Reversing the Chemical and Structural Changes of Prussian White After Exposure to Humidity to Enable Aqueous Electrode Processing for Sodium-ion Batteries

The accelerated materials exposure test used in this study was first described by Sicklinger et al. for lithium-ion layered oxide positive electrode materials. ²⁹ Briefly, Prussian White with different sodium content (Na_1.3MnHCF and Na_1.8MnHCF) was stored for up to one week over an open water bath at 25 °C, located in a vessel that contained a lid with a small hole to allow intrusion of ambient air. This resulted in a controlled atmosphere with high relative humidity (RH ≈ 100%) and ∼400 ppm CO₂. Materials referred to as "washed" were immersed in purified water at a 1:5 by weight PW:H₂O ratio and stirred with a magnetic stirrer for 20 min Afterwards, the washed PW was vacuum-filtered. Na_1.8MnHCF powder undergoing this procedure changes its color from white after chemical sodiation to dark blue after washing and back to white after drying. Exposed samples were dried in a heated antechamber for at least 4 h at 70 °C under dynamic vacuum (∼10⁻⁴ mbar) to remove physisorbed H₂O, and subsequently stored in an Ar-filled glove box without exposure to ambient air after drying. Note that we did not observe any Na⁺/H⁺-exchange during the washing procedure. The water solution did not change its pH and the washed and reheated material did not show a reduced first charge capacity (see coin cell experiments). A summary of the various samples is given in Table I.

Table I. Prussian White materials used in this study.

Sample	Stoichiometry	Treatment	Duration
pristine Na1.3MnHCF	Na1.3Mn0.8Fe0.2[Fe(CN)6]0.90H2O	vacuum-dried at 150 °C	14 h
stored Na1.3MnHCF	Na1.3Mn0.8Fe0.2[Fe(CN)6]0.9zH2O	storage at ∼100% RH and ∼400 ppm CO2	1 h, 3 h, 6 h, 12 h, 24 h, 48 h, 1 week
pristine Na1.8MnHCF	Na1.8Mn0.8Fe0.2[Fe(CN)6]0.90H2O	vacuum-dried at 150 °C	14 h
stored Na1.8MnHCF	Na1.8Mn0.8Fe0.2[Fe(CN)6]0.9zH2O	storage at ∼100% RH and ∼400 ppm CO2	1 h, 3 h, 6 h, 24 h, 48 h, 1 week
washed Na1.8MnHCF	Na1.8Mn0.8Fe0.2[Fe(CN)6]0.9zH2O	immersed in water (1:5 by weight PW:H2O)	20 min

Pristine and stored PW samples were analyzed on a SDT Q600 TGA system (TA Instruments, USA). First a conditioning step at 25 °C for 15 min with an argon flow rate of 100 ml min⁻¹ was done to remove any air introduced in the TGA sample chamber during the transfer. Afterwards the argon flow rate was reduced to 20 ml min⁻¹, the temperature was increased from 25 to 190 °C with a constant heat ramp rate of 5 K min⁻¹ and the final temperature of 190 °C was held for 15 min Beyond this temperature hydrogen cyanide (HCN) can be released from the decomposition of the MnHCF framework. ^19,30 To quench any HCN, the TGA carrier gas stream was diverted to a sodium hypochlorite wash bottle instead of being sent to a mass spectrometer.

FTIR spectra were collected using a Cary 630 FTIR spectrometer (Agilent Technologies) equipped with a germanium crystal attenuated total reflectance (ATR) accessory. Sixteen scans were collected for each background and sample measurement at a resolution of 1 cm⁻¹ using the MicroLab PC software. All measurements were performed inside an Ar-filled glovebox to minimize exposure to air.

XRD patterns of pristine and storage PW powders were collected with a Bruker D8 diffractometer equipped with a Cu target X-ray tube and a diffracted beam monochromator. Samples were measured in the scattering angle (2θ) range of 10°–70° with 0.02° steps. Air-sensitive samples that were previously dried or only exposed for a controlled amount of time were measured in a custom-made air-tight XRD-sample holder with an acquisition time of 20 s/step to improve the signal-to-noise ratio (total measurement time of 17 h/sample). ^31,32 All other samples were measured in a regular XRD-sample holder with a shorter acquisition time of 3 s/step (total measurement time 2.5 h sample).

Positive electrodes were made from PW in various conditions (see Table I) mixed with C65 carbon black and PVDF at a weight ratio of 90:5:5 in NMP. Negative electrodes were made from hard carbon (HC, Kuranode, Type II, BET surface area 9 m²/g, Kuraray, Japan) mixed with C65 and PVDF at a weight ratio of 94:1:5 in NMP. H₂O-based positive electrodes were made from 90% PW mixed with 5 wt% C65 conductive carbon and a mixture of 2.5 wt% carboxymethyl cellulose (CMC) and 2.5 wt% sodium polyacrylate (NaPAA, Sigma-Aldrich, USA) binders. H₂O-based negative electrodes were made from 94.5% hard carbon mixed with 1 wt% C65 conductive carbon and a mixture of 2.5 wt% CMC and 2 wt% styrene-butadiene rubber (SBR) binders. Before making the H₂O-based positive electrodes, the voltage stability of the various binders was tested by cyclic voltammetry (see below). For that, carbon electrodes consisting of 75 wt% C65 and 25 wt% of either CMC, SBR or NaPAA were made. The slurry viscosities were adjusted by adding appropriate amounts of water. For reference, an NMP-based carbon electrode with 75 wt% C65 and 25 wt% PVDF was made. The slurries were cast onto aluminum foil with a 150 μm notch bar spreader as described by Marks et al. ³³ The electrodes were dried in a convection oven at 120 °C for 3 h. Dried electrodes were then calendared at a pressure of 2000 atm, positive electrodes were punched with a diameter of 11.25 mm, and negative electrodes with a diameter of 12.75 mm. Before coin cell fabrication, the electrodes were further dried under vacuum (∼10⁻⁴ mbar) for 14 h at 150 °C (for materials referred to as "reheated") or 120 °C (for all other materials). After drying, the positive active material loadings were 10–12 mg cm⁻², negative active material loadings were 6–7 mg/cm².

Galvanostatic cycling is carried out in 2325-type coin cells in a temperature-controlled box at 30 °C using battery testers from Neware (Shenzhen, China). Half cells were assembled with a PW positive electrode, two pieces of Celgard 3501 as the separators with a PP-mesh between them, and metallic sodium foil as the counter-electrode. These cells were filled with 50 mg of 1.5 M NaPF₆ in FEC:DMC (1:4 by weight) electrolyte (Shenzhen Capchem, China). Full cells were assembled with a PW positive electrode, one piece of BMF (blown microfiber, 3 M, Canada) as the separator, and a HC negative electrode. These cells were filled with 50 mg of 1.5 M NaPF₆ in EC:DMC (1:4 by weight) electrolyte + 4 wt% vinylene carbonate (VC) additive. After assembly, all cells rested for 1 h prior to charge/discharge cycling in order to allow complete wetting; C-rates refer to the nominal specific capacity of 170 mAh/g for sodiated PW. Cells were formed at C/20 between 2 and 4 V for half cells and 1.5 and 4 V for full cells. Half cells were cycled in constant current constant voltage (CCCV) mode with a CV hold at top of charge to a C/50 current cut-off, full cells had an additional CV step on discharge to prevent the accumulation of sodium in the overhang region. After formation, cells were cycled with a similar protocol, but at C/5 rate with C/20 current cut-off in the CV steps and C/20 checkup cycles every 50 cycles.

Cyclic voltammograms were recorded to test the voltage stability of various binders (see above) in 2325-type coin cells with 12.75 mm diameter carbon working-electrodes, two pieces of Celgard 3501 as the separators with a PP-mesh between them, and metallic sodium foil as the counter-electrode. Three subsequent CV cycles were measured with a VMP3 potentiostat (Bio-Logic, France) at 0.1 mV s⁻¹ and 25 °C between 3 and 4.5 V vs Na⁺/Na.

Figure 1a shows the relative mass increase of partially sodiated Na_1.3MnHCF (black) and fully sodiated Na_1.8MnHCF (yellow) during an accelerated materials exposure test at 100% RH and 25 °C in air. The right y-axis shows the number, z, of water molecules per formula unit calculated from the relative mass increase. For reference, the dashed lines show the mass increase for equal numbers of sodium and water per formula unit (z = x). Both materials were vacuum-dried at 150 °C for 14 h prior to storage to make sure they are mostly dehydrated (z ≈ 0).

Zoom In Zoom Out Reset image size

Over the first day of storage, a rapid mass increase is observed for, both, Na_1.3MnHCF and Na_1.8MnHCF. After that, the mass of the Na_1.3MnHCF sample plateaus at 7.8% increase relative to its starting mass at the 24 h mark and only increases slightly to 8.2% after 1 week of storage. Similarly, the Na_1.8MnHCF sample plateaus at a mass increase of 18.6% after 24 h and increases slightly to 19.5% after 1 week. This data shows a clear correlation between the sodium content of PW and the mass increase during storage at high relative humidity, with the fully sodiated material exhibiting more than twice the water uptake of the partially sodiated material. Interestingly, for Na_1.3MnHCF, the number of water molecules per formula unit (z = 1.3) calculated from the mass increase is close to the number of sodium atoms (x = 1.3). However, for Na_1.8MnHCF this correlation does not hold, and the mass increase would suggest ∼3 H₂O per formula unit. The observed mass increase of almost 20% for Na_1.8MnHCF is significantly larger than the mass increase by <1% for LiNi_0.8Mn_0.1Co_0.1O₂ in an identical exposure test. ²⁶ Other PW materials like Na_1.80Fe[Fe(CN)₆] showed a similar mass increase when stored at high RH. ²⁰ Such large mass increase indicated the presence of interstitial water in PW after storage, which has been shown to lead to structural changes. ¹⁹

Figure 1b shows pictures of the Na_1.8MnHCF powder after exposure to 100% RH and 25 °C in ambient air for 5 min, 1 h, 3 h and 24 h. The material is white in its dehydrated state but gradually changes to dark blue the longer the exposure time. A timelapse of the first 24 h of storage can be found in the Supplementary Information showing a clear color change of PW powder from white to dark blue associated with the uptake of interstitial water.

Figure 2 shows the FTIR transmission spectra of pristine Na_1.8MnHCF (black) and Na_1.8MnHCF stored at 100% RH and 25 °C for 1 h, 3 h, 24 h, 48 h and 1 week (168 h, yellow). The FTIR measurements were performed on Na_1.8MnHCF since it showed more mass increase than partially sodiated PW (see Fig. 1) and would likely show more IR active surface species. The vertical lines in Fig. 2 mark the IR absorption peaks associated with the CN⁻-bridges in MnHCF at 2050 cm⁻¹, the OH⁻-vibration of surface hydroxide, adsorbed H₂O, or interstitial H₂O at 1620 cm⁻¹, and the CO₃ ²⁻-vibration at 1300 cm⁻¹ (main carbonate group vibration mode, side peaks of the asymmetric vibrations are marked in light grey at 1460 cm⁻¹ and 1780 cm⁻¹). ^19,34–39 It has to be noted that it is not possible to differentiate surface impurities containing OH⁻-groups from interstitial H₂O inside of the crystal structure or from H₂O adsorbed on the PW surface with FTIR as the vibrational modes of these species are very similar.

Zoom In Zoom Out Reset image size

Pristine Na_1.8MnHCF shows only one absorption peak at 2050 cm⁻¹, which can be assigned to the CN⁻-bridge vibration of PW. ^19,34–36 No other peaks are visible indicating the absence of IR active surface impurities, e.g., carbonates and hydroxides. PW stored for 1 h shows a slight shift of the CN⁻ absorption band to a higher wavenumber of 2060 cm⁻¹, which will be discussed later. Furthermore, this sample shows a significant peak at 1620 cm⁻¹, which can be assigned to the OH⁻-vibration in surface hydroxide or H₂O. ^19,37–39 Storing the material for 3 h leads to a large OH⁻ peak at 1620 cm⁻¹. Interestingly, after 24 h of storage, the OH⁻-feature at 1620 cm⁻¹ decreased and new absorption bands appeared at 1300 cm⁻¹,1470 cm⁻¹ and 1780 cm⁻¹, which can be assigned to CO₃ ²⁻. ^29,40–42 This could indicate that the OH⁻-feature mostly stems from surface hydroxide which converts into carbonate by reaction with atmospheric CO₂ during prolonged storage. ²⁹ The cation associated with these anions is likely Na⁺, which is removed from the PW surface during storage. Note that these samples were heated to 70 °C in a vacuum chamber before bringing them into a glovebox for FTIR measurements, which can explain the absence of adsorbed H₂O, but would likely not affect surface hydroxide or carbonate. PW stored for 48 h shows more pronounced CO₃ ²⁻ peaks than the 24 h sample. The sample stored for 1 week shows similar CO₃ ²⁻ peaks and a strong pronounced OH⁻-absorption band from H₂O or hydroxide.

Figure 3 shows the powder X-ray diffractograms of pristine Na_1.3MnHCF and Na_1.8MnHCF (black), as well as Na_1.8MnHCF stored at 100% RH and 25 °C for 1 h, 3 h, 24 h and 1 week (168 h, yellow). Vertical lines indicate the characteristic peaks for the monoclinic/cubic phase of PW at 16.9° and the rhombohedral phase of PW at 18.1°.

Zoom In Zoom Out Reset image size

Pristine Na_1.3MnHCF shows a cubic structure in Fig. 3, which is in agreement with Song et al. ¹⁹ who showed that partially sodiated PW exhibits a cubic crystal structure in, both, hydrated and dehydrated state. In contrast, fully sodiated Na_1.8MnHCF was shown to exhibit a rhombohedral structure in its dehydrated state, which converts into a monoclinic structure upon water uptake. ¹⁹ Indeed, pristine Na_1.8MnHCF shows a rhombohedral phase in Fig. 3, but after only 1 h of storage a peak at 16.9° along with other smaller peaks indicates the presence of a partial monoclinic phase. After 3 h of storage the peak at 16.9° increased in intensity and some peak splitting is observed at higher scattering angles indicating the presence of, both, a rhombohedral and monoclinic phase. After 24 h of storage, Na_1.8MnHCF completely converted into a monoclinic phase. The cubic and monoclinic phase of PW have a similar diffraction pattern, which makes it difficult to differentiate them with our in-house XRD. However, it can be observed that the monoclinic phase has slightly broader peaks than the cubic phase and synchrotron XRD work by Song et al. ¹⁹ has confirmed that fully sodiated PW exhibits a monoclinic crystal structure. After 1 week of storage, the material is still in its monoclinic phase with no change compared to the 24 h spectrum, indicating that the structural change is completed after at least 24 h.

Figure 4 shows the normalized mass loss (a) and the derivative of the mass loss (b) as a function of temperature during a TGA experiment with pristine Na_1.8MnHCF (black), Na_1.8MnHCF stored at 100% RH and 25 °C for 6 h (purple), and 1 week (yellow), as well as Na_1.8MnHCF after immersion in water for 20 min (blue) and subsequent filtering and drying at 70 °C. The TGA mass loss data can be compared to the mass increase shown in Fig. 1 and allows to probe if interstitial water can be removed by heat treatment.

Zoom In Zoom Out Reset image size

Pristine Na_1.8MnHCF shows only a small mass loss of 3% up to 150 °C, which likely stems from adsorption of moisture during sample transfer to the TGA machine. After 6 h of storage Na_1.8MnHCF shows a mass loss of 6.3% at 150 °C, which fits well with the mass increase of 5.7% after 6 h in Fig. 1, suggesting that physisorbed and interstitial water can be removed by heating to 150 °C. Na_1.8MnHCF stored for 1 week shows 15.9% mass loss at 150 °C, which is again in reasonable agreement with the mass increase of 19.5% observed in Fig. 1. The difference could be explained by loss of adsorbed water during the 70 °C drying prior to the TGA experiment (see Experimental section) and the fact that the carbonate surface impurities formed during storage (see Fig. 2) would not be removed at 150 °C. Immersing Na_1.8MnHCF in water for 20 min allows to deconvolute the combined effect of moisture and atmospheric CO₂ from simple exposure to water during aqueous electrode processing. The washed sample shows 10% mass loss at 150 °C, which is smaller than the 15.9% observed for the sample stored for 1 week, potentially due to the significantly shorter exposure time. Interestingly, the washed material shows a plateau in the mass loss curve at 150 °C, which corresponds to a minimum in the derivative of the mass loss curve. Thus, the removal of physisorbed and interstitial water typically observed in this temperature range seems to be completed at 150 °C for the immersed sample, ^19,30 making it a promising drying temperature for PW electrodes made in an aqueous slurry process.

Heating PW to 190 °C-200 °C would lead to a sharp weight loss due to the decomposition of the MnHCF framework. ^19,30 These temperatures need to be avoided in a practical drying procedure, since the material would release toxic HCN gas. ^19,30 However, since the weight loss does not completely cease beyond 150 °C, one could explore drying temperatures between 150 and 200 °C or drying at higher vacuum, which will be part of future studies.

Figure 5 shows the FTIR spectra of pristine Na_1.8MnHCF (black), Na_1.8MnHCF stored at 100% RH and 25 °C for 1 week (168 h, yellow), the same material reheated at 150 °C for 14 h in vacuum (red), as well as Na_1.8MnHCF washed in water (blue) and reheated at 150 °C for 14 h in vacuum (light blue). Note that pristine and 1 week stored Na_1.8MnHCF were already shown in Fig. 2.

Zoom In Zoom Out Reset image size

Material that was stored for 1 week and subsequently reheated shows a significant decrease in the OH⁻ absorption band at 1620 cm⁻¹, which can be interpreted as a removal of adsorbed and interstitial water. The small remaining peak at 1620 cm⁻¹ could be from residual NaOH, which would only decompose above 318 °C. ⁴³ There are no changes to the CO₃ ²⁻ bands upon heating since surface carbonates would only decompose at much higher temperatures, e.g., 900 °C in the case of Na₂CO₃. ⁴⁴ Interestingly, washed Na_1.8MnHCF shows a somewhat smaller OH⁻ absorption band than the sample stored for 1 week and no CO₃ ²⁻ bands. Although semi-quantitative, the fact that less interstitial water and fewer surface species are observed for the washed samples compared to the sample stored for 1 week is consistent with the smaller weight loss in Fig. 4. After heating the washed sample to 150 °C, the OH⁻ band is completely gone and the sample looks identical to pristine Na_1.8MnHCF. This is further proof that vacuum-drying of washed PW at 150 °C can remove absorb and interstitial water. It also suggests that no hydroxide or carbonate is formed during washing, but rather adsorbed water and interstitial water is found. This means that Na⁺/H⁺ exchange needed for the formation of NaOH or Na₂CO₃ is not observed during the short 20 min immersion of PW in water. The absence of NaOH and Na₂CO₃ could also be explained by the immediate dissolution of these surface species during washing. However, as we will see later, reheated PW after washing does not exhibit a cubic phase (see Fig. 6), nor does it have reduced charge capacity (see Fig. 8). Thus, there is no significant Na loss during washing.

Zoom In Zoom Out Reset image size

The inset in Fig. 5 highlights the shift of the CN⁻ absorption band of PW to a higher wavenumber of 2070 cm⁻¹ upon washing. After drying the washed sample, the CN⁻ band shifts back to 2050 cm⁻¹ and overlaps with that of pristine material. This shift is very similar to the one observed by operando FTIR spectroscopy of the desodiation process of Prussian blue analogs. ³⁴ It has been shown that interstitial H₂O molecules are located close to intercalated Na⁺-ions, which could mean that H₂O shields the electrostatic interaction between Na⁺-ions and the CN⁻-bridges and make the CN⁻ vibration mode similar to the one observed in desodiated PW. ^19,45 At any rate, the complete reversal of the CN⁻ shift is further proof that all interstitial H₂O has been removed from the material.

Figure 6 shows the XRD patterns of pristine Na_1.8MnHCF (black), Na_1.8MnHCF stored at 100% RH and 25 °C for 1 week (168 h, yellow), the same material reheated at 150 °C for 14 h in vacuum (red), as well as Na_1.8MnHCF washed in water (blue) and reheated at 150 °C for 14 h in vacuum (light blue). Note that pristine and 1 week stored Na_1.8MnHCF were already shown in Fig. 3.

Material that was stored for 1 week and subsequently reheated shows a change from a monoclinic to a rhombohedral crystal structure. Apart from a small peak at 16.9° and slight peak splitting at higher angles, the diffraction pattern for reheated material resembles the one of pristine Na_1.8MnHCF. After 20 min of washing, Na_1.8MnHCF shows a monoclinic structure, which resembles the crystal structure of material stored for 1 week, indicating that both, storage at high RH and immersion in H₂O, lead to the same phase change from rhombohedral to monoclinic. Upon heating the washed material, the phase change can be fully reversed and a rhombohedral crystal structure is obtained that is identical to the one of pristine Na_1.8MnHCF.

Combining the information from FTIR (see Fig. 5) and XRD (see Fig. 6), it can be concluded that PW absorbs water in its crystal structure regardless if it is water vapor during ambient storage or liquid water during aqueous electrode processing. Furthermore, the interstitial water can be easily removed by heating the material to 150 °C in vacuum. Unfortunately, hydroxide and carbonate surface species formed during ambient storage cannot be removed by heating the material to 150 °C. However, these surface contaminants do not form during ∼20 min immersion in water, which suggests that they would also not form during brief aqueous electrode processing. Thus, vacuum drying of PW electrodes should fully reverse any chemical and structural changes experienced by the material during immersion in water and enable a low-cost, environmentally friendly electrode manufacturing process. It should be noted, however, that slurry mixing of large batches in industry would take longer than 20 min, which warrants future studies of long-term water exposure of PW.

Appropriate binders have to be selected for aqueous PW electrodes that provide electrode cohesion and adhesion to the Al current collector. In particular, the voltage stability of the polymer binders needs to be tested since PW/HC full cells are charged to 4.0 V, which means the polymer binder in the positive electrode experiences ∼4.1 V vs Na⁺/Na, which would be equivalent to ∼4.4 V vs Li⁺/Li. Graphite negative electrodes and LiFePO₄ positive electrodes for lithium-ion batteries can be made in an aqueous slurry process. ^46,47 A typical binder system for aqueous graphite negative electrodes is CMC combined with SBR. ⁴⁸ For aqueous LiFePO₄ positive electrodes, combinations of CMC, SBR, PTFE, PAA and other binders have been explored. ^49,50 LiFePO₄/graphite cells typically charge to 3.65 V, ⁵¹ which is significantly lower than the voltage experienced by binders in PW positive electrodes.

Figure 7 shows CVs of C65 carbon electrodes prepared from a H₂O-based slurry with either 25 wt% CMC (red), SBR (blue), or NaPAA (light blue), as well as carbon electrodes made from an NMP-based slurry with 25 wt% PVDF (black dashed line) for reference. The CVs were recorded in 2325 coin cells with a Na metal counter-electrode to test the voltage stability of the three different binders in a range from 3.0 to 4.5 V vs Na⁺/Na.

Zoom In Zoom Out Reset image size

The PVDF-bonded carbon electrode shows minimal current at 4.1 V vs Na⁺/Na (max. positive electrode voltage in PW/HC full cells) and ∼0.1 mA m⁻² carbon surface area normalized current density at 4.5 V vs Na⁺/Na (≈4.8 V vs Li⁺/Li), which is in agreement with Jung et al. ⁵² Similarly small currents are observed with CMC-bonded carbon electrodes indicating high voltage stability of this binder. SBR-bonded carbon electrodes show 0.2 mA m⁻² at 4.1 V vs Na⁺/Na and >1.3 mA cm⁻² at 4.5 V vs Na⁺/Na, highlighting the susceptibility of this polymer to oxidative decomposition. A possible reason for the voltage instability of SBR could be the presence of carbon-carbon double bonds in the polymer chain from which electrons can be withdrawn. This is consistent with a report by Yabuuchi et al., which shows that the unsaturated C=C of butadiene chains in SBR is irreversibly oxidized at voltages above 4.2 V. ⁵³ In addition to cyclic voltammetry, the authors use a chemical staining technique by osmium tetraoxide to proof that the double bonds of the butadiene unit in SBR can be electrooxidized. ⁵³ SBR is typically used in combination with CMC to improve cohesion of graphite negative electrodes, ^48,54 but the data in Fig. 7 suggests that it cannot be used at the high voltages experience in PW positive electrode. An alternative to SBR with better voltage stability could be NaPAA. NaPAA-bonded carbon electrodes show similarly low currents as electrodes with CMC up to 4.1 V vs Na⁺/Na making it a suitable binder for combination with CMC in aqueous PW electrodes.

Figure 8 shows the first cycle voltage profiles and differential capacity vs voltage plots of PW half cells with electrodes made from (a) pristine Na_1.8MnHCF (black) and Na_1.8MnHCF stored for 1 week at 100% RH and 25 °C (yellow), as well as (b) Na_1.8MnHCF stored for 1 week at 100% RH and 25 °C heated to 150 °C before slurry making (red) and Na_1.8MnHCF electrodes made from a H₂O-based slurry with CMC/NaPAA binder dried at 150 °C in vacuum (light blue).

Zoom In Zoom Out Reset image size

The voltage profile of pristine Na_1.8MnHCF shows a single plateau during charge and discharge at 3.5 V and 3.35 V vs Na⁺/Na, respectively (see panel a-1). Accordingly, only a single dQ/dV peak is visible for this material (see panel a-2). This is in agreement with Song et al. ¹⁹ who showed that dehydrated rhombohedral PW has a single voltage plateau. ¹⁹ Pristine Na_1.8MnHCF achieves 150 mAh g⁻¹ discharge capacity (theoretical capacity 170 mAh g⁻¹) and only shows a small irreversible capacity of 2 mAh g⁻¹. After 1 week of storage, Na_1.8MnHCF shows a step in its voltage profile and two distinct dQ/dV peaks centered around 3.4 V and 3.65 V vs Na⁺/Na. The first cycle discharge capacity is reduced to 140 mAh g⁻¹. Electrodes from this material have been dried at 120 °C, which means that residual interstitial water is likely remaining in the material leading to the characteristic voltage step of hydrated PW and the reduced specific capacity (7 wt% H₂O would be sufficient to explain the observed 10 mAh g⁻¹ reduction in specific capacity, due to the mass increase of the electrode). ¹⁹ In contrast, 1 week stored Na_1.8MnHCF reheated to 150 °C (see Fig. 8b) shows only a single voltage plateau and dQ/dV peak at 3.5 V vs Na⁺/Na characteristic of dehydrated material and a relatively high first cycle discharge capacity of 157 mAh g⁻¹, which is better than pristine material heated to 120 °C. The improved performance of reheated PW is in agreement with the removal of interstitial water shown by FTIR and the rhombohedral phase found by XRD. Surface hydroxide and carbonate which remained on the exposed material after reheating (see Fig. 5) may account for the difference between experimental and theoretical capacity. Finally, Na_1.8MnHCF processed in H₂O with CMC/NaPAA binder, which was subsequently vacuum-dried at 150 °C shows a single voltage plateau and dQ/dV peak at 3.5 V vs Na⁺/Na confirming that it is mostly dehydrated even after immersion in H₂O. Its specific discharge capacity of 155 mAh g⁻¹ is similar to PW dehydrated after storage and the voltage profile as well as dQ/dV features show no indication of CMC/NaPAA binder instability.

Figure 9 shows voltage profiles, differential capacity and capacity retention data for PW/HC full cells with both, the positive and negative electrode processed in H₂O (light blue), and both electrodes processed in NMP (black). The H₂O-based slurries had CMC/NaPAA binder for the positive electrode and CMC/SBR binder for the negative electrode. The NMP-based slurries had PVDF binder for both electrodes. Aqueous PW electrodes were vacuum-dried at 150 °C, all other electrodes were vacuum-dried at 120 °C. These full cells will allow to compare the capacity retention of NMP and H₂O-processed PW in the absence of sodium-metal, which is prone to dendrite formation and low coulombic efficiency over extended cycling. ^55,56

Zoom In Zoom Out Reset image size

The voltage profiles of both full cell types differ from the half cell voltage profile, due to the characteristic voltage of HC upon sodium insertion and removal, with a sloped high voltage region and a low voltage plateau around 0.15 V vs Na⁺/Na. ^57,58 Full cells with NMP-based electrodes show a first charge capacity of 154 mAh g⁻¹ (see Fig. 8a), which is similar to half cells with pristine Na_1.8MnHCF (see Figs. 7a–71). Upon discharge, however, a high irreversible capacity of 52 mAh g⁻¹ is observed leading to a specific discharge capacity of 102 mAh g⁻¹. Note that the NP-ratio of 1.13 in these cells was not optimized and future work will focus on improving the first cycle efficiency of HC, e.g., by selecting appropriate electrolyte additives. Full cells with H₂O-based electrodes achieve a slightly lower first charge capacity of 149 mAh g⁻¹, but also a smaller irreversible capacity loss of 34 mAh g⁻¹, resulting in a higher specific discharge capacity of 115 mAh g⁻¹. The improved first cycle efficiency could be related to a difference in surface groups on hard carbon immersed in water compared to hard carbon that reacted with moisture, N₂, O₂, and CO₂ in the air. ^59–61 Furthermore, Dahbi et al. found an improvement of the first cycle efficiency and better cycling performance if a water-based CMC/SBR binder was used for hard carbon anodes instead of PVDF binder. ⁶² The dQ/dV vs V curves in Fig. 9b show that both cells have their main dQ/dV peak at ∼3.3 V. The absence of a second peak towards higher voltage suggests that both PW electrode materials are dehydrated.

Figure 9c shows the capacity retention of full cells with NMP- and H₂O-based electrodes. The former shows a specific discharge capacity of 102 mAh g⁻¹ at C/20 and 97 mAh g⁻¹ at C/5 (error bars indicate standard deviation of pair cells). The latter show a slightly higher specific discharge capacity of 110 mAh g⁻¹ at C/20 and 103 mAh g⁻¹ at C/5 due to their higher first cycle efficiency (see Fig. 9a). Both cells show a similar capacity retention of 75% after 70 cycles with unoptimized carbonate electrolyte. While this capacity retention is clearly not sufficient for practical application, it shows that full cells with H₂O-based PW electrodes are feasible if chemical and structural changes due to water uptake are reversed by effective drying. It should be pointed out that the manganese hexacyanoferrate material used in this study can cycle 40,000 times at 25 °C to 80% of its starting capacity when couple with a manganese hexacyanomanganate negative electrode. ²⁷ Also, CATL announced a long cycle life sodium-ion battery that is based on Prussian White and hard carbon. ⁶³ This underlines the opportunity for careful electrolyte optimization in order to improve the cycle life of sodium-ion full cells with Prussian White positive electrodes and hard carbon negative electrodes.