The current generation of Li-ion batteries (LIBs) use carbonaceous anodes, a mature technology capable of achieving capacites close to its theoretical limit of 372 mAh/g. There has been much interest in increasing the capacity of LIBs by using anode materials with greater native capacities. Materials such as silicon or tin have higher theoretical capacities than carbon (4000 and 991 mAh/g, respectively). However, the lithiation cycle causes these materials to expand and contract, by as much as 400%, leading to cracking and degrading performance. One approach to mitigate the effects of volumetric expansion is to include a non-Li metal in an intermetallic compound. In our lab, we investigate the use of Ni (Li inactive) with Sn to improve the cycleability of LIB electrodes.1
It has been shown2 that the composition of electrodeposited alloys can be controlled via current density. The plating bath should have a significantly smaller concentration of the metal with the more positive reduction potential. While this one preferentially plates at small current densities, its ionic current is limited at large current densities, leading to higher a higher deposition rate of the counterpart metal.
In this project, I have used the scanning electron microscope (SEM) to examine the affect of current density and bath composition the morphology of several electrodeposited samples of Ni-Sn and Cu-Sn (copper is also inactive toward Li). The elemental compositions of the samples were evaluated using energy dispersive spectroscopy (EDS), and a cross section of one sample was created using the focused ion beam (FIB) to examine the grain structure of the deposit through its thickness.
Figures 4 through 8 show Ni-Sn electrodes prepared using the modified Watts bath, at different current densities. Even though these electrodes were prepared for the same time as the Ni samples, these films appeared much thinner. The micrographs show that rather than achieving uniform deposition, plating occured from isolated nucleation spots. The area density of nucleation sites increased with increasing current density. High magnification details (Figures 7 and 8) show that high current densities promote dendritic growth.
EDS spectra were collected of the modified Watts bath samples (Figure 9). Apart from the signal from the Cu substrate, in the lower current density samples (2 and 4 mA/cm2) the only element with a measurable quantity was Sn. In the high current density sample (20 mA/cm2), we begin to see a small signal for Ni at about 7.5 keV. High current density favors the deposition of Ni in our system, though the amount of Ni deposited relative to Sn was expected to be much higher.
The samples prepared from the bath with the added surfactant are shown in Figures 10 and 11. It is clear that a small concentration of surfactant has a large affect of the morphology of the deposit, givng both large flakes and thin, fibrous deposits.
Figures 12 through 14 depict the Cu-Sn samples plated at low current density (0.08 mA/cm2); Figures 15 through 17 show the high current density samples (4 mA/cm2). It is shown that low current densities favor the growth of larger crystallites that higher current densities.
Here we can also demonstrate the advantages of using the different detectors in the SEM. Figures 12 and 15 were taken with the in-lens detector; Figures 13 and 16 were taken with the SE2 detector, and the backscatter detector was used in Figures 14 and 17. We can see that the topography of larger features such as in the low current density samples look more natural using the SE2 detector. However, the higher contrast achievable by the in-lens detector is particularly useful for finely detailed grain structures as in the high current density sample. It is interesting to note that any difference in local atomic concentration was not shown by the backscatter detector.
An EDS map of the low current density sample was collected and shown in Figures 18 through 20. It can be seen that the deposit was not entirely uniform, with a Sn-rich region appearing in the bottom portion of Figure 20.
The FIB was used to mill a cross section in low current density Cu-Sn sample (Figures 21 and 22). The Cu substrate is the dark gray portion in the bottom of Figure 22. It is possible that the lighter region above this is a region of diffusion of Cu into the deposit. ImageJ was used to determine a deposit thickness of 1,300 nm, with an average grain diameter of 340 nm.
I would like to thank Hanyuan Zhang and Brian McIntyre for their time and their help in learning the techniques presented here.