84
85
are lithium deficient compared to the Li3PO4 and the deposition of Li helps to compensate this deficiency. In a second step, the surface of Li3PO4 then will decompose in contact with Li and form the binary compounds Li2O and Li3P. In Figure 26, it can also be seen that for longer lithium deposition times, the slope of the curves for different components are different. This is another indication that more than one reaction takes place. Whereas the reaction of Li3PO4 seems to be finished (the slope of P(Li3PO4) is almost the same as the slope of the P(LiPO) signal ), the slope of P(Li3P) is less steep, indicating that although the total intensity is attenuated by the evolving overlayer, more Li3P is formed.
The examination for LiPON–low and LiPON–high shows similar results. Again, Li3PO4 is formed prior to Li3P. In addition, simultaneous to the reduction of „LiPON“ to Li3PO4 and Li2O, Li3N is formed. For LiPON–low – in contrast to the “LiPO” sample – the decay of the intensities already sets in after 600 – 900 s of lithium deposition and the formed interphase is expected to be thinner than that of “LiPO”. The reaction of LiPON–high happens for much longer Li deposition times: Not until lithium has been deposited for 2000 s, the intensities related to SEI species begin to decrease. Therefore, we conclude that the interphase on top of the nitrogen–rich sample is thicker than for the other two solid electrolytes.
It is important to note that unavoidable water and/or oxygen impurities in the residual gas atmosphere of the vacuum chamber heavily affect the in situ XPS experiment and lead to the partial oxidation of the freshly deposited lithium metal. Most importantly, the interpretation of the oxygen signal is complicated as Li2O can either be formed as reaction product of Li and „LiPON“ or due to the reaction of “LiPO(N)” with H2O or O2 from the gas phase. This is reflected in the significantly less steep slope in the Li2O profile as well as in the O 1s spectra (see Figure A 7, Appendix D), which comprise an Li2O component with almost constant intensity even after long Li deposition times – confirming the re–oxidation of the Li film.
Moreover, additional components from LiOH or from Li2O (reaction of Li with H2O or O2
residual gas) contributing to the O 1s signal cannot be excluded. However, expected binding energy differences to signals from these species are too small to be resolved and, due to the blindness of (lab–scale) XPS for hydrogen, the existence of LiOH cannot be proven. The incorporation of oxygen can also cause difficulties in determining the SEI thickness and the lithium deposition rate of lithium (see sections below): If instead of pure lithium metal a
86
mixture of lithium and lithium oxide is deposited on top of the electrolyte, not the entire lithium layer can contribute to the interface reaction. Consequently, the data evaluation will reveal a smaller growth rate than expected because more lithium is needed to achieve a certain SEI thickness. Secondly, lithium oxide with its poor transport properties can lead to a passivation of the SEI and impede the SEI growth. If that happens, even electrolytes that would undergo full decomposition in contact with lithium metal might be stabilized.
Nevertheless, being aware of these external influences, the contribution of additional Li2O can be corrected by a careful data treatment, and a reliable interpretation of the data is still possible.
Figure 26: Change of the intensities of SE and SEI signals as a consequence of the step–wise deposition of lithium in dependence on the deposition time. Substrate signals are shown as empty data points. The dotted line depicts the maximum intensity of P(Li3PO4) the dashed line depicts the maximum intensity of P(Li3P).
10 100 1000
Intensity / Counts
Lithium deposition time / s
LiPO
1000 2000 3000 4000 10 100 1000
Lithium deposition time / s
Intensity / Counts
LiPON-low
1000 2000 3000 4000
0 1000 2000 3000 4000 5000
10 100
Intensity / Counts
Li deposition time / s
LiPON-high
N(double) N(Li3N)
O(„LiPO(N)“) O(Li2O)
P(“LiPO(N)”) P(Li3PO4) P(Li3P)
87
The time–dependent evolution of the relative fractions of different phosphorous and nitrogen species reveals further interesting insights on the reaction. Figure 27 shows the evolution of the phosphorous signals of the three different samples. In the beginning, almost exclusively P(LiPO(N))can be found as no reaction took place yet. Directly after the first Li deposition step the P(LiPO(N))signal decreases and the P(Li3PO4) signal increases. The formation of Li3P sets in later. As can be seen, the P(Li3PO4) signal results in a constant fraction after 1000 s of lithium deposition whereas the P(LiPO(N)) signal decreases continuously and the Li3P signal increases. This finding suggests that Li3PO4 is formed from
“LiPO(N)” and then, upon further lithiation, Li3PO4 is transformed into Li3P. At the same time the formation of Li3PO4 continues. The reduction of Li3PO4 to Li3P seems to occur without large amounts of intermediate products as the fraction of partially reduced LixPy
remains very small (< 3 %). The same observations can also be made for the nitrogen–
containing „LiPON“ samples. The Li3PO4 formation always occurs prior to the formation of Li3P. However, for the nitrogen–containing samples, a constant ratio of the different phosphorous species is achieved after 3000 s of lithium deposition. Note, that only the absolute intensities of all signals decrease (compare Figure 26) and that, once the relative ratio of different phosphorous species is constant, no further decomposition of the electrolyte or SEI formation occurs. This finding suggests that the SEI formation ceases and the lithium|electrolyte interphase is passivated.
0 1000 2000 3000 4000 0
20 40 60 80 100
Phosphorous fraction / %
LiPO
0 1000 2000 3000 4000 LiPO/N Li3PO4 LixPy Li3P LiPON-high
0 1000 2000 3000 4000 Lithium deposition time / s LiPON-low
Figure 27: Evolution of the relative fractions of phosphorous species in the “LiPO(N)” samples.
88
In addition to phosphorous, also the change of the relative intensities of nitrogen is interesting (Figure 28). Exemplarily shown for LiPON–high, the relative intensities of Nd, Nt
and N(Li3N) also change upon lithiation. In the case of LiPON–high, the initial Nt signal vanishes instantly after the beginning of the lithium deposition. Also Nd decreases but Li3N is formed. It is interesting to note that the relative fraction of N(Li3N) begins to cease around the time when the Li3PO4 formation reaches a constant value, suggesting that the decomposition of „LiPON“ stops.