- Email: [email protected]
Higher voltage plateau cubic Prussian White for Na-ion batteries
Journal of Power Sources 324 (2016) 766e773
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Higher voltage plateau cubic Prussian White for Na-ion batteries Piernas-Mun ~ oz a, Elizabeth Castillo-Martínez a, *, Oleksandr Bondarchuk a, María Jose a filo Rojo a, b, ** Michel Armand , Teo a b
gico de Alava, ~ ano, Spain CICenergigune, Parque Tecnolo Albert Einstein 48, ED. CIC, 01050 Min nica, Universidad del País Vasco UPV/EHU, P.O. Box 664, 48080 Bilbao, Spain Departamento de Química Inorga
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Cubic Prussian White, A2-xFe2(CN)6 (A ¼ Na, K), are synthesized by a new method. Above 140 mAh g1 are achieved at 1C when tested as cathodes vs. Naþ/ Na. An interesting voltage increase of 0.35 V is observed for the potassium phase. Hybrid Na/K insertion seems to be the main cause of the voltage increase. In addition, it exhibits a high capacity retention (80% after 500 cycles).
a r t i c l e i n f o
a b s t r a c t
Article history: Received 23 February 2016 Received in revised form 8 May 2016 Accepted 11 May 2016 Available online 8 June 2016
Cubic sodium Prussian White, Na2-xFe2(CN)6$yH2O, and potassium Prussian White, K2-xFe2(CN)6$yH2O, are prepared following a mild synthetic methodology. While cubic symmetry is confirmed by XRD and TEM, IR and XPS show characteristic features different from Prussian Blue compositions. When investigated as cathode materials in sodium ion batteries, both compounds exhibit reversible capacities above 140 mAh g1 at 1C (ca. 80 mA g1). While sodium Prussian White shows better high rate capability (10C/ 0.1C ¼ 0.64), potassium Prussian White exhibits longer cycle stability, with up to 80% of capacity retention after 500 cycles. Interestingly, the potassium Prussian White phase also provides an increase of 0.35 V in the high voltage redox peak compared to the sodium Prussian White analogue ascribed to the preferential insertion of Kþ ions instead of Naþ, resulting in an increment of the gravimetric energy density. On the other hand, the insertion of Naþ seems to occur at the lower voltage plateau. This hybrid Naþ and Kþ insertion in the framework of potassium Prussian White is most likely the responsible of the long cycle stability as a consequence of synergistic effects. © 2016 Elsevier B.V. All rights reserved.
Keywords: Electrochemical energy storage Na-ion batteries Prussian white Prussian blue related Hybrid battery
1. Introduction * Corresponding author. Current address: Department of Chemistry, University of Cambridge, Lensfield Road, CB2 3ED, Cambridge, UK. ** Corresponding author. Departamento de Química Inorg anica, Universidad del País Vasco UPV/EHU, P.O. Box 664, 48080 Bilbao, Spain. E-mail addresses: [email protected], [email protected] (E. CastilloMartínez), [email protected] (T. Rojo). http://dx.doi.org/10.1016/j.jpowsour.2016.05.050 0378-7753/© 2016 Elsevier B.V. All rights reserved.
Na-ion batteries (SIB) are attracting the attention of the scientific community as alternative to the already well established Li-ion batteries [1]. Several cathodic compounds, including layered oxides, olivine type NaFePO4 structures and polyanions, among others, are under study with the aim of finding the most suitable
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
material for this new approach [2,3]. Undoubtedly, an interesting material for this type of emerging technology is the metal-organic open framework called Prussian Blue AM[M0 (CN)]6 (A ¼ alkali metals and M, M0 ¼ transition metals; sometimes, M ¼ M0 ; typically M0 ¼ Fe), in which alkali ions occupy half of the empty cubic cavities created by a cubic network of alternating M and M0 ions in the vertices and eC^Ne ligands on the edges [4e8]. These materials exhibit interesting properties such as low cost, easiness of synthesis and good stability. Even of more technological importance is the alkali-rich reduced phase, known as Prussian White (PW hereafter), Na2M[M0 (CN)]6. Unlike Prussian Blue (PB from now on), PW has all the cubic cavities occupied by alkali ions thereby preventing first cycle inefficiencies and the need of providing an additional source of alkali ions in the anode [4]. Several phases of the Prussian White, also called Everitt’s salt, have been recently investigated as low-cost cathode material for SIB. Moritomo et al. explored Na2-dMxFey(CN)6 (M ¼ Co/Mn, d 0.4), which showed capacities up to 135 mAh g1 and fast Naþ intercalation/de-intercalation with no structural phase transition [8e10]. Nonetheless, when M ¼ Mn and d 0.3, a structural transformation from cubic to rhombohedral occurs. However similar capacities to those previously reported are obtained [11]. On the other hand, a monoclinic Mn-based sodium rich PW, Na1.96Mn [Mn(CN)6]0.99∙▫0.01 ∙2H2O, studied by Cui et al., exhibited up to 209 mAh g1 at 40 mA g1 due to the possible insertion of 3 Naþ in the structure and the reduction from MnIIIeN^CeMnIII to MnIIeN^CeMnI. A structural phase transition from the assynthesized monoclinic PW to an orthorhombic phase along the de-insertion process has also been observed, to finally reach a cubic structure at the end of charge [12]. Besides, in an attempt to combine the good specific capacity observed in Na1.8Mn [Fe(CN)6]0.99∙▫0.01 and the coulombic efficiency of Na1.86Ni [Fe(CN)6]0.96∙▫0.04, Ma and coworkers prepared and tested the ternary Prussian blue analogue (PBA), Na1.76Ni0.12Mn0.88[Fe(CN)6]0.98∙▫0.02. It displayed a specific capacity of 123.3 mAh g1, coulombic efficiency of 92.71% in the first cycle (and above 98% after 5 cycles) and 83.3% of capacity retention after 800 cycles when cycled at 100 mA g1 [13]. Also, a highly crystalline rhombohedral CoeFe based Prussian White, Na1.85Co[Fe(CN)6]0.99 ∙▫0.01∙ 1.9H2O, delivers high capacity of 150 mAh g1 and cyclability of 90% over 200 cycles at 100 mA g1 [14]. Interestingly, the purely iron based PW cathode material, Na2xFe2(CN)6, has been recently reported only by Guo et al. and Goodenough et al. This compound presents a rhombohedral structure and delivers capacities of 150 mAh g1 at 10e25 mA g1 and fairly good cycling performance. About 90% of the initial capacity is retained after 200 cycles at 25 mA g1 in Guo’s report and 80% is observed after 750 cycles in Goodenough’s study, when it is charged at 75 mA g1 and discharged at 300 mA g1 [15,16]. Among the PW and PB framework compounds, those with potassium as alkali ion were the first known to form [4]. Indeed, the large ionic radii (1.38 Å for Kþ compared to 1.02 Å for Naþ) [17] fits well the size of the cavity and stabilizes the framework. Thus, the use of K-PW in a Na-ion cell is appealing for simplicity reasons. Nevertheless, an extra step would be required to avoid any possible electrolyte contamination with K ions, unless the use of a hybrid electrolyte is not detrimental to the electrochemical activity. The utilization of dual electrolytes containing two cations has been successfully deployed for NaeMg batteries, [18] as well as for Li-ion and Na-ion batteries. In fact, Li-ion activation facilitates Na-ion insertion into Ge-electrodes [19] and other sodium compounds such as Na3V2(PO4)2F3 have also been used in Li-ion cells [20]. Even aqueous mixed Li/Na electrolytes have been used in hybrid operation Li/Na-ion batteries in which each electrode only inserts one specific cation [21]. Consequently, the use of potassium PW in a Na-
767
ion cell seems an interesting approach towards low cost SIB. In this article, we describe a new synthetic method for obtaining nanoparticles (NPs) of the cubic alkali-rich phases sodium Prussian White (Na-PW from now on), Na2-xFe2(CN)6$yH2O, and potassium Prussian White (K-PW hereafter), K2-xFe2(CN)6$yH2O, by coprecipitation in water and heating under mild conditions (60 C). A comparative study on the electrochemical performance of both compounds as cathode materials for SIB is also reported. An advantageous increase in the gravimetric energy and long-cycle stability is reached when K-PW is used instead of Na-PW. Since these materials are only iron-based and, therefore, the more sustainable and less costly among the PW analogues, they are promising cathode materials to be used in SIB. 2. Experimental section 2.1. Synthesis The A-PW (A ¼ Na, K) phases were prepared by using a different methodology to that reported by Jiang et al. [22] for the synthesis of K-PW, in order to avoid the formation of some Fe2O3 as impurity. The new synthetic route we developed is described below. Na-PW NPs, ideally Na2FeII[FeII(CN)6]$xH2O, were obtained using a single step method. In inert atmosphere, 100 mL of a freshly prepared pale yellow Na4[Fe(CN)6]$10H2O 40 mM aqueous solution were directly mixed with 100 mL of a pale yellow-greenish FeCl2 40 mM aqueous solution, forming a light blue suspension, according to the reaction described in equation (1). The partial iron oxidation from Fe2þ to Fe3þ was avoided by adding a small amount (approximately 2 mg) of ascorbic acid in the solution. The ascorbic acid acts as reducing agent, preserving the Prussian White phase and hindering the evolution towards Prussian Blue. Then, the suspension was mixed with an excess of NaCl to increase the Na content in the structure. After ca. 4 h heating at 60 C under N2 flow, the mixture was filtered, washed with H2O (3 10 ml) and acetone (3 10 ml) and dried, obtaining pale blue crystal-like aggregates of Na-PW NPs, Na2-xFeII1þ(x/2)[FeII(CN)6]$yH2O (see equation (1)). (1 þ (x/2)) FeCl2 þ Na4[Fe(CN)6]$10H2O / Na2-xFeII [FeII(CN)6]$yH2O þ (2 þ x) NaCl (aq.)
1þ(x/2)
(1)
K-PW nanoparticles, were prepared following the same procedure deployed for obtaining Na-PW NPs, just replacing the starting sodium ferrocyanide by the potassium salt, K4[Fe(CN)6]$3H2O and NaCl by KCl. In this case, a light green suspension is formed but the final product presents a pale blue colour. The reaction that takes place is analogous to that of equation (1) (see equation (2)). (1 þ (x/2)) FeCl2 þ K4[Fe(CN)6]$3H2O / K2-xFeII yH2O þ (2 þ x) KCl (aq.)
II 1þ(x/2)[Fe (CN)6]$
(2)
2.2. Structural, morphological and physico-chemical characterization PXRD patterns were recorded on a Bruker D8 Discover X-ray diffractometer, using lCu-Ka ¼ 1.54056 Å radiation in the 5e80 2q range, with a step width of 0.0194 . IR spectra were collected in the range of wavenumber ¼ 4000-650 cm1 in absorption mode in an Agilent Cary 630 FTIR spectrometer (Agilent Technologies), placed inside an Ar-filled glove box. SEM images were obtained in a FEI Quanta 200 F SEM operated at 30 kV and equipped with an Apollo 10 SSD Energy Dispersive X-ray (EDX). TEM images and electron diffraction patterns were taken using a Philips CM200. Chemical
768
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
compositions of the materials were determined using, respectively, atomic absorption and emission spectroscopy to calculate the Na: Fe and K: Fe ratios. Elemental analyses have been performed in a Euro Elemental Analyzer (CHNS), to check the H, C and N percentage present in the samples. TGA (NETZSCH STA 449 F3 Jupiter) was used to collect the thermogravimetric curves. Experiments were performed in the temperature range from 30 to 325 C using N2 atmosphere. XPS spectra were acquired in transmission mode with pass energy of 20 eV in an UHV system equipped with a hemispherical electron energy analyzer PHOIBOS 150 (SPECS) and a twin Al/Mg anode - X-ray source XR50 (SPECS) operated at 12 keV and power 100 W. The size of the acceptance area of the analyzer was ca. 2 mm in diameter and its energy scale was calibrated against Cu 2p3/2, Ag 3d5/2 and Au4f7/2 lines of standard AueAgeCu sample. The powder samples were fixed onto the Mo sample plate using double-sided carbon tape. To avoid contact of the samples with ambient, the samples were transferred from the glove box where they were stored to the XPS chamber in an Ar filled air-tight container. Raman analyses were conducted in the range 100e4000 cm1 with resolution 4 cm1 using a Ranishaw InVia spectrometer equipped with a 532 nm green laser operated at a power of 0.5 mW. The laser spot size was approximately 1 mm in diameter. In order to avoid oxidation of the samples due to exposure to the high power density laser radiation the measurements were performed also in an Ar-filled compartment. 2.3. Electrochemical characterization The electrochemical measurements of both compounds A-PW (A ¼ Na, K) were performed in CR2032 coin cells, using a Maccor® series 4000 battery tester. Positive electrodes consisted of 70 wt (%) of active material (either Na-PW or K-PW) with 20 wt (%) conductive C (Ketjen black® EC-600JD) and 10 wt (%) binder (PVDF - Solef ®), mixed with Nmethyl-2-pyrrolidone (NMP). The slurries were homogeneously coated on an Al current collector and dried overnight at 80 C under vacuum. Electrodes were cut as discs 13 mm in diameter and pressed at 5 tons for few seconds. For measurements carried out with powder, positive electrodes were prepared by simply mixing 80 wt (%) of the active material (either Na-PW or K-PW) with 20 wt (%) of conductive C (Ketjen black® EC-600JD) through grinding. Then, they were directly deposited in the bottom casing of the cell. Unlike positive electrodes described in the previous paragraph, powder materials are binder-free and current collector free and no calendering or extra processing, other than vacuum drying, were implemented. When the feasibility of Na-PW and K-PW was explored vs. Naþ/ Na, i.e. for SIB, measurements were performed in the voltage range from 2.25 to 4.25 V. Metallic sodium was used as counter and reference electrode. The electrolyte deployed for both A-PW (A ¼ Na, K) was 1 M NaPF6 in EC: PC: FEC 49:49:2% vol [23], impregnating a Whatmann® GF-D separator. To study the nature of the higher voltage plateau observed in KPW, its electrochemical properties as cathode in K-ion batteries (KIB) were also investigated. Metallic potassium was used instead of sodium as counter and reference electrode. 1 M KPF6 in EC: DMC 50:50% vol. was the selected electrolyte. No FEC was added in this case, as no long-term cyclability was required. 3. Results and discussion 3.1. Morphological, structural and physico-chemical study Light blue Na-PW and K-PW NPs were prepared following the procedure described in the experimental section. To calculate the
Na: Fe ratio of Na-PW and K: Fe of K-PW, atomic absorption and emission spectroscopy techniques were deployed, respectively. We further confirmed that the chemical formula of both PW is consistent with the CNH analysis. The accurate chemical composition achieved was Na1.70Fe2.15(CN)6 for Na-PW and K1.59Fe2.20(CN)6 for K-PW. The powder X-ray diffractograms of Na-PW and K-PW are displayed in Fig. 1a and b. Unlike the purely iron based Prussian White previously reported in the literature [15,16], that possess a rhombohedral lattice (R 3) (see Fig. 1c), the materials herein synthesized seem to present a cubic structure profile. Rietveld refinements support this statement (see Figure S1). The fittings have been performed refining lattice parameters, constrained atomic positions and occupancies of a starting cubic Prussian blue model and results in a lattice parameter a ¼ 10.350(5) Å for Na-PW and a ¼ 10.117(6) Å for K-PW. To corroborate the formation of the Na-rich phase Prussian White and discard that cubic Prussian Blue was obtained instead, other spectroscopic and diffraction techniques were deployed. IR spectra (Fig. 1d, e) show that cyanide (eC^N) stretching vibrational bands appear at 2068 and 2067 cm1, respectively for Na-PW and K-PW. This deviation towards lower wavenumber with regard to Prussian blue (ca. 2078 cm1) [25] suggests longer C^N distances when the two iron atoms approach to oxidation state 2þ in fully sodiated/potassiated PW (the Fe(II) covalently bonded to C (FeIIeC) and the Fe2þ ionically linked to N (Fe2þeN)). These results are in agreement with those presented by Takachi [8] and Wu [28], who observed a higher wavenumber of the eC^N band when iron oxidation state increases in a Prussian blue structure. In addition, the IR spectra revealed the presence of a small amount of water in both pristine materials. H, C, N analyses point to an approximate H2O content of ca. 4 and 2 molecules per formula unit, accordingly for the sodiated and potassiated phases. However, this water content could be influenced by the exposure of the samples to air when the analysis is carried out. Indeed, a negligible amount of H2O is estimated both in Na-PW and K-PW from the TGA curves (see Figure S2). SEM images show the morphology of Na and K-PW (Fig. 1f, g) and highlight a heterogeneous particle size distribution that ranges from few micrometers (2e5 mm) to 100 mm. TEM images, nonetheless, provide a more accurate particle size value and prove the formation of primary nanoparticles which are aggregated into the larger grains observed by SEM. Crystalline domains of 35e40 and 20 nm are observed, respectively, for Na-PW and K-PW, which agrees with the very broad reflections displayed by PXRD. Besides, electron diffraction patterns were recorded to determine a possible distortion from cubic to rhombohedral symmetry in individual particles. Indexation of the ring-type electron diffraction patterns shown by the nanocrystals of Na-PW (Fig. 1h) and K-PW (Fig. 1i) confirmed the cubic structure (F m 3 m) in both materials. EDX performed in the TEM to the individual crystallites analyzed by electron diffraction yielded approximate Na: Fe and K: Fe ratios of 1.3: 2 and 1.67: 2. The cubic symmetry must be related to the lower than 2 Na/K content on Na/K-PW. This assumption is consistent with what Guo et al. have reported, who observed that the crystal structure evolved from cubic to rhombohedral when changing from Na1.24Fe1.89(CN)6 to Na1.63Fe1.89(CN)6, demonstrating that the PB framework tends to form lower symmetry structures the greater the amount of Na it contains [15]. This change was also observed in the Mn-based Prussian White Na2-dMn[Fe(CN)6], in which the phase transition from cubic to rhombohedral occurred at d 0.3 [11]. In fact, the formula of our Na-PW compound, Na1.70Fe2.15(CN)6 or Na1.58Fe2(CN)5.58, is very close to that of Guo’s rhombohedral phase but still on the cubic side of the transition. On the other hand, the fully potassiated K-PW phase is reported to be monoclinic [22]. Consequently, despite having obtained alkali rich PW phases, the
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
769
Fig. 1. X-ray diffractograms of Na-PW (a) and K-PW (b). To compare the experimental diffractograms obtained with that already reported for Sodium Prussian White, Na-PW, a profile with lattice parameters and space group R-3 from ref. [24] was simulated (c). IR spectra of Na-PW (d) and K-PW (e), highlighting the more significant bands (C^N (cyanide) stretching vibrational band and OeH (hydroxyl) stretching vibrational and deformation band). SEM images of Na-PW (f) and K-PW (g) morphology. Electron diffraction patterns of Na-PW (h) and K-PW (i).
alkali content in our compounds is far from reaching the full site occupancy of 2 and thus they maintain the cubic structure. Raman spectra of the pristine Na-PW and K-PW compounds were collected in the wavenumber range 1900e2300 cm1, to evaluate the eC^N stretching band. The results are depicted in Fig. 2a and the deconvolution of the Raman spectra are shown in Fig. S3 (Supplementary data). The Raman scattering frequencies change both position and relative intensities when comparing NaPW and K-PW. A couple of distinctive bands are clearly distinguished at 2116 and 2148 cm1 and at 2090 and 2131 cm1, for NaPW and K-PW, respectively. A shoulder at lower wavenumber can also be guessed at ca. 2060e2075 cm1. Red shift on the bands when changing from Na-PW to K-PW could be indicative of the C^N bond softening due to the larger ionic radii of K ions. Although the presence of three bands in the Raman spectra is usually found
in rhombohedral Fe-based PW, the values obtained in this case are slightly different from those observed in previous works at 2070, 2110 and 2135 cm1 [15,16,26]. This suggests that despite the average crystal structure is cubic as determined from PXRD and electron diffraction, local rhombohedral distortions might be present in both compounds. Additional analyses were conducted. X-ray photoelectron spectroscopy (XPS) spectra provided determinant evidence of the formation of Prussian White phases distinct from Prussian Blue (see Fig. 2b). XPS analyses at the Fe 2p edge were performed on A-PW and A-PB (A ¼ Na, K) samples. The spectra of both Na- and K-PW are virtually identical as it can be observed in Fig. 2b, indicating the same electronic configuration of the iron in both materials. Analogously, both Na- and K-PB spectra are as well identical between them. Thus, the substitution of alkali metals (Na by K) does not
Fig. 2. Raman spectra focusing on the cyanide (C^N) band region (a) and XPS at the Fe 2p edge (b) of Na-PW (black) and K-PW (purple). XPS spectra of Na-PB (black) and K-PB (purple) are also included in dashed lines in Fig. 2b. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
770
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
affect the shape of Fe 2p spectra in both types of materials. For all the phases, A-PW and A-PB (A ¼ Na, K), the Fe 2p region has two split orbit components 2p3/2 and 2p1/2 with maxima at 709 and 722 eV, coincident with those previously reported for low spin Fe2þ of [Fe(CN)6]4- in Prussian Blue species [27,28]. The satellite pattern, however, varies when we compare PB to PW. The presence of a small peak at ca. 713 eV in Na/K-PB has been assigned in the literature either to high spin Fe3þ [27,29] despite the intensity is not as large as that of the Fe2þ signal, or to Fe2þ bonded to oxygen [30]. The latter can be attributed to the presence of water molecules. The peak appearing at 713 eV is however not observed for Na/K-PW which agrees with the absence of high spin Fe3þ and much lower content of H2O (see Figure S4 for the full XPS spectra: at Na 1s edge, K 2p edge, Fe 2p edge, C 1s edge, N 1s edge and O 1s edge). Instead, another satellite at ca. 724 eV, which could be related to the high spin state of the Fe2þ ions, can be observed for the PW species. Therefore, the PB and PW phases can be distinguished by the Fe 2p data obtained from XPS as we are showing here for the first time. 3.2. Electrochemical performance The electrochemical behavior of the materials Na-PW and K-PW was investigated for SIB (vs. Naþ/Na) through galvanostatic measurements in the cathodic voltage range from 2.25 to 4.25 V. Fig. 3a compares the voltage profiles corresponding to the first galvanostatic charge-discharge of Na-PW and K-PW in a sodium half-cell, using 1 M NaPF6 in EC: PC: FEC 49:49:2% vol as electrolyte [23]. Typical OCV values range from ca. 2.85e2.9 V for Na-PW to 2.9e2.95 V for K-PW. The lower OCV obtained compared to the usual 3.2 V observed in Prussian Blue is indicative of the higher alkali content in the PW phases. Therefore, the electrochemical measurements begin with a charge process. The large first low voltage plateau corroborates an almost complete filling of the alkali ion cavities in both cases. For Na-PW, two slopy plateaus centered at 3.0 and 3.3 V are distinguished along the charge (see Fig. 3a, in black). Upon discharge, the higher voltage redox process keeps on occurring centered at 3.3 V (see dQ/dV vs. V curve in the inset of Fig. 3a). Nonetheless, the redox reaction at lower voltage experiences a small shift towards lower potential values, from 3.0 to 2.8 V, probably due to the insulating character of the sodiated phase that leads to a higher polarization. These voltage values above mentioned for the redox processes are in-between those reported by Guo-Guo et al., who showed mainly a single plateau at 2.8e2.9 V [15], and those found by Goodenough et al., where a couple of
plateaus are visible in the range from 3.0 to 3.3 V [16]. For K-PW, a couple of plateaus are also observed while oxidizing at 3.0 and 3.7 V and the same two plateaus appear at 2.85 and 3.65 V upon discharge. Again, the lower voltage redox process presents a higher polarization. It is worth mentioning the increase occurring in the voltage value of the upper voltage process when moving from Na-PW to KPW. A significant difference of 0.35 V is noticed in the dQ/dV vs. voltage curve (see Fig. 3a inset and Fig. 3b). In addition, these plots manifest the complexity of the redox processes that are taking place (in some cases, each redox process consisted of two step reaction), as well as the repeatability and reproducibility of these along the cyclability. To explain the convenient voltage difference existing in the upper voltage redox process between Na-PW and K-PW, two hypotheses can be considered. On one hand, the presence of a different cation in the materials could be responsible of this positive potential shift. The effect of the cation on the increasing cell voltage, and therefore the energy density, has already been reported for other systems, such as quinones [31] In our particular case, for the K-PW electrodes, once the material is completely charged and consequently the lattice emptied of Kþ, the alkali metal being inserted when starting to discharge could be either Naþ or Kþ. Although the amount of Naþ present is several fold higher with respect to Kþ (i.e. 26.5 times bigger, calculated from the volume of the electrolyte used and the weight of electrode material), the selectivity of the PB framework for Kþ is likely higher [32]. PB is known as a decontaminant for 137Cs, being selective in biofluids for this even larger cation [33]. Probably in our case, the bigger ionic size of Kþ stabilizes somehow the molecular structure of the metal-organic framework. However, the low voltage redox process of K-PW vs. Naþ/Na occurs at similar voltage to that observed for Na-PW electrodes. This fact suggests that either Naþ insertion is taking place during this reaction or that the Kþ second intercalation happens at the same potential value of Naþ. On the other hand, structural variances in the lattice, from Na-PW to K-PW, could also be responsible of the different voltage plateau. Although both starting phases are cubic according to PXRD and electron diffraction (Fig. 1), the data obtained from Raman spectra (Fig. 2a) suggest that different local distortions appear and those could be translated into all the de-intercalated/intermediate phases. We can rule out an effect coming from the negative electrode, as the reduction potential of Kþ is more negative than that of Naþ, and thus the only redox equilibrium taking place is Naþ þ e ⇔ Na .
Fig. 3. a) First cycle voltage profile, (inset) first cycle dQ/dV vs. voltage curves and b) second cycle dQ/dV vs. voltage curves of Na-PW (in black) and K-PW (in violet), when cycling vs. Naþ/Na in the voltage range of 2.25e4.25 V, using 1 M NaPF6 in EC: PC: FEC 49:49:2% vol. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
To investigate whether the difference in the upper voltage plateau between Na-PW and K-PW was due to the insertion of a different alkali metal despite both materials being cycled vs. metallic sodium (Naþ/Na) using a Naþ containing electrolyte, the next procedure was followed: A K-PW electrode was electrochemically charged up to 4.2 V, with the purpose of reaching the alkali-empty phase (Berlin Green, BG). Once this phase was attained, the cell was opened, the electrode was washed with the corresponding carbonate solvent (PC) and when dried, it was reassembled again with 1 M NaPF6 EC: PC: FEC vs. Naþ/Na. As it can be seen in Fig. 4, the electrochemical performance of this BG phase is almost identical to that exhibited by Na-PW. This fact indicates that when K-PW is directly cycled against Naþ/Na, it is not Naþ but Kþ the alkali ion inserting and de-inserting into/from the structure, at least in the high voltage redox process. Therefore, it can be stated that the difference in the voltage value corresponds to the different alkali ions intercalated. To determine then the nature of the lower voltage plateau, KPW was cycled vs. metallic potassium (Kþ/K). Fig. 5 illustrates the dQ/dV vs. voltage profile of K-PW tested against Naþ/Na and Kþ/K. The difference in the standard reduction potentials between Naþ/ Na and Kþ/K is 0.211 V in aqueous solution [34]. Although the voltage difference may change in organic media, this 0.211 V shift has been considered to recalculate the voltage shown by K-PW when cycled vs. Naþ/Na for comparison with that of K-PW cycled in a K-ion battery (Fig. 5). The voltage values are presented both vs. Naþ/Na and Kþ/K. It is clear that the redox peaks are somehow coincident in the higher voltage plateau, which agrees with Kþ insertion, as demonstrated previously. Nonetheless, in the lower voltage plateau, the redox process occurs at different potential values, revealing that the species inserted during this process might be more likely Naþ or a mixture of both cations, Naþ and Kþ as two redox peaks appear when cycling vs. Naþ/Na and only 1 peak when cycling vs. Kþ/K. In addition, it is worth noting the excellent properties obtained when K-PW is tested vs. Kþ/K, proving its viability to be used as cathode in K-ion batteries (KIB), as Eftekhari [35] had reported for Prussian Blue in 2004. EDX analyses in K-PW electrodes cycled vs. Naþ/Na and strategically stopped at certain points of the electrochemical discharge curve have been conducted to chemically confirm the hybrid nature of the cation insertion in this material. Prior to the EDX collection,
Fig. 4. dQ/dV vs. V profile comparative between the electrochemical behavior of NaPW (continuous line in black), K-PW (dashed line in purple) and K-PW charged, washed with PC and cycled again vs. Naþ/Na (red dots). The electrolyte used in all experiments here plotted was 1 M NaPF6 in EC: PC: FEC 49:49:2% vol. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
771
Fig. 5. dQ/dV vs. V profile comparative between the electrochemistry behavior of KPW against Kþ/K (continuous line in purple) and Naþ/Na (dashed line in purple). The electrolyte used for KIB was 1 M KPF6 in EC: DMC 50:50% vol. while 1 M NaPF6 in EC: PC: FEC 49:49:2% vol was the mixture selected for SIB. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
the electrodes were properly rinsed with PC for several times to remove any residual Naþ trace that could stem from the electrolyte salt and, then, dried overnight at 80 C under vacuum. The K/Na ratios thus increase from BG to PB phase supporting our hypothesis that Kþ is the metal being mainly inserted at the high voltage redox process, and decrease from PB to PW, which agrees with Naþ being the alkali mostly inserted at the low voltage plateau (see Supplementary data, section F, for further details). The hypothesis of structural variances in the lattice and/or different structural transformations upon cycling is not discarded, however, since both effects could be additive to the cation effect already demonstrated. Future work in understanding the mechanism behind it is necessary to determine whether structural effects are contributing. In addition, further insights into the mechanism of reaction could reveal the nature of the cation (either Naþ or a mixture of Naþ/Kþ) which is being inserted/de-inserted in the lower voltage plateau of K-PW. This would agree with the beneficial cooperative effects which have already been observed for other cation combinations, such as Li and Na [19e21] or Mg and Na [18]. Notwithstanding, this assumption cannot be confirmed before analyzing the reaction mechanism of these phases. C-rate capabilities of the two Na-PW and K-PW against metallic sodium (Naþ/Na) are shown in Fig. 6a and b. When cycled in powder form (Fig. 6a), both materials exhibit reversible capacities up to 145e150 mA h g1 at C/10 (being C ¼ 85.4 mAh g1 per 1Naþ insertion per formula unit for Na-PW and 77.5 mAh g1 for K-PW) with coulombic efficiencies of ca. 95e97% and 97e99.4%, respectively. These specific discharge capacities are analogous to those already reported in the literature for the rhombohedral purely iron based Na-PW [15,16]. Capacities between 115 and 120 mA g1 are achieved at moderate rate, as 1C. At higher current densities, such as 10C (nearly 1 A g1), the capacity retention displayed by Na-PW, 75 mAh g1, is larger than that for K-PW, 50 mAh g1. However, both materials demonstrate a good stability when cycled back at C/ 10. C-rate capabilities of Na-PW and K-PW, when the materials are coated over Al current collector forming electrodes are displayed in Fig. 6b. In this case, although specific capacities similar to those achieved in powder are reached, a negative effect in the coulombic efficiency is observed at low current densities, C/10. This inefficiency issue is reduced, nonetheless, once the materials operate
772
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
Fig. 6. C-rate capabilities of Na-PW (black squares) and K-PW (violet circles) vs. Naþ/Na cycled in powder (a) and coated over Al current collector (b). c) Long-term cyclabilities (capacities and coulombic efficiency values) of Na-PW (black squares) and K-PW (violet circles) when cycling vs. Naþ/Na in the cut off voltage DV ¼ 2.25e4.25 V. The theoretical capacity 1C corresponds to 85.4 mA g1 or 77.5 mA g1/1Naþ insertion per f.u., respectively for Na-PW and K-PW. 1 M NaPF6 in EC: PC: FEC 49:49:2% vol. was the electrolyte utilized in all experiments here displayed. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
at moderate and higher capabilities, finding analogous values than those in powder. In addition, when contrasted to the powder form, better capacity retention is evident for K-PW at 1C (achieving 126 mAh g1), and for Na-PW (88 mAh g1) and K-PW (72 mAh g1) at 10C. The adverse effect on the coulombic efficiency at low current densities could tentatively be attributed to Al current collector corrosion, though PF 6 usually protects the native Al2O3 layer. The addition of ACl (A ¼ Na, K) to synthesize the material, although beneficial to raise the alkali content, can imply the risk of corroding the Al used as substrate from trace Cl. Consequently, it is paramount to thoroughly rinse the material with H2O to remove all the Cl. In our case, no Cl presence was detected in pristine K-PW and only some traces in some isolated particles at Na-PW. The good performance of Na-PW and K-PW along the cycle-life at moderate current density (see Fig. 6c) is also remarkable. Roughly 140 mAh g1 are achieved during the first cycles at 1C. In addition, 63.4% of the initial capacity is retained after 500 cycles with >98% of coulombic efficiency for Na-PW. Although the capacity retention obtained for the cubic Na-PW is lower than that presented by Guo-Guo [15] and Goodengouh [16], the coulombic efficiency is far superior. Besides, comparing cubic Na-PW with other Prussian White systems, such as those Mn-Fe [10,11], Co-Fe [9,14] or Mn-Mn [12] based, the capacity retention and coulombic efficiency values are higher as well. On the other hand, the hybrid K-PW exhibits enhanced properties compared to Na-PW when cycled vs. Naþ/Na, keeping 80% of the initial capacity retention after 500 cycles with >97.5% of coulombic efficiency during the first 30 cycles and >99.5% in cycle 500. In addition, it shows a fairly good stability in a long-term cyclability of ca. 1000 h ¼ 42 days of autonomy. The coulombic efficiency obtained, far from the optimum >99.9%, could be fairly improved however by simply assembling full cells, since the deployment of Na metallic as negative and counter electrode is known to react detrimentally with the electrolyte forming dendrites, resulting in inefficiency [36e39]. Another option to bear in mind could be to replace the Al current collector deployed in here by a carbon cloth current collector substrate frequently used in Prussian Blue systems [12]. These tasks will be addressed in a future work. In any case, the interesting behavior exhibited at moderate current densities, which are typical for market applications, allow us to consider K-PW as suitable and desirable electrode material for a real use in a hybrid K/Na-ion battery.
the formation of materials with cubic structure for purely iron based PW, unlike those reported previously in the literature. The cubic symmetry is confirmed by PXRD and electron diffraction experiments in the TEM while IR and XPS show the PW character distinct from PB. The benefit of obtaining cubic symmetry is major, since the species formed during the cycling process (PB and BG) are cubic as well, and no structural transitions are required in principle. The materials exhibit fairly good electrochemical properties, when cycled vs. Naþ/Na in the cathodic range from 2.25 to 4.25 V using 1 M NaPF6 in EC: PC: FEC. Specific capacities over 140 mA h/g at 1C with reasonably high coulombic efficiencies values (above 98%) were exhibited by Na-PW and K-PW. Full cell constructions and the use of carbon based current collectors could help to improve the good but limited coulombic efficiency here obtained with Al current collector and with the use of metallic Na as counter and reference electrode. In addition, it is important to highlight that an interesting increase in the gravimetric energy density is observed when K-PW replaces Na-PW, due to a raise of 0.35 V in the potential at which the high voltage redox process occurs. The reason behind is so far ascribed to the insertion and de-insertion of Kþ instead of Naþ, but further studies on the reaction mechanism must be performed to either discard or confirm an additional effect due to a structural change and synergetic effects related to the presence of both cations in the medium. This higher voltage, along with the better capacity retention (80% after 500 cycles), makes K-PW an even more attractive cathode material than Na-PW for real market applications.
Acknowledgements This work was financially supported through projects LINABATT (ENE2013-44330-R) and Etortek 14 CIC Energigune. SGIker technical and human support (UPV/EHU, MICINN, GV/EJ, ESF) is gratefully acknowledged for the help with AAS and elemental analysis (H, C, N). The authors would also like to acknowledge Jon Ajuria and Frederic Aguesse, for their help with SEM images collection and Piernas Mun ~ oz Ana Martinez Amesti for recording TEM. Ma Jose thanks the Basque Government for the grant corresponding to “Nuevas Becas y Renovaciones para el Programa Predoctoral de n de Personal Investigador” (PRE.2013.1.790, MOD A). Formacio
4. Conclusions Appendix A. Supplementary data Na-PW and K-PW have been synthesized following a new mild synthetic route, at low temperature (60 C), in aqueous media and avoiding the addition of acidic products. This methodology allows
Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2016.05.050.
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
References [1] V. Palomares, P. Serras, I. Villaluenga, K. Hueso, J. Carretero-Gonzalez, T. Rojo, Energy Environ. Sci. 5 (2012) 5884e5901. [2] V. Palomares, M. Casas-Cabanas, E. Castillo-Martínez, M.H. Han, T. Rojo, Energy Environ. Sci. 6 (2013) 2312e2337. [3] N. Yabuuchi, K. Kubota, M. Dahbi, S. Komaba, Chem. Rev. 114 (2014) 11636e11682. [4] J.F. Keggins, F.D. Miles, Nature (1936) 577. [5] D.B. Brown, D.F. Shriver, Inorg. Chem. 18 (1969) 37e42. [6] D. Yang, J. Xu, X-Z. Liao, H. Wang, Y.-S. He, X.-F. Ma, Chem. Comm 51 (2015) 8181e8184. [7] Y. Lu, L. Wang, J. Cheng, J.B. Goodenough, Chem. Comm 48 (2012) 6544. [8] M. Takachi, T. Matsuda, Y. Moritomo, Jpn J. Appl. Phys. 52 (2013) 090202. [9] M. Takachi, T. Matsuda, Y. Moritomo, Appl. Phys. Express 6 (2013) 025802. [10] M. Takachi, T. Matsuda, Y. Moritomo, Chem. Commun 49 (2013) 2750e2752. [11] L. Wang, Y. Lu, J. Liu, M. Xu, J. Cheng, D. Zhang, J.B. Goodenough, Angew. Chem. Int. Ed 52 (2013) 1e5. [12] H.-W. Lee, R.Y. Wang, M. Pasta, S.W. Lee, N. Liu, Y. Cui, Nat. Commun 5 (2014) 5280. [13] D. Yang, J. Xu, X.-Z. Liao, Y.-S. He, H. Li, Z.-F. Ma, Chem. Commun 50 (2014) 13377e13380. [14] X. Wu, C. Wu, C. Wei, L. Hu, J. Qian, Y. Cao, X. Ai, J. Wang, H. Yang, ACS Appl. Mat. Interfaces (2016), http://dx.doi.org/10.1021/acsami.5b12620. [15] Y. You, X.-Q. Yu, Y.-X. Yin, K.W. Nam, Y.-G. Guo, Nano Res. 8 (2015) 117e128. [16] L. Wang, J. Song, R. Qiao, L.A. Wray, M.A. Hossain, Y.-D. Chuang, W. Yang, Y. Lu, D. Evans, J.-J. Lee, S. Vail, X. Zhao, M. Nishijima, S. Kakimoto, J.B. Goodenough, JACS 137 (2015) 2548e2554. [17] R.D. Shannon, Acta Cryst A32 (1976) 751e767. n ~ ez, M.V. Kovalenko, Chem. Mater 27 (2015) [18] M. Walter, K.V. Kravchyk, M. Iba 7452e7458. [19] A. Kohandehghan, K. Cui, M. Kupsta, J. Ding, E.M. Lotfabad, W.P. Kalisvaart, D. Mitlin, Nano Lett. 14 (2014) 5873e5882.
773
[20] W. Song, X. Ji, Z. Wu, Y. Zhu, F. Li, Y. Yao, C.E. Banks, RSC Adv 4 (2014) 11375e11383. [21] L. Chen, Q. Gu, X. Zhow, X. Lee, Y. Xia, Z. Liu, Sci. Rep 3 (2013) 1946. [22] M. Hu, J.S. Jiang, Mat. Res. Bull. 46 (2011) 702e707. ~ oz, E. Castillo-Martínez, J.L. Go mez-C [23] M.J. Piernas-Mun amer, T. Rojo, Electrochim. Acta 200 (2016) 123e130. [24] J. Song, L. Wang, Y. Lu, J. Liu, B. Guo, P. Xiao, J.-J. Lee, X.-Q. Yang, G. Henkelman, J.B. Goodenough, JACS 137 (2015) 2658e2664. [25] L. Xia, R.L. McCreery, Electrochem. Soc. 146 (1999) 3696e3701. [26] Y. You, X.-L. Wu, Y.-X. Yin, Y.-G. Guo, Energy Environ. Sci. 7 (2014) 1643e1647. [27] M. Datta, A. Datta, J. Phys. Chem. 94 (1990) 8203e8207. [28] Q. Wu, G. Wu, L. Wang, W. Hu, H. Wu, Mat. Sci. Semicond Process 30 (2015) 476e481. [29] E. Desimoni, B. Brunetti, Chemosensors 3 (2015) 70e117. [30] X. Li, J. Liu, A.I. Rykov, H. Han, C. Jin, X. Liu, App. Catal. B Environ. 179 (2015) 196. ndez-Burgos, G.G. Rodríguez-Calero, W. Zhow, S.E. Burkhardt, [31] K. Herna ~ a, J. Am. Chem. Soc. 135 (2013) 14532e14535. H.D. Abrun [32] V. Krishnan, A.L. Xidis, V.D. Neff, Anal. Chim. Acta 239 (1990) 7e12. [33] J.M. Arnal, M. Sancho, B. García-Fayos, Desalination 321 (2013) 22e27. [34] D.R. Lide, W.M. Haynes, CRC Handbook of Chemistry and Physics (Internet version), 90th Ed., CRC press, Boca Raton (Florida) e USA, 2010, pp. 8e22 (1212). [35] A. Eftekhari, J. Power Sources 126 (2004) 221e228. [36] D.I. Iermakova, R. Dugas, M.R. Palacín, A.J. Ponrouch, Electrochem. Soc. 162 (2015) A7060eA7066. [37] A. Rudola, D. Aurbach, P. Balaya, Electrochem. Commun. 46 (2014) 56. [38] N. Weadock, N. Varongchayakul, J. Wan, S. Lee, J. Seog, L. Hu, Nano Energy 2 (2013) 713. [39] A. Ponrouch, D. Monti, A. Boschin, B. Steen, P. Johansson, M.R. Palacín, J. Mater. Chem. A 3 (2015) 22.
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Higher voltage plateau cubic Prussian White for Na-ion batteries Piernas-Mun ~ oz a, Elizabeth Castillo-Martínez a, *, Oleksandr Bondarchuk a, María Jose a filo Rojo a, b, ** Michel Armand , Teo a b
gico de Alava, ~ ano, Spain CICenergigune, Parque Tecnolo Albert Einstein 48, ED. CIC, 01050 Min nica, Universidad del País Vasco UPV/EHU, P.O. Box 664, 48080 Bilbao, Spain Departamento de Química Inorga
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Cubic Prussian White, A2-xFe2(CN)6 (A ¼ Na, K), are synthesized by a new method. Above 140 mAh g1 are achieved at 1C when tested as cathodes vs. Naþ/ Na. An interesting voltage increase of 0.35 V is observed for the potassium phase. Hybrid Na/K insertion seems to be the main cause of the voltage increase. In addition, it exhibits a high capacity retention (80% after 500 cycles).
a r t i c l e i n f o
a b s t r a c t
Article history: Received 23 February 2016 Received in revised form 8 May 2016 Accepted 11 May 2016 Available online 8 June 2016
Cubic sodium Prussian White, Na2-xFe2(CN)6$yH2O, and potassium Prussian White, K2-xFe2(CN)6$yH2O, are prepared following a mild synthetic methodology. While cubic symmetry is confirmed by XRD and TEM, IR and XPS show characteristic features different from Prussian Blue compositions. When investigated as cathode materials in sodium ion batteries, both compounds exhibit reversible capacities above 140 mAh g1 at 1C (ca. 80 mA g1). While sodium Prussian White shows better high rate capability (10C/ 0.1C ¼ 0.64), potassium Prussian White exhibits longer cycle stability, with up to 80% of capacity retention after 500 cycles. Interestingly, the potassium Prussian White phase also provides an increase of 0.35 V in the high voltage redox peak compared to the sodium Prussian White analogue ascribed to the preferential insertion of Kþ ions instead of Naþ, resulting in an increment of the gravimetric energy density. On the other hand, the insertion of Naþ seems to occur at the lower voltage plateau. This hybrid Naþ and Kþ insertion in the framework of potassium Prussian White is most likely the responsible of the long cycle stability as a consequence of synergistic effects. © 2016 Elsevier B.V. All rights reserved.
Keywords: Electrochemical energy storage Na-ion batteries Prussian white Prussian blue related Hybrid battery
1. Introduction * Corresponding author. Current address: Department of Chemistry, University of Cambridge, Lensfield Road, CB2 3ED, Cambridge, UK. ** Corresponding author. Departamento de Química Inorg anica, Universidad del País Vasco UPV/EHU, P.O. Box 664, 48080 Bilbao, Spain. E-mail addresses: [email protected], [email protected] (E. CastilloMartínez), [email protected] (T. Rojo). http://dx.doi.org/10.1016/j.jpowsour.2016.05.050 0378-7753/© 2016 Elsevier B.V. All rights reserved.
Na-ion batteries (SIB) are attracting the attention of the scientific community as alternative to the already well established Li-ion batteries [1]. Several cathodic compounds, including layered oxides, olivine type NaFePO4 structures and polyanions, among others, are under study with the aim of finding the most suitable
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
material for this new approach [2,3]. Undoubtedly, an interesting material for this type of emerging technology is the metal-organic open framework called Prussian Blue AM[M0 (CN)]6 (A ¼ alkali metals and M, M0 ¼ transition metals; sometimes, M ¼ M0 ; typically M0 ¼ Fe), in which alkali ions occupy half of the empty cubic cavities created by a cubic network of alternating M and M0 ions in the vertices and eC^Ne ligands on the edges [4e8]. These materials exhibit interesting properties such as low cost, easiness of synthesis and good stability. Even of more technological importance is the alkali-rich reduced phase, known as Prussian White (PW hereafter), Na2M[M0 (CN)]6. Unlike Prussian Blue (PB from now on), PW has all the cubic cavities occupied by alkali ions thereby preventing first cycle inefficiencies and the need of providing an additional source of alkali ions in the anode [4]. Several phases of the Prussian White, also called Everitt’s salt, have been recently investigated as low-cost cathode material for SIB. Moritomo et al. explored Na2-dMxFey(CN)6 (M ¼ Co/Mn, d 0.4), which showed capacities up to 135 mAh g1 and fast Naþ intercalation/de-intercalation with no structural phase transition [8e10]. Nonetheless, when M ¼ Mn and d 0.3, a structural transformation from cubic to rhombohedral occurs. However similar capacities to those previously reported are obtained [11]. On the other hand, a monoclinic Mn-based sodium rich PW, Na1.96Mn [Mn(CN)6]0.99∙▫0.01 ∙2H2O, studied by Cui et al., exhibited up to 209 mAh g1 at 40 mA g1 due to the possible insertion of 3 Naþ in the structure and the reduction from MnIIIeN^CeMnIII to MnIIeN^CeMnI. A structural phase transition from the assynthesized monoclinic PW to an orthorhombic phase along the de-insertion process has also been observed, to finally reach a cubic structure at the end of charge [12]. Besides, in an attempt to combine the good specific capacity observed in Na1.8Mn [Fe(CN)6]0.99∙▫0.01 and the coulombic efficiency of Na1.86Ni [Fe(CN)6]0.96∙▫0.04, Ma and coworkers prepared and tested the ternary Prussian blue analogue (PBA), Na1.76Ni0.12Mn0.88[Fe(CN)6]0.98∙▫0.02. It displayed a specific capacity of 123.3 mAh g1, coulombic efficiency of 92.71% in the first cycle (and above 98% after 5 cycles) and 83.3% of capacity retention after 800 cycles when cycled at 100 mA g1 [13]. Also, a highly crystalline rhombohedral CoeFe based Prussian White, Na1.85Co[Fe(CN)6]0.99 ∙▫0.01∙ 1.9H2O, delivers high capacity of 150 mAh g1 and cyclability of 90% over 200 cycles at 100 mA g1 [14]. Interestingly, the purely iron based PW cathode material, Na2xFe2(CN)6, has been recently reported only by Guo et al. and Goodenough et al. This compound presents a rhombohedral structure and delivers capacities of 150 mAh g1 at 10e25 mA g1 and fairly good cycling performance. About 90% of the initial capacity is retained after 200 cycles at 25 mA g1 in Guo’s report and 80% is observed after 750 cycles in Goodenough’s study, when it is charged at 75 mA g1 and discharged at 300 mA g1 [15,16]. Among the PW and PB framework compounds, those with potassium as alkali ion were the first known to form [4]. Indeed, the large ionic radii (1.38 Å for Kþ compared to 1.02 Å for Naþ) [17] fits well the size of the cavity and stabilizes the framework. Thus, the use of K-PW in a Na-ion cell is appealing for simplicity reasons. Nevertheless, an extra step would be required to avoid any possible electrolyte contamination with K ions, unless the use of a hybrid electrolyte is not detrimental to the electrochemical activity. The utilization of dual electrolytes containing two cations has been successfully deployed for NaeMg batteries, [18] as well as for Li-ion and Na-ion batteries. In fact, Li-ion activation facilitates Na-ion insertion into Ge-electrodes [19] and other sodium compounds such as Na3V2(PO4)2F3 have also been used in Li-ion cells [20]. Even aqueous mixed Li/Na electrolytes have been used in hybrid operation Li/Na-ion batteries in which each electrode only inserts one specific cation [21]. Consequently, the use of potassium PW in a Na-
767
ion cell seems an interesting approach towards low cost SIB. In this article, we describe a new synthetic method for obtaining nanoparticles (NPs) of the cubic alkali-rich phases sodium Prussian White (Na-PW from now on), Na2-xFe2(CN)6$yH2O, and potassium Prussian White (K-PW hereafter), K2-xFe2(CN)6$yH2O, by coprecipitation in water and heating under mild conditions (60 C). A comparative study on the electrochemical performance of both compounds as cathode materials for SIB is also reported. An advantageous increase in the gravimetric energy and long-cycle stability is reached when K-PW is used instead of Na-PW. Since these materials are only iron-based and, therefore, the more sustainable and less costly among the PW analogues, they are promising cathode materials to be used in SIB. 2. Experimental section 2.1. Synthesis The A-PW (A ¼ Na, K) phases were prepared by using a different methodology to that reported by Jiang et al. [22] for the synthesis of K-PW, in order to avoid the formation of some Fe2O3 as impurity. The new synthetic route we developed is described below. Na-PW NPs, ideally Na2FeII[FeII(CN)6]$xH2O, were obtained using a single step method. In inert atmosphere, 100 mL of a freshly prepared pale yellow Na4[Fe(CN)6]$10H2O 40 mM aqueous solution were directly mixed with 100 mL of a pale yellow-greenish FeCl2 40 mM aqueous solution, forming a light blue suspension, according to the reaction described in equation (1). The partial iron oxidation from Fe2þ to Fe3þ was avoided by adding a small amount (approximately 2 mg) of ascorbic acid in the solution. The ascorbic acid acts as reducing agent, preserving the Prussian White phase and hindering the evolution towards Prussian Blue. Then, the suspension was mixed with an excess of NaCl to increase the Na content in the structure. After ca. 4 h heating at 60 C under N2 flow, the mixture was filtered, washed with H2O (3 10 ml) and acetone (3 10 ml) and dried, obtaining pale blue crystal-like aggregates of Na-PW NPs, Na2-xFeII1þ(x/2)[FeII(CN)6]$yH2O (see equation (1)). (1 þ (x/2)) FeCl2 þ Na4[Fe(CN)6]$10H2O / Na2-xFeII [FeII(CN)6]$yH2O þ (2 þ x) NaCl (aq.)
1þ(x/2)
(1)
K-PW nanoparticles, were prepared following the same procedure deployed for obtaining Na-PW NPs, just replacing the starting sodium ferrocyanide by the potassium salt, K4[Fe(CN)6]$3H2O and NaCl by KCl. In this case, a light green suspension is formed but the final product presents a pale blue colour. The reaction that takes place is analogous to that of equation (1) (see equation (2)). (1 þ (x/2)) FeCl2 þ K4[Fe(CN)6]$3H2O / K2-xFeII yH2O þ (2 þ x) KCl (aq.)
II 1þ(x/2)[Fe (CN)6]$
(2)
2.2. Structural, morphological and physico-chemical characterization PXRD patterns were recorded on a Bruker D8 Discover X-ray diffractometer, using lCu-Ka ¼ 1.54056 Å radiation in the 5e80 2q range, with a step width of 0.0194 . IR spectra were collected in the range of wavenumber ¼ 4000-650 cm1 in absorption mode in an Agilent Cary 630 FTIR spectrometer (Agilent Technologies), placed inside an Ar-filled glove box. SEM images were obtained in a FEI Quanta 200 F SEM operated at 30 kV and equipped with an Apollo 10 SSD Energy Dispersive X-ray (EDX). TEM images and electron diffraction patterns were taken using a Philips CM200. Chemical
768
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
compositions of the materials were determined using, respectively, atomic absorption and emission spectroscopy to calculate the Na: Fe and K: Fe ratios. Elemental analyses have been performed in a Euro Elemental Analyzer (CHNS), to check the H, C and N percentage present in the samples. TGA (NETZSCH STA 449 F3 Jupiter) was used to collect the thermogravimetric curves. Experiments were performed in the temperature range from 30 to 325 C using N2 atmosphere. XPS spectra were acquired in transmission mode with pass energy of 20 eV in an UHV system equipped with a hemispherical electron energy analyzer PHOIBOS 150 (SPECS) and a twin Al/Mg anode - X-ray source XR50 (SPECS) operated at 12 keV and power 100 W. The size of the acceptance area of the analyzer was ca. 2 mm in diameter and its energy scale was calibrated against Cu 2p3/2, Ag 3d5/2 and Au4f7/2 lines of standard AueAgeCu sample. The powder samples were fixed onto the Mo sample plate using double-sided carbon tape. To avoid contact of the samples with ambient, the samples were transferred from the glove box where they were stored to the XPS chamber in an Ar filled air-tight container. Raman analyses were conducted in the range 100e4000 cm1 with resolution 4 cm1 using a Ranishaw InVia spectrometer equipped with a 532 nm green laser operated at a power of 0.5 mW. The laser spot size was approximately 1 mm in diameter. In order to avoid oxidation of the samples due to exposure to the high power density laser radiation the measurements were performed also in an Ar-filled compartment. 2.3. Electrochemical characterization The electrochemical measurements of both compounds A-PW (A ¼ Na, K) were performed in CR2032 coin cells, using a Maccor® series 4000 battery tester. Positive electrodes consisted of 70 wt (%) of active material (either Na-PW or K-PW) with 20 wt (%) conductive C (Ketjen black® EC-600JD) and 10 wt (%) binder (PVDF - Solef ®), mixed with Nmethyl-2-pyrrolidone (NMP). The slurries were homogeneously coated on an Al current collector and dried overnight at 80 C under vacuum. Electrodes were cut as discs 13 mm in diameter and pressed at 5 tons for few seconds. For measurements carried out with powder, positive electrodes were prepared by simply mixing 80 wt (%) of the active material (either Na-PW or K-PW) with 20 wt (%) of conductive C (Ketjen black® EC-600JD) through grinding. Then, they were directly deposited in the bottom casing of the cell. Unlike positive electrodes described in the previous paragraph, powder materials are binder-free and current collector free and no calendering or extra processing, other than vacuum drying, were implemented. When the feasibility of Na-PW and K-PW was explored vs. Naþ/ Na, i.e. for SIB, measurements were performed in the voltage range from 2.25 to 4.25 V. Metallic sodium was used as counter and reference electrode. The electrolyte deployed for both A-PW (A ¼ Na, K) was 1 M NaPF6 in EC: PC: FEC 49:49:2% vol [23], impregnating a Whatmann® GF-D separator. To study the nature of the higher voltage plateau observed in KPW, its electrochemical properties as cathode in K-ion batteries (KIB) were also investigated. Metallic potassium was used instead of sodium as counter and reference electrode. 1 M KPF6 in EC: DMC 50:50% vol. was the selected electrolyte. No FEC was added in this case, as no long-term cyclability was required. 3. Results and discussion 3.1. Morphological, structural and physico-chemical study Light blue Na-PW and K-PW NPs were prepared following the procedure described in the experimental section. To calculate the
Na: Fe ratio of Na-PW and K: Fe of K-PW, atomic absorption and emission spectroscopy techniques were deployed, respectively. We further confirmed that the chemical formula of both PW is consistent with the CNH analysis. The accurate chemical composition achieved was Na1.70Fe2.15(CN)6 for Na-PW and K1.59Fe2.20(CN)6 for K-PW. The powder X-ray diffractograms of Na-PW and K-PW are displayed in Fig. 1a and b. Unlike the purely iron based Prussian White previously reported in the literature [15,16], that possess a rhombohedral lattice (R 3) (see Fig. 1c), the materials herein synthesized seem to present a cubic structure profile. Rietveld refinements support this statement (see Figure S1). The fittings have been performed refining lattice parameters, constrained atomic positions and occupancies of a starting cubic Prussian blue model and results in a lattice parameter a ¼ 10.350(5) Å for Na-PW and a ¼ 10.117(6) Å for K-PW. To corroborate the formation of the Na-rich phase Prussian White and discard that cubic Prussian Blue was obtained instead, other spectroscopic and diffraction techniques were deployed. IR spectra (Fig. 1d, e) show that cyanide (eC^N) stretching vibrational bands appear at 2068 and 2067 cm1, respectively for Na-PW and K-PW. This deviation towards lower wavenumber with regard to Prussian blue (ca. 2078 cm1) [25] suggests longer C^N distances when the two iron atoms approach to oxidation state 2þ in fully sodiated/potassiated PW (the Fe(II) covalently bonded to C (FeIIeC) and the Fe2þ ionically linked to N (Fe2þeN)). These results are in agreement with those presented by Takachi [8] and Wu [28], who observed a higher wavenumber of the eC^N band when iron oxidation state increases in a Prussian blue structure. In addition, the IR spectra revealed the presence of a small amount of water in both pristine materials. H, C, N analyses point to an approximate H2O content of ca. 4 and 2 molecules per formula unit, accordingly for the sodiated and potassiated phases. However, this water content could be influenced by the exposure of the samples to air when the analysis is carried out. Indeed, a negligible amount of H2O is estimated both in Na-PW and K-PW from the TGA curves (see Figure S2). SEM images show the morphology of Na and K-PW (Fig. 1f, g) and highlight a heterogeneous particle size distribution that ranges from few micrometers (2e5 mm) to 100 mm. TEM images, nonetheless, provide a more accurate particle size value and prove the formation of primary nanoparticles which are aggregated into the larger grains observed by SEM. Crystalline domains of 35e40 and 20 nm are observed, respectively, for Na-PW and K-PW, which agrees with the very broad reflections displayed by PXRD. Besides, electron diffraction patterns were recorded to determine a possible distortion from cubic to rhombohedral symmetry in individual particles. Indexation of the ring-type electron diffraction patterns shown by the nanocrystals of Na-PW (Fig. 1h) and K-PW (Fig. 1i) confirmed the cubic structure (F m 3 m) in both materials. EDX performed in the TEM to the individual crystallites analyzed by electron diffraction yielded approximate Na: Fe and K: Fe ratios of 1.3: 2 and 1.67: 2. The cubic symmetry must be related to the lower than 2 Na/K content on Na/K-PW. This assumption is consistent with what Guo et al. have reported, who observed that the crystal structure evolved from cubic to rhombohedral when changing from Na1.24Fe1.89(CN)6 to Na1.63Fe1.89(CN)6, demonstrating that the PB framework tends to form lower symmetry structures the greater the amount of Na it contains [15]. This change was also observed in the Mn-based Prussian White Na2-dMn[Fe(CN)6], in which the phase transition from cubic to rhombohedral occurred at d 0.3 [11]. In fact, the formula of our Na-PW compound, Na1.70Fe2.15(CN)6 or Na1.58Fe2(CN)5.58, is very close to that of Guo’s rhombohedral phase but still on the cubic side of the transition. On the other hand, the fully potassiated K-PW phase is reported to be monoclinic [22]. Consequently, despite having obtained alkali rich PW phases, the
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
769
Fig. 1. X-ray diffractograms of Na-PW (a) and K-PW (b). To compare the experimental diffractograms obtained with that already reported for Sodium Prussian White, Na-PW, a profile with lattice parameters and space group R-3 from ref. [24] was simulated (c). IR spectra of Na-PW (d) and K-PW (e), highlighting the more significant bands (C^N (cyanide) stretching vibrational band and OeH (hydroxyl) stretching vibrational and deformation band). SEM images of Na-PW (f) and K-PW (g) morphology. Electron diffraction patterns of Na-PW (h) and K-PW (i).
alkali content in our compounds is far from reaching the full site occupancy of 2 and thus they maintain the cubic structure. Raman spectra of the pristine Na-PW and K-PW compounds were collected in the wavenumber range 1900e2300 cm1, to evaluate the eC^N stretching band. The results are depicted in Fig. 2a and the deconvolution of the Raman spectra are shown in Fig. S3 (Supplementary data). The Raman scattering frequencies change both position and relative intensities when comparing NaPW and K-PW. A couple of distinctive bands are clearly distinguished at 2116 and 2148 cm1 and at 2090 and 2131 cm1, for NaPW and K-PW, respectively. A shoulder at lower wavenumber can also be guessed at ca. 2060e2075 cm1. Red shift on the bands when changing from Na-PW to K-PW could be indicative of the C^N bond softening due to the larger ionic radii of K ions. Although the presence of three bands in the Raman spectra is usually found
in rhombohedral Fe-based PW, the values obtained in this case are slightly different from those observed in previous works at 2070, 2110 and 2135 cm1 [15,16,26]. This suggests that despite the average crystal structure is cubic as determined from PXRD and electron diffraction, local rhombohedral distortions might be present in both compounds. Additional analyses were conducted. X-ray photoelectron spectroscopy (XPS) spectra provided determinant evidence of the formation of Prussian White phases distinct from Prussian Blue (see Fig. 2b). XPS analyses at the Fe 2p edge were performed on A-PW and A-PB (A ¼ Na, K) samples. The spectra of both Na- and K-PW are virtually identical as it can be observed in Fig. 2b, indicating the same electronic configuration of the iron in both materials. Analogously, both Na- and K-PB spectra are as well identical between them. Thus, the substitution of alkali metals (Na by K) does not
Fig. 2. Raman spectra focusing on the cyanide (C^N) band region (a) and XPS at the Fe 2p edge (b) of Na-PW (black) and K-PW (purple). XPS spectra of Na-PB (black) and K-PB (purple) are also included in dashed lines in Fig. 2b. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
770
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
affect the shape of Fe 2p spectra in both types of materials. For all the phases, A-PW and A-PB (A ¼ Na, K), the Fe 2p region has two split orbit components 2p3/2 and 2p1/2 with maxima at 709 and 722 eV, coincident with those previously reported for low spin Fe2þ of [Fe(CN)6]4- in Prussian Blue species [27,28]. The satellite pattern, however, varies when we compare PB to PW. The presence of a small peak at ca. 713 eV in Na/K-PB has been assigned in the literature either to high spin Fe3þ [27,29] despite the intensity is not as large as that of the Fe2þ signal, or to Fe2þ bonded to oxygen [30]. The latter can be attributed to the presence of water molecules. The peak appearing at 713 eV is however not observed for Na/K-PW which agrees with the absence of high spin Fe3þ and much lower content of H2O (see Figure S4 for the full XPS spectra: at Na 1s edge, K 2p edge, Fe 2p edge, C 1s edge, N 1s edge and O 1s edge). Instead, another satellite at ca. 724 eV, which could be related to the high spin state of the Fe2þ ions, can be observed for the PW species. Therefore, the PB and PW phases can be distinguished by the Fe 2p data obtained from XPS as we are showing here for the first time. 3.2. Electrochemical performance The electrochemical behavior of the materials Na-PW and K-PW was investigated for SIB (vs. Naþ/Na) through galvanostatic measurements in the cathodic voltage range from 2.25 to 4.25 V. Fig. 3a compares the voltage profiles corresponding to the first galvanostatic charge-discharge of Na-PW and K-PW in a sodium half-cell, using 1 M NaPF6 in EC: PC: FEC 49:49:2% vol as electrolyte [23]. Typical OCV values range from ca. 2.85e2.9 V for Na-PW to 2.9e2.95 V for K-PW. The lower OCV obtained compared to the usual 3.2 V observed in Prussian Blue is indicative of the higher alkali content in the PW phases. Therefore, the electrochemical measurements begin with a charge process. The large first low voltage plateau corroborates an almost complete filling of the alkali ion cavities in both cases. For Na-PW, two slopy plateaus centered at 3.0 and 3.3 V are distinguished along the charge (see Fig. 3a, in black). Upon discharge, the higher voltage redox process keeps on occurring centered at 3.3 V (see dQ/dV vs. V curve in the inset of Fig. 3a). Nonetheless, the redox reaction at lower voltage experiences a small shift towards lower potential values, from 3.0 to 2.8 V, probably due to the insulating character of the sodiated phase that leads to a higher polarization. These voltage values above mentioned for the redox processes are in-between those reported by Guo-Guo et al., who showed mainly a single plateau at 2.8e2.9 V [15], and those found by Goodenough et al., where a couple of
plateaus are visible in the range from 3.0 to 3.3 V [16]. For K-PW, a couple of plateaus are also observed while oxidizing at 3.0 and 3.7 V and the same two plateaus appear at 2.85 and 3.65 V upon discharge. Again, the lower voltage redox process presents a higher polarization. It is worth mentioning the increase occurring in the voltage value of the upper voltage process when moving from Na-PW to KPW. A significant difference of 0.35 V is noticed in the dQ/dV vs. voltage curve (see Fig. 3a inset and Fig. 3b). In addition, these plots manifest the complexity of the redox processes that are taking place (in some cases, each redox process consisted of two step reaction), as well as the repeatability and reproducibility of these along the cyclability. To explain the convenient voltage difference existing in the upper voltage redox process between Na-PW and K-PW, two hypotheses can be considered. On one hand, the presence of a different cation in the materials could be responsible of this positive potential shift. The effect of the cation on the increasing cell voltage, and therefore the energy density, has already been reported for other systems, such as quinones [31] In our particular case, for the K-PW electrodes, once the material is completely charged and consequently the lattice emptied of Kþ, the alkali metal being inserted when starting to discharge could be either Naþ or Kþ. Although the amount of Naþ present is several fold higher with respect to Kþ (i.e. 26.5 times bigger, calculated from the volume of the electrolyte used and the weight of electrode material), the selectivity of the PB framework for Kþ is likely higher [32]. PB is known as a decontaminant for 137Cs, being selective in biofluids for this even larger cation [33]. Probably in our case, the bigger ionic size of Kþ stabilizes somehow the molecular structure of the metal-organic framework. However, the low voltage redox process of K-PW vs. Naþ/Na occurs at similar voltage to that observed for Na-PW electrodes. This fact suggests that either Naþ insertion is taking place during this reaction or that the Kþ second intercalation happens at the same potential value of Naþ. On the other hand, structural variances in the lattice, from Na-PW to K-PW, could also be responsible of the different voltage plateau. Although both starting phases are cubic according to PXRD and electron diffraction (Fig. 1), the data obtained from Raman spectra (Fig. 2a) suggest that different local distortions appear and those could be translated into all the de-intercalated/intermediate phases. We can rule out an effect coming from the negative electrode, as the reduction potential of Kþ is more negative than that of Naþ, and thus the only redox equilibrium taking place is Naþ þ e ⇔ Na .
Fig. 3. a) First cycle voltage profile, (inset) first cycle dQ/dV vs. voltage curves and b) second cycle dQ/dV vs. voltage curves of Na-PW (in black) and K-PW (in violet), when cycling vs. Naþ/Na in the voltage range of 2.25e4.25 V, using 1 M NaPF6 in EC: PC: FEC 49:49:2% vol. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
To investigate whether the difference in the upper voltage plateau between Na-PW and K-PW was due to the insertion of a different alkali metal despite both materials being cycled vs. metallic sodium (Naþ/Na) using a Naþ containing electrolyte, the next procedure was followed: A K-PW electrode was electrochemically charged up to 4.2 V, with the purpose of reaching the alkali-empty phase (Berlin Green, BG). Once this phase was attained, the cell was opened, the electrode was washed with the corresponding carbonate solvent (PC) and when dried, it was reassembled again with 1 M NaPF6 EC: PC: FEC vs. Naþ/Na. As it can be seen in Fig. 4, the electrochemical performance of this BG phase is almost identical to that exhibited by Na-PW. This fact indicates that when K-PW is directly cycled against Naþ/Na, it is not Naþ but Kþ the alkali ion inserting and de-inserting into/from the structure, at least in the high voltage redox process. Therefore, it can be stated that the difference in the voltage value corresponds to the different alkali ions intercalated. To determine then the nature of the lower voltage plateau, KPW was cycled vs. metallic potassium (Kþ/K). Fig. 5 illustrates the dQ/dV vs. voltage profile of K-PW tested against Naþ/Na and Kþ/K. The difference in the standard reduction potentials between Naþ/ Na and Kþ/K is 0.211 V in aqueous solution [34]. Although the voltage difference may change in organic media, this 0.211 V shift has been considered to recalculate the voltage shown by K-PW when cycled vs. Naþ/Na for comparison with that of K-PW cycled in a K-ion battery (Fig. 5). The voltage values are presented both vs. Naþ/Na and Kþ/K. It is clear that the redox peaks are somehow coincident in the higher voltage plateau, which agrees with Kþ insertion, as demonstrated previously. Nonetheless, in the lower voltage plateau, the redox process occurs at different potential values, revealing that the species inserted during this process might be more likely Naþ or a mixture of both cations, Naþ and Kþ as two redox peaks appear when cycling vs. Naþ/Na and only 1 peak when cycling vs. Kþ/K. In addition, it is worth noting the excellent properties obtained when K-PW is tested vs. Kþ/K, proving its viability to be used as cathode in K-ion batteries (KIB), as Eftekhari [35] had reported for Prussian Blue in 2004. EDX analyses in K-PW electrodes cycled vs. Naþ/Na and strategically stopped at certain points of the electrochemical discharge curve have been conducted to chemically confirm the hybrid nature of the cation insertion in this material. Prior to the EDX collection,
Fig. 4. dQ/dV vs. V profile comparative between the electrochemical behavior of NaPW (continuous line in black), K-PW (dashed line in purple) and K-PW charged, washed with PC and cycled again vs. Naþ/Na (red dots). The electrolyte used in all experiments here plotted was 1 M NaPF6 in EC: PC: FEC 49:49:2% vol. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
771
Fig. 5. dQ/dV vs. V profile comparative between the electrochemistry behavior of KPW against Kþ/K (continuous line in purple) and Naþ/Na (dashed line in purple). The electrolyte used for KIB was 1 M KPF6 in EC: DMC 50:50% vol. while 1 M NaPF6 in EC: PC: FEC 49:49:2% vol was the mixture selected for SIB. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
the electrodes were properly rinsed with PC for several times to remove any residual Naþ trace that could stem from the electrolyte salt and, then, dried overnight at 80 C under vacuum. The K/Na ratios thus increase from BG to PB phase supporting our hypothesis that Kþ is the metal being mainly inserted at the high voltage redox process, and decrease from PB to PW, which agrees with Naþ being the alkali mostly inserted at the low voltage plateau (see Supplementary data, section F, for further details). The hypothesis of structural variances in the lattice and/or different structural transformations upon cycling is not discarded, however, since both effects could be additive to the cation effect already demonstrated. Future work in understanding the mechanism behind it is necessary to determine whether structural effects are contributing. In addition, further insights into the mechanism of reaction could reveal the nature of the cation (either Naþ or a mixture of Naþ/Kþ) which is being inserted/de-inserted in the lower voltage plateau of K-PW. This would agree with the beneficial cooperative effects which have already been observed for other cation combinations, such as Li and Na [19e21] or Mg and Na [18]. Notwithstanding, this assumption cannot be confirmed before analyzing the reaction mechanism of these phases. C-rate capabilities of the two Na-PW and K-PW against metallic sodium (Naþ/Na) are shown in Fig. 6a and b. When cycled in powder form (Fig. 6a), both materials exhibit reversible capacities up to 145e150 mA h g1 at C/10 (being C ¼ 85.4 mAh g1 per 1Naþ insertion per formula unit for Na-PW and 77.5 mAh g1 for K-PW) with coulombic efficiencies of ca. 95e97% and 97e99.4%, respectively. These specific discharge capacities are analogous to those already reported in the literature for the rhombohedral purely iron based Na-PW [15,16]. Capacities between 115 and 120 mA g1 are achieved at moderate rate, as 1C. At higher current densities, such as 10C (nearly 1 A g1), the capacity retention displayed by Na-PW, 75 mAh g1, is larger than that for K-PW, 50 mAh g1. However, both materials demonstrate a good stability when cycled back at C/ 10. C-rate capabilities of Na-PW and K-PW, when the materials are coated over Al current collector forming electrodes are displayed in Fig. 6b. In this case, although specific capacities similar to those achieved in powder are reached, a negative effect in the coulombic efficiency is observed at low current densities, C/10. This inefficiency issue is reduced, nonetheless, once the materials operate
772
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
Fig. 6. C-rate capabilities of Na-PW (black squares) and K-PW (violet circles) vs. Naþ/Na cycled in powder (a) and coated over Al current collector (b). c) Long-term cyclabilities (capacities and coulombic efficiency values) of Na-PW (black squares) and K-PW (violet circles) when cycling vs. Naþ/Na in the cut off voltage DV ¼ 2.25e4.25 V. The theoretical capacity 1C corresponds to 85.4 mA g1 or 77.5 mA g1/1Naþ insertion per f.u., respectively for Na-PW and K-PW. 1 M NaPF6 in EC: PC: FEC 49:49:2% vol. was the electrolyte utilized in all experiments here displayed. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
at moderate and higher capabilities, finding analogous values than those in powder. In addition, when contrasted to the powder form, better capacity retention is evident for K-PW at 1C (achieving 126 mAh g1), and for Na-PW (88 mAh g1) and K-PW (72 mAh g1) at 10C. The adverse effect on the coulombic efficiency at low current densities could tentatively be attributed to Al current collector corrosion, though PF 6 usually protects the native Al2O3 layer. The addition of ACl (A ¼ Na, K) to synthesize the material, although beneficial to raise the alkali content, can imply the risk of corroding the Al used as substrate from trace Cl. Consequently, it is paramount to thoroughly rinse the material with H2O to remove all the Cl. In our case, no Cl presence was detected in pristine K-PW and only some traces in some isolated particles at Na-PW. The good performance of Na-PW and K-PW along the cycle-life at moderate current density (see Fig. 6c) is also remarkable. Roughly 140 mAh g1 are achieved during the first cycles at 1C. In addition, 63.4% of the initial capacity is retained after 500 cycles with >98% of coulombic efficiency for Na-PW. Although the capacity retention obtained for the cubic Na-PW is lower than that presented by Guo-Guo [15] and Goodengouh [16], the coulombic efficiency is far superior. Besides, comparing cubic Na-PW with other Prussian White systems, such as those Mn-Fe [10,11], Co-Fe [9,14] or Mn-Mn [12] based, the capacity retention and coulombic efficiency values are higher as well. On the other hand, the hybrid K-PW exhibits enhanced properties compared to Na-PW when cycled vs. Naþ/Na, keeping 80% of the initial capacity retention after 500 cycles with >97.5% of coulombic efficiency during the first 30 cycles and >99.5% in cycle 500. In addition, it shows a fairly good stability in a long-term cyclability of ca. 1000 h ¼ 42 days of autonomy. The coulombic efficiency obtained, far from the optimum >99.9%, could be fairly improved however by simply assembling full cells, since the deployment of Na metallic as negative and counter electrode is known to react detrimentally with the electrolyte forming dendrites, resulting in inefficiency [36e39]. Another option to bear in mind could be to replace the Al current collector deployed in here by a carbon cloth current collector substrate frequently used in Prussian Blue systems [12]. These tasks will be addressed in a future work. In any case, the interesting behavior exhibited at moderate current densities, which are typical for market applications, allow us to consider K-PW as suitable and desirable electrode material for a real use in a hybrid K/Na-ion battery.
the formation of materials with cubic structure for purely iron based PW, unlike those reported previously in the literature. The cubic symmetry is confirmed by PXRD and electron diffraction experiments in the TEM while IR and XPS show the PW character distinct from PB. The benefit of obtaining cubic symmetry is major, since the species formed during the cycling process (PB and BG) are cubic as well, and no structural transitions are required in principle. The materials exhibit fairly good electrochemical properties, when cycled vs. Naþ/Na in the cathodic range from 2.25 to 4.25 V using 1 M NaPF6 in EC: PC: FEC. Specific capacities over 140 mA h/g at 1C with reasonably high coulombic efficiencies values (above 98%) were exhibited by Na-PW and K-PW. Full cell constructions and the use of carbon based current collectors could help to improve the good but limited coulombic efficiency here obtained with Al current collector and with the use of metallic Na as counter and reference electrode. In addition, it is important to highlight that an interesting increase in the gravimetric energy density is observed when K-PW replaces Na-PW, due to a raise of 0.35 V in the potential at which the high voltage redox process occurs. The reason behind is so far ascribed to the insertion and de-insertion of Kþ instead of Naþ, but further studies on the reaction mechanism must be performed to either discard or confirm an additional effect due to a structural change and synergetic effects related to the presence of both cations in the medium. This higher voltage, along with the better capacity retention (80% after 500 cycles), makes K-PW an even more attractive cathode material than Na-PW for real market applications.
Acknowledgements This work was financially supported through projects LINABATT (ENE2013-44330-R) and Etortek 14 CIC Energigune. SGIker technical and human support (UPV/EHU, MICINN, GV/EJ, ESF) is gratefully acknowledged for the help with AAS and elemental analysis (H, C, N). The authors would also like to acknowledge Jon Ajuria and Frederic Aguesse, for their help with SEM images collection and Piernas Mun ~ oz Ana Martinez Amesti for recording TEM. Ma Jose thanks the Basque Government for the grant corresponding to “Nuevas Becas y Renovaciones para el Programa Predoctoral de n de Personal Investigador” (PRE.2013.1.790, MOD A). Formacio
4. Conclusions Appendix A. Supplementary data Na-PW and K-PW have been synthesized following a new mild synthetic route, at low temperature (60 C), in aqueous media and avoiding the addition of acidic products. This methodology allows
Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2016.05.050.
~ oz et al. / Journal of Power Sources 324 (2016) 766e773 M.J. Piernas-Mun
References [1] V. Palomares, P. Serras, I. Villaluenga, K. Hueso, J. Carretero-Gonzalez, T. Rojo, Energy Environ. Sci. 5 (2012) 5884e5901. [2] V. Palomares, M. Casas-Cabanas, E. Castillo-Martínez, M.H. Han, T. Rojo, Energy Environ. Sci. 6 (2013) 2312e2337. [3] N. Yabuuchi, K. Kubota, M. Dahbi, S. Komaba, Chem. Rev. 114 (2014) 11636e11682. [4] J.F. Keggins, F.D. Miles, Nature (1936) 577. [5] D.B. Brown, D.F. Shriver, Inorg. Chem. 18 (1969) 37e42. [6] D. Yang, J. Xu, X-Z. Liao, H. Wang, Y.-S. He, X.-F. Ma, Chem. Comm 51 (2015) 8181e8184. [7] Y. Lu, L. Wang, J. Cheng, J.B. Goodenough, Chem. Comm 48 (2012) 6544. [8] M. Takachi, T. Matsuda, Y. Moritomo, Jpn J. Appl. Phys. 52 (2013) 090202. [9] M. Takachi, T. Matsuda, Y. Moritomo, Appl. Phys. Express 6 (2013) 025802. [10] M. Takachi, T. Matsuda, Y. Moritomo, Chem. Commun 49 (2013) 2750e2752. [11] L. Wang, Y. Lu, J. Liu, M. Xu, J. Cheng, D. Zhang, J.B. Goodenough, Angew. Chem. Int. Ed 52 (2013) 1e5. [12] H.-W. Lee, R.Y. Wang, M. Pasta, S.W. Lee, N. Liu, Y. Cui, Nat. Commun 5 (2014) 5280. [13] D. Yang, J. Xu, X.-Z. Liao, Y.-S. He, H. Li, Z.-F. Ma, Chem. Commun 50 (2014) 13377e13380. [14] X. Wu, C. Wu, C. Wei, L. Hu, J. Qian, Y. Cao, X. Ai, J. Wang, H. Yang, ACS Appl. Mat. Interfaces (2016), http://dx.doi.org/10.1021/acsami.5b12620. [15] Y. You, X.-Q. Yu, Y.-X. Yin, K.W. Nam, Y.-G. Guo, Nano Res. 8 (2015) 117e128. [16] L. Wang, J. Song, R. Qiao, L.A. Wray, M.A. Hossain, Y.-D. Chuang, W. Yang, Y. Lu, D. Evans, J.-J. Lee, S. Vail, X. Zhao, M. Nishijima, S. Kakimoto, J.B. Goodenough, JACS 137 (2015) 2548e2554. [17] R.D. Shannon, Acta Cryst A32 (1976) 751e767. n ~ ez, M.V. Kovalenko, Chem. Mater 27 (2015) [18] M. Walter, K.V. Kravchyk, M. Iba 7452e7458. [19] A. Kohandehghan, K. Cui, M. Kupsta, J. Ding, E.M. Lotfabad, W.P. Kalisvaart, D. Mitlin, Nano Lett. 14 (2014) 5873e5882.
773
[20] W. Song, X. Ji, Z. Wu, Y. Zhu, F. Li, Y. Yao, C.E. Banks, RSC Adv 4 (2014) 11375e11383. [21] L. Chen, Q. Gu, X. Zhow, X. Lee, Y. Xia, Z. Liu, Sci. Rep 3 (2013) 1946. [22] M. Hu, J.S. Jiang, Mat. Res. Bull. 46 (2011) 702e707. ~ oz, E. Castillo-Martínez, J.L. Go mez-C [23] M.J. Piernas-Mun amer, T. Rojo, Electrochim. Acta 200 (2016) 123e130. [24] J. Song, L. Wang, Y. Lu, J. Liu, B. Guo, P. Xiao, J.-J. Lee, X.-Q. Yang, G. Henkelman, J.B. Goodenough, JACS 137 (2015) 2658e2664. [25] L. Xia, R.L. McCreery, Electrochem. Soc. 146 (1999) 3696e3701. [26] Y. You, X.-L. Wu, Y.-X. Yin, Y.-G. Guo, Energy Environ. Sci. 7 (2014) 1643e1647. [27] M. Datta, A. Datta, J. Phys. Chem. 94 (1990) 8203e8207. [28] Q. Wu, G. Wu, L. Wang, W. Hu, H. Wu, Mat. Sci. Semicond Process 30 (2015) 476e481. [29] E. Desimoni, B. Brunetti, Chemosensors 3 (2015) 70e117. [30] X. Li, J. Liu, A.I. Rykov, H. Han, C. Jin, X. Liu, App. Catal. B Environ. 179 (2015) 196. ndez-Burgos, G.G. Rodríguez-Calero, W. Zhow, S.E. Burkhardt, [31] K. Herna ~ a, J. Am. Chem. Soc. 135 (2013) 14532e14535. H.D. Abrun [32] V. Krishnan, A.L. Xidis, V.D. Neff, Anal. Chim. Acta 239 (1990) 7e12. [33] J.M. Arnal, M. Sancho, B. García-Fayos, Desalination 321 (2013) 22e27. [34] D.R. Lide, W.M. Haynes, CRC Handbook of Chemistry and Physics (Internet version), 90th Ed., CRC press, Boca Raton (Florida) e USA, 2010, pp. 8e22 (1212). [35] A. Eftekhari, J. Power Sources 126 (2004) 221e228. [36] D.I. Iermakova, R. Dugas, M.R. Palacín, A.J. Ponrouch, Electrochem. Soc. 162 (2015) A7060eA7066. [37] A. Rudola, D. Aurbach, P. Balaya, Electrochem. Commun. 46 (2014) 56. [38] N. Weadock, N. Varongchayakul, J. Wan, S. Lee, J. Seog, L. Hu, Nano Energy 2 (2013) 713. [39] A. Ponrouch, D. Monti, A. Boschin, B. Steen, P. Johansson, M.R. Palacín, J. Mater. Chem. A 3 (2015) 22.