Comparison of Surface Properties and Flotation Performance of Products of SAG + Ball Mill and HPGR + Tower Mill

The modification of mineral surface properties during comminution plays an important role in determining subsequent flotation behavior. This study compares the surface characteristics and flotation performance of hematite ore products obtained from two comminution circuits: semi-autogenous grinding plus ball milling (SAG + BM) and high-pressure grinding rolls plus tower milling (HPGR + TM). SAG products with a P95 of 1 mm were collected from an operating concentrator and further ground in a laboratory ball mill, while run-of-mine ore was processed by HPGR to the same P95 and subsequently milled in a tower mill to an equivalent final fineness. Particle size distribution, microcrack development, and surface morphology were characterized using laser particle sizing, scanning electron microscopy (SEM), and atomic force microscopy (AFM), respectively. Surface roughness and specific surface area were quantified by root mean square (RMS) roughness Rq, power spectral density (PSD), and Brunauer–Emmett–Teller (BET) analyses.
Results indicate that HPGR + TM products exhibit a higher density of intra- and intergranular microcracks, increased surface roughness, and larger BET surface areas compared with SAG + BM products. Flotation tests show that the HPGR + TM circuit produced a concentrate with 64.62% TFe at a recovery of 85.06%, while the SAG + BM circuit achieved 61.71% TFe at 86.10% recovery. These correspond to a measurable improvement in concentrate grade of approximately 2.9 percentage points under the investigated conditions. The observed difference is associated with comminution-induced surface modification, which may improve particle hydrophobicity and bubble–particle attachment behavior.
Overall, the study highlights the coupling between comminution-induced surface characteristics and flotation response, emphasizing the importance of considering interactions between interconnected processes at the flowsheet level. Further statistical validation is required to fully quantify the observed performance difference and to support process-scale optimization.