In this work, we have examined the predictive capability of a machine leaning model based on the XGBoost framework as regards the solubility of active pharmaceutical ingredient-like molecules in organic solvents over a wide range of temperatures. A total of 30 binary mixtures has been investigated. The dataset was divided in two sets, with one set for training, testing and validation including solubility data for four solute compounds (butyl paraben, fenofibrate, risperidone, fenoxycarb) consisting of a total of 224 data points, and the second set used for prediction consisting of the solubility data for butamben, with 50 data points in total. The calculated root mean square errors (RMSLE) for the calculated solubility (train, test, validation) were 0.05, 0.09, 0.13 and 0.15, respectively, while the average RMSLE for the predicted solubility of butamben was 0.41. A total of 10 descriptors were considered in this work, comprising parameters for solute (heat of fusion, melting temperature, heat capacity and Hansen solubility parameter), two descriptors representing the solvent (dielectric constant and boiling temperature) as well as temperature, and drug and solvent names. The temperature-dependence of solubility has been captured accurately by setting a constraint on the XGBoost algorithm. A comparison between the performance of the machine learning model proposed and evaluated in this work, and the semi-predictive Flory-Huggins and the temperature-dependent NRTL-SAC models on the other hand, shows that for all the studied compounds, the machine learning model can deliver significantly improved capability to model as well as predict solubility.

The growing demand for sustainable energy has accelerated the recycling of critical materials from spent batteries. In this context, this study investigates eutectic freeze crystallization (EFC) in a synthetic aqueous NiSO₄–CoSO₄–H₂O system, with the long-term goal of applying the method to real industrial lithium-ion battery (LiB) leachate streams. Binary phase diagrams for NiSO₄-H₂O and CoSO₄-H₂O were identified experimentally at eutectic points. In mixed systems, eutectic points were determined from temperature–concentration variations by adding 1–12 wt% CoSO₄ to a ≈ 20 wt% NiSO₄ solution. Similarly, 1–12 wt% NiSO₄ was incrementally added to a ≈ 20 wt% CoSO₄ solution to examine the influence of nickel. The eutectic points of both systems converged as the concentration of the added metal increased.

Experimental analysis (XRD, SEM-EDS, ICP-OES) revealed the formation of separate NiSO₄·7H₂O and CoSO₄·7H₂O crystal phases, and notably, CoNi(SO₄)₂·12H₂O was detected for the first time at eutectic points. Comparative simulations deviated from experimental findings, highlighting the need for improved low-temperature models. After washing, the generated ice contained <0.2 wt% Ni and Co, enabling its reuse as process water. These findings demonstrate EFC's potential for high-purity metal salt recovery and provide the necessary thermodynamic framework for its application to real-world industrial battery recycling.

The chemical equilibria of the liquid-phase sorbitol double dehydration to isosorbide are studied at 403.15–463.15 K using three acidic macroreticular ion-exchange resins: Amberlyst™45, Amberlyst™35, and Purolite®CT482. Liquid molar heat capacities of sorbitol, isosorbide, 1,4-sorbitan, and 2,5-sorbitan were experimentally measured by differential scanning calorimetry at temperatures relevant to the reaction conditions. The product distributions revealed that 1,4-sorbitan, 2,5-sorbitan, and isosorbide were produced in the two-step sorbitol dehydration. The equilibrium constant values show that sorbitol dehydration to sorbitans (first step) is more favoured than the dehydration of 1,4-sorbitan to isosorbide (second step). The reactions thermodynamic properties, crucial towards industrial biorefinery reactor design transforming biomass into value-added chemicals, were determined experimentally and compared to theoretical estimations and to the virtually non-existent literature data. Thermochemical analysis of the reactions revealed that dehydration of sorbitol to 1,4-sorbitan is exothermic with ΔrH°= – (13 ± 2) kJ/mol, while the sorbitol dehydration to 2,5-sorbitan (ΔrH°= 15 ± 3 kJ/mol) and the 1,4-sorbitan dehydration to isosorbide (ΔrH°= 17 ± 7 kJ/mol) are both endothermic. The corresponding entropy changes for these reactions are (13 ± 5), (74 ± 8), (53 ± 16) J/(mol·K), respectively. From these values, the computed standard enthalpy change of formation for isosorbide, 1,4-sorbitan, and 2,5-sorbitan are –755.8, –1045.4, and –1028.0 kJ/mol, respectively. The associated absolute entropy values are 306.4, 296.7, and 303.6 J/(mol·K), respectively. Sorbitol conversion of ∼99.5% with a selectivity to isosorbide of ∼72.3% confirm the potential of ion-exchange resins as catalysts for the studied process at industrial scale.

Recycling of rare earth elements (REEs) has become of paramount importance for permanent magnets used in electric vehicles, wind turbines and motors. There is an imbalance in supply and demand of this commodity and the REE has been identified as critical raw materials by the European Union. This study focuses on the recovery of REEs (La, Pr, Nd, Dy, Y) from sulfuric acid leach solutions using antisolvent crystallization, scaling up the process by hybrid process modelling, and performing environmental impact assessment. Ethanol is used as an antisolvent to crystallize REE2(SO4)3∙8H2O. The impact of inorganic impurities including Cu(II), Co(II), Al(III), B(III), Fe(II) and Fe(III) on the quality of the crystal product, in terms of purity and morphology, has been investigated. Higher purity (above 99%) is obtained for seeded experiments, and the purity is higher for higher seed loading and lower antisolvent dosing rate. Furthermore, the REEs have a tendency to be precipitate as a mixed phase, i.e. REE2(SO4)3∙8H2O. By balancing the addition of antisolvent and seed loading the optimum conditions in terms of high purity and productivity can be found. Scale-up of the process to 100 kg/batch, with solvent recovery of ethanol and H2SO4 (aq.), waste heat reutilization and trade-off analysis, potentially reduces the global warming potential from 40 kg CO2 eq to 0.96 kg CO2 eq /kg REE2(SO4)3∙8H2O. The results can provide valuable insights to understand and optimize the recovery of REEs from sulfate media as a pure concentrate from impure leach liquors. The potential for scaling up the process is also demonstrated resulting in relatively low impacts on global warming.

The thermodynamics and kinetics of ethanol antisolvent crystallization of rare earths from sulfate solutions has been explored, with a view towards separating the rare earths as part of a NdFeB magnet recycling process. The solubility of single and binary metal (Nd, Pr, Fe) phases in aqueous ethanol solutions has been determined. The impact of Fe and Pr on the crystallization of Nd is evaluated, the oxidation kinetics of Fe(II) to Fe(III) quantified, and the influence of Fe oxidation state on the thermodynamics and kinetics of crystallization investigated. Oxidation to Fe(III) is slow, with a half life of approx. 600 h. For pure Nd, the solubility of the obtained, stable sulphate octahydrate decreases exponentially with increased molar organic:aqueous (O/A) ratio, and is well described by the OLI model until O/A=0.2. Pr crystallizes as an isostructural octahydrate with similar solubility. Fe(II) precipitates as a mixed solid phase, with a solubility approximately 40 times higher than the rare earths at O/A=0.2. Fe(III) solutions exhibit liquid–liquid phase separation without precipitation at all evaluated concentrations. Nd and Pr coprecipitate together in proportion to their relative concentrations, with Pr precipitating at concentrations well below its pure component solubility. Fe(II) does not precipitate with Nd at O/A≤0.2 even at high concentration, with significant precipitation as separate particles at higher O/A for all concentrations. The crystallization kinetics and the morphology of the Nd phase is affected by the Fe oxidation state. The work highlights the potential of antisolvent crystallization for selective and efficient separation of REE from Fe.

In this study L‐(+)‐tartaric acid was used to extract metals from either pure cathode material (NMC111) or black mass from spent lithium‐ion batteries. The leaching efficiencies of Li, Co, Ni, and Mn from NMC111 are >87 % at 70 °C, with an initial solid to liquid ratio of 17, and >72.4±1.0 % from black mass under corresponding conditions. The metals tend to form mixed phases in antisolvent crystallization and seeding has a minimal effect on the final solid composition. Impurities influence both crystal nucleation and growth. By controlling the antisolvent addition rate crystal growth can be promoted. The theoretical dielectric constant of the solution is shown to correlate excellently to the recovery efficiency across different antisolvents, where a value <52 results in over 95 % total transition metal recovery efficiency. The correlation can be a powerful tool for quantitative prediction of optimal solvent composition for effective antisolvent crystallization.

Nucleation, the first step of crystallization, is stochastic at small volumes and becomes increasingly deterministic with increasing volume. Translating crystallization processes from the lab to commercial scale remains challenging due to the stochastic nature of nucleation at smaller scales, differences between crystallizer designs, and inconsistencies that arise during scale-up. This study examines how solution volume and agitation affect nucleation kinetics in the crystallization of griseofulvin in methanol. Induction time experiments were conducted at 1 and 100 mL scales using Crystal16 and EasyMax setups, respectively. These results were compared to previously reported experiments conducted at a 20 mL scale using a setup of magnetically stirred solutions, where the induction time was determined through visual analysis of video recordings. GSF nucleated as stable Form I under all investigated conditions. There is a transition from stochastic nucleation observed at 1 and 20 mL scales to a more deterministic process at the 100 mL scale. The nucleation rate exhibits a correlation with differing energy dissipation and solution volume conditions.

Griseofulvin represents a rare case of a close-packed organic apohost that can clathrate selected volatile guests in a solid-gas fashion. Inclusion mechanisms and solvent exchange were investigated by a combination of single crystal and powder X-ray diffraction, coupled to optical microscopy and thermal analyses. In particular, gas diffusion and dissolution/recrystallization are alternatively observed, depending on the host polymorph, as well as the chemical nature of the guest and its physical state.

Rare earth elements (REEs) are important for permanent magnets used in for example wind turbines and motors. There is an imbalance in supply and demand of this commodity and the REE have been identified as critical raw materials by the European Union. This study focuses on recovery of REEs from sulfuric acid solutions using antisolvent crystallization in recycling of magnet waste. Ethanol is used as an antisolvent to crystallize Nd2(SO4)3∙8H2O and (Nd/Dy)2(SO4)3∙8H2O. The impact of the presence of Fe in ferrous and ferric states, and of different seeding strategies, on the quality of the crystal product in terms of purity, crystal size, morphology and agglomeration has been investigated. Higher purity (above 99%) is obtained for seeded experiments and the purity is higher for higher seed loading and lower antisolvent dosing rate. Furthermore, Fe(III) has a higher tendency to be incorporated into the pure Nd phase compared to the Nd phase containing 10% of Dy, while Fe(II) is not detected in any of the phases. By balancing the addition of antisolvent and seed loading the optimum conditions in terms of high purity and productivity can be found. The results provide insights to improve the recovery of REEs as a pure concentrate.

The effect of solvent on active pharmaceutical ingredient (API) nucleation behavior is system-dependent. A better understanding of the role of the solvent in nucleation could help predict and control crystallization. In this work, induction time experiments, spectroscopic analysis, and dynamic light scattering were used to explore the influence of solvent on the polymorphic landscape and the nucleation behavior of griseofulvin (GSF), a medium-sized, flexible, model API. Based on a total of 2960 induction time experiments, the relative ease of nucleation was characterized in three solvents commonly used in the pharmaceutical industry: methanol (MeOH), acetonitrile (ACN), and n-butyl acetate (nBuAc). GSF crystallized as stable Form I in MeOH and as solvated forms in both ACN and nBuAc. GSF nucleated most easily in ACN, followed by nBuAc, while nucleation was most difficult in MeOH. This order was found to correlate with increasing interfacial energy, which was found to be lower in ACN, intermediate in nBuAc, and higher in MeOH, based on a classical evaluation. However, in contrast to classical nucleation theory, which suggests that higher nucleation rates are associated with larger pre-exponential factors, the pre-exponential factor was found to be highest in MeOH, while it remained comparable in ACN and nBuAc. An analysis of the GSF solutions used in the nucleation studies confirmed the presence of mesoscale clusters in ACN and in nBuAc, but not in MeOH. The size and concentration of mesoscale clusters in ACN solution were higher than those in nBuAc, which could explain the higher nucleation rate observed in ACN if the nonclassical nucleation pathway is considered for these solvents.