Assessment of GAFF and OPLS Force Fields for Urea: Crystal and Aqueous Solution Properties.

Molecular simulations such as Monte Carlo, molecular dynamics, and metadynamics have been used to provide insight into crystallization phenomena, including nucleation and crystal growth. However, these simulations depend on the force field used, which models the atomic and molecular interactions, to adequately reproduce relevant material properties for the phases involved. Two widely used force fields, the General AMBER Force Field (GAFF) and the Optimized Potential for Liquid Simulations (OPLS), including several variants, have previously been used for studying urea crystallization. In this work, we investigated how well four different versions of the GAFF force field and five different versions of the OPLS force field reproduced known urea crystal and aqueous solution properties. Two force fields were found to have the best overall performance: a specific urea charge-optimized GAFF force field and the original all-atom OPLS force field. It is recommended that a suitable testing protocol involving both solution and solid properties, such as that used in this work, is adopted for the validation of force fields used for simulations of crystallization phenomena.

Rapid Assessment of Crystal Nucleation and Growth Kinetics: Comparison of Seeded and Unseeded Experiments.

In this work, we outlined an experimental workflow enabling the rapid assessment of primary and secondary nucleation and crystal growth kinetics. We used small-scale experiments in agitated vials with in situ imaging for crystal counting and sizing to quantify nucleation and growth kinetics of α-glycine in aqueous solutions as a function of supersaturation at isothermal conditions. Seeded experiments were required to assess crystallization kinetics when primary nucleation is too slow, especially at lower supersaturations often encountered in continuous crystallization processes. At higher supersaturations, we compared results from seeded and unseeded experiments and carefully analyzed interdependencies of primary and secondary nucleation and growth kinetics. This approach allows for the rapid estimation of absolute values of primary and secondary nucleation and growth rates without relying on any specific assumptions about functional forms of corresponding rate expressions used for estimation approaches based on fitting population balance models. Quantitative relationships between nucleation and growth rates at given conditions provide useful insights into crystallization behavior and can be explored to rationally manipulate crystallization conditions for achieving desirable outcomes in batch or continuous crystallization processes.

Machine Learning-Derived Correlations for Scale-Up and Technology Transfer of Primary Nucleation Kinetics.

Scaling up and technology transfer of crystallization processes have been and continue to be a challenge. This is often due to the stochastic nature of primary nucleation, various scale dependencies of nucleation mechanisms, and the multitude of scale-up approaches. To better understand these dependencies, a series of isothermal induction time studies were performed across a range of vessel volumes, impeller types, and impeller speeds. From these measurements, the nucleation rate and growth time were estimated as parameters of an induction time distribution model. Then using machine learning techniques, correlations between the vessel hydrodynamic features, calculated from computational flow dynamic simulations, and nucleation kinetic parameters were analyzed. Of the 18 machine learning models trained, two models for the nucleation rate were found to have the best performance (in terms of % of predictions within experimental variance): a nonlinear random Forest model and a nonlinear gradient boosting model. For growth time, a nonlinear gradient boosting model was found to outperform the other models tested. These models were then ensembled to directly predict the probability of nucleation, at a given time, solely from hydrodynamic features with an overall root mean square error of 0.16. This work shows how machine learning approaches can be used to analyze limited datasets of induction times to provide insights into what hydrodynamic parameters should be considered in the scale-up of an unseeded crystallization process.

Digital Design of Filtration and Washing of Active Pharmaceutical Ingredients via Mechanistic Modeling.

To facilitate integrated end-to-end pharmaceutical manufacturing using digital design, a model capable of transferring material property information between operations to predict product attributes in integrated purification processes has been developed. The focus of the work reported here combines filtration and washing operations used in active pharmaceutical ingredient (API) purification and isolation to predict isolation performance without the need of extensive experimental work. A fixed Carman-Kozeny filtration model is integrated with several washing mechanisms (displacement, dilution, and axial dispersion). Two limiting cases are considered: case 1 where there is no change in the solid phase during isolation (no particle dissolution and/or growth), and case 2 where the liquid and solid phases are equilibrated over the course of isolation. In reality, all actual manufacturing conditions would be bracketed by these two limiting cases, so consideration of these two scenarios provides rigorous theoretical bounds for assessing isolation performance. This modeling approach aims to facilitate the selection of most appropriate models suitable for different isolation scenarios, without the requirement to use overly complex models for straightforward isolation processes. Mefenamic acid and paracetamol were selected as representative model compounds to assess a range of isolation scenarios. In each case, the objective of the models was to identify the purity of the product reached with a fixed wash ratio and minimize the changes to the crystalline particle attributes that occur during the isolation process. This was undertaken with the aim of identifying suitable criteria for the selection of appropriate filtration and washing models corresponding to relevant processing conditions, and ultimately developing guidelines for the digital design of filtration and washing processes.

Tuning Interfacial Concentration Enhancement through Dispersion Interactions to Facilitate Heterogeneous Nucleation.

Classical molecular dynamics simulations were used to investigate how dispersion (van der Waals) interactions between non-polar, hydrophobic surfaces and aqueous glycine solutions affect the solution composition, molecular orientation, and dynamics at the interface. Simulations revealed that dispersion interactions lead to a major increase in the concentration of glycine at the interface in comparison with the bulk solution, resulting from a competition between solute and solvent molecules to be or not to be near the interface. This can then lead to kinetic and/or structural effects facilitating heterogeneous nucleation of glycine at non-polar surfaces, in agreement with recent observations for tridecane, graphene, and polytetrafluoroethylene. A novel parameterization process was developed to map a model surface with tunable dispersion interactions to heptane, tridecane, and graphite materials. The model surface was capable of reproducing the solution structure observed in fully atomistic simulations with excellent agreement and also provided good agreement for dynamic properties, at a significantly reduced computational cost. This approach can be used as an effective tool for screening materials for heterogeneous nucleation enhancement or suppression, based on non-specific dispersion interactions based on bulk material molecular properties, rather than interfacial functional groups, templating or confinement effects.

Molecular Dynamics Investigation of Clustering in Aqueous Glycine Solutions.

Recent experiments with undersaturated aqueous glycine solutions have repeatedly exhibited the presence of giant liquid-like clusters or nanodroplets around 100 nm in diameter. These nanodroplets re-appear even after careful efforts for their removal and purification of the glycine solution. The composition of these clusters is not clear, although it has been suggested that they are mainly composed of glycine, a small and very soluble amino acid. To gain insights into this phenomenon, we study the aggregation of glycine in aqueous solutions at concentrations below the experimental solubility limit using large-scale molecular dynamics simulations under ambient conditions. Three protonation states of glycine (zwitterion = GLZ, anion = GLA, and cation = GLC) are simulated using molecular force fields based on the 1.14*CM1A partial charge scheme, which incorporates the OPLS all-atom force field and TIP3P water. When initiated from dispersed states, we find that giant clusters do not form in our simulations unless salt impurities are present. Moreover, if simulations are initiated from giant cluster states, we find that they tend to dissolve in the absence of salt impurities. Therefore, the simulation results provide little support for the possibility that the giant clusters seen in experiments are composed purely of glycine (and water). Considering that strenuous efforts are made in experiments to remove impurities such as salt, we propose that the giant clusters observed might instead result from the aggregation of reaction products of aqueous glycine, such as diketopiperazine or other oligoglycines which may be difficult to separate from glycine using conventional methods, or their co-aggregation with glycine.

Temperature Dependence of Solubility Predicted from Thermodynamic Data Measured at a Single Temperature: Application to α, ß, and γ-Glycine.

Understanding of solid-liquid equilibria for polymorphic systems is crucial for rational design and efficient operation of crystallization processes. In this work, we present a framework to determine the temperature dependent solubility based on experimentally accessible thermodynamic data measured at a single temperature. Using this approach, we investigate aqueous solubility of α, ß, and γ-glycine, which, despite numerous studies, have considerable quantitative uncertainty, in particular for the most stable (γ) and the least stable (ß) solid forms. We benchmark our framework on α-glycine giving predictions in excellent agreement with direct solubility measurements between 273-340 K, using only thermodynamic data measured at the reference temperature (298.15 K). We analyze the sensitivity of solubility predictions with respect to underlying measurement uncertainty, as well as the excess Gibbs free energy models used to derive required thermodynamic quantities before providing solubility predictions for ß and γ-glycine between 273-310 and 273-330 K, respectively. Crucially, this approach to predict solubility as a function of temperature does not rely on measurement of solute melting properties which will be particularly useful for compounds that undergo thermal decomposition or polymorph transition prior to melting.

Modeling Diffusive Mixing in Antisolvent Crystallization.

Diffusion controls local concentration profiles at interfaces between segregated fluid elements during mixing processes. This is important for antisolvent crystallization, where it is intuitively argued that local concentration profiles at interfaces between solution and antisolvent fluid elements can result in significant supersaturation overshoots over and above that at the final mixture composition, leading to poorly controlled nucleation. Previous work on modeling diffusive mixing in antisolvent crystallization has relied on Fickian diffusion, where concentration gradients are the driving force for diffusion. This predicts large overshoots in the supersaturation at interfaces between solution and antisolvent, as is often intuitively expected. However, chemical potential gradients provide a more physically realistic driving force for diffusion, and in highly nonideal solutions, such as those in antisolvent crystallization, this leads to nonintuitive behavior. In particular, as solute diffusion toward antisolvent is severely hindered, it can diffuse against its concentration gradient away from antisolvent. We apply thermodynamically consistent diffusion model based on the multicomponent Maxwell-Stefan formulation to examine diffusive mixing in a nonideal antisolvent crystallization system. Large supersaturation overshoots above that at the final mixture composition are not found when a thermodynamically consistent approach is used, demonstrating that these overshoots are modeling artifacts and are not expected to be present in physical systems. In addition, for certain conditions, localized liquid-liquid spinodal demixing is predicted to occur during the diffusive mixing process, even when the final mixture composition is outside the liquid-liquid phase separation region. Intermittent spinodal demixing driven by diffusive mixing may provide a novel explanation for differences of nucleation behaviors among various antisolvents.

The unexpected dominance of secondary over primary nucleation.

It is still a challenge to control the formation of particles in industrial crystallization processes. In such processes, new crystals can be generated either by primary or secondary nucleation. While in continuous stirred tank crystallization processes, secondary nucleation is thought to occur due to the shear or attrition of already present larger crystals; in antisolvent crystallization processes, where mixing at the inlets locally causes high supersaturations, primary nucleation is understood to be the main mechanism. We aim to show here that secondary nucleation is the dominant nucleation mechanism, even under conditions that are generally considered to be dominated by primary nucleation mechanisms. Measurements of primary and secondary nucleation rates under similar industrial crystallization conditions of sodium bromate in water, sodium chloride in water, glycine in water and isonicotinamide in ethanol show that the secondary nucleation rate is at least 6 orders of magnitude larger in all these systems. Furthermore, seeded fed-batch and continuous antisolvent crystallizations of sodium bromate under high local supersaturation, seeded with crystals of a specific handedness, result in a close to chirally pure crystalline product with the same handedness. This shows that indeed, enantioselective secondary nucleation is the dominant mechanism in these antisolvent crystallizations. It is even possible to use the enantioselective secondary nucleation mechanism to control the product chirality in such a process, making antisolvent crystallization a viable crystallization-enhanced deracemization technique, having a superior productivity compared to other crystallization-enhanced deracemization methods. Our finding of a dominant secondary nucleation mechanism, rather than primary nucleation, will have a strong impact on nucleation control strategies in industrial crystallization processes.

Heat Transfer and Residence Time Distribution in Plug Flow Continuous Oscillatory Baffled Crystallizers.

Heat transfer coefficients in a continuous oscillatory baffled crystallizer (COBC) with a nominal internal diameter of 15 mm have been determined as a function of flow and oscillatory conditions typically used under processing conditions. Residence time distribution measurements show a near-plug flow with high Peclet numbers on the order of 100-1000 s, although there was significant oscillation damping in longer COBC setups. Very rapid heat transfer was found under typical conditions, with overall heat transfer coefficients on the order of 100 s W m-2 K-1. Furthermore, poor mixing in the COBC cooling jacket was observed when lower jacket flow rates were implemented in an attempt to decrease the rate of heat transfer in order to achieve more gradual temperature profile along the crystallizer length. Utilizing the experimentally determined overall heat transfer coefficients, a theoretical case study is presented to investigate the effects of the heat transfer rate on temperature and supersaturation profiles and to highlight potential fouling issues during a continuous plug flow cooling crystallization.

Computer Aided Design of Solvent Blends for Hybrid Cooling and Antisolvent Crystallization of Active Pharmaceutical Ingredients.

Choosing a solvent and an antisolvent for a new crystallization process is challenging due to the sheer number of possible solvent mixtures and the impact of solvent composition and crystallization temperature on process performance. To facilitate this choice, we present a general computer aided mixture/blend design (CAMbD) formulation for the design of optimal solvent mixtures for the crystallization of pharmaceutical products. The proposed methodology enables the simultaneous identification of the optimal process temperature, solvent, antisolvent, and composition of solvent mixture. The SAFT-γ Mie group-contribution approach is used in the design of crystallization solvents; based on an equilibrium model, both the crystal yield and solvent consumption are considered. The design formulation is implemented in gPROMS and applied to the crystallization of lovastatin and ibuprofen, where a hybrid approach combining cooling and antisolvent crystallization is compared to each method alone. For lovastatin, the use of a hybrid approach leads to an increase in crystal yield compared to antisolvent crystallization or cooling crystallization. Furthermore, it is seen that using less volatile but powerful crystallization solvents at lower temperatures can lead to better performance. When considering ibuprofen, the hybrid and antisolvent crystallization techniques provide a similar performance, but the use of solvent mixtures throughout the crystallization is critical in maximizing crystal yields and minimizing solvent consumption. We show that our more general approach to rational design of solvent blends brings significant benefits for the design of crystallization processes in pharmaceutical and chemical manufacturing.

Influence of Ions and pH on the Formation of Solid- and Liquid-like Melanin.

Melanin is a natural pigment with broadband absorption and effective ability to dissipate the energy absorbed. The macromolecular structure of melanin shows a delicate balance between short-range ordered and disordered structures without being a random aggregate. The presence of ions or the variation in pH or ionic strength can alter the self-assembly process which subsequently changes the structure of melanin. To understand these relationships, this study investigates the influence of ions and pH in melanin formation. The types of ions present and pH have a profound influence on the formation and structure of melanin particles, while only minor changes are observed in the absorption and excitation-emission analysis. In some conditions, the formation of discernible particles with significant refractive index contrast is avoided while retaining the spectroscopic characteristics of melanin, leading to liquid-like melanin. These findings identify potential pathways which can be used to manipulate the melanin macromolecular structure while providing the desired spectral properties to enable novel bio-engineering applications.

Chain-End Modifications and Sequence Arrangements of Antimicrobial Peptoids for Mediating Activity and Nano-Assembly.

Poly(N-substituted glycine) "peptoids" are an interesting class of peptidomimics that can resist proteolysis and mimic naturally found antimicrobial peptides (AMPs), which exhibit wide spectrum activity against bacteria. This work investigates the possibility of modifying peptoid AMP mimics (AMPMs) with aliphatic lipid "tails" to generate "lipopeptoids" that can assemble into micellar nanostructures, and evaluates their antimicrobial activities. Two families of AMPMs with different distributions of hydrophobic and cationic residues were employed-one with a uniform repeating amphiphilicity, the other with a surfactant-like head-to-tail amphiphilicity. To further evaluate the interplay between self-assembly and activity, the lipopeptoids were variously modified at the AMPM chain ends with a diethylene glycol (EG2) and/or a cationic group (Nlys-Nlys dipeptoid) to adjust amphiphilicity and chain flexibility. Self-assembly was investigated by critical aggregation concentration (CAC) fluorescence assays and dynamic light scattering (DLS). The structure of a key species was also verified by small-angle X-ray scattering (SAXS) and cryo-electron microscopy (cryo-EM). To screen for antibacterial properties, we measured the minimum inhibitory concentrations (MIC) against S. aureus, E. coli, and P. aeruginosa. We found that certain combinations of lipid tail and AMPM sequences exhibit increased antibacterial activity (i.e., decreased MICs). Perhaps counter-intuitively, we were particularly interested in increased MICs in combination with low CACs. Concealing antimicrobial interactions due to packing of AMPMs in nano-assemblies could pave the way to AMPMs that may be "inert" even if unintentionally released and prevent microbes from gaining resistance to the lipopeptoids. Overall, incorporation of EG2 significantly improved lipopeptoids packing while the hydrophobic tail length was found to have a major influence over the MIC. One particular sequence, which we named C15-EG2-(kss)4, exhibited a very low CAC of 34 µM (0.0075 wt.%) and a significantly increased MIC above values for the unmodified AMPM. With the sequence design trends uncovered from this study, future work will focus on discovering more species such as C15-EG2-(kss)4 and on investigating release mechanisms and the potency of the released lipopeptoids.

ProteinVR: Web-based molecular visualization in virtual reality.

Protein structure determines biological function. Accurately conceptualizing 3D protein/ligand structures is thus vital to scientific research and education. Virtual reality (VR) enables protein visualization in stereoscopic 3D, but many VR molecular-visualization programs are expensive and challenging to use; work only on specific VR headsets; rely on complicated model-preparation software; and/or require the user to install separate programs or plugins. Here we introduce ProteinVR, a web-based application that works on various VR setups and operating systems. ProteinVR displays molecular structures within 3D environments that give useful biological context and allow users to situate themselves in 3D space. Our web-based implementation is ideal for hypothesis generation and education in research and large-classroom settings. We release ProteinVR under the open-source BSD-3-Clause license. A copy of the program is available free of charge from http://durrantlab.com/protein-vr/, and a working version can be accessed at http://durrantlab.com/pvr/.

Interfacial Concentration Effect Facilitates Heterogeneous Nucleation from Solution.

Crystal nucleation from solution plays an important role in environmental, biological, and industrial processes and mainly occurs at interfaces, although the mechanisms are not well understood. We performed nucleation experiments on glycine aqueous solutions and found that an oil-solution interface dramatically accelerates glycine nucleation compared to an air-solution interface. This is surprising given that nonpolar, hydrophobic oil (tridecane) would not be expected to favor heterogeneous nucleation of highly polar, hydrophilic glycine. Molecular dynamics simulations found significantly enhanced vs depleted glycine concentrations at the oil-solution vs air-solution interfaces, respectively. We propose that this interfacial concentration effect facilitates heterogeneous nucleation, and that it is due to dispersion interactions. This interface effect is distinct from previously described mechanisms, including surface functionalization, templating, and confinement and is expected to be present in a wide range of solution systems. This work provides new insight that is essential for understanding and controlling heterogeneous nucleation.

Thermal inactivation of jack bean urease.

Thermal inactivation of oligomeric enzymes results in complex structural changes. This work deals with thermal inactivation of a native hexamer, jack bean urease. In order to find the mechanism and kinetics of thermal inactivation corresponding well with the modification of tertiary and quaternary structure of this enzyme, several types of experiments were carried out in the temperature range of 65-85 °C. Inactivation data exhibited the characteristic biphasic character. Dynamic light scattering experiments revealed a significant increase of the mean hydrodynamic radius of urease with temperature and time. A significant contribution to understanding the mechanism of inactivation was provided by native gel electrophoresis data of inactivated samples. Simultaneous fit of inactivation data verified a two-step mechanism composed of reversible unfolding/folding reaction followed by a relatively fast aggregation of the denatured urease form. A complex reaction scheme containing numerous oligomeric forms was thus described by a relatively simple model which suitably represents the main types of reactions involved in the urease activity loss.