nature materials

Article https://doi.org/10.1038/s41563-025-02287-7

Solvent co-intercalation in layered cathode

active materials for sodium-ion batteries

Received: 11 June 2024 Yanan Sun 1,2,8 , Gustav Åvall 1,3,8, Shu-Han Wu 1, Guillermo A. Ferrero1,2,
Annica Freytag 1,2, Pedro B. Groszewicz4,5, Hui Wang 1,2,
Accepted: 13 June 2025
Katherine A. Mazzio 1,2, Matteo Bianchini 6, Volodymyr Baran 7,
Published online: xx xx xxxx Sebastian Risse 2 & Philipp Adelhelm 1,2

Check for updates

Solvent co-intercalation, that is, the combined intercalation of ions and
solvent molecules into electrode materials, is an additional but much
less explored lever for modifying the properties of metal-ion battery
electrodes (metal = Li, Na, Mg, etc.). Knowledge on solvent co-intercalation
is relatively scarce and largely limited to graphite anodes, for which in
sodium-ion batteries, the co-intercalation of glyme molecules is fast and
highly reversible. The use of co-intercalation for cathode active materials
(CAMs) remains much less explored. Here we investigate for a series of
sodium-layered sulfide CAMs (NaxMS2, M = Ti, V, Cr and mixtures) under
which conditions solvent co-intercalation occurs and how this process
impacts the phase behaviour, electrode breathing, redox potential and
cycle life compared to ‘Na+-only’ intercalation. Co-intercalation is a complex
process that can, for example, cause opposing fluxes, meaning that solvents
intercalate into the CAMs while sodium ions simultaneously deintercalate.
Co-intercalation leads to layered structures that can include different
amounts of confined solvated ions, ions and unbound solvent molecules.
It is an approach to designing structurally diverse, layered materials with
potential applications for batteries and beyond.

Lithium-ion and sodium-ion batteries (LIBs, SIBs) typically rely on detrimental because it leads to degradation of the electrodes7,8. In
intercalation reactions, where lithium or sodium ions are stored in current LIBs, the solid-electrolyte interphase (SEI) prevents the
the layered structures of the electrodes and exchanged between them co-intercalation of solvents9. Nevertheless, co-intercalation reactions
during charging and discharging1–4. The electrodes are separated by a can exhibit high reversibility and rapid kinetics, often enduring for
liquid electrolyte in which the ions are solvated, that is, the ions carry thousands of cycles10,11. A prominent example is the co-intercalation
a solvation shell. Intercalation of the ions from the electrolyte into the of Na+ and diglyme (2G) into graphite electrodes12–14. It is important to
electrode hence requires desolvation5. Likewise, solvation occurs when realize that the co-intercalation of solvents provides a unique oppor-
the ions deintercalate from the electrodes. In some cases, incomplete tunity for designing electrode reactions15. For example, as the solvents
stripping of the solvation shell allows solvent molecules to also interca- become part of the electrode reaction itself, solvent co-intercalation
late into the electrodes6. This process is referred to as co-intercalation allows a targeted modification of the electrode potential over few hun-
(ions and solvents jointly intercalate) and is typically seen as being dred millivolts depending on the type of co-intercalated solvent(s)16.

1
Institut für Chemie, Humboldt-Universität zu Berlin, Berlin, Germany. 2Joint Research Group Operando Battery Analysis (CE-GOBA), Helmholtz-Zentrum
Berlin für Materialien und Energie (HZB), Berlin, Germany. 3SEEL Swedish Electric Transport Laboratory, Gothenburg, Sweden. 4SE-ASPIN, Helmholtz-Zentrum
Berlin für Materialien und Energie (HZB), Berlin, Germany. 5Department of Radiation Science and Technology, Delft University of Technology, Delft,
Netherlands. 6Bavarian Center for Battery Technology (BayBatt), Bayreuth, Germany. 7Deutsches Elektronen-Synchrotron (DESY), Hamburg, Germany.
8
These authors contributed equally: Yanan Sun, Gustav Åvall. e-mail: [email protected]; [email protected]

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

a 1st cycle
b 1st cycle
c 1st cycle
2.7 2.7 2.7
2nd cycle 2nd cycle 2nd cycle

Potential (V versus Na+/Na)

Potential (V versus Na+/Na)

Potential (V versus Na+/Na)
3rd cycle 3rd cycle 3rd cycle
2.4 2.4 2.4

2.1 2.1 2.1

1.8 1.8 1.8

1.5 1.5 1.5

0 50 100 150 0 50 100 150 0 50 100 150

Specific capacity (mAh g–1) Specific capacity (mAh g–1) Specific capacity (mAh g–1)

d
110P2

101P2 103P2 112P2

002P2 106P2
8 104P2 P2
102P2
Discharge

110OP4 P2 + OP4
6 101OP4 105OP4
004OP4
114OP4
Time (h)

OP4
110Al 110O2
110Na 112O2
4 O2 ~18%
002O2 101O2

110OP4
004OP4 101OP4 105OP4 OP4
114OP4
P2 + OP4
2
Charge

002P2 101P2 103P2 106P2 112 P2

104P2 P2

102P2 110P2

3.0 2.4 1.8 1.2 0.8 1.2 1.6 2.0 4.0 4.5 5.5 6.0 6.5 7.0 7.5 8.0 3.40 3.45 3.50 6 7
Cell voltage (V versus Na+/Na) 2θ (deg), λ = 0.2073 Å a (Å) Avg. interlayer
spacing (Å)

Fig. 1 | Electrochemical behaviour of P2-NaxTiS2. a–c, Voltage profiles of P2-NaxTiS2 in 1 M NaPF6 in EC/DMC electrolyte during the first cycle at 0.125 C.
P2-NaxTiS2 in different electrolytes (1 M NaPF6 in EC/DMC (a), 0.5 M NaPF6 Right side, pink symbols, corresponding a lattice parameter; right side, green
in PC (b) and 1 M NaPF6 in 2G (c)) for the first three cycles at 0.1 C. Different symbols, corresponding average interlayer spacing; grey dahsed line, fully
electrochemical behaviours are observed with the three studied electrolytes. desodiated state. A typical intercalation reaction was detected when using
Three-electrode measurements in Swagelok-type cells were carried out to the EC/DMC-based electrolyte. A reversible phase transition of P2–OP4–O2 is
minimize the influence of counter-electrode polarization. d, Operando XRD of demonstrated in layered sulfide cathodes for the first time.

Another advantage is that the charge-transfer resistance during studies encourage the exploration of new electrode reactions, TiS2
intercalation, which is normally dominated by the energy needed to is not a true CAM because it lacks sodium (or other alkali earth ions)
strip the solvation shell17, can be minimized or even bypassed, favour- from the start, which would be required from a commercial perspective
ing high energy efficiency reactions and high rate capability18. Nota- to allow cell assembly in the discharged state24.
bly, despite their much larger size, the diffusion of ‘solvated Na+’ in Therefore, this study addresses solvent co-intercalation reactions
graphite is faster than the diffusion of Li+ (refs. 19,20). A drawback of for a series of NaxMS2 (where M = Ti, V, Cr or mixtures) compounds using
solvent co-intercalation is that the larger size of the solvated ion leads different solvents (2G, PC and an ethylene carbonate (EC)/dimethyl
to larger electrode breathing, and in the case of graphite, a decrease carbonate (DMC) mixture). The co-intercalation of Na+ in CAMs for
in specific capacity by more than two-thirds compared with conven- SIBs is demonstrated, and a reaction mechanism indicating an
tional intercalation and hence energy density. The need for excess opposite flux of ions and solvents is identified. Guided by theoreti-
electrolyte is considered a practical challenge too but electrolyte cal calculations and experimental validation, an interlayer binding
and electrode optimization is possible to minimize the amount of elec- energy–interlayer free volume model is proposed to predict solvent
trolyte required, for example, by bypassing the irreversible electrolyte co-intercalation in layered CAMs, where the co-intercalation behaviour
consumption during SEI formation. is governed by phase structure, sodium content, transition metal/anion
Although most co-intercalation studies focus on graphite anodes, species and solvent properties.
co-intercalation in cathode active materials (CAMs) has received very
little attention, and understanding of its properties and feasibility in Co-intercalation investigation in P2-NaxTiS2
batteries remains limited21. Interestingly, co-intercalation was recently Due to the elemental abundance of titanium, P2-type NaxTiS2 (P63/mmc
demonstrated by Ferrero et al. and Park et al. for layered titanium space group) synthesized by a high-temperature solid-state method
disulfide (TiS2) which is able to take up Na+ along with 2G molecules18,22. was chosen as the prototype structure for detailed investigation (Sup-
Tchitchekova et al. showed co-intercalation in TiS2 for Mg2+ and Ca2+ plementary Figs. 1–3 and Supplementary Tables 1 and 2). Defined
with propylene carbonate (PC), but high temperatures ≥100 °C potential steps are visible in the voltage profile with the EC/DMC-based
are required and cell cycling remains challenging23. Although these electrolyte, while in the PC- and 2G-based electrolytes, the voltage

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

a 18 b e 98
2.8
10 002P2 96

Potential (V versus Na+/Na)

16 002P2
2.4
94

Thickness (µm)
Discharge
Discharge

14
8 002co-intercalation 92
002co-intercalation 2.0
12
–7%
004co-intercalation 90

1.6
10 6 88
Time (h)

Time (h)
004co-intercalation
86 1.2
8 10 15
006co-intercalation Time (h)
4
6 f 130

Charge
2.8
Charge

120

Potential (V versus Na+/Na)

4
2
002P2 002P2 110 2.4

Thickness (µm)
2
100
2.0
90 +66%
3.0 2.4 1.8 1.2 0.5 1.0 1.5 2.0 3.0 2.4 1.8 1.2 0.4 0.8 1.2 1.6 2.0
Cell voltage 2θ (deg), λ = 0.2073 Å Cell voltage 2θ (deg), λ = 0.2073 Å 1.6
80
(V versus Na+/Na) (V versus Na+/Na)
70
1.2
c XRD pattern at the fully charged state
d XRD pattern at the fully charged state 8 12
Calculated line Calculated line Time (h)
Difference Difference
Bragg positions of the PC co-intercalated structure Bragg positions of the 2G co-intercalated structure g 95
2.8
Rwp = 0.991% Rwp = 0.760%

Potential (V versus Na+/Na)

Intensity (a.u.)
Intensity (a.u.)

90
χ2 = 1.36 χ2 = 1.01
2.4

Thickness (µm)
85

2.0
+28%
80

1.6
75

70 1.2
2 4 6 8 2 4 6 8
10 15
2θ (deg), λ = 0.2073 Å 2θ (deg), λ = 0.2073 Å
Time (h)

Fig. 2 | Large lattice expansion due to co-intercalation in PC- and 2G-based electrolytes (aluminium and sodium peaks omitted). e–g, Three-electrode
electrolytes. a,b, Operando XRD of P2-NaxTiS2 in 0.5 M NaPF6 in PC (a) and 1.0 M operando electrochemical dilatometry results for P2-NaxTiS2 in EC/DMC-based
NaPF6 in 2G (b) electrolytes for the first cycle at 0.125 C. Compared to the (e), PC-based (f) and 2G-based (g) electrolytes for the first cycle at 0.1 C. The blue
EC/DMC electrolyte (Fig. 1d), a much larger expansion of the structure is curves show the thickness of the P2-NaxTiS2 electrode during cycling. Charging
observed in the PC and 2G electrolytes. c,d, Le Bail refinement of the XRD pattern (desodiation) in the EC/DMC electrolyte leads to a minor contraction of the
in the fully desodiated state in the first cycle for PC-based (c) and 2G-based (d) electrode, whereas a large expansion is found for the PC and 2G electrolytes.

plateaus smear out and the profiles become more sloping (Fig. 1a–c Na+/Na resulted in two well-defined increases in the position of the
and Supplementary Fig. 4). Notably, an additional pair of plateaus 002/004 peak (2θ shifted from 1.69° to 1.83° and then to 2.04°), which
located at 2.02 V (desodiation) and 1.77 V (sodiation) appears with 2G. is related to the successive formation of OP4 and O2 phases, as cor-
The long-term cycling stability and high current density rate capabil- roborated by Rietveld refinement results (Supplementary Figs. 10 and
ity of P2-NaxTiS2 in 2G is better than that in EC/DMC (Supplementary 11 and Supplementary Tables 3 and 4). Both the a lattice parameter
Figs. 5 and 6). Interestingly, the appearance of this additional pair of and the average interlayer spacing (∼18%) decrease during the first
plateaus (2.02/1.77 V versus Na+/Na) in 2G is very reversible and is main- desodiation process. Given the small overall changes in lattice para­
tained even after 2,000 cycles, while almost all plateaus with EC/DMC meters, a conventional intercalation mechanism seems likely.
degrade gradually upon cycling. Additionally, a much narrower voltage In contrast to EC/DMC, the phase behaviour of P2-NaxTiS2 cycled
gap (128 mV) after 200 cycles is found using 2G (Supplementary Fig. 7). in PC- and 2G-based electrolytes is different. During the first deso-
In comparison, the PC-based electrolyte leads to an inferior capacity diation process, the pristine P2 phase is maintained until the cell is
retention upon long-term cycling. The reaction mechanisms with the desodiated to 2.43 V and 2.47 V versus Na+/Na for PC (Fig. 2a and Sup-
three electrolytes were therefore studied using synchrotron operando plementary Fig. 12) and 2G (Fig. 2b and Supplementary Fig. 13), respec-
X-ray diffraction (XRD). tively. After this, the P2 phase disappears and a new phase emerges
Figure 1d shows the operando XRD results for P2-NaxTiS2 in the with the 002 peaks shifting to smaller angles, which is indicative of
EC/DMC-based electrolyte using an in-house designed operando a substantial interlayer expansion. A Le Bail refinement of the XRD
cell (Supplementary Figs. 8 and 9). Charging to 2.3 V and 2.45 V versus pattern in the fully oxidized state of the first cycle produces a perfect

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

match with an expanded structure in both the PC (Fig. 2c) and 2G Na+. After saturation, sodiation proceeds via bare Na+ intercalation,
(Fig. 2d) cases. At these extreme interlayer spacings with PC and with pronounced shifts attributed to through-space pseudocontact
2G electrolytes, which are 163% (18.3880 Å) and 106% (14.3458 Å) effects30 from Ti3+ paramagnetic centres in NaxTiS2. This interaction
larger than the pristine P2 phase (6.9794 Å), respectively, PC- or also influences the 13C NMR spectra (Fig. 3b, Supplementary Figs. 22–26
2G-solvated ions have ample space to fit into the structure (Supple- and Supplementary Tables 5–8), revealing three distinct 2G environ-
mentary Figs. 12c and 13c). Ex situ scanning electron microscopy (SEM) ments: bulk electrolyte outside the layered structure, a paramagnetic
and XRD further confirm the formation of cracks and an expanded shift assigned to solvated Na+ in the interlayer space, and broadened
structure for the cases of PC and 2G (Supplementary Figs. 14–18), resonances assigned to free 2G solvent in the interlayer space. Follow-
which is typical for a solvent co-intercalation mechanism. During ing the initial co-intercalation of solvent, there are, at any SoC, always
subsequent sodiation, the expanded structure is maintained in Na+, solvated Na+ and free solvents in the structure.
both PC and 2G electrolytes, with only partial reappearance of the To investigate this further, the mass of the electrodes was moni-
original P2 phase. However, the reversibility using 2G is better, indicat- tored (Fig. 3c and Supplementary Fig. 27). Before any electrochemical
ing a greater structural flexibility. Furthermore, the 002-peak position driving force was applied, the increase in electrode mass was similar
using 2G (0.83° 2θ) is higher than when using PC (0.65° 2θ), indicating for all three electrolytes, attributable to electrode wettability at OCV.
that the co-intercalation of PC molecules causes a 30% larger expansion During desodiation, the trend for EC/DMC is as expected, that is, deso-
than when 2G is used. diation leads to a continuous reduction of the electrode mass. For 2G
Operando electrochemical dilatometry (ECD) continuously and especially for PC, however, a substantial mass increase is found
probes the thickness change of the entire electrode and, as a comple- after desodiation to 2.3 V versus Na+/Na, followed by a slight decrease
ment to XRD, is sensitive to the formation of, or any change in, amor- upon further desodiation to 2.8 V versus Na+/Na. This change is well
phous phases25,26. Using EC/DMC (Fig. 2e), the electrode thickness in line with the operando ECD and XRD results discussed above, that
continuously decreases/increases during desodiation/sodiation with is, solvents diffuse into the crystal structure leading to a substantial
a change in the single-digit range, which is typical for intercalation increase in mass, while a slight decrease in electrode thickness was
reactions27,28. A quite different behaviour is found using PC (Fig. 2f) and also observed in ECD when reaching high redox potentials. Overall,
2G (Fig. 2g); in these electrolytes the electrode substantially expands this suggests that three stages occur during desodiation (as illustrated
during desodiation. We were initially very surprised at this behaviour in Fig. 3d and Supplementary Fig. 28): in stage 1, conventional deinter-
because, during desodiation, sodium ions leave the structure and one calation of Na+ occurs. Stage 2 begins when the sodium content drops
would therefore expect a contraction rather than a substantial expan- below a critical threshold, at which point Na+ deintercalation continues
sion of the electrode. The electrode expansion peaks before the end concurrently with solvent intercalation. This defines a ‘co-intercalation
of desodiation at around 2.5 V versus Na+/Na and amounts to 66% and with opposite flux’ mechanism, where both Na+ and solvent molecules
28% for PC and 2G, respectively. Upon further desodiation to higher coexist within the structure. In stage 3, both species are extracted
potentials, the electrode thickness slightly decreases again, that is, (de-cointercalation), although a fraction of solvent may remain trapped
the electrode expands and contracts during the first desodiation pro- within the host lattice.
cess, indicating a very complex behaviour. Subsequently during the Because free solvent molecules are electrically neutral, there
sodiation/desodiation process (Supplementary Fig. 19), the electrode is no electrochemical driving force for them to enter the P2-NaxTiS2
remains expanded in PC and 2G, which aligns well with the operando structure. Instead, it is likely that a chemical process or simply absorp-
XRD findings. These observations are understood to arise from solvent tion is the cause. To prove this, the electrode desodiated to 2.3 versus
co-intercalation processes that occur in cells with electrolytes contain- Na+/Na in EC/DMC was disassembled from the cell and soaked in
ing PC or 2G, where solvents are unexpectedly entering the structure pure PC solvent for 2 days. This procedure surprisingly results in a
as the ions are exiting during the initial desodiation. substantial mass increase (150%) and an expanded structure, as evi-
denced by the corresponding XRD pattern (Fig. 3e). This confirms
Mechanism of co-intercalation in layered sulfide that the expansion of the P2-NaxTiS2 lattice and the electrode dur-
cathodes ing desodiation in PC and 2G electrolytes is indeed the result of the
Detailed information on how the intercalated species (Na+ and 2G) chemical uptake of these solvents.
interact with each other and with the host structure was obtained by Notably, stage 2 only emerges when the cell is desodiated above
23
Na and 13C solid-state NMR measurements at different states of charge a threshold potential. A reduced voltage window of 1.3–2.1 V versus
(SoCs) and using P2-NaxTiS2 as host. The 23Na NMR spectra (Fig. 3a and Na+/Na is applied to test whether solvent uptake at lower voltages
Supplementary Figs. 20 and 21) provide evidence for both solvated can be excluded. The identical voltage behaviours and ex situ XRD
Na+ (at +7 ppm)20 and bare Na+ (+50 to +200 ppm) occupying the inter- results (Fig. 3f–h and Supplementary Figs. 29 and 30) confirm that
layer space. The signal at –8.6 ppm is due to Na+ from the bulk elec- only sodium (de)intercalation takes place in the low-voltage regime in
trolyte29. The peak area of solvated Na+ reaches a maximum at 1.75 V the three electrolytes, in contrast to the solvent uptake of PC and 2G
and remains similar at 1.30 V, suggesting saturation of co-intercalated when reaching a sufficiently high redox potential. Thus, a low sodium

Fig. 3 | Mechanism of co-intercalation in layered sulfide cathodes. a, 23Na of Na+ occurs. For PC or 2G, Na+ deintercalation also takes place but simultaneous
solid-state NMR spectra (20 kHz) of P2-NaxTiS2 at different SoCs in 1 M NaPF6 in 2G solvent intercalation occurs as soon as the sodium content in P2-NaxTiS2 becomes
electrolyte. Paramagnetic interaction with Ti3+ centres result in signals between sufficiently low. The intercalated solvent prefers to interact with sodium ions in the
+50 and +200 ppm for intercalated Na+, with the remaining environments related layered structure, forming solvated sodium ions. Towards the end of desodiation,
to solvated Na+ in the interlayer space (+7 ppm) and bulk electrolyte (−8.6 ppm). de-cointercalation, that is, combined deintercalation of Na+ and solvents, takes
b, 13C solid-state NMR spectrum of the P2-NaxTiS2 of the first desodiation at place. e, XRD patterns for a P2-NaxTiS2 electrode desodiated using an EC/DMC
2.8 V versus Na+/Na. Red, experimental spectrum; blue, sum of fitted lines; light electrolyte (green line) and after subsequently soaking this electrode in PC for 48 h
blue, free solvent in the layered structure; orange, bulk electrolyte outside (orange line). The measurements show that the expansion is caused by an uptake
layered structure; green, solvated Na+ in layered structure. c, Relative mass of of specific solvents such as PC and occurs by chemical diffusion. f–h, Voltage
the P2-NaxTiS2 electrodes in the pristine state (before cell assembly), at OCV profiles of P2-NaxTiS2 in EC/DMC (f), PC (g) and 2G (h) electrolytes in a reduced
(electrodes are wetted), and after desodiation to 2.3 V and 2.8 V versus Na+/Na voltage window of 1.3–2.1 V versus Na+/Na at 0.1 C (three-electrode measurements
in the different electrolytes. d, Schematic diagram of solvent uptake and (de-) in Swagelok-type cells). The identical shapes demonstrate that solvent
co-intercalation in P2-NaxTiS2 during charging. For EC/DMC, only deintercalation intercalation does not take place for high sodium contents in NaxTiS2 (x > 0.6).

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

a Pristine
b Exp. spectrum
Fitted line
1st [email protected] V +
Solvated Na (in layered structure)
1st [email protected] V
Free solvent (in layered structure)
1st [email protected] V
Bulk electrolyte (outside layered structure)

+ +
Bare Na Solvated Na Bulk electrolyte
200 150 100 50 0 80 78 76 74 72 70 68 66 64 62 60 58 56 54
23 13
Na chemical shift (ppm) C chemical shift (ppm)

c d
Solvent uptake of the electrodes
PC or 2G
170% EC/DMC
PC

O layer Shrinkage
18% Expansion
2.8 V 2.8 V
Mass change

167% for PC
2G O layer
137% for 2G

EC/DMC O2 phase

Co-intercalation
structure
100% Charging

Pristine OCV Desodiated Desodiated P layer

electrode (in cell) at 2.3 V at 2.8 V Solvent
2.32 V 2.38 V
uptake
O layer
e Na+
Na+
Solvent uptake deintercalation
deintercalation
OP4 phase
Electrodes charged in 1 M NaPF6 in EC/DMC Na+ deintercalation
+
Solvent uptake

Charge to 2.3 V and soak in PC

Intensity (a.u.)

→ Structure expanded
P layer

2.23 V 2.1 V
P layer

Na+ Na+
Charge to 2.3 V deintercalation deintercalation
P2 phase
P2 phase
→ No structure change (only deintercalation)
1st charge

10 20 30 40 50 60
Na ions Ti ions S ions Solvent molecules Solvated Na ions
2θ (deg), λ = 1.5406 Å

f g h
2.7 1st cycle 2.7 1st cycle 2.7 1st cycle
2nd cycle 2nd cycle 2nd cycle
No solvent 3rd cycle Region of solvent 3rd cycle Region of solvent 3rd cycle
Potential (V versus Na+/Na)

Potential (V versus Na+/Na)

Potential (V versus Na+/Na)

2.4 uptake/release 2.4 uptake/release 2.4 uptake/release

2.1 2.1 2.1

1.8 Only Na+ 1.8 Only Na+ 1.8 Only Na+

intercalation intercalation intercalation
1.3–2.1 V 1.3–2.1 V 1.3–2.1 V
1.5 1.5 1.5

0 50 100 150 0 50 100 150 0 50 100 150

Specific capacity (mAh g–1) Specific capacity (mAh g–1) Specific capacity (mAh g–1)

Nature Materials
 Article https://doi.org/10.1038/s41563-025-02287-7

Co-intercalation in other layered cathodes

Graphite A generalization of the concept was explored by studying a series of
Interlayer binding energy

–0.1
layered sulfides, NaxMS2 (where M = Ti, V, Cr or mixtures). The P3-type
TiS2 NaxVS2, O3-type NaxTiS2 and P3-NaTi0.5V0.5S2 show similar characteristic
(eV per atom)

–0.2 P2-NaxTiS2 features in their voltage profiles and phase behaviours during cycling,
O3-NaxTiS2 that is, co-intercalation occurs with the PC- and 2G-based electrolytes,
P3-NaxVS2 but not when using EC/DMC (Fig. 5a–e and Supplementary Figs. 38–41).
–0.3 O3-NaxCrS2
Notably, P3-NaxVS2 demonstrates improved cycling stability with the
O3-NaxCoO2
co-intercalation reaction (Supplementary Fig. 42). The interlayer
P2-NaxMnO2
–0.4
spacing of various co-intercalated CAMs depends more on the inter-
0 10 20 30 40 50 60 70 calated solvent, as shown in Supplementary Fig. 43, than on the type
Interlayer free volume (Å3) of layered structure/compound. Upon solvent uptake, the interlayer
Fig. 4 | Interlayer binding energy and interlayer free volume of a variety of forces between transition metal layers are weakened as the solvent
layered sulfides and oxides. The calculated interlayer binding energy and molecules occupy the interlayer space. However, the voltage hysteresis
interlayer free volume of the optimized structures in their fully (x = 1, full of the layered CAMs with co-intercalation depends on both the elec-
symbols) and partially (x = 0.25, hollow symbols) sodiated states. The layered trolyte solvent and the type of the layered structure (Supplementary
structures of the sulfide cathodes are much more open and easier to expand than Figs. 44–46). The P3- or O3-type layered sulfides show a similar degree
those of the layered oxides. Moreover, the structures with low sodium content of voltage hysteresis for intercalation and co-intercalation processes.
are much easier to expand and contain a larger degree of empty space for solvent Equilibration measurements over a period of 160 h showed that there is
uptake. The interlayer binding energies for graphite and TiS2 are added for no notable difference in equilibration voltage or self-discharge between
comparison.
the electrodes being intercalated and co-intercalated (Supplemen-
tary Fig. 47). Notably, an exception is found for the O3-type NaxCrS2,
where no co-intercalation is found with the three different electrolytes
content (x < 0.6 in P2-NaxTiS2) is crucial for enabling the uptake of (EC/DMC, PC and 2G; Supplementary Fig. 48). This is also the sulfide that
PC or 2G solvent to expand the structure (as illustrated in Fig. 3d and has the lowest interlayer free volume in its sodium-deficient structure
supported by Supplementary Fig. 31). (Fig. 4). It therefore seems that the transition metal has a major influ-
ence on the ability for solvent intercalation to occur. In fact, we find
Theoretical considerations that for mixed layered sulfides containing more than one transition
Density functional theory (DFT) was used to further investigate metal, the behaviour depends on the ratio of the transition metals. For
the mechanism. Structural optimization using the Vienna Ab initio a series of NaCryTi1−yS2 compounds with y = 0.33, 0.5 and 0.67 (Fig. 5f
Simulation Package (VASP) was carried out on P2-Na0.25TiS2, expanded and Supplementary Figs. 49 and 50), PC or 2G co-intercalation occurs
in the c direction to match the XRD observations, with a solvation when titanium dominates (y = 0.33), while only intercalation takes place
shell placed between TiS2 layers (Supplementary Fig. 32). However, when chromium dominates (y = 0.67). For an equal content of titanium
during structural optimization, the structure contracts as the solvents and chromium (y = 0.50), however, solvent co-intercalation takes place
break away from the solvated ion and instead begin to coordinate in the case of 2G but not with EC/DMC or PC. Overall, the results indicate
to the sodium layer, that is, the solvents begin to coordinate to Na+ that solvent intercalation can occur in many (but not all) layered sulfides
from the interior of the sodium layer, revealing that a single solva- and depends on the type of solvent, the type of transition metal(s), and
tion shell is not sufficient to generate a stable expanded structure the sodium content, providing a large chemical playground for further
as observed in operando XRD. Thus, more solvent is required in the studies. A summary of the co-intercalation behaviour of layered sulfide
interlayer to maintain the expanded structure, and we have previously cathodes is given in Table 1. In comparison, the representative layered
shown that there is a large amount of free solvent present in materials oxides, NaMnO2, P2-Na0.7CoO2, O3-NaCoO2 and NaNiO2, show only an
where solvent co-intercalation occurs20,24,29,31. However, in contrast to intercalation reaction for the three different solvents (Supplementary
previously studied anodes, there is no electrochemical driving force Figs. 51–54). These observations are in good agreement with the DFT
for the solvent to enter the bulk of the material during a desodiation results shown in Fig. 4, which indicate that co-intercalation may well
process. Rather, this process appears to occur by simple diffusion of occur (next to graphite) in a variety of layered sulfides, while the lattice
solvent into the structure which subsequently expands. For such a of layered oxides is probably too dense and rigid to allow for such a
process to occur, materials with weakly bound layers and large voids, process. Layered oxides, even those with low sodium contents, exhibit
that is, materials that have large interlayer free volumes and are easy to relatively high binding energies (even higher than or comparable to TiS2)
expand, should be prime candidates for solvent uptake. The interlayer and in some cases the interlayer free volume is too low to favour the
binding energy and the interlayer free volume are thus proposed as occurrence of the co-intercalation reaction. However, layered sulfide
vital parameters to study co-intercalation. Computing the interlayer cathodes are capable of 2G or PC co-intercalation.
binding energy and the interlayer free volume for several layered The main solvent properties responsible for promoting solvent
sulfides, and for two-layered oxide cathodes, shows that in general co-intercalation are the stability of the solvation shell (desolvation
the layered sulfides (in both sodium-deficient and sodium-replete free energy), the size of the solvation shell and the oxidative/reduc-
cases) are much easier to expand and contain more free volume than tive stability of the electrolyte. The more stable the solvation shell,
the layered oxides (Fig. 4 and Supplementary Fig. 33). Crucially, how- the higher the activation barrier for stripping the solvation shell
ever, all the structures become much easier to expand and contain and hence an increased likelihood of solvent co-intercalation32.
more free volume at low sodium content. The sodium-deficient layered Considering size effects, solvents that can form small solvation shells
sulfides require very low interlayer energies of 0.066–0.096 eV per (such as water), or solvents with the ability to wrap around the cation
atom to expand, comparable to graphite (0.057 eV per atom), and have (such as glymes), are more likely to co-intercalate as less energy is
very large interlayer free volumes. In particular, the free volume for needed to expand the interlayer space of the layered structure (Sup-
P2-Na0.25TiS2 increases by more than 700% compared with P2-NaTiS2 plementary Fig. 55)24. Finally, the electrolyte needs to be sufficiently
and has the largest free volume. In comparison, TiS2, characterized by stable to avoid decomposing before the electrode reaction is com-
a higher interlayer energy of 0.129 eV per atom, exhibited no chemical pleted. Overall, layered sulfides show favourable properties for solvent
uptake of PC or 2G solvents (Supplementary Figs. 34–37)18. co-intercalation because their redox potential is typically in a safe

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

a b EC/DMC PC 2G d
1st cycle

1st discharge
1st discharge

1st discharge
2.7
2nd cycle
3rd cycle
2.4

003P3
2.1

Intensity (a.u.)
Co-intercalation
phase 2G
1.8
EC/DMC
001O1 Co-intercalation
1.5 phase PC

2.7 1st cycle

1st charge
Potential (V versus Na+/Na)

1st charge

1st charge
2nd cycle EC/DMC
3rd cycle
2.4
003P3 10 20 30 40 50 60
003P3 2θ (deg), λ = 1.5406 Å
2.1

1.8 0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0 e
PC 2θ (deg), 2θ (deg), 2θ (deg),
1.5 λ = 0.2073 Å λ = 0.2073 Å λ = 0.2073 Å PC
c
Solvent uptake of the electrodes
2.7 1st cycle
170%
2nd cycle

Intensity (a.u.)
PC
3rd cycle 3 5 7
2.4
2θ (deg), λ = 1.5406 Å
Mass change

2G
2.1 2G
PC
1.8 EC/DMC
EC/DMC
2G
1.5 100% Charging
0 50 100 150 200 10 20 30 40 50 60
Pristine OCV Desodiated Desodiated
Specific capacity (mAh g–1) electrode (in cell) at 2.3 V at 2.8 V 2θ (deg), λ = 1.5406 Å

Co-intercalation No co-intercalation
Ti dominates Cr dominates
f
3.3 3.3 3.3
NaCr0.33Ti0.67S2 NaCr0.5Ti0.5S2 NaCr0.67Ti0.33S2
EC/DMC 3.0 EC/DMC 3.0 EC/DMC
Cell voltage (V versus Na+/Na)

Cell voltage (V versus Na+/Na)

3.0
Cell voltage (V versus Na+/Na)

PC PC PC
2G 2G 2G
2.7 2.7 2.7

Intensity (a.u.)
Intensity (a.u.)
Intensity (a.u.)

2.4 2.4 2.4

2.1 2.1 2.1

3 4 5 6 7 8 9
3 4 5 6 7 8 9 2θ (deg),
3 4 5 6 7 8 9
2θ (deg), λ = 1.5406 Å
1.8 2θ (deg), 1.8 1.8
λ = 1.5406 Å
λ = 1.5406 Å

1.5 1.5 1.5

0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Normalized capacity Normalized capacity Normalized capacity

Fig. 5 | Co-intercalation in other layered sulfides. a, Voltage profiles of P3- and 2G indicates solvent intercalation. d,e, Ex situ XRD patterns of O3-NaxTiS2
NaxVS2 in EC/DMC, PC and 2G electrolytes at 0.1 C showing different behaviours. (d) and NaTi0.5V0.5S2 (e) for the first discharge at 2.19 V versus Na+/Na (sodiation)
Three-electrode measurements in Swagelok-type cells were used to minimize for different electrolytes. Inset e: enlarged region of 3°–8° of the XRD pattern
the influence of counter-electrode polarization. b, Results from operando XRD of NaTi0.5V0.5S2 in the PC-based electrolyte. Both compounds form an expanded
of P3-NaxVS2 electrodes cycled in EC/DMC, PC and 2G at 0.125 C. Similar to P2- structure when using PC or 2G. f, Normalized voltage profiles (second cycle) and
NaxTiS2, a large lattice expansion indicative of solvent intercalation occurs with ex situ XRD patterns at 2.19 V versus Na+/Na (sodiation) in insets for NaCryTi1−yS2
PC and 2G. c, Relative mass of P3-NaxVS2 electrodes in the pristine state (before (y = 0.33, 0.5 or 0.67) in different electrolytes. Blue circles in insets: the peak of
cell assembly), at OCV (electrodes are wetted), and after desodiation to 2.3 V and the co-intercalation phase. The co-intercalation behaviour changes with the
2.8 V versus Na+/Na in the different electrolytes. The large mass increase for PC dominance of different transition metals (titanium or chromium).

region (1.3–3.5 V) where many solvents are still stable and because similarity. However, it should be noted that this difference in behaviour
the interlayer energy is relatively weak compared with, for example, has been known for many years in the context of lithium intercalation
layered oxides. The fact that PC can co-intercalate in some layered into graphite. Several possible explanations have been presented,
sulfides while EC cannot appears surprising considering their chemical such as the methyl group lending more flexibility to the PC molecule

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

Table 1 | Ability for co-intercalation reactions of layered 5. Xu, K., von Cresce, A. & Lee, U. Differentiating contributions to ‘ion
sulfide cathodes transfer’ barrier from interphasial resistance and Li⁺ desolvation at
electrolyte/graphite interface. Langmuir 26, 11538–11543 (2010).
Compounds Phase Co-intercalation 6. Besenhard, J. O. & Fritz, H. P. The electrochemistry of black
EC/DMC PC 2G carbons. Angew. Chem. Int. Ed. 22, 950–975 (1983).
NaxTiS2 P2 × ✓ ✓ 7. Wagner, M. R., Albering, J. H., Moeller, K.-C., Besenhard, J. O. &
Winter, M. XRD evidence for the electrochemical formation of
NaxVS2 P3 × ✓ ✓
Li⁺(PC)yCn− in PC-based electrolytes. Electrochem. Commun. 7,
NaxTiS2 O3 × ✓ ✓ 947–952 (2005).
NaTi0.5V0.5S2 P3 × ✓ ✓ 8. Houdeville, R. G., Black, A. P., Ponrouch, A., Palacín, M. R. &
Fauth, F. Operando synchrotron X-ray diffraction studies on TiS2:
NaxCrS2 O3 × × ×
the effect of propylene carbonate on reduction mechanism.
NaCr0.67Ti0.33S2 O3/P3 × × × J. Electrochem. Soc. 168, 030514 (2021).
NaCr0.5Ti0.5S2 O3/P3 × × ✓ 9. Chung, G.-C. et al. Origin of graphite exfoliation: an investigation
NaCr0.33Ti0.67S2 O3/P3 × ✓ ✓ of the important role of solvent cointercalation. J. Electrochem.
Soc. 147, 4391–4398 (2000).
10. Park, J., Xu, Z.-L. & Kang, K. Solvated ion intercalation in graphite:
compared with EC9,33,34. However, it should also be noted that EC is sodium and beyond. Front. Chem. 8, 432 (2020).
solid at room temperature and is therefore typically mixed with other 11. Guo, H., Elmanzalawy, M., Sivakumar, P. & Fleischmann, S.
linear solvents such as DMC. This results in a less stable solvation shell, Unifying electrolyte formulation and electrode nanoconfinement
making co-intercalation in an EC-based electrolyte less likely. More design to enable new ion–solvent cointercalation chemistries.
importantly, the two solvents differ in their ability to form a stable SEI Energy Environ. Sci. 17, 2100–2116 (2024).
layer on graphite, with EC forming an SEI layer that effectively prevents 12. Jache, B. & Adelhelm, P. Use of graphite as a highly reversible
solvent co-intercalation35. electrode with superior cycle life for sodium-ion batteries by
making use of co-intercalation phenomena. Angew. Chem. Int. Ed.
Conclusions 53, 10169–10173 (2014).
Solvent co-intercalation can be used to tune the properties of some 13. Kim, H. et al. Sodium storage behavior in natural graphite using
layered CAMs for Na-ion batteries and depends on the specific ether-based electrolyte systems. Adv. Funct. Mater. 25, 534–541
CAM–solvent combination. For NaxMS2 (M = Ti, V, Cr or mixtures), (2015).
intercalation occurs in EC/DMC electrolytes, while for M = Ti and 14. Goktas, M. et al. Graphite as cointercalation electrode for
V, co-intercalation occurs in PC and 2G electrolytes. Notably, while sodium-ion batteries: electrode dynamics and the missing
co-intercalation in graphite anodes typically lowers the specific capacity solid electrolyte interphase (SEI). Adv. Energy Mater. 8, 1702724
(around 110 mAh g−1 for co-intercalation compared with, for example, (2018).
372 mAh g−1 for lithium intercalation), the penalty for the CAMs studied 15. Ferrero, G. A. et al. Solvent co-intercalation reactions for batteries
here is very small. The process of co-intercalation in sulfide CAMs is and beyond. Chem. Rev. 125, 3401–3439 (2025).
found to be very complex and depends on the specific combination 16. Jache, B., Binder, J. O., Abe, T. & Adelhelm, P. A comparative
of CAMs, solvents and the SoC of the electrode. The interlayer binding study on the impact of different glymes and their derivatives as
energy and the interlayer free volume can be used as descriptors to electrolyte solvents for graphite co-intercalation electrodes in
predict whether co-intercalation can take place. Initial charging causes lithium-ion and sodium-ion batteries. Phys. Chem. Chem. Phys. 18,
a mechanism with an opposite flux, meaning that when Na+ deinter- 14299–14316 (2016).
calates, solvent molecules intercalate and expand the lattice once a 17. Abe, T., Fukuda, H., Iriyama, Y. & Ogumi, Z. Solvated Li-ion transfer
threshold potential (≥2.1 V versus Na+/Na for P2-NaxTiS2) is exceeded. at interface between graphite and electrolyte. J. Electrochem.
Depending on the SoC, the layered structure contains confined solvated Soc. 151, A1120 (2004).
ions, ions and unbound (‘free’) solvent molecules. Overall, this study 18. Alvarez et al. Co-intercalation batteries (CoIBs): role of TiS2 as
shows that co-intercalation of ions and solvents into layered structures electrode for storing solvated Na ions. Adv. Energy Mater. 12,
provides an alternative and versatile approach for designing materials. 2202377 (2022).
19. Jung, S. C., Kang, Y.-J. & Han, Y.-K. Origin of excellent rate and
Online content cycle performance of Na⁺-solvent cointercalated graphite vs.
Any methods, additional references, Nature Portfolio reporting sum- poor performance of Li⁺-solvent case. Nano Energy 34, 456–462
maries, source data, extended data, supplementary information, (2017).
acknowledgements, peer review information; details of author con- 20. Leifer, N., Greenstein, M. F., Mor, A., Aurbach, D. & Goobes, G.
tributions and competing interests; and statements of data and code NMR-detected dynamics of sodium co-intercalation with diglyme
availability are available at https://doi.org/10.1038/s41563-025-02287-7. solvent molecules in graphite anodes linked to prolonged
cycling. J. Phys. Chem. C 122, 21172–21184 (2018).
References 21. McKinnon, W. R. & Dahn, J. R. How to reduce the cointercalation
1. Manthiram, A. A reflection on lithium-ion battery cathode of propylene carbonate in LixZrS2 and other layered compounds.
chemistry. Nat. Commun. 11, 1550 (2020). J. Electrochem. Soc. 132, 364–366 (1985).
2. Yabuuchi, N. et al. P2-type Nax[Fe1/2Mn1/2]O2 made from Earth- 22. Park, J., Kim, S. J., Lim, K., Cho, J. & Kang, K. Reconfiguring sodium
abundant elements for rechargeable Na batteries. Nat. Mater. 11, intercalation process of TiS2 electrode for sodium-ion batteries
512–517 (2012). by a partial solvent cointercalation. ACS Energy Lett. 7, 3718–3726
3. Saha, S. et al. Exploring the bottlenecks of anionic redox in Li-rich (2022).
layered sulfides. Nat. Energy 4, 977–987 (2019). 23. Tchitchekova, D. S. et al. Electrochemical intercalation of
4. Wang, X. et al. Achieving a high-performance sodium-ion pouch calcium and magnesium in TiS2: fundamental studies related
cell by regulating intergrowth structures in a layered oxide to multivalent battery applications. Chem. Mater. 30, 847–856
cathode with anionic redox. Nat. Energy 9, 184–196 (2024). (2018).

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

24. Åvall, G. et al. In situ pore formation in graphite through solvent 32. Yoon, G., Kim, H., Park, I. & Kang, K. Conditions for reversible Na
co-intercalation: a new model for the formation of ternary intercalation in graphite: theoretical studies on the interplay
graphite intercalation compounds bridging batteries and among guest ions, solvent, and graphite host. Adv. Energy Mater.
supercapacitors. Adv. Energy Mater. 13, 2301944 (2023). 7, 1601519 (2017).
25. Escher, I., Hahn, M., Ferrero, G. A. & Adelhelm, P. A practical guide 33. Xu, K. Electrolytes, Interfaces and Interphases: Fundamentals and
for using electrochemical dilatometry as operando tool in battery Applications in Batteries (Royal Society of Chemistry, 2023).
and supercapacitor research. Energy Technol. 10, 2101120 (2022). 34. Chung, G.-C., Kim, H.-J., Jun, S.-H. & Kim, M.-H. New cyclic
26. Palaniselvam, T. et al. Sodium storage and electrode dynamics carbonate solvent for lithium ion batteries: trans-2,3-butylene
of tin–carbon composite electrodes from bulk precursors for carbonate. Electrochem. Commun. 1, 493–496 (1999).
sodium-ion batteries. Adv. Funct. Mater. 29, 1900790 (2019). 35. Xu, K. Electrolytes and interphases in Li-ion batteries and beyond.
27. Nayak, P. K. et al. Investigation of Li1.17Ni0.20Mn0.53Co0.10O2 Chem. Rev. 114, 11503–11618 (2014).
as an interesting Li- and Mn-rich layered oxide cathode
material through electrochemistry, microscopy, and in situ Publisher’s note Springer Nature remains neutral with regard to
electrochemical dilatometry. ChemElectroChem 6, 2812–2819 jurisdictional claims in published maps and institutional affiliations.
(2019).
28. Spingler, F. B., Kücher, S., Phillips, R., Moyassari, E. & Jossen, A. Open Access This article is licensed under a Creative Commons
Electrochemically stable in situ dilatometry of NMC, NCA and Attribution 4.0 International License, which permits use, sharing,
graphite electrodes for lithium-ion cells compared to XRD adaptation, distribution and reproduction in any medium or format,
measurements. J. Electrochem. Soc. 168, 040515 (2021). as long as you give appropriate credit to the original author(s) and the
29. Escher, I., Freytag, A. I., Lopez del Amo, J. M. & Adelhelm, P. source, provide a link to the Creative Commons licence, and indicate
Solid-state NMR study on the structure and dynamics of graphite if changes were made. The images or other third party material in this
electrodes in sodium-ion batteries with solvent co-intercalation. article are included in the article’s Creative Commons licence, unless
Batter. Supercaps 6, e202200421 (2023). indicated otherwise in a credit line to the material. If material is not
30. Pell, A. J., Pintacuda, G. & Grey, C. P. Paramagnetic NMR in included in the article’s Creative Commons licence and your intended
solution and the solid state. Prog. Nucl. Magn. Reson. Spectrosc. use is not permitted by statutory regulation or exceeds the permitted
111, 1–271 (2019). use, you will need to obtain permission directly from the copyright
31. Gotoh, K. et al. Structure and dynamic behavior of sodium– holder. To view a copy of this licence, visit http://creativecommons.
diglyme complex in the graphite anode of sodium ion battery org/licenses/by/4.0/.
by 2H nuclear magnetic resonance. J. Phys. Chem. C 120,
28152–28156 (2016). © The Author(s) 2025

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

Methods Ex situ XRD patterns of the electrodes were collected at room

Materials synthesis temperature on a Bruker D2 Phaser diffractometer equipped with Cu Kα
The appropriate amounts of sodium sulfide (Na2S; Alfa Aesar), radiation (λ = 1.5406 Å) at a voltage of 30 kV and a current of 10 mA, in
sulfur (Sigma Aldrich) and elemental transition metal (titanium, Alfa step-scan mode with a step size of 0.02° (2θ) and step time of 0.1–1.0 s in
Aesar, 325 mesh, 99%; vanadium, abcr, 325 mesh, 99.5%; chromium, a 2θ range of 3°–80°. The morphological information of the active mate-
Sigma Aldrich, 100 mesh, 99.5%) were well ground and mixed with rials was collected from scanning electron microscopy (SEM; Thermo
a mortar and pestle in an argon-filled glovebox (MBRAUN) with an Fisher Scientific, Phenom Pharos Desktop SEM). Inductively coupled
oxygen and water content of <0.1 ppm. The mixture was then placed plasma optical emission spectroscopy (ICP-OES) was performed with
in a quartz tube which was sealed under vacuum and heated slowly to a Thermo Fisher system (ICAP 7000) using QTEGRA software.
a target temperature in a muffle furnace (Nabertherm). After being Synchrotron radiation operando XRD was measured at the P02.1
kept at the target temperature for a few hours/days, the quartz tube beamline at DESY with a photon energy of 60 keV (photon wavelength,
was slowly cooled down or quenched in air. The target temperature ∼0.2073 Å)40. All operando experiments were tested using an in-house
for preparing P2-NaxTiS2 and P3-NaxVS2 was 750 °C, whereas that designed two-electrode operando cell with 12-mm-diameter electrodes
for O3-NaxTiS2 was 430 °C. The NaTiyV1−yS2, NaCryTi1−yS2 and NaxCrS2 as working electrodes and 12-mm-diameter metallic sodium discs as
were prepared according to previous literature methods36,37. After counter-electrodes. Mylar foil with a thickness of 23 μm was used as the
transferring the quartz tube into the glovebox, the quartz tube window material. Data were collected in transmission geometry using
was broken and the obtained sample was ground. For the synthesis the VATEX CT4343 detector at a distance of 1,150 mm. Calibration was
of layered oxide compounds, Na2O2 and CoO or NiO2 were used as pre- done using LaB6 (NIST 660c) powder. Data integration was done using
cursors, and the compounds were synthesized at high temperatures pyFAI software41. For the synchrotron radiation ex situ XRD measure-
(450 or 850 °C) in air or in an oxygen atmosphere38,39. Because the ments, all powder materials were sealed in capillaries, mounted on
samples are air- and moisture-sensitive, all the processes are carried a brass pin secured in a special holder, and loaded on a high-speed
out in an argon atmosphere unless otherwise noted, and the quartz spinner. An open-source software General Structure Analysis System-II
tubes were predried before use. (GSAS II) was used for the Rietveld and Le Bail refinement42.
Operando ECD was measured with a three-electrode ECD-3-nano
Electrochemistry cell from EL-CELL. Metallic sodium was used as counter and reference
Electrode preparation was completed in an argon-filled glovebox. electrodes, while 10-mm electrodes were used as working electrodes.
Slurries consisting of 70:20:10 (wt%) mixtures of active material, C65 About 250 μl of electrolyte was employed in each operando ECD cell.
carbon black (MTI) and polyvinylidene fluoride (MTI) were cast onto GCPL tests were performed on an SP-50 battery test station (BioLogic).
carbon-coated aluminium foil (MTI) using a 250-μm-thick doctor Single-pulse NMR spectra were acquired at room temperature on
blade (mtv messtechnik). The prepared film was then dried overnight a Bruker Avance 400 NMR spectrometer (23Na and 13C Larmor frequen-
at 60 °C under vacuum in a glass oven (BÜCHI, B585). The mass loading cies of 105.86 and 100.63 MHz, respectively). NMR parameters used
of the active material was between 0.8 and 2.6 mg cm−2. For electrolyte for magic-angle spinning (MAS) NMR data collection are shown in
preparation, sodium hexafluorophosphate (NaPF6; E-lyte, >99%) was Supplementary Table 9. Zirconia rotors (2.5 mm) were spun at 20 kHz.
used directly without drying. EC (Sigma Aldrich, 99%), DMC (Sigma NaF was used as a reference for 23Na (7.4 ppm) and adamantane powder
Aldrich, ≥99%), PC (Sigma Aldrich, 99.7%) and 2G (Sigma Aldrich, 99.5%) for 13C (29.5 ppm). Samples for MAS NMR were prepared by disas-
were dried with a mixture of 3-Å and 4-Å molecular sieves (Carl Roth) sembling the cycled coin cells in an argon-filled glovebox and filling
before use. The prepared electrolytes were stirred for at least 6 h before the rotor with the cycled material (coin cells were cycled with powder
application. materials instead of electrode sheets). The cycled material was not
Two-electrode measurements were done in coin cells (CR2032; washed before being packed into the NMR rotors.
MTI). For cell assembly, the prepared electrodes were used as the Solution-state 13C (not decoupled) NMR spectra were collected
working electrodes, while 12-mm-diameter sodium (BASF) discs were on a Bruker 500 MHz NMR spectrometer, on a 1 M NaPF6 sample in
used as the counter-electrodes. Two 16-mm-diameter glass microfibre 2G and d6-THF.
filters (Whatman, GF/A) were used as separators. Three-electrode The mass change experiments were conducted following the
measurements were performed using Swagelok-type cells with procedure developed by Åvall et al.24 Similar electrodes were cycled
sodium as both counter and reference electrodes and the prepared at constant current in coin cells to a specific potential and held for
12-mm-diameter electrodes as working electrode. All cells were assem- a minimum of 6 h. The cells were taken to an argon-filled glovebox
bled in an argon-filled glovebox with an oxygen and water content and were disassembled. Immediately after the cell was opened, the
of <0.1 ppm. Galvanostatic cycling with potential limitation (GCPL) electrodes were carefully peeled from the separator and placed in a
tests were performed with a BCS battery testing system (BioLogic) weighing boat to measure the mass on a Pioneer PX225D balance with
and a Neware battery testing system. All GCPL tests were performed an accuracy of 0.01 mg.
with two-electrode coin cells unless the use of three-electrode cells
is explicitly stated. The 1-C rate corresponds to a specific current of Simulation and theoretical calculation
0.2 A g−1. All electrochemical characterization measurements were Complete solvation shells of [Na:EC 6]+, [Na:PC6]+ and [Na:2G2]+
performed at room temperature in a temperature-controlled room. were optimized using the Gaussian 16 suite at the B3LYP/6-311-G(d,p)
level of theory, with the SMD implicit solvent model, and a subse-
Characterization quent frequency calculation was carried out to ensure the optimized
High-temperature in situ XRD was conducted on a Stoe StadiP diffrac- structure is a local minimum in the energy landscape43–47. These opti-
tometer using a Stoe ST2K furnace. The diffractometer is equipped mized solvation shells were placed in the interlayer of Na0.25TiS2 with
with monochromatic Mo Kα1 (λ = 0.709319 Å) radiation and with a an interlayer distance set to the interlayer distance found from XRD.
MYTHEN 1K detector. A mixture of the reactants in an appropriate Structural optimization was then carried out with VASP 5.2 using
ratio was sealed in a 1-mm-diameter quartz capillary, and heated in the the Perdew–Burke–Ernzerhof functional, with van der Waals disper-
ST2K furnace at a rate of 1 °C min−1. Simultaneously, the in situ XRD pat- sion correction through the Tkatchenko–Scheffler method, and a
terns were collected. Patterns were recorded in the 2θ range of 2°–37° plane-wave energy cut-off of 540 eV (refs. 48–53). The ionic positions
with a step size of 0.015° for 10 min at each temperature. The furnace and unit cell vectors were allowed to relax and the energies and charge
has a rocking motion to improve powder averaging. den­sities were computed with a 111 Γ-centred k-point mesh. Due to the

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

large number of atoms in the structures, converged structures using 40. Dippel, A.-C. et al. Beamline P02.1 at PETRA III for high-resolution
a larger number of k-points could not be obtained. Instead, one of the and high-energy powder diffraction. J. Synchrotron Radiat. 22,
solvation shells was removed and only a single layer was expanded, 675–687 (2015).
which allowed convergence of structures with a 444 Γ-centred 41. Kieffer, J., Valls, V., Blanc, N. & Hennig, C. New tools for calibrating
k-point mesh. diffraction setups. J. Synchrotron Radiat. 27, 558–566 (2020).
The starting structures of the layered sulfides (NaxTiS2 (P2), Nax- 42. Toby, B. H. & Von Dreele, R. B. GSAS-II: the genesis of a modern
TiS2 (O3), NaxVS2 (P3) and NaxCrS2 (O3)) and oxides (NaxCoO2 (O3) and open-source all-purpose crystallography software package.
NaxMnO2 (P2)) were taken from the materials database54, with x = 1 or J. Appl. Crystallogr. 46, 544–549 (2013).
0.25. Four possible starting locations for the sodium cations in NaTiS2 43. Frisch, M. J., et al. Gaussian 16 Revision C.01 (2016).
were investigated with respect to k-point sampling and energy cut-off 44. Becke, A. D. A new mixing of Hartree–Fock and local density-
(Supplementary Fig. 56). All choices of k-point sampling and cut-off functional theories. J. Chem. Phys. 98, 1372–1377 (1993).
energy show that a starting configuration where the sodium cations 45. Becke, A. D. Density-functional exchange-energy approximation
are not placed directly underneath titanium are preferable, and this with correct asymptotic behavior. Phys. Rev. A 38, 3098–3100
structure was chosen as the basic structure used for interlayer energy (1988).
calculations. Moreover, the energy converged with respect to k-point 46. Lee, C., Yang, W. & Parr, R. G. Development of the Colle–Salvetti
sampling and plane-wave energy cut-off using an 884 Γ-centred k-point correlation-energy formula into a functional of the electron
mesh with a cut-off of 500 eV. Similarly, the convergence with respect density. Phys. Rev. B 37, 785–789 (1988).
to k-point sampling and energy cut-off was tested for the expanded 47. Marenich, A. V., Cramer, C. J. & Truhlar, D. G. Universal solvation
NaTiS2 structure, yielding the same result. Consequently, an energy model based on solute electron density and on a continuum
cut-off of 500 eV and an 884 Γ-centred k-point mesh was chosen for model of the solvent defined by the bulk dielectric constant and
all further calculations. For the partially sodiated structures, the atomic surface tensions. J. Phys. Chem. B 113, 6378–6396 (2009).
shortest distance between sodium ions inside the unit cell was 48. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient
chosen because these produced structures with the lowest ener- approximation made simple. Phys. Rev. Lett. 77, 3865–3868
gies. The following VASP-recommended, projector-augmented wave (1996).
potentials where used: C, O, Na_pv, S, Ti_sv, V_sv, Cr_pv, Mn_pv, Co. 49. Kresse, G. & Furthmüller, J. Efficiency of ab-initio total energy
The interlayer binding energy EIB was computed as calculations for metals and semiconductors using a plane-wave
basis set. Comput. Mater. Sci. 6, 15–50 (1996).
Eopt − Eexp
EIB = 50. Kresse, G. & Furthmüller, J. Efficient iterative schemes for
n
ab initio total-energy calculations using a plane-wave basis set.
Phys. Rev. B 54, 11169–11186 (1996).
where Eopt is the energy of the optimized unit cell, Eexp is the energy 51. Kresse, G., Furthmüller, J. & Hafner, J. Theory of the crystal
where the interlayer distance of one layer of the optimized unit cell structures of selenium and tellurium: the effect of generalized-
was increased by 20 Å, and n is the number of atoms in the unit cell. gradient corrections to the local-density approximation.
The free volume inside the optimized structures was computed as Phys. Rev. B 50, 13181–13185 (1994).
52. Kresse, G. & Hafner, J. Ab initio molecular dynamics for liquid
⃗ − V − mV
Vfree = d√||a⃗ × b|| Na S/O
metals. Phys. Rev. B 47, 558–561 (1993).
53. Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the
projector augmented-wave method. Phys. Rev. B 59, 1758–1775
where d is the distance between the sulfur/oxygen layers, a⃗ × b⃗ is the
(1999).
cross product of the unit cell vectors a⃗ and b⃗ , √||a⃗ × b||⃗ is the area of
54. Zagorac, D., Müller, H., Ruehl, S., Zagorac, J. & Rehme, S. Recent
the parallelogram spanned by a⃗ and b⃗ , VNa is the volume of sodium based developments in the Inorganic Crystal Structure Database:
on its ionic radius (1.16 Å) and VS/O is the volume of a sulfur/oxygen atom theoretical crystal structure data and related features. J. Appl.
based on their ionic radii (1.70 Å and 1.26 Å, respectively)55; m is 1 for Crystallogr. 52, 918–925 (2019).
the fully sodiated structures and 4 for the partially sodiated structures, 55. Holleman, A. F., Wiberg, E. & Wiberg, N. in Lehrbuch der
that is, those with 0.25 sodium atoms per transition metal. anorganischen Chemie 2035–2038 (1985).
56. Sun, Y., Åvall, G. & Adelhelm, P. Solvent co-intercalation in layered
Data availability cathode active materials for sodium-ion batteries. figshare
The data that support the findings of this study are present in the https://doi.org/10.6084/m9.figshare.29143679 (2025).
paper and in the Supplementary Information. Computational data are
available via figshare at https://doi.org/10.6084/m9.figshare.29143679 Acknowledgements
(ref. 56). Source data are provided with this paper. This project received funding from the European Research Council
(ERC) under the European Union’s Horizon 2020 research and
References innovation programme (grant agreement number 864698, SEED,
36. Wang, T. et al. Anionic redox reaction in layered NaCr2/3Ti1/3S2 P.A.) and over the joint research group on operando battery analysis
through electron holes formation and dimerization of S–S. between Humboldt-University Berlin and Helmholtz-Zentrum
Nat. Commun. 10, 4458 (2019). Berlin (CE-GOBA). The ICP-OES measurements were carried out by
37. Wang, T. et al. Anomalous redox features induced by strong the Solar Fuel Testing Facility laboratory of the Helmholtz Energy
covalency in layered NaTi1−yVyS2 cathodes for Na-ion batteries. Materials Foundry (HEMF). We thank P. Bogdanoff and U. Michalczik
Angew. Chem. Int. Ed. 61, e202205444 (2022). for their support. We acknowledge DESY, a member of the Helmholtz
38. Bianchini, M. et al. The interplay between thermodynamics and Association HGF, for the provision of experimental facilities. Parts
kinetics in the solid-state synthesis of layered oxides. Nat. Mater. of this research were carried out at beamline P02.1 at PETRA III.
19, 1088–1095 (2020). Beamtime was allocated for proposals I-20221303 (Y.S., P.A.) and
39. Vassilaras, P., Ma, X., Li, X. & Ceder, G. Electrochemical properties I-20230377 (Y.S., P.A.). H.W. and K.A.M. acknowledge DESY, a member
of monoclinic NaNiO2. J. Electrochem. Soc. 160, A207–A211 of the Helmholtz Association HGF, for support with travel costs.
(2013). The authors gratefully acknowledge the computing time granted

Nature Materials
Article https://doi.org/10.1038/s41563-025-02287-7

by the North-German Supercomputing Alliance (HLRN, P.A., G.Å.), Funding

financial support from the Dutch Research Council (NWO) for the Open access funding provided by Humboldt-Universität zu Berlin.
ECCM Tenure Track funding under project number ECCM.TT.ECCM.006
(P.B.G.), and the German Federal Ministry of Education and Research Competing interests
(grant reference number 03SF0565A, P.B.G.). The authors thank The authors declare no competing interests.
K. Scheurell for assistance with the solid-state NMR measurements.
Additional information
Author contributions Supplementary information The online version
P.A. and Y.S. conceived and designed the project. P.A., Y.S. and G.Å. contains supplementary material available at
developed the concept and planned the experiments. P.A., Y.S., G.Å. https://doi.org/10.1038/s41563-025-02287-7.
and G.A.F. contributed to research discussions and the development
of ideas. Y.S. synthesized, characterized and electrochemically tested Correspondence and requests for materials should be addressed to
the investigated compounds with support from S.-H.W., G.A.F. and Yanan Sun or Philipp Adelhelm.
H.W. G.Å. carried out structural simulations and DFT calculations.
A.F. and P.B.G. performed and analysed the NMR measurements. Peer review information Nature Materials thanks Simon Fleischmann
Synchrotron XRD measurements were conducted by Y.S., K.A.M. and and the other, anonymous, reviewer(s) for their contribution to the
H.W., with assistance from V.B. M.B. performed the high-temperature peer review of this work.
in situ XRD measurements. S.R. designed the operando test cell. The
manuscript was written by Y.S., G.Å. and P.A. All authors discussed the Reprints and permissions information is available at
results and provided feedback on the manuscript. www.nature.com/reprints.

Nature Materials