Experimental and Computational Studies of Choline Chloride-Based

Article pubs.acs.org/jced

Experimental and Computational Studies of Choline Chloride-Based Deep Eutectic Solvents Sasha L. Perkins, Paul Painter,* and Coray M. Colina* Department of Materials Science and Engineering, The Pennsylvania State University, 324 Steidle Building, University Park, Pennsylvania 16802, United States

J. Chem. Eng. Data 2014.59:3652-3662. Downloaded from pubs.acs.org by EASTERN KENTUCKY UNIV on 09/01/18. For personal use only.

S Supporting Information *

ABSTRACT: Choline-chloride based deep eutectic solvents (DES) have been used for several different applications (e.g., solubility, electrochemistry, and purifications) due to their relative inexpensive and readily available nature. In this work, three choline chloride-based DESs are simulated using molecular dynamics to study the hydrogen bonding interactions of the system. Three hydrogen bond donors (HBD) are studied in order to determine the changes in the hydrogen bonding interactions when the HBD is different in the DES. One dicarboxylic acid and two polyols (with different number of OH groups) were chosen as the HBDs of interest. First, the simulations are validated by comparing simulated and experimental thermodynamic and transport properties, when possible. Then, for maline (choline chloride/malonic acid), the more anomalous system studied here, molecular simulations complement results obtained from an FTIR spectroscopic study in order to further understand this unique system. Good agreement with experimental values was obtained for simulated density, heat capacity, and transport properties. A high relative percent of hydrogen bonding is observed for interactions between the anion and the HBD for the three main systems studied here, consistent with the nature of how these moieties interact in DESs. Comparison is also done with a previous DES studied in our group. From the infrared spectroscopic study conducted on maline films, band assignments were discussed highlighting a “free” carbonyl group of the carboxylic acid group in the eutectic mixture when the OH group is hydrogen bonded to something else. Additionally, a band is assigned to a hydrogen bonded carbonyl group. These band assignments are consistent with findings in the molecular simulations and highlight the predominant interactions of the system.

■

ILs in the past decade. A subsequent search for “deep eutectic solvents” primarily highlights experimental studies showing an increasing interest in the field of DESs in the past 10 years, seen in Figure S1 as well. Less than a handful of publications7−9 on molecular simulations of DES were found through more specific searches. It is hence expected that the number of publications using computational and experimental studies will increase for DESs with several reviews10−15 recently published on the promising applications of DESs. One of these reviews15 recognized that the top three applications for DESs are currently electrochemistry, material preparation, and synthesis. The authors also suggest an increasing interest for other applications, in particular separation processes. For example, choline chloride/ethylene glycol (referred to as ethaline) has been used4 with high current efficiency for the electropolishing of stainless steel. Choline chloride/glycerol (also known as glyceline) has remarkably proved5 to purify biodiesel. It has also been reported2 that choline chloride/

INTRODUCTION As inexpensive and readily available substances, deep eutectic solvents (DESs) have been shown to be interesting alternatives to organic solvents and ionic liquids (ILs) in recent studies of CO2 solubility,1 metal oxide solubility,2 electrochemistry,3,4 and purifications.5 In contrast to typical ILs that are made of discrete bulky ions, DES systems are interesting since the components of the mixture have high melting points in their pure state and become liquid at room temperature after they are mixed. In particular, choline chloride-based eutectic solvents have been used for many applications. However, compared to typical ILs there is a lack of experimental and theoretical data on fundamental thermodynamic and transport properties for this family of solvents. Moreover, binary and ternary mixtures that form these eutectic mixtures are potentially numerous and not straightforward to predict.6 A search on Web of Science showed that as of May 23, 2014 over 51,670 papers contained the topic of “ionic liquids”. This number is reduced to just almost 2,450 in an additional search for papers published on “ionic liquid + molecular simulations”. A steady increase of publications from 2003 to 2013, as shown in Figure S1, demonstrates the growing interest in simulating © 2014 American Chemical Society

Received: June 16, 2014 Accepted: September 24, 2014 Published: October 7, 2014 3652

dx.doi.org/10.1021/je500520h | J. Chem. Eng. Data 2014, 59, 3652−3662

Journal of Chemical & Engineering Data

Article

three DESs to understand how these types of HBDs participate in forming complexes within each system.

malonic acid (referred to as maline) provides a higher metal oxide solubility than other choline chloride-based eutectic mixtures. In maline, malonic acid’s anomalous transport behavior has been attributed16 to greater hydrogen bond donor (HBD) interactions in this system than in other DESs. For this and other choline chloride-based systems, understanding their behavior can be vital in predicting their advantages in current and prospective applications. Molecular simulations can provide insight into which specific interactions impact the behavior of these novel solvents. Furthermore, since the ability to tailor the thermodynamic and transport properties of DESs experimentally is still very limited, molecular simulations can provide efficient screening to predict the properties of the solvents. Choline chloride, a readily available quaternary ammonium salt, was chosen for all the eutectic mixtures studied here in order to determine the effect of the HBD. We thus assume that changes in the system are due to the different HBDs used. One dicarboxylic acid (malonic acid) and two monomeric polyols (ethylene glycol and glycerol), all shown in Figure 1, were

■

METHODOLOGY Computational Details. As in our previous work9 for reline (choline chloride/urea), the GAFF parameters were used for the bonded terms using the GAFF functional form. Each ion and neutral component was optimized in an isolated state in order to be able to transfer the charge model to other eutectic mixture combinations. A multiconformation procedure using the RESP charge derivation method17 in R.E.D. Tools18−21 was used to obtain the final charges for these flexible HBDs, hence minimizing geometry dependence during charge derivation. Conformations were obtained from the literature, when possible, and those used in this work are shown in Figures S2 through S-5 of the Supporting Information. Due to the nature of carboxylic groups, van der Waals term for the hydrogen atoms in the hydroxyl groups were slightly modified as in the work of Maginn et al.,22 because these parameters are originally zero for this atom type. Thus, the vdW radius was set to 0.112 Å and the well depth was set to 0.001 kcal/mol to provide a small hard core potential. This prevented some overlapping from occurring in malonic acid, which was not previously seen9 for reline. For consistency, this procedure was applied to all hydrogen atoms in any hydroxyl group for all the molecular simulations discussed in this work. All other GAFF LennardJones parameters were kept the same. Preliminary results, following a similar analysis as performed in ref 9 for densities and transport properties, showed better agreement with experimental values for these properties when using a reduced charge of ± 0.9 e for the ionic species. This is different from the reline system previously studied,9 which showed more promising results when using ± 0.8 e charges on the ionic species. Tables S-1 through S-4 in Supporting Information contain partial charges and Lennard-Jones parameters used for the choline cation and the HBD molecules applied in this work. For maline, 350 choline chloride ion pairs and 350 malonic acid molecules were randomly inserted in a low density simulation box corresponding to the eutectic composition. For ethaline and glyceline, 300 choline chloride ion pairs and 600 HBDs were placed in a low-density box at random in order to simulate the eutectic composition of these systems in a 1:2 molar ratio. Molecular dynamics were performed using AMBER10 and 12.23−26 After minimization, these systems were equilibrated using a compression and decompression scheme,27 previously used9 in our group, which efficiently worked in equilibrating the reline system. This is one possible method to achieve proper equilibration, since ILs and DESs have slow dynamics. Equilibrium runs of 3 ns were performed for ethaline and glyceline, while maline was equilibrated for about 15 ns. Production runs for all three systems were approximately 20 ns long. Langevin dynamic thermostats were applied with a collision frequency of 5 ps−1. The SHAKE algorithm was used to perform bond constraints on all hydrogen atoms. Simulations were carried out at 0.98 bar and at (298, 315, and 330) K with periodic boundary conditions and a nonbonded cutoff of 15 Å. Long-range electrostatics were calculated using the PME procedure. For transport properties, the NVE ensemble was used after the NPT production runs. In order to conserve energy, PME and SHAKE tolerances of 10−6 and 10−8, respectively, were applied for NVE runs. NPT and NVE production runs used a step size of (2 and 1) fs, respectively. For thermodynamic and transport property

Figure 1. Structure of (a) ethylene glycol, (b) glycerol, (c) malonic acid, and (d) choline.

chosen as the HBDs in order to study the effects on the hydrogen bonding network of these systems. Looking at changes in eutectic mixtures when using a dicarboxylic acid and two alcohols with two different functional groups can highlight differences in changing the functional group type (i.e., OH versus COOH) and quantity (i.e., two OH groups versus three OH groups). All three of these eutectic mixtures have a freezing point below room temperature and can thus be used as solvents in many room temperature applications. First, densities, thermodynamic properties, and self-diffusion coefficients were obtained for ethaline and glyceline through molecular dynamic (MD) simulations and compared to experimental values. Next, the hydrogen bond network in ethaline and glyceline are compared to each other to show the effects of changing the number of functional groups. For maline, the DES with the highest viscosity studied here, the focus was on thermodynamic properties, as well as vibrational spectroscopy, to determine which interactions are dominant in the system. A detailed structural and hydrogen bonding analysis was performed for all 3653



Article

Figure 2. Experimental and simulated density averaged over three simulation boxes. Solid blue line with a blue dot (a) simulated ChCl:EG 1:2, (b) simulated ChCl:Gly 1:2, and (c) simulated ChCl:MA 1:1; green triangle, Ciocirlan;28 red circle, D’Agostino;16 red square, Abbott, Harris;29 orange triangle, Leron;30 □, Shahbaz (calculated from the Rackett equation)31

Table 1. Volume Expansivity and Heat Capacity between 298 and 330 K for the DES in This Work eutectic mixture

simulated volume expansivity (104 K−1)

simulated heat capacity (J·mol−1 K−1)

experimental heat capacity (J·mol−1 K−1) for (303.15−333.15) K

ethaline glyceline maline

6.45 ± 0.05 6.09 ± 0.07 4.91 ± 0.29

209.27 ± 1.55 259.15 ± 2.87 229.32 ± 6.71

190.8−199.2 237.7−246.9 N/A

Table 2. Average Self-Diffusion Coefficients for Two DES Studied in This Work average transport (D·1011 m2 s−1) at 298 K eutectic mixture

cation (exp)

16

cation (sim)

anion (sim)

HBD (exp)16

1.81 ± 0.07 2.50 ± 0.28 0.30 ± 0.02 0.46 ± 0.09 average transport (D·1011 m2 s−1) at 330 K

4.77 0.52

HBD (sim) 3.75 ± 0.28 0.45 ± 0.02

ethaline glyceline

2.62 0.38

eutectic mixture

cation (exp*)16

cation (sim)

anion (sim)

HBD (exp*)16

HBD (sim)

ethaline glyceline

9.79 1.79

7.44 ± 0.17 2.11 ± 0.29

9.71 ± 0.10 3.09 ± 0.01

16.4 2.45

15.3 ± 0.20 3.12 ± 0.12

*

Experimental values are at 328 K.

attributed to Ciocirlan et al. were obtained by using a best fit line from Figure 1 in ref 28. Error bars were not included for clarity, but standard deviations ranged from (5.7·10−5 to 1.1· 10−3) g/cm3 between the average values of the simulation boxes. Tables 1 and 2 contain simulated and experimental volume expansivities, heat capacities, and self-diffusion coefficients for these systems. Volume expansivity values were obtained from the slopes of the density versus temperature lines in Figure 2, after dividing by the initial density. The trend is in good agreement with values calculated from the slopes of the experimental density values. The heat capacities were calculated following the methods developed by Cadena et al.32,33 (also used9 previously for reline). For both ethaline and glyceline, this method overestimates experimental values by only 7 %. Even though empirical correction factors have been applied in some IL systems34 to improve the agreement with experimental values, for the systems here studied no correction factors were used for consistency, due to the lack of experimental data for maline. In other words, no correction factors were implemented for the ideal heat capacity values. In considering the relative values of the different DES, Leron and Li35 concluded that as the molecular weight of the DES increases the molar heat capacity increases as well. This trend is observed for DES systems that have a 1:2 molar ratio. Maline, which was not reported in ref 35, did not follow this trend. It was expected that, since the molecular weight of maline (121.8 g/mol) is the highest among these three DESs, it would have had the highest molar heat capacity. However, the simulated value of heat capacity for maline is not the highest value among these DESs. Since the

analyses, values were averaged over three simulation boxes after equilibration, each with different starting configurations. RDF curves were also averaged over three simulation boxes during a 20 ns long production run. The hydrogen bonding analyses were performed over a 1 ns trajectory, with 2 ps between frames at the end of the production runs. Then, relative percent occupancies and percent of hydrogen bonding types were obtained. Percent occupancy is described as the total time a unique hydrogen bond is present during the analyzed trajectory. All percent occupancies were included in this study. Experimental Details. Malonic acid (Sigma-Aldrich) and Choline chloride (Alfa Aesar) were mixed in a 1:1 molar ratio over a hot plate (∼80 °C) until a homogeneous liquid was formed. An absorbance spectrum of maline on a polished KBr window was obtained using a Nicolet 6700 FTIR spectrometer with a resolution of 2.0 cm−1 over 200 scans. Diffuse reflectance infrared Fourier transform (DRIFT) spectra were recorded for malonic acid. All spectra were taken at ambient conditions.

■

RESULTS AND DISCUSSION We use a complementary computational and experimental approach to underline the important interactions that affect the behavior of DESs. We first look at the simulation results of ethaline, glyceline, and maline. Then, a spectroscopic study of maline is reported in order to better understand this unique system. Computational Studies. Force Field Validation. For all three systems studied here, the simulated densities are in very good agreement with experimental values when using ± 0.9 e for the ionic species, as shown in Figure 2. Note that the values 3654



Article

Figure 3. Atom−atom RDFs of hydrogen atoms on different groups around the (a) anion and (b) O3 in ethaline.

Figure 4. COM RDF for ethaline. (a) Blue: anion−anion; brown: anion−cation; cyan: anion−HBD; (b) blue: cation−cation; brown: HBD−HBD; cyan: HBD−cation.

species in our simulations provides reasonable agreement with both simulated self-diffusion coefficients and density values. For glyceline, transport properties are underestimated by 14−20 % of experimental values at 298 K. At 330 K, the simulated values are overestimated by 17−27 % compared to the experimental values. Average and standard deviations shown in Table 2 were obtained using three different simulation boxes per temperature. Structural Analysis. A structural analysis for each DES was performed, with the results referring to the atom numbers shown in Figure 1. First, we will discuss atom−atom and center of mass (COM) radial distribution functions (RDF), showing the contact distances between different moieties. Then a more discriminant hydrogen bond analysis depicting the hydrogen bond network in these DESs will be described. Choline Chloride:Ethylene Glycol 1:2 (Ethaline). In the first DES studied here, ethaline has two close contacts involving the anion of the system. The first is between the chlorine anion and the hydroxyl hydrogen atom of choline, and the second is between the anion and the hydroxyl hydrogen atoms in ethylene glycol, as seen in Figure 3a. H18 refers to the hydroxyl hydrogen atom in choline. Figure 3a also shows contact distances between the OH hydrogen atoms of ethylene glycol and the anion. These will not be included in the hydrogen bond analysis in the following section, since they occur at long distances compared to the hydroxyl OH/OH hydrogen bonds in the system. Figure 3b illustrates the contact distances between the oxygen atom (O3) in the ethylene glycol molecule with other hydroxyl hydrogen atoms in the system. Note that the contact distances between the oxygen atom (O4) in the ethylene glycol molecule with other hydroxyl hydrogen atoms in the system are not shown for brevity, since a similar plot to Figure 3b is obtained. The tall and narrow peaks in this image

heat capacity value of maline has not yet been validated with experiments, rough estimates of heat capacity values for eutectic mixtures at 298 K were obtained by simply calculating a theoretical heat capacity from the heat capacities of the HBDs obtained from the NIST webbook.36 From this crude heat capacity estimate, a trend is observed for the eutectic mixtures that matches the trend determined in this simulation study, ethaline < maline < glyceline. The self-diffusion coefficients for maline were not simulated, since the computational time scales available would not lead to accurate transport properties of this highly viscous system. However, about 40 and 80 ns of trajectory were analyzed for ethaline and glyceline, respectively, to estimate the transport properties of these systems. Self-diffusion coefficients were calculated only in the region where a diffusive regime could be observed, by calculating the so-called β parameter previously discussed9 and summarized here for completeness. In short, to determine the appropriate diffusive regime, autocorrelation curves were analyzed. The so-called beta-parameter, β(t) = d log10⟨Δr(t)2⟩/d log10 t, as used by Del Popolo and Voth37 was obtained to determine the location of the diffusive regime. If β < 1, the system is in the subdiffusive regime, and the diffusive regime is obtained at β =1. Thus, self-diffusion coefficients were calculated when β =1. It can be seen in Table 2 that for ethaline the transport properties are still underestimated by 20−30 % for simulations at 298 K and 5−25 % for simulations at 330 K. Maginn et al.22 also observed underestimated values by a factor of 2 for transport properties of ionic liquids using GAFF parameters and reduced charges. However, some preliminary work on this system (not shown here) indicated that using partial charges of ± 0.8e overestimated transport properties as well as underestimated density values. Thus, using ± 0.9e for the ionic 3655



Article

Figure 5. Atom-atom RDFs of hydrogen atoms on different groups around (a) anion, (b) O3, and (c) O1 in glyceline.

are due to covalently bonded O−H interactions. The interactions around 2 Å for both panels b and c in Figure 3 are presumably due to inter- or intramolecular hydrogen bonded interactions. The relative contributions from inter- and intramolecular interactions are not calculated here. However, Figure S-6 in the Supporting Information shows examples of inter- and intramolecular hydrogen bond distances that would contribute to the RDF. However, a detailed hydrogen bond analysis in the following section will conclude which type of hydrogen bond is predominant. Peaks at distances at or greater than 3 Å are likely non-hydrogen bonded intramolecular interactions and are not expected to contribute to a hydrogen bond network. Possible interactions occurring at these distances are shown in Figure S-6 of the Supporting Information. The COM RDFs between different molecules in the system show important anion−cation and anion-HBD contacts. With these curves, different relative orientations can be discussed. For the anion−anion/cation/HBD RDFs shown in Figure 4a, the anion-HBD and anion−cation contacts predominate at distances less than 6 Å, so that interactions between these groups prevail. The anion-HBD COM RDF has a very peculiar shape with two main peaks and a small shoulder between these peaks. One possible explanation for these two main peaks involves the relative orientation of ethylene glycol with respect to the anion, which we will describe in terms of three different orientations, labeled type 1, type 2, and type 3. A representation of these three arrangements can be found in Figure S-7 in Supporting Information. Type 1 occurs at the peak near 3.4 Å in Figure 4a. This orientation can be described by picturing a line from the anion to the COM of ethylene glycol intersecting the C−C bond of ethylene glycol nearly perpendicularly. This contact occurs with both OH groups of ethylene glycol in a position to interact with the anion. Type 2 orientations are those where the line from the anion to the COM of ethylene

glycol intersects the C−C bond of ethylene glycol at an angle close to 0° (i.e., not quite collinearly). This contact occurs when only one OH group in ethylene glycol is close enough to hydrogen bond to the anion. Type 3 contacts occur when this line intersects the C−C bond of ethylene glycol close to 90°, similar to type 1 orientations, but at longer distances than type 1 contacts, thus probably not involving hydrogen bonds. This occurs when the ethylene glycol molecule has CH, instead of OH groups, facing the anion. Both type 2 and type 3 orientations are present at distances near 4.6 Å, shown as the highest intensity peak in Figure 4a for the anion-HBD interaction. The relative contributions from each type of interaction were not calculated. Figure 4b shows additional contacts, with some being more interesting than others. The most likely interactions to occur within 6 Å involve HBD−HBD and HBD−cation contacts, as expected. An analysis on the hydrogen bonds in this system will be discussed later, showing which of the two contacts has a greater contribution to the hydrogen bond network. Choline Chloride:Glycerol 1:2 (Glyceline). The second DES containing a polyol as a HBD is glyceline, thus a similar analysis as done with ethaline is conducted. Figure 5a presents different contact distances for the interactions of interest around the anion and the hydroxyl hydrogen atom in a glycerol molecule. In this system, distances involving H18, H1, H7, and H8 are likely to occur near 2 Å. Figure 5b,c illustrates the contact distance between OH groups in the system. Similarly to ethaline, the tallest and thinnest peaks around 1 Å are due to covalently bonded O−H groups. Possible hydrogen bonded interactions are highlighted by the peaks around 2 Å. A hydrogen bond analysis that will be discussed later will show if there is a predominant inter- or intramolecular interaction in these DES. Peaks in the RDF curves at distances at or greater 3656



Article

Figure 6. COM RDF for glyceline. (a) Blue: anion−anion; brown: anion−cation; cyan: anion−HBD; (b) Blue: cation−cation; brown: HBD−HBD; cyan: HBD−cation.

Figure 7. Atom−-atom RDFs of hydrogen atoms on different groups around (a) anion and (b) O5 and in maline. Blue: H18 (choline); magenta: H19/H22 (MA); green: H20/H21 (MA).

Figure 8. COM RDF for maline. (a) Blue: anion−anion; brown: anion−cation; cyan: anion−HBD; (b) blue: cation−cation; brown: HBD−HBD; cyan: HBD−cation.

interactions do, however, occur at shorter distances than same-charge interactions, i.e., anion−anion and cation−cation correlations. Choline Chloride:Malonic Acid 1:1 (Maline). The last DES discussed in this work is maline. We start by first focusing on distances between the anion and other atoms in maline. Figure 7a shows the closest contacts near 2.3 Å around the anion to be from the hydroxyl hydrogen atom of choline and malonic acid. Hydroxyl hydrogen atoms covalently bonded to carbon atoms are not expected to contribute to a hydrogen bond network, due to the longer Cl--H distances. Figure 7b illustrates the contact distances around the carbonyl oxygen atom. Two possible interactions occur with either the hydroxyl groups from the cation or groups in other malonic acid molecules. The highest intensity peaks in these two figures, near (2.0 and 2.95) Å, are likely due to intramolecular distances

than 3 Å are likely non-hydrogen bonded intramolecular interactions within the same molecule. The bimodal distribution observed in the COM RDFs of reline (see ref 9) and ethaline (Figure 4a) is also present in Figure 6a, suggesting that this is typical for choline−chloride interactions in eutectic mixtures, due to the possible orientations of the choline cation near the chlorine ion. The anion−HBD RDF in glyceline lacks the first sharp peak observed for the same interaction in reline (see ref 9) and ethaline (Figure 4a), potentially due to the wider variety of orientations possible for glycerol−anion interactions (depending on which of the three functional groups makes hydrogen bonds with the anion). HBD−HBD and HBD−cation COM distances, shown in Figure 6b, can also influence the structure of the hydrogen bond network, although at longer distances than the interactions involving the anion. These two 3657



Article

Figure 9. Relative hydrogen bond per frame (using DH···A) for (a) ethaline, (b) glyceline, (c) reline, and (d) maline.

Hydrogen Bond Analysis. The RDFs analyzed above show possible relationships between atoms that can participate in hydrogen bonding. In this section, a distance and angle cutoff criteria was used to identify the strongest hydrogen bonds present. The following criteria was used: heavy atom distance cutoff of 3.5 Å and a minimum angle cutoff for ∠DHA of 150°, where D is the hydrogen donor atom, H is the hydrogen atom, and A is the hydrogen acceptor atom. This hydrogen bond analysis shows both the number and types of hydrogen bonds present over the time analyzed. Hydrogen bond interactions in ethaline, glyceline, reline, and maline are analyzed in this section and compared to one another. For each system, the fraction of H-bond shown was normalized by the total number of H-bonds in the system for the trajectory analyzed. This was done so the hydrogen bond types can be compared relative to each other (i.e., in the y-axis) as well as the presence of that type of hydrogen bond in the trajectory analyzed (i.e., in the xaxis). The data for reline was obtained from the work presented in ref 9. Figure 9 shows the relative contribution of hydrogen bonds for a given hydrogen bond type. For all systems except reline (Figure 9c), the largest fraction of hydrogen bond interactions occur between the HBD and the anion, as shown in Figure 9a,b,d. Reline still exhibits a great amount of urea−urea interactions, because the oxygen atom in urea also acts as a relatively strong hydrogen bond acceptor. Additionally, the two polyols (Figure 9a,b) have similar trends for the different hydrogen bond types (i.e., both systems have the largest contribution from HBD−anion interactions). The main difference being a larger contribution from the HBD−HBD hydrogen bonds in glyceline than in ethaline, due to the additional OH functional group present in glycerol. Maline exhibits an extensive amount of HBD−anion and cation−anion interactions (Figure 9d). As previously mentioned, maline tends to have a different transport behavior relative to other DESs. In this study, the HBD−HBD interactions fall short of being the largest contributor in this DES. Another possible reason for its anomalous behavior, as

within the same malonic acid molecule. The peak at distances slightly less than 2 Å in Figure 7b (a shoulder in the magenta curve) is indicative of pairs of atoms that would likely participate in hydrogen bonded interactions. A schematic is shown in Figure S-8 in the Supporting Information showing the distances for different cases of inter- and intramolecular distances and possible interactions. Additional interactions that would probably not contribute to a hydrogen bond network are shown in RDF curves in Figure S-9 of the Supporting Information. A more detailed hydrogen bond analysis will show the relative contributions of the type of interactions that could affect the hydrogen bond network. As done for ethaline and maline, the COM RDF curves are also analyzed in order to determine distances for typical orientations in this system. The first COM RDF analysis is done for the components around the anion. It can be seen from Figure 8a that there is a bimodal distribution of distances between the COM of the anion and cation, with the closest contact near 4.3 Å and a second peak near 5.2 Å. In between these two peaks, there is a contact distance involving the HBD and the anion. There are shoulders at shorter distances for this RDF curve that could be due to less preferred orientations of the malonic acid molecules around the anion. As expected, the anion−anion separations are likely to occur at much longer distances, as are cation−cation separations, shown in Figure 8b. In this figure, the two contacts of importance are those between HBD molecules and between HBD molecules and the cation. Both curves involve similar intermolecular distance and studies involving a hydrogen bond analysis provide more insight. From the structural analysis of these three DESs, it is interesting that all three DESs have very similar anion−cation, anion−anion, cation−cation, HBD−HBD, and HBD−anion distances. This is potentially a signature feature of DESs. The anion−HBD RDFs are the ones that change the most from DES to DES, presumably due to the different number and types of functional groups present and their ability to interact with the anion. 3658



Article

Figure 10. Hydrogen bond percent occupancies (using DH···A) for (a) ethaline, (b) glyceline, (c) reline, and (d) maline.

Figure 11. Spectra of neat choline chloride, malonic acid, and their eutectic mixture in the OH region (left) and carbonyl stretch region (right).

showing a larger number of hydrogen bonds at higher percent occupancies have a higher viscosity (ethaline < glyceline < reline < maline). The trend is not as easily observed for selfdiffusion coefficients (except at the extremes), because the experimental values for glyceline and reline are very similar. Overall, in agreement with the idea of a complexed anion of some form (i.e., single or multiple HBD interacting with an anion), one of the most significant interactions for all these DESs involves the anion−HBD hydrogen bond type. It is also important to note that using the hydrogen bonding criteria discussed previously for these systems, a negligible number of intramolecular hydrogen bonding (∼0−1 %) was determined in these mixtures. This percent is already included into the HBD− HBD relative fraction values. This suggests that intermolecular hydrogen bonds are the main contributors to the hydrogen bond network. Additionally, HBD−cation interactions seem to have the smallest contribution to this extensive network and

shown here, is cation−anion interactions coupled with strong HBD−anion interactions, which drastically increase the extent of the hydrogen bond network and decrease the mobility of malonic acid molecules within the system. Regarding the presence of unique hydrogen bonds during a short trajectory, maline has a larger number of hydrogen bonding interactions with greater percent occupancies (Figure 10d) than the other DESs, suggesting that these hydrogen bonds are present for longer times. This is related to the fact that maline has the lowest self-diffusion coefficients (highest viscosity) among these eutectics. Ethaline, with the fastest selfdiffusion coefficients (lowest viscosity), does not have many hydrogen bonds with more than a 60% occupancy for the 1 ns trajectory (seen in Figure 10a). The greatest percent occupancies for glyceline, shown in Figure 10b, falls somewhere between those of ethaline and reline. After analyzing percent occupancies from Figure 10a−d, it is evident that the systems 3659



Article

Figure 12. Carbonyl stretch region of malonic acid (left) and maline (right) spectra with bands obtained by curve-resolving.

center of each of the two CO bands is separated by about 2.8 Å (obtained from our optimized geometry calculations). Transition dipole coupling varies as the inverse cube of the distance apart, so one would anticipate a greater splitting or separation of the two carbonyl bands depicted in the spectrum of malonic acid, as observed. The OH stretching region of the spectrum of the eutectic mixture displays the broad bands below 3100 cm−1 seen in malonic acid, but the sharper superimposed bands are no longer apparent, possibly because of the amorphous, liquid state of the material. In addition, a new broad band appears at higher wavenumbers, near 3327 cm−1, indicating OH hydrogen bonded groups with a weaker hydrogen bond strength than the cyclic dimers found in pure malonic acid. Furthermore, the prominent OH stretching mode associated with the choline chloride OH···Cl− hydrogen bond seems to disappear. The broad nature of these bands makes specific assignments difficult, but are consistent with the results of simulations, which suggest that the OH stretching region reflects the presence of multiple types of hydrogen bonds in the form of OH (malonic acid)···Cl−, OH(choline)···Cl−, and malonic acid−malonic acid hydrogen bonds. The carbonyl stretching region provides more information. Upon forming the DES, the vibrational mode near 1704 cm−1 becomes much less intense and is shifted or replaced by modes at higher wavenumbers, overlapping significantly with any bands near 1740 cm−1 and resulting in a broad, almost featureless profile. However, it is possible to curve-resolve this region into component bands in order to obtain some insight. The infrared spectrum of pure malonic acid was first curveresolved into its component bands following the procedure described in our previous work.9 The results are shown in Figure 12 (left). As might be expected, there are two main bands, centered near (1741 and 1704) cm−1, with very weak bands at the edge of this spectral region. These may be real or artifacts of a poorly placed baseline. Our experience is that they are usually real. If one expands the absorbance scale of any midinfrared region of the spectrum of most materials, a multitude of very weak overtone/combination bands are often revealed. If

thus the focus for future DES should be to maximize anion− HBD interactions. Experimental Spectroscopic Studies. The infrared spectra of malonic acid, choline chloride, and the DES mixture are compared in Figure 11. The spectra on the left show the OH stretching region, while the spectra on the right show the CO stretching region. The spectrum of choline chloride was discussed previously.9 Here, the spectrum of malonic acid will be examined first, and then the changes that occur upon forming the DES are studied. The OH stretching region of malonic acid is typical of carboxylic acids that form strongly hydrogen bonded dimer rings, with broad overlapping bands centered at frequencies lower than 3100 cm−1. Superimposed upon these broad bands are sharper modes that can be assigned to CH stretching fundamentals and/or overtone/combination modes. The carbonyl stretching region is characterized by two strong modes centered near (1740 and 1704) cm−1. In older work,38,39 it was suggested that the two hydrogen bonded dimer rings found in each unit cell, illustrated in Figure 1 of ref 37, have somewhat different strengths, accounting for the two bands. However, in subsequent publications, Bougeard et al.40 noted that neutron scattering work found that the two dimer rings are similar, with almost equal hydrogen bond lengths. This would be unlikely to result in difference of 35 cm−1 in the position of the CO stretching modes. We suggest an alternative assignment, transition dipole coupling between the two dimers found in each cell. It is wellknown that interactions between CO groups in each dimer lead to a splitting, but because of symmetry considerations only one of the modes is infrared active, while the other is Raman active. In addition, this particular splitting is probably not solely due to transition dipole interactions.41 However, the interactions between CO groups in different hydrogen bonded pairs can lead to significant splittings, about 20 cm−1 in stearic acid42 and 15 cm−1 in alkoxybenzoic acids.43 The splitting observed in alkoxybenzoic acids has been attributed to interactions between carbonyl group in adjacent stacked layers, which are about 4 Å apart. In malonic acid, on average, the 3660



Article

these modes are not included in a curve fitting procedure, poor fits and bands with abnormally large widths at half height are generated. A detailed discussion of our procedure can be found elsewhere44 but using this methodology a band is resolved near 1716 cm−1. This mode is also apparent in a broadening of the 1704 cm−1 mode on the high frequency side in the original spectrum. Again, this may be an overtone or combination mode that gains some intensity as a result of Fermi resonance interactions with the fundamentals. The curve resolved spectrum of the DES (right of Figure 12) is dominated by two bands, near (1751 and 1724) cm−1, together with a number of weaker modes. The former can be assigned to “free” or non-hydrogen bonded carbonyl groups. One has to distinguish two types of free groups, those where both the CO and OH groups of a carboxylic acid are not hydrogen bonded, as in very dilute solutions or the gaseous state, and free carbonyl groups in a carboxylic acid where the OH group is hydrogen bonded to some other moiety. In polymers containing methacrylic acid groups, for example, the former groups absorb near 1740 cm−1, while the latter is found at lower frequencies, near 1730 cm−1.45 On this basis, we assign the band near 1751 cm−1 to “free” carbonyls in carboxylic acids where the OH group is hydrogen bonded. Simulations (Figure 10d) show that most of the OH groups of malonic acid are hydrogen bonded to the Cl− anion which would show the presence of “free” carbonyl groups when this interaction occurs. These results are consistent with the ability of these mixtures to form a DES. The band at 1724 cm−1 can be assigned to carbonyl groups hydrogen bonded to OH groups, consistent with the simulations that show that most of the remaining OH groups in the system are hydrogen bonded to carboxylic CO groups. The weaker modes are probably associated with overtone or combination modes or due to transition dipole coupling between COOH groups. In the liquid state, malonic acid groups can take on various conformations and the geometry and proximity of carboxylic acids within a molecule would then vary. Two clusters of molecules for maline showing different possible interactions are shown in Figure S-10 of the Supporting Information.

transition dipole coupling in pure malonic acid are now attributed to “free” carbonyl groups, where OH(malonic acid) interacts with another moiety and the carbonyl group in the COOH group is “free”, and hydrogen bonded CO groups. This assignment is consistent with molecular simulations which show OH(malonic acid)···Cl− interactions (causing the carboxylic CO group to be “free”), as well as OH groups in the system (from both choline and malonic acid) hydrogen bonded to the carboxylic CO groups.

■

ASSOCIATED CONTENT

S Supporting Information *

Complete list of force field parameters and conformations. Additional helpful images for different orientations and interactions are also illustrated. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

* E-mail: [email protected]. * E-mail: [email protected]. Funding

This work was supported in part through the National Science Foundation (DMR-0901180) and Penn State’s MatSE ButtonWaller awards. High performance computational resources were provided by instrumentation funded by the National Science Foundation through Grant OCI-0821527, as well as Penn State’s Research Computing and Cyberinfrastructure Group. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Li, X.; Hou, M.; Han, B.; Wang, X.; Zou, L. Solubility of CO2 in a Choline Chloride + Urea Eutectic Mixture. J. Chem. Eng. Data 2008, 53, 548−550. (2) Abbott, A. P.; Capper, G.; Davies, D. L.; McKenzie, K. J.; Obi, S. U. Solubility of Metal Oxides in Deep Eutectic Solvents Based on Choline Chloride. J. Chem. Eng. Data 2006, 51, 1280−1282. (3) Abbott, A. P.; Capper, G.; McKenzie, K. J.; Ryder, K. S. Voltammetric and impedance studies of the electropolishing of type 316 stainless steel in a choline chloride based ionic liquid. Electrochim. Acta 2006, 51, 4420−4425. (4) Abbott, A. P.; Capper, G.; Swain, B. G.; Wheeler, D. A. Electropolishing of stainless steel in an ionic liquid. Trans. Inst. Met. Fin. 2005, 83, 51−53. (5) Abbott, A. P.; Cullis, P. M.; Gibson, M. J.; Harris, R. C.; Raven, E. Extraction of glycerol from biodiesel into a eutectic based ionic liquid. Green Chem. 2007, 9, 868. (6) Maugeri, Z.; Domínguez de María, P. Novel choline-chloridebased deep-eutectic-solvents with renewable hydrogen bond donors: levulinic acid and sugar-based polyols. RSC Adv. 2012, 2, 421. (7) Rimsza, J. M.; Corrales, L. R. Adsorption complexes of copper and copper oxide in the deep eutectic solvent 2:1 urea−choline chloride. Comput. Theor. Chem. 2012, 987, 57−61. (8) 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, 2433−41. (9) Perkins, S. L.; Painter, P. C.; Colina, C. M. Molecular Dynamic Simulations and Vibrational Analysis of an Ionic Liquid Analogue. J. Phys. Chem. B 2013, 117, 10250−10260. (10) Zhang, Q.; Oliveira Vigier, K.; Royer, S.; Jérôme, F. Deep eutectic solvents: syntheses, properties and applications. Chem. Soc. Rev. 2012, 41, 7108−7146.

■

CONCLUSIONS Molecular simulations performed on several DES showed very good agreement with experimental densities and thermodynamic properties. Simulating transport properties proved to be a nontrivial task. Reasonable agreement was obtained with experimental self-diffusion coefficients for ethaline and glyceline. Since the experimental self-diffusion coefficients for maline are 1 and 2 orders of magnitude smaller than those for glyceline and ethaline, respectively, these were not validated here due to insufficient computational resources. A structural and hydrogen bond analysis highlighted the importance of anion−HBD interactions for all four systems. In three of the systems (reline, ethaline, and glyceline), HBD−HBD interactions also play an important role. Infrared spectra of pure malonic acid and maline showed changes in the OH stretch region. The broad nature of the OH stretching bands make it difficult to make specific band assignments. However, results from the simulations suggest multiple types of hydrogen bonds occurring in this region of the spectra, including OH(malonic)···Cl−, OH(choline)···Cl−, and malonic acid···malonic acid interactions. In the carbonyl stretching region, the two main bands observed for malonic acid either disappear or shift frequency values. The split bands in this region that were attributed to a 3661



Article

from 298.15 to 333.15K. J. Taiwan Inst. Chem. Eng. 2012, 43, 551− 557. (31) Shahbaz, K.; Baroutian, S.; Mjalli, F. S.; Hashim, M. A.; AlNashef, I. M. Densities of ammonium and phosphonium based deep eutectic solvents: Prediction using artificial intelligence and group contribution techniques. Thermochim. Acta 2012, 527, 59−66. (32) Cadena, C.; Zhao, Q.; Snurr, R. Q.; Maginn, E. J. Molecular modeling and experimental studies of the thermodynamic and transport properties of pyridinium-based ionic liquids. J. Phys. Chem. B 2006, 110, 2821−2832. (33) Cadena, C.; Maginn, E. J. Molecular simulation study of some thermophysical and transport properties of triazolium-based ionic liquids. J. Phys. Chem. B 2006, 110, 18026−18039. (34) Gutowski, K. E.; Gurkan, B.; Maginn, E. J. Force field for the atomistic simulation of the properties of hydrazine, organic hydrazine derivatives, and energetic hydrazinium ionic liquids. Pure Appl. Chem. 2009, 81, 1799−1828. (35) Leron, R. B.; Li, M.-H. Molar heat capacities of choline chloridebased deep eutectic solvents and their binary mixtures with water. Thermochim. Acta 2012, 530, 52−57. (36) Linstrom, P. J., Mallard, W. G.; Eds. NIST Chemistry WebBook, NIST Standard Reference Database Number 69. (37) Del Pópolo, M. G.; Voth, G. A. On the Structure and Dynamics of Ionic Liquids. J. Phys. Chem. B 2004, 108, 1744−1752. (38) Pigenet, C.; Lucazeau, G.; Novak, A. Vibrational spectra and structure of malonic acid. J. Chim. Phys. 1976, 73, 141−145. (39) Ganguly, S.; Fernandes, J. R.; Desiraju, G. R.; Rao, C. N. . Phase transition in malonic acid: an infrared study. Chem. Phys. Lett. 1980, 69, 227−229. (40) Bougeard, D.; De Villepin, J.; Novak, A. Vibrational spectra and dynamics of crystalline malonic acid at room temperature Internal vibrations. Spectrochimistry 1988, 44A, 1281−1286. (41) Bosi, P.; Zerbi, G. Splitting of carbonyl stretching frequencies from transition dipole coupling. Chem. Phys. Lett. 1976, 38, 571−573. (42) Vergoten, G.; Fleury, G. Transition dipole-dipole coupling interactions in the B form of stearic acid single crystals. Chem. Phys. Lett. 1984, 112, 272−274. (43) Painter, P. C.; Cleveland, C.; Coleman, M. M. An Infrared Spectroscopic Study of p-n-Alkoxybenzoic Acids. Mol. Cryst. Liq. Cryst. 2000, 348, 269−293. (44) Painter, P. C.; Zhao, H.; Park, Y. Vibrational Relaxation and Dynamical Transitions in Atactic Polystyrene. Macromolecules 2009, 42, 435−444. (45) Cleveland, C. S.; Guigley, K. S.; Painter, P. C.; Coleman, M. M. Infrared Spectroscopic Studies of Poly(styrene-co-methacrylic acid) Blends with Polytetrahydrofuran. Macromolecules 2000, 33, 4278− 4280.

(11) Carriazo, D.; Serrano, M. C.; Gutiérrez, M. C.; Ferrer, M. L.; Monte, F. Deep-eutectic solvents playing multiple roles in the synthesis of polymers and related materials. Chem. Soc. Rev. 2012, 41, 4996−5014. (12) Tang, S.; Baker, G. a; Zhao, H. Ether- and alcohol-functionalized task-specific ionic liquids: attractive properties and applications. Chem. Soc. Rev. 2012, 41, 4030−4066. (13) Domínguez de María, P.; Maugeri, Z. Ionic liquids in biotransformations: from proof-of-concept to emerging deep-eutectic-solvents. Curr. Opin. Chem. Biol. 2011, 15, 220−225. (14) Ruß, C.; König, B. Low melting mixtures in organic synthesis − an alternative to ionic liquids? Green Chem. 2012, 14, 2969−2982. (15) Francisco, M.; Van den Bruinhorst, A.; Kroon, M. C. LowTransition-Temperature Mixtures (LTTMs): A New Generation of Designer Solvents. Angew. Chem., Int. Ed. 2013, 52, 3074−3085. (16) 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. (17) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. A WellBehaved Electrostatic Potential Based Method Using Charge Restraints for Deriving. J. Phys. Chem. 1993, 97, 10269−10280. (18) Dupradeau, F.-Y.; Pigache, A.; Zaffran, T.; Savineau, C.; Lelong, R.; Grivel, N.; Lelong, D.; Rosanski, W.; Cieplak, P. The R.E.D. tools: advances in RESP and ESP charge derivation and force field library building. Phys. Chem. Chem. Phys. 2010, 12, 7821−39. (19) Vanquelef, E.; Simon, S.; Marquant, G.; Garcia, E.; Klimerak, G.; Delepine, J. C.; Cieplak, P.; Dupradeau, F.-Y. R.E.D. Server: a Web Service for Deriving RESP and ESP Charges and Building Force Field Libraries for New Molecules and Molecular Fragments. Nucl. Acids Res. 2011, W511−W517. (20) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A.; Vreven, J. T.; Kudin, K. N.; Burant, J. C. et al. Gaussian 03, revision E.01; Gaussian, Inc.: Wallingford, CT, 2004. (21) Macke, T. J.; Svrcek-Seiler, W. A.; Brown, R. A.; Kolossvary, I.; Bomble, Y. J.; Anandakrishnan, R.; Case, D. A.; Walker, R. C.; Crowley, M. F.; Brozell, S. et al. AmberTools 1.5 Manual, 2011. (22) Liu, H.; Maginn, E. J.; Visser, A. E.; Bridges, N. J.; Fox, E. B. Thermal and Transport Properties of Six Ionic Liquids: An Experimental and Molecular Dynamics Study. Ind. Eng. Chem. Res. 2012, 51, 7242−7254. (23) Case, D. A.; Darden, T. A.; T.E. Cheatham, I.; Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Crowley, M.; Walker, R. C.; Zhang, W. et al. Amber 10, 2008. (24) Case, D. A.; Darden, T. A.; Cheatham, T. E., III; Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Walker, R. C.; Zhang, W.; Merz, K. M. et al. Amber 12, 2012. (25) Götz, A. W.; Williamson, M. J.; Xu, D.; Poole, D.; Le Grand, S.; Walker, R. C. Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 1. Generalized Born. J. Chem. Theory 2012, 8, 1542−1555. (26) Le Grand, S.; Götz, A. W.; Walker, R. C. SPFP: Speed without compromiseA mixed precision model for GPU accelerated molecular dynamics simulations. Comput. Phys. Commun. 2012, 184, 374−380. (27) Larsen, G. S.; Lin, P.; Hart, K. E.; Colina, C. M. Molecular Simulations of PIM-1-like Polymers of Intrinsic Microporosity. Macromolecules 2011, 44, 6944−6951. (28) Ciocirlan, O.; Iulian, O.; Croitoru, O. Effect of Temperature on the Physico-chemical Properties of Three Ionic Liquids Containing Choline Chloride. Rev. Chim (Bucharest) 2010, 61, 721−723. (29) 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, 4910−3. (30) Leron, R. B.; Soriano, A. N.; Li, M.-H. Densities and refractive indices of the deep eutectic solvents (choline chloride+ethylene glycol or glycerol) and their aqueous mixtures at the temperature ranging 3662


Experimental and Computational Studies of Choline Chloride-Based

Recommend Documents