Experimental and molecular modeling evaluation of the physicochemical properties of proline-based deep eutectic solvents
Citation for published version (APA):
van den Bruinhorst, A., Spyriouni, T., Hill, J. R., & Kroon, M. C. (2018). Experimental and molecular modeling evaluation of the physicochemical properties of proline-based deep eutectic solvents. Journal of Physical Chemistry B, 122(1), 369-379. https://doi.org/10.1021/acs.jpcb.7b09540
DOI:
10.1021/acs.jpcb.7b09540
Document status and date:
Published: 11/01/2018
Document Version:
Accepted manuscript including changes made at the peer-review stage
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne Take down policy
If you believe that this document breaches copyright please contact us at:
Experimental and Molecular Modeling Evaluation of the Physicochemical Properties of Proline-based
Deep Eutectic Solvents
Adriaan van den Bruinhorst,a Theodora Spyriouni,b Jörg-Rüdiger Hill,b and Maaike C. Kroona,c,*
a. Eindhoven University of Technology, Dept. Chemical Engineering and Chemistry Separation Technology Group, Het Kranenveld, Bldg. 14 (Helix), 5612 AZ Eindhoven, The Netherlands;
b. Scienomics GmbH, Bürgermeister-Wegele-Str. 12, D-86167 Augsburg, Germany;
c. Khalifa University of Science and Technology, Petroleum Institute, Dept. of Chemical Engineering, P.O. Box 2533, Abu Dhabi, United Arab Emirates.
ABSTRACT
The liquid range and applicability of deep eutectic solvents (DESs) are determined by their physicochemical properties. In this work, the physicochemical properties of glycolic acid:proline and malic acid:proline were evaluated experimentally and with MD simulations at five different ratios. Both DESs exhibited esterification upon preparation, which affected the viscosity in particular. In order to minimize oligomer formation and water release, three different experimental preparation methods were explored, but none could prevent esterification. The experimental and calculated densities of the DESs were found to be in good agreement. The measured and modeled glass transition temperature showed similar trends with composition, as did the experimental viscosity and the calculated diffusivities. The MD simulations provided additional insight at the atomistic level, showing that at acid-rich compositions, the acid-acid hydrogen bonding (HB) interactions prevail. Malic acid-based DESs show stronger acid-acid HB interactions than glycolic acid-based ones, possibly explaining its extreme viscosity. Upon the addition of proline, the inter- species interactions become predominant, confirming the formation of the widely assumed HB network between the DESs constituents in the liquid phase.
INTRODUCTION
Organic eutectic mixtures have been extensively studied over the past decades. They have been used to e.g. improve the uptake of pharmaceuticals in the body,1 store latent heat,2–4 or to perform solvent-free organic syntheses.5 In 2003, the term Deep Eutectic Solvents (DESs) was introduced6 and eutectic mixtures were subsequently presented as alternatives to ionic liquids (ILs).7 DESs are therefore only recently acknowledged as solvent class. A DES is generally composed of two components that show a large melting point depression upon mixing, resulting in a liquid at temperatures far below the melting temperatures of both individual pure components.8 These liquids exhibit characteristics similar to those of ILs, but can be prepared by the direct combination of its (often cheap, non-toxic, and biodegradable) counterparts without the need of further purification.8,9
During the last decade, research to DESs emerged and the solvents have been applied to a wide variety of fields. Although a large amount of distinct DESs has been explored, the majority of the applied DESs are based on a combination of choline chloride (ChCl) and urea, glycerol or ethylene glycol.9,10 These commercially available DESs are probably widely applied because they are cheap, easy to prepare and well characterised.6,11–13 Hence, established physicochemical properties are essential for the applicability of new solvents. New combinations of DESs are screened empirically and the experimental characterization of multiple DESs is laborious and time- consuming. Moreover, DESs typically show high viscosities (>0.1 Pa∙s) at ambient temperatures, complicating their characterization and industrial applicability. It would be more efficient to predict the physicochemical properties of DESs based on their counterparts, prior to experimental analysis. In literature, various methods have been described for predicting DES properties. A
mixing rules to predict the critical properties and normal boiling temperatures of several DESs.15–
17 These properties have not been evaluated experimentally, because most DESs start to decompose before critical or boiling conditions are met.15–17 Based on the calculated properties, the density of the studied DESs have been predicted using a modified Rackett equation.15–17 Similarly, the refractive index,18 speed of sound19 and surface tension20 have been estimated.
Density functional theory (DFT) has been used to roughly estimate the melting temperature of DESs.21 A relationship has been found between the DES’s melting point and the total charge densities of its so-called cage critical points (CCP). These CCPs have been determined through the optimization of minimalclusters of the DES components. It should be noted that different types of DESs yield slightly different relationships.
Molecular dynamics (MD) simulations are an alternative route to predict the properties of DESs.
MD can provide valuable insight into the atomistic structure and interactions, guiding the selection of appropriate DES constituents. Although ILs have been extensively studied with MD,22 publications on DES have only emerged recently.23–28 The main properties that have been studied are volumetric properties, radial distribution functions and other structural distributions, hydrogen bonding (HB) analysis, and diffusivities. These studies focus on ChCl-based DESs, but many DESs described in literature do not contain halide or ammonium ions.29 Hence, it would be interesting to probe the properties and interactions of other types of DESs, such as amino-acid- based DESs. These showed high solubility of natural colorants and polymers and enhanced their stability as compared to common solvents.30,31 Also, they might play a role in the defence mechanism of plants against drought-stress.32,33 Recently, L-proline-based DESs were applied to the oxidative desulfurization of diesel, showing these mixtures can be of industrial and biological
In this work, the physicochemical properties of L-proline-based DESs were evaluated, both experimentally and computationally. Malic acid:proline (MalA:Pro) and glycolic acid:proline (GlyA:Pro) mixtures were studied at several ratios to explore the tunability of their properties by varying the composition. The chemical structure and abbreviation of the components are shown in Figure 1. The used DESs are referred to by their abbreviated counterparts followed by the molar ratio e.g. GlyA:Pro 3:1.
The influence of the preparation method and water content on the chemical stability and physicochemical properties of the studied DESs were evaluated. The DESs were prepared at various water contents with the heating,7 freeze-drying,35 and grinding methods36. Additionally, MD simulations were conducted to calculate volumetric (density), structural (radial distribution function, HB), thermal (glass transition temperature, Tg), and transport (diffusion) properties. The experimental data obtained using the most common preparation method, the heating method, were compared to the properties calculated from MD simulations.
EXPERIMENTAL SECTION Deep eutectic solvent preparation
Glycolic acid (purity, 99%), DL-Malic acid (purity, ≥99%) and L-Proline (purity, ≥98.5%) were purchased from Sigma-Aldrich. All chemicals were dried under vacuum in a desiccator with dry silica gel for at least 12 h prior to DES preparation. Three different methods were used for DES preparation; typically batches of 15 g DES were prepared.
The heating method comprised the addition of the DES’s components at the desired molar ratio to a capped round bottom flask using an analytical balance, and the subsequent heating while stirring at 50 rpm using a magnetic stirring and heating plate (accuracy ±0.5 K, precision ±0.1 K).
Malic acid-based DESs were prepared at 373 K, glycolic acid-based DESs at 323 K. The solid mixtures were heated until a clear liquid was obtained. Afterwards the DESs were submitted to a vacuum line for 15 to 60 min, in order to remove excess water or any gas introduced to the liquid during stirring.
The freeze drying method was adapted from Gutiérrez et al.35 125 g of 5 wt% aqueous solution of the DES’s components was divided over five round bottom flasks. The bottom of each flask was submersed in liquid nitrogen to freeze its content. Then each flask was packed with aluminium foil and immediately connected to a freeze-drier setup (Salmenkipp Christ Alpha 1-4 LD+ with Edwards RV8 vacuum pump). When the samples were defrosted (typically overnight), one of the round bottom flasks was removed for analysis. The remaining flasks were subjected to another freeze-drying step, following the above described procedure.
The grinding method was applied following Florindo et al.36 The DESs and samples for DES characterization were prepared in a glove-box with water and oxygen contents below 2 ppm and
with a pestle until a liquid or a white/colorless paste was obtained (~30 min). To remove the gas introduced in the DESs during grinding, the samples were centrifuged using a Sigma 2-16KHL at 293 K (accuracy ±3 K, precision ±1 K) and 3500 rpm for 30 min.
Deep eutectic solvent characterization
All samples were prepared at least in duplicate, any variation in sample preparation is expressed in the standard deviation of analysis, since systematic errors were minimized.
Water content. The water content of the DESs was determined by means of Karl-Fischer (KF) titration (Metrohm 899 Coulometer, 10 μg H2O precision and detection limit). Before titration, the DESs were diluted with ethanol in a 1:4 mass ratio to lower their viscosity and thus to decrease the dissolution time in the titration medium. Each titration was performed in triplicate.
NMR spectroscopy. 1H-Nuclear Magnetic Resonance (NMR) and 13C-NMR spectra were recorded on a Bruker BZH 400/52 spectrometer, with 16 and 512 scans, respectively, and a relaxation time of 5 s and 1 s, respectively. The samples were prepared using deuterium oxide (D2O) as solvent. The spectra were phased, baseline corrected and manually integrated with MestreNova (version 9.1.0-14011). All integrals were normalized to the integral of the CH2 group next to the NH group in proline.
Thermal analysis. Thermal behavior was reviewed using a TA instruments Q100 Differential Scanning Calorimeter (DSC) with autosampler. Aluminum Tzero hermetic pans were filled with 4-10 mg of DES, sealed, and scanned in 3 cycles from 303 to 193 K and back at 5 K∙min-1. The Tg
was taken as the midpoint of the transition upon heating. The equipment was calibrated and verified with a high purity (>99.99 %) indium standard (accuracy ±0.5 K, precision ±0.1 K).
Viscosity and density measurements. The density was determined using an AccuPyc II 1340 gas (helium) pycnometer with a 35 cm3 chamber and a glass sample cup. The temperature was not
actively controlled and varied between 293 and 296 K. Each sample was purged 10 times followed by 5 equilibrium cycles. A Brookfield CAP-2000+ equipped with spindle CAP05 was used to analyze the viscosity at 323, 348 or 353 K at several rotational speeds. The Anton Paar SVM Stabinger 3000/G2 was used for determining the viscosity and density of samples with viscosities lower than approximately 2 Pa∙s at 273 K. All viscosity and density measurements were performed in duplicate. The water content of the samples was determined before each measurement, following the previously described method. The calibration of the instruments was verified with a high viscosity reference liquid (S2000) from Anton Paar.
COMPUTATIONAL SECTION
Systems. The mixtures MalA:Pro and GlyA:Pro were studied with MD simulations at molar ratios 3:1, 2:1, 1:1, 1:2 and 1:3 to cover the composition range from acid-rich to proline-rich mixtures. For MalA:Pro these are the stoichiometric ratios previously reported to form a liquid that is stable at room temperature.31 In order to compare the two systems, the same ratios were applied for GlyA:Pro. It was assumed that protonation takes place in these acid-base systems: a proton is transferred from the acid groups to proline. Hence, the deprotonated acid turns into an anion (COO-) and the protonated proline into a cation (NH2+). In the present work, the minority component is modeled as fully (de)protonated. Other options, such as to model partial deprotonation or the zwitterionic form of proline, known to be predominant in solutions, were not considered in this work. Partial deprotonation has been reported for other DESs.37 Additionally, models using the zwitterionic form of proline can be found in the literature.38,39 However, without supporting spectroscopic data, any choice for partial (de)protonation or zwitterion formation would be arbitrary.
Malic acid, which has two acid groups, was modeled with either one (MalA:Pro 3:1, 2:1, and 1:1) or both (MalA:Pro 1:2 and 1:3) acid groups deprotonated. The formed anions and cations have a net charge of -1e (or -2e for malic acid) and +1e, respectively. At ratios rich in acid or proline, the excess component was modeled in its neutral form to maintain the mixture neutral. For example, for GlyA:Pro and MalA:Pro 3:1, the composition was 1 anion : 1 cation : 2 neutral acid molecules, while for MalA:Pro 1:3, the mixture was composed of 1 anion (with charge -2e), 2 cations and 1 neutral proline.
Force field. Judicious force field selection is of primary importance for the reliability of the simulations. Typically, several force fields or force field parameters are screened against primarily
experimental densities but also transport properties. The estimation of charges plays a central role for systems like ILs and DES. Charges for ILs have been calculated with DFT.40 However, the selection of the functional is not straightforward. Another common approach is to use charges that fit the electrostatic potential of a compound.41
Figure 1. Structures of (a) malic acid, (b) glycolic acid, (c) proline, (d) malic anion with both acid groups deprotonated, (e) glycolic anion, (f) proline cation. The labels shown on the atoms refer to the tabulated partial charges on the atoms. They are presented together with the non-bonded force field parameters in SI1.
In this work, the Amber-Cornell force field42 was used and the partial atomic charges were calculated according to the Generalized Amber Force-Field (GAFF)43 procedure in order to be consistent with the force field. The following calculations were executed within the MAPS platform of Scienomics (Materials and Processes Simulations Platform, Version 4.0, Scienomics SARL, Paris, France). The geometry of the isolated molecules was optimized using MAPS. Α post-Hartree–Fock method, the Møller–Plesset perturbation theory, was used with the 6-31G*
basis set. Subsequently, a single-point energy calculation was performed using the Hartree-Fock method and the 6-31G* basis set for calculating the charges with the Mertz-Singh-Kollman scheme.44,45 The partial charges are listed in tables S1-S6 (SI1).
The initial configurations were built by using the Theodorou and Suter approach46 implemented in MAPS. This is a modification of the rotational isomeric state theory that uses conditional probabilities for the non-bonded interactions between the atom to be placed and the rest of the system, and subsequently minimizes the energy progressively.
Four to six different initial configurations were created for each mixture at each composition.
The amorphous systems were prepared at a low initial density of 1 g/cm3 at 450 K. Then, they were minimized by initially applying steepest descent, followed by conjugate gradient minimization. The minimized configurations were subjected to short NVT MD runs for 200 ps at 293 K, followed by NPT simulations at 293 K and ambient pressure for 1 ns. Since these systems carry a net charge on some atoms, they have low mobility and cannot relax holes, created initially, during simulation runs of some ns. Thus, they tend to converge to lower density. For this reason, a compression/decompression cycle was implemented to reach equilibrium density. In the literature, several variations of this procedure are described.24 In the present work, the configurations were compressed to high pressure at 1000 bar for 1 ns to eliminate the holes. Then, they were subjected to short NVT runs for 200 ps, and, finally, to productive NPT runs at ambient pressure and 293 K for 4 ns. The densities resulting from the final NPT simulations were typically higher than those from the initial NPT runs (before the compression/ decompression cycle), indicating that holes were destroyed following this procedure. Visual inspection confirmed that the molecules occupied the volume uniformly.
Other settings. The Coulomb interactions were treated with the particle-mesh Ewald method.47 A cut-off distance of 12 Å was used for the van der Waals interactions and tail corrections were taken beyond that distance. The MD calculations were carried out using LAMMPS28 through the MAPS interface.
Glass transition temperature. Tg was derived from the average volume as a function of temperature; generally, the volume decreases linearly with temperature and changes slope at Tg.48 Therefore, the average volume was calculated at a series of temperatures ranging from 293 to 203 K with steps of 15 K. At each temperature, the equilibrium configuration was subjected to an NVT simulation for 200 ps. The resulting configuration was run in the NPT ensemble for 4 ns for the system to reach the equilibrium average density. At the end of each NPT simulation, the average volume was calculated from the equilibrated part of the trajectory, usually the last 2 ns. The average volume was plotted versus temperature and Tg was estimated as the intersection point of two linear fits above and below Tg.
Self-diffusion coefficient. The calculation of the self-diffusion coefficient of the ions and neutral species in the mixtures was performed using the Einstein relationship:
𝐷𝐷 = 1 6 lim𝑡𝑡→∞
𝑑𝑑
𝑑𝑑𝑑𝑑〈|𝑟𝑟𝑖𝑖(𝑑𝑑) − 𝑟𝑟𝑖𝑖(0)|2〉
where |𝑟𝑟𝑖𝑖(𝑑𝑑) − 𝑟𝑟𝑖𝑖(0)| is the displacement of the position of species’ centre-of-mass over time 𝑑𝑑 and the brackets denote an average over all species along the trajectory. The diffusion coefficients were calculated in the Fickian regime as the slope of the linear part of the average mean-square displacement plot over time. Two different initial configurations equilibrated at 323 K for GlyA:Pro and 353 K for MalA:Pro were run in the NVT ensemble for 8 ns. The temperatures were chosen so that they match the temperatures of the viscosity results to permit comparison and cross- interpretation of the observations.
Hydrogen bonding. Hydrogen bonds (HBs) were defined all pairs of a hydrogen atom bonded to N or O atom with another N or O atom found in a distance smaller than 2 Å. Angle criteria between the three atoms were not considered.
RESULTS AND DISCUSSION
Influence of preparation method on properties and DES stability
Three preparation methods were explored in this work, because Florindo et al. previously showed that the method used to prepare a DES can be vital for its chemical stability and might affect its physicochemical properties.36 The DESs were first prepared with the most common method, the heating method. Directly after preparation with the heating method, the malic-acid- based DESs showed water contents between 2 and 5 wt%. Water influences the properties of DESs considerably, e.g. the Tg of MalA:Pro 1:1 decreases by 55 K if the water content is increased from 2.1 wt% (this study) to 17.8 wt%.29,49 Thus, water should either be incorporated in the MD simulations, or eliminated from the DES samples to make a legitimate comparison between predicted and experimentally determined properties.
Aiming for water elimination, all studied DESs were re-prepared using the heating method in a water-free environment with dried constituents. Most DESs exhibited water contents above 1 wt%.
Hence, water had to be released during DES preparation. It is likely that ester formation occurred through the condensation of the hydroxyl and carbonyl groups of the DESs components. This esterification reaction was confirmed by 1H-NMR spectroscopy; typical spectra and the peak assignments can be found in Figures S1-S5 (SI2). Glycolic acid was esterified with both glycolic acid and proline after preparation of GlyA:Pro 3:1. The esterification was also confirmed with Liquid Chromatography Mass Spectrometry (LC-MS) (SI3), as oligomers with several chain lengths could be distinguished (Figure S7). The reaction could be suppressed by adding water to the mixture before DES preparation, the excess of water forces the equilibrium to its reactants (Figure S8). However, too high water contents lead to the disintegration of a DES into its
water-solvated counterparts.23,29 Taking this into account, other preparation methods were explored to prevent esterification and water formation.
The freeze-drying method combines the addition of water with a lower preparation temperature, potentially further suppressing the esterification reaction. A DES is prepared through the sublimation of water from a frozen aqueous solution of its components.26 The water content was drastically decreased upon freeze-drying (see Figure S9). However, in order to obtain a DES with a low water content (< 0.5 wt%) many more steps would be required. Moreover, esterification was observed after the first freeze-drying step, although to a lesser extent than when using the heating method (Figure S10). Since reaction occurs despite the low temperatures and because of the low drying rates, the freeze-drying method was not further studied.
The grinding method27 is the third preparation method that was tested to reduce the initial water content and to prevent esterification. The two dried components of a DES were ground together at ambient temperature in a water-free environment, resulting in a white liquid paste for GlyA:Pro 3:1 and 2:1. Both pastes were added to a centrifuge tube and their water contents were <0.1 wt%.
After leaving them for 72 h, their appearance was less pasty and more liquid. Both liquids were centrifuged in order to remove the (dry nitrogen) bubbles introduced into the sample during grinding. This resulted in a clear liquid for GlyA:Pro 2:1 and a clear liquid with small crystals at the bottom for GlyA:Pro 3:1 (see Figure S12). The water contents of both clear liquid phases were significantly higher than 0.1 wt%, i.e. 1.2 and 2.5 wt% for GlyA:Pro 2:1 and 3:1, respectively.
Figure 2. Water content of GlyA:Pro 2:1 in a water-free environment over time, determined with two methods: Karl-Fischer (KF) titration (triangles), calculation from 1H-NMR integrals of esterified groups, corrected for the initial water content (solid circles). The water contents calculated from the 1H-NMR integrals of the GlyA dimer (GG, half open circles) and the GlyA–
Pro (GP, open circles) ester are also presented.
GlyA:Pro 2:1 was selected to study the production of water over time, because it yielded a clear liquid after centrifugation. Samples were taken periodically from a batch of DES to monitor the water content and follow the reaction with 1H-NMR spectroscopy, using the equation described in SI 2.2. Figure 2 shows that the calculated water content virtually follows the same trend as the measured water content, confirming that esterification takes place at ambient conditions. Although some outliers are present, a clear linear increase of the water content in time was observed. The esterification reaction of glycolic acid with proline and the self-esterification reaction of glycolic acid contribute linearly to the total increase in water content. A water content of approximately 1 wt% (~85 h) corresponds to a conversion of approximately 11 mol% monomeric glycolic acid and 5.5 mol% proline. Since esterification is an equilibrium reaction, it is expected that the water production would gradually suppress the reaction rate, and that a limit would be reached over time.
The impact of oligomer formation on the density, viscosity and Tg was evaluated at various water contents and molar ratios of GlyA:Pro DESs. A sample was prepared with the heating and the grinding method at the same composition. Water was added directly after grinding in order to approximate the water released by esterification when using the heating method while avoiding oligomer formation. The densities of the heating and grinding method samples were in the same order of magnitude, and no clear trend with water content could be identified (Figure S13). The high viscosity at ambient temperature (~298 K) led to difficulties in sample handling, which could not be overcome by heating the sample, since that would induce esterification.
Figure 3. Influence of water content and preparation method/oligomerization on viscosity at 323 K, water content is depicted above each column in percentage (wt%). The DESs were prepared in duplicate at three different molar ratios using the heating and grinding method. The final mole fractions of the DESs prepared with the grinding method are given in Table S10 (SI6), as they differ slightly from the aimed molar ratios.
Contrary to the density, the viscosity was clearly affected by the oligomer formation. In Figure 3 can be seen that the viscosity of the heating method samples increases with increasing water
viscosity. The opposite trend was exhibited by samples that were prepared with the grinding method: the viscosity decreased with water content at all compositions (see also Figure S14).
Hence, the viscosity is increased stronger by oligomer formation than it is lowered by the water released during esterification. No obvious effect of the esterification reaction on the Tg could be identified (Figure S15). The Tg of the grinding method samples was generally slightly lower than those of the heating method samples. A higher water content led to a significant decrease in Tg for all samples, also for those with oligomers, although the decrease in Tg was more pronounced for the samples where water was added.
Apart from hampering esterification and lowering the DES’s viscosity and Tg, the addition of water seems to strongly promote liquid formation. A clear liquid phase was obtained when small amounts (~0.5-2 wt%) of water were added to the GlyA:Pro DES prepared with the grinding method (Figure S16). This is in line with results published for the urea : ChCl 2:1 DES, which shows a melting temperature depression upon water addition.50 For the DESs studied here, no melting transition could be observed with DSC. Both a high viscosity and a high water content can inhibit direct crystal formation, making it challenging to detect first order phase transitions of DESs on the time-scale of a DSC measurement. It can take hours, or days, before crystallization can be observed.29 Hence, only glass transitions could be recognized.
For all measurements, bubble-free samples are required. With the grinding method this was impossible to achieve without lowering the viscosity of the samples by adding water (see Figure S16). This allowed for easier removal of the gas introduced in the DESs during grinding and settling of any particles. Without added water, the liquid phase would either contain bubbles, or be indistinguishable from the opaque bottom phase. The bottom phase resembles a liquid saturated with crystals. 1H-NMR analysis of both phases after water addition shows that the glycolic acid
content in the bottom phase is significantly higher than in the top phase (Table S11). This is in line with the centrifuge method described in a recent publication.51 Hence, the clear liquid top phase would not have the aimed composition after centrifugation, complicating comparison with the MD simulation data. The heating method induces oligomer formation, but the oligomers only affect the viscosity strongly. The density and Tg were much less affected, and those are the properties that were directly calculated from the MD simulations. Therefore, the MD simulation results were compared to experimental data of DESs prepared with the heating method, omitting that water and oligomers are present after preparation.
Physical properties: experiment and simulation
Figure 4. Density of the MalA:Pro (circles) and GlyA:Pro (triangles) mixtures versus the mole fraction of proline (ratios are shown at top). The closed symbols are experimental data from this work. The open symbols are results from MD simulations. The error bars of the experimental data represent the standard deviation for multiple samples.
Density. Figure 4 shows the density of the studied DESs. The density decreases with increasing molar content of proline. Good agreement was obtained between the experimental and calculated
preparation and, therefore, could not be characterized experimentally. Hence, the liquid range of GlyA:Pro DESs was narrower than that of MalA:Pro DESs.
For MalA:Pro, the calculated densities are 0.77 to 2.2% lower than the experimentally obtained values, with the exception of MalA:Pro 1:1 which lies below the linear trend observed for the other ratios. For GlyA:Pro, the deviations are smaller (0.25 to 1.14 %) and the calculated densities are slightly higher than the experimental densities. The good agreement of the experimental and simulation data indicates that the proposed simulation model can reliably reproduce the cohesion of the mixtures giving rise to high densities. Preliminary calculations of neutral MalA:Pro and GlyA:Pro systems resulted into much lower densities, as much as -15% or more. This illustrates the significance of the charges for the simulation model.
Figure 5. Viscosity (η) versus the mole fraction of proline (ratios are shown at top)for MalA:Pro (circles) and GlyA:Pro (triangles) DES at 348 and 323 K, respectively. MalA:Pro 1:3 was determined at 353 K due to measuring scope limitations.
Transport properties. The experimental viscosities of MalA:Pro at 348 K and GlyA:Pro at 323 K are presented in Figure 5. Τhe viscosity increases with the proline content. Especially the GlyA
should be noted however, that for similar water contents, the viscosity of GlyA:Pro 1:1 obtained by the grinding method was a factor 2 lower than the viscosity obtained with the heating method (Figure 3). The viscosities of the MalA DESs were all extremely high. At ambient temperature, these DESs were glassy and immobile; at 348 K (or 353 for MalA:Pro 1:3), they would flow but exhibit viscosities > 10 Pa∙s.
Figure 6. Simulation results for the self-diffusion coefficients (D) of ions and neutral species contained in MalA:Pro at 353 K (top) and GlyA:Pro at 323 K (bottom) versus the proline mole fraction (ratios are shown at top).
Self-diffusivities were calculated at the same temperatures as the viscosity measurements to permit comparison with the viscosity data, namely at 353 K for MalA:Pro and 323 K for GlyA:Pro.
Also, because higher temperatures enhance mobility and give more reliable estimates. The results are shown in Figure 6, and listed in Table S12 and S13 (SI8). The neutral acids have higher
mixtures, the proline cation has slightly higher mobility than the malic acid for the 3:1 ratio and approximately the same mobility for the other ratios. Proline has a cyclic structure that makes it difficult to diffuse, and malic acid has two acid groups that develop polar interactions especially through the modeled anions and this hampers diffusion.
Figure 7.Glass transition temperature Tg of MalA:Pro and GlyA:Pro mixtures versus the proline mole fraction (ratios are shown at top). Experimental data (filled symbols) and results from MD simulations (open symbols) are presented.
The diffusion coefficients of MalA:Pro mostly corroborate the viscosity results. At ratios 3:1 and 2:1, only small differences can be observed for the measured viscosities, as well as for the diffusivities of the proline cation and malic acid anion. However, at both ratios also neutral malic acid is present in the modeled system, showing a clear decrease in mobility with increasing proline content. This is not expressed in an increase of the viscosity or experimental Tg (Figure 7). As proline is added, the diffusion coefficients drop and the viscosity increases for ratio 1:1 and further for ratio 1:2. The slower diffusion and increased viscosity can be attributed to the increasing strength of the modeled cation/anion interactions as proline is added. This is especially evident for
For the 1:3 ratio, neutral proline is present in the mixture and this results in a slight increase of diffusion and a decrease of viscosity and Tg (Figure 7).
For the GlyA:Pro mixtures, it is observed that the glycolic anions diffuse faster than the proline cations due to their smaller size. The difference decreases as the amount of proline in the mixture increases and more species are present as ions than neutral species. The difference is the smallest for the 1:1 ratio, where the system contains only anions and cations. The results are consistent with the viscosity observations, especially considering the impact the esterification reaction has on the viscosity results. With the addition of proline the diffusion coefficients decrease, and the restrained mobility is expressed in an increase in viscosity and Tg.
The self-diffusion coefficients of the GlyA:Pro and MalA:Pro mixtures were generally lower than those reported in literature for other DESs.52–54 This is best explained by the high viscosities of the DESs studied here, which were typically more than an order of magnitude higher. The malonic acid-ChCl 1:1 DES has lower viscosities than the studied GlyA:Pro and MalA:Pro, but contains similar functionalities. Its self-diffusion coefficients are in the same order of magnitude as those of the mixtures studied here.52 Only one mixture has been reported with similar viscosities (η > 10 Pa∙s), namely a concentrated mixture of glycerol and sodium acetate.54 The reported self- diffusivities for this system (~2-4∙10-8 cm2∙s-1 at 322 K) are in the same range as those obtained from the MD simulation results for the GlyA:Pro and MalA:Pro mixtures, confirming the reliability of the force-field parameters.
All prepared DESs showed a glass transition; as previously discussed, melting or crystallization could not be observed with DSC. The experimental and simulated Tg are shown in Figure 7. The calculated values are averages over approximately four different initial configurations. The large
error bars of the simulation results reflect the big fluctuations of volume in these systems due to polar and charged species.
For MalA:Pro, no clear overall trend of Tg with composition could be observed for either the experimental or simulation results. The experimental Tg’s reflect the relative differences in viscosity and self-diffusivities. The MalA:Pro DESs show significantly higher Tg than the GlyA:Pro mixtures, which is in accordance with the viscosity and diffusivity results. The simulation results for the Tg are lower than the measured values, and the absolute deviation varies from 6 to 15K. The underestimation of Tg for MalA:Pro might be partly attributed to the lower density of the simulation or the presence of oligomers in the DSC samples.
The Tg of GlyA:Pro increases with proline content, following the same trend as for the viscosity.
This could be ascribed to the increasing intermolecular interactions between the acid and proline that results in restricted motion, and thus a higher Tg and viscosity. The calculated Tg’s show the same trend, but lie above the experimental values. The absolute deviation is 15 K for GlyA:Pro 3:1 and decreases to 6 K for GlyA:Pro 1:1. The density deviations between experiment and simulation are small for these systems and could not explain the observed differences for Tg. Also, the water content and esterification did not have a very significant effect on Tg. Instead, these differences could originate from the large fluctuations of the calculated volume, resulting into large error bars.
Structure and hydrogen bonding
Figure 8. Hydrogen bonds (HB) between the acid (neutral or anion) and proline (neutral or cation) (circles), the acid (neutral or anion) with itself (squares), proline (neutral or cation) with itself (triangles), and all molecules (diamonds), normalized by the amount of molecules in the MalA:Pro (A) and GlyA:Pro (B) mixtures versus the molar fraction of proline (ratios are shown at top). The results are averages over 4 trajectories of different initial configurations.
To further investigate the intermolecular interactions at the atomistic level, the intermolecular HB was calculated at 293 K at all ratios for both mixtures. HBs formed between the acid (neutral or anion) and proline (neutral or cation) are hereafter called inter-species HB. The HBs between all species are shown in Figure 8. The total amount of HBs shows a continuous decrease up to 1:2 ratio for MalA:Pro where it reaches a plateau. For GlyA:Pro, a weak minimum is observed at 1:1 ratio. Both trends are the combination of increasing inter-species interactions and decreasing inter-
the inter-acid HBs start from significantly higher values for MalA:Pro, since it has two acid groups.
It can be seen that for all ratios, the inter-acid HB of malic acid is 2 to 3 times larger than that of glycolic acid, and that the inter-species HB per molecule increases as proline is added. For GlyA:Pro it reaches a plateau at 1:2, while for MalA:Pro it continues to increase since malic acid possesses an extra site to form HBs as compared to glycolic acid. As proline is added, the amount of inter-acid HB drops significantly for both acids, because the intermolecular HB with proline becomes prominent. The proline-proline HB interactions are negligible, except for GlyA:Pro 1:2 and 1:3. These are also the mixtures for which no clear liquid could be obtained. If the solubility of solid proline in the liquid at the eutectic composition is low, stronger proline-proline interactions might be an explanation for the observed precipitation.
Figure 9. Hydrogen bonds (HB) between the acid (neutral or anion) and proline (neutral or cation) molecules (inter-species) normalized by the number of proline (neutral or cation) molecules in the MalA:Pro (circles) and GlyA:Pro (triangles) mixtures versus the molar fraction of proline (ratios are shown at top). The results are averages over 4 trajectories of different initial configurations.
The intermolecular HB per proline (neutral or cation) molecule is around 1 at all ratios (see Figure 9). This means that each proline engages with approximately one acid molecule at each ratio, even at high proline ratios like 1:3. As proline is added, the inter-acid HB decreases and the
Proline is thus the molecule that activates the inter-species HB. Since the inter-species HB is supposed to liquefy the solids upon mixing,6–8 it is important to know the HB at different ratios for each mixture. From Figure 8 can be deduced that having malic acid, instead of the weaker glycolic acid, has no significant impact on the inter-species HB. Moreover, malic acid engages in more acid-acid HBs than glycolic acid. These extra short-distance interactions might result in a tighter cohesion for MalA:Pro than for GlyA:Pro mixtures and explain the higher densities observed. The higher acid-acid interactions probably also contribute to the extreme viscosities of the MalA:Pro systems. As proline is added, the inter-species interactions become predominant compared to the inter-acid interactions. This was also observed in an experimental study towards the nanostructure of a DES based on malic acid and ChCl. The acid-acid dimers typically observed for liquefied dicarboxylic acids contributed to the liquid structure to a lesser extent than the inter- species interactions.55 The increased network of inter-species HB could be at the origin of increased viscosity and hampered diffusion at higher proline contents.
Figure 10. Pair distribution functions g(r) at 293 K between the centers of mass of acid (neutral or anion) - proline (neutral or cation) (MalA-Pro/GlyA-Pro, solid lines), acid (neutral or anion) with itself (MalA-MalA/GlyA-GlyA, dashed lines) and between proline (neutral or cation) with itself (Pro-Pro, dash-dot lines) for the MalA:Pro (top) and GlyA:Pro (bottom) mixtures at ratios 3:1, 1:1, and 1:3 (from left to right). The results are averages over 4 trajectories from different initial configurations.
Similar observations to the above can be made from the radial distribution functions g(r) of the molecules centers of mass for both mixtures at ratios 3:1, 1:1 and 1:3, at 293 K, shown in Figure 10. The g(r)’s of the neutral molecules and ions have been averaged together for the same species (acid or proline). Long range order is observed. The highest peak represents the interactions between the acid (neutral or anion) and proline (neutral or cation) for both mixtures in all ratios, stressing out that the most important interactions are the intermolecular ones between different species. The second highest peak is between the acid molecules for compositions rich in acid. This was also observed in literature for an equimolar mixture of malic acid and ChCl, where the acid- acid interactions show a lower peak at a larger distance than the inter-species interactions.55 When two distinct humps appear in the position of the first peak, the second hump is due to the neutral molecules. If the two mixtures for the 3:1 ratio are compared, it can be seen that the acid-acid peak for malic acid is stronger than for glycolic acid. This difference is even more pronounced for the 1:1 ratio. The first acid-acid peak appears at shorter distances for the GlyA:Pro mixtures than for the Mala:Pro mixtures, due to the smaller size of glycolic acid as compared to malic acid. At high proline ratio 1:3, the peak of the acids is low and appears at larger distances. The first peak between the acids and proline becomes broader. The first acid-acid peak generally appears at smaller
by the composition of the mixtures. For ratios richer in proline, the explanation should lie in the bulkier proline molecule.
CONCLUSIONS
Two DESs, GlyA:Pro and MalA:Pro, were evaluated experimentally and with MD simulations at five different ratios. The mixtures were prepared with three methods, namely the heating, freeze- drying, and grinding method. The grinding method initially resulted in negligible esterification, but even at room temperature the reaction progressed slowly. Since esterification is an equilibrium reaction, the same final degree of esterification is expected for all preparation methods. Hence, although esterification itself could not be prevented, the reaction rate was affected by the conditions (temperature, water content, time) under which the DES was prepared and stored. The oligomers resulting from the condensation reaction increased the DES’s viscosity stronger than it was lowered by the released water. Tg and density were less affected. Experimental and modeled densities showed satisfactory agreement. Calculated diffusivities were qualitatively compared to the experimental viscosity data and exhibited similar trends. For both mixtures, the viscosity increased with proline content. It must be pointed out that the viscosities were very high for both DESs; MalA:Pro was glassy and immobile at ambient temperatures. The chemical instability owing to esterification and the high viscosity of the studied mixtures strongly limit their practical applications. No crystallization or melting temperatures could be observed for any of the mixtures due to their high viscosity. Instead, all mixtures exhibited a glass transition. The measured and modeled Tg showed similar trends, with deviations up to 15 K. GlyA:Pro mixtures showed a clear increase of Tg with the addition of proline up to the 1:1 ratio, matching the strong increase in viscosity. At the atomistic level, the HB interactions were calculated from the MD simulations.
Proline was found to activate the inter-species HB by disrupting the acid-acid interactions. The inter-species HB was calculated to increase with the proline content. Glycolic acid was found to be equally efficient as malic acid to engage proline, having weaker acid-acid interactions compared
to malic acid, probably leading to its advantageous physicochemical properties. The above was also confirmed by the radial distribution functions.
The low chemical stability and high viscosity of the studied DESs complicated their property analysis. Nevertheless, the experimentally obtained physicochemical properties were successfully modeled using MD simulations. Moreover, the MD simulations gave further insight in the intermolecular interactions at the atomistic level.
ASSOCIATED CONTENT
Supplementary Information (SI) available “Supplementary Information for Experimental and molecular modeling evaluation of the physicochemical properties of proline-based deep eutectic solvents.pdf”. SI1, Partial atom charges and non-bonded force-field parameters; SI2, Verification of esterification with NMR; SI3, Verification of esterification with LC-MS; SI4, Influence of preparation method on esterification reaction; SI5, Grinding method and centrifugation; SI6, Influence of oligomer formation on physicochemical properties; SI7, Composition of DESs prepared with the grinding method; SI8, Experimental data of DESs prepared with the heating method.
AUTHOR INFORMATION
Corresponding Author
*Phone: +971-26075317. E-mail: mkroon@pi.ac.ae
ACKNOWLEDGEMENT
Financial support from Netherlands Organization for Scientific Research (NWO) and the company Paques B.V. is gratefully acknowledged. This work is part of the research program “MES meets DES”, with project number STW-Paques 12999.
REFERENCES
(1) Goldberg, A. H.; Gibaldi, M.; Kanig, J. L. Increasing Dissolution Rates and Gastrointestinal Absorption of Drugs Via Solid Solutions and Eutectic Mixtures I. J. Pharm. Sci. 1965, 54, 1145–1148.
(2) Lorsch, H. G.; Kauffman, K. W.; Denton, J. C. Thermal Energy Storage for Solar Heating and off-Peak Air Conditioning. Energy Convers. 1975, 15, 1–8.
(3) Rathod, M. K.; Banerjee, J. Thermal Stability of Phase Change Materials Used in Latent Heat Energy Storage Systems: A Review. Renew. Sustain. Energy Rev. 2013, 18, 246–258.
(4) Diarce, G.; Quant, L.; Campos-Celador, Á.; Sala, J. M.; García-Romero, A. Determination of the Phase Diagram and Main Thermophysical Properties of the Erythritol–Urea Eutectic Mixture for Its Use as a Phase Change Material. Sol. Energy Mater. Sol. Cells 2016, 157, 894–906.
(5) Rothenberg, G.; Downie, A. P.; Raston, C. L.; Scott, J. L. Understanding Solid/Solid Organic Reactions. J. Am. Chem. Soc. 2001, 123, 8701–8708.
(6) Abbott, A. P.; Capper, G.; Davies, D. L.; Rasheed, R. K.; Tambyrajah, V. Novel Solvent Properties of Choline Chloride/Urea Mixtures. Chem. Commun. 2003, No. 1, 70–71.
(7) Abbott, A. P.; Boothby, D.; Capper, G.; Davies, D. L.; Rasheed, R. K. Deep Eutectic Solvents Formed Between Choline Chloride and Carboxylic Acids: Versatile Alternatives to Ionic Liquids. J. Am. Chem. Soc. 2004, 126, 9142–9147.
(8) Zhang, Q.; De Oliveira Vigier, K.; Royer, S.; Jérôme, F. Deep Eutectic Solvents: Syntheses,
(9) Paiva, A.; Craveiro, R.; Aroso, I.; Martins, M.; Reis, R. L.; Duarte, A. R. C. Natural Deep Eutectic Solvents – Solvents for the 21st Century. ACS Sustain. Chem. Eng. 2014, 2, 1063–
1071.
(10) Tang, B.; Row, K. H. Recent Developments in Deep Eutectic Solvents in Chemical Sciences. Monatshefte für Chemie - Chem. Mon. 2013, 144, 1427–1454.
(11) Abbott, A. P.; Harris, R. C.; Ryder, K. S. Application of Hole Theory to Define Ionic Liquids by Their Transport Properties †. J. Phys. Chem. B 2007, 111, 4910–4913.
(12) Leron, R. B.; Wong, D. S. H.; Li, M.-H. Densities of a Deep Eutectic Solvent Based on Choline Chloride and Glycerol and Its Aqueous Mixtures at Elevated Pressures. Fluid Phase Equilib. 2012, 335, 32–38.
(13) Leron, R. B.; Li, M.-H. High-Pressure Density Measurements for Choline Chloride: Urea Deep Eutectic Solvent and Its Aqueous Mixtures at T=(298.15 to 323.15)K and up to 50MPa. J. Chem. Thermodyn. 2012, 54, 293–301.
(14) Valderrama, J. O.; Robles, P. A. Critical Properties, Normal Boiling Temperatures, and Acentric Factors of Fifty Ionic Liquids. Ind. Eng. Chem. Res. 2007, 46, 1338–1344.
(15) Mirza, N. R.; Nicholas, N. J.; Wu, Y.; Kentish, S.; Stevens, G. W. Estimation of Normal Boiling Temperatures, Critical Properties, and Acentric Factors of Deep Eutectic Solvents.
J. Chem. Eng. Data 2015, 60, 1844–1854.
(16) Mjalli, F. S.; Shahbaz, K.; Alnashef, I. M. Modified Rackett Equation for Modelling the Molar Volume of Deep Eutectic Solvents. Thermochim. Acta 2015, 614, 185–190.
Solvents Densities at Different Temperatures. Thermochim. Acta 2011, 515, 67–72.
(18) Shahbaz, K.; Bagh, F. S. G.; Mjalli, F. S.; AlNashef, I. M.; Hashim, M. A. Prediction of Refractive Index and Density of Deep Eutectic Solvents Using Atomic Contributions. Fluid Phase Equilib. 2013, 354, 304–311.
(19) Mjalli, F. S.; Abdel Jabbar, N. M. Acoustic Investigation of Choline Chloride Based Ionic Liquids Analogs. Fluid Phase Equilib. 2014, 381, 71–76.
(20) Shahbaz, K.; Mjalli, F. S.; Hashim, M. A.; AlNashef, I. M. Prediction of the Surface Tension of Deep Eutectic Solvents. Fluid Phase Equilib. 2012, 319, 48–54.
(21) García, G.; Atilhan, M.; Aparicio, S. An Approach for the Rationalization of Melting Temperature for Deep Eutectic Solvents from DFT. Chem. Phys. Lett. 2015, 634, 151–155.
(22) Maginn, E. J. Molecular Simulation of Ionic Liquids: Current Status and Future Opportunities. J. Phys. Condens. Matter 2009, 21, 373101.
(23) Perkins, S. L.; Painter, P.; Colina, C. M. Molecular Dynamic Simulations and Vibrational Analysis of an Ionic Liquid Analogue. J. Phys. Chem. B 2013, 117, 10250–10260.
(24) Perkins, S. L.; Painter, P.; Colina, C. M. Experimental and Computational Studies of Choline Chloride-Based Deep Eutectic Solvents. J. Chem. Eng. Data 2014, 59, 3652–3662.
(25) Sun, H.; Li, Y.; Wu, X.; Li, G. Theoretical Study on the Structures and Properties of Mixtures of Urea and Choline Chloride. J. Mol. Model. 2013, 19, 2433–2441.
(26) Shah, D.; Mjalli, F. S. Effect of Water on the Thermo-Physical Properties of Reline: An
16, 23900–23907.
(27) Ferreira, E. S. C.; Voroshylova, I. V.; Pereira, C. M.; D. S. Cordeiro, M. N. Improved Force Field Model for the Deep Eutectic Solvent Ethaline: Reliable Physicochemical Properties.
J. Phys. Chem. B 2016, 120, 10124–10137.
(28) Kaur, S.; Gupta, A.; Kashyap, H. K. Nanoscale Spatial Heterogeneity in Deep Eutectic Solvents. J. Phys. Chem. B 2016, 120, 6712–6720.
(29) Dai, Y.; van Spronsen, J.; Witkamp, G.-J.; Verpoorte, R.; Choi, Y. H. Natural Deep Eutectic Solvents as New Potential Media for Green Technology. Anal. Chim. Acta 2013, 766, 61–
68.
(30) Dai, Y.; Verpoorte, R.; Choi, Y. H. Natural Deep Eutectic Solvents Providing Enhanced Stability of Natural Colorants from Safflower (Carthamus Tinctorius). Food Chem. 2014, 159, 116–121.
(31) Francisco, M.; van den Bruinhorst, A.; Kroon, M. C. New Natural and Renewable Low Transition Temperature Mixtures (LTTMs): Screening as Solvents for Lignocellulosic Biomass Processing. Green Chem. 2012, 14, 2153–2157.
(32) Choi, Y. H.; van Spronsen, J.; Dai, Y.; Verberne, M.; Hollmann, F.; Arends, I. W. C. E.;
Witkamp, G.-J.; Verpoorte, R. Are Natural Deep Eutectic Solvents the Missing Link in Understanding Cellular Metabolism and Physiology? Plant Physiol. 2011, 156, 1701–1705.
(33) Swarcewicz, B.; Sawikowska, A.; Marczak, Ł.; Łuczak, M.; Ciesiołka, D.; Krystkowiak, K.; Kuczyńska, A.; Piślewska-Bednarek, M.; Krajewski, P.; Stobiecki, M. Effect of Drought
Untargeted GC–MS Profiling. Acta Physiol. Plant. 2017, 39, 158.
(34) Hao, L.; Wang, M.; Shan, W.; Deng, C.; Ren, W.; Shi, Z.; Lü, H. L -Proline-Based Deep Eutectic Solvents (DESs) for Deep Catalytic Oxidative Desulfurization (ODS) of Diesel. J.
Hazard. Mater. 2017, 339, 216–222.
(35) Gutiérrez, M. C.; Ferrer, M. L.; Mateo, C. R.; del Monte, F. Freeze-Drying of Aqueous Solutions of Deep Eutectic Solvents: A Suitable Approach to Deep Eutectic Suspensions of Self-Assembled Structures. Langmuir 2009, 25, 5509–5515.
(36) Florindo, C.; Oliveira, F. S.; Rebelo, L. P. N.; Fernandes, A. M.; Marrucho, I. M. Insights into the Synthesis and Properties of Deep Eutectic Solvents Based on Cholinium Chloride and Carboxylic Acids. ACS Sustain. Chem. Eng. 2014, 2, 2416–2425.
(37) Griffin, P. J.; Cosby, T.; Holt, A. P.; Benson, R. S.; Sangoro, J. R. Charge Transport and Structural Dynamics in Carboxylic-Acid-Based Deep Eutectic Mixtures. J. Phys. Chem. B 2014, 118, 9378–9385.
(38) Kapitán, J.; Baumruk, V.; Kopecký, V.; Pohl, R.; Bouř, P. Proline Zwitterion Dynamics in Solution, Glass, and Crystalline State. J. Am. Chem. Soc. 2006, 128, 13451–13462.
(39) Yang, G.; Zhou, L.; Chen, Y. Stabilization of Zwitterionic Versus Canonical Proline by Water Molecules. Springerplus 2016, 5, 19.
(40) Dommert, F.; Wendler, K.; Berger, R.; Delle Site, L.; Holm, C. Force Fields for Studying the Structure and Dynamics of Ionic Liquids: A Critical Review of Recent Developments.
ChemPhysChem 2012, 13, 1625–1637.
Electrostatic Potentials. The Need for High Sampling Density in Formamide Conformational Analysis. J. Comput. Chem. 1990, 11, 361–373.
(42) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.;
Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. A Second Generation Force Field for the Simulation of Proteins, Nucleic Acids, and Organic Molecules. J. Am. Chem.
Soc. 1995, 117, 5179–5197.
(43) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. Development and Testing of a General Amber Force Field. J. Comput. Chem. 2004, 25, 1157–1174.
(44) Singh, U. C.; Kollman, P. A. An Approach to Computing Electrostatic Charges for Molecules. J. Comput. Chem. 1984, 5, 129–145.
(45) Besler, B. H.; Merz, K. M.; Kollman, P. A. Atomic Charges Derived from Semiempirical Methods. J. Comput. Chem. 1990, 11, 431–439.
(46) Theodorou, D. N.; Suter, U. W. Detailed Molecular Structure of a Vinyl Polymer Glass.
Macromolecules 1985, 18, 1467–1478.
(47) Ewald, P. P. Die Berechnung Optischer Und Elektrostatischer Gitterpotentiale. Ann. Phys.
1921, 369, 253–287.
(48) Ediger, M. D.; Angell, C. A.; Nagel, S. R. Supercooled Liquids and Glasses. J. Phys. Chem.
1996, 100, 13200–13212.
(49) Dai, Y.; Witkamp, G. J.; Verpoorte, R.; Choi, Y. H. Tailoring Properties of Natural Deep Eutectic Solvents with Water to Facilitate Their Applications. Food Chem. 2015, 187, 14–
(50) Meng, X.; Ballerat-Busserolles, K.; Husson, P.; Andanson, J.-M. Impact of Water on the Melting Temperature of Urea + Choline Chloride Deep Eutectic Solvent. New J. Chem.
2016, 40, 4492–4499.
(51) van den Bruinhorst, A.; Kollau, L.; Kroon, M. C.; Esteves, C. The Centrifuge Method: A New Method to Construct Eutectic Phase Diagrams. to be Submitt. 2017.
(52) D’Agostino, C.; Harris, R. C.; Abbott, A. P.; Gladden, L. F.; Mantle, M. D. Molecular Motion and Ion Diffusion in Choline Chloride Based Deep Eutectic Solvents Studied by 1H Pulsed Field Gradient NMR Spectroscopy. Phys. Chem. Chem. Phys. 2011, 13, 21383–
21391.
(53) D’Agostino, C.; Gladden, L. F.; Mantle, M. D.; Abbott, A. P.; Ahmed, Essa, I.; Al- Murshedi, A. Y. M.; Harris, R. C. Molecular and Ionic Diffusion in Aqueous – Deep Eutectic Solvent Mixtures: Probing Inter-Molecular Interactions Using PFG NMR. Phys.
Chem. Chem. Phys. 2015, 17, 15297–15304.
(54) Abbott, A. P.; D’Agostino, C.; Davis, S. J.; Gladden, L. F.; Mantle, M. D. Do Group 1 Metal Salts Form Deep Eutectic Solvents? Phys. Chem. Chem. Phys. 2016, 18, 25528–25537.
(55) Hammond, O. S.; Bowron, D. T.; Jackson, A. J.; Arnold, T.; Sanchez-Fernandez, A.;
Tsapatsaris, N.; Garcia Sakai, V.; Edler, K. J. Resilience of Malic Acid Natural Deep Eutectic Solvent Nanostructure to Solidification and Hydration. J. Phys. Chem. B 2017, 121, 7473–7483.
TOC Graphic
