Many industrial separations of chemically-similar elements are achieved by solvent extraction, exploiting differences in speciation and solubility across aqueous-organic interfaces. We recently identified [OM4(OH)6(SCN)12]4- (OM4, M=Zr/HfIV) tetrahedral oxoclusters as the main species in industrial processes that produce nuclear-grade Zr and Hf from crude ore. However, isostructural/isoelectronic OM4-oxoclusters do not explain selective extraction of Hf into the organic phase. Here we have characterized heterometal Hf-Zr clusters in solution and the solid-state yielding key information about their fundamentally different chemistry that engenders separation. Clusters prepared with both ammonium (industrial process) and tetramethylammonium counter cations revealed that 1) heterometal clusters (instead of a mixture of homometal clusters) assemble, and 2) Hf-rich OM4 selectively precipitates over Zr-rich OM4, providing a separation process that does not require an organic extractant. Mass spectrometry, small-angle X-ray scattering, solution-state 1H nuclear magnetic resonance (NMR) spectroscopy, and solid-state 17O NMR evidence both mixed-metal speciation and selective Hf-precipitation. Raman spectroscopy suggests greater Zr-ligand lability than Hf-ligand lability, consistent with higher aqueous solubility of Zr-rich clusters, enabling both extraction and precipitation-based separation. Fundamentally, we also identify a key difference between these chemically similar elements that has enabled diversification of Zr-polyoxocation chemistry over the last decade, while Hf-polyoxocation chemistry lags.