V5+O2·4H2O]+ and formed oligomeric species, which polymerized further. Hence, our observations of a reduction of the V2O5 signal during the course of the reaction might be explained by a homogeneous distribution of V4+ in close vicinity of V5+ formed via polymerization rather than the consumption of V2O5 into other species. Our proposed reaction pathway, which is in agreement with Pozarnsky and McCormick,[23,41] is illustrated in Figure . Alternatively, as Livage discussed, [V4+O·5H2O]2+ could intercalate between the V2O5 layers.[21] This alternative hypothesis would be possible if [V4+O·5H2O]2+ would be homogeneously distributed.

Figure 6

Proposed reaction pathways occurring during the synthesis of V2O5·nH2O nanosheets.

It should also be noted that the synthesis reaction failed when aged VO2(M) was used (Figure c, blue). This aged compound had been stored under ambient conditions and was consequently altered after being in contact with air. To understand the reason behind this phenomenon, we compared the 1H, 51V NMR, ESR spectra, and XRD patterns of the fresh and aged VO2(M).

The XRD pattern of the fresh sample agreed well with the standard pattern of monoclinic VO2(M). By contrast, the XRD pattern of the aged sample had fewer peaks, which complicated the assignment of the formed VO phase (Figure S4). Interestingly, the semilogarithmic plot revealed 001 and 003 reflections of the V2O5·nH2O phase in the pattern of the aged VO2(M), which matched well with previous reports on the instability of VO2(M) under ambient conditions.[44] In comparison with the 51V MAS ss-NMR spectrum of fresh VO2(M) (see Figure S3, red) in which broad peaks resonating between 1800 and 3000 ppm were attributed to VO2(M), the aged VO2 showed no such peaks (see Figure S5, black), presumably due to oxidation of the vanadium species to V2O5. On the other hand, the ESR spectrum (see Figure S2) indicated that the aged VO2 still possessed a reasonable measurable quantity of V4+. Furthermore, the 1H NMR (data not shown) displayed a broad 1H NMR peak for the aged VO2(M) as compared to the fresh one. The increased 1H NMR signal intensity suggested strongly that concurrent H2O uptake had occurred. Hence, as was reported by Etman et al.,[7] the relative fraction of VO2(M)–V2O5 used in the synthesis was crucial for successful exfoliation.

Conclusions

To summarize, real-time solid-state and solution-state 51V NMR studies were performed to follow the transformation of VO2 and V2O5 in aqueous dispersion into nanosheets of V2O5·nH2O. During exfoliation, a loss of the 51V NMR signal of V2O5 was observed, which was attributed to a homogeneous distribution of V4+ that is in close contact with V5+ bleaching their signals. Taken together, our findings were consistent with a hypothesis that both V2O5 and VO2 had been dissolved and VO2 formed as [V4+O·5H2O]2+ cations, which were oligomerized with [V5+O2·4H2O]+ species from V2O5 and then polymerized further. Another explanation could have been intercalation of [V4+O·5H2O]2+ between the layers of V2O5. Additional future studies could include real-time ESR experiments with stirring; however, such were out of the scope of this current study.

Materials and Methods

Materials

The V2O5·nH2O or V2O5·nD2O nanosheets were synthesized as described in ref (7). In a typical synthesis, a mixture of 1:4 (weight ratio) of V2O4 (Fisher Scientific, UK) and V2O4 (Sigma-Aldrich, Germany, purity 99.9%), denoted by VO2(M), was used as the precursors. The mixture of oxides was dispersed in water or D2O (CortecNet, 99.8%) by sonication for 10 min and then heated under reflux at 80–90 °C for 8–24 h. At the end of heating process, a greenish black suspension of V2O5·nH2O nanosheets was formed, which was then dried in air at 80 °C for 5 h to obtain the V2O5·nH2O nanosheets.

For the real-time ss-NMR experiments, 0.3 mg of VO2, 1.2 mg of V2O5, and 20 μL of H2O were placed in a Kel-F insert, which can be sealed with screws. After sealing, the mixture was sonicated for a minute. The insert was then placed in a 4 mm rotor, which was inserted into the spectrometer.

For the real-time solution-state NMR experiments, VO2 and V2O5 with a mass ratio of 1:4 were blended with 550 μL of H2O and 50 μL of D2O. The synthesis was performed at 70 °C, and the sample tube was spun at 20 Hz.

Methods

51V MAS NMR data were acquired on 14.1 T (51V Larmor frequency of −157.9 MHz) and 9.4 T (Larmor frequency of −105.2 MHz) Bruker AVANCE-III spectrometers equipped with a 4 mm or 3.2 mm triple-resonance MAS probe. Real-time MAS 51V NMR spectra were recorded at a MAS rate of 7 kHz, while rotors containing solely solid compounds were spun at the rate of 14 or 24 kHz. The isotropic chemical shift was determined by comparing 51V NMR spectra recorded at two different spinning frequencies. An aqueous solution of sodium metavanadate (1 mol/L) was used to externally calibrate the 51V NMR chemical shift to −574.38 ppm.[45] For solid samples, the length and strength of the radio frequency (rf) pulse were estimated using the reference solution, and a rf pulse and a nominal flip angle of 10° were used for single pulse acquisition. For real-time MAS experiments on the reaction mixture, the rf pulse was calibrated on the sample itself and a 45° nominal flip angle was used. Applied rf fields between 80 and 95 kHz and spectral widths between 2500 and 5000 kHz were used. The carrier frequency was placed on resonance at the isotropic shift for V2O5 as well at VO2(M) and others. The 51V longitudinal relaxation time constants T1 for the solid compounds were estimated to be less than a second at room temperature and at 85 °C. Nevertheless, a repetition delay of 5 s was used for the real-time MAS NMR measurements on the reaction mixture at 85 °C. Static 2H NMR was carried out on V2O5·nD2O nanosheets prepared in D2O instead of H2O. A small piece of the V2O5·nD2O nanosheets was glued in place in a glass insert, which was inserted into the rotor. The 2H NMR spectra were recorded without spinning, and a quadrupolar echo pulse sequence was used. For real-time solution-state NMR measurements, V2O5 and VO2 with a mass ratio of 4:1 were blended with 550 μL of H2O and 50 μL of D2O in a 5 mm NMR tube. The synthesis was performed at 70 °C, and the sample tube was spun at 20 Hz.

A moderately wide Gaussian function, which was optimized to reach close to 0 at the time point the signal had decayed, was pairwise multiplied with the recorded free induction decays before Fourier transformation. Unless otherwise stated, the spectral intensities were normalized to the mass of the sample packed in the rotor. Processing including baseline corrections was performed with in-house scripts (MATLAB, Mathworks).

All ESR experiments were carried out on a Bruker ELEXYS (X-band) spectrometer at room temperature or 80 °C. Samples were studied in a glass capillary, and the magnetic field was stepped from 1000 to 6000 gauss using two or eight signal accumulations. For heat treatment, capillaries were flame sealed prior to storage in the oven at 80 °C.

XRD measurements were conducted in the 2θ range of 5°–45° using a PANalytical diffractometer (Cu Kα1 radiation) and a synchrotron source (SLS beamline, λ = 0.7766 Å). The morphology of the nanosheets and their SAED patterns were studied by a transmission electron microscope (JEOL JEM-2100LaB6) using an accelerating voltage of 200 kV.

pH measurements were performed with a Hanna instrument (model-HI2210). A Quantum Design MPMS XL SQUID (superconducting quantum interference device) magnetometer was used for magnetic characterization. Magnetization versus temperature measurements in a magnetic field of 1 kOe were performed in the temperature range of 2–107 K in steps of 3 K.