2. Current Insights
Additional information about the electrochemical processes involved during Na (de)intercalation is provided by dQ/dV curves at the C/20 rate (black curve in
Figure 2a). The first cycle dQ/dV curve shows a broad cathodic peak at 3.3–3.4 V corresponding to the Na extraction, and an incomplete process of further Na extraction at 4.6 V. The first cathodic peak matches well with the (de)intercalation potential reported for γ-Na
xV
2O
5 [7][9]. Anodic dQ/dV curve at the first discharge reveals Na insertion at potentials of 0.85 V, 0.48 V and 0.25 V. The origin of a small broad peak near 1.5 V is ambiguous and can be tentatively interpreted as a minor amount of embedded Na. dQ/dV curves for further cycles confirm the irreversible character of the Na insertion during the first discharge and no anodic peaks are observed anymore. At the same time, the cathodic curves for second and third cycles reveal a sharp peak at 3.9 V which becomes broader and shifts towards 4.0 V and 4.2 V at fourth and fifth cycles, respectively. These well-pronounced cathodic peaks indicate quantitative sodium extraction from the structure formed upon the first discharge.
Figure 2. dQ/dV curves of Na9V14O35 in Na half-cell for the first five cycles at C/20 (a) and C/5 (b) current rates.
The degradation of the crystal structure supposed from galvanostatic curves is corroborated by ex situ SXRD (
Figure 3) of Na
9V
14O
35 at different states of discharge. The SXRD pattern of the material discharged to 1 V still looks identical to that of the pristine material and reveals close unit cell parameters
a = 15.206(1) Å,
b = 5.0325(8)(1) Å,
c = 20.781(2) Å, β = 109.166(5)°, V = 1502.1(4) Å
3 (
Figure S5). The difference of unit cell volumes for the pristine and 1-V-discharged compounds is about 0.6 Å
3 and falls into the range of two standard deviations, keeping in mind the unit cell volume of 1500 Å
3 (
Table 1). The absence of a significant change of the unit cell volume corresponds to a low discharge capacity of ~60 mAh g
−1 registered at 1 V, since no significant amount of sodium has been intercalated into the structure at this potential. Moreover, [010] SAED pattern and HAADF-STEM image taken from the 1-V-discharged material (
Figure S6) are obviously identical to those of pristine Na
9V
14O
35. Both SXRD and TEM data correlate with the dQ/dV curve in
Figure 2a, showing that no considerable Na insertion occurs above 1 V.
Figure 3. SXRD profiles of Na9V14O35 electrodes at different state of charge: pristine (black), discharged to 1 V (green), 0.25 V (violet) and 0.1 V vs. Na+/Na (turquoise). Wavelength λ = 0.20736 Å.
The SXRD pattern of the 0.25-V-discharged material (
Figure 3, violet curve) demonstrates suppressed but still visible reflections, the positions of which do not match those of pristine Na
9V
14O
35. Although the quality of the SXRD pattern is insufficient for the Rietveld refinement, the reflections still can be indexed with the
P2/
c unit cell of Na
9V
14O
35, but with a much larger
b-parameter resulting in an about 31% increase of the unit cell volume:
a = 15.347(3) Å,
b = 6.301(1) Å,
c = 21.367(3) Å, β = 107.253(9)°, V = 1973.3(8) Å
3 (
Figure S7). The expansion of the structure along the
a- and
c-axes is small (0.9% and 2.8%, respectively), but the expansion along the
b-axis is huge and amounts to ~25.2%. This difference reflects the rigidity of the (010) layers formed by tightly interlinked VO
5 square pyramids and VO
4 tetrahedra, in which the expansion in the
a-
c plane occurs through elongation of the V-O bonds upon reduction of vanadium cations with increase in their ionic radius (r(V
2+) = 0.79 Å, r(V
3+) = 0.64 Å, r(V
4+) = 0.58 Å, r(V
5+) = 0.54 Å, CN = 6)
[24]. Large increase in the interlayer separation is in line with the necessity to provide enough space to accommodate large amounts of Na, as the specific capacity at 0.25 V exceeds 400 mAh g
−1. This must cause structure instability, and indeed, the crystals with another symmetry were found in the 0.25 V-discharged material, as one can see from the SAED patterns and high-resolution HAADF-STEM images, which are typical for a disordered rock-salt (DRS) structure with
F-centered cubic lattice, a unit cell parameter
a ~ 4.7 Å (
Figure 4) and a Na:V ≈ 1:1 atomic ratio (
Table 2). This observation is in agreement with the formation of the DRS structure, in which Na and V atoms randomly occupy the same crystallographic positions. The formation of the DRS structure was observed earlier in the related Li-ion system
[18], in which it demonstrated a stable cycling at anodic potentials. However, the conversion process continues further in Na
9V
14O
35 up to 0.1 V resulting in a complete amorphization as indicated by absence of any reflections in the corresponding SXRD pattern (
Figure 3, turquoise curve). Regarding the Na and V distribution, the 0.1-V-discharged sample is strongly inhomogeneous. It still contains particles with the Na:V ≈ 1:1 atomic ratio (
Table 2), but another phase, strongly enriched with Na up to Na:V ≈ 6.7:1 (
Table 2,
Figure S8), also appears in the sample, and is probably responsible for the high capacity of 490 mAh g
−1. The multicomponent nature of active material at 0.1 V and progressing structural degradation upon further cycles cause the fast decrease of discharge capacity even at low current density (
Figure 5d).
Figure 4. SAED patterns (a–c) indexed in the F-centered cubic lattice and [011] HAADF-STEM image (d), corresponding to DRS structure formed during discharge of Na9V14O35 to 0.25 V vs. Na+/Na.
Figure 5. Galvanostatic charge-discharge curves of Na9V14O35 in Na cells at C/20 (a), C/10 (b) and C/5 (c) current rates. (d) Dependence of the specific discharge capacity on cycle number for different C-rates.
To draw the correlation between the structural transformation and change in the oxidation state of vanadium upon electrochemical cycling, we recorded EELS spectra in the vicinity of V-L
3,2 edge (
Figure 6). The spectra were interpreted in terms of correlation between the vanadium oxidation state and the onset of the V-L
3 edge as proposed by Tan et al.
[25]. The empirical dependence between the V-L
3 edge onset E
V determined at 10% of the maximum height of the V-L
3 edge, and the vanadium formal oxidation V
V state demonstrates a linear increase of E
V with V
V. We have used the linear E
V-V
V equation derived from a set of standard materials by Tan et al.
[25] to estimate the vanadium oxidation state in the samples under investigation. In the pristine material, the onset energy is at E
V = 515.2 eV that corresponds to V
V = +4.3, in good agreement with the average oxidation state of +4.36 from the chemical composition Na
9V
4.1+10O
19(V
5+O
4)
4. Charging to 4.6 V vs. Na
+/Na slightly shifts the V-L
3,2 edge towards a higher energy loss resulting in E
V = 515.4 eV and V
V = +4.5 that corresponds to extraction of 3 Na per Na
9V
14O
35 formula unit as deduced from the electrochemical data and EDX analysis (Na
6V
14O
35, V
V = +4.57). Upon discharge to 1 V and 0.25 V, the V-L
3 edge onset energy is reduced to E
V = 514.8 and 514.3 eV, respectively, providing V
V = +3.9 and +3.5. It should be noted that the vanadium reduction from +4.5 to +3.5 upon discharge from 4.6 V to 0.25 V accounts for only ~250 mAh g
−1 capacity that corresponds to the end of the first discharge plateau. Thus, the capacity of the second and third discharge plateaus should be attributed to a conversion reaction with the formation of the Na-rich phase. Unfortunately, the two-phase nature of the sample discharged to 0.1 V and strong Na inhomogeneity even within the same crystallite (
Figure S8) prevent sensible EELS measurement of V
V.
Figure 6. EELS spectra of Na9V14O35 at different states of charge in the vicinity of the V-L3,2 edge.