Direct characterization of the Li intercalation mechanism into α-V₂O₅ nanowires using in-situ transmission electron microscopy Free

Lithium-based rechargeable batteries are widely used in portable electronics and transportation sectors.¹ To be considered as a good cathode candidate for Li-ion batteries, a material needs to demonstrate several features, such as having a cation that can be easily oxidized/reduced, having high capacity and high voltage, which would lead to good energy storage capabilities and good cycling ability through, preferably, an intercalation type reaction with lithium involving minimal structural transformation.² 

Vanadium Pentoxide (V₂O₅) is a well-known cathode candidate for Li ion intercalation and has been studied since the 1970s.^3–5 The α-V₂Ο₅ polymorph has a layered structure consisting of two dimensional sheets of distorted square pyramids of VO₅ sharing corners and edges held together by weak Van der Waal's interactions.^2,6 Due to the weak interlayer bonding, α-V₂O₅ can accommodate ions along the interlayer direction to form intercalated compounds. It crystallizes in an orthorhombic structure with the space group Pmmn. Vanadium exhibits the +5 valence state in V₂O₅, and so, it can be reduced easily to lower valence states with subsequent ionic intercalation.⁷ Also, α-V₂O₅ exhibits a high redox potential vs Lithium (3.5 V)^8,9 and high theoretical capacity (440 mAh/g) upon intercalation of 3 Li ions.¹⁰ 

There has been a wide range of studies of the lithiation behavior of bulk V₂O_5. It has been established that Li insertion into orthorhombic $α−V2O5$ leads to the formation of various intermediate phases depending on temperature, as well as the degree of Li intercalation.^2,6 However, nanoscale electrode materials for lithium ion batteries have several advantages including providing shorter Li diffusion path lengths and accommodating strain upon lithiation better, which can lead to better cycling performace.^11,12 There have been some investigations on the lithiation mechanism of nanoscale V₂O₅.^13,14 In 2007, Chan et al.¹⁴ employed a systematic TEM investigation of chemically lithiated V₂O₅ nanowires and reported the formation of $γ−Li2V2O5$ and $ω−Li3V2O5$ phases upon ex-situ chemical lithiation using electron diffraction and electron energy loss spectroscopy (EELS). It was found that the lithiated nanowire retained its structure, suggesting an intercalation type reaction and various lithiated phases, as well as the pristine V₂O₅ phase, were found to co-exist in different areas of the nanowires that were investigated.

Recently, Strelcov et al.¹⁵ employed in-situ SEM and investigated the morphological changes in a V₂O₅ nanobelt battery during in-situ lithiation using a LiCoO₂ counter electrode and an ionic liquid electrolyte. They found that the V₂O₅ nanobelt can undergo reversible intercalation with Li showing minimal shape distortion and no fracture/crack formation. However, the lithiated V₂O₅ phases were inferred from the ex-situ cycling data, and no direct identification of the distribution or evolution of different phases was reported. It will, therefore, be of great scientific interest to use an in-situ TEM approach for investigating the lithiation process of the V₂O₅ nanowire and to directly determine the structural and phase evolution of the nanowire upon progressive lithiation.

Since its introduction by Huang et al.¹⁶ and Wang et al.,¹⁷ the “open-cell nano-battery” inside the transmission electron microscope has been extensively used to study the lithiation mechanism in anode materials such as SnO₂,^18,19 Si,^20–22 and Ge²² and cathodes such as MnO₂.²³ This method allows for real time observation of structural changes induced by in-situ lithiation, which can be captured via imaging and electron diffraction and getting chemical information using electron energy loss spectroscopy (EELS). This method can, therefore, provide direct and unique insights into the lithiation mechanism of the α-V₂O₅ nanowire cathode.

In this paper, in-situ lithiation studies of single crystalline α-V₂O₅ nanowires are performed using a combination of selected area electron diffraction and electron energy loss spectroscopy to investigate the phase changes and structural evolution upon progressive lithiation. The characterization and in-situ lithiation experiments were performed using a JEOL JEM 3010 operating at 300 keV as well as a probe-side aberration-corrected JEOL JEM ARM200CF with a cold field emission gun operating at 200 keV. For both imaging and EELS, the probe convergence semiangle of 28 mrad was used with a probe current of 19 pA and a probe size of 0.78 Å. The EELS spectrometer collection angle was 45 mrad, and the dispersion was set to 0.1 eV/channel. The raw diffraction data were converted to rotationally averaged line-profiles using the plug-in to the DigitalMicrograph Software Suite.²⁴ 

The α-V₂O₅ nanowires were synthesized via a hydrothermal route.²⁵ For use as cathode in the in-situ lithiation experiments, the yellow powder like nanowires were sonicated in isopropyl alcohol (IPA) for uniform dispersion. The in-situ lithiation experiments are performed using a Nanofactory TEM-STM biasing holder that has a Li metal placed on the STM tungsten tip and the V₂O₅ nanowires glued to a gold tip using conducting epoxy. More discussion on the details of the in-situ setup can be found elsewhere.^{19,20,22,26,27} The Li metal anode was formed by scratching the corresponding foil with a tungsten (W) tip inside an Ar filled glove box and transferred to the holder in an Ar environment. During loading of the tips and subsequent transfer to the TEM, the Li metal is partially oxidized to form a lithium oxide (Li₂O) layer on the surface. Before the tips are brought into contact, EEL spectra were collected from the V₂O₅ nanowire representative of the pristine (unlithiated) phase. Using the piezo-control of the TEM-STM sample stage, the V₂O₅ nanowire was then brought into contact with the Li source and a negative bias was applied on the V₂O₅ nanowire to initiate lithiation. The initial condition was 0 V bias, and subsequently, bias voltages of −0.5 V, −1 V, and −1.3 V were applied. At every intermediate bias step, the system was allowed to reach equilibrium by waiting for around 20 min, and then, the electron diffraction pattern from the V₂O₅ nanowire was captured. At −1.3 V, EEL spectra were collected from the nanowire which is representative of the dominant lithiated phase at that bias condition.

The evolution of the vanadium valence state as a consequence of the in-situ lithiation is quantified using the V L- and O K-edges; however, the overlap of these edges makes it difficult to use the conventional white line ratio method²⁸ for quantification of the vanadium valence. Previous work by Laffont et al.²⁹ and Gallasch et al.^30,31 have established a linear relationship of the V L- and O K-edge energy onsets with the vanadium valence state, in accordance with Kunzl's law.³² This makes it straightforward to compare the experimentally obtained difference in the energies between V L₃ and O K-edge for pristine and lithiated V₂O₅ nanowires and extract information about the Vanadium valence state. In this paper, we use the previously published reference dataset³¹ to calibrate our vanadium valence measurements.

The diameter of the synthesized α-V₂O₅ nanowires is in the range of 50 nm to 130 nm, and the length can be up to several centimeters [see Fig. 1(a)]. For the in-situ experiments, we typically choose ∼600 nm long nanowires, and so, we can establish reliable contact with the tungsten (W) tip containing a Li metal source. The X-ray diffraction (XRD) data [Fig. 1(b)] can be indexed according to α-V₂O₅ [PDF # 01-072-0433]. The high-resolution TEM image in Fig. 1(c) shows distinct lattice fringes with 0.454 nm spacing which are consistent with the spacing between (010) planes of orthorhombic α-V₂O₅. The inset in Fig. 1(c) shows the indexed selected area electron diffraction pattern. In Fig. 1(d), the atomic-resolution high angle annular dark field (HAADF) image clearly shows the vanadium atomic columns, again confirming the single crystalline nature of the nanowires.

Figure 2 shows snapshots taken during the in-situ lithiation process. More specifically, the TEM images shown in Figs. 2(b) and 2(c) clearly demonstrate the formation of a surface layer/coating on the V₂O₅ nanowire. It can be seen that the diameter of the nanowire core and shell taken together grows with applied bias voltage and time [Fig. 2(d)]. This core-shell formation for nanowire electrodes during in-situ lithiation has been previously reported in the case of Si and Ge anodes and identified as a combination of a Li_xSi or Li_xGe alloy and lithium oxide.^20–22 Since V₂O₅ is an intercalation compound, we do not expect any alloying to occur in our experiments. However, we can also identify a very thin layer around the pristine V₂O₅ nanowire in Fig. 2(a), which is most likely very thin V₂O₅, as also verified by further EELS analysis of a pristine V₂O₅ nanowire having similar core-shell appearance (Fig. S3 in the supplementary material). So, it is possible that the surface shell that grows with bias voltage and time is a mixture of Li₂O and lithiated V₂O₅ phases. Li₂O serves as a solid state electrolyte, facilitating the movement of Li⁺ ions into the nanowire core, and so, the lithiation proceeds radially in this case. Our raw diffraction data have been presented in the supplementary material (Fig. S1), while Fig. 3 presents the corresponding rotationally averaged line-profiles. The diffraction profiles from the V₂O₅ nanowire at different bias voltage points can be indexed according to lithium oxide (Li₂O), lithiated V₂O₅ $(α−Li0.04V2O5$ at −0.5 V and −1 V and $γ−Li2V2O5$ at −1.3 V), and unreacted, pristine V₂O₅. As the bias voltage decreases from −0.5 V to −1.3 V, the dominant lithiated phase changes from $α−Li0.04V2O5$ to $γ−Li2V2O5$⁠. Due to the overlap of phases, primary colors were chosen to indicate the three dominant phases (pristine V₂O₅, lithiated V₂O₅, and lithium oxide) and secondary colors mark the overlapped phases [Figs. 3(b)–3(d)]. In each case, the arrowhead color represents the phase and the (hkl) value is indicated which represents the corresponding lattice plane in the case of primary phases or the more intense reflection in the case of secondary/overlapped phases.

The EELS analysis of the V₂O₅ nanowire during the in-situ lithiation at −1.3 V is shown in Fig. 4. Figure 4(a) shows the comparison between the EEL spectra collected from the pristine V₂O₅ nanowire (blue) and EEL spectra collected from the nanowire after in-situ lithiation (red). The V L₃-edge shifts to a lower energy value upon lithiation, suggesting a reduction in the Vanadium valence state. The energy difference between V L₃- and O K-edge is found to be 12.7 eV for pristine spectra (shown in blue) and 13.7 eV for lithiated spectra (shown in red).

The experimentally measured energy difference between the V L₃- and O K-edges, determined by EELS spectra, as presented in Fig. 4(a) can be directly correlated with the valence state of vanadium. The calibration data from the literature³¹ have been shown in Fig. 4(b), and we find that the energy difference of 12.7 eV corresponds to the V⁵⁺ valence state (highlighted in Fig. 4(b) with a blue line) consistent with pristine V₂O₅, and the energy difference of 13.7 eV (highlighted in Fig. 4(b) with a red line) corresponds to the V⁴⁺ valence state, consistent with $γ−Li2V2O5$⁠. Another set of EELS data acquired from a different in situ lithiation experiment was analyzed using a different calibration method, comparing the energy difference between the V L₃-edge and O K-edge onset (peak to trough).^30,33,34 This EELS comparison data and the corresponding calibration data are presented in the supplementary material (Fig. S2).

From our results and analysis presented here, we can conclude that the in-situ lithiation mechanism of single-crystal α-V₂O₅ nanowires can be explained by the following scheme [see Fig. 4(c)]. The surface layer of Li₂O, which forms on the nanowire, acts as a solid state electrolyte that facilitates Li ion diffusion into the center of the nanowire electrode, and the bulk of the nanowire undergoes transformation to the $γ−Li2V2O5$ phase at a bias voltage of −1.3 V that can be identified by electron diffraction. Furthermore, EELS verifies the valence change of Vanadium from V⁵⁺ to V⁴⁺, consistent with the transformation from pristine V₂O₅ to lithiated phase $γ−Li2V2O5$ at −1.3 V. We did not find any evidence, suggesting the formation of higher lithiated phases such as $ω−Li3V2O5$ phase that has been previously reported to form upon ex-situ lithiation of V₂O₅ (Refs. 9 and 35) in our in-situ lithiation experiments.

Our findings are consistent with previous studies of in-situ SEM observation of V₂O₅ nanobelts¹⁵ and TEM characterization of chemically lithiated V₂O₅ nanowires¹⁴ as amorphization or fracturing of the lithiated V₂O₅ nanowires at several hundred nanometers scale were not seen. Importantly, our study elucidates the lithiation scheme of the V₂O₅ nanowire cathode dynamically and provides the direct identification of lithiated phases by combining structural and spectroscopic characterization techniques. Our conclusions suggest that the V₂O₅ nanowires can accommodate the lithium ions without the loss of structural integrity, which should lead to stable cycling performance and longer cycling lifetimes for the battery. These in-situ lithiation experiments with α-V₂O₅ nanowires could also further pave the way for in-situ experiments with Mg employing a suitable ionic liquid electrolyte which would be explored in future studies.

See supplementary material for raw electron diffraction data acquired during the in situ lithiation experiment, an independent set of EELS data investigating the in situ lithiation of the V₂O₅ nanowire, EELS data for the pristine α- V₂O₅ nanowire having core-shell appearance, and EELS line scan data for the α-V₂O₅ nanowire with different pixel dwell times.

This work was supported by the Joint Center for Energy Storage Research (JCESR), an Energy Innovation Hub funded by the U.S. Department of Energy (DOE), Office of Science, Basic Energy Sciences. R. Shahbazian-Yassar and H. Asayesh-Ardakani acknowledge the funding support from the National Science Foundation (NSF-DMR-1620901) for their efforts on in-situ TEM. The acquisition of UIC JEOL JEM ARM200CF was supported by an MRI-R2 grant from the National Science Foundation (Grant No. DMR-0959470). The use of instrumentation at the UIC Research Resources Center (RRC-East) is acknowledged. Dr. Alan Nicholls from UIC Electron Microscopy Service (EMS), RRC East, is also acknowledged for his help and support.