Article https://doi.org/10.1038/s41467-022-35376-1 Lithiated Prussian blue analogues as positive electrode active materials for stable non-aqueous lithium-ion batteries Ziheng Zhang1,2, Maxim Avdeev3, Huaican Chen4,5, Wen Yin4,5, Wang Hay Kan 4,5 & Guang He 1,6 Prussian blue analogues (PBAs) are appealing active materials for post-lithium electrochemical energy storage. However, PBAs are not generally suitable for non-aqueous Li-ion storage due to their instability upon prolonged cycling. Herein, we assess the feasibility of PBAs with various lithium content for non- aqueous Li-ion storage. We determine the crystal structure of the lithiated PBAs via neutron powder diffraction measurements and investigate the influence of water on structural stability and Li-ion migration through oper- ando X-ray diffraction measurements and bond valence simulations. Further- more, we demonstrate that a positive electrode containing Li2-xFeFe(CN)6⋅nH2O (0 ≤ x ≤ 2) active material coupled with a Li metal elec- trode and a LiPF6-containing organic-based electrolyte in coin cell configura- tion delivers an initial discharge capacity of 142 mAh g−1 at 19mAg−1 and a discharge capacity retention of 80.7% after 1000 cycles at 1.9 A g−1. By repla- cing the lithiummetal with a graphite-based negative electrode, we also report a coin cell capable of cycling for more than 370 cycles at 190mAg−1 with a stable discharge capacity of about 105 mAh g−1 and a discharge capacity retention of 98% at 25 °C. Non-aqueous lithium-ion batteries (LIBs) have become a dominant power source for portal electronic devices, power tools, electric vehicles, and other renewable energy storage systems1. Albeit its popularity, the thermal runaway induced accidents are occasionally happened all over the world2,3. In commercialized lithium-ion bat- teries, the layered transition-metal (TM) oxides, represented by a general formula of LiMO2, have been widely used as higher energy density positive electrode materials due to their appealing elec- trochemical performance namely the specific gravimetric capacity, rate capacity and energy density4,5. However, the inadequate ther- mal stability of the cell at charge states seems inevitable for such layered structures due to their O 2p and M 3d orbital overlapping. Consequently, prolonged cycling might induce undesired struc- tural defects of cationic migration and oxide-ion vacancy. Alter- natively, polyanion-type cathodes such as LiFePO4 demonstrate greatly promoted thermal stability due to the strong covalent P–O bonds in the structures6–8. Furthermore, the inductive effect (an electronic effect due to the polarisation of σ bonds within a mole- cule or ion) brings a higher working potential in polyanion cathodes in comparison to layered lithium metal oxide cathodes. None- theless most polyanion cathodes suffer from lower theoretical capacities as compared to their layered counterparts. It’s been a Received: 14 November 2021 Accepted: 29 November 2022 Check for updates 1Tianjin Key Laboratory of Advanced Functional Porous Materials, Institute for New Energy Materials and Low-Carbon Technologies, School of Materials Science and Engineering, Tianjin University of Technology, Tianjin 300384, China. 2Renewable Energy Conversion and Storage Center (RECAST), Haihe Laboratory of Sustainable Chemical Transformations, Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), College of Chemistry, Nankai University, Tianjin 300071, China. 3AustralianNuclear Science and TechnologyOrganization (ANSTO), LucasHeights, NSW2234, Australia. 4Spallation Neutron Source Science Center, Dalang, Dongguan 523803, China. 5Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China. 6Tianneng Co. Ltd, Huzhou 313100, China. e-mail: jianhx@ihep.ac.cn; heguang@tjut.edu.cn Nature Communications | (2022) 13:7790 1 12 34 56 78 9 0 () :,; 12 34 56 78 9 0 () :,; http://orcid.org/0000-0002-1663-2999 http://orcid.org/0000-0002-1663-2999 http://orcid.org/0000-0002-1663-2999 http://orcid.org/0000-0002-1663-2999 http://orcid.org/0000-0002-1663-2999 http://orcid.org/0000-0003-4167-053X http://orcid.org/0000-0003-4167-053X http://orcid.org/0000-0003-4167-053X http://orcid.org/0000-0003-4167-053X http://orcid.org/0000-0003-4167-053X http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-35376-1&domain=pdf http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-35376-1&domain=pdf http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-35376-1&domain=pdf http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-35376-1&domain=pdf mailto:jianhx@ihep.ac.cn mailto:heguang@tjut.edu.cn fundamental task to explore new cathode materials for the devel- opment of LIBs. Prussian blue analogues (PBAs) have open channel structure that is suitable for alkali ion de/intercalation, and in certain circumstance two-electron reaction per formula unit could occur with optimized compositions9–15. In the past few years, intercalation behaviors of dif- ferent metal ions have been studied including Li+, Na+, K+, Rb+, NH4+, Mg2+, Zn2+ etc., amongwhichNa-PBA is considered themost promising applications in Na-ion batteries (SIBs) with balanced capacities, cell voltages, rate capability and cycling life16–19. For example, the Prussian white Na2FeFe(CN)6 has a theoretical capacity of 170 mAh g−1 and average cell discharge voltage of ~3 V vs. Na/Na+, both of which are competitive among various Na cathodes20. Practically, high discharge capacity (>150 mAh g−1), stable cycling life (>1000 cycles) and good rate capability (>1.9A g−1) have been demonstrated by research groups in Na-FePBA cells21,22. Despite the great of PBA materials in SIBs, no commercial SIBs are commercially available yet. On the other hand, the applications of PBA in LIBs are not optimistic due to the following concerns. First, Li intercalation potential is generally higher than for Na inmost polyanion frameworks, i.e., FePO4 and VOPO4, but this advantage is significantly diminished for Li-PBA cathodes. Furthermore, the replacement of Na- ions by Li-ions could cause structural decomposition of PBAs. Good- enough et al. found the shorter Pauli repulsion radius of Li-ionmakes it prefer to stay closer to N as compared to Na in the MnFe(CN)6 host, which leads to structural evolution of MnN6 octahedra into LiN4 and MnN4 tetrahedra23. Ling et al. also demonstrate that the stable inter- stitial site will convert from face-centered site to body-centered site with the ion radius increases24. All these findings suggest the large PBA voids may not match the smaller size of Li-ions. The PBA framework as the cathode materials for lithium-ion sto- rage was early reported in 199925, the Fe4[Fe(CN)6]3·5.89H2O delivered a discharge capacity of 90 mAh g−1 in the first ten cycles at a constant current density of 0.1mA cm−2 (lithium coin cells with 1M LiClO4 in a propylene carbonate/1,2-dimethoxy ethane as electrolyte, 1:1 vol/vol). After that, there are only few research work using PBAs for lithium insertion such as KMnMn(CN)6, FeFe(CN)6 26–28. The earliest results demonstrated poor cyclability and practicality of Li-PBA materials. Upon the rise of Na-PBA materials in recent years, there have been much attention on the compositional and structural optimization to attain better performance inSIBs, which is likely to be applied in Li-PBA cathodes. In particular, the two-electron reaction provides very appealing capacities once the M site is rationally designed for the Li2MFe(CN)6 cathodes. For example, the Li2FeFe(CN)6 has a theoretical capacity of 190mAh g−1 (vs. 170mAh g−1 for LiFePO4), even comparable to high-energy layered compounds (LiCoO2 and LiNi0.6Co0.2Mn0.2O2.). Also, the previous research works indicate less concerns of mass transfer for Na-PBAs, and similar results are expected for Li-PBAs. This is a difference for PBAmaterials from other Li polyanion cathodes that usually require a nano engineering process before use, because the nanoscale structural design can effectively improve ion or electron diffusion andmitigate themechanical stress at the cathode side29–31. In terms of the comparison of Li-PBA and Na-PBA, the graphite anode is crucial to boost the energy density in cell-level in LIBs due to the different intercalation potentials between graphite and hard carbon10,32. Supplementary Fig. 1 summarizes the key parameters of lithium-based Prussian blue and other materials, highlighting the necessity to revisit the electrochemical behaviors of tailored PBA cathodes in LIBs, in particular for the water content. In this work, we prepared lithium-containing Prussian blue hexacyanoferrate materials with different synthetic routes. The structures of samples were determined by powder neutron dif- fraction, and the dehydration and the chemical degradation process of Li2−xFeFe(CN)6 (denoted as LiFeHCF) samples were observed by operando variable temperature X-ray diffraction (XRD) and combined Thermogravimetry-infrared spectrometry (TGA-IR), confirming the influence of zeolitic water in terms of material thermal stability. Compared to other sizes of LiFeHCF samples, the LiFeHCF-1 sample with micron size and single crystal morphology exhibits a high reversible capacity of 142 mAh g−1 at 19 mA g−1. Fur- ther electrochemical evaluation shows appealing electrochemical energy storage performance of the LiFeHCF-1 sample under differ- ent tests such as high area loading (10mg cm−2), high specific cur- rent (1.9 A g−1), long lifespan (over 1000 cycles) and wide temperature windows (−20 °C to 55 °C). The graphite||LiFeHCF-1 coin cell shows the stable cycling life with a capacity retention of 98% after 300 cycles at 25 °C, highlighting the great potentials of Prussian cathodes for practical applications in LIBs. Results and discussions Materials synthesis and characterizations Herein, a series of LiFeHCF samples with different defects and water contents were systemically prepared by different synthetic methods. The specific methods procedures are described in the experimental section. Briefly, LiFeHCF-1, LiFeHCF-2 and LiFeHCF-3 samples were obtained through ion-exchange process, while LiFeHCF-4 and LiFeHCF-5 samples were synthesized through the self-oxidation/pre- cipitation method. Figure 1a shows the schematic illustration of the synthetic route for LiFeHCF-1 by chemical lithiation of FeFe(CN)6-1 (FeHCF-1), which was first prepared via chemically extracting sodium from NaFeFe(CN)6-1 (NaFeHCF-1). LiFeHCF-2 and LiFeHCF-3 samples were obtained using NaFeHCF-2 and NaFeHCF-3 as the precursors that were prepared by a citrate-assisted co-precipitation method20,22. To investigate the chemical desodiation/lithiation process of LiFeHCF-1, LiFeHCF-2 and LiFeHCF-3, the colors of samples at various preparation stages were recorded (Supplementary Fig. 2). Upon the chemical extraction of sodium, the blue NaFeHCF-1 and NaFeHCF-2 were turned to greenish due to the oxidation of Fe2+ to Fe3+. The blue color was recovered after microwave-assisted treatment with LiI, con- firming the chemical intercalation of Li ions into the lattice. The other two LiFeHCF samples LiFeHCF-4 and LiFeHCF-5 were obtained using Li4Fe(CN)6 as both Li and Fe source (Supplementary Fig. 3). The XRD patterns confirm all LiFeHCF samples are well crystallizedwith the cubic structures (Supplementary Fig. 4). TGA (see Supplementary Fig. 5), element analyzer (EA) and inductively coupled plasma mass spectro- metry (ICP-MS) (see Supplementary Table 1) were employed to calculate the chemical compositions of different LiFeHCF samples, the results show the LiFeHCF-1, LiFeHCF-2, LiFeHCF-3, LiFeHCF-4 and LiFeHCF-5 samples can be denoted as Li1.05Fe[Fe(CN)6]0.97□0.03·2.6 H2O, Li1.36Fe[Fe(CN)6]0.96□0.04·2.9 H2O, Li1.14Fe[Fe(CN)6]0.91□0.09·4.7 H2O, Li0.63Fe[Fe(CN)6]0.96□0.04·2.7 H2O and Li1.0Fe[Fe(CN)6]0.90□0.10·4.9H2O, respectively. To get the insights of the defect concentration and the location of watermolecules, sampleswere further analyzed by neutron diffraction (Fig. 1b–d, Supplementary Figs. 6 and 7). The long-range ordering of Prussian Blue has been previously studied by Herren, Yusuf and others33,34. We built our initial model based on the Herren research work inwhichFe ions are octahedrally coordinatedwithCN ligands in a cubic structure with a space group of Fm-3m. At this stage, no H2O molecules was put into the unit cell as their locations will be deter- mined by the Fourier-Transform different maps after the initial refinement cycles. Constrain conditions were created such that the molar ratio of Fe and CN was consistent with the ICP result in the investigated samples. Nonetheless, the occupancy parameter was allowed to change, in addition to the parameters associated with profile, cell parameter, thermal factor, background, sample displace- ment, and transparency. To reduce the number of variables in the refinement, the thermal parameters of C andNwere assumed to be the same. In addition, the two crystallographic distinct Fe ions were also tightened to be the same. After the above setting, all parameters were Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 2 allowed to change until Rietveld analysis went to convergent. This initial step was used to determine the [Fe(CN)6] 3− vacancy concentra- tion in the PBA structure. Next, a Fourier-Transform neutron different map was conducted to find out the locations of coordinated and uncoordinated water molecules. It is also reported in the literature that Prussian Blue compounds comprise of various defects including [Fe(CN)6] 3- vacan- cies and coordinated/uncoordinated water molecules, which could affect the structures. Based on the literature33,34, H2Omolecules can be placed in theCN vacancy (Wyckoff site 24e). A localmaximumwas also found to be located at theWyckoff site 32f (0.366, 0.366, 0.366) which is the part of the void in the structure. O ions were then added to the N defect locations (coordinated water) and the 32 f sites (uncoordinated water) inside the unit cells. New constrains were created such that the O atomswere allowed tomove between the above two sites while their total quantity was remaining constant. Once the Rietveld analysis was convergent, lithium ions position was determined by the Fourier- Transform neutron different map. Another local maximum was found to be located at the Wyckoff site 32f (0.25, 0.25, 0.25). Finally, lithium ions were added at the 32 sites in the unit cell. All of the Rietveld analysis can be convergent, and the agreement factor wRp (weighted R-factor profile) values were reasonably low (2.62%). All refinement results were summarized into Supplementary Tables 2–5. As shown in Fig. 1e, the LiFeHCF-1 sample demonstrates the lattice parameter increases from 10.2134 to 10.2296Å as compared to LiFeHCF-3, which means higher diffusion coefficient for lithium ions35. In addition, the higher occupancy of water and the longer Fe–O bond distance were observed in LiFeHCF-3 samples. The water molecules located in vacancy can impede Li-ion migration, but suitable water closed to Fe(CN)6 octahedramay helpmaintain structure and avert the severe Li displacement in the <111> direction. Electron paramagnetic resonance (EPR) is an effective character- ization technique that can be used to verify the iron species in sub- stances, especially when studying the coordination environment and oxidation state of transition metals36. Figure 2a and Supplementary Fig. 8a illustrate that the EPR spectra of LiFeHCF-1, LiFeHCF-2 and LiFeHCF-3 samples have characteristic peaks at g-factor = 4.3 and g- factor = 2.03, respectively. The peak at g-factor = 4.3 is assigned to the four-coordinated or distorted four-coordinated isolated Fe3+, while the peak with g-factor = 2.03 was assigned to the highly symmetrical six- coordinated isolated Fe3+ or the characteristic peak of polymeric iron ions37,38. Different LiFeHCF samples show distinct coordination envir- onment of ironbound toFe–C6octahedra or FeN6octahedra. LiFeHCF- 1 has a narrow peak at g-factor = 4.3, which is indicative of that highly symmetrical six-coordinated ferric iron ions and hence the high crys- tallinity and low defects. In contrast, LiFeHCF-2 and LiFeHCF-3 have relatively broad peaks at g-factor = 4.3, which suggests presence of the polymerized Fe ion with increased defects in lattice. The fragmented Fe(CN)6 group hinders Li-ion migration in the crystal structure. The Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy (Raman) were also utilized to study the structural var- iation of LiFeHCF samples. Supplementary Fig. 8b shows the FTIR spectra of different LiFeHCF materials in the range of 500–4000 cm−1. The characteristic peaks at 494 cm−1 and 608 cm−1 are assigned to the a c d b e Fe-C6 octahedra Fe-N6 octahedra O(H2O) Li N C LiFeHCF Framework LiFeHCF-3 LiFeHCF-1 a = 10.2296 Å LiFeHCF-3 a = 10.2134 Å 2.05531 Å Fe O 0.3252(7) occupancy LiFeHCF-1 2.37048 Å Fe O 0.5228(7) occupancy LiFeHCF-3 c b c b a c b a NaFeHCF-1 FeHCF-1 Stirred 60 oC LiFeHCF-1 Na4Fe(CN)6 HCl Self-oxidation Chemical lithiation Chemical desodiation Microwave-assisted solvothermal Nitrogen protection N C Fe Fig. 1 | Material synthesis strategy and refinement results. a Schematic illus- trations of the synthetic route for the LiFeHCF-1 sample. b Crystal structure of lithiatedPrussianblue.Neutrondiffractionpatterns andRietveld refinement results of the c LiFeHCF-1 and d LiFeHCF-3 samples. e Comparison of the unit cell parameters and coordination environment between zeolitic water and Fe(CN)6 octahedra for the LiFeHCF-1 and LiFeHCF-3 samples. Lithium, carbon, nitrogen, iron (coordinatedwith carbon), iron (coordinatedwith nitrogen) andoxygenatoms are shown in green, blue, yellow, purple, dark blue and red colors, respectively. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 3 out-of-plane bending vibration peak and the in-plane bending vibra- tion peak of the Fe–C chemical bonds in the Fe–C≡N group, respectively39–41. In addition, the peaks at 1638 cm−1 and 3462 cm−1 are O–H bond stretching vibration peaks and in-plane bending vibration peaks, and the peaks at 2082 cm−1 represents the stretching vibration peak of the C≡N bond that is related to the transition metal ions bonded to the CN− groups. Notably, the stretching vibration peak of the C≡N bond, as shown in Fig. 2b, reveals the vibrations of Fe–C≡N group of LiFeHCF samples. Compared to LiFeHCF-3, LiFeHCF-1 has a blue shift andhigherpeakof Fe–C≡Nat608 cm−1, which is attributed to the stronger C≡N chemical bond and more stable Fe–C≡N group. In Fig. 2c, Raman spectra (1800–2550 cm−1) show the characteristic dif- fraction peak attributed to the C≡N groups bonded to iron ions of different valences in the lattice21,42. The gradual increase of peaks at 2151 cm−1 and 2094 cm−1 indicates that the increased symmetry of the FeC6 and FeN6 octahedra, which is consistent with the EPR result. Based on the above analyses, Fig. 2d schematically illustrates the dis- tribution of the Fe–C≡Ngroup in LiFeHCF-1, LiFeHCF-2, and LiFeHCF-3. LiFeHCF-1 has more symmetric FeC6 or FeN6 octahedra, good crys- tallinity and lower defects among all samples, whereas the massively incomplete Fe–C≡N group in LiFeHCF-2 and LiFeHCF-3 possibly cause irreversible insertion/extraction of lithium ions and the increased space for zeolitic water. The different structures of LiFeHCF samples were further verified with the Nitrogen adsorption/desorption mea- surements and analyses. As shown in Fig. 2e, f, LiFeHCF-3 shows higher specific surface area (55m2 g−1 vs. 15m2 g−1) as well asmoremicropores (<2 nm) than for LiFeHCF-1. The high surface area is partially due to the reduced particle size of LiFeHCF-3 (see Supplementary Fig. 9), but also an indication of abundant defects resulted from the synthetic process. The Nitrogen adsorption/desorption measurements and analyses of LiFeHCF-2, LiFeHCF-4, and LiFeHCF-5 samples were also presented for comparison (Supplementary Fig. 10). To further probe the local environment, X-ray photoelectron spectroscopy (XPS) measurements was performed with LiFeHCF samples (Supplementary Fig. 11). For high-resolution XPS spectra, the Fe 2p peaks are located at the binding energy of 724.7 eV and 709.8 eV, corresponding to Fe3+ 2p3/2 and Fe3+ 2p1/2, respectively43,44. Within the binding energy regions of 721.2 eV and 708.4 eV, all LiFeHCF spectra show the characteristic peaks of Fe2+, but the signals of Fe3+ ions are not different from each other. The LiFeHCF-3 and LiFeHCF-5 samples have a bulging at ~712 eV caused by polymeric iron ions, which is consistent with the ERP results. It’s generally recognized that water content has significant influ- ence on electrochemical performance for Prussian cathodes10,32. For this reason, thermogravimetric analysis (TGA) was performed to determine the presence and content of water in the lattice of LiFeHCF samples. In Supplementary Fig. 5, distinct weight loss between below 300 °C is attributed to the release of the adsorbed water and the coordinated water. The results were summarized in Fig. 3a revealing the mass loss of each LiFeHCF sample as 98.1%, 95.2%, 89.7%, 94.9%, and 81.8% at 150 °C, respectively.When the temperaturewas increased to 300 °C, the values were turned to 80.1%, 73.8%, 70.3%, 80.5%, and 68.6%, respectively. These results indicate less coordinated water in LiFeHCF-1 and LiFeHCF-4 samples. The more symmetric FeC6 or FeN6 octahedra in these samples lead to a decrease of water in lattice and themore stable structure; In contrast, the gradual increased defects in LiFeHCF-2, LiFeHCF-3, and LiFeHCF-5 samples lead to thermal instability, which is consistent with the previous conclusions. The dehydration process and chemical degradation process of LiFeHCF samples were further studied with operando variable tem- perature XRD performed from 25 to 450 °C (Supplementary Fig. 12). Supplementary Fig. 13 is the schematic illustration of the stage with LiFeHCF powder in the heating module. Figure 3b, c (Supplementary Figs. 14 and 15) displays the TGA (DTA, differential thermal analysis) curves and the contour plots of the (200) and (220) diffraction peak of LiFeHCF-1 and LiFeHCF-3. The heating and the decomposition process can be sketched into three stages. In the first stage (stage a, from 25 °C to 100 °C), the (200) peak gradually shifts to higher angles and the (220) peak becomes weak, corresponding to the extraction of zeolitic water (absorbed water) from lattice. Then the remaining zeolitic water (absorbed water) continuously evaporate from defect sites, and the (200) and (220) peaks remain shifting and weakening from 100 °C to 250 °C/290 °C. When the temperature is increased to 300 °C/330 °C (stage b), the coordination water is deeply extracted from lattice and LiFeHCF-2 LiFeHCF-3 LiFeHCF-1 Vacancy M(Li, Na, K…) C FeHS(N) N FeLS(C) a b c a b c a b c c b c a a b d b c f e a Fig. 2 | Structural characterization of different LiFeHCF samples. a Electron paramagnetic resonance (EPR) spectra of the LiFeHCF-1, 2, and 3 samples. b FTIR spectra of various LiFeHCF materials. The inset shows the enlarged plots of the spectra between 550 and 700 cm−1. c Raman spectra (1800–2550 cm−1) of LiFeHCF showing the evolutions of the (C≡N)− group. d Schematic illustrations of the char- acteristics of LiFeHCF samples and vacancy correlations. N2 adsorption–desorption isotherms (e) and Pore-size distribution (f) of the LiFeHCF-1, LiFeHCF-3 samples. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 4 the samples evolve into an anhydrous phase. The disappearance of (200) and (220) peaks occurred at the 300 °C/330 °C due to the structural decomposition. We further applied the combined Thermogravimetry-infrared spectrometry (TGA-IR) technology for the LiFeHCF sample under flowing Ar protection (Supplementary Fig. 16), confirming the decomposition products include CO2, NO2 and HCN45,46. Significantly, the LiFeHCF-3 sample exhibits relatively broa- dened and weakened diffraction peaks at stage a, indicating more zeolitic water was removed from the crystal lattice, resulting in severe lattice distortion and volume reduction. Figure 3d shows more detailed comparison of the cell parameter variation during heating. Due to the release of zeolitic water and coordination water, the cell parameter is decreased from 10.218 Å to 10.129 Å for LiFeHCF-3, and the unit cell volume is contracted by 2.7%. In comparison, LiFeHCF-1 has a little shrinking of 0.9% as the cell parameters is stable from 25 °C to 290 °C (10.201 Å vs. 10.23368Å). Afterwards, the two LiFeHCF samples exhibit the lattice expansion with volume change of ~1.2% and ~2.1% until the samples start to decompose. Collectively, the dehydration behavior and the chemical degra- dation process of LiFeHCF samples is summarized in the schematic illustration of Fig. 3e. Different types of water correspond to various extraction temperatures and lattice changes. The results also suggest that the presence of zeolitic water in the crystal lattice has negative effects on thermal stability, which was in fact dominated by the increase of asymmetric Fe–C6/Fe–N6 octahedra. The correlations between thermal stability and zeolitic water content and the integrity of structural framework of LiFeHCF materials are not usually reported and discussed in the literature. The operando variable temperature XRD patterns of LiFeHCF-2 and LiFeHCF-4 show similar results as the LiFeHCF-1 and LiFeHCF-3 samples in Supplementary Figs. 17–20. The scanning electron microscope (SEM) images for LiFeHCF-1, LiFeHCF-2, LiFeHCF-3, LiFeHCF-4 and LiFeHCF-5 powder samples are shown in Supplementary Fig. 9. LiFeHCF samples preparedbydifferent methods possesses similar cubic morphologies but different particles sizes (from nanometer to micrometer). In Supplementary Fig. 9a, d, the cubic features with smooth interface of LiFeHCF-1 and LiFeHCF-4 samples can be attributed to the self-decomposition synthesis meth- ods with intrinsic slow nucleation rate (using Na4Fe(CN)6, Li4Fe(CN)6 as sodium source and lithium source, respectively). In contrast, LiFeHCF-2 and LiFeHCF-3 exhibit different morphologies with agglomerations from the other two samples, relating to the randomly aggregated process by the modified citrate-assisted co-precipitation method. Figure 4a–f shows the typical crystallite sizes of LiFeHCF samples. LiFeHCF-1 has largest size of ~ 4 μm and LiFeHCF-4 cube are only ~650–850nm. LiFeHCF-1 sample possess a single-crystal mor- phology with high crystallinity, while LiFeHCF-2 and LiFeHCF-3 Fig. 3 | Characterizations of different LiFeHCF samples during the heating process. a The variation of weight in different LiFeHCF samples. Structural evo- lutionofb LiFeHCF-1 and c LiFeHCF-3 powder samples duringheating from25 °C to 450 °C. d Comparison of the cell parameters changes of distorted framework of LiFeHCF samples during heating process. The error bars represent the range of cell parameters for the Prussian blue samples upon heat treatment. e Schematic illus- tration of the dehydration behavior and the chemical degradation mechanism of LiFeHCF samples. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 5 samples has the average crystallite sizes of ~2μm and ~200nm. Their precursors were synthesized by same coprecipitation method expect for using different content of sodium citrate. LiFeHCF-5 sample pre- pared by simple coprecipitation method (using Li4Fe(CN)6) has the smallest size of ~10 nm.All samples have different nucleation rates, and the fast crystallization may result in the presence of vacancies and the large content of zeoliticwater, hindering the growthof crystallite sizes. We also attempted to directly observe the location of defects and water in the lithium-based Prussian blue lattice, but the samples too sensitive for high-resolution electron microscopy and spectroscopy, for this reason, PBA high-resolution images have rarely been reported47. Electrochemical performances of LiFeHCF samples LiFeHCF electrodes were evaluated in coin cell configuration using Li metal as counter electrode with 1M LiPF6 in ethylene carbonate (EC) and diethyl carbonate (DEC) electrolyte. Figure 4g–k shows the Galva- nostatic cycling curves of LiFeHCF-1, LiFeHCF-2, LiFeHCF-3, LiFeHCF-4 and LiFeHCF-5 samples at a specific current of 19mAg−1. The cells deliver discharge capacity of 126 mAh g−1, 117 mAh g−1, 122 mAh g−1, 103mAh g−1 and 80mAh g−1 in the second cycle, respectively. It is worth noting that the LiFeHCF-1 sample with the largest particle size has an activation process and the highest discharge capacity of 143 mAh g−1 appears after 30 cycles (Fig. 4k). Supplementary Fig. 21a give a com- parison between crystallite size and electrochemical energy storage performance of LiFeHCF samples. It has been demonstrated that the crystallite size is closely related to the electrochemical behaviors of potassium Prussian white materials (K1.7Fe[Fe(CN)6]0.9) 48. The K-based Prussian blue sample with a small size (~20nm) shows the best elec- trochemical performance; however, the micron-sized Prussian blue samples seem more suitable for lithium-ion batteries. The typical discharge voltage plateau at 3.1 V can be observed for all samples, but the LiFeHCF-1, LiFeHCF-2 also have a second plateau at 3.8 V, which correlate with the reduction reactions Fe3+/Fe2+ couples during the lithiation process. Cyclic voltammetry (CV) test was per- formed to understand the reasons for various voltage plateau of LiFeHCF samples. As shown in Fig. 5a, the cathodic peaks at 3.1 V belongs to the redox reactions of high-spin Fe3+/Fe2+ couples coordi- nated with nitrogen, and the high-potential peaks at 3.8 V are related to the low-spin Fe3+/Fe2+ couples coordinated with carbon49. The CV curves of other LiFeHCF samples under the same test conditions are shown in Supplementary Fig. 21b–d. The peaks do not appear in high voltage range for LiFeHCF-3, indicating there is correlation between Fe–C6 octahedra and redox potentials upon lithium ions insertion/extraction. Rate capability of LiFeHCF electrodes were furthered investigated in non-aqueous Limetal coin cell configuration as shown in Fig. 5b and Supplementary Fig. 22. Li metal cells with LiFeHCF-1-based positive electrodes show the best performance among all samples. The capa- cities are 135, 131, 128, 122, 107, and 97mAh g−1 at 38, 95, 190, 380, 950, and 1900 mA g−1, respectively. Also, the LiFeHCF-1 electrode demon- strates superior cycling performance at 190mAg−1, delivering a reversible capacity of 109 mAh g−1 and good retention of ~90% after 650 cycles. In comparison, the capacities of LiFeHCF-2, LiFeHCF-3, LiFeHCF-4, and LiFeHCF-5 electrodes at 190mAg−1 are 87, 41, 61, 29 mAh g−1 and capacity retention of 82.3%, 40%, 70%, and 47.4%, respectively (Fig. 5c). Significant capacity fading occurs after 100 500 nm 500 nm 500 nm 500 nm 1 μm a b e d c 4 μm f g h j i k 3 μm 500 nm 200 nm Particle size LiFeHCF-1 LiFeHCF-5 LiFeHCF-3 LiFeHCF-4 LiFeHCF-2 10 nm 2 μm Fig. 4 | Influence of particle size on electrochemical performance of LiFeHCF samples. Particle size of a LiFeHCF-5, b LiFeHCF-3, c LiFeHCF-4, d LiFeHCF-2, and e LiFeHCF-1 powder samples. All images are taken from ex situ SEMmeasurements. f Schematic illustration of a microcrystalline crystallite size of LiFeHCF sample. g–k The corresponding capacity−voltage profiles of the LiFeHCF samples at spe- cific current of 19mAg−1. All LiFeHCF electrodes were evaluated in coin cell con- figuration using Li metal as counter electrode and tested at 25 °C. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 6 cycles for LiFeHCF-3 and LiFeHCF-5. The poor stability is possibly related to the large number of vacancies that hinder Li-ion migration through the <100> direction. Besides, the zeolitic water in the struc- tures is gradually extracted during the charge/discharge process, fur- ther accelerating electrolyte decomposition and reducing coulombic efficiency. The LiFeHCF-1 electrode exhibits noticeable long-term cycling performance at a high specific current of ~ 1.9 A g−1. As shown in Fig. 5d, the cell deliveres a discharge capacity of 118.9 mAh g−1 at the initial cycle and maintains 80.7% capacity over 1000 cycles (0.019% capacity fading per cycle) with a high coulombic efficiency of 99.3%. Even positive electrodes with a mass loading of 10mgcm−2, the Li metal coin cell has a capacity retention of 98% after 300 cycles at 190mAg−1 (Fig. 5e); it also shows good performance at 55 °C and −20 °C (Fig. 5f and g). Supplementary Table 6 summarizes the elec- trochemical performances of representative Na-PBA cathodes and our LiFeHCF, highlighting its potentials for the application in LIBs. To better evaluate the relation between discharge and lithium-ion diffusion properties, galvanostatic intermittent titration technique (GITT) measurement was performed by inserting Li+ into LiFeHCF samples at a specific current of 190mAg−1 between 2.0 and 4.3 V, and the cell was discharged from 4.3 V for 10min followed by a rest of 120min. As shown in Supplementary Fig. 23, the sloping regions at 3.8–3.5 V and 2.7–2.0 V are indicative of solid solution process, while the overpotential between 3.2–2.7 V (corresponding to lithium inser- tion from Li0.5FeFe(CN)6 to Li1.8FeFe(CN)6) and the flat sloping region indicates two-phase mechanism of the insertion of more lithium49,50. Investigation of phase transitions of LiFeHCF-1 sample during cycling Operando XRD was conducted to study the phase transitions during Li+ insertion and extraction in the LiFeHCF-1 sample. The digital image and schematic illustrations of the operando XRD cell and instrument are shown in Supplementary Fig. 24a, b. In Fig. 6a, the peaks at 17.8° and 35.5° are gradually shifted to lower angles and the peak at 25° is disappeared (Supplementary Fig. 24c), indicating the increases of the volume of the unit cell on lithiation. The peaks are recovered at the initial position after 2nd discharge, suggesting the formation of Li-rich Li1+xFeFe(CN)6. It is worth mentioning that this phenomenon is dif- ferent from the observation of irreversible lattice shrinking with Na Prussianbluematerials50. The cubic unit cell and the ease ofmovement of lithium ions for LiFeHCF-1 material have a highly reversible phase transition that helps long cycle life (over 1000 cycles at 1.9 A g-1). The expanded views of the (200), (220), and (400) peak regions are shown in Fig. 6b. The (200) peak gradually shifts from 17.8° to 17.52° upon the first discharge process, which indicates the evolution of the cell parameters during the second lithium intercalation process. The full insertion of Li+ to Li1+x FeFe(CN)6 lattice evolved a new phase when the voltage reached 2.0V. The subsequent charge shows the recovery of peak, corresponding to the deintercalation of Li+ vs. Li2−xFeFe(CN)6 and FeFe(CN)6. Notably, the (200) and (400) diffrac- tion peaks are recovered after re-discharged to 2.0 V, confirming the high reversibility of the Li2−xFeFe(CN)6 phase during Li intercalation/ dentercalation. This highly reversible Li migration process is likely to d b c f e g a 1.9 A g-1 Over 1000 cycles Fig. 5 | Electrochemical evaluation of LiFeHCF-1 samples. a Cyclic voltammetry curves of LiFeHCF-1 sample at different scanning rates. b Rate performance of LiFeHCF samples at various specific currents. c Cycling performance of LiFeHCF samples at 190mAg−1.d Long-termcycling of LiFeHCF-1material at 1.9 A g−1. eHigh- loading performance (10mg cm−2). f, gHigh and low temperature tests at 55 °C and −20 °C. All LiFeHCF electrodes were evaluated in coin cell configuration using Li metal as counter electrode and tested at 25 °C (exceptwhere differently indicated). The specific capacity values were calculated base on themass of positive electrode active material. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 7 be attributed to the more symmetric Fe–C6/Fe–N6 octahedra of the LiFeHCF-1 sample. Also, the (220) diffraction at 25° splits into two peaks after discharged to 2.0 V, which is another evidence of high lithiation degree of the LiFeHCF-1-containing positive electrode. As schematically illustrated in Fig. 6c, it is assumed that the onset of Li diffusion is through the <100> direction for the cubic FeIIIFeIII(CN)6. Upon the subcube sites are half-occupied, the diffusion path is chan- ged to the <111> direction due to the volume expansion accompanied by the lattice transformation from cubic LiFeIIIFeII(CN)6 to rhombohe- dral structure Li2Fe IIFeII(CN)6, which is in agreement with the GITT results. Lithium-ion and water molecules diffusion behavior To understand the Li-ion diffusion behaviors, bond valence energy landscape (BVEL) calculations was performedwith LiFeHCFs. Since the C and N atoms are covalently bonded together in the CN ligands, the correct charge on C and N could not be easily determined. None- theless, we arbitrarily assigned C and N with three different scenarios: 1. C4+ and N3−, and 2. C3+ and N3−, and 3. C4+ and N3−. We expect that the exact energy barrier would be estimated with uncertainty. However, such amethod could still be useful to illustrate the topology of the pathways of lithium ions. As shown in Fig. 7a, we reveal that Li- ion diffusion can proceed through the a 3.2 × 2.0 Å2 ellipsoid pore windowcentered at (0.25, 0.25, 0) into the 3D-interconnected cubic 3 × 3 × 3 Å3 nano-cage centered at (0.25, 0.25, 0.25). The diffusion of lithium ions was mainly driven by two factors: 1. the columbic inter- action between the negative charged framework and the position of the charge on the lithium ions when the Fe ions were undergone reduction reaction to change their valence states from 3+ into 2+, upon the discharging process; 2. the chemical potential different between the PBA compounds and its surrounding environment. In SupplementaryFig. 25, since theoccupancyofH2Omolecules is 0.32 at the 32f site (0.25, 0.25, 0.25) in the unit cell, the probability to H2O molecule at this sitewas 32%. The topological pathwaywas very similar to the simulation without putting an addition water molecule in the unit cell (Supplementary Figs. 26, 27 andTable 7). However, thedensity of the percolation network was found to shrink significantly, as we compared the same bond valence mismatch value of 1. This indicated that the water molecule can reduce very significantly on the con- ductivity of the lithium ions in the unit cell. If the water content is increased in the framework, the apparent percolation network will shrink significantly. Therefore, the lithium-ion conductivity will be dropped dramatically. Recently, a few computational studies have tried to explain how water molecules can stabilized the structure of PBAs upon electro- chemical cycling51,52. This enlighten us to further investigate diffu- sion behaviors of water molecules in PBAs as water molecules could be exchanged between the non-aqueous electrolyte solution and the PBA structure during the charge/discharge processes. As shown in Fig. 7b, c, the BVS mismatch map on PBA indicates the diffusion pathway of water molecules are resemble but wider spread behavior as the Li-ion percolation network. This suggests that water mole- cules could easily block the diffusion pathway of Li-ions if their concentration is high (>1.8mol% or 9.8 wt.%). Therefore, minimizing the water content and the defects in PBAs could help to optimize the electrochemical energy storage performances of Li-based cells with PBAs-containing positive electrodes. As shown in Supplementary Fig. 28, additional supercell analysis was performed to enlarge the unit cells along x by 2 and along y by 2 to provide better understanding on the effect of H2O molecules in space as they can either occupy or un-occupy into the 32f sites. Of 32 sites in the supercells, 10 of them will be filled with H2O molecules. This will T im e 4.3 V 2.0 V (2 00 ) (2 00 ) (2 02 ) (2 02 ) (2 20 ) (4 00 ) (2 20 )/ (0 24 ) (2 20 ) (4 04 ) (4 00 ) (4 04 ) 2.0 V a b c In te ns ity ( a. u. ) Li+ Li+ Fe Fe (CN)6 LiFe FeII(CN)6 Li2FeIIFeII(CN)6 b Li+ Li+ Li+ Li+ Li+ Li+ a c c b b a c b a c c b c b Fig. 6 | Structural evolution of the LiFeHCF-1 sample during charge and dis- charge. a Operando XRD patterns. b Detailed information of the (200), (220)/ (024), and (400) diffractions. c Schematic of the phase evolutions upon Li insertion/extraction. The Operando battery for XRD testing was carried out by stainless steel cell with beryllium window at specific current of 9.5mAg−1 at 25 °C. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 8 come with a total combination of 64512240. Thanks to the supercell program, the columbic energies of all supercells can be calculated and ranked based on their energies for outputting. Below is the additional BVEL analysis on the lowest energy supercell to show the effect of H2O on the lithium-ion diffusion property. First, it is interesting that water molecules are not evenly distributed inside the supercell, while some of them like to aggregate to form water clusters. With a small value of BVEL, 3D percolation network can be observed but they are formed around the water clusters. As we progressively increase the BVEL values, lithium ion diffusion network can also form as they pass through the void where water molecules are absent. Interestingly, the increase in the density of 3D percolation network is mainly due to the new pathways with void, while the original pathways, with water clusters nearby, is hardly changed. Therefore, we can conclude that an excess of watermolecules will decrease the over density of lithium-ion percolation network in the structure. Electrochemical performance of LiFeHCF-1-based positive elec- trodes in Li-ion coin cell configuration To further validate the practical application of the LiFeHCF-1 cathode, Li- ion cells were assembled and tested with commercial graphite as the negative electrode active material (Fig. 8a and Supplementary Fig. 29). At a specific current of 38mAg−1 (considered the mass of positive electrode active material), the graphite||LiFeHCF−1 coin cell gave a reversible capacity of 106 mAh g−1 (Fig. 8b) with an average discharge voltage around 2.9V, no capacity attenuation was appeared after 5 cycles over the voltage range of 1.5 ~ 4.0V. In Fig. 8c, d, the full cell showed good reversible capacity 104, 102, 99, 95 mAh g−1 at different rates of 95mAg−1, 190mAg−1, 380mAg−1 and 950mAg-1, respectively, and the specific capacity could be recovered to 109 mAh g−1 at specific current of 38mAg−1. Interestingly, the cell can be cycled up to 1.9 Ag−1 delivering a specific discharge capacity of about 90 mAh g−1. Further- more, Fig. 8e shows a high capacity of 105 mAh g−1 and capacity reten- tion of 98% (average coulombic efficiency ≈ 99.6%) over 370 cycles. In summary, we systematically studied the preparation of various LiFeHCF samples and compared their Li-ion storage performance in a non-aqueous environment. The impacts of zeolitic and crystal water in terms of thermal stability and structural framework integrity have been revealed by using advanced technologies including powder neutron diffraction, thermogravimetry-infrared spectrometry, operando variable temperature X-ray diffraction and Bond Valence Energy Landscape calculations. The tailored LiFeHCF material with micron size and single crystal morphology exhibits high capacity of >140mAh g−1 (at 19mAg−1) and long lifespan over 1000 cycles, as well as appealing performance under practical conditions such as high loading (10mgcm−2), wide temperature (−20 to 55 °C) window and Li-ion cell configuration (graphite||LiFeHCF). Methods Materials synthesis LiFeHCF-1, LiFeHCF-2 and LiFeHCF-3were prepared by the ion-exchange method from NaFeHCFs53,54. Typically NaFeFe(CN)6−1 (NaFeHCF-1) was first obtained via self-oxidation/precipitation55,56. First, 2mmol sodium ferrocyanide (Na4Fe(CN)6, 99.7%, Aladdin) were dissolved in 100ml deionized water to afford clear solution under stirring (400 rpm). Next, 1mL of hydrochloric acid (37wt% solution in water, Acros) was added dropwise to the solution, and then the solution was heated at 60 °C under vigorous stirring (600 rpm) for 4 h. The resulting dark blue sus- pensionwas collectedby centrifugation (8000 rpm, 5min), washedwith water (40ml) and ethanol (40ml) three times, and dried in vacuum oven at 100 °C for 24h to obtain NaFeHCF-1. FeFe(CN)6-1 (FeHCF-1) was prepared by chemical desodiation using tetrafluoroborate (NO2BF4, 96%, Alfa, excess 50%) under nitrogen atmosphere with ultra-dry acet- onitrile medium (99.9%, water ≤ 30 ppm, Innochem). LiFeFe(CN)6-1 (LiFeHCF-1) was synthesized by chemical lithiation through a microwave-assisted solvothermal (MW-ST) process (using Anton Paar Monowave 400, 600 rpm) at 80 °C. LiI (99.9%metals basis, Aladdin)was served as both lithium source and reducing agent. The obtained sample were collected by centrifugation (8000 rpm, 5min), washed by acet- onitrile anhydrous and vacuum-dried 12 h at 100 °C. The LiFeHCF-2 and LiFeHCF-3 samples were prepared by the same route except for the different NaFeHCF-2 and NaFeHCF-3. Here both NaFeHCF samples were synthesized by a modified citrate-assisted co- precipitation method20,57. Briefly, 6mmol iron sulfate heptahydrate (FeSO4·7H2O, 99.5%, Acros) and 25 g (NaFeHCF-2) or 5 g (NaFeHCF-3) of sodium citrate (C6H5Na3O7, 98%, Aladdin) were dissolved in 100mL of a b Bond Valence Sum Mismatch map (BVS) for oxide ions (100) (111) (110) Selective Projection on (100) for lithium ions Z = 0-0.1 Z = 0.1-0.2 Z = 0.2-0.3 Z = 0.3-0.4 Z = 0.4-0.5 (100) (111) (110) Bond Valence Energy Landscript (BVEL) for lithium ions a b c b c a a b c a b a c b b a c a b Fig. 7 | Theoretical prediction of the pathways of Li-ions for PBAs. a Bond Valence Energy Landscape (BVEL) calculation for Li-ions in LiFeHCF samples. b Bond Valence Sum Mismatch map (BVS) calculation for oxide ions in LiFeHCF samples. The selected projection on (100), (110) and (111) planes were presented, indicating that Li-ions can be transported 3-dimensionally through the void of PBA compounds. c BVEL with different Z values on the (100) planes. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 9 deionized water, stirring (400 rpm) until a clear solution A was formed. 4mmol Na4Fe(CN)6 was added to 100mL deionized water, labeled as solution B. Then, solution A was slowly added to solution B under vigorous stirring (600 rpm) and a milky white precipitate formed immediately. The mixture was stirred 12 h, and the product were col- lected by centrifugation (8000 rpm, 5min), washed by deionized water (40ml) and ethanol (40ml) and vacuum-dried 24 h at 100 °C. LiFeFe(CN)6-4 and LiFeFe(CN)6-5 (LiFeHCF-4 and LiFeHCF-5) samples were prepared using lithium ferrocyanide (Li4Fe(CN)6). Li4Fe(CN)6 samples was synthesized following the same methodology as reported in previous literature58,59. Briefly, 56mmol potassium fer- rocyanide (K4Fe(CN)6, 99%, Acros) was dissolved in 25ml deionized water, and 14mmol lithium perchlorate (LiClO4, 99.99%, Aladdin) solution were mixed with continuous magnetic stirring (400 rpm). A white precipitate of potassium perchlorate was formed, which was removed by centrifugation (8000 rpm, 3min), while the solution was heated to 70 °C to remove partial water. Next, the obtained solution was placed in a refrigerator overnight to allow potassium perchlorate was precipitated. The above steps were repeated several times until no KClO4 precipitation was observed. The Li4Fe(CN)6 was dried at 150 °C. LiFeHCF-4 and LiFeHCF-5 was prepared by the same method as NaFeHCF-1 and NaFeHCF-3 except used Li4Fe(CN)6 without citrate used during the synthesis. Materials characterizations The obtained LiFeHCF samples were investigated by powder XRD analysis (RigakuMiniflex 600) equipped using Cu kα radiation at 40 kV and 20mA. The crystal structure of LiFeHCF samples was analyzed by neutron diffraction. The diffraction data (LiFeHCF-1, 3 and 4 samples) was collected at the ECHIDNA high-resolution powder diffractometer with a monochromatic wavelength of 1.6215 Å in Australian Nuclear Science and Technology Organization (ANSTO). TOF neutron e d c b a LiFeHCF e Charge Graphite - + e- Discharge Al Cu Li+ Fig. 8 | Electrochemical performance of LiFeHCF-1-based positive electrodes coupledwith graphite-basednegative electrodes. a The schematic illustration of the graphite | |LiFeHCF full cell. b Galvanostatic charge–discharge voltage profiles of the full-cell at 38mAg−1. Voltage profiles (c) and rate capability (d) at different current densities from 38mAg−1 to 1900mA g−1. e The demonstration of long-term cycling of the cell at 190mAg−1. All LiFeHCF graphite||LiFeHCF full cell were tested at 25 °C and the specific capacity values were calculated based on the mass of positive electrode active material. Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 10 diffraction measurement of LiFeHCF-2 was conducted at Multiple Physics Instrument (MPI) in China Spallation Neutron Source (CSNS). 2–3 g of powders were put into vanadium cans and the measurement time was about 3 h for each sample. The diffraction data was subse- quently analyzed by GSAS EXPGUI60. The content of Li, Fe, C, N ele- ments in the LiFeHCF samples were identified by ICP analysis (OPTIMA 8000DV Optical Emission Spectrometers) and Element analyzer (Vario EL Cube), the relative standard deviation of the measured samples was less than 1.5%. Scanning electron microscopy (SEM) images were per- formed by FEI Verios 460 L. X-ray photoelectron spectroscopy (XPS) was taken on an ESCALAB250Xi (Thermo Scientific) spectrometer equipped with X-ray source (hv = 1486.6 eV, monochromatic Al Ka, 150W). Fourier transform infrared (FTIR) spectra were carried out on a Fourier transform infrared spectrometer (Perkin Elmer) with KBr disk method. Raman spectra were measured at a Raman microscope with a 325 nm excitation laser (HORIBA JOBIN YVON S.A.S.). Nitrogen adsorption/desorption measurements were performed on Autosorb- iQ-MP (Quantachrome), the micropore surface areas, pore size dis- tribution and pore volumes were determined from nitrogen (N2) adsorption branch at 77K with Brunauer–Emmett–Teller (BET) theory and the Barrett–Joyner–Halenda (BJH) model. EPR spectra were recor- ded from Bruker EMXplus-6/1 EPR spectrometer (9065.8MHz, X band, 300K). Thermogravimetry coupled with infrared spectrometry (TGA- IR) techniques (Perkin Elmer, Frontier Mid-IR FTIR/STA6000-TL9000- Clarus SQ8) was employed to confirm the decomposition products. Operando variable temperature X-ray diffraction measurement was performed on Rigaku Miniflex 600 from 25 °C to 450 °C, the powders sample was placed in heating module (Supplementary Fig. 13). Electrochemical measurements The LiFeHCF electrodes slurry were coated from N-methyl-2- pyrrolidone (NMP, 99.9%, Aladdin) onto aluminum collectors using polyvinylidene fluoride (PVDF, 99.9%, Arkema) as the binder. The typical electrode was made with a formula of active material (LiFeHCF powder): Super P (99.9%, Lion corporation): PVDF = 70:20:10. Then the electrodes were vacuum dried at 100 °C for 12 h and the active mass loading was ~1.5 mg cm−2. CR2032 lithium coin cells were fabricated inside an Ar-filled glovebox (H2O < 0.1 ppm, O2 < 0.1 ppm) with 1M LiPF6 in diethyl carbonate (DEC) and ethylene carbonate (EC) (1:1 vol/vol) electrolyte (water content <10 ppm), a metallic lithium (0.45mm, 99.9%, Innochem) negative electrode, and celgard separator (25 μm, 55% porosity, 0.064 μm pore size). The charge-discharge test was carried out on Land CT2001A battery cycler and tested in constant temperature test room (25 ± 1 °C, except where differently indicated). The high and low-temperature tests were investigated on the high-low temperature test chamber (SHIPAC, operating temperature range: - 75 °C~150 °C ± 1 °C). The specific capacity was obtained based on the mass of positive active material and at least two parallel cells were tested for individual electrochemical experiment. Cyclic voltammetry (CV) tests were conducted on an Ivium electrochemical Workshop between 2.0–4.3 V vs. Li/Li + at various scan rates of 0.1 mV s−1, 0.2mV s−1, 0.4mV s−1, 0.6mV s−1, 0.8mV s−1, and 1.0mV s−1. GITT testing of coin cell at 9.5 mA g−1 between 2.0–4.3 V, in which the cell was discharged for 10min followed by 120min resting. The Operando XRD test was carried out with a stainless steel electrochemical cell with beryllium window at 9.5 mA g−1 from 2.0 to 4.0 V, which was assembled with metallic lithium negative electrode, Celgard separator, and positive electrode. The full cell was assembled with commercial graphite (50~90 μm, 99.7%, Shenzhen Kejing Star) as the anode, and the graphite electrodes were made by commercial graphite (90wt%) and PVDF (10 wt%). The loading mass of LiFeHCF and graphite was ~1.5 mg cm−2 and the LiFeHCF|Graphite mass ratio is 1:0.8. The slurry in NMP was coated on copper foil (9 μm, 99.8%, Shenzhen Kejing Star) and dried under vacuum at 120 °C. Calculation method The lithium-ion migration behavior was estimated by 3DBVSMAPPER. In particular, Bond Valence Energy Landscapist (BVEL) was calculated by lithium-ion migration pathways in two situations: 1. Valence states of C, N, O, and Fewere assigned as +2+, +3, −2 and +3; 2. Valence states of C, N, O and Fe were assigned as +4, +3, −2, and +3, respectively. The two calculated percolation networks were very similarwith each other, indicating the deviation of valence state for C between 2 and 4was not sensitive to the diffusion pathway. Similarly, the water diffusion pathway was also analyzed by the same methods. Two Valence states of C, N, O were separately considered as +4, +3, −2 and +2, +3, −2, respectively. The diffusion percolation networks were found to be non-sensitive with the valence states of C and N atoms. Data availability The data that support the findings of this study are available from the corresponding author upon reasonable request. Source data are pro- vided with this paper. References 1. Tarascon, J.-M. & Armand, M. Issues and challenges facing rechargeable lithium batteries. Nature 414, 359–367 (2001). 2. Armand, M. & Tarascon, J.-M. Building better batteries. Nature 451, 652–657 (2008). 3. Goodenough, J. B. & Kim, Y. Challenges for rechargeable Li bat- teries. Chem. Mater. 22, 587–603 (2010). 4. Ohzuku, T., Ueda, A. & Nagayama, M. Electrochemistry and struc- tural chemistry of LiNiO2 (R3m) for 4 volt secondary lithium cells. J. Electrochem. Soc. 140, 1862–1870 (1993). 5. Noh, H.-J., Youn, S., Yoon, C. S. & Sun, Y.-K. Comparison of the structural and electrochemical properties of layered Li[NixCoyMnz] O2 (x = 1/3, 0.5, 0.6, 0.7, 0.8 and 0.85) cathodematerial for lithium- ion batteries. J. Power Sources 233, 121–130 (2013). 6. Chen, Z. & Dahn, J. R. Reducing carbon in LiFePO4/C composite electrodes tomaximize specific energy, volumetric Energy, and tap density. J. Electrochem. Soc. 149, A1184 (2002). 7. Padhi, A. K., Nanjundaswamy, K. S. & Goodenough, J. B. Phospho‐ olivines as positive‐electrode materials for rechargeable lithium batteries. J. Electrochem. Soc. 144, 1188–1194 (1997). 8. Ryu, H.-H., Park, K.-J., Yoon, C. S. & Sun, Y.-K. Capacity fading of Ni- Rich Li[NixCoyMn1–x–y]O2 (0.6 ≤ x ≤ 0.95) cathodes for high-energy- density lithium-ion batteries: bulk or surface degradation? Chem. Mater. 30, 1155–1163 (2018). 9. Zhou, A. et al. Hexacyanoferrate-type Prussian blue analogs: prin- ciples and advances toward high-performance sodium and potas- sium ion batteries. Adv. Energy Mater. 11, 2000943 (2021). 10. Hurlbutt, K., Wheeler, S., Capone, I. & Pasta, M. Prussian blue ana- logs as battery materials. Joule 2, 1950–1960 (2018). 11. Kraft, A. Some considerations on the structure, composition, and properties of Prussianblue: a contribution to thecurrent discussion. Ionics 27, 2289–2305 (2021). 12. Ivanov, V. D. Four decades of electrochemical investigation of Prussian blue. Ionics 26, 531–547 (2020). 13. Ma, L. et al. Accommodating diverse ions in Prussian blue analogs frameworks for rechargeable batteries: the electrochemical redox reactions. Nano Energy 81, 105632 (2021). 14. Li,W. J. et al. Chemical properties, structural properties, and energy storage applications of Prussian blue analogues. Small 15, 1900470 (2019). 15. Buser, H. J., Schwarzenbach, D., Petter, W. & Ludi, A. The crystal structure of Prussian blue: Fe4[Fe(CN)6]3.xH2O. Inorg. Chem. 16, 2704–2710 (1977). 16. Lu, Y., Wang, L., Cheng, J. & Goodenough, J. B. Prussian blue: a new framework of electrode materials for sodium batteries. Chem. Commun. 48, 6544–6546 (2012). Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 11 17. Zhang, L., Chen, L., Zhou, X. & Liu, Z. Towardshigh-voltageaqueous metal-ion batteries beyond 1.5 V: the zinc/zinc hexacyanoferrate system. Adv. Energy Mater. 5, 1400930 (2015). 18. Lee, H. W., Pasta, M., Wang, R. Y., Ruffo, R. & Cui, Y. Effect of the alkali insertion ion on the electrochemical properties of nickel hexacyanoferrate electrodes. Faraday Discuss. 176, 69–81 (2014). 19. Kim, D.-M. et al. Co-intercalation of Mg2+ and Na+ in Na0.69Fe2(CN)6 as a high-voltage cathode for magnesium batteries. ACS Appl. Mater. Interfaces 8, 8554–8560 (2016). 20. Liu, Y. et al. Sodium storage in Na-rich NaxFeFe(CN)6 nanocubes. Nano Energy 12, 386–393 (2015). 21. Li, W. J. et al. Facile method to synthesize Na-enriched Na1+xFeFe(CN)6 frameworks as cathode with superior electro- chemical performance for sodium-ion batteries. Chem. Mater. 27, 1997–2003 (2015). 22. You, Y., Yu, X., Yin, Y., Nam, K.-W. & Guo, Y. G. Sodium iron hex- acyanoferrate with high Na content as a Na-rich cathode material for Na-ion batteries. Nano Res. 8, 117–128 (2015). 23. Xiao, P., Song, J., Wang, L., Goodenough, J. B. & Henkelman, G. Theoretical study of the structural evolution of a Na2FeMn(CN)6 cathode uponNa intercalation.Chem.Mater. 27, 3763–3768 (2015). 24. Ling, C., Chen, J. & Mizuno, F. First-principles study of alkali and alkaline earth ion intercalation in iron hexacyanoferrate: the important role of ionic radius. J. Phys. Chem. C. 117, 21158–21165 (2013). 25. Imanishi, N. et al. Lithium intercalation behavior into iron cyanide complex as positive electrode of lithium secondary battery. J. Power Sources 79, 215–219 (1999). 26. Asakura, D. et al. Fabrication of a cyanide-bridged coordination polymer electrode for enhanced electrochemical ion storage abil- ity. J. Phys. Chem. C. 116, 8364–8369 (2012). 27. Shen, L., Wang, Z. & Chen, L. Prussian blues as a cathode material for lithium ion batteries. Chem. Eur. J. 20, 12559–12562 (2014). 28. Wu, X. et al. Low defect FeFe(CN)6 framework as stable host material for high performance Li-ion batteries. ACS Appl. Mater. Interfaces 8, 23706–23712 (2016). 29. Wang, B. et al. Gradient substitution: an intrinsic strategy towards highperformance sodiumstorage in Prussian blue-based cathodes. J. Mater. Chem. A 6, 8947–8954 (2018). 30. Tang, X. et al. Hierarchical sodium-rich Prussian blue hollow nanospheres as high-performance cathode for sodium-ion bat- teries. Nano Res. 11, 3979–3990 (2018). 31. Liu, Y. et al. A heterostructure coupling of bioinspired, adhesive polydopamine, and porous Prussian blue nanocubics as cathode for high-performance sodium-ion battery. Small 16, 1906946 (2020). 32. Wang, B. et al. Prussian blue analogs for rechargeable batteries. iScience 3, 110–133 (2018). 33. Herren, F., Fischer, P., Ludi, A. & Haelg,W. Neutron diffraction study of Prussian Blue, Fe4[Fe(CN)6]3·xH2O. Location of water molecules and long-range magnetic order. Inorg. Chem. 19, 956–959 (1980). 34. Kumar, A., Yusuf, S. M. & Keller, L. Structural and magnetic prop- erties of Fe4[Fe(CN)6]·4H2O. Phys. Rev. B 71, 054414 (2005). 35. Takachi, M., Fukuzumi, Y. & Moritomo, Y. Na+ diffusion kinetics in nanoporous metal-hexacyanoferrates. Dalton Trans. 45, 458–461 (2016). 36. Yu, Z. Y. et al. Unconventional CN vacancies suppress iron-leaching in Prussian blue analogue pre-catalyst for boosted oxygen evolu- tion catalysis. Nat. Commun. 10, 2799 (2019). 37. Pérez-Ramírez, J. et al. Evolution of isomorphously substituted iron zeolites during activation: comparison of Fe-beta and Fe-ZSM-5. J. Catal. 232, 318–334 (2005). 38. Dzwigaj, S., Stievano, L., Wagner, F. E. & Che, M. Effect of pre- paration andmetal content on the introduction of Fe in BEA zeolite, studied by DR UV–vis, EPR and Mössbauer spectroscopy. J. Phys. Chem. Solids 68, 1885–1891 (2007). 39. Chong, S. et al. Potassium nickel iron hexacyanoferrate as ultra- long-life cathode material for potassium-ion batteries with high energy density. ACS Nano 14, 9807–9818 (2020). 40. Zhang, C. et al. Potassium Prussian blue nanoparticles: a low-cost cathodematerial for potassium-ion batteries. Adv. Funct. Mater. 27, 1604307 (2017). 41. He, X. et al. A novel highly crystalline Fe4(Fe(CN)6)3 concave cube anodematerial for Li-ionbatterieswith high capacity and long life. J. Mater. Chem. A 7, 11478–11486 (2019). 42. Huang, Y. et al. A novel border-rich Prussian blue synthetized by inhibitor control as cathode for sodium ion batteries. Nano Energy 39, 273–283 (2017). 43. Pajerowski, D. M., Watanabe, T., Yamamoto, T. & Einaga, Y. Elec- tronic conductivity in berlin green and Prussian blue. Phys. Rev. B 83, 153202 (2011). 44. Qian, J., Zhou,M., Cao, Y., Ai, X. & Yang,H. NanosizedNa4Fe(CN)6/C composite as a low-cost and high-rate cathodematerial for sodium- ion batteries. Adv. Energy Mater. 2, 410–414 (2012). 45. Ma, Y. et al. High-entropy metal–organic frameworks for highly reversible sodium storage. Adv. Mater. 33, 2101342 (2021). 46. Griffith, D.W. T. &Galle, B. Fluxmeasurements ofNH3,N2OandCO2 using dual beam FTIR spectroscopy and the flux–gradient techni- que. Atmos. Environ. 34, 1087–1098 (2000). 47. Shang, Y. et al. Unconventional Mn vacancies in Mn–Fe Prussian blue analogs: suppressing Jahn-Teller distortion for ultrastable sodium storage. Chem 6, 1804–1818 (2020). 48. He, G. & Nazar, L. F. Crystallite size control of Prussian white ana- logues for nonaqueous potassium-ion batteries. ACS Energy Lett. 2, 1122–1127 (2017). 49. You, Y. et al. Subzero-temperature cathode for a sodium-ion bat- tery. Adv. Mater. 28, 7243–7248 (2016). 50. Wang, W. et al. Reversible structural evolution of sodium-rich rhombohedral Prussian blue for sodium-ion batteries. Nat. Commun. 11, 980 (2020). 51. Guo, X. et al. Water contributes to higher energy density and cycling stability of Prussian blue analogue cathodes for aqueous sodium-ion batteries. Chem. Mater. 31, 5933–5942 (2019). 52. Peng, J. et al. Defect-free-induced Na+ disordering in electrode materials. Energy Environ. Sci. 14, 3130–3140 (2021). 53. He, G., Huq, A., Kan, W. H. &Manthiram, A. β-NaVOPO4 obtained by a low-temperature synthesis process: a new 3.3 V cathode for sodium-ion batteries. Chem. Mater. 28, 1503–1512 (2016). 54. Zhang, Z. H., Ni, Y., Avdeev, M., Kan, W. H. & He, G. Dual-ion inter- calation to enable high-capacity VOPO4 cathodes for Na-ion bat- teries. Electrochim. Acta 365, 137376 (2021). 55. You, Y., Wu, X. L., Yin, Y. X. &Guo, Y. G. High-quality Prussian blue crystals as superior cathode materials for room-temperature sodium-ion batteries. Energy Environ. Sci. 7, 1643–1647 (2014). 56. Wu, X. et al. Single-crystal FeFe(CN)6 nanoparticles: a high capacity and high rate cathode for Na-ion batteries. J. Mater. Chem. A 1, 10130–10134 (2013). 57. Wang, L. et al. A superior low-cost cathode for a Na-ion battery. Angew. Chem. Int. Ed. 52, 1964–1967 (2013). 58. Mandal, B., Basumallick, I. & Ghosh, S. Synthesis, characterization and electrochemical studies on Li4Fe(CN)6 as cathode material for lithium batteries. J. Pure Appl. Chem. 5, 30–42 (2015). 59. Mandal, B. A novel aqueous Li4Fe(CN)6 cathode and hydrophobic ionic liquid electrolyte combined lithium-ion battery. Adv. Mater. Proc. 2, 745–748 (2017). 60. Toby, B. H. EXPGUI, a graphical user interface for GSAS. J. Appl. Crystallogr. 34, 210–213 (2001). Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 12 Acknowledgements This work was supported by the National Key Research and Develop- ment Program of China (2019YFA0205700), the National Natural Sci- ence Foundation of China (51602219); Tianjin Municipal Science and Technology Commission (No. 17JCYBJC42000); W.H.K. acknowledges the National Natural Science Foundation of China (11805034, 21704105, and U1930102), and the Natural Science Foundation of Guangdong Province (2017A030313021). The authors gratefully acknowledge the beamtime in ECHIDNA granted from Australian Centre for Neutron Scattering (ACNS) in ANSTO. Author contributions G.H. and Z.Z. conceived the idea and designed the experiments. Z.Z. conducted the characterizations, performed the electrochemical mea- surements, analyzed the results, and drafted the manuscript. H.C. and W.Y. performed the powder neutron diffraction for materials. M.A. pro- videdcomments on themanuscript.W.H.K. contributed to the theoretical calculations and analyses of powder neutron diffraction for materials. G.H. proposed the research, revised the manuscript and supervised this work. All the authors contributed to the interpretation of the results. Competing interests The authors declare no competing interests. Additional information Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41467-022-35376-1. Correspondence and requests for materials should be addressed to Wang Hay Kan or Guang He. Peer review information Nature Communications thanks Swastika Banerjee, Vilas Pol, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Reprints and permissions information is available at http://www.nature.com/reprints Publisher’s note Springer Nature remains neutral with regard to jur- isdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ licenses/by/4.0/. © The Author(s) 2022 Article https://doi.org/10.1038/s41467-022-35376-1 Nature Communications | (2022) 13:7790 13 https://doi.org/10.1038/s41467-022-35376-1 http://www.nature.com/reprints http://creativecommons.org/licenses/by/4.0/ http://creativecommons.org/licenses/by/4.0/ Lithiated Prussian blue analogues as positive electrode active materials for stable non-aqueous lithium-ion batteries Results and discussions Materials synthesis and characterizations Electrochemical performances of LiFeHCF samples Investigation of phase transitions of LiFeHCF-1 sample during cycling Lithium-ion and water molecules diffusion behavior Electrochemical performance of LiFeHCF-1-based positive electrodes in Li-ion coin cell configuration Methods Materials synthesis Materials characterizations Electrochemical measurements Calculation method Data availability References Acknowledgements Author contributions Competing interests Additional information