Automated microbatch-under-oil phase diagrams to rationalize serial crystallography sample preparation

Serial crystallography relies on the reproducible production of high-density suspensions of microcrystals, yet sample optimization remains a resource-intensive bottleneck. While phase diagrams provide a theoretical framework for controlling crystal size and number, experimental mapping is traditionally hindered by relatively high sample consumption. We present an automated microbatch-under-oil crystallization approach that rapidly maps phase boundaries using only 15-60 µl (∼0.15-3.8 mg) of protein. While this workflow is ideally suited for refining existing hits, it serves as a standalone platform for characterizing the crystallization landscape of new protein targets. The power of this approach lies in the integration of three distinct strategies that exploit the stable chemical environment of microbatch-under-oil. Firstly, we utilize an ingenious diagonal sampling strategy that traverses the phase boundary parallel to the solubility curve by systematically varying protein-to-precipitant ratios, identifying primary nucleation zones with far greater efficiency than traditional orthogonal grids. Secondly, we employ a linked variation of multiple precipitants to reveal morphology-specific regions, such as the rod versus plate transitions crucial for time-resolved experiments. Finally, we incorporate automated seed-stock titration to precisely define the metastable zone, enabling the predictive rescue of nucleation-limited systems. The synergy of these three strategies enables the systematic decoupling of nucleation from growth, providing a rational route to optimize microcrystal density, size and lattice order. Crucially, by eliminating the evaporation-related variables inherent in vapor diffusion, this method ensures that the chemical coordinates identified during screening remain constant during scale-up to larger volumes. This workflow transforms empirical serial crystallography sample preparation into a rational, reproducible and highly efficient process applicable to both the optimization of known conditions and the de novo development of microcrystal suspensions, tailored to the rigorous demands of modern serial diffraction experiments.