Investigating the Particle Formation Mechanism during Precipitation of Ni-Rich Hydroxide Precursors for Cathode Active Materials for Lithium-Ion Batteries

Abstract

The industrial production of lithium-nickel-cobalt-manganese-oxides (LiNi x Co y Mn z O2, with x+y+z = 1) as cathode active material (CAM) for lithium-ion batteries is conducted by precipitation of a mixed metal hydroxide (Ni x Co y Mn z (OH)2) precursor (referred to as pCAM) in a stirred tank reactor, followed by a calcination of the pCAM with a lithium compound at elevated temperatures in a roller hearth kiln.1 Thereby, the physical properties of the resulting CAM powders are significantly affected by those of the associated pCAM.2 Independent of the material composition, the fluidity and filterability of pCAM and CAM powders deteriorates drastically with decreasing particle size, with non-uniformity of the particle size distribution, and with a deviation from a spherical particle shape.3,4 However, a fluent form of the pCAM and CAM powders is imperative for their further processing in tonnage quantities during pCAM and CAM manufacturing to prevent clogging of the powders in the technical equipment. As frequent cleaning and/or pulverization of the materials between process steps is cost-intensive, an economical and efficient manufacturing of pCAM and CAM powders requires a precise control of particle size and sphericity, which in turn requires an in-depth knowledge of the pCAM particle formation mechanism during the precipitation reaction.To gain a fundamental mechanistic understanding about the pCAM particle formation mechanism, the development of the secondary particle size and morphology during the semi-batch precipitation of Ni0.8Co0.1Mn0.1(OH)2 at various stirring speeds (between 550 and 1350 rpm) was monitored by light scattering and SEM imaging. The evolution of the volume-based median particle size (d 50) over the course of the run time (t run) at the different stirring speeds is depicted in Figure 1a. The d 50 of initial particles generated after 5 minutes as well as the d 50 after 8.75 h depends inversely on the applied stirring speed. After an initial nucleation time, the analysis in Figure 1b shows that the size of the approximatively spherical particles at a given stirring speed (for constant feed flow rates) increases linearly with respect to the third square root of run time/run time to the power of one third (t run1/3 ). Thus, the course of particle growth in each experiment can be divided into two distinct regimes: i) for t run1/3 < 1.0 (stage I in Figure 1b), the particle size remains essentially constant; ii) after this initial phase, for t run1/3 ≥ 1.0 (stage II in Figure 1b), the d 50 increases perfectly linearly with. It is concluded that during stage I, the seeding phase, an increasing number of particles with a constant size are being generated, since the overall volume of formed solid is increasing. In contrast, in stage II, the growth phase, the continuing precipitation occurs via deposition on already existing particles without any further increase in particle number, reflected by the linear relationship between particle size and t run1/3 .In light of these results, it is further demonstrated by SEM imaging that, during the seeding phase (stage I), irregularly shaped secondary particle agglomerates are formed, which in the subsequent growth phase (stage II) increase in size by growth of individual plate-like shaped primary particles. Moreover, it is shown that, at constant process parameters, the degree of turbulence quantified by the average energy input and the Reynolds number, both depending on the stirring speed, governs the initial agglomerate size formed during seeding. This not only dictates the growth rate, thus the final particle size, but also the ability of particles to become spherical over t run. Finally, a particle formation mechanism is proposed and compared to mechanisms described in the literature. Figure 1