Exploring the Effect of V2O5 and Nb2O5 Content on the Structural, Thermal, and Electrical Characteristics of Sodium Phosphate Glasses and Glass–Ceramics

Na-V-P-Nb-based materials have gained substantial recognition as cathode materials in high-rate sodium-ion batteries due to their unique properties and compositions, comprising both alkali and transition metal ions, which allow them to exhibit a mixed ionic–polaronic conduction mechanism. In this study, the impact of introducing two transition metal oxides, V2O5 and Nb2O5, on the thermal, (micro)structural, and electrical properties of the 35Na2O-25V2O5-(40 − x)P2O5 − xNb2O5 system is examined. The starting glass shows the highest values of DC conductivity, σDC, reaching 1.45 × 10−8 Ω−1 cm−1 at 303 K, along with a glass transition temperature, Tg, of 371 °C. The incorporation of Nb2O5 influences both σDC and Tg, resulting in non-linear trends, with the lowest values observed for the glass with x = 20 mol%. Electron paramagnetic resonance measurements and vibrational spectroscopy results suggest that the observed non-monotonic trend in σDC arises from a diminishing contribution of polaronic conductivity due to the decrease in the relative number of V4+ ions and the introduction of Nb2O5, which disrupts the predominantly mixed vanadate–phosphate network within the starting glasses, consequently impeding polaronic transport. The mechanism of electrical transport is investigated using the model-free Summerfield scaling procedure, revealing the presence of mixed ionic–polaronic conductivity in glasses where x < 10 mol%, whereas for x ≥ 10 mol%, the ionic conductivity mechanism becomes prominent. To assess the impact of the V2O5 content on the electrical transport mechanism, a comparative analysis of two analogue series with varying V2O5 content (10 and 25 mol%) is conducted to evaluate the extent of its polaronic contribution.


Introduction
The rapid progression of battery technology is reshaping the landscape of energy storage, wherein sodium-ion batteries (SIBs) are emerging as a promising alternative to their lithium-ion counterparts.This shift is attributed to the cost-effectiveness, abundance, and widespread availability of sodium compared with lithium [1,2].Despite the lower energy density resulting from sodium's heavier mass, SIBs are being explored for grid-level applications where cost considerations take precedence over ultra-high energy density demands.In this context, a critical factor influencing the performance of SIBs is the cathode material [1,2].The strategic design and development of such material, capable of delivering high capacity and voltage while maintaining electrochemical, thermal, and mechanical stability, significantly contribute to enhancing the energy density of the battery.
In the exploration of diverse cathode materials, particular emphasis is placed on systems incorporating both sodium and transition metal (TM) ions, such as V, Mn, Fe, Co, and Ni, among others.These combinations yield a range of materials, such as those with a 2D layered transition metal oxide (TMO) structure (e.g., NaCoO 2 , NaNi 0.5 Mn 0.5 O 2 ) or polyanionic compounds featuring a 3D structure (e.g., NaFePO 4 , Na 3 V 2 (PO 4 ) 3 ) capable of hosting and releasing Na + ions [1,2].An additional advantage of Na-TM systems is their potential to exhibit mixed ionic-electronic conductivity, wherein the electronic contribution relies on the existence of a TM n+ /TM (n+1)+ redox pair, adhering to the principles of the small polaron hopping (SPH) theory [3].Each of the two contributions plays a significant role, with the redox centres facilitating charge transfer during the processes of Na + intercalation/deintercalation and the ionic contribution promoting faster diffusion of Na + ions [4].
Among the aforementioned cathode materials, NASICON-structured compounds stand out for their outstanding electrochemical performance, structural stability, and high ionic conductivity [4,5].Particularly noteworthy are those derived from the Na-V-P-based system, such as Na 3 V 2 (PO 4 ) 3 , which exhibits conductivity on the order of ~10 −8 Ω −1 cm −1 at 298 K [6].Nevertheless, its rate performance is significantly impeded by the intrinsic limitation of low electronic conductivity.Various strategies have been devised to tackle this challenge.Notably, doping with ions like Ni 2+ [7], Nb 5+ [8][9][10], Mo 6+ [11], and W 6+ [12] has emerged as an effective approach.This technique aids in the formation of V mixed valence states, thereby enhancing electronic conductivity.Furthermore, the partial substitution of Na and/or V with ions possessing larger radii can further promote Na + mobility by expanding the migration channels for Na + .The addition of Nb 5+ ions has proven particularly effective, markedly enhancing the intrinsic electronic conductivity and Na + mobility in Na 3 V 2 (PO 4 ) 3 , thereby reducing electrode polarization and enhancing rate capability [8][9][10].
In addition to Na-V-P crystalline materials, their glassy and glass-ceramic counterparts play a prominent role in the research and development of cathode materials.Noteworthy for their unique properties, these materials exhibit characteristics such as the absence of grain boundaries, isotropic conductivity, and exceptional compositional flexibility [13,14].Moreover, Na-V-P glass and glass-ceramic systems provide an additional advantage with their capability to control the mechanism of electrical conductivity via straightforward adjustments in composition [15,16].Specifically, research into the electrical properties of the ternary A-V-P system (A = Li, Na, K) revealed that the conductivity mechanism varies with the ratio r = [A 2 O]/([V 2 O 5 ] + [P 2 O 5 ]) [17][18][19][20][21]. Glasses featuring smaller r values demonstrate the highest conductivity and a predominantly polaronic mechanism, while those with higher r values tend to be predominantly ionic conductors and those in between display mixed ionic-electronic (polaronic) conductivity.
Building upon the aforementioned discoveries regarding both crystalline and glassy Na-V-P-based materials, and drawing on our proficiency in exploring the electrical properties of diverse alkali-TMO-containing phosphate-based systems manifesting distinct electrical conductivity mechanisms (ionic [22], polaronic [23], mixed ionic-polaronic [24][25][26]), we extended our research scope to encompass Na-V-P-based glass-(ceramic) systems incorporating Nb 2 O 5 .In our prior research, we examined the influence of Nb 2 O 5 addition on the thermal, (micro)structural, and electrical characteristics of 35Na 2 O-10V 2 O 5 -(55 − x)P 2 O 5 − xNb 2 O 5 (x = 0-40, mol%) glass and glass-ceramic series [27].Solid-state impedance spectroscopy (SS-IS) measurements unveiled a purely ionic mechanism of electrical conductivity, despite the presence of 10 mol% of V 2 O 5 .Non-monotonic trends observed in DC conductivity and activation energy were ascribed to the facilitating effect of Nb 2 O 5 on Na + ion transport, indicating the mixed glass former effect (MGFE).Furthermore, it has been shown that this series mirrors the behaviour demonstrated by the majority of A-P-Nb (A = Li [28], Na [29][30][31][32][33], K [34]) glass and glass-ceramic systems, where the addition of Nb 2 O 5 enhances both the thermal and electrical properties.
To further elucidate the impact of V 2 O 5 content on the observed trends resulting from the gradual substitution of P 2 O 5 with Nb 2 O 5 in Na-V-P-based systems, we decided to investigate 35Na 2 O-25V 2 O 5 -(40 − x)P 2 O 5 − xNb 2 O 5 (x = 0-35, mol%) glass and glass-ceramic series with higher V 2 O 5 content.This study investigates the impact of introducing V 2 O 5 and Nb 2 O 5 on the glass-forming region along with the thermal, (micro)structural, and electrical properties of the given system.SS-IS reveals the initial glass as the optimal sample, showcasing the highest DC conductivity, σ DC (1.45 × 10 −8 Ω −1 cm −1 at 303 K), and glass transition temperature, T g (371 • C).The incorporation of Nb 2 O 5 gives rise to non-monotonic trends in both σ DC and T g , with the lowest values at x = 20 mol%.Insights from vibrational spectroscopy indicate that these trends originate from the disruption of the mixed vanadate-phosphate network upon the addition of Nb 2 O 5 , thereby impeding polaronic transport.Using the model-free Summerfield scaling procedure, mixed ionic-polaronic conductivity is observed for x < 10 mol%, while for x ≥ 10 mol%, ionic conductivity prevails.A comparative analysis of two analogue Na-Nb-P series, with 10 mol% V 2 O 5 [27] and 25 mol% V 2 O 5 , evaluates the impact of V 2 O 5 content on the electrical transport mechanism and the extent of its polaronic contribution.

Results and Discussion
The glass-forming region (GFR) in the quaternary system 35Na 2 O-25V 2 O 5 -(40 − x)P 2 O 5 − xNb 2 O 5 (25V series) is observed to be relatively wide, incorporating Nb 2 O 5 up to 25 mol%.Beyond the GFR, spontaneous crystallization occurs, leading to the formation of partially crystallized (glass-ceramic) samples, 25V-30Nb and 25V-35Nb.Table 1 presents the batch compositions of the prepared glasses and glass-ceramics, accompanied by selected physical properties.
Table 1.Batch composition and selected properties for all glass and glass-ceramic samples.

General Properties and Thermal Behaviour
The density, ρ, of the samples demonstrates an increasing trend with the addition of Nb 2 O 5 ; see Figure 1a.Interestingly, the molar volume, V M , exhibits a non-monotonic pattern as the content of Nb 2 O 5 increases, reaching its peak for glass at 25V-20Nb; see Figure 1a.This observed behaviour is surprising, given that V M typically follows the opposite trend to that of ρ [29][30][31][32][33][34].Changes in ρ and V M are commonly attributed to two factors: (i) the difference in molar masses-Nb 2 O 5 (265.81g/mol) having a greater molar mass than P 2 O 5 (141.943g/mol) and (ii) the difference in bond strengths-Nb-O bond being stronger (772 kJ/mol [35]) compared to the P-O bond (599 kJ/mol [35]).While these factors straightforwardly account for the increasing trend in ρ, comprehending the unconventional V M trend is more intricate.This phenomenon can be explained as a result of incorporating Nb 2 O 5 in the form of voluminous octahedral NbO 6 structural units [36], which disrupt the initially compact mixed vanadate-phosphate (V-P) network, where vanadate and phosphate units are mutually interconnected via V-O-P bonds.This transformation leads to a more "open" network, as indicated by the rising V M values.
Nb2O5 in the form of voluminous octahedral NbO6 structural units [36], which disrupt the initially compact mixed vanadate-phosphate (V-P) network, where vanadate and phosphate units are mutually interconnected via V-O-P bonds.This transformation leads to a more "open" network, as indicated by the rising VM values.The DTA curves for each sample exhibit an endothermic effect indicative of the glass transition, Tg, along with one or more exothermic signals associated with the crystallization process, Tc; see Figure 1b.The temperature of the onset of Tg and the peak temperature of the first Tc for all the samples are determined and listed in Table 1, along with the quantity (Tc − Tg), representing the thermal stability (TS) of the glasses.The dependence of Tg on Nb2O5 content in the 25V series is illustrated in Figure 1c.Similar to VM, the Tg trend follows a non-monotonic pattern, gradually decreasing until reaching a minimum for the 25V-20Nb glass, after which it begins to rise.These findings diverge from the anticipated thermal behaviour observed in similar glass systems, where Tg typically increases nearly linearly with the addition of Nb2O5 [27,[29][30][31][32][33][34].In contrast to the anticipated rise in Tg with the incorporation of Nb2O5, the 25V series displays a minimum, and Tg values exhibit only small changes upon the introduction of Nb2O5.
Variations in Tg typically arise from the same factors influencing ρ and VM, such as the replacement of weaker P-O bonds with stronger Nb-O bonds and the formation of a more condensed mixed niobate-phosphate (Nb-P) network, leading to an increase in Tg in previously reported series [27,[29][30][31][32][33][34].However, the non-monotonic trend in Tg within the 25V series indicates the unfavourable influence of Nb2O5 on Tg.As elucidated concerning VM, this behaviour stems from the incorporation of niobate units that disrupt the initial mixed V-P network.As the predominantly mixed niobate-vanadate (Nb-V) network forms for x > 20 mol%, a slight elevation in Tg is observed, with the addition of more Nb2O5 exerting a positive influence on the Tg values.On the contrary, the TS of the glasses, assessed through the quantity (Tc − Tg) [37], increases upon the initial incorporation of Nb2O5 up to 15 mol% and stabilizes for glasses with 15 ≤ x ≤ 25 (see Figure 1c), revealing a beneficial influence of Nb2O5 on the TS of the investigated glass compositions.Moreover, the DTA curves of the partially crystalline samples, 25V-30Nb and 25V-35Nb, reveal a glass The DTA curves for each sample exhibit an endothermic effect indicative of the glass transition, T g , along with one or more exothermic signals associated with the crystallization process, T c ; see Figure 1b.The temperature of the onset of T g and the peak temperature of the first T c for all the samples are determined and listed in Table 1, along with the quantity (T c − T g ), representing the thermal stability (TS) of the glasses.The dependence of T g on Nb 2 O 5 content in the 25V series is illustrated in Figure 1c.Similar to V M , the T g trend follows a non-monotonic pattern, gradually decreasing until reaching a minimum for the 25V-20Nb glass, after which it begins to rise.These findings diverge from the anticipated thermal behaviour observed in similar glass systems, where T g typically increases nearly linearly with the addition of Nb 2 O 5 [27,[29][30][31][32][33][34].In contrast to the anticipated rise in T g with the incorporation of Nb 2 O 5 , the 25V series displays a minimum, and T g values exhibit only small changes upon the introduction of Nb 2 O 5 .
Variations in T g typically arise from the same factors influencing ρ and V M , such as the replacement of weaker P-O bonds with stronger Nb-O bonds and the formation of a more condensed mixed niobate-phosphate (Nb-P) network, leading to an increase in T g in previously reported series [27,[29][30][31][32][33][34].However, the non-monotonic trend in T g within the 25V series indicates the unfavourable influence of Nb 2 O 5 on T g .As elucidated concerning V M , this behaviour stems from the incorporation of niobate units that disrupt the initial mixed V-P network.As the predominantly mixed niobate-vanadate (Nb-V) network forms for x > 20 mol%, a slight elevation in T g is observed, with the addition of more Nb 2 O 5 exerting a positive influence on the T g values.On the contrary, the TS of the glasses, assessed through the quantity (T c − T g ) [37], increases upon the initial incorporation of Nb 2 O 5 up to 15 mol% and stabilizes for glasses with 15 ≤ x ≤ 25 (see Figure 1c), revealing a beneficial influence of Nb 2 O 5 on the TS of the investigated glass compositions.Moreover, the DTA curves of the partially crystalline samples, 25V-30Nb and 25V-35Nb, reveal a glass transition.The T g values associated with these transitions are markedly lower in comparison with the glasses within the same series.As detailed later in the text, both glass-ceramics feature significant amounts of the Na 13 Nb 35 O 94 crystalline phase.Thus, it can be inferred that the diminished T g values stem from the residual glassy matrix undergoing depletion of sodium and niobium, which participate in the formation of a confirmed crystalline phase.

PXRD and SEM-EDS Analysis
The PXRD method validates the successful preparation of glasses over a broad composition range (0 ≤ x ≤ 25), indicating their amorphous nature as evidenced by the absence of diffraction maxima, with only the characteristic amorphous "halo" present in the diffraction patterns; see Figure 2a.However, in melts with higher Nb 2 O 5 amounts (x ≥ 30), spontaneous crystallization occurs during cooling, yielding glass-ceramic, a composite material consisting of both crystalline and amorphous phases.Both partially crystallized samples, 25V-30Nb and 25V-35Nb, contain a crystalline Na 13 Nb 35 O 94 phase (24882-ICSD) [38] in addition to the amorphous phase.Quantitative phase analysis of both glass-ceramics performed using the Rietveld refinement method shows the 25V-30Nb sample contains 29 wt.% of the crystalline phase Na transition.The Tg values associated with these transitions are markedly lower in comparison with the glasses within the same series.As detailed later in the text, both glass-ceramics feature significant amounts of the Na13Nb35O94 crystalline phase.Thus, it can be inferred that the diminished Tg values stem from the residual glassy matrix undergoing depletion of sodium and niobium, which participate in the formation of a confirmed crystalline phase.

PXRD and SEM-EDS Analysis
The PXRD method validates the successful preparation of glasses over a broad composition range (0 ≤ x ≤ 25), indicating their amorphous nature as evidenced by the absence of diffraction maxima, with only the characteristic amorphous "halo" present in the diffraction patterns; see Figure 2a.However, in melts with higher Nb2O5 amounts (x ≥ 30), spontaneous crystallization occurs during cooling, yielding glass-ceramic, a composite material consisting of both crystalline and amorphous phases.Both partially crystallized samples, 25V-30Nb and 25V-35Nb, contain a crystalline Na13Nb35O94 phase (24882-ICSD) [38] in addition to the amorphous phase.Quantitative phase analysis of both glass-ceramics performed using the Rietveld refinement method shows the 25V-30Nb sample contains 29 wt.% of the crystalline phase Na13Nb35O94, with the remaining 71 wt.% being amorphous glass matrix; see Figure 2b.With further introduction of Nb2O5, the proportion of the crystalline phase Na13Nb35O94 embedded into the glass matrix for the 25V-35Nb glassceramic sample increases to 45 wt.%.SEM-EDS microscopy is employed to gain insight into the morphology and elemental composition of the prepared samples.In the case of the glassy samples, the SEM micrograph confirms the amorphous structure and uniform distribution of the elements.Conversely, for the partially crystallized ones obtained by spontaneous crystallization beyond the GFR boundary, the SEM micrograph reveals areas with distinct microstructural characteristics.As illustrated in Figure 3a, the 25V-30Nb glass-ceramic features grains that are uniformly embedded in the amorphous glassy matrix.These grains assume the shape of spheres, with a diameter ranging from ~500 nm to ~2 µm.SEM-EDS microscopy is employed to gain insight into the morphology and elemental composition of the prepared samples.In the case of the glassy samples, the SEM micrograph confirms the amorphous structure and uniform distribution of the elements.Conversely, for the partially crystallized ones obtained by spontaneous crystallization beyond the GFR boundary, the SEM micrograph reveals areas with distinct microstructural characteristics.As illustrated in Figure 3a, the 25V-30Nb glass-ceramic features grains that are uniformly embedded in the amorphous glassy matrix.These grains assume the shape of spheres, with a diameter ranging from ~500 nm to ~2 µm.
Furthermore, the SEM micrograph of the 25V-35Nb glass-ceramic shown in Figure S1 in the Supplementary Materials (SM) reveals that with the addition of more Nb 2 O 5 , the morphology further develops with the ongoing crystallization process.Fused grains take on the form of plate-shaped grains, varying in size from approximately 2 to 20 µm in diameter.These grains are also uniformly embedded in the remaining amorphous glassy matrix.
The EDS maps depicted in Figures 3b,c and S1b,c in the SM visually illustrate the distribution and variation of the chemical elements across the two glass-ceramic samples.Examination of the maps for the individual elements O, Na, V, Nb, and P reveals distinct compositional differences between the grains and the residual glass matrix area, as shown in Figure 3d,e and Figure S1d,e in the SM.In the grain areas, the elements Na, Nb, and O are concentrated, while V and P are present only in trace amounts on the surface.This observation aligns with the EDS analysis (see Figure S1f in the SM), which indicates that the composition of the observed grains closely matches that of the Na 13 Nb 35 O 94 crystalline phase.As the identified crystalline phase progressively crystallizes with the increasing Nb 2 O 5 content, the amorphous glass matrix is depleted of Na, Nb, and O elements, while the elements V and P remain in the glass matrix.This is particularly evident in the 25V-35Nb sample (see Figure S1c in the SM), where the EDS analysis reveals that the composition of the grains (area 2) corresponds to that of the Na 13 Nb 35 O 94 crystalline phase, while the glassy matrix (area 3) becomes enriched in V, as illustrated in Figure S1g in the SM.The EDS maps depicted in Figures 3b,c and S1b,c in the SM visually illustrate distribution and variation of the chemical elements across the two glass-ceramic samp Examination of the maps for the individual elements O, Na, V, Nb, and P reveals dist compositional differences between the grains and the residual glass matrix area, as sho in Figure 3d,e and Figure S1d,e in the SM.In the grain areas, the elements Na, Nb, an are concentrated, while V and P are present only in trace amounts on the surface.observation aligns with the EDS analysis (see Figure S1f in the SM), which indicates the composition of the observed grains closely matches that of the Na13Nb35O94 crysta phase.As the identified crystalline phase progressively crystallizes with the increa Nb2O5 content, the amorphous glass matrix is depleted of Na, Nb, and O elements, w the elements V and P remain in the glass matrix.This is particularly evident in the 2 35Nb sample (see Figure S1c in the SM), where the EDS analysis reveals that the com sition of the grains (area 2) corresponds to that of the Na13Nb35O94 crystalline phase, w the glassy matrix (area 3) becomes enriched in V, as illustrated in Figure S1g in the SM

Vibrational Spectroscopy-Raman and IR-ATR Studies
The effect of the gradual substitution of P2O5 with Nb2O5 on the structural charac istics of the glass and glass-ceramic samples from the studied 25V series is investig using two complementary vibrational spectroscopy techniques-Raman and attenu total reflectance infrared (IR-ATR) spectroscopy.Figure 4 illustrates the changes in Ram

Vibrational Spectroscopy-Raman and IR-ATR Studies
The effect of the gradual substitution of P 2 O 5 with Nb 2 O 5 on the structural characteristics of the glass and glass-ceramic samples from the studied 25V series is investigated using two complementary vibrational spectroscopy techniques-Raman and attenuated total reflectance infrared (IR-ATR) spectroscopy.Figure 4 illustrates the changes in Raman and IR-ATR spectra resulting from the incorporation of Nb 2 O 5 , and a detailed list of observed bands and their assignments, sourced from the literature data [29][30][31][32][33][34]36,, can be found in Table S1 in the SM.
The Raman and IR-ATR spectra for the initial 25V-0Nb glass reveal a mixed glass network composed of phosphate and vanadate units.The bands observed between 1045 and 1285 cm −1 indicate the presence of Q 2 and Q 1 phosphate units, while the most prominent signal, occurring at ~965 cm −1 , originates from the symmetric stretching vibrations of the double V=O bond.The latter band may also include a signal corresponding to isolated Q 0 phosphate units, whose prevalence is expected for the O/P ratio of 4.5; see Table 1.However, the strong Raman scattering tendency of transition metal oxides typically masks the Raman response from phosphate units, thereby concealing the Q 0 signals beneath those originating from the vanadate units.Furthermore, the signal at ~850 cm −1 signifies the stretching of the V-O bonds within VO x units, whereas the signal at ~750 cm −1 corresponds to the stretching of the bridging V-O-V bonds within mutually interconnected VO x units.The presence of a band at ~640 cm −1 , associated with V-O-P bonding, confirms the formation of a mixed V-P glass network.Additionally, the signal at ~470 cm −1 aligns with the stretching vibrations of the V-O-V bonds in metavanadate chains, suggesting a substantial level of crosslinking among the vanadate structural units.observed bands and their assignments, sourced from the literature data [29][30][31][32][33][34]36,, can be found in Table S1 in the SM.The Raman and IR-ATR spectra for the initial 25V-0Nb glass reveal a mixed glass network composed of phosphate and vanadate units.The bands observed between 1045 and 1285 cm −1 indicate the presence of Q 2 and Q 1 phosphate units, while the most prominent signal, occurring at ~965 cm −1 , originates from the symmetric stretching vibrations of the double V=O bond.The latter band may also include a signal corresponding to isolated Q 0 phosphate units, whose prevalence is expected for the O/P ratio of 4.5; see Table 1.However, the strong Raman scattering tendency of transition metal oxides typically masks the Raman response from phosphate units, thereby concealing the Q 0 signals beneath those originating from the vanadate units.Furthermore, the signal at ~850 cm −1 signifies the stretching of the V-O bonds within VOx units, whereas the signal at ~750 cm −1 corresponds to the stretching of the bridging V-O-V bonds within mutually interconnected VOx units.The presence of a band at ~640 cm −1 , associated with V-O-P bonding, confirms the formation of a mixed V-P glass network.Additionally, the signal at ~470 cm −1 aligns with the stretching vibrations of the V-O-V bonds in metavanadate chains, suggesting a substantial level of crosslinking among the vanadate structural units.
The changes in Raman spectra resulting from the incorporation of Nb2O5 are depicted in Figure 4a.Upon the initial introduction of Nb2O5 (x = 5), only a subtle increase in the intensity of the signal at 965 cm −1 is observed, and the spectra of the two glasses, 25V-0Nb and 25V-5Nb, exhibit nearly identical features.More significant changes in the spectra become evident for x ≥ 10, as the intensities of the bands within the 520-1030 cm −1 range progressively increase.The most noticeable alteration in shape occurs in the cases of 25V-15Nb and 25V-20Nb glasses, where the signal at ~850 cm −1 emerges as the strongest one.The changes in Raman spectra resulting from the incorporation of Nb 2 O 5 are depicted in Figure 4a.Upon the initial introduction of Nb 2 O 5 (x = 5), only a subtle increase in the intensity of the signal at 965 cm −1 is observed, and the spectra of the two glasses, 25V-0Nb and 25V-5Nb, exhibit nearly identical features.More significant changes in the spectra become evident for x ≥ 10, as the intensities of the bands within the 520-1030 cm −1 range progressively increase.The most noticeable alteration in shape occurs in the cases of 25V-15Nb and 25V-20Nb glasses, where the signal at ~850 cm −1 emerges as the strongest one.In contrast, the shape of the Raman spectrum of the 25V-10Nb glass closely resembles those of the starting 25V-0Nb glass and 25V-5Nb glasses, suggesting that this composition marks the point where the structural transition occurs.The spectrum of the 25V-25Nb glass differs from the others, as the signal at ~680 cm −1 exhibits a significant increase in intensity, establishing itself as the dominant signal alongside the one at ~850 cm −1 .The changes in spectral shape and signal intensities observed with the increasing content of Nb 2 O 5 can be ascribed to distinct structural motifs formed by Nb 2 O 5 within the glass structure [32][33][34]61].These structural features give rise to vibrations categorized into three distinct domains: (i) short Nb-O bonds within highly distorted NbO 6 units (900-1030 cm −1 , Domain I), (ii) NbO 6 octahedra linked into chains via Nb-O-Nb bonds (750 to 900 cm −1 , Domain II), and (iii) 3D network of less distorted corner-shared NbO 6 units (520-750 cm −1 , Domain III).
As mentioned earlier, in the case of glasses with 10 ≤ x ≤ 20, the most prominent signal is observed within Domain II.Despite the overlap with signals attributed to the V-O bonds in VO x units, its intensity consistently amplifies with rising Nb 2 O 5 content, thus suggesting that niobate units predominantly participate in the formation of a chain-like structure.Moreover, a band at approximately ~450 cm −1 , indicative of P-O-Nb bonding, becomes more prominent, thereby confirming cross-linking between the PO 4 and NbO 6 units.However, with the addition of more Nb 2 O 5 , the intensity of this band gradually decreases.Meanwhile, the signal within Domain III broadens and strengthens, suggesting a rising number of less-distorted corner-shared NbO 6 units connected into chains via Nb-O-Nb bonds.Additionally, it is worth noting that signals arising from mixed V-O-Nb bridging bonds may exist in this range; nevertheless, the overlapping signals describing distinct vibration modes of the vanadate and niobate units make their detection challenging.In the spectrum of the 25V-25Nb glass, a change in the intensity ratio between the two maxima in Domains II and III occurs, with the signal in Domain III becoming the strongest, indicative of the predominant participation of NbO 6 units in the formation of 3D clusters.The Raman spectra of the 25V-30Nb and 25V-35Nb glass-ceramics further highlight a pronounced presence of the signal in Domain III.This observation is anticipated, as both compositions encompass substantial amounts of the Na 13 Nb 35 O 94 crystal phase, distinguished by its crystal structure featuring a 3D network of edge-and corner-shared NbO 6 octahedra.
According to these findings, the investigated glasses-(ceramics) can be categorized into three compositional regions based on their structural characteristics: (i) a predominantly mixed V-P network (x < 10, Region I); (ii) a mixed Nb-V-P glass network (10 ≤ x ≤ 20, Region II), where NbO 6 units participate in the formation of chains through Nb-O-Nb bonds; and (iii) a predominantly mixed Nb-V network (x ≥ 25, Region III), where NbO 6 units tend to cluster, forming a 3D network (see Figure 4b-g).
The results of IR-ATR analysis reveal a structural evolution similar to that observed in the Raman spectra (see Figure 4b-g), showcasing a transition from a predominantly mixed V-P network to a predominantly mixed Nb-V network upon the substitution of P 2 O 5 with Nb 2 O 5 .Notably, a majority of the bands are present in both the IR-ATR and Raman spectra.Nevertheless, while the Raman spectra are characterized by strong, narrow bands linked to symmetric stretching vibrations, the IR-ATR spectra take on a more complex form, where broad yet intense bands, attributed to asymmetric stretching and bending motions, prevail [62,63].This is apparent when comparing the Raman and IR-ATR spectra of the initial 25V-0Nb glass (see Figure 4b), where the two strongest signals in the IR-ATR spectrum correspond to the deformation modes of the phosphate and vanadate units and the asymmetric stretching of the bridging oxygen atoms.Because Raman spectra often obscure signals associated with phosphate units beneath those emanating from TMO polyhedra owing to their notably greater polarizability [33,34], utilising IR-ATR spectroscopy, which preserves information concerning phosphate units, serves as an additional avenue for achieving a more comprehensive understanding of the structure of the studied glasses.

Electrical Properties 2.4.1. Complex Impedance Plane and Direct Current (DC) Conductivity
Figure 5a presents the conductivity spectra at various temperatures for glass 25V-20Nb, chosen as a representative for the entire series.The spectra reveal two distinct regions: (i) a frequency-independent segment referred to as the DC plateau, σ DC , which is linked to the long-range motion of charge carriers, and (ii) a frequency-dependent region (AC part) observed at lower temperatures and higher frequencies, i.e., the dispersion that arises from the localized motion of charge carriers over shorter distances.Additionally, upon a thorough examination of the conductivity spectra, it is evident that glasses where x ≥ 10 exhibit a slight decline in conductivity at the highest temperatures and the lowest frequencies, which points to the emergence of the electrode polarization effect.This phenomenon, indicative of ionic conductivity, arises due to the accumulation of mobile Na + ions at the surface of the blocking gold electrodes and is further validated through the analysis of the complex impedance spectra depicted in Figure 5b.
upon a thorough examination of the conductivity spectra, it is evident that glasses where x ≥ 10 exhibit a slight decline in conductivity at the highest temperatures and the lowest frequencies, which points to the emergence of the electrode polarization effect.This phenomenon, indicative of ionic conductivity, arises due to the accumulation of mobile Na + ions at the surface of the blocking gold electrodes and is further validated through the analysis of the complex impedance spectra depicted in Figure 5b.Specifically, the presentation of experimental data through complex impedance spectra provides a comprehensive insight into the diverse processes in the electrical response of the investigated samples, with all impedance spectra analysed by modelling with the corresponding electrical equivalent circuit (EEC) using the complex non-linear leastsquares (CNLLSQ) method.At 303 K, the impedance spectra for all samples exhibit a depressed semicircle, indicating the bulk process where the intercept on the Z' axis corresponds to the DC resistance, R, at the given temperature.These impedance spectra are described using the EEC model featuring a parallel combination of the resistor (R) and a constant phase element (CPE).However, at higher temperatures, the formation of an additional low-frequency "spur" can be observed in the impedance spectra of samples where x ≥ 10, a characteristic typical of ionically conducting glasses, which corresponds to the aforementioned electrode polarization effect.The EEC model for these spectra comprises a parallel R-CPE combination connected in series with the CPE.As illustrated in Figure 5b, the experimental data closely align with the theoretical curves, and the DC conductivity, σDC, is calculated using the formula σDC = t/(A × R), where t is the sample thickness, A is the electrode area, and R is the electrical resistance.The corresponding σDC values at 303 K, listed in Table 2, demonstrate close agreement between the DC conductivity derived from the DC plateau and the values obtained through EEC modelling.Specifically, the presentation of experimental data through complex impedance spectra provides a comprehensive insight into the diverse processes in the electrical response of the investigated samples, with all impedance spectra analysed by modelling with the corresponding electrical equivalent circuit (EEC) using the complex non-linear leastsquares (CNLLSQ) method.At 303 K, the impedance spectra for all samples exhibit a depressed semicircle, indicating the bulk process where the intercept on the Z' axis corresponds to the DC resistance, R, at the given temperature.These impedance spectra are described using the EEC model featuring a parallel combination of the resistor (R) and a constant phase element (CPE).However, at higher temperatures, the formation of an additional low-frequency "spur" can be observed in the impedance spectra of samples where x ≥ 10, a characteristic typical of ionically conducting glasses, which corresponds to the aforementioned electrode polarization effect.The EEC model for these spectra comprises a parallel R-CPE combination connected in series with the CPE.As illustrated in Figure 5b, the experimental data closely align with the theoretical curves, and the DC conductivity, σ DC , is calculated using the formula σ DC = t/(A × R), where t is the sample thickness, A is the electrode area, and R is the electrical resistance.The corresponding σ DC values at 303 K, listed in Table 2, demonstrate close agreement between the DC conductivity derived from the DC plateau and the values obtained through EEC modelling.

DC Conductivity and Activation Energy
Figure 6a illustrates the temperature-dependent variation in σ DC for different Nb 2 O 5 content, demonstrating that the DC conductivity is temperature-activated and exhibiting Arrhenius temperature dependence with characteristic activation energy, as shown in Figure 6b.The activation energy for DC conductivity, E DC , is calculated from the slope of the log(σ DC T) vs. 1000/T, utilising the Arrhenius equation (Equation ( 1)), where T represents the temperature in K, σ 0 * is the pre-exponential factor, and k B is the Boltzmann constant.The σ DC at 303 K and the E DC values for all samples are listed in Table 2, and they closely align with those reported for sodium phosphate-based glasses containing niobium [30,31] or vanadium [15,16] with compositions similar to those in this study.
25V  where T represents the temperature in K, σ0* is the pre-exponential factor, and kB is the Boltzmann constant.The σDC at 303 K and the EDC values for all samples are listed in Table 2, and they closely align with those reported for sodium phosphate-based glasses containing niobium [30,31] or vanadium [15,16] with compositions similar to those in this study.Within the 25V series, both the DC conductivity and activation energy exhibit a nonmonotonic trend.While σ DC shows a decreasing trend with increasing Nb 2 O 5 content, E DC exhibits an opposite trend compared with σ DC , as anticipated.The highest σ DC value corresponds to the starting glass, 25V-0Nb, and with the introduction of Nb 2 O 5 , it undergoes a reduction by approximately three orders of magnitude, reaching a minimum for the 25V-20Nb glass.This decrease is most pronounced for glasses with x ≤ 10, showing a drop in σ DC of around 2.5 orders of magnitude.Beyond this point, σ DC values exhibit less substantial changes.Similar observations apply to E DC , showing a progressive increase for x ≤ 10 and stabilization for 15 ≤ x ≤ 25; see Table 2. Furthermore, the σ DC values for two glass-ceramic samples, 25V-30Nb and 25V-35Nb, exhibit a noticeable decrease when compared with those of glasses from the 25V series.This decrease in conductivity can be attributed to the significant presence of the Na 13 Nb 35 O 94 phase in the two glass-ceramic samples, as its crystallization leads to the depletion of sodium ions within the residual glassy matrix.Moreover, the insufficient interconnection among grains of the Na 13 Nb 35 O 94 crystal phase hinders the creation of an easy conductive pathway for the transport of Na + ions, further contributing to the observed drop in conductivity.
Returning to the non-monotonic trends in σ DC and E DC , it is noteworthy that the observed behaviour strikingly resembles the conductivity minimum seen in mixed ionicpolaronic systems containing both alkali and vanadium ions [15][16][17][18][19][20][21].However, unlike these systems, where the gradual transition in the conduction mechanism from predominantly one mechanism (ionic-polaronic) to predominantly another is induced by changes in the amount of transition metal oxide and/or alkali metal oxide, in the 25V series, the quantities of both Na 2 O and V 2 O 5 remain constant.This is reflected in the constant number density of the sodium ions, N V (Na + ), calculated from the glass composition and density; see Table 2 and Figure 6c.Nonetheless, the number density of the polarons arising from the presence of reduced vanadium ions in the +4 oxidation state, N V (V 4+ /V total ), calculated as the product of the number density of vanadium ions and the fraction of V 4+ ions, is observed to gradually decrease with increasing Nb 2 O 5 content, having a negative impact on the total number density of the charge carriers, N V (Na + ) + N V (V 4+ ).See Table 2 and Figure 6c.It is important to note that the main factor influencing the observed downward trend in the V 4+ /V total ratio is determined through EPR measurements.As demonstrated in Table 2, there is a notable decrease in the fraction of V 4+ ions with increasing Nb 2 O 5 content.This observed downward trend aligns well with the existing literature [18,64,65] and can be influenced by synthesis parameters and the overall optical basicity of the glass.Given that increased temperatures and prolonged melting times result in a higher proportion of vanadium in the lower oxidation state, all melts in this study were uniformly held at the highest temperature for the same duration during synthesis.In addition to the synthesis parameters, the presence of V 4+ ions is induced by the acidic environment created through a high P 2 O 5 content, promoting the reduction of V 5+ ions.On the other hand, the introduction of Nb 2 O 5 , known for its intrinsic basic nature and high electronic polarizability, results in a significant increase in basicity [34,61].This, in turn, stabilizes a higher oxidation state of vanadium ions (V 5+ ) [66], consequently leading to a notable decrease in the relative amount of the V 4+ ions.See Table 2. Considering that the small polaron hopping mechanism is heavily dependent on the fraction of TM ions in a lower oxidation state [3,67], the decreasing trends observed in the fraction of V 4+ (i.e., a decrease in the number of V 5+ -V 4+ pairs) primarily contributes to the decreasing trend in σ DC .This serves as a significant indication that vanadium plays a key role in achieving the highest σ DC values for 25V-0Nb and 25V-5Nb, with the subsequent decrease in conductivity stemming from the diminishing contribution of polaronic conductivity originating from the vanadium species in this mixed-conductive system.Specifically, the mixed-conduction mechanism in both the 25V-0Nb and 25V-5Nb glasses is evident in the markedly lower E DC values of ~0.5 eV (see Table 2).These values are close to those of pure polaronic V-P glasses [68,69], indicating a significant polaronic contribution.Furthermore, utilising the method of internal friction, Barczy ński and Murawski investigated the relaxation processes involving the migration of Na + ions and the hopping of polarons between V 4+ and V 5+ ions in a Na-V-P glass system [17].They established that the glass-sharing composition similar to that of the 25V-0Nb glass in this study displays one large, mixed electronicionic peak.This feature indicates a substantial presence of both mobile ions and polarons in the glass, reinforcing the observation that these glasses exhibit mixed conductivity.Nevertheless, with the gradual addition of Nb 2 O 5 , a progressive rise in E DC is noted, reaching a steady value of ~0.7 eV (see Table 2), suggesting an emerging prominence of the ionic conductivity mechanism.This phenomenon is also evident in the previously discussed complex impedance spectra (see Figure 5b) where the 25V-0Nb and 25V-5Nb glasses (Region I), displaying mixed conductivity with a notable polaronic contribution, exhibit no noticeable effect of electrode polarization even at the elevated temperature of 513 K.In contrast, for samples where x ≥ 10 (Regions II and III), the "spur", typical of ionically conductive materials, begins to emerge, suggesting an increasing influence of the ionic conductivity mechanism.All the aforementioned suggests that the addition of Nb 2 O 5 triggers a transition in the conduction mechanism from mixed conductivity in Region I to predominantly ionic conductivity in Regions II and III.Furthermore, another essential factor influencing the observed decrease in polaronic contribution to the overall conductivity is the modification of the glass structure caused by the introduction of Nb 2 O 5 .Results from Raman and IR spectroscopy (see Figure 4) indeed indicate that within the 25V series, the introduction of Nb 2 O 5 disrupts the structure of the initial, predominantly mixed V-P network, leading to its unravelling and the breakdown of the V-O-V bonds connecting the vanadate units.Considering that a low degree of crosslinking between vanadate structural units is recognized to hinder electron hopping along V 4+ -O-V 5+ bonds [18][19][20], the introduction of Nb 2 O 5 , leading to the rupture of V-O-V bonds and an increased average distance between the vanadium ions, subsequently disrupts the pathways for electron hopping.
In the following section, the Summerfield scaling procedure is applied to conductivity spectra across a broad spectrum of frequencies and temperatures.This aims to explore the dynamics of charge carriers and acquire a more profound understanding of the electrical transport mechanism.

Scaling Properties of Conductivity Spectra
The Summerfield scaling method is one of the most straightforward and widely adopted model-free scaling techniques, relying on two scaling parameters: the DC conductivity, σ DC , and the temperature, T [70].It is mathematically represented by the equation (σ ′ (ν, T)/σ DC (T)) = F(ν/σ DC (T)T), where σ ′ denotes the real component of conductivity and the other physical quantities maintain their conventional meanings.This method, recognized as mobility scaling, attains validity by affirming that temperature impacts charge carrier dynamics without modifying the conduction mechanism.Thus, the successful superposition of individual conductivity isotherms scaled by the common factor σ DC T in a double log-log plot of σT vs. ν indicates that the Summerfield scaling procedure is effective, signifying that the conduction mechanism remains unchanged with varying temperatures.On the contrary, the deviation from Summerfield scaling, as observed in mixed ion-polaron glasses with significant amounts of both ion and polaron charge carriers, is ascribed to differently thermally activated mobilities of ions and polarons [26].
The 25V-0Nb and 25V-5Nb glasses precisely exhibit this behaviour, with their conductivity spectra failing to overlap, as depicted in Figure 7a.Conversely, the described approach successfully generates perfect conductivity master curves for all glasses with x ≥ 10, as illustrated in Figure 7b.This result verifies the validity of time-temperature superposition (TTS), affirming the stability of the conduction mechanism regardless of temperature variations.The inability to scale the conductivity spectra of glasses with x < 10 aligns well with the previously mentioned findings, collectively suggesting that these glasses demonstrate mixed conductivity.On the other hand, the successful construction of master curves for glasses where x ≥ 10 additionally confirms that the incorporation of Nb 2 O 5 leads to the prevalence of the ionic conduction mechanism in these glasses.
An additional validation of the Summerfield scaling results can be attained through the following approach.Specifically, this scaling method is considered to be satisfied if the slope of the line log(σ'T) vs. logν 0 , where ν 0 represents the onset frequency of conductivity dispersion defined by the equation σ'(ν 0 ) = 2σ DC , is equal to 1. Indeed, all glasses for which master curves are successfully constructed adhere to this criterion, as illustrated in Figure S2 in the Supplementary Materials.On the other hand, for the glass samples 25V-0Nb and 25V-5Nb, the observed deviation from a slope of 1 aligns with expectations, given that Summerfield scaling fails to produce master curves.
In the next step, individual conductivity master curves are superimposed to investigate the influence of glass composition and structure on conductivity dispersion.It should be noted that two glass-ceramic samples, 25V-30Nb and 25V-35Nb, are excluded from the analysis due to partial crystallization.Additionally, two glasses-specifically, 25V-0Nb and 25V-5Nb-are omitted from the analysis as well, because they do not comply with the scaling criteria of the Summerfield procedure.As depicted in Figure 7c, the attempt to construct a super master curve for glasses with 10 ≤ x ≤ 25 from the 25V series proves unsuccessful.Instead, the master curve for the 25V-10Nb glass displays a distinct shape compared with the other three glasses, all of which share the same shape.While the master curves of the 25V-15Nb and 25V-20Nb glasses perfectly overlap, the master curve for the 25V-25Nb glass exhibits a slight shift toward lower values of log(σ DC T).Nevertheless, it aligns well with the other two master curves when shifted along the scaled frequency axis; see the inset of Figure 7c.Here, it is worth noting the correlation between the scaling outcomes for the investigated glasses and their σ DC and E DC values.Specifically, the shape of the master curves is observed to change concurrently with the changes in σ DC , mirroring phenomena observed in our previous study on another series within the Na-V-P-Nb system [27].Furthermore, these alterations in the master curves coincide with the transition of the conductivity mechanism from mixed conductive in glasses with x < 10 to predominantly ionic in glasses with x ≥ 10.In the case of the initial two glasses from the 25V series, 25V-0Nb and 25V-5Nb, this is evidenced by the inability to obtain master curves because of their mixed ion-polaron conductivity.Furthermore, for the 25V-10Nb glass, which exhibits the largest drop in σ DC compared with the initial glass and marks the composition of significant structural modifications induced by Nb 2 O 5 introduction, the master curve assumes a distinctive shape differing from the other successfully constructed master curves.On the other hand, the 25V-15Nb, 25V-20Nb, and 25V-25Nb glasses, exhibiting successful super master construction, also show similar σ DC and E DC (see Table 2), indicating that they possess similar local structural environments for Na + transport.Based on all of the aforementioned observations, the following conclusions can be drawn: (i) the extent of the mixed ion-polaron conduction mechanism is strongest for 25V-0Nb and 25V-5Nb glasses, indicated by the inability to construct their master curves; (ii) the successful construction of a master curve for 25V-10Nb implies a diminishing contribution of polaronic conductivity and an increasing influence of ionic conductivity as the predominant mechanism of electrical conductivity; and (iii) the 25V-15Nb, 25V-20Nb, and 25V-25Nb glasses exhibit predominantly ionic conduction mechanisms and possess similar local structural environments for Na + transport, as evidenced by the same-shaped master curves.An additional validation of the Summerfield scaling results can be attained through the following approach.Specifically, this scaling method is considered to be satisfied if the slope of the line log(σ'T) vs. logν0, where ν0 represents the onset frequency of conductivity dispersion defined by the equation σ'(ν0) = 2σDC, is equal to 1. Indeed, all glasses for which master curves are successfully constructed adhere to this criterion, as illustrated in

Influence of V 2 O 5 Content on the Mechanism of Electrical Conductivity
As consistently detailed throughout the text, the 25V series distinguishes itself with unique behaviour, differentiating it from other alkali-niobate-phosphate (A-Nb-P) systems where P 2 O 5 is gradually replaced with Nb 2 O 5 [27][28][29][30][31][32][33][34].This distinct effect is evident not only in its general and thermal properties but also in its electrical characteristics.Specifically, V M , T g , and σ DC exhibit non-monotonic trends resulting from structural changes induced by the addition of Nb 2 O 5 .While T g and σ DC show a decreasing trend, reaching their minimum values in the 25V-20Nb glass, V M follows a positive trend, reaching its peak with the same glass composition.It is intriguing because the addition of Nb 2 O 5 typically has a beneficial impact on the thermal properties of alkali-phosphate-based systems manifested by an increase in T g and a reduction in V M , indicative of the formation of a more compact mixed Nb-P network [27,[29][30][31][32][33][34].Additionally, the incorporation of Nb 2 O 5 is usually recognized for its ability to enhance electrical conductivity in alkali-phosphate-based systems by forming mixed P-O-Nb bonds, thereby facilitating the transport of alkali ions through the glass network [28].However, the 25V series deviates from this behaviour, showing an unfavourable impact of Nb 2 O 5 on both thermal and electrical properties.What sets the 25V series apart from other A-Nb-P systems is the notable presence of a relatively high quantity of V 2 O 5 (25 mol%), which may directly account for the distinctive trends observed in this series.
To shed light on the influence of V 2 O 5 content on the observed trends arising from the gradual substitution of P 2 O 5 with Nb 2 O 5 in this series, a valuable comparison can be made between the 25V series and the outcomes reported in the previously published 35Na 2 O-10V 2 O 5 -(55 − x)P 2 O 5 − xNb 2 O 5 (10V) series [27].Both of these systems consist of the same oxides and maintain constant proportions of Na 2 O (35 mol%) and V 2 O 5 (10/25 mol%), with P 2 O 5 being replaced by Nb 2 O 5 .Hence, changing the concentration of V 2 O 5 content allows for the assessment of the extent of its influence on the properties under examination.A comprehensive exploration of the general, thermal, structural, and electrical properties of the 10V series in Ref. [27] reveals that this series exhibits trends consistent with the majority of the A-Nb-P systems documented in the literature [27][28][29][30][31][32][33][34].In this series, V 2 O 5 is recognized to play dual roles within the structure, acting as both a glass modifier and a network former; nevertheless, owing to its relatively small quantity, it exerts no substantial influence on the overall trends, unlike in the 25V series.Indeed, the latter series exhibits behaviour that is entirely contrasting to that of the 10V series, as illustrated in the following.
The addition of Nb 2 O 5 has different effects on the change in V M in two series, attributable to structural modifications induced by its incorporation.In series 10V, the addition of Nb 2 O 5 leads to a more compact structure, reflected in the downward trend of V M [27].This is further substantiated by the vibrational spectroscopy results, unveiling a structural evolution from predominantly phosphate (Region I) through predominantly mixed Nb-P (Region II) to a predominantly niobate network (Region III).Conversely, in the 25V series, the incorporation of Nb 2 O 5 has an opposing effect, disrupting the initially compact mixed V-P network, which leads to a more "open" network, as evidenced by the increasing V M values.Interestingly, the Raman/IR-ATR spectroscopy finding reveals some similarities between the 10V and 25V series; nevertheless, notable distinctions exist, with the most significant being that in the 25V series, Region I corresponds to a predominantly mixed V-P network.As the Nb 2 O 5 content increases, the signals describing NbO 6 units connected into chains via Nb-O-Nb bridging bonds progressively dominate (Region II).At the highest Nb 2 O 5 content, these units tend to cluster, forming a 3D network in Region III.Due to the lower initial amount of P 2 O 5 (a consequence of a higher V 2 O 5 content), the GFR in the 25V series is narrower compared with that in the 10V series.For the same reason, introducing Nb 2 O 5 into the initially mixed V-P network in the 25V series triggers the formation of 3D clusters at a lower Nb 2 O 5 concentration (25 mol%), in contrast to the 40 mol% Nb 2 O 5 concentration required for niobate clustering to dominate in the 10V series.
The electrical transport mechanism is also found to be markedly affected by the V 2 O 5 content, and comparing the two series enables assessing the extent of its polaronic contribution.Specifically, an in-depth examination of the electrical properties in the 10V series reveals that V 2 O 5 , at a concentration of 10 mol%, does not actively participate in the overall conduction process through polaronic transport.The mechanism of electrical conductivity in the 10V series is identified as purely ionic, and the observed non-monotonic trends in σ DC and E DC are ascribed to the facilitating effect of Nb 2 O 5 on the transport of Na + ions, resulting in the positive mixed glass former effect [27].See Figure 8.On the contrary, the 25V series displays an opposite trend, wherein the substitution of P 2 O 5 by Nb 2 O 5 decreases σ DC (see Figure 8).This observation is intricately tied to the substantial amounts of two types of charge carriers, ionic (35 mol% Na 2 O) and, especially, polaronic (25 mol% V 2 O 5 ), leading to mixed conductivity in the 25V-0Nb and 25V-5Nb glasses.With the addition of more Nb 2 O 5 , the polaronic contribution diminishes as the number of V 5+ -V 4+ pairs decreases and the mixed V-P network, crucial for enabling favourable electron transfer pathways, undergoes disruption and transitions to a mixed Nb-V-P network.An effective approach to visualize the complexities of mixed conductive glass systems like the 25V series involves employing the Meyer-Neldel (M-N) formalism [71].According to this rule, the pre-exponential factor σ0* can be linked to EDC, using the equation log σ0* = a EDC + b, where a and b are constants.A negative slope indicates electronic conductivity, while a positive slope suggests ionic conductivity.This allows for distinguishing various conduction mechanisms, as seen in studies across different materials, including glasses with ionic, electronic, or mixed electronic-ionic conduction [24].When examining the pre-exponential factor, σ0*, in relation to composition (see Figure 9a) and EDC (see An effective approach to visualize the complexities of mixed conductive glass systems like the 25V series involves employing the Meyer-Neldel (M-N) formalism [71].According to this rule, the pre-exponential factor σ 0 * can be linked to E DC , using the equation log σ 0 * = a E DC + b, where a and b are constants.A negative slope indicates electronic conductivity, while a positive slope suggests ionic conductivity.This allows for distinguishing various conduction mechanisms, as seen in studies across different materials, including glasses with ionic, electronic, or mixed electronic-ionic conduction [24].When examining the pre-exponential factor, σ 0 *, in relation to composition (see Figure 9a) and E DC (see Figure 9b), a positive relationship is observed for samples with 10 ≤ x ≤ 35, indicating predominant ionic electrical transport.However, glasses with x < 10, displaying a noteworthy contribution of polaronic conductivity, deviate from this trend and exhibit lower values for both σ 0 * and E DC .A comparison with different glass series, including the 10V series-showcasing ionic [22,27], electronic [23], or mixed electronic-ionic conduction [24][25][26]-reveals that the glasses-(ceramics) with 10 ≤ x ≤ 35 fall within Area I, signifying glasses with dominant ionic conductivity (see Figure 9c).On the other hand, 25V-0Nb glass and 25V-5Nb are positioned in Area II, corresponding to glasses with a dominant polaronic conduction mechanism.

Materials and Methods
Glass samples within the quaternary system 35Na2O-25V2O5-(40 − x)P2O5-xNb2O5 (25V), x = 0-35, mol%, are synthesized using the conventional melt quenching technique in 3 g batches.The precursors, Na2CO3, NH4H2PO4, Nb2O5, and V2O5, are precisely weighed and mixed in the desired stoichiometric proportions.Homogenization is achieved through 15 min of grinding in an agate mortar.The prepared reaction mixtures undergo calcination at 700 °C, followed by melting in a platinum crucible under ambient air within the temperature range of 1100 to 1200 °C.After a 40 min hold time at a specific temperature, black and opaque products are obtained by pouring them into a stainlesssteel mould at RT.Given a weight loss of less than 1.5%, the composition of the glass batch is considered an accurate representation of the actual glass composition.The samples are designated according to the mol% of V2O5 and Nb2O5 in the batch.For example, the glass labelled as 25V-20Nb comprises 25 mol% V2O5 and 20 mol% Nb2O5, as outlined in Table 1.
The determination of sample density, ρ, for bulk samples is conducted at RT using Archimedes' method, with 96% ethanol employed as the immersion liquid.The molar

Materials and Methods
Glass samples within the quaternary system 35Na  1.
The determination of sample density, ρ, for bulk samples is conducted at RT using Archimedes' method, with 96% ethanol employed as the immersion liquid.The molar volume, V M , is calculated as V M = M/ρ, where M represents the average molar weight of the glass.
The thermal characteristics of the prepared samples are investigated using differential thermal analysis (DTA) utilising the Mettler TGA/DSC 3+ thermobalance.Powder samples (~30 mg) are placed in a Pt crucible in an oxygen atmosphere, with a heating rate of 20 • C min −1 over the temperature range of 25-1000 • C. The results are analysed using the Mettler STARe 9.01 software, and the glass transition temperature (T g ) is determined for all the samples along with the first observable crystallization peak temperature (T c ).
Powder X-ray diffraction (PXRD) data are obtained using a Bruker D8 Discover diffractometer equipped with a LYNXEYE XE-T detector (Bruker AXS GmbH, Karlsruhe, Germany).Data are recorded in Bragg-Brentano geometry in the 2θ range of 10 • to 70 • using CuKα radiation (1.5418 Å).Rietveld analysis of the diffraction patterns is performed with the HighScore X'pert HighScore Plus 3.0 program (Malvern Panalytical, Almelo, The Netherlands).The amorphous structure of the glasses is confirmed, and the partially crystallized samples with high Nb 2 O 5 content (25V-30Nb and 25V-35Nb) are analysed both qualitatively and quantitatively.The quantitative analysis employs the internal standard method, using crystalline ZnO as an internal standard, with weight fractions of the amorphous and crystalline phases determined with Rietveld refinement [72].In a system with a known amount of standard S, the weight fraction of the crystalline phase P was determined according to the following equation (Equation ( 2)): where S is the scale factor, Z is the number of the formula unit within the unit cell, M is the formula unit mass, and V is the volume of the unit cell.The weight fraction of the amorphous phase was determined by the following expression (Equation (3)): where W S , W P , and W A are the weight percentages of the standard, crystalline, and amorphous phases, respectively.Raman spectra are measured from bulk samples at RT across the spectral range of 1500-60 cm −1 using a Thermo Scientific DXR Raman spectrometer with a 532 nm solid-state (Nd: YAG) diode-pumped laser.The acquired Raman spectra exhibit a complex shape, analysed using a least-square fitting procedure, assuming a Gaussian shape for all bands.From the deconvoluted Raman spectra, the position and intensity of individual components are determined.The Q n notation, denoting the number of bridging oxygen atoms per PO 4 tetrahedron (n = 0-3), is used to represent phosphate units.Additionally, attenuated total reflectance infrared (IR-ATR) spectra of powder samples are recorded using a Perkin Elmer Spectrum Two FT-IR Spectrometer equipped with a diamond universal attenuated total reflectance (UATR) accessory within a spectral range of 4000-400 cm −1 .
Electron paramagnetic resonance (EPR) spectra of bulk samples are acquired at RT utilising a cw-ESR EMXmicro spectrometer (Bruker), functioning in the X band (~9.83 GHz) and equipped with Xenon software.Sample weights are determined to the nearest hundredth of a milligram and the obtained EPR spectra are double-integrated, while experimental parameters are employed to normalize the areas of all spectra.Utilising Mn 2+ as a standard, the spin concentration is determined as spin/g for all samples, and the V 4+ /V total ratio is calculated from the obtained data.
The Axia™ ChemiSEM™ Scanning Electron Microscope (Thermo Fisher Scientific, Waltham, MA, USA) with an energy-dispersive X-ray spectroscopy (EDS) system is employed to analyse the microstructure and elemental composition of the prepared samples.Solid-state impedance spectroscopy (SS-IS) is utilised for electrical measurements.The annealed samples are crafted into ~1 mm thick disks with gold electrodes (5.4 mm in diameter) sputtered on both sides using a Sputter Coater SC7620.Electrical properties are determined by measuring complex impedance with a Novocontrol Alpha-AN Dielectric Spectrometer (Novocontrol Technologies GmbH & Co., KG, Hundsangen, Germany) across a wide range of frequencies (0.01 Hz-1 MHz) and temperatures (183 K to 513 K), maintaining a temperature control accuracy of ±0.2 K.The obtained impedance spectra are analysed through modelling with the appropriate electrical equivalent circuit (EEC) using the complex non-linear least-squares (CNLLSQ) method and WinFIT software (version 3.2, Novocontrol Technologies GmbH & Co., KG, Hundsangen, Germany).Impedance spectra displaying a depressed semicircle are fitted using the EEC model, which includes a parallel combination of a resistor (R) and a constant phase element (CPE).In cases where impedance spectra exhibit an additional low-frequency "spur" attributed to the electrode polarization effect, the corresponding EEC model integrates a parallel R-CPE combination connected in series with another CPE.

Conclusions
This study investigates the influence of introducing two transition metal oxides, V 2 O 5 and Nb 2 O 5 , on the thermal, (micro)structural, and electrical properties of a mixed conductive glass and glass-ceramic system with the nominal composition 35Na 2 O-25V 2 O 5 -(40 − x)P 2 O 5 − xNb 2 O 5 (x = 0-35, mol%).Solid-state impedance spectroscopy measurements and differential thermal analysis indicate that the initial glass is the optimal sample, exhibiting the highest values for both direct current (DC) conductivity and glass transition temperature.Our findings highlight that the gradual substitution of P 2 O 5 with Nb 2 O 5 impacts both DC conductivity and glass transition temperature, resulting in non-linear trends, with the lowest values observed for the glass containing 20 mol% Nb 2 O 5 .An extensive analysis of the electrical properties, coupled with electron paramagnetic resonance measurements and vibrational spectroscopy results, indicates that the observed non-monotonic trend in DC conductivity arises from the transition of the electrical conductivity mechanism from mixed conductive to predominantly ionic.This is associated with a decrease in the contribution of polaronic conductivity due to two factors: the reduction in the relative number of V 4+ ions and the structural modifications of the glass network resulting from the introduction of Nb 2 O 5 .While the former results in a reduction in the V 5+ -V 4+ pairs available for participation in the small polaron hopping mechanism, the latter disrupts the predominantly mixed vanadate-phosphate network in the initial glasses, causing the breakage of the V-O-V bonds and an increased average distance between vanadium ions.This ultimately impedes electron hopping pathways and hinders polaronic transport.The mechanism of electrical transport is additionally examined through the model-free Summerfield scaling procedure, revealing that glasses characterized by a predominantly mixed vanadate-phosphate network (x < 10 mol%, Region I) display mixed conductivity.Meanwhile, glasses featuring a mixed Nb-V-P glass network (10 mol% ≤ x ≤ 20 mol%, Region II) and a predominantly mixed Nb-V network (x ≥ 25 mol%, Region III) reveal the dominance of the ionic conductivity mechanism.Furthermore, a comparative analysis of two analogue series with varying V 2 O 5 content (10 mol% and 25 mol%) is conducted to assess the impact of V 2 O 5 content on the electrical transport mechanism and to evaluate the extent of its polaronic contribution.We illustrate that at 10 mol%, V 2 O 5 does not actively partake in the overall conduction process through polaronic transport, given its low concentration.However, elevating the V 2 O 5 content to 25 mol% leads to a substantial enhancement in DC conductivity (a jump by more than two orders of magnitude), attributed to the significant contribution of V 2 O 5 to the total conductivity via polaron transport.The results of this study offer valuable insights into the mixed conductive glass and glass-ceramic Na-V-P-Nb-based systems.They showcase the capability to modify the mechanism of electrical conductivity through straightforward adjustments in composition, a crucial aspect in the

Figure 1 .
Figure 1.(a) Dependence of density, ρ, and molar volume, VM, on Nb2O5 content; (b) DTA curves; and (c) dependence of glass transition temperature, Tg, and the parameter (Tc − Tg) on Nb2O5 content.The lines connecting data points in (a,c) are a guide for the eye.

Figure 1 .
Figure 1.(a) Dependence of density, ρ, and molar volume, V M , on Nb 2 O 5 content; (b) DTA curves; and (c) dependence of glass transition temperature, T g , and the parameter (T c − T g ) on Nb 2 O 5 content.The lines connecting data points in (a,c) are a guide for the eye.
13 Nb 35 O 94 , with the remaining 71 wt.% being amorphous glass matrix; see Figure 2b.With further introduction of Nb 2 O 5 , the proportion of the crystalline phase Na 13 Nb 35 O 94 embedded into the glass matrix for the 25V-35Nb glass-ceramic sample increases to 45 wt.%.

Figure 2 .
Figure 2. (a) PXRD patterns of glasses and glass-ceramics from this study; (b) Rietveld refinement of 25V-30Nb and 25V-35Nb glass-ceramics mixed with powder of ZnO used as an internal standard for amorphous phase quantification.Experimental data are given by the blue line, the calculated pattern is shown in black, teal vertical marks show the positions of diffraction lines belonging to the dominant Na13Nb35O94 phase, while the positions of ZnO lines are given as maroon vertical marks.

Figure 2 .
Figure 2. (a) PXRD patterns of glasses and glass-ceramics from this study; (b) Rietveld refinement of 25V-30Nb and 25V-35Nb glass-ceramics mixed with powder of ZnO used as an internal standard for amorphous phase quantification.Experimental data are given by the blue line, the calculated pattern is shown in black, teal vertical marks show the positions of diffraction lines belonging to the dominant Na 13 Nb 35 O 94 phase, while the positions of ZnO lines are given as maroon vertical marks.

Figure 3 .
Figure 3. (a) SEM micrographs of 25V-30Nb glass-ceramic; (b) count X-ray mapping; (c) quantit X-ray mapping; (d) count and (e) quantitative elemental mapping of O, Na, V, Nb, and P fro selected area of the given glass-ceramic.Furthermore, the SEM micrograph of the 25V-35Nb glass-ceramic shown in Fig S1inthe Supplementary Materials (SM) reveals that with the addition of more Nb2O5 morphology further develops with the ongoing crystallization process.Fused grains on the form of plate-shaped grains, varying in size from approximately 2 to 20 µm in ameter.These grains are also uniformly embedded in the remaining amorphous gl matrix.The EDS maps depicted in Figures3b,cand S1b,c in the SM visually illustrate distribution and variation of the chemical elements across the two glass-ceramic samp Examination of the maps for the individual elements O, Na, V, Nb, and P reveals dist compositional differences between the grains and the residual glass matrix area, as sho in Figure3d,e and FigureS1d,e in the SM.In the grain areas, the elements Na, Nb, an are concentrated, while V and P are present only in trace amounts on the surface.observation aligns with the EDS analysis (see FigureS1fin the SM), which indicates the composition of the observed grains closely matches that of the Na13Nb35O94 crysta phase.As the identified crystalline phase progressively crystallizes with the increa Nb2O5 content, the amorphous glass matrix is depleted of Na, Nb, and O elements, w the elements V and P remain in the glass matrix.This is particularly evident in the 2 35Nb sample (see FigureS1cin the SM), where the EDS analysis reveals that the com sition of the grains (area 2) corresponds to that of the Na13Nb35O94 crystalline phase, w the glassy matrix (area 3) becomes enriched in V, as illustrated in FigureS1gin the SM

Figure 3 .
Figure 3. (a) SEM micrographs of 25V-30Nb glass-ceramic; (b) count X-ray mapping; (c) quantitative X-ray mapping; (d) count and (e) quantitative elemental mapping of O, Na, V, Nb, and P from a selected area of the given glass-ceramic.

Figure 4 .
Figure 4. (a) Raman spectra of glasses and glass-ceramics from the 25V series and (b-g) comparison of IR-ATR and Raman spectra for individual glass samples.Shading in (b-g) corresponds to the band assignment in (a).The orange shade denotes the vibration of phosphate units, the green shade represents vanadate units, and the blue shade indicates niobate units.The most intense colored regions correspond to the dominant signals.

Figure 4 .
Figure 4. (a) Raman spectra of glasses and glass-ceramics from the 25V series and (b-g) comparison of IR-ATR and Raman spectra for individual glass samples.Shading in (b-g) corresponds to the band assignment in (a).The orange shade denotes the vibration of phosphate units, the green shade represents vanadate units, and the blue shade indicates niobate units.The most intense colored regions correspond to the dominant signals.

Figure 5 .
Figure 5. (a) Conductivity spectra at different temperatures for 25V-20Nb glass and (b) the complex impedance spectra for all glass samples at 483 K. Empty symbols represent the experimental data, while the lines depict the fit obtained through EEC modelling.

Figure 5 .
Figure 5. (a) Conductivity spectra at different temperatures for 25V-20Nb glass and (b) the complex impedance spectra for all glass samples at 483 K. Empty symbols represent the experimental data, while the lines depict the fit obtained through EEC modelling.

Figure 6a illustrates the
Figure 6a illustrates the temperature-dependent variation in σDC for different Nb2O5 content, demonstrating that the DC conductivity is temperature-activated and exhibiting Arrhenius temperature dependence with characteristic activation energy, as shown in Figure 6b.The activation energy for DC conductivity, EDC, is calculated from the slope of the log(σDCT) vs. 1000/T, utilising the Arrhenius equation (Equation (1)), σDCT = σ0*exp(-EDC/kBT), (1)

Figure 6 .
Figure 6.(a) Compositional and temperature dependence of DC conductivity, σDC; (b) Arrhenius plot of DC conductivity; and (c) compositional dependence of number density, NV, of Na + ions, V 4+ ions, and the total number density of charge carriers.Empty symbols represent glasses, whereas full symbols denote glass-ceramics.The lines connecting data points in (a,c) are a guide for the eye, while the lines in (b) are obtained through linear regression.

Figure 6 .
Figure 6.(a) Compositional and temperature dependence of DC conductivity, σ DC ; (b) Arrhenius plot of DC conductivity; and (c) compositional dependence of number density, N V , of Na + ions, V 4+ ions, and the total number density of charge carriers.Empty symbols represent glasses, whereas full symbols denote glass-ceramics.The lines connecting data points in (a,c) are a guide for the eye, while the lines in (b) are obtained through linear regression.

23 Figure 7 .
Figure 7. Summerfield scaling of conductivity spectra of (a) 25V-5Nb glass and (b) 25V-20Nb glass.(c) Construction of super master curve of the conductivity isotherms obtained by applying the Summerfield scaling procedure to all the investigated glasses; inset: individual master curves of glasses with 10 ≤ x ≤ 25 shifted along the x-axis to overlap with the reference master curve of 25V-10Nb glass.

Figure 7 .
Figure 7. Summerfield scaling of conductivity spectra of (a) 25V-5Nb glass and (b) 25V-20Nb glass.(c) Construction of super master curve of the conductivity isotherms obtained by applying the Summerfield scaling procedure to all the investigated glasses; inset: individual master curves of glasses with 10 ≤ x ≤ 25 shifted along the x-axis to overlap with the reference master curve of 25V-10Nb glass.

23 Figure 9 .
Figure 9.The dependence of the pre-exponential factor (σ0*) as a function of (a) composition and (b) activation energy (EDC); and (c) a comparison of the glasses and glass-ceramics from this study with those from 10V series [27] and systems Na-W-P [22], Na-Mo-Zn-P [25], Li-W-Zn-P [25] and Na-V-P [23] from the literature.The line in (b) is obtained through linear regression.The dotted circles in Figure 9c highlight the samples from this study.

Figure 9 .
Figure 9.The dependence of the pre-exponential factor (σ 0 *) as a function of (a) composition and (b) activation energy (E DC ); and (c) a comparison of the glasses and glass-ceramics from this study with those from 10V series [27] and systems Na-W-P [22], Na-Mo-Zn-P [25], Li-W-Zn-P [25] and Na-V-P [23] from the literature.The line in (b) is obtained through linear regression.The dotted circles in Figure 9c highlight the samples from this study.
2 O-25V 2 O 5 -(40 − x)P 2 O 5 − xNb 2 O 5 (25V), x = 0-35, mol%, are synthesized using the conventional melt quenching technique in 3 g batches.The precursors, Na 2 CO 3 , NH 4 H 2 PO 4 , Nb 2 O 5 , and V 2 O 5 , are precisely weighed and mixed in the desired stoichiometric proportions.Homogenization is achieved through 15 min of grinding in an agate mortar.The prepared reaction mixtures undergo calcination at 700 • C, followed by melting in a platinum crucible under ambient air within the temperature range of 1100 to 1200 • C.After a 40 min hold time at a specific temperature, black and opaque products are obtained by pouring them into a stainless-steel mould at RT.Given a weight loss of less than 1.5%, the composition of the glass batch is considered an accurate representation of the actual glass composition.The samples are designated according to the mol% of V 2 O 5 and Nb 2 O 5 in the batch.For example, the glass labelled as 25V-20Nb comprises 25 mol% V 2 O 5 and 20 mol% Nb 2 O 5 , as outlined in Table