The galvanostatic intermittent titration technique for silicon-based li-ion battery anodes : theory and experimental validation

Lithium ion batteries (LIBs) are the battery of choice for a number of applications, and the LIB market is growing rapidly. Good kinetic abilities is important for many applications, such as electric vehicles, grid storage and portable electronics. The galvanostatic intermittent titration technique (GITT), is a method used to determine the diffusion constant of Li in electrode materials. Three experimental parameters, pulse time, current rate and relaxation time, need to be chosen for GITT experiments. In this study, the influence of these parameters on the calculated diffusion coefficients are explored.\n\nThe effect of these parameters were studied in a main experiment where a series of GITT charge/discharge cycles were conducted with the pulse length being doubled in each iteration. The experiment was run with three current rates, C/10, C/5 and C/2, on different batteries. The cells used in the experiment were half cells with silicon thin films as working electrodes. The thin films had a thickness of 60 nm and 80 nm. A total of 18 cells where used, 9 with each thin film thickness. A secondary experiment was performed on two of the cells, one of each thickness, to test the influence of a longer relaxation time.\n\nThe data resulting from the GITT experiments was analysed using the programming language Python. The diffusion coefficients determined by GITT experiments were found to vary significantly with choice of parameter values. The largest change in results was observed in the second experiment with longer relaxation time. A sufficiently long relaxation time was therefore deemed important to achieve good results with GITT.\n\nThe validity of two simplifications made during in the derivation of the GITT method were tested for different pulse lengths. The pulse lengths where these simplifications are valid depend on the current rate utilized. A lower current rate allows for longer pulse times. For the cells used in this thesis a pulse length between 8 - 64 s at a current rate of C/10 were deemed a good choice, as the simplifications are valid in this range and too much noise at low pulse lengths is avoided.