“Accelerated discovery of artificial minerals from machine-supported slag admixture and liquid-state processing”


This project targets research areas A and B of PP 2315 “Engineered Artificial Minerals” (EnAM) with a primary focus on the formulation and experimental screening of slag states, liquid-phase processing and crystallization, and the related predictive tools. We will employ high-throughput methodology towards machine-supported discovery of complex slag formulations from material admixture and adapted liquid processing timescales. This will involve high-throughput synthesis and thermal treatment of multi-material slag mixtures in their liquid state, combinatorial analyses of physical property data on solid EnAMs, and the deciphering of up-stream descriptors in a machine workflow and learning environment. EnAMs comprise of a crystalline phase in which critical elements  accumulate, usually precipitated from and embedded in a (residual) amorphous phase. According to the overall concept of PP 2315, this crystal phase – formed from the liquid state – enables fractionation through solid processing; EnAMs are generated from primary slags, slag admixture and/or further additives so that critical elements can be recovered from waste (slag) streams through down-stream mechanical processing. In terms of EnAM formulation, this poses a series of fundamental problems, which we aim to address by this proposal. By nature, slags are chemically complex multi-component materials outside of thermodynamic equilibrium; for this high complexity, their direct (thermodynamic or atomistic) modelling is elusive. EnAMs, on the other hand, may require additional component admixture, which further enhances complexity. Admixture may be by combining different slag/waste streams or by using non-waste additives; in both cases, the employed relations between mixed components must be highly optimized in terms of achieving a useful EnAM state for a given element to be recovered, but also in terms of thermal treatment in the liquid state, down-stream efficiency of component fractionation, sustainability and availability of waste streams, and various other factors (including cost). All these aspects require deep understanding of EnAM formation on the basis of consistent datasets; suitable descriptors which combine physical properties and down-stream processability are urgently needed for EnAM discovery.

As a particular challenge, we aim to overcome the trade-off between high-throughput generation of laboratory materials on the one side, and achieving sufficiently large slag samples to allow for the generation of datasets, which are representative for real world EnAMs on the other side. To this end, we intend to employ combinations of multi-material printing and thermal treatment in gradient environments. The former provides a unique way to fabricate mm-scale slag samples at a processing rate, which cannot be achieved through conventional laboratory experiments (e.g., melt-casting). At the same time, it overcomes the constraints of sample geometry (thickness) and accessible chemistry, which preclude the use of established high-throughput techniques for the lack of reproducing the microstructural complexity and length scale of complex slags. We will implement a wafer-based approach by which up to ~ 100 EnAM slag samples are generated at a time from variable mixtures of primary slags and/or further additives at representative size, processed and characterized for relevant physical data in parallel, and from this, will conceive of holistic descriptors of EnAM discovery.