Differential Voltage Analysis of Composite Silicon Anodes for Lithium Ion Cells

Monday, 27 July 2015
Hall 2 (Scottish Exhibition and Conference Centre)
M. J. Lain, S. D. Beattie, M. J. Loveridge, R. Dashwood, and R. Bhagat (Warwick University)
1.            Introduction

Silicon remains as an attractive but challenging anode material for lithium ion cell anodes. It has a high capacity compared to graphite (3579 : 372 mA hr g-1), but significantly greater expansion when fully lithiated (280 : 10 %). Significant advances in the understanding of the lithiation in silicon process have been made through techniques like in situ and ex situ NMR [1]. The initial crystalline silicon is converted into amorphous silicon clusters, then discrete silicon atoms, and then finally crystalline Li15Si4. Information from these spectroscopic techniques can be used to interpret data from more routine cycling methods, during the development of silicon electrodes.

Differential voltage analysis has been applied to various lithium ion cells, to study degradation reactions as a function of cycle number [2, 3]. Standard cycling data is used to calculate either dQ / dV or dV / dQ, which is then plotted against the voltage. In dQ / dV plots, the peaks represent phase equilibria, whereas in dV / dQ plots, the peaks represent phase transitions. This report uses differential voltage analysis to investigate composite silicon electrodes under different test conditions.

 2.            Results

Composite silicon anodes were prepared using a 3 mm silicon powder, with partially neutralised poly acrylic acid as binder. Silicon half cells were cycled at fixed lithiation capacities (1200 and 1800 mA hr g-1), and also at full capacity down to 0.005 V. Figure 1 shows a typical result for full capacity cycling, and Figure 2 the corresponding differential dQ / dV plot. The peak at 0.1 V / 0.3 V was barely present at 1200 mA hr g-1, and declined steadily at higher capacities. The 0.1 V peak is associated with the formation of isolated silicon atoms, via small clusters e.g. Si2 dumb-bells. The peak at 0.2 V is from an earlier stage of the reaction, when amorphous silicon is converted to an extended network, and then larger clusters.

 Cycling at full capacity causes more expansion than the binder can withstand. In turn, this leads to a relatively rapid decline in capacity, and a corresponding increase in the cycling rate. The disappearance of the 0.1 V lithiation peak is associated with a much more rapid drop in electrode voltage. The delithiation peak initially at 0.3 V declines, and the peak voltage drifts to higher values. From approximately 35 cycles onwards, the delithiation voltage jumps straight to 0.5 V. The lithiation and hence subsequent delithiation reactions are rate dependent [1], which may explain the results observed here.

 Further results with three electrode cells will be presented at the conference.

 3              Conclusions

Silicon electrodes in practical cells at realistic current densities show similar multi-stage electrochemical processes as those observed in specialised spectroscopy cells at low current densities. Using dQ / dV analysis in half cells and three electrode cells allows the peaks in simple two electrode full cells to be interpreted.


[1]          C. P. Grey et al  J. Am. Chem. Soc. 131 (2009) p. 9239, Nature Comm. 5 (2014) p. 3217

[2]          I. Bloom et al      J. Power Sources 139 (2005) p. 295, p. 304, 157 (2006) p. 537

[3]          J. R. Dahn et al   J. Electrochem. Soc. 159 (2012) p. A705