The effect of the triblock copolymer Pluronic P123 (PP123) on the growth of succinic acid crystals from aqueous solutions is reported at two batch process scales: 10 and 350 mL. The presence of small quantities of PP123 is shown to modify the crystal morphology from plate-like crystals to block-like crystals, in a fully reproducible manner. Increasing the quantity of polymer present, or the concentration of succinic acid used, produces needle-like crystals that are less favorable for processing. In-line process analytical tools (FBRM, PVM, and Raman) were implemented for the larger volume batch processes, allowing the crystallization to be monitored in real time. The effect of the polymer on the metastable zone width (MSZW) has also been determined in designing the crystallization experiments and is presented. In addition, the effect of the individual blocks of the copolymer, poly(ethylene glycol) and poly(propylene glycol), on the crystal morphology was examined, and these findings, together with face indexing and knowledge of the underlying crystal structure, have allowed a possible mechanism to be constructed for the interaction of the polymer with the crystal surface. This mechanism is supported by subsequent recrystallization experiments following washing of the block-like crystals with a nonpolar solvent.

The development of efficient CO₂ capture and utilization technologies driven by renewable energy sources is mandatory to reduce the impact of climate change. Herein, seven imidazolium-based ionic liquids (ILs) with different anions and cations were tested as catholytes for the CO₂ electrocatalytic reduction to CO over Ag electrode. Relevant activity and stability, but different selectivities for CO₂ reduction or the side H₂ evolution were observed. Density functional theory results show that depending on the IL anions the CO₂ is captured or converted. Acetate anions (being strong Lewis bases) enhance CO₂ capture and H₂ evolution, while fluorinated anions (being weaker Lewis bases) favour the CO₂ electroreduction. Differently from the hydrolytically unstable 1-butyl-3-methylimidazolium tetrafluoroborate, 1-Butyl-3-Methylimidazolium Triflate was the most promising IL, showing the highest Faradaic efficiency to CO (>95%), and up to 8 h of stable operation at high current rates (−20 mA & −60 mA), which opens the way for a prospective process scale-up.

We studied the performance of fine-tuning open-source LLMs for a range of different chemical questions. We benchmark their performances against “traditional” machine learning models and find that, in most cases, the fine-tuning approach is superior.