Insight into the Structural Disorder in Honeycomb-Ordered Sodium-Layered Oxide Cathodes.

Honeycomb-layered phases Na3M2XO6 (M = Ni, Cu, Co; X = Sb, Bietc.) have been intensively pursued as high-voltage and high-rate capability cathode materials for Na-ion batteries (NIBs), but the crystal structure is not well elucidated. Herein, structural analysis was conducted on pristine Na3Ni2SbO6 material using electron microscopy and associated spectroscopies to reveal its crystallographic features. Experimental observations along multiple zone axes indicate that structural disorder is intrinsic in the pristine Na3Ni2SbO6, characteristic of randomly stacked layers with three variants of monoclinic structure. Stacking disorder is demonstrated by the non-vertical relationship of adjacent Ni2SbO6 layers in [100] zone axis, the different Ni/Sb atomic arrangements in [010] zone axis, and the Ni/Sb random overlap in [001] zone axis. The insight on the structural disorder may inspire studies on their phase transformations upon cycling and provide some clues to potentially solve the voltage/capacity decay problems of these honeycomb-layered materials.

Introduction

Currently, with the rapid increase of lithium-ion battery (LIB) applications in portable devices and electric vehicles, the present lithium resources can hardly meet the growing global demand (Dunn et al., 2011, Hwang et al., 2017, Masquelier and Croguennec, 2013, Ong et al., 2011). To tackle this problem, sodium-ion batteries (NIBs) have been pursued as a promising alternative due to their low cost, abundant supply as well as the electrochemical similarities to LIBs (Hwang et al., 2017, Yabuuchi et al., 2014). Distinct from the common lithium layered oxides, different Na+ coordination in prismatic and octahedral site leads to two different types of layered oxides—P2-type (prismatic) and O3-type (octahedral) —and both have been extensively explored as NIB cathode materials (Delmas and Hagenmuller, 1980, Han et al., 2015, Hwang et al., 2016a, Hwang et al., 2016b, Yabuuchi et al., 2012). Although P2-type cathode materials can deliver a high discharge capacity of more than 190 mAh/g, the low initial Na content and poor cycling stability make them impractical to meet the demands of Na full-cells (Han et al., 2016, Kalluri et al., 2014, Kang et al., 2016, Wang et al., 2015, Wang et al., 2016). Therefore, O3-type excels P2 type due to the sufficient Na content when considered as an ideal Na+ reservoir to realize a practical Na-ion battery (deBoisse et al., 2015, Guo et al., 2015, Yu et al., 2014).

Among these O3-type cathode materials, NaNiO2 is a model intercalation material that possesses high operating voltage and theoretical capacity, but the irreversible phase transformation during the initial cycling leads to rapid capacity fading and poor cycling stability, hindering its practical application in NIBs (Huon Han et al., 2014, Wang et al., 2017a, Wang et al., 2017b, Wang et al., 2014). With this perspective, researchers have tried to introduce 1/3 Sb5+cations substitution of Ni2+ to form the honeycomb-ordered NaNi2/3Sb1/3O2 (Politaev et al., 2010, Seibel et al., 2013). Given the ionic radius and charge difference between Ni2+ and Sb5+, the 2:1 ratio gives rise to a distinct ordered cation arrangement in the Ni2SbO6 layers where each SbO6 octahedron is surrounded by six NiO6 octahedrons, forming a honeycomb-ordered superstructure (Wang et al., 2018, Zvereva et al., 2015). The partial substitution of Ni by Sb enables the stable low valence state of Ni(+2), inducing an intense electrostatic repulsion between Ni2+ and Sb5+ as well as a promoted voltage profile (Wang et al., 2017a, Wang et al., 2017b, You et al., 2017, Yuan et al., 2014) and improved air and thermal stabilities of Na3Ni2SbO6 material (Wang et al., 2019). Interestingly, the imperfection of honeycomb ordering that relates to the Ni2SbO6 layer stacking has been widely noticed, which is sensitive to the synthesis conditions (Dai et al., 2017, Liu et al., 2016, Ma et al., 2015) and similar to that reported in lithium-rich layered oxides (Zhao et al., 2020, Shukla et al., 2015), but the crystallographic characteristics of this structural disorder is still ambiguous.

Direct imaging of this honeycomb disordering at the atomic scale is crucial to correlate the structure and electrochemical performance in these materials. Here we apply atomic-resolution scanning transmission electron microscopy (STEM) to explicitly understand the in-plane honeycomb-ordering of Ni and Sb atoms in the Ni2SbO6 layers and the stacking disorder of Ni2SbO6 layers along c-axis in pristine Na3Ni2SbO6 material. We discover that the structural disorder is a result of random stacking of the Ni2SbO6 layers with three different orientations along the c-axis, which generates disordered sequences and the overlap of Ni/Sb atoms that are evidenced by atomic-resolution STEM images in different zone axes.

Results

The stoichiometric Na3Ni2SbO6 material, as confirmed by inductively coupled plasma-optical emission spectrometry (ICP-OES) analysis (Table S1) and X-ray photoelectron spectroscopy (XPS) results (Figure S1), was synthesized via a simple solution combustion method. Typical scanning electron microscopy (SEM) image, high angle annular dark field image (HAADF-STEM) images, and corresponding energy dispersive X-Ray spectroscopy (EDX) elemental mappings demonstrate its plate-like morphology and uniform element distribution (Figure S2). Figure 1A and Tables S2 show the X-Ray diffraction (XRD) pattern and Rietveld refinement of pristine Na3Ni2SbO6 material with all diffraction peaks that can be well indexed to a monoclinic C2/m structure, as schematically illustrated in Figure 1C. The weak diffraction peaks located at 18–32° 2θ angles represent the honeycomb Ni/Sb ordering superstructure within ab plane (Viciu et al., 2007, Zvereva et al., 2012), where each SbO6 octahedron is surrounded by six NiO6 edge-sharing octahedron in the Ni2SbO6 planes (Wang et al., 2018). The experimental electrochemical cycle (Figure 1B) shows good agreement with calculated characteristic voltage profiles (Liu et al., 2016, Ma et al., 2015, Wang et al., 2019). The selected area electron diffraction (SAED) patterns in [010] and [001] zone axes were collected to provide information on in-plane honeycomb order and out-of-plane stacking disorder (Figures 1D and 1E). Extra streaks and varying degrees of scattering are observed in Figure 1D as the characteristics of stacking disorder along c-axis, indicating the deviation from ordered monoclinic stacking sequence along c-axis (Liu et al., 2016, Meng et al., 2005). As shown in Figure 1E, the simple Rm rhombohedral cell is inadequate for describing the SAED patterns collected in [001]R zones, whereas all observed diffraction spots can be well indexed by the √3 × √3 × 1 P312 trigonal supercell, illustrating the existence of in-plane honeycomb-ordered superstructure (Ma et al., 2015). For Na3Ni2SbO6 material, most investigations claimed that stacking disorder is induced by lateral displacement of the TM layers (Mortemard de Boisse et al., 2016, Wang et al., 2019), yet the explicit atomic-scale structure analysis considering different crystal orientations is scarce.

Figure 1

Preliminary Characterization of Na3Ni2SbO6 Material

(A) The X-ray diffraction (XRD) pattern and Rietveld plot of pristine Na3Ni2SbO6 material.

(B) Calculated and experimental voltage platform of Na3Ni2SbO6.

(C) Crystal structure models of Na3Ni2SbO6 viewed in [001] zone axis (top) and [010] zone axis (bottom).

(D) and (E) Selected area electron diffraction (SAED) patterns taken from the Na3Ni2SbO6 material along the [010]P and [001]P zone axes respectively.“R” subscript indicates Rm space group and “P” subscript indicates P312 space group.

To have a deeper insight into the stacking relationship between layers, detailed atomic-resolution structural characterization of Na3Ni2SbO6 material in [100] zone axis was performed. Figures 2A and 2B show the representative low-magnification and atomic-resolution HAADF-STEM images and fast Fourier transform (FFT) pattern. The bright spots and the relatively darker spots correspond to the Sb and Ni columns, respectively, showing the Sb–Ni–Ni–Sb sequence along b-axis (Lu et al., 2014). In principle, a perfect ordered monoclinic stacking sequence, viewed in the [100] zone axis, exhibits the same Ni and Sb atomic positions along c-axis, as shown in the first crystal model of Figure 2C. In the honeycomb-ordered superstructure that one SbO6 octahedron is surrounded by six NiO6 octahedrons in ab plane, the Ni2SbO6 layer overlaps after each rotation of 60° given the Sb site as the center of symmetry. Therefore, only three different crystal variants exist in ab plane, including [100], [110], and [10] projections, as shown in Figure 2C. Evidently, projections corresponding to [100], [110], and [10] zone axes of the monoclinic phase are observed and distinguished by orange, green, and blue bars in the enlarged STEM image (Figure 2D), which deviate from the perfect vertical alignment of Ni and Sb atoms in the simulated HAADF-STEM image along the [100] zone axis (Figure 2E). Based on the crystal orientation relationships between [100], [110], and [10] zone axes, [100] can directly convert to [110] or [10] by virtue of rotating 60° clockwise or counterclockwise in ab plane (Figure S3A), suggesting that the stacking disorder in pristine Na3Ni2SbO6 material may originate from the rotation of ordered monoclinic Ni2SbO6 layers. This new observation differs from the stacking fault mechanism that was proposed in previous studies, involving the insertion and removal of layers and the deconstruction of ABC close-packed sequences (Aidhy et al., 2016, Ma et al., 2015).

Figure 2

Structural Analysis along [100] Zone Axis

(A and B) (A) A low-magnification image and (B) corresponding magnified HAADF-STEM image of Na3Ni2SbO6 at the [100] zone axis.

(C) Structure models showing the [100], [110], and [10] projections, respectively.

(D) Colour-coded HAADF-STEM image showing the random distribution of the [100], [110], and [10] projections.

(E) Simulated HAADF-STEM image along the [100] zone axis, showing perfect alignments of Ni and Sb atoms.

To verify whether the stacking disorder is induced by rotation of the Ni2SbO6 layers, it is necessary to check the projections in other orientations. Figure 3A exhibits a representative HAADF-STEM image taken from the Na3Ni2SbO6 material tilted to the [010] zone axis. Interestingly, two different projections of monoclinic phase present in the same field of view are distinguished using different color bars, showing different measured angles (72.1° and 80.4°) for ABC stacking sequences (Figure 3A). The atomic arrangements of these two projections are well consistent with the simulated HAADF-QSTEM images along [010] and [310]/[0] zone axes (Figures 3B and 3C). The [310] and [0] zone axes show the same atomic distribution, making them indistinguishable in the HAADF-STEM image. As depicted schematically in Figure 3D, the atomic arrangements corresponding to [010], [0], and [310] orientations will fully overlap after 60 degrees of rotation, which further confirm that the stacking disorder is caused by the rotation of the Ni2SbO6 layers (Figure S3B).

Figure 3

Structural Analysis along [010] Zone Axis

(A) Typical HAADF-STEM image and corresponding FFT pattern of Na3Ni2SbO6 material along [010] zone axis.

(B and C) Simulated HAADF-STEM images of Na3Ni2SbO6 material in [010] and [310]/[0] zone axes showing different atomic arrangements.

(D) Crystal models of the [010], [0], and [310] zone axes.

The two directions that are parallel to ab plane show the existence of different variants of monoclinic phase via the rotation of Ni2SbO6 layers in the Na3Ni2SbO6 material (Figures 2 and 3), but the interpretation of STEM images taken in the direction vertical to ab plane is still ambiguous in the literature. Figure 4A shows an HAADF-STEM image and corresponding FFT pattern in the [001] zone axis. In contrast, unlike those shown in the other two orientations, the atomic arrangements in the [001] zone axis exhibit a perfect honeycomb or hexagon pattern. The in-plane honeycomb structure indicates a single variant of monoclinic phase at first glance. It is worth noting that, however, uniform spot contrast that corresponds to Ni and Sb atoms can be observed in the HAADF-STEM image, which is in contradiction to the perfect honeycomb-ordered structure. The honeycomb-ordered arrangement of Ni and Sb atoms in ab plane can be seen in the simulated HAADF-STEM image (Figure 4B), where the brighter spots represent Sb atoms and the darker spots represent Ni atoms (marked by yellow open circle and blue open circle, respectively). The observed honeycomb-disordered structure (Figure 4A) can be ascribed to random overlap of Ni and Sb atoms in each column when viewed along c-axis, as illustrated schematically in Figure 4C. This Ni/Sb overlap feature is achieved by the rotation of some Ni2SbO6 layers, due to the six-fold overlap feature of the Ni2SbO6 layers in monoclinic structure (Figure S3C). In a word, the stacking disorder is demonstrated by the non-vertical relationship of adjacent Ni2SbO6 layers in [100] zone axis, the different Ni/Sb atomic arrangements in [010] zone axis, and the Ni/Sb random overlap in [001] zone axis.

Figure 4

Structural Analysis along [001] Zone Axis

(A) HAADF-STEM image of Na3Ni2SbO6 at [001] zone axis. Inset: FFT pattern.

(B) Simulated HAADF-STEM image of Na3Ni2SbO6 material in [001] zone axis.

(C) The corresponding model structure of stacking disorder and stacking order.

To appreciate the chemical states of O/Ni/Sb ions across different Ni2SbO6 layers, atomic-resolution electron energy loss spectroscopy (EELS) line-scan analysis was also conducted on the pristine Na3Ni2SbO6 material. Figure 5A shows an HAADF-STEM image with the EELS line-scan ranging from the surface to the bulk along the [010] zone axis. The corresponding O K-, Ni L-, and Sb M-edges spectra were plotted in Figure 5B. The energy loss peaks of O2−, Ni2+, and Sb5+ locate at the same position without noticeable chemical shift (Figures 5B and 5C), implying uniform chemical environments from the surface to Na3Ni2SbO6 bulk. Based on the three-dimensional crystallographic evidence from HAADF-STEM images in various zone axes and chemical information from EDS- and EELS-STEM results, it is reasonable to conclude that the pristine Na3Ni2SbO6 material is consisted of a single monoclinic phase, and the intrinsic structural disorder is associated with three different crystallographic variants.

Figure 5

EELS Analysis of Na3Ni2SbO6 Material

(A) HAADF-STEM image along [010] zone axis showing electron energy loss spectroscopy (EELS) spectrum taken from the horizontal yellow arrow.

(B) Corresponding EELS spectra for O K-, Ni L-, and Sb M-edges.

(C) Corresponding energy profile of Ni L3 edge from the surface to the bulk.

Discussion

In summary, we unambiguously unravel the intrinsic structure disorder in pristine honeycomb-ordered Na3Ni2SbO6 material by means of electron microscopy and associated spectroscopy. We discover that the structural disorder is associated with three different crystallographic variants of a single monoclinic phase, showing in-plane honeycomb structure and out-of-plane stacking disorder induced by the rotation of Ni2SbO6 layers in ab plane. We anticipate that the insight on pristine honeycomb-ordered layered oxides would suggest new clues to researchers exploring the phase transitions occurred in this kind of materials upon cycling and potentially solving their voltage fading and capacity loss issues.

Limitations of the Study

We believe that the structural disorder is a result of random stacking of the Ni2SbO6 layers with three different orientations along the c-axis, which is also applicable to other honeycomb-ordered oxide material, such as Na3Ni2BiO6, Na3Mg2SbO6, and Na2RuO3, etc. However, direct atomic-resolution HAADF-STEM images of them are lacking. Meanwhile, the relationship between the structure and phase transformation is still unrevealed, and some structure analysis on the materials upon cycling is also lacking.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.