Characterization of Electrode Materials for Lithium Ion and Sodium Ion Batteries Using Synchrotron Radiation Techniques

Figure 2 shows a typical sequence used for an in situ experiment. After synthesis and characterization of active material powders, composite electrodes are prepared from slurries containing the active material, a binder such as polyvinylidene fluoride (PVDF) and conductive additives such as carbon black or graphite suspended in N-methylpyrrolidinone (NMP), cast onto either aluminum or copper foil current collectors. Aluminum is used for lithium ion battery cathodes and all sodium ion battery electrodes, and copper is used for lithium ion battery anodes. After the electrodes are dried, cut, and weighed, cells are assembled in an inert atmosphere glovebox using microporous separators, the appropriate electrolytic solutions and negative electrodes consisting of either Li or Na foils. These components are then hermetically sealed in a protective pouch, made of polyester, which keeps out air and is reasonably X-ray transparent. Aluminum and nickel tabs are used to make electrical contacts to the positive and negative electrodes, respectively. The Al tabs are ultrasonically welded to the cathode current collectors, while the soft Li or Na metal used as the anode is simply pressed around the Ni tab to make contact. To maintain pressure, the pouch cell is compressed between two metal plates with 2 mm holes cut into them to allow the transmission of X-rays. Poor contact between cell components may result in high overpotentials and premature cutoff, particularly if voltage limits are not adjusted to accommodate the additional overpotential encountered in this configuration. Excessive pressure, on the other hand, may cause cell shorting and failure of the experiment. Better pressure control is achieved when components are first assembled into a coin cell with small holes drilled into the casings and spacers, which is then sealed into the protective pouch after tabs are attached. Wave washers and spacers are used to fill any extra volume in the device, maintain pressure, and ensure good contact among the components.

A small portable potentiostat/galvanostat and laptop computer are then used to perform the electrochemical experiment and collect data at the beam line. One charge-discharge cycle typically takes about 20 hr to complete. The cycle is usually performed galvanostatically (i.e. using constant current) between preselected voltage limits. The sample may either be kept stationary, rocked (left/right or up/down) or rotated around the beam axis in the beam line. Advantages to the last two are that results are obtained over a somewhat larger area of the electrode, effects of preferred orientation in powder-containing electrodes are minimized, and counting statistics are improved.

Transmission XRD ring patterns (see Figure 2, step 5) can be obtained in about 10 sec, with a data readout time of about 1-2 min. Integration of calibrated image patterns yields line scans (intensity vs. Q). Beam line 11-3 at the Stanford Synchrotron Radiation Lightsource uses a single Si(311) monochromator, generating an incident wavelength of approximately 0.97 Å (12,735 eV), though energy fluctuations on the order of a few eVs (~0.01%) due primarily to diurnal cycling (daily temperature fluctuations) are often observed over the course of the lengthy charge and discharge measurements. Thus, image calibration for each scan is essential to de-convolute the diffraction pattern changes. Calibration is performed with The Area Diffraction Machine software developed in conjunction with the 11-3 beam line (http://code.google.com/p/areadiffractionmachine/).

Figure 3 shows in situ XRD data obtained on a Li/Li_x[Ni_0.45Mn_0.45Co_0.05Al_0.05]O₂ cell which underwent charge (in black) and discharge (in green), after calibration and conversion of the ring patterns to line scans. Peaks arising from cell components including the Al current collector, polyester pouch, and polypropylene separator are marked with red and blue dots (lithium metal is essentially transparent to X-rays, but additional peaks will arise if sodium metal anodes are used). Indexed reflections attributable to the Li_x[Ni_0.45Mn_0.45Co_0.05Al_0.05]O₂ active material are marked on the patterns. Because the unit cell parameters changed as a function of x (Li content), peaks due to this phase and the Al current collector overlapped in some of the patterns. The interference from cell components presented significant challenges both for perfect background subtraction and Rietveld refinement of the entire diffraction patterns. To circumvent this problem, backgrounds were manually subtracted, and a limited set of peaks that did not overlap with cell components were selected for the fitting. Unit cell parameters at various states-of-charge were subsequently calculated by a least-squares refinement using the available peak positions and the program CelRef (http://www.ccp14.ac.uk/tutorial/lmgp/celref.htm). The degree to which cell components interfere in patterns obtained from in situ experiments vary depending on the nature of the material under study, and these problems are not always encountered. In that case, any convenient refinement or fitting program may be used to analyze data (GSAS, PowderCell, RIQAS, FullProf, etc.).

Due to time constraints, it is sometimes preferable to perform synchrotron XRD experiments ex situ. It is obviously not practical to perform multiple cycles over a long time in the beam line, for example. Instead, electrodes can be removed from cycled cells, rinsed with solvent to remove residual electrolyte salt, dried, and covered with Kapton film to provide protection from air, for later examination. Additionally, it can be useful to study a few electrodes at different states-of-charge harvested from electrochemical cells, to give an idea of what to expect from a more involved in situ experiment performed later. These experiments are much simpler to carry out and much less time-consuming; several samples can usually be run in an hour. An additional benefit to the ex situ work is the absence of most interfering cell components, although signals from the current collector, binder and conductive additives are usually still observed and the Kapton itself contributes to the background. Caveats for ex situ work are that washing and long or improper storage may change or degrade the sample. In worst-case scenarios, the data obtained ex situ may not even provide relevant information because of these problems. If appropriate caution is maintained, however, ex situ work can still be of some value, although direct observation of processes using in situ configurations is clearly the most desirable option whenever time constraints allow.

Because XAS experiments are element-specific, interference from cell components other than the electrode material of interest are not as problematic as with XRD (assuming that cell hardware does not contain the metals of interest). Only one absorption edge (element) can be measured at a time, however. While switching to a new energy takes only seconds, tuning, changing gains and offsets on the ion chambers, changing reference foils, and purging with gas may take up to an additional ten minutes. Switching from one element to another during an in situ run may result in some loss of data. Meaningful EXAFS data can be difficult to obtain during in situ work, because the structural changes that are occurring often have similar time constants to that of the measurements themselves. Another consideration is that XAS beam lines are often heavily subscribed, meaning limited time for each user. For these reasons, it is generally more practical to carry out XAS experiments on ex situ samples rather than perform in situ work (although see reference 7 for an example of in situ work). Obtaining data on ex situ samples can take anywhere from a few minutes to one hour depending on how many elements are being studied and the facility at which the work is carried out. During each edge measurement, a similar metal foil (e.g. Ni, Mn, or Co) spectrum should be recorded for energy reference. This is carried out simultaneously with the sample measurement. In addition, the user may wish to record data on reference materials containing the metals of interest with known oxidation states, separately, particularly if unusual redox states are involved in the electrochemistry. For example, Li₃MnO₄ was used as a reference for a recent study of a series of lithium manganese oxynitride electrode materials to verify the presence of tetrahedrally coordinated Mn⁵⁺.⁸

Most XAS experiments directed towards studying bulk processes in electrode materials are run in transmission mode, which is suitable when molar concentrations of the elements of interest are above about 5-10% (http://xafstraining.ps.bnl.gov). Best results are obtained when the thickness of the sample, x, is adjusted so that μx<3 above the absorption edge. If the absorption coefficient (μ) is not known (e.g. for complex materials, which includes many battery electrode materials), it can be useful to start with a very small amount of powder sprinkled onto the sticky side of a piece of scotch tape. One or more additional pieces of powdered scotch tape can be attached to the original to increase the signal to the point where the optimum response is obtained (typically, corresponding to one absorption length). For materials where the absorption coefficient is known, the sample can be diluted with BN so that the correct absorption is obtained at a given thickness.

At SSRL, Ni, Mn, and Co K-edges can be studied at beam line 4.1, while Ti and S edges are investigated at beam line 4.3. Detuning the double crystal monochromator by about 30% eliminates higher order harmonics. Calibration is carried out using the first peak in the derivative of the absorption spectra of the reference metals. Duplicate scans can be run and merged after alignment to improve the quality of the data. Artemis/Athena from the software package IFEFFIT are used for analysis⁹. After merging like-scans, the background contribution is subtracted and the data is normalized. EXAFS data is isolated using the AUTOBK function, and is Fourier transformed. Least squares fitting to the Fourier transformed spectrum in R or k space is then used to extract structural information. An example of XAS data, taken at the Mn K edge, is shown in Figure 2, step 5 and the XANES and EXAFS regions are marked on the spectrum.

Figure 1. Schematic of a Li-ion battery with a graphite anode and layered metal oxide cathode undergoing discharge. Used with permission from reference 3.

Figure 2. Typical sequence of an in situ experiment. Steps include 1) preparation and characterization of the sample, 2) preparation of composite electrodes, 3) assembly of pouch cells, 4) set up of an in situ experiment in the beamline, and 5) data acquisition and analysis.

Figure 3. Line scans obtained by integrating image scans on a Li/Li_x[Ni_0.45Mn_0.45Co_0.05Al_0.05]O₂ cell undergoing charge (black) and discharge (green). Reflections attributed to the Al current collector and polymeric cell components (pouch and microporous separator) are marked with blue and red dots, respectively.

Table 1. Table of Materials.

Table 2. Table of Equipment.