Bing
Wu
*a,
Vlastimil
Mazánek
a,
Min
Li
b,
Martin
Veselý
c,
Qiliang
Wei
d,
Luxa
Jan
a,
Filipa M.
Oliveira
a,
Lei
Zheng
a,
Heng
Li
a,
Vojtech
Kundrat
e,
Jakub
Zálešák
f,
Jakub
Regner
a,
Rui
Gusmão
a,
Junjie
He
b,
Tomáš
Hartman
a,
Saeed
Ashtiani
a,
Yulong
Ying
g and
Zdenek
Sofer
*a
aDepartment of Inorganic Chemistry, University of Chemistry and Technology Prague, Technická 5, 166 28 Prague, Czech Republic. E-mail: wui@vscht.cz; zdenek.sofer@vscht.cz
bDepartment of Physical and Macromolecular Chemistry, Faculty of Science, Charles University in Prague, Prague 12843, Czech Republic
cDepartment of Organic Technology, University of Chemistry and Technology Prague, Technická 5, 166 28 Prague, Czech Republic
dInstitute of Micro/Nano Materials and Devices, Ningbo University of Technology, Ningbo, 315211, P.R. China
eDepartment of Molecular Chemistry and Materials Science, Weizmann Institute of Science, Rehovot 7610001, Israel
fChemistry and Physics of Materials, University of Salzburg, Jakob-Haringer-Strasse 2A, 5020 Salzburg, Austria
gSchool of Materials Science and Engineering, Zhejiang Sci-Tech University, Hangzhou 310018, PR China
First published on 30th January 2025
Two-dimensional (2D) layered thiophosphates have garnered attention for advanced battery technology due to their open ionic diffusion channels, high capacity, and unique catalytic properties. However, their potential in energy storage applications remains largely unexplored. In this study, we report a 2D transition metal thiophosphate (Nb4P2S21) with high sulfur content, synthesized via chemical vapor transport (CVT). The bulk material, exhibiting a layered quasi-one-dimensional (quasi-1D) structure, can be exfoliated into high-quality nanoplates using glue-assisted grinding. Density functional theory (DFT) calculations reveal a direct bandgap of 1.64 eV (HSE06 method) for Nb4P2S21, aligning with its near-infrared (NIR) photoluminescence at 755 nm. Despite an initial discharge capacity of 1500 mA h g−1, the material shows low reversible capacity and rapid capacity decay at 0–2.6 V. In situ Raman confirms the formation of polysulfides during cycling. Given its high sulfur content, the material was evaluated at 0.5–2.6 V, 1.0–2.6 V, and 1.5–2.6 V to assess its sulfur-equivalent cathode performance. In carbonate-based electrolytes, electrochemical performance is hindered by polysulfide formation and side reactions, but switching to ether-based electrolytes improves initial reversible capacity and coulombic efficiency due to additional LixS conversion above 2.2 V. EDS and TOF-SIMS analyses of cycled electrodes show a significant sulfur loss, worsening the polysulfide shuttle effect and leading to battery failure. Adapting strategies from lithium–sulfur batteries, such as polar host catalysts, could enhance the material's performance.
Two-dimensional (2D) materials have emerged as leading candidates to address these challenges, with research over the past decade demonstrating their potential to revolutionize energy storage.3 These materials, defined by their atomically thin structure and high surface-to-volume ratio, have shown exceptional promise due to their unique physicochemical properties. Moreover, their layered structure facilitates van der Waals interactions that allow for efficient ion intercalation and deintercalation, a crucial mechanism for battery operation. Theoretical and experimental studies have highlighted their ability to facilitate fast electron and ion transport, and offer novel mechanisms for charge storage.4 These materials, including graphene, transition metal dichalcogenides, and hexagonal boron nitride, each contribute to a deeper understanding of 2D physics and chemistry while pushing the boundaries of energy storage capabilities.5
Within the realm of 2D materials, layered thiophosphates are particularly noteworthy. These materials are part of a broader family of transition metal phosphorus trichalcogenides MPCh3 (M = transition metal, Ch = S, Se), which have been extensively studied for their electronic and magnetic properties.6 Recent advances highlight the burgeoning potential of these MPCh3 materials in the domain of energy storage.7 The relationships between the structural characteristics of these thiophosphates and their electrochemical performance underscore the importance of the van der Waals gap and electronic conductivity in dictating the rate capabilities and cycling stability of the resulting batteries. They are recognized for their utility as (1) electrode materials for alkali metal-ion batteries, exemplified by FePS3, FePSe3, SnP2S6 and high-entropy (CoVMnFeZn)PS3 in lithium/sodium/potassium-ion systems;7d,8 (2) catalysts facilitating swift conversion of metal (poly)sulfides in metal–sulfur batteries, with bimetallic (FeMn)PS3 and (FeCo)PS3 being a case in point;7c,9 and (3) solid-state electrolytes in multivalent metal batteries, as seen with ZnPS3 as zinc-ion solid electrolytes.7b,10 It is worth noting that recently the layered van der Waals material V2PS10 has gained attention due to its unique properties in lithium and magnesium-ion batteries.11 However, other 2D thiophosphate compounds with varying phosphorus and sulfur content, such as Nb4P2S21, may offer distinct advantages for battery applications similar to V2PS10, yet they remain unreported. The sulfur-rich composition of these materials provides a high theoretical capacity as a result of the multiple electron-exchange reactions that sulfur atoms can undergo. Additionally, materials with rich sulfur structures, such as MoSx (3 ≤ x ≤ 7),12 NbSx (3 ≤ x ≤ 5),13 (Fe/Co/Ni)2S7,14 Co2S9,14 Fe3S8,14 WS5,15 and the P4S10+n series,16 often exhibit advantages similar to those in lithium–sulfur batteries as sulfur-equivalent electrode materials. These include a high voltage plateau, faster electron conduction in active materials, and more rapid reaction kinetics between alkali metal and active materials. Additionally, such materials, when subjected to high cutoff voltages in LIBs, may not fully discharge, leading to the formation of metallic compounds (such as non-stoichiometric metal sulfides) other than Li2S. These compounds can catalyze the subsequent conversion of Li2S, functioning similarly to catalysts in lithium–sulfur battery systems. Consequently, they are considered promising high-capacity sulfur-equivalent cathode materials for metal–sulfur battery systems.17
Building on this foundation, our research introduces the niobium thiophosphate Nb4P2S21, a sulfur-rich 2D material synthesized through an innovative chemical vapor transport (CVT) process. By employing density functional theory (DFT) calculations, we reveal the direct band gap feature of this material with 1.64 eV (based on the HSE06 method), as well as exhibiting near-infrared (NIR) photoluminescence at 755 nm. The synthesis and subsequent delamination of this material into high-quality nanoplates via glue-assisted grinding exfoliation represent a significant breakthrough. This method not only retains the structural integrity of the thiophosphate but also makes the benefits of its narrow 2D layered structure accessible for ion diffusion, leading to potentially superior rate performance over conventional materials. Prior research has underscored the challenge of high sulfur content in electrode materials, particularly concerning electrolyte interactions and the stability of the cathode during cycling. Our study addresses these challenges by presenting a detailed electrochemical analysis of Nb4P2S21, examining its performance across different voltage windows and comparing it with various electrolyte systems from carbonate-based to ether-based electrolytes. Post-cycling failure analysis of the batteries using X-ray diffraction (XRD), Scanning Electron Microscopy/Energy Dispersive X-ray spectroscopy (SEM-EDS), and TOF-SIMS identified the polysulfide shuttle effect, like that in lithium–sulfur batteries, which involves the migration of soluble lithium polysulfides formed during discharge between the anode and cathode, leading to capacity loss. The presented results not only shed light on the electrochemical performance but also guide further enhancements for sulfur-rich materials.
For comparison, the bulk powder was obtained through the same procedure, excluding the addition of CMC glue solution.
Electrochemical testing included discharge/charge performance analysis using a Neware battery test system (Neware BTS 8.0, Shenzhen, China). Additionally, cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were conducted at room temperature, employing a CorrTest CS Studio electrochemical workstation (Wuhan CorrTest Instrument Corp).
In Fig. 1c, the bulk crystals with a quasi-1D needle morphology are shown, which gives an idea about the highly crystallized Nb4P2S21 single crystal. The SEM image in Fig. 1d reveals the microstructure of the material, and the high magnification inset shows the crystal quality and layered structure, which are swift ion intercalation processes and guest intercalations. Fig. S3† depicts the EDS analysis results of an Nb4P2S21 crystal. The elemental maps for Nb, P, and S show a homogeneous distribution across the crystal. The EDS spectrum quantitatively supports the material composition, revealing a stoichiometry nearly matching the anticipated ratio of Nb:
P
:
S at 4
:
2
:
21. The Raman spectrum of the Nb4P2S21 single crystal is presented in Fig. 1e. The split bands at 531.1 cm−1 and 562.2 cm−1 are from asymmetric stretching of PS4 unit, while symmetric stretching vibration is centered at 413.1 cm−1 and swinging vibration is at 329.4 cm−1. The signals in the 150–300 cm−1 region can be attributed to the symmetric and asymmetric bending modes of PS4 tetrahedral overlaps with NbS8 polyhedron.24 The observation of a signal centred at 488.1 cm−1 is indicative of S3 bridge bonding within a unit layer.25Fig. 1f–h further provides XPS spectra that display the binding states of Nb 3d, P 2p, and S 2p, respectively. Based on the structural analysis of the material described earlier, Nb and P should exhibit single bonding states corresponding to NbS8 polyhedral and PS4 tetrahedral units, respectively. Nb3d spectrum exhibits a doublet at 203.7 and 206.5 eV that indicates a Nb4+ oxidation state.26 P2p spectrum shows a doublet at 131.2 and 132.0 eV which is typical for PS4 tetrahedra. On the other hand, sulfur exhibits a very complex chemical environment, with at least four different binding states expected: Nb2S, NbSP, PSS, and SSS. Additionally, there is a significant variety in the Nb–S and P–S bond lengths that can further broaden the S 2p peak. Such a complex nature causes the broadening of S 2p peak which makes it challenging to fit the spectrum correctly since individual states can have very small differences in binding energies (tenths of eV); therefore, the fitting of S 2p was not done. However, the peak is within 159–164 eV which is below the elemental sulfur, thus corresponding to sulfides.
Fig. 2b presents exfoliated Nb4P2S21 nanoplates in DMF, showing the dark orange-yellow color of the suspension. The right side under laser illumination exhibits the Tyndall effect, confirming a stable colloidal dispersion of the exfoliated material. Fig. 2c presents a STEM image of exfoliated Nb4P2S21, showing the thin, evenly distributed nanoplates, indicative of a successful exfoliation to a few-layer thickness. The corresponding elemental maps for Nb, P, S, and O in Fig. 2d across a nanoplate, confirm uniform composition and suggest slight surface oxidation, which is typical for such materials when exposed to air. After exfoliation, its EDS spectrum in Fig. S4† exhibits a close stoichiometric ratio of all elements to Nb4P2S21. In addition, no peak for Na Ka at 1.041 was observed, indicating that CMC has been completely removed from the material. Simultaneously, in our sample, we can observe specimens with a lateral size of approximately 300 nm, which exhibit a thickness of ∼1.7 nm in the AFM characterizations shown in Fig. S5.† This thickness indicates a single layer of the multi-atomic layered structure consistent with van der Waals crystals, as demonstrated in Fig. 1a. Furthermore, the elemental mapping presented in Fig. S6† confirms that these thin flake materials are Nb4P2S21. Fig. 2f and g further provides comparative SEM images displaying Nb4P2S21 samples without and with CMC glue during grinding, respectively. The results show Nb4P2S21 without CMC glue for exfoliation, where the material exhibits a bulk morphology with irregularly shaped particles and a broad size distribution. The particles have a more three-dimensional shape, with visible facets and edges, indicating a typical unprocessed bulk material. While the Nb4P2S21 exfoliated with CMC glue, shows the morphology of thinner, more uniform nanoplates, displaying a needle-like or flake-like structure. These structures are characteristic of the delamination effect induced by the exfoliation process, where the CMC glue plays a crucial role. The UV-vis absorption spectra (Fig. S7a†) show two absorption edges for both bulk and exfoliated Nb4P2S21. In the exfoliated sample, edges at 599 nm and 638 nm suggest two electronic transitions: the 599 nm edge likely corresponds to the primary bandgap, while the 638 nm edge may involve excitonic or defect-related transitions. The bulk sample shows edges at 556 nm and 641 nm. Tauc plots estimate an optical bandgap of 2.03 eV for both samples, aligning with HSE-calculated results (Fig. S1b†). Photoluminescence (PL) spectra (Fig. S7b†) reveal a stronger PL peak at 755 nm for exfoliated Nb4P2S21 due to quantum confinement and defect states, while bulk shows a weaker peak at 767 nm, likely due to non-radiative centers.28 Raman spectra (Fig. S7c†) indicate shifts in interlayer interactions or lattice strain in exfoliated nanoplates.
While the SEM images of bulk Nb4P2S21 reveal the facets and edges, indicating the layered stacking direction (i.e., the a-axis), a more direct confirmation is needed through cross-sectional HRTEM or electron diffraction analysis. We initially attempted to analyze lamellae cut along the major axis of the needle-like Nb4P2S21 for TEM analysis; however, the lamellae in this orientation proved unstable under a high-voltage electron beam, making it challenging to obtain information related to the crystal structure. Consequently, we opted to cut a chunk along the minor axis, which was then lifted out and transferred onto a FIB grid. The sample was further thinned on a compustage within a TF Helios 5 FX dual-beam microscope equipped with a T-Pix detector for electron diffraction collection, as demonstrated in Fig. S8.† The results of the cross-section, shown in Fig. 2h, align well with the simulated material with lattice orientation [001]. The slight rotational offset of the diffraction pattern can be attributed to the manual insertion of the FIB grid into the compustage. This indicates that the needle-like Nb4P2S21 is stacked along the etching direction with van der Waals forces connecting the a-axis. In Fig. 2j, k and S9,† HRTEM and SAED analyses were performed on the edges of the exfoliated nanoplatelets, revealing the (020) crystal planes parallel to the long axis and the (004) planes parallel to the short axis. These findings confirm that both the rod-like bulk and exfoliated flake materials are stacked along the a-axis direction, with the short b-axis and long c-axis forming the needle-like Nb4P2S21, as schematically represented in Fig. 2l. Fig. S10a and b† compare the refined XRD spectra of both the bulk powder and the exfoliated material, with both spectra fitting well with the C2/c space group of Nb4P2S21. However, a slight increase in the lattice parameters of the exfoliated material suggests a reduction in stress and a consequent expansion of the lattice post-exfoliation. When comparing the I(400)/I(310) crystal plane intensity ratio, the bulk powder exhibits a value of 9.6, larger than the 5.6 observed in the exfoliated nanoplates. In Fig. S10c–i,† we present the positions of the (400) and (310) crystal planes. The reduction in the intensity ratio might be attributed to a relative decrease in material thickness, leading to a reduced number of effective (400) planes along the a-axis, as depicted in Fig. S10c-ii and iii.†
For this material, it is possible to form metal phosphides, metal sulfides, or other types of metal phosphosulfides in different partial discharge ranges. The presence of these in situ formed substances may facilitate the conversion of lithium sulfides, achieving excellent electrochemical performance, akin to the role of catalysts in lithium–sulfur battery systems. Therefore, we first assessed the capacity contribution potential of this material across different charge–discharge ranges. As shown in Fig. 3a, the 5 cycles of CV recorded at different voltage ranges begin to exhibit two partially overlapping reduction peaks when the initial reduction voltage is between 1.0–1.7 V. Due to the high intensity and breadth of these peaks and considering the discharge curve from the complete discharge range of 0–2.6 V (Fig. 3b), the discharge capacity within this voltage range approaches half of the total discharge capacity. According to the Interface reactions tools with GGA/GGA+U (mixed) from the Materials Project, generating different proportions of reactants (Fig. 3c), we attribute this peak to the combination of lithiation and conversion reactions of Nb4P2S21:
Nb4P2S21 + 4 Li → 2LiNb2PS10 + Li2S, (2.0 V → 1.6 V) |
(16 + 4x)Li + Nb4P2S21 → 5Li2S + 4LixNbS2 + 2Li3PS4 |
(26 + 4x)Li + Nb4P2S21 → 13 Li2S + 4 LixNbS2 + 2P, (0 ≤ x ≤ 1, 2.0 V → 1.0 V) |
LixNbS2 + Li → Nb + Li2S |
P + Li → Li3P, (1.0 V → 0 V) |
Within the 0–2.6 V range, although the material initially displays a discharge capacity of up to 1500 mA h g−1, the first reversible capacity is only 630 mA h g−1, with a rapid capacity fade to about 350 mA h g−1 within the first 50 cycles. Additionally, in situ EIS spectra within the 0–2.6 V voltage range were recorded, as shown in Fig. S13.† As the material transitions from OCP to a cathodic 1.0 V, the corresponding charge resistance's depressed semicircle gradually decreases, indicating lithiation activation. From 1.0 V to 0.5 V, an additional depressed semicircle in the low-frequency region forms, corresponding to SEI formation. This semicircle continues to enlarge during the anodic scan to 2.6 V and decreases during the cathodic scan to 0 V.
As shown in Fig. 3d, when first discharged to 1.4 V, the vibrational information of the PS4 unit and the –S–S–S– linking two Nb2PS10 quickly disappears, and no observable vibration signal strength appears during the charging process. Instead, the appearance of LiSx, NbSx, S–S, S4−, and S3˙− is noted.32 The presence of S3˙− suggests the equilibrium of Li2S4 → Li2S.32b Throughout the entire charge–discharge process, this vibrational peak persists, indicating the continuous conversion process between Li2S and polysulfides. Additionally, we recorded the Nb 3d XPS spectra of exfoliated Nb4P2S21 before the electrochemical cycle, after the first discharge to 0 V, and after continuing delithiation charging to 2.6 V, as shown in Fig. 3e. Compared to Nb4P2S21 bulk crystal, exfoliated Nb4P2S21 shows a higher binding energy doublet at 204.9/207.7 eV, possibly due to partial oxidation of the material's surface leading to the formation of Nb4+. When discharged to 0 V, mainly two sets of doublets at 204.7/207.5 eV and 202.5/205.3 eV are observed, representing Nb4+ and Nb metal, respectively. The former may be due to oxidation during sample transfer or partial non-conversion to metallic NbSx, while the latter represents the Nb metal formed upon full lithiation. Continuing to charge to 2.6 V, three sets of doublets appear at 205.5/208.3 eV, 204.1/206.9 eV, and 202.4/206.2 eV, representing Nb5+, Nb2+, and Nb metal, respectively. The presence of Nb5+ may be due to the formation of NbSx (x > 2), similar to XPS results reported in analogous VSx.33 The emergence of Nb2+ might be due to the formation of Nb–P–S compounds similar to the initial structure. The existence of Nb metal in this charged state could be due to incomplete reactions preventing Nb from being fully oxidized.
As shown in Fig. 3f, the material is structurally open not only in the van der Waals layer but also in the S3 bridge connection layer. These two positions are the most likely sites for the insertion of a single lithium atom into the material. As expected from the analysis, the DFT calculations confirmed that the binding energies of Li1 (S3 bridge layer) and Li2 (van der Waals layer) are very close as depicted in Fig. 3g, with values of −3.583 eV and −3.666 eV, respectively. This indicates that lithium atoms can be embedded simultaneously at both sites during the initial stage of lithium insertion.
The long-term charge–discharge curves and recorded cycling data for various ranges under carbonate-based electrolytes are depicted in Fig. 4b and S14.† The test results show that in the 1.0–2.6 V range, the material exhibits a first discharge capacity of 756 mA h g−1, with a first reversible capacity of only 300 mA h g−1. Interestingly, as cycling progresses, this range also sees a capacity increase and the emergence of a new self-improving discharge plateau around 1.8 V, potentially indicating higher energy density as a cathode material. Over 100 cycles, the material exhibits a reversible capacity of 314 mA h g−1, performing better than measured under the potential range of 0.5–2.6 and 1.5–2.6 V (as compared in Fig. S14c†). However, further understanding and improvement are needed regarding the issue of high initial irreversible capacity for these materials to be considered promising sulfur-equivalent cathodes.
In lithium–sulfur battery systems, ether-based electrolytes, rather than carbonate-based ones, are often employed to inhibit the irreversible reaction of carbonates with polysulfides.34 Considering the high sulfur content of our material and the presence of polysulfides during the charge–discharge process as indicated by in situ Raman, the low reversible capacity observed could be related to the use of carbonate-based electrolytes in the test system. Thus, we conducted electrochemical cycling of the material in a high voltage range using an ether-based electrolyte. As shown in Fig. 4c, under the potential of 1.0–2.6 V, the material demonstrates a higher reversible capacity compared to that in a carbonate-based electrolyte, mainly due to a longer sulfur evolution plateau at 2.2 V, which remains stable through the fifth cycle. However, during extended charge–discharge cycling, the charging curves become extremely unstable, possibly due to the shuttle effect of polysulfides formed during the electrochemical reaction of sulfur-rich materials and the onset of electrode instability.35 Further recording of the material's charge–discharge curves at 0.5–2.6 V and 1.5–2.6 V in Fig. S15† reveals that the wider voltage window, the earlier the instability in the charging curves appears. Failure starts from the first cycle in the 0.5–2.6 V window, while in the 1.5–2.6 V window, electrode failure begins after 28 cycles.
Additionally, we disassembled the batteries after 100 cycles in the 1.0–2.6 V window under both carbonate-based and ether-based electrolytes. As shown in Fig. 4d and e, black material is observed on the lithium foil corresponding to the electrode material area in both cases, with a more pronounced and larger black area in the ether electrolyte, and the separator is significantly yellowed. This is because, although ether-based electrolytes can mitigate the side reaction between the electrolyte and polysulfides, they do not address the shuttle effect of polysulfides. This leads to a significant shuttle of active components to the lithium side, causing charging curve oscillations and battery failure. In the carbonate-based electrolyte, there are fewer side reactions on the lithium foil, likely because most formed polysulfides react with the carbonate-based electrolyte on the electrode material side. In Fig. 4f and g, the post-reaction material still appears flaky, but EDS measurements exhibit the element ratios of Nb:
P
:
S is 48
:
27
:
25 reacted under carbonate-based electrolyte and 46
:
24
:
30 under ether-based electrolyte, indicating a significantly lower sulfur content compared to the original material of Nb4P2S21, further verifying the polysulfide shuttle effect similar to the lithium–sulfur system, which major reason for the low reversible capacity and electrode failure. This may also explain why, although we can observe from the SEM that the material maintains its stacked layers after cycling, the crystal structure of the material has undergone significant changes due to the loss of sulfur components.
However, the results of EDS analysis represent a cumulative summation of elemental content. To observe the specific local distribution of elements from the surface to the bulk, TOF-SIMS analysis is required. The TOF-SIMS was utilized to analyze the materials after 100 cycles, examining the etching depth and distribution of elements Nb, P, and S in Nb4P2S21. Fig. 4i–k reveal that the distributions of Nb and P largely overlap; however, significant accumulations of S are evident in certain areas. This segregation of S occurs due to its extraction from the material structure during repeated lithiation processes. The aggregation of S not only leads to poor conductivity in the aggregated areas, rendering the active components underutilized but also exacerbates the shuttle effect, subsequently leading to the failure of the entire battery. Further reducing the material dimensions and adopting strategies similar to those used in lithium–sulfur batteries, such as incorporating adsorptive catalytic media like MXene and C3N4 mixtures, might effectively unleash the actual energy storage potential of these materials.36 The positive polarity profiles presented in Fig. 4l demonstrate a lower concentration of Nb on the surface compared to the bulk, primarily due to the surface being predominantly covered by the SEI layer. Additionally, the signal intensity of P in the positive mode is weak and in the negative mode is much stronger, which is attributed to the original P(5+) in the material being transformed into Li3P(3−) during the lithiation process. However, the relative intensity of S from the surface to the bulk is significantly lower than the initial proportion in Nb4P2S21, which further demonstrates the shuttling effect of sulfur components in the ether-based electrolyte. In comparison, as shown in Fig. 4m, the S component in the bulk of the material is a relatively reasonable ratio in the carbonate-based electrolyte. This indicates that the lower sulfur component detected by EDS in Fig. 4f is likely due to the side reactions between the S on the surface and the carbonate electrolyte during the lithiation process, resulting in its loss.
Footnote |
† Electronic supplementary information (ESI) available [DETAILS]. See DOI: https://doi.org/10.1039/d4na01060d |
This journal is © The Royal Society of Chemistry 2025 |