Prediction Put in Practice

The experimental and theoretical paper on The behaviour of lithium electrodes in propylene and ethylene carbonate by Aurbach et al. in 1992 [35] was likely the first to predict electrochemical stabilities of lithium battery materials specifically -using both energy differences and the HOMO-LUMO approach. The energies of EC, PC, and both linear and cyclic radical anions thereof were calculated free or coordinated to LP (Fig. 9.4). From the reduction energies the authors suggested EC to be more prone to reduction. However, different interpretations were possible depending on the assumptions made about the reduction products. As visualized in 

In 2000, Endo et al. returned to a subset of solvents (HF) [15]. A notable difference to the earlier work was a more or less reversed fw trend, with the most spectacular result for EC - the least stable previously - now the most stable. Surprisingly, the authors did not address these qualitative differences. With the compntational power at hand in 2014, it is easy to check these matters and therefore we here repeated the HF calculations, and obtained a different reduction energy for [Li(EC)l (-3.51 eV) compared to the reported, clearly erroneous, value (-2.95 eV). However, as evident from Fig. 9.5, the trends for some of the solvents were still different between the 1998 and 2000 approaches for the solvent-LF pairs. As this was not the case for the solvents, this must originate from the LF interaction description. 

As a bit more complex model, Li and Balbuena in 2000 investigated Ered of EC and PC (DFT/PCM) [10] by a TD cycle. A potential difference identical to the experimental result referred to Aurbach et al. [35] was obtained. However, the main objective was to investigate an updated two-electron reduction mechanism of EC, following nucleophilic attack from groups at the electrode surface [10]. Here also transition state theory was used to estimate reaction activation barriers, equihbtium constants, and rate constants. The results suggested LijCOs as the favourable product overall, together with lithium diethylene carbonate at high EC concentrations. 

With this short expose into the early works, we have shown the scattering in approaches even initially to cover most of the possible types of strategies, methods, and models. With this in mind, we move on to the modem works with a maintained scatter in approaches, but adding also new dimensions in targets and aims for the predictive calcnlations - most notable are the work on redox reaction mechanisms and the trend to introdnee more complex solvent surroundings. 

To test the models against experimentally determined red for EC, TD cycles were used. The predicted red varied between -0.95 to +1.05 V vs. Li/LF [11, 37], depending on the model. Underestimated - negative or close to zero - were obtained for the smaller [Li(EC) ] (n=l-2) super-molecules, while the cluster-continuum models and the [Li(EC) ] ( =4) super-molecule with E,ea of 1.05 V and 

---