Abstract
Silicon is a very promising anode material for Li-ion batteries, thanks to its ~ x3 and x10 higher volumetric and specific capacity respectively, compared to commercially available graphite anode. However, this capacity comes from the high alloyable lithium content (up to c-Li15Si4), inducing a proportional volume expansion (up to ~ 280%) at the origin of electrode failure mechanisms, which are:
-The fragmentation and delamination of the electrode coating, because of binder failure;
-The continuous electrolyte degradation, stemming from unprotected active material surfaces, because of SEI destabilization or silicon particles morphology evolution;
-The silicon particle fracture with the accumulation of stress through lithiation/delithiation and an evolution towards a porous particle structure.
In this context, the optimization of the binder properties used for the electrode processing is crucial to mitigate the mechanical degradations, but also the chemical degradation as the binder may act as artificial SEI. In the present work, the crosslinking ability of poly(carboxylic acid) binder through metallic cation-carboxylate coordination bonds is exploited. Among all possible benefits, the highest expectation is that those bonds will endow the reticulated binder network with self-healing properties, allowing a better stress dissipation during cycling.
Experimentally, a poly(carboxylic acid) binder solution was made of sodium carboxymethylcellulose (NaCMC) dissolved in pH ~ 3 buffer solution. Then, zinc(II) salt or oxide was dissolved in the medium, before mixing the as-prepared solution with silicon submicrometric particles (obtained by high-energy ball milling) and graphene nanoplatelets (as conductive additive). Furthermore, electrodes were prepared by tape casting the slurry onto copper current collector. Importantly, inks viscosity was measured to be the same between different formulations, as it might impact electrode microstructure. Finally, electrodes were punched (active mass loading ~ 1.75 mgsi.cm-²), dried and transferred into a glove box for Swagelok cell assembly (LP30 + 10 w% FEC is used as an electrolyte).
The presence of the zinc-carboxylate bonds was confirmed from IR spectroscopy. Moreover, at the silicon particle level, zinc was homogeneously dispersed in a covering and bridging binder phase, as evidenced by STEM-EDX mappings (see Figure 1, where carbon and sodium act as binder markers), further testifying of the binder phase reticulation.
Electrochemical results (Figure 2) for reference (no zinc) and coordinated binder samples clearly show highest discharge capacity retention for the latter. Moreover, zinc containing samples presented most of the time a higher coulombic efficiency, especially at the first cycle.
In order to confirm the assumption that improved cycling with coordinated binders comes from improved mechanical behavior of the electrodes, scratch test and operando dilatometry were performed. The former allows an evaluation of the adhesive (with the current collector) and cohesive strengths of the electrode coating. Results show that coordinated sample was resistant to a ~ 4 times higher scratch loading than the reference. The latter technique probes the thickness variation of the electrode during cycling. It shows that the uncoordinated sample was not able to withstand full lithiation as electrode delamination occurs and that during partial lithiation/delithiation experiments, its relative thickness variation was more than 3 times higher than a coordinated sample (Figure 3, first cycle).
The impact of the reticulated binder on the SEI was also evaluated through impedance spectroscopy, RMN quantification of diamagnetic Li and F, as well as STEM-EDX mappings of electrodes after 1 cycle.
To sum up, the addition of zinc(II) to a poly(carboxylic acid) binder creates metal-carboxylate bonds which further reticulates the binder network. This resulted in improved mechanical properties and integrity of the electrodes, which translates to a better capacity retention. However, silicon electrode failures do not solely come from mechanical degradations thus the impact of the coordinated binder on the SEI will also be discussed.
Figure 1