Identifying the Structure of the Intermediate, Li2/3CoPO4, Formed during Electrochemical Cycling of LiCoPO4

In situ synchrotron diffraction measurements and subsequent Rietveld refinements are used to show that the high energy density cathode material LiCoPO4 (space group Pnma) undergoes two distinct two-phase reactions upon charge and discharge, both occurring via an intermediate Li2/3(Co2+)2/3(Co3+)1/3PO4 phase. Two resonances are observed for Li2/3CoPO4 with intensity ratios of 2:1 and 1:1 in the 31P and 7Li NMR spectra, respectively. An ordering of Co2+/Co3+ oxidation states is proposed within a (a × 3b × c) supercell, and Li+/vacancy ordering is investigated using experimental NMR data in combination with first-principles solid-state DFT calculations. In the lowest energy configuration, both the Co3+ ions and Li vacancies are found to order along the b-axis. Two other low energy Li+/vacancy ordering schemes are found only 5 meV per formula unit higher in energy. All three configurations lie below the LiCoPO4–CoPO4 convex hull and they may be readily interconverted by Li+ hops along the b-direction.

: Impurity phases including Li 3 PO 4 , LiPO 3 and Co 2 P were observed in the X-ray diffraction (XRD) pattern obtained after the first 6-hour heating step of the solid-state synthesis of LiCoPO 4 .  The same coloring scheme is used for the arrows (in (a)) and the XRD patterns they point to in (b) -(d).  Figure S4: The first two galvanostatic cycles of LiCoPO 4 at a charge rate of C/20. The blue crosses in (a) indicate when an XRD pattern was taken. The capacities obtained were 149 mAh/g and 139 mAh/g (89 and 83 % of the theoretical capacity) for the first and second cycle, respectively. In Figure S3(b) the charge profiles have been aligned, so that they end at the same capacity. This results in aligning the plateaus and the first period of side reactions (shown in Figure 1(b-d) by red crosses) decreases by 17 mAh/g. Table S1: The unit cell parameters of the end member phases of the two two-phase reactions which take place when LiCoPO 4 is cycled. The standard deviations, resulting from the Rietveld refinements are quoted in the brackets.   The introduction of large CSA interactions (or hyperfine couplings) results in significant decay of the signal, even without homonuclear couplings (due to the finite time length of RF pulses). When homonuclear couplings are introduced, limited magnetization transfer is observed (due to the introduction of an additional decay factor). However, in order to detect significant magnetization transfer between the spins longer recoupling times are required (80 rotor cycles corresponding to 2ms). This is very challenging in practice, due to the fast T 1 relaxation of the intermediate phase.

Experimental Refined
The 31 P T 2 ' values obtained for "Pristine" and "Charged" LiCoPO 4 ( Figure 2), were 1.188 ms and 0.458 ms, respectively. The 7 Li T 2 ' values obtained for "Pristine" and "Charged" LiCoPO 4 , were 1.624 ms and 1.279 ms, respectively. The spectra "Ch185" in Figure 2(a) displays the best resolution of the 31 P peaks of the intermediate phase, whereas the "Charged" spectra in in Figure 2(b) has the best resolution of the 7 Li intermediate peaks, owing to the smaller contribution of the "CoPO 4 " peaks in the 7 Li spectra. For "Ch185" (i.e. LiCoPO 4 charged to a capacity of 185 mAh/g) the 31 P T 2 ' value was 0.708 ms. In order to more accurately calculate the relative integrated intensities of the peaks in the Li 2/3 CoPO 4 spectra, the individual 31 P T 2 ' values were found to be 0.877 ms and 0.576 ms for the peaks at 2610 ppm and 2210 ppm, respectively (with 1 and 3 Co 3+ around P, respectively). In the "Charged" spectra (i.e. fully charged LiCoPO 4 ) the 7 Li T 2 ' values were 1.543 ms and 1.516 ms for the peaks at -125 and 69 ppm, respectively (corresponding to 0 and 3 Co 3+ in the first coordination shell, respectively). The environments with more Co 3+ in the first coordination shell, in both the 31 P and 7 Li spectra, have a faster relaxation, attributed to the additional unpaired electron in Co 3+ compared with Co 2+ and subsequent faster fluctuations of the electronic moment. Figure S7: The different Co first coordination shells around the P atoms when the unit cell is transformed to the following supercells: (2axbxc), (ax2bxc), (axb2xc),  Figure S9: The only three possible (and symmetrically equivalent) Co 2+ /Co 3+ orderings that give rise to two 31 P peaks in the 31 P NMR spectra in a 2:1 ratio (Co 2+ and Co 3+ are in blue and magenta, respectively).   The only three possible (and symmetrically equivalent) Co 2+ /Co 3+ orderings that give rise to two 31 P peaks in the 31 P NMR spectra in a 2:1 ratio (Co 2+ and Co 3+ are in blue and magenta, respectively).  The spin/magnetic states which did not reach electronic convergence are denoted by "nc", while those we did not investigate are denoted by "--". The (**) symbol indicates that the low spin antiferromagnetic state of CoPO 4 is equivalent to the low spin CoPO 4 ferromagnetic state, and its energy is therefore not reproduced in the table. The lowest energy spin and magnetic states, in bold, are those depicted in Figure 7 of the paper. Table S2 summarizes the outcome of all calculations performed on the Li x CoPO 4 (x = 0, 2/3, 1) phases.
First, we note that the relative energy differences between the different spin states and magnetic configurations in the end member structures are highly dependent on the U parameter. Yet, for calculations on the optimized structures, the antiferromagnetic high spin state is always lowest in the energy, whatever the U value used.
Besides, GGA+U results reveal that the influence of magnetic ordering on the energy is quite small. This suggests that we can limit ourselves to calculations on ferromagnetically aligned cells in the GGA case and compare ferromagnetic GGA and GGA+U data without loss of generality.  in pure GGA, an antiferromagnetic state with the Co 3+ ions in a high spin state is favored, suggesting that their may be a delicate balance between the lowest energy Co 3+ spin state and electronic configurational entropy. Nevertheless, configuration b remains second lowest energy irrespectively of the U value used, which indicates that its stabilization is also driven by favorable electrostatics, whereby single Li vacancies (as in configurations a, b, and c) are favored over Li divacancies (as in configurations d, e, and f), in contrast to what has been observed for Li 0.6 FePO 4 . 10 Figure S14: The spin density map of configuration a in the antiferromagnetic high spin state obtained in GGA+U (U eff = 5.48 eV), plotted with an isosurface spin density of 0.01e/Å 3 . The Co 3+ (Co 2+ ) symbols indicate rows of Co 3+ (Co 2+ ) ions.
The spin density map of the lowest energy configuration obtained in GGA+U (U eff = 5.48 eV) shown above indicates that spin density is delocalized more effectively from the Co 3+ ions onto the neighboring O atoms i.e. there is more efficient hybridization between the Co 3+ 3d and O 2p orbitals. This phenomenon can also be deduced by a careful look at the integrated spin densities at the Co 3+ and Co 2+ nuclei obtained in the output of the VASP calculations.