Copper Electroactivity in Prussian Blue Based ... - ACS Publications

Article pubs.acs.org/JPCC

Cite This: J. Phys. Chem. C 2018, 122, 15868−15877

Copper Electroactivity in Prussian Blue-Based Cathode Disclosed by Operando XAS Angelo Mullaliu,† Giuliana Aquilanti,‡ Paolo Conti,§ Jasper R. Plaisier,‡ Marcus Fehse,∥ Lorenzo Stievano,*,⊥,# and Marco Giorgetti*,† †

Department of Industrial Chemistry “Toso Montanari”, University of Bologna, Viale Risorgimento 4, 40136 Bologna, Italy Elettra − Sincrotrone Trieste, ss 14, km 163.5, 34149 Basovizza, Trieste, Italy § School of Science, Chemistry Division, University of Camerino, Via S. Agostino 1, 62032 Camerino, Macerata, Italy ∥ Dutch-Belgian (DUBBLE), ESRFThe European Synchrotron, CS 40220 Grenoble Cedex 9, France ⊥ Institut Charles Gerhardt Montpellier, CNRS UMR 5253, Université de Montpellier, 34090 Montpellier, France # Réseau sur le Stockage Electrochimique de l’Energie (RS2E), CNRS FR3459, 80039 Amiens, France

Downloaded via UNIV OF LOUISIANA AT LAFAYETTE on October 12, 2018 at 07:35:23 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: The electronic and structural evolution of copper hexacyanoferrate (CuHCF) cathode material was studied by operando X-ray absorption spectroscopy (XAS) simultaneously at both Fe and Cu K edges during a full galvanostatic cycle. The full set of XAS data collected during the electrochemical process was analyzed by a combined chemometric approach using the multicurve resolution analysis with the alternate least squares algorithm. Using this joint approach and by applying a simultaneous multipleedge fitting procedure, it was possible to clarify the participation of both copper and iron centers to the redox processes and to analyze their local environment. The structural modifications occurring in CuHCF along with the redox processes are entirely reversible, with the steady multiplicity of Fe−C−N−Cu linear chains evidencing the structural stability of the material during cycling. copper hexacyanoferrate (CuHCF) can be electroactive,14 the belief that iron is in fact the only active species is widely spread in the battery community.1,15−18 In some cases, not only the metal sites, but also the ligands are capable of participating in the redox processes, as already observed for copper nitroprusside, Cu[Fe(CN)5(NO)], an analogue of copper hexacyanoferrate.19 In copper nitroprusside, the nitrosyl ligand originally in the oxidized −NO+ form can accept one electron, be reduced to the radical −NO, and allow the reaction of one extra lithium equivalent per unit formula. In this work, we demonstrate that also copper is active in copper hexacyanoferrate during the electrochemical reaction with lithium and that it plays a remarkable role in the redox process. To reach this goal, operando X-ray absorption spectroscopy (XAS)20,21 was carried out at both metal centers. Indeed, XAS is a powerful tool that can be tuned to a chosen element and performed in operando mode, i.e., by collecting several spectra during electrochemical cycling in ad hoc developed in situ cells.21 In this way, the physicochemical properties and the local structure of the selected element can

1. INTRODUCTION Prussian blue analogues (PBAs), and in particular, metal hexacyanoferrates, have gained considerable attention among insertion materials due to the ease of preparation and separation, effectiveness as electrode materials, and wide versatility toward several ions, ranging from monovalent ions, such as lithium,1 sodium,2−6 and potassium,7,8 to divalent and trivalent ions, for instance, calcium,9 magnesium,10 and aluminum.11 The ability of accommodating such a variety of ions originates in the structure of PBAs, which is characterized by a three-dimensional cubic network (although other crystal symmetries are found) of repeating −NC−Fe−CN−M−NC− units, where iron and M sites are typically octahedral in the socalled “soluble” structure, displayed in Chart 1. The sites at the cube center (8c positions) can be occupied by countercations and water molecules to achieve charge neutrality. The cube size, which is twice the Fe−M distance, is about 10 Å and guarantees enough space for insertion/release of ions inside the zeolitic channels of roughly 3.2 Å, beyond cavities of around 5 Å arising from vacancies. Our group has determined selectivity constants for alkali-metal insertion using the nickel hexacyanoferrate analogue12 and has studied the electroactivity of both metals in hexacyanoferrates, for instance, for H2O2 sensing.13 Although the literature reports that both metals in © 2018 American Chemical Society

Received: April 11, 2018 Revised: May 31, 2018 Published: June 20, 2018 15868

DOI: 10.1021/acs.jpcc.8b03429 J. Phys. Chem. C 2018, 122, 15868−15877

Article

The Journal of Physical Chemistry C

Chart 1. Sketch Summarizing the Repetitive Atomic Fragments of the Copper Hexacyanoferrate Structure with Their Relative Structural Parametersa

At the bottom, a list of the multiple scattering (MS) n-body signals used for the extended fine-structure part of the spectrum (EXAFS) data analysis are shown. The degeneracy of the MS path is indicated in brackets. a

consecutively in the range 5° < 2θ < 70°, with steps of 0.01° and an acquisition time of 1 s/step. The crystal structure was refined using FullProf Suite,25 adopting the structure reported by Ayrault et al.26 as model and a Thompson−Cox−Hastings pseudo-Voigt profile shape. Rietveld refinement was performed by considering two different copper sites, one structural in the 4b position and one interstitial in the 8c crystallographic position. Moreover, the potassium/iron ratio was constrained to be 0.34 according to the X-ray fluorescence (XRF) analysis. Graphic representation of the crystal structures was achieved using VESTA software.27 The self-supported electrode for the XAS operando measurement was obtained by mixing the pure active material (80%), 15% Super-P carbon black, and 5% poly(tetrafluoroethylene) in a mortar. The collected electrode pellet was dried overnight in a Büchi oven under vacuum at 80 °C. The mass loading of the electrode was about 8 mg/cm2 of active material. 2.2. X-ray Absorption Spectroscopy (XAS). XAS experiments were conducted at ELETTRA Sincrotrone Trieste, Basovizza (Italy), at the XAFS beamline.21 The storage ring was operated at 2.0 GeV in top-up mode with a typical current of 300 mA. Data were recorded at Fe and Cu K edges in transmission mode using ionization chambers filled with a mixture of Ar, N2, and He to have 10, 70, and 95% of absorption in the I0, I1, and I2 chambers. An internal reference of iron and copper foil was used for energy calibration in each scan. This allowed a continuous monitoring of the energy during consecutive scans. No energy drifts of the monochromator were observed during the experiments. Spectra at Fe and Cu K edges were collected with a constant k-step of 0.03 Å−1 with 3 s/point acquisition time. Data were collected alternatingly from 6900 to 8000 eV and from 8750 to 9830 eV at the Fe and Cu K edges, respectively. The energies were calibrated by assigning the first inflection point of the spectra of the metallic iron and copper to 7112 and 8980.3 eV, respectively. The white beam was monochromatized using a fixed exit monochromator equipped with a pair of Si(111) crystals. Harmonics were rejected by using the cutoff of the reflectivity of the platinum mirror placed at 3 mrad with

be monitored at all moments during the charge and discharge processes. Moreover, the data treatment on the extended finestructure part of the spectrum (EXAFS) can be accomplished by using both single-edge and multiple-edge approaches,22 in the latter case by refining simultaneously the spectral data at two or more edges, which increases the reliability of the structural results. Combined with operando XAS, MCRalternate least squares (ALS) (multivariate curve resolution using a constrained alternating least squares algorithm) was used to recover the spectra of the pure chemical species. This chemometric approach is able to determine the number of “pure” components present in the system under study and to identify the evolution of their concentration. This analysis was first employed for battery systems by Conti et al. in 2010,23 where a joint chemometric dynamic XAS permitted a complete understanding of the cell under investigation.

2. EXPERIMENTAL SECTION 2.1. Synthesis, Characterization, and Electrode Material Preparation. Following a reported synthesis1 and adapting it slightly, copper hexacyanoferrate was synthesized by co-precipitation at 40 °C, by mixing 250 mL of a 0.04 M aqueous solution of CuSO4·5H2O and 250 mL of aqueous 0.02 M K4[Fe(CN6)]. Both solutions were prepared using doubledistilled water (ddH2O) and previously thermostated at 40 °C. Chemicals were purchased from Carlo Erba and used without further purification. The batch was aged for 2 days in the dark at room temperature. The red precipitate was separated by filtration on a Whatman 42 paper filter, washed twice with ddH2O, and dried at 30−40 °C under vacuum. The obtained solid (2.98 g) was finally ground in an agate mortar. X-ray fluorescence (XRF) was performed by means of a PANalytical AxiomAX spectrometer on boric acid pellets with a 26% mass concentration of active material to determine K/Cu/Fe ratio. Powder X-ray diffraction (PXRD) data were recorded on the synthesized powder using a monochromatic X-ray beam with a wavelength of 1 Å at the MCX beamline in ELETTRA Sincrotrone Trieste, Basovizza (Italy).24 Data were collected on the sample in a capillary geometry, setting the spinner at 300 rpm. The X-ray diffraction (XRD) pattern was collected 15869


Article


the two edges (with a degeneracy given by CN8c). Given the small scattering contribution of Li+ ions, and after having verified the weak dependency of the two-body contribution on the 8c site cation, the phase shift of copper was used for M8c. According to this model, the total number of parameters used in the fitting procedure (including the structural and nonstructural terms, namely, E0 and S02, and the experimental resolution) was 11 for single-edge fitting (Fe K edge) and 19 for double-edge fitting (jointly at Fe and Cu K edges). Besides, it is worth mentioning that in all cases the number of fitting parameters did not exceed the estimated “number of 2·δk·δR independent data points” Nind = π + 2, thus ensuring that the fit is well constrained and does not lead to parameters with very large errors, which, in turn, confirms the reliability of the minimization. The phase shifts for the photoabsorber and backscatterer atoms were calculated starting from the structural model by Ayrault et al.26 They were calculated according to the muffintin approximation and allowing 10−15% overlap between the muffin-tin spheres. The Hedin−Lundqvist complex potential32 was used for the exchange-correlation potential of the excited state. The core-hole lifetime, Γc, was fixed to the tabulated value33 and was included in the phase shift calculation. The experimental resolution used in the fitting analysis was about 1 eV, in agreement with the stated value for the beamline. Values of the amplitude correction factor S02 were identified for both Fe and Cu K edges using multiple-edge fitting procedure of pristine sample. The obtained values were 0.84(4) and 0.75(7) for the Fe and Cu K edges, respectively. The structural refinements of the EXAFS spectra during the operando scan were performed using those values for the amplitude correction factor.

respect to the beam upstream the monochromator and by detuning the second crystal of the monochromator by 30% of the maximum. A suitable in situ electrochemical cell28 for the operando XAS experiment was used. The cell was assembled in an Arfilled glovebox with the CuHCF electrode mounted against lithium metal foil negative electrode. The two electrodes were kept apart by a Celgard separator soaked with a 1 M LiPF6 in ethylene carbonate/propylene carbonate/3-dimethyl carbonate electrolyte. Galvanostatic cycling was conducted in the 2.0 < E < 4.3 V versus Li+/Li potential window at C/30 current rate, whereas 1C refers to the insertion of 1 mol Li into 1 mol CuHCF. 2.3. XAS Data Analysis. The EXAFS analysis was performed using the GNXAS package,29,30 which is based on the multiple scattering (MS) theory. Data analysis is performed by minimizing a χ2-like residual function that compares the theoretical signal, αmod(E), to the experimental one, αexp(E). The method is based on the decomposition of the EXAFS signal into a sum of several contributions, namely, n-body terms. The theoretical signal is calculated ab initio and contains the relevant two-body γ(2), three-body γ(3), and fourbody γ(4) multiple scattering terms.31 The two-body terms are associated with pairs of atoms and probe their distances and variances. The three-body terms are associated with triplets of atoms and probe angles and bond−bond and bond−angle correlations. The four-body terms are associated to chains of four atoms; probe distances and angles in-between; and bond− bond and bond−angle correlations. Due to the linearity of the −Fe−C−N−Cu− chains, all of the angles are set to be 180° and hence the actual number of parameters used to define the three-body and four-body terms is reduced by symmetry (deviations from the planarity can be neglected and therefore the corresponding correlations are set to zero). The use of all of the correlations concerning the four-body contribution is beyond the aim of the present work, and more details can be found in refs 13, 31. Joint fitting procedures at both Fe and Cu K edges were conducted in multiple-edge mode. The following set of MS paths (referring to the reported Chart 1) were included: the two-atom contributions γ1(2) Fe−C with degeneracy of 6, the three-body contribution η1(3) Fe−C−N with degeneracy of 6, and the four-body contribution η1(4) Fe− C−N−Cu with degeneracy of 6. It is worth noting that the inclusion of the three-body term η1(3) allows monitoring the shells beyond the first one by using the same three-atom coordinates for both the two-atom and three-atom contributions. In fact, the three-body signal η1(3) Fe−C−N includes both γ(2) Fe−N and γ(3) Fe−C−N signals. Similarly, the η1(4) MS signal includes two three-body contributions, γ(3) Fe−C··· Cu and γ(3) Fe···N−Cu and the interaction Fe···Cu through the γ(2) Fe···Cu signal. At Cu K edge, the MS signals included in the fit were: γ1(2) Cu−N; η1(3) Cu−N−C; η1(4) Cu−N−C− Fe. The degeneracy of those signals was allowed to float, using a parameter CNchain. It is interesting to note that, due to the multiple-edge fitting approach, only the three distances Fe−C, C−N, and Cu−N depicted in Chart 1 are necessary for the description of the above-mentioned MS signals, making this fitting approach highly reliable. In addition, a two-body γ(2) Cu−O with degeneracy of 2 was necessary at the Cu K edge to take into account the “insoluble” structure of metal hexacyanoferrates, whereas a two-body signal, γ(2) Fe(Cu)− M8c, due to the contribution of the interstitial cations (K+ and Cu2+ during charge, Li+ during discharge) was considered at

3. RESULTS AND DISCUSSION The stoichiometry of the resulting copper hexacyanoferrate can be written as K0.34Cu1.96Fe(CN)6 after XRF analysis, which provided a K/Cu/Fe ratio of 0.34/1.96/1. The Cu/Fe ratio agrees with the weighed ratio of the edge jump values at Cu and Fe K edges, which resulted 1.9 as well. The structure of hexacyanoferrates is related to the wellknown “soluble” and “insoluble” forms. Both feature a rigid cubic framework of linear −Fe−C−N−Cu− chains, where Fe and Cu ions occupy 4a and 4b positions, respectively, and are octahedrally coordinated to −CN and −NC groups, respectively. In case of the “insoluble” form, 1/3 of the Fe(CN)64− units is vacant, with the empty nitrogen positions occupied by water molecules to complete the coordination shell of Cu so that its average coordination would be CuN4O2. Conversely, every 4a-Fe atom is coordinated octahedrally by six carbon atoms. Moreover, according to the selected model,26 we considered a fraction of additional Cu atoms occupying interstitial 8c positions so that the final structure arising from the model can be simplified as Cu0.5Cu1.5Fe(CN)6. The obtained pattern for copper hexacyanoferrate, depicted in Figure 1, well matches with the reported model,26 described by a CuII/FeII ratio of 2. The refinement (χ2 = 1.45) led to a cubic structure (space group: Fm3m) characterized by the lattice parameter a = 9.997 Å. By refining occupancies of both interstitial and structural copper, and of iron as well, we identified the individual amount of metallic content per site so 4b that we could write the chemical formula as K0.34Cu8c 0.4Cu1.6Fe6, where 0.4 is the amount of interstitial copper (8c), the value 15870


Article


called the “edge resonance” is associated to the transition to the continuum. The features of the overall XANES region (labeled as C) arise from multiple scattering resonances of the photoelectrons and also from the interatomic distances and coordination geometry.13,34 The XANES image of CuHCF at the Fe K edge is distinctive of Fe in an octahedral environment with the Fe−C−N ligation.35 The edge position is typical of FeII formal oxidation state in the pristine material. For Cu K edge, XANES image is characterized by the presence of an intense peak before the principal edge, which is assigned to the 1s−4p transition.32 This occurs at about 8984 eV, a value that indicates the presence of a CuI formal oxidation state in the pristine sample, whereas its intensity suggests the fourfold coordination geometry of the Cu site.36 This is a reasonable hypothesis based on other evidence available in the literature, where, for CuI, the transition shows the highest intensity for linear two-coordinated complexes and becomes less intense and broadened as the coordination number increases and/or the symmetry is lowered.37 Overall, this result confirms the local geometry as seen by powder XRD, where the absence of 1/3 of the [Fe(CN)6]2− ions leads to fourfold coordination of the Cu site by N ends of the cyanide group. Regarding the associated charge of Cu in the pristine sample, the position of the 1s−4p transition indicates mainly CuI. This result is unexpected, if one considers the precursor salt (CuSO4) used for the synthesis of the material, but is in line to a study of a copper-based Prussian blue analogue,13,38 suggesting that the formal charge at the copper site for CuHCF is rather close to CuI. Briefly, the driving force of CuII to receive electrons in its 3d hole to establish an electron configuration close to 3d10 cooperates with the ability of the CN group to donate electrons from its 5σ orbital, resulting in a suppression of the associated positive charge of Cu. The multiple-edge fitting mode has proven to be the most efficient way to analyze the EXAFS image in metal hexacyanoferrates,13,31,39 since the electron scattering takes place in a strongly correlated system. This allows not only to double the number of experimental points, while using the same structural parameters, which in turn are more constrained by the two absorption channels, but also to check the number

Figure 1. XRD pattern of synthesized copper hexacyanoferrate, illustrating experimental, calculated, and residual lines. Miller indexes for the main reflections are also shown.

1.6 refers to structural copper (4b), while potassium ions were placed in 8c positions. In this case, the refined total amount of ferrocyanide vacancies is about 37%. Although the total amount of copper with respect to iron was not held fixed, the final ratio well agrees with both XRF- and XAS-derived ratios. This also helps understanding the coordination numbers used for a straightforward EXAFS analysis at the Cu and Fe K edges. 3.1. XAS Analysis of the Pristine Material at Both Fe and Cu K Edges. Due to the sensitivity of the XAS technique for the atomic species, XAS images at both Cu and Fe K edges of the studied material at the pristine state give information of the charge of the absorbing atom and its coordination geometry pointed of the absorbing atom. Figure 2 displays both X-ray absorption near edge structure (XANES) and EXAFS (experimental best fit) signals as well as their corresponding Fourier transform (FT) curve of Fe and Cu K edges. XANES images are characterized by pre-edge features due to a transition (labeled as A) to bound states,13 which becomes an actual pre-edge peak at Cu K edge. Cu K edge corresponds to the transition 1s−4p, while Fe K edge to the dipole forbidden transition 1s−3d. A main peak (labeled as B)

Figure 2. Fe and Cu K edges. Fit on k2-extracted EXAFS signal and corresponding Fourier transforms (FTs). 15871


Article

The Journal of Physical Chemistry C Table 1. Selected Structural Parameters from EXAFS Fitting Results during Operando Scan of CuHCF/Li Batterya spectrum charge pristine Fe K edge only Fe−C (Å) σ2(Fe−C) (Å2) CN (Å) σ2(CN) (Å2) Cu−N (Å) σ2(Cu−N) (Å2) CN8c M−K/Cub CNchain (Cu−N) a/2c (Å)

1.857(5) 0.0010(4) 1.19(1) 0.008(2) 1.91(3) 0.019(4) 2 fixed

discharge

pristine

10th

18th

27th

33rd

40th

48th

54th

62nd

1.853(7) 0.003(1) 1.189(4) 0.011(2) 1.900(15) 0.009(3) 2 fixed 4.5(5) 4.942

1.855(8) 0.003(1) 1.189(6) 0.011(2) 1.91(1) 0.009(2) 2.0(9) 4.3(3) 4.954

1.862(9) 0.003(1) 1.186(4) 0.009(3) 1.930(6) 0.008(3) 1.6(9) 4.4(2) 4.978

1.870(8) 0.003(1) 1.182(7) 0.011(2) 1.954(6) 0.006(3) 1.5(10) 4.2(2) 5.006

1.887(7) 0.003(1) 1.17(1) 0.012(3) 1.954(4) 0.005(2) 0.5 fixed 4.3(3) 5.011

1.874(10) 0.003(1) 1.178(8) 0.011(5) 1.949(5) 0.006(1) 0.4(3) 4.4(3) 5.001

1.858(9) 0.003(1) 1.189(5) 0.012(4) 1.929(5) 0.010(2) 0.4(3) 4.5(2) 4.976

1.858(7) 0.004(1) 1.189(4) 0.008(3) 1.91(1) 0.010(3) 1.4(10) 4.5(3) 4.957

1.856(7) 0.004(1) 1.189(4) 0.011(3) 1.91(1) 0.011(3) 1.5(10) 4.6(2) 4.955

a

The estimated parameter errors are indicated in parentheses. They have been obtained by contour plots.40 bMixed contribution due to potassium and copper during charge (extraction). During lithiation process, the previous metallic species were still considered to obtain a trend, since Li+ ions slightly contribute to the scattering signal. cData obtained from geometrical evaluation.

Figure 3. Graphic representation and comparison of iron K edge and copper K edge during electrochemical cycling.

occupation of 1/4 of the 8c structural site. More details about the EXAFS analysis will be presented in the following operando XAS section. A better fit is of course obtained by analyzing the Fe edge in single-edge mode. As seen in Figure S2, the agreement between the experimental and theoretical curves is excellent. Table 1 reports the structural parameters and their corresponding associated errors as obtained by the fitting procedures. The single bond length distances of the Fe− C−N−Cu fragment appear similar in the two-fitting modality, but a quite large uncertainty of 0.03 Å results for the Cu−N distance in the single-edge mode. Also, the single edge does not allow the determination of the degree of vacancy in the structure (see below), which is particularly important during the operando scan. Using the multiple-edge fitting, the Fe−C interatomic distance is 1.853(7) Å, a value slightly higher than that quoted for this class of compounds.13,31 The Cu−N bond length results 1.900(15) Å, whereas the CN is 1.189(4) Å. Structurally speaking, the occurrence of 1/3 of [Fe(CN)6]2− ion vacancies reduces the number of Cu−N−C−Fe chains to four from the Cu site, unlike the Fe one, where the number of Fe−C−N−Cu fragments is 6. This could be considered only in the multiple-edge fitting mode, where the best fit is obtained with a value of 4.5(5), in agreement with XRD that returned 37% of ferrocyanide vacancies. Eventually, these results

of vacancies through the coordination number associated to the Cu−N−C−Fe linear chains. This strategy has already been used by our group, and more details about it are available in previous works.13,22 Starting from the structural information from PXRD, the obtained best fit, in terms of k2-extracted EXAFS signal and corresponding Fourier transforms (FTs), is shown in Figure 2 for both Fe and Cu K edges. The good agreement between the experimental curves and the refined ones indicates the reliability of this approach, even though a little discrepancy is observed for the Cu centers (FT above 3 Å). This can be explained by the stoichiometry of the pristine sample, which affects the Cu site exclusively. As shown above by XRD, the chemical formula of CuHCF can be written as 4b K0.34[Cu8c 0.4Cu1.6]Fe(CN)6, where most of the Cu (i.e., 1.6 Cu per unit formula) occupies the 4b site and is involved in the formation of the CuHCF cubic network, whereas the remaining 0.4 Cu per unit of formula occupies the 8c zeolitic site.26 This has a twofold effect: (i) the Cu K edge site is characterized by the presence of a secondary structural site for the Cu and this spoils a little the EXAFS signals as seen before in the FT curve; (ii) additional MS signals at both Cu and Fe K edges due to a Cu···Cu or Fe···Cu interaction at more than 4 Å have been included in the fitting procedure with a coordination number of 2, which corresponds to the 15872


Article


Figure 4. Comparison between pristine, charged, and discharged states at both Fe and Cu K edges. The insets show the magnification of the preedge region.

Figure 5. Pre-edge map for both iron and copper metals during cycling.

confirm that the multiple-edge fitting is the option of choice for this class of compounds. 3.2. Operando XAS Investigation. Figure 3 displays the electrochemical signature as obtained during the operando XAS experiment together with the respective XANES curves at both Fe and Cu K edges. The battery was charged first and then discharged. During charging, the cations in the 8c position (both potassium and copper ions) are extracted from the hexacyanoferrate structure, making feasible the Li+ insertion in the successive discharge step (cf. Supporting Information (SI)). We have recorded a total of 64 spectra, 34 during the charge and 30 in the subsequent discharge. As shown in Figure 4, large changes take place the Cu K edge spectra, particularly in the pre-edge region. The peak at 8984 eV (1s−4p transition) decreases during charging and is restored during discharging (see Figure 5). Simultaneously, a new peak grows during charge at slightly higher energy

(around 8989 eV) corresponding to the 1s−4p transition in CuII, which disappears during the following discharge. These findings are consistent with the reversible oxidation of CuI to CuII during charging and its subsequent reduction upon discharging. This information can only be qualitative due to a lack of suitable reference materials for these systems. On the other hand, almost no substantial modification is visible in the XANES images at the Fe K edge, except for a slight shift of the absorption edge to higher energy. A finestructure analysis of the 1s−3d pre-edge peak, however, adds complementary information. As shown in the inset in Figure 4 (left; see also SI), the peak at 7114 eV shifts toward higher energy at the end of charge and a new weak peak grows at 7111 eV. These modifications are in line with a different occupancy of the t2g levels of the 3d orbitals in a low-spin octahedral environment, which is caused by the oxidation of FeII to FeIII. 15873


Article


Figure 6. Concentration profile (CS,F) and obtained pure spectra (AW,F) of the normalized XANES data taken at the Cu K edge of CuHCF material.

Figure 7. Graphical representation of the main EXAFS parameters trend during cycling. Variation of (a) Fe−C and Cu−N bond lengths and (b) Debye−Waller factors. (c) Copper coordination number to N, or rather the number of Cu−N−C−Fe linear chains. (d) Coordination number to 8c interstitial positions.

resolution analysis (MRC) with the alternate least squares (ALS) algorithm can be used to analyze a full set of XAS data during an electrochemical process in a battery.23 Using this approach, not only the number of pure spectral components but also their existence range during the whole electrochemical process can be revealed. In analytical terms, this means that the operando data matrix XS,W can be decomposed into two matrices CS,F and AW,F, in agreement with the following relationship (eq 1)

These results clearly point out that both Fe and Cu are electroactive in this material. This fact represents a new physical insight with respect to the report of Wessells et al.15 on CuHCF material prepared by co-precipitation, who only observed FeII/FeIII activity, and a confirmation of the mechanism proposed (but not proved) by Okubo et al.41 on the basis of the equivalents of reacted Li. However, Asakura et al.42 previously observed a CuI/CuII activity by ex situ XANES during the lithiation of core@shell CuHCF nanoparticles. 3.3. MCR-ALS Analysis of Spectra Taken at the Cu K Edge. A chemometric approach using the multicurve

X S,W = CS,F ·A W,F 15874

(1) DOI: 10.1021/acs.jpcc.8b03429 J. Phys. Chem. C 2018, 122, 15868−15877

Article


Interestingly, Debye−Waller factors give further insight into the redox system. Indeed, σ2(Fe−C) does not vary almost at all during cycling, assuming a constant value throughout charge and discharge and revealing very little modifications of the Fe local environment. On the other side, σ2(Cu−N) drops during charge, meaning presumably that the ion-extraction process might diminish the copper local asymmetry, whereas it increases during discharge, i.e., ion insertion, mirroring once again the reversibility of the system. Figure 7c displays the Cu−N coordination number, which is an indicator of the Cu−N−C−Fe linear chains and [Fe(CN)6]2− vacancies. Values oscillate smoothly around the constant value of 4.5, strongly suggesting only a small structural strain in the material during cycling. This result agrees with the long cycle life of copper hexacyanoferrate as cathode material, already reported in the literature,15 since the absence of large distortions during cycling is expected to lead to a better cyclability. Figure 7d, finally, correlates the amount of 8c interstitial cation to the cycling advancement. The trend that we observe, even considering the large associated error, agrees with charge/ discharge processes; in other words, the system experiences a decrease in cation occupancy (likely due to both potassium and copper extraction) during charge, and the opposite trend is recorded during discharge. This trend has qualitative value, especially for the extraction process, and derives from a mixed contribution of the interstitial cations. A last remark on EXAFS fitting parameters concerns the a/2 values, that is, the half-cell length (cf. Table 1). In the pristine sample, this value (4.942 Å; a = 9.884 Å) is in fair agreement with the value of 4.998 Å obtained by XRD. This small discrepancy is most probably due to the dehydration of the electrode before the operando XAS measurement, in contrast to the powder employed for XRD measurement. As a matter of fact, negative thermal expansion is common in PBA materials.43 Besides, the lattice expands during ion extraction (charge) and contracts during ion insertion (discharge), which fully agrees with other works in the literature.44,45

where CS,F contains the pure concentration profiles, and AW,F the pure spectral components. In this case, the MCR-ALS algorithm was applied to the whole series of operando Cu K edge XANES images. The graphic results of the two matrices, obtained hypothesizing two components for the electrochemical reaction, the pristine and the charged sample, are shown in Figure 6. Further tests considering three species were also computed, but the results of the chemometric procedure set back to only two components (cf. SI). A complementary analysis was also performed on the Fe XANES images, but the data set was less suitable for MRC-ALS analysis due to the weak dependency of the XANES traces to the intercalation/release reaction. Therefore, it was decided to rely only on the Cu K edge data. The MRC-ALS analysis indicates that: (a) the two components (pristine and charged electrode) progressively transform into each other during the electrochemical reaction; (b) by inverting the electrode polarity, the initial pristine species is restored; (c) the XANES images of the pure species computed by the MRC-ALS analysis are very similar to those measured experimentally (cf. Figures 4 and SI); and (d) because of the existence of only two spectral components, some isosbestic points should be present in the experimental spectra. Indeed, Figure S3 confirms the appearance of this condition at about 8994 and 9008 eV. Finally, the concentration profile described above, where a simple, almost linear and mutual transformation of exclusively two species takes place, demonstrates the weak strain of the material CuHCF, in line with its high cycling stability in the repeated intercalation and release of Li+ ions.15 3.4. Operando EXAFS. To investigate the local structure transformation at the two active metallic centers during the operando scan, an extensive analysis of the EXAFS data was performed. Since the MRC-ALS analysis presented above for the Cu K edge XANES images indicates that the full charge/ discharge cycle can be described as the progressive reversible transformation of only two species, the selection of a limited number of representative spectra for the EXAFS analysis seems largely justified. Therefore, only spectra #1, #10, #18, #27, and #33 during charge and #40, #54, and #62 in the subsequent discharge were considered. Figure S4 displays the best-fit plots of the k2-extracted EXAFS signals for spectra #1 and #33 at both K edges, corresponding to the pristine material and to the end of the first charge, respectively. The theoretical curves match well with the experimental ones, demonstrating the reliability of the data analysis. Table 1 reports some relevant structural parameters (cf. SI for full details of the fitting parameters), such as the interatomic distances and the corresponding Debye−Waller factors as well as the number of Cu−N−C−Fe fragments, which in turn reveals the degree of vacancies into the open framework structure of copper hexacyanoferrate cathode during the operando scan. Figure 7 reports the trends of the main calculated EXAFS parameters during charge and discharge. Figure 7a,b illustrates the variation in bond length and Debye−Waller factor, respectively, of Fe−C and Cu−N. Both Fe−C and Cu−N interatomic distances assume different values in the redox process: in pristine CuHCF (spectrum #1), Fe−C and Cu−N are 1.853(7) and 1.900(15) Å, respectively, whereas in the charged state (spectrum #33), they both stretch to 1.887(7) and 1.954(4) Å. The reverse, instead, is observed at the end of the discharge process (spectrum #62), where these interatomic distances return to 1.856(7) and 1.91(1) Å, respectively.

4. CONCLUSIONS Copper hexacyanoferrate was studied for its interesting performance as positive electrode material in alkali-ion batteries. We focused our work on disclosing the participation of the different metal centers during lithiation/delithiation electrochemical processes, which is often misinterpreted. The electroactivities of copper and iron centers were proven by means of a joint operando XAS/MCR-ALS approach during cycling: operando XAS was used to probe simultaneously the chemical environment of both Cu and Fe in CuHCF, while the application of MCR-ALS to the collected data allowed us to discern the presence of two interconvertible spectral components in the charge/discharge processes. A thorough analysis of XANES data confirmed the reversible redox activity of Cu and Fe in CuHCF, the edge and pre-edge signals being strongly influenced by the oxidation state and electron spin configuration. On the other hand, EXAFS fitting underlines the steady multiplicity of Fe−C−N−Cu linear chains, which evidences the structural stability and low strain of CuHCF during cycling, in agreement with the excellent cyclability reported in the literature. To conclude, it was proven that both metals are electrochemically active during the lithiation/delithiation process of 15875


Article


Material for Nonaqueous Magnesium-Ion Batteries. J. Power Sources 2017, 363, 269−276. (11) Liu, S.; Pan, G. L.; Li, G. R.; Gao, X. P. Copper Hexacyanoferrate Nanoparticles as Cathode Material for Aqueous Al-Ion Batteries. J. Mater. Chem. A 2015, 3, 959−962. (12) Giorgetti, M.; Scavetta, E.; Berrettoni, M.; Tonelli, D. Nickel Hexacyanoferrate Membrane as a Coated Wire Cation-Selective Electrode. Analyst 2001, 126, 2168−2171. (13) Giorgetti, M.; Guadagnini, L.; Tonelli, D.; Minicucci, M.; Aquilanti, G. Structural Characterization of Electrodeposited Copper Hexacyanoferrate Films by Using a Spectroscopic Multi-Technique Approach. Phys. Chem. Chem. Phys. 2012, 14, 5527. (14) Makowski, O.; Stroka, J.; Kulesza, P. J.; Malik, M. A.; Galus, Z. Electrochemical Identity of Copper Hexacyanoferrate in the SolidState: Evidence for the Presence and Redox Activity of Both Iron and Copper Ionic Sites. J. Electroanal. Chem. 2002, 532, 157−164. (15) Wessells, C. D.; Huggins, R. A.; Cui, Y. Copper Hexacyanoferrate Battery Electrodes with Long Cycle Life and High Power. Nat. Commun. 2011, 2, No. 550. (16) Huggins, R. A. ReviewA New Class of High Rate, Long Cycle Life, Aqueous Electrolyte Battery Electrodes. J. Electrochem. Soc. 2017, 164, A5031−A5036. (17) Jiang, P.; Shao, H.; Chen, L.; Feng, J.; Liu, Z. Ion-Selective Copper Hexacyanoferrate with an Open-Framework Structure Enables High-Voltage Aqueous Mixed-Ion Batteries. J. Mater. Chem. A 2017, 5, 16740−16747. (18) Pasta, M.; Wessells, C. D.; Huggins, R. A.; Cui, Y. A High-Rate and Long Cycle Life Aqueous Electrolyte Battery for Grid-Scale Energy Storage. Nat. Commun. 2012, 3, No. 1149. (19) Mullaliu, A.; Sougrati, M.-T.; Louvain, N.; Aquilanti, G.; Doublet, M.-L.; Stievano, L.; Giorgetti, M. The Electrochemical Activity of the Nitrosyl Ligand in Copper Nitroprusside: A New Possible Redox Mechanism for Lithium Battery Electrode Materials? Electrochim. Acta 2017, 257, 364−371. (20) Giorgetti, M.; Stievano, L. X-Ray Absorption Spectroscopy Study of Battery Materials. In X-ray Characterization of Nanostructured Energy Materials by Synchrotron Radiation; Khodaei, M., Petaccia, L., Eds.; InTech: Rijeka, 2017; pp 51−75. (21) Aquilanti, G.; Giorgetti, M.; Dominko, R.; Stievano, L.; Arčon, I.; Novello, N.; Olivi, L. Operando Characterization of Batteries Using X-ray Absorption Spectroscopy: Advances at the Beamline XAFS at Synchrotron Elettra. J. Phys. D Appl. Phys. 2017, 50, No. 074001. (22) Giorgetti, M.; Berrettoni, M. Structure of Fe/Co/Ni Hexacyanoferrate as Probed by Multiple Edge X-Ray Absorption Spectroscopy. Inorg. Chem. 2008, 47, 6001−6008. (23) Conti, P.; Zamponi, S.; Giorgetti, M.; Berrettoni, M.; Smyrl, W. H. Multivariate Curve Resolution Analysis for Interpretation of Dynamic Cu K-Edge X-Ray Absorption Spectroscopy Spectra for a Cu Doped V2O5 Lithium Battery. Anal. Chem. 2010, 82, 3629−3635. (24) Rebuffi, L.; Plaisier, J. R.; Abdellatief, M.; Lausi, A.; Scardi, A. P. Mcx: A Synchrotron Radiation Beamline for X-Ray Diffraction Line Profile Analysis. Z. Anorg. Allg. Chem. 2014, 640, 3100−3106. (25) Rodríguez-Carvajal, J. Recent Advances in Magnetic Structure Determination by Neutron Powder Diffraction. Phys. B 1993, 192, 55−69. (26) Ayrault, S.; Jimenez, B.; Garnier, E.; Fedoroff, M.; Jones, D. J.; Loos-Neskovic, C. Sorption Mechanisms of Cesium on Cu2IIFeII(CN)6 and Cu3II[FeIII(CN)6]2 Hexacyanoferrates and Their Relation to the Crystalline Structure. J. Solid State Chem. 1998, 141, 475−485. (27) Momma, K.; Izumi, F. VESTA 3 for Three-Dimensional Visualization of Crystal, Volumetric and Morphology Data. J. Appl. Crystallogr. 2011, 44, 1272−1276. (28) Hannauer, J.; Scheers, J.; Fullenwarth, J.; Fraisse, B.; Stievano, L.; Johansson, P. The Quest for Polysulfides in Lithium-Sulfur Battery Electrolytes: An Operando Confocal Raman Spectroscopy Study. ChemPhysChem 2015, 16, 2755−2759.

CuHCF and that the system relies on a great electronic and structural stability during a full battery cycle.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.8b03429. Details of the XANES, EXAFS, and MCR-ALS data analysis (Figures S1−S7 and Table 1) and electrochemical curve recorded during the operando experiment (Figure S8) (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. Phone: +33 4 67 14 33 46. *E-mail: [email protected]. Phone: +39 051 20 93 666. Fax: +39 051 20 93 690. ORCID

Angelo Mullaliu: 0000-0003-2800-2836 Marco Giorgetti: 0000-0001-7967-8364 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS XAS measurements at ELETTRA Sincrotrone Trieste were supported by proposal # 20130225 (principal investigator M.G.). The authors acknowledge RFO funding from the University of Bologna.

■

REFERENCES

(1) Wessells, C. D.; Peddada, S. V.; McDowell, M. T.; Huggins, R. A.; Cui, Y. The Effect of Insertion Species on Nanostructured Open Framework Hexacyanoferrate Battery Electrodes. J. Electrochem. Soc. 2012, 159, A98. (2) Moritomo, Y.; Urase, S.; Shibata, T. Enhanced Battery Performance in Manganese Hexacyanoferrate by Partial Substitution. Electrochim. Acta 2016, 210, 963−969. (3) Song, J.; Wang, L.; Lu, Y.; Liu, J.; Guo, B.; Xiao, P.; Lee, J.-J.; Yang, X.-Q.; Henkelman, G.; Goodenough, J. B. Removal of Interstitial H2O in Hexacyanometallates for a Superior Cathode of a Sodium-Ion Battery. J. Am. Chem. Soc. 2015, 137, 2658−2664. (4) Liu, Y.; Wei, G.; Ma, M.; Qiao, Y. Role of Acid in Tailoring Prussian Blue as Cathode for High-Performance Sodium-Ion Battery. Chem. − Eur. J. 2017, 23, 15991−15996. (5) Liu, Y.; He, D.; Han, R.; Wei, G.; Qiao, Y. Nanostructured Potassium and Sodium Ion Incorporated Prussian Blue Frameworks as Cathode Materials for Sodium-Ion Batteries. Chem. Commun. 2017, 53, 5569−5572. (6) Marzak, P.; Yun, J.; Dorsel, A.; Kriele, A.; Gilles, R.; Schneider, O.; Bandarenka, A. S. Electrodeposited Na2Ni[Fe(CN)6] Thin-Film Cathodes Exposed to Simulated Aqueous Na-Ion Battery Conditions. J. Phys. Chem. C 2018, 122, 8760−8768. (7) Wessells, C. D.; Peddada, S. V.; Huggins, R. A.; Cui, Y. Nickel Hexacyanoferrate Nanoparticle Electrodes for Aqueous Sodium and Potassium Ion Batteries. Nano Lett. 2011, 11, 5421−5425. (8) Eftekhari, A. Potassium Secondary Cell Based on Prussian Blue Cathode. J. Power Sources 2004, 126, 221−228. (9) Shiga, T.; Kondo, H.; Kato, Y.; Inoue, M. Insertion of Calcium Ion into Prussian Blue Analogue in Nonaqueous Solutions and Its Application to a Rechargeable Battery with Dual Carriers. J. Phys. Chem. C 2015, 119, 27946−27953. (10) Chae, M. S.; Hyoung, J.; Jang, M.; Lee, H.; Hong, S.-T. Potassium Nickel Hexacyanoferrate as a High-Voltage Cathode 15876


Article

The Journal of Physical Chemistry C (29) Filipponi, A.; Di Cicco, A.; Natoli, C. R. X-Ray-Absorption Spectroscopy and n-Body Distribution Functions in Condensed Matter. I. Theory. Phys. Rev. B 1995, 52, 15122−15134. (30) Filipponi, A.; Di Cicco, A. X-Ray-Absorption Spectroscopy and n-Body Distribution Functions in Condensed Matter. II. Data Analysis and Applications. Phys. Rev. B 1995, 52, 15135−15149. (31) Giorgetti, M.; Berrettoni, M.; Filipponi, A.; Kulesza, P. J.; Marassi, R. Evidence of Four-Body Contributions in the EXAFS Spectrum of Na2Co[Fe(CN)6]. Chem. Phys. Lett. 1997, 275, 108− 112. (32) Hedin, L.; Lundqvist, B. I. Explicit Local Exchange-Correlation Potentials. J. Phys. C: Solid State Phys. 1971, 2064−2083. (33) Krause, M. O.; Oliver, J. H. Natural Widths of Atomic K And L levels Kα X-ray lines and several KLL Auger lines. J. Chem. Phys. Ref. Data 1979, 8, 329−337. (34) Westre, T. E.; Kennepohl, P.; DeWitt, J. G.; Hedman, B.; Hodgson, K. O.; Solomon, E. I. A Multiplet Analysis of Fe K-Edge 1s → 3d Pre-Edge Features of Iron Complexes. J. Am. Chem. Soc. 1997, 119, 6297−6314. (35) Hayakawa, K.; Hatada, K.; D’Angelo, P.; Longa, S. D.; Natoli, C. R.; Benfatto, M. Full Quantitative Multiple-Scattering Analysis of X-Ray Absorption Spectra: Application to Potassium Hexacyanoferrat(II) and -(III) Complexes. J. Am. Chem. Soc. 2004, 126, 15618−15623. (36) Tisato, F.; Marzano, C.; Peruzzo, V.; Tegoni, M.; Giorgetti, M.; Damjanovic, M.; Trapananti, A.; Bagno, A.; Santini, C.; Pellei, M.; Porchia, M.; Gandin, V. Insights into the Cytotoxic Activity of the Phosphane Copper(I) Complex [Cu(Thp)4][PF6]. J. Inorg. Biochem. 2016, 165, 80−91. (37) Pickering, I. J.; George, G. N.; Dameron, C. T.; Kurz, B.; Winge, D. R.; Dance, I. G. X-Ray Absorption Spectroscopy of Cuprous-Thiolate Clusters in Proteins and Model Systems. J. Am. Chem. Soc. 1993, 115, 9498−9505. (38) Chaboy, J.; Muñoz-Páez, A.; Carrera, F.; Merkling, P.; Marcos, E. S. Ab Initio X-Ray Absorption Study of Copper K-Edge XANES Spectra in Cu(II) Compounds. Phys. Rev. B: Condens. Matter Mater. Phys. 2005, 71, No. 134208. (39) Giorgetti, M.; Aquilanti, G.; Ciabocco, M.; Berrettoni, M. Anatase-Driven Charge Transfer Involving a Spin Transition in Cobalt Iron Cyanide Nanostructures. Phys. Chem. Chem. Phys. 2015, 17, 22519−22522. (40) Filipponi, A. Statistical Errors in X-Ray Absorption FineStructure Data Analysis. J. Phys. Condens. Matter 1995, 7, 9343−9356. (41) Okubo, M.; Asakura, D.; Mizuno, Y.; Kudo, T.; Zhou, H.; Okazawa, A.; Kojima, N.; Ikedo, K.; Mizokawa, T.; Honma, I. IonInduced Transformation of Magnetism in a Bimetallic CuFe Prussian Blue Analogue. Angew. Chem., Int. Ed. 2011, 50, 6269−6273. (42) Asakura, D.; Li, C. H.; Mizuno, Y.; Okubo, M.; Zhou, H.; Talham, D. R. Bimetallic Cyanide-Bridged Coordination Polymers as Lithium Ion Cathode Materials: Core@Shell Nanoparticles with Enhanced Cyclability. J. Am. Chem. Soc. 2013, 135, 2793−2799. (43) Adak, S.; Daemen, L. L.; Hartl, M.; Williams, D.; Summerhill, J.; Nakotte, H. Thermal Expansion in 3d-Metal Prussian Blue Analogs - A Survey Study. J. Solid State Chem. 2011, 184, 2854−2861. (44) Ojwang, D. O.; Grins, J.; Wardecki, D.; Valvo, M.; Renman, V.; Häggström, L.; Ericsson, T.; Gustafsson, T.; Mahmoud, A.; Hermann, R. P.; Svensson, G. Structure Characterization and Properties of KContaining Copper Hexacyanoferrate. Inorg. Chem. 2016, 55, 5924− 5934. (45) Renman, V.; Ojwang, D. O.; Valvo, M.; Gómez, C. P.; Gustafsson, T.; Svensson, G. Structural-Electrochemical Relations in the Aqueous Copper Hexacyanoferrate-Zinc System Examined by Synchrotron X-Ray Diffraction. J. Power Sources 2017, 369, 146−153.

15877


Copper Electroactivity in Prussian Blue Based ... - ACS Publications

Recommend Documents