Unraveling the (De)sodiation Mechanisms of BiFeO3 at a High Rate with Operando XRD

Development of new anode materials for Na-ion batteries strongly depends on a detailed understanding of their cycling mechanism. Due to instrumental limitations, the majority of mechanistic studies focus on operando materials’ characterization at low cycling rates. In this work, we evaluate and compare the (de)sodiation mechanisms of BiFeO3 in Na-ion batteries at different current densities using operando X-ray diffraction (XRD) and ex situ X-ray absorption spectroscopy (XAS). BiFeO3 is a conversion-alloying anode material with a high initial sodiation capacity of ∼600 mAh g–1, when cycled at 0.1 A g–1. It does not change its performance or cycling mechanism, except for minor losses in capacity, when the current density is increased to 1 A g–1. In addition, operando XRD characterization carried out over multiple cycles shows that the Bi ⇋ NaBi (de)alloying reaction and the oxidation of Bi at the interface with the Na–Fe–O matrix are detrimental for cycling stability. The isolated NaBi ⇋ Na3Bi reaction is less damaging to the cycling stability of the material.


Section S1: Structural and Morphological Characterization
BiFeO 3 takes a distorted perovskite structure described in space group R3c using a hexagonal unit cell (Figure S1b, Table S1).Due to the inert lone pair of the 6s orbitals of Bi 3+ the FeO 6 and the BiO 12 polyhedra are significantly distorted compared to the ideal perovskite structure.The secondary particles of the synthesized BiFeO 3 had a wide distribution in particle sizes between 1−500 µm in diameter, as shown by scanning electron microscopy (SEM), and they consisted of primary particles with sizes ranging from 100−500 nm (Figure S1c and d).To improve the electrochemical performance of the material, the sample was ball milled for 20 min at 250 rpm to pulverize the largest secondary particles without drastically changing the nanostructures of the material.The SEM images of the ball milled sample showed that the largest secondary particles after ball milling were <20 µm in diameter, while the primary particles were of the same size as the pristine sample (Figure S1e and f).The XRD pattern of the ball-milled sample was similar to the pristine sample (Figure S1a).The only significant difference was some peak broadening, which is probably due to strain from the ball milling.

Section S2: Electrochemical Characterization
To evaluate the electrochemical performance of ball-milled BiFeO 3 in NIBs, we conducted several cyclic voltammetry (CV) and galvanostatic cycling (GC) measurements in half cells vs Na/Na + .CV measurements with a sweep rate of 0.1 mV s −1 showed two distinct peaks during sodiation at ∼0.6 and ∼0.3 V (Figure S4a).The peak at ∼0.6 V is larger during the first sodiation than in the following cycles, and it has a clear shoulder on the right side starting around 1 V.This peak (and shoulder) should correspond to the conversion reaction from BiFeO 3 to Bi and the first step of the alloying reaction forming NaBi (reaction (1)).In addition, some solid electrolyte interface (SEI) formation likely occurs in this region.The second peak is characteristic of the second alloying step where NaBi transforms into Na 3 Bi (reaction (2)).
During desodiation, the peaks corresponding to the dealloying reaction of Na 3 Bi to NaBi is present at ∼0.7 V and the peak at ∼0.8 V corresponds to the formation of Bi (Figure S4a).In addition to these two peaks, there is a bump between 1.00 and 1.25 V that is related to the further oxidation of Bi [2].At the higher sweep rate of 1 mV s −1 , the peaks are broader and shifted towards lower voltages during sodiation and higher voltages for desodiation compared to the measurement at 0.1 mV s −1 (Figure S4b).This is most pronounced during the first sodiation where the curve peaks around 0.1 V (shifted with 0.5 V), indicating that the initial conversion reaction has limited kinetics.The peaks corresponding to the alloying reactions during the following cycles are not shifted as much (roughly 0.1 V), indicating a lower overpotential and better kinetics.
To further evaluate the electrochemical performance of BiFeO 3 at different rates, GC measurements at 0.1 and 1 A g −1 were performed (Figure S4c and d).The initial sodiation capacity of BiFeO 3 was ∼600 mAh g −1 at 0.1 A g −1 .The measured capacity is higher than the theoretical capacity of the conversion and alloying to Na 3 Bi and a Na−Fe−O matrix with Fe 3+ , which is 514 mAh g −1 .Some of this deviation between theoretical and experimental capacity could be explained by redox activity of Fe, which is known to happen for BiFeO 3 [2].The rest of the deviation probably comes from SEI formation, sodiation of the conductive carbon and uncertainty from weighing of the electrodes.The contribution from the conductive carbon is estimated to be ∼15 mAh g −1 and the uncertainty of the weighing is ∼10% for the electrodes used in this study [3].The most pronounced difference at the different cycling rates is the conversion reaction during the first sodiation where the reaction starts at ∼1.0 V in the 0.1 A g −1 measurement compared to ∼0.5 V at 1 A g −1 .This is consistent with the CV measurements (Figure S4a and b), which also showed a significant difference in the first sodiation.Apart from this, the measurements at the two current densities are very similar, showing that the material can handle high current densities without compromising on performance.The cycling stability of BiFeO 3 is challenging as the capacity drops from ∼450 mAh g −1 at the second cycle to approximately ∼130 mAh g −1 after 25 cycles (Figure S4e).Following this, the capacity is reasonably stable up to at least 100 cycles, where the capacity ends up at ∼120 mAh g −1 and ∼70 mAh g −1 for 0.1 A g −1 and 1 A g −1 , respectively (Figure S4f).We have ∼12 wt% Bi 2 O 3 in our sample, which is known to be electrochemically active in NIBs and show similar performance as BiFeO 3 [4][5][6].The Bi 2 O 3 impurities introduces some uncertainties to the specific capacities presented in this paper, but should not alter the key findings.

Figure
Figure S1: a) XRD with Rietveld refinement of pristine BiFeO3 and the sample after ball milling for 20 min at 250 rpm.b) Visualization of the crystal structure of BiFeO3 (space group: R3c).SEM images of c) pristine BiFeO3 particles with magnification of 100 x, d) pristine BiFeO3 at 25 000 x, e) ball milled BiFeO3 at 100 x, f) ball milled BiFeO3 at 25 000 x.All SEM images are generated by secondary electrons.

Figure S2 :
Figure S2: XRD pattern fitted with Rietveld refinement of pure BiFeO3 obtained by sol-gel synthesis followed by leaching with an aqueous solution of 10% HNO3.

Figure S3 :
Figure S3: Galvanostatic cycling of leached BiFeO3.a) (De)sodiation curves plotted as voltage vs specific capacity and b) specific capacity per cycle plot.

Figure S4 :
Figure S4: Electrochemical characterization of ball-milled BiFeO3.a) and b) cyclic voltammograms obtained from CV measurements with a voltage range of 0.01−2.00V vs Na/Na + and sweep rates of 0.1 mV s −1 and 1 mV s −1 , respectively.c) and d) (de)sodiation curves of selected cycles derived from GC measurements with current densities of 0.1 A g −1 and 1 A g −1 , respectively.e) and f) capacity per cycle plot comparing the cycling behavior at the different current rates, derived from the GC measurements.

Figure S6 :
Figure S6: XAS measurements of the Bi L3 edge.a) XANES and b) FT EXAFS spectra of BiFeO3, Bi2O3 and Bi-foil.c) XANES spectra of BiFeO3 samples after 1 st sodiation and desodiation compared to Bi2O3 and Bimetal foil as references.d) Corresponding FT EXAFS spectra of 1 st sodiation and desodiation.

Figure S7 :
Figure S7: XANES spectra of the Fe K edge.a) Pristine BiFeO3 and BiFeO3 after 1 st sodiation compared to Fe-oxide references and b) cycled BiFeO3 samples at different stages of cycling.

Figure S8 :
Figure S8: Operando XRD visualized as a contour plot (left) with corresponding (de)sodiation curves (right) derived from a measurement of BiFeO3 over the course of 27 cycles with current density of 0.2 A g −1 and voltage range of 0.01−2.00V vs Na/Na + .

Figure S9 :
Figure S9: dQ/dV plot derived from (de)sodiation curves presented in Figure 2b in the main article.

Figure S10 :
Figure S10: Maximum normalized phase fractions for Bi and NaBi vs cycle number extracted from surface Rietveld refinement on operando XRD measurement performed with a specific current of 0.1 A g −1 and a voltage range of 0.01−2.00V vs Na/Na + from our previous study on Bi2MoO6 [3].

Figure
Figure S11: a) (De)sodiation curves and b) capacity per cycle plot from GC measurement on BiFeO3 cycled between 0.01-0.90V with a current density of 0.1 A g −1 .