You are on page 1of 14

The Journal

ARTICLE scitation.org/journal/jcp
of Chemical Physics

A comprehensive computational and principal


component analysis on various choline
chloride-based deep eutectic solvents to reveal
their structural and spectroscopic properties
Cite as: J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569
Submitted: 31 March 2021 • Accepted: 30 June 2021 •
Published Online: 29 July 2021

Mahmudul Islam Rain,1,2 Humayun Iqbal,1,2 Mousumi Saha,1,3 Md Ackas Ali,1,2


4 5,a)
Harmeet Kaur Chohan, Md Sajjadur Rahman, and Mohammad A. Halim4,a)

AFFILIATIONS
1
Division of Quantum Chemistry, The Red-Green Research Center, BICCB, 16, Tejkunipara, Tejgaon, Dhaka 1215, Bangladesh
2
Department of Chemistry, Jahangirnagar University, Savar, Dhaka 1342, Bangladesh
3
Department of Chemistry, University of Dhaka, Curzon Hall, Dhaka 1000, Bangladesh
4
Department of Physical Sciences, University of Arkansas-Fort Smith, Fort Smith, Arkansas 72913-3649, USA
5
Department of Chemistry and Biochemistry, South Dakota State University, Brookings, South Dakota 57007, USA

Note: This paper is part of the JCP Special Topic on Chemical Physics of Deep Eutectic Solvents.
a)
Authors to whom correspondence should be addressed: rahman.sajjadur@sdstate.edu and mohammad.halim@uafs.edu.
Tel.: 1-479-788-7741. Fax: +1-479-424-6741

ABSTRACT
In this study, the quantum chemical properties, nonbonding interactions, and spectroscopic insights of a wide variety of choline chloride
(ChCl)-based deep eutectic solvents were investigated employing molecular dynamics (MD), density functional theory, and spectroscopic
analyses. Nine experimentally reported ChCl-based deep eutectic solvents (DESs) were selected for this study where ChCl was common in
all the DESs and the hydrogen bond donors (HBDs) were varied. The most energetically favorable cluster was selected using MD simula-
tion followed by density functional theory calculation. The most stable cluster structures were fully optimized, and their quantum chemical
properties and IR spectra were computed at the ωB97XD/6-31G++(d,p) level of theory. Principal component analysis was performed to
distinguish their behavioral differences and to find out if any correlation exists among the 1:1 and 1:2 clusters. The atom–atom radial dis-
tribution functions based on MD simulations revealed that several hydrogen bonds were formed among the donor and acceptor molecules.
However, the most prominent hydrogen bonds were found to be N–HHBD ⋅ ⋅ ⋅Cl− for ChCl:U, ChCl:TU, and ChCl:Ace and O–HHBD ⋅ ⋅ ⋅Cl−
for ChCl:Glu, ChCl:Ma, ChCl:Ox, ChCl:Gly, and ChCl:Phe. Both N–HHBD ⋅ ⋅ ⋅Cl− and O–HHBD ⋅ ⋅ ⋅Cl− were major interactions for ChCl:Pro,
where Cl− worked as a bridge between Ch+ and the respective donors. In addition, the –OH of Ch+ showed strong intermolecular interactions
with the acceptor groups of the donor molecules, such as C=O and O–H. This study has tried to extract a pattern of the contributions of HBDs
by comparing the structural, spectroscopic, and thermodynamic properties of ChCl-based DESs, which have also been successfully correlated
with the intermolecular interactions.
Published under an exclusive license by AIP Publishing. https://doi.org/10.1063/5.0052569

I. INTRODUCTION a new formulation with a lower melting point.1–3 Eutectic mixtures,


such as ionic liquids (ILs), are prepared by mixing ionic species, and
There is an immense need for finding green and eutectic sol- they have a wide range of solvation properties along with higher
vents due to the emerging global demand of alternative solvents in conductivity and polarity, large heat capacity, and excellent ther-
the last few decades. Generally, an eutectic mixture is represented as mal stability.4–8 Several disadvantages of these solvents have been
a mixture of two components at a certain molar ratio, which gives reported, such as higher cost for bulk applications, low moisture

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-1


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

tolerance, relative toxicity, low biodegradability and recyclability, would be even more difficult to find correlations and major dif-
and requirement of expensive materials to formulate.9 ferences. Unsupervised multivariate analysis, principal component
Deep eutectic solvents (DESs) are the advanced version of ILs, analysis (PCA), can make the complex dataset more simple.24,25 It
which are nontoxic, greener, and biodegradable compared to ionic reduces a large number of variables into a smaller set of variables,
liquids.10,11 They are cheaper to formulate as their production do not still keeping the majority of the important information. It reduces
require any expensive materials. The formulation procedure is quite the chance of overfitting of the model by eliminating features with
simple as well without the need for any delicate instruments.3,12 The high correlation.
synthesis of DESs typically involves mixing of hydrogen bond accep- The structural analysis of the radial distribution function (RDF)
tors (HBAs), such as quaternary ammonium, sulfonium, or phos- based on the snapshots of MD simulation is a common way to study
phonium salts, and complexing agents or hydrogen bond donors DESs to get insights into their microstructure.26 It is an example of
(HBDs). The HBDs almost always form an intermolecular and long a pair correlation function, which describes how density varies as
range hydrogen bond network with salt and thus considerably lower a function of distance from a reference particle.27 Hammond et al.
the melting point of the DES.13 For example, the DES of ChCl recently used empirical potential structure refinement to interpret
and urea (1:2) has a melting point of 12 ○ C, whereas both ChCl neutron diffraction measurements and reported the first experimen-
and urea have significantly higher melting points (302 and 133 ○ C, tally validated RDF for reline.28 RDF analysis could provide the
respectively).14 probability of finding one labeled atom of a HBA to another atom
Due to their mild environmental footprint and huge poten- of a HBD at a certain distance.26 Thus, it could highlight meaning-
tial in several fields, including nanotechnology, nanoscale synthesis, ful reasonings responsible for the formation of the large number of
electrochemistry, biochemistry, biofuel production, and materials nonbonding interactions in a DES.29
science, it is enormously necessary to understand the atomic level Some recent works on DESs involving Pulsed Field Gradient
interactions that occur among the functional groups of the HBA and (PFG) NMR diffusion experiments revealed new insights at a micro-
HBD molecules in DES samples.15,16 Compared with the molecular scopic level. For example, D’Agostino et al. applied PFG NMR to
solvents, the interactions within a DES system are quite complicated study the molecular mobility of three ChCl-based DESs (reline,
because of their formulation using a large variety of components. glyceline, and ethaline) in the aqueous phase in order to eluci-
Thus, the structural characteristics of each DES and interactions date the effect of water on motion and inter-molecular interactions
among the components could be extremely diverse. However, com- among the species in the mixtures, i.e., the Ch+ cation and hydro-
putational methods are able to simulate the molecular structures gen bond donor (HBD).30 In another study, they used PFG NMR
accurately and can analyze their thermodynamic, structural, and to study self-diffusion, ion pairing, and ionic aggregation in 1-ethyl-
spectral properties and reveal their mechanism of formation pre- 3-methylimidazolium-based ionic liquids with the effect of temper-
cisely.17,18 Over the past few years, characterization of the DESs has ature, ion size, and glucose dissolution.31 This approach helped to
been conducted employing common spectroscopic tools, such as answer the question that group 1 metal salts cannot form true eutec-
FTIR spectrometer, nuclear magnetic resonance (NMR) spectrom- tic solvents with glycerol.32 Along with RDF calculations, this type of
eter, and Raman spectrometer. However, there is a lack of detailed dynamic spectroscopic technique could add more realistic insights
work on the correlation of structural, thermochemical, and IR spec- to elucidate a DES structure.
troscopic properties. Two of our recent studies on 1:1 ChCl:Aspirin Although various computational studies are performed on one
and 1:1 Menthol (Men):Acetic Acid (AA) helped to explain how or two DES systems, no comprehensive computational and statisti-
the experimental and computed IR could explain the interactions cal studies are reported on the series of ChCl-based DESs. Keeping
among the components of the DES.19,20 the hydrogen bond acceptors (HBAs) same and varying only the
Density functional theory (DFT)-based structures and their hydrogen bond donors (HBDs) could provide a broader perspective
quantum chemical (QC) calculations and IR spectra have already and deeper understanding of the structures, nonbonding interac-
been established as important tools to investigate the nanostructural tions, and formations of the ChCl-based DESs. In this study, we
properties and noncovalent interactions of DESs, which are com- reported comprehensive molecular dynamics (MD), radial distribu-
parable to experimental results.21,22 Our previous studies showed tion function (RDF) analysis, density functional theory (DFT), and
that computational works can show important structural and spec- principal component analysis (PCA) on nine different ChCl-based
troscopic features of a DES. In addition to our studies, Zhu et al. DESs keeping the HBA (ChCl) constant and varying the HBDs every
investigated experimental vibrational spectra and the formational time. The main aim of this study is to comprehensively investigate
mechanism of three deep eutectic solvents, namely, ChCl:urea, the structural, thermochemical, and spectroscopic features of a large
ChCl:glycerol, and ChCl:acetic acid with a molar ratio of 1:2, 1:1, and number of ChCl-based DES samples to find out a pattern from the
1:1, respectively, and reproduced the spectral data by using DFT.23 extracted information, which could help future researchers to pick
Our latest study on the structure elucidation of the 1:1 Men:AA DES up a suitable HBD for their work.
using mass spectrometry reveals that along with that specific 1:1
Men:AA clusters, there are a lot of other stable clusters possible in
a DES.20 II. METHODOLOGY
Although IR spectra in a higher frequency region provide
important information about the noncovalent interactions of an A. Computational details
eutectic system, the complex fingerprint region is also crucial, but The initial structures of the HBA and HBDs were optimized
the spectral data of this region are sometimes difficult to interpret. in the gas phase using the ωB97XD/6-31++G (d,p) level of the-
When the data are extracted from a large number of samples, it ory employing the Gaussian 09 software package.33 This method

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-2


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

was chosen based on our recently published work on ChCl-based For simulating 1:1 DESs, a molar ratio of 10:10 of ChCl with
DESs, where it was explained elaborately.19 The optimized struc- oxalic acid, malic acid, and D-glucose was used, and for 1:2 DESs,
tures of the HBA and HBD molecules were used in the following a molar ratio of 10:20 was used (Table I) employing the YASARA
molecular dynamics (MD) simulation to generate spontaneous DES Dynamics package.34 The ratios were also selected based on our
structures for further quantum calculations. However, to minimize previous study. An AMBER 1435 force field and the Particle Mesh
time and resource expenses, the simplest and most predominant Ewald (PME) method36 were used for calculating the long-range
cluster structures with the lowest H-bond distances were chosen, electrostatic interactions. All the systems were simulated for 50 ns.
which could be representative of the whole bulk cluster structures of Atom–atom radial distribution functions and the center of mass
each DES. radial distribution functions were also calculated to reveal additional

TABLE I. Types of the studied ChCl-based deep eutectic solvents (DESs) that are selected based on different hydrogen bond donor (HBD) molecules. Tm represents the melting
point of the pure HBDs, and Te represents the eutectic points of the reported DESs, where AT means ambient temperature.

Name of the HBD


Type (based on the HBD) Formula (HBD) (abbreviation) DES (molar ratio) Tm (○ C) Te (○ C) References

1. Urea based Urea (U) ChCl:U (1:2) 134 12 10, 14, and 40

2. Thiourea based Thiourea (TU) ChCl:TU (1:2) 175 69 10, 14, and 40

3. Amino acid based L-proline (Pro) ChCl:Pro (1:2) 221 AT 41

4. Organic acid based Oxalic acid (Ox) ChCl:Ox (1:1) 190 34 2, 10, and 40

5. Organic acid–alcohol based Malic acid (Ma) ChCl:Ma (1:1) 131 −56 42

6. Aromatic alcohol based Phenol (Phe) ChCl:Phe (1:2) 41 −30 40 and 43

7. Aliphatic alcohol based Glycerol (Gly) ChCl:Gly (1:2) 17.8 −40 10 and 44

8. Amide based Acetamide (Ace) ChCl:Ace (1:2) 80 51 10, 14, and 40

9. Sugar based D-glucose (Glu) ChCl:Glu (1:1) 133 31 45

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-3


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

insights into the structures of DESs in the gas phase. The periodic DESs since their constituents are highly hygroscopic. The spectra
boundary conditions (cell box size of 53.54 × 53.54 × 53.54 Å3 ) of the DESs were recorded in the spectral region between 500 and
and a temperature of 298 K were considered using NVT ensem- 4000 cm−1 at ambient temperature with the resolution of 4 cm−1
bles for all the simulations in the gas phase. The default macro- and 50 scans, and the spectra were further analyzed using Origin
script of YASARA Dynamics was used to minimize and equilibrate Lab 2021.38
the systems. The time step of 2.5 fs was used, and the simulation
snapshots were captured at every 100 ps. The most stable cluster
conformers of the DES were isolated depending on the single point III. RESULTS AND DISCUSSION
energy calculation (Table S1) and distances of nonbonding inter- A. Quantum chemical properties
actions using the BIOVIA Discovery Studio software package. The
The Gibbs free energy is one of the most important thermody-
isolated conformers were optimized, and their spectra were calcu-
namic properties for the characterization of a DES system.46 For a
lated using the same level of theory. All the spectral data were scaled
condensed phase system like DESs, the change in Gibbs free energy
using the NIST scaling factor of 0.952.37 The FT-IR spectra and
(ΔG) denotes the free energy of complexation (separate gas phase
principal component analysis of them were drawn and analyzed
molecules → gas phase complex) at a particular temperature and
using the Origin Pro (version 2021) software.38 Furthermore, from
pressure.
the Gaussian output files, thermodynamic properties, including the
The ΔG and ∆H values were calculated using the following
change in Gibbs free energy (ΔG), the change in enthalpy (ΔH),
equation: ∆Xcluster = Xcluster − Xdonor − XChCl , where X is either the
the change in entropy (ΔS), and equilibrium constant (Keq ), were
Gibbs free energy (G) or enthalpy (H). The change in entropy (∆S)
calculated.
was calculated using the following formula: ΔS = ΔH−ΔG , and the
In our study, crucial experimental solvatochromic properties, T
equilibrium constant (Keq ) was calculated using the following for-
such as hydrogen bond donor acidity and hydrogen bond accep- −ΔG
tor basicity, have been considered to make the correlation analysis mula: Keq = e RT , where R and T stand for the universal gas constant
−1 −1
more realistic. PCA is a dimensionality-reduction method, which (8.31 J mol K ) and the temperature of 298.15 K, respectively.
can reduce the dimensionality of large datasets, such as spectral The calculation of ΔG values of the nine studied DES sys-
data of a large number of compounds, by identifying their princi- tems reveals that ChCl:TU and ChCl:U form a DES at a ratio of
pal source of variation.39 PCA decomposes the multivariate response 1:2 as they have higher ΔG values: −98.75 kJ/mol for ChCl:TU and
arranged in an X matrix into a product of two new matrices as indi- −84.005 kJ/mol for ChCl:U. The corresponding equilibrium con-
cated by the following equation: X = T k Pk T + E, where Tk represents stant (Keq ) value of ChCl:TU (1.99 × 1017 ) is higher than that of
how samples are related to each other, Pk signifies how variables ChCl:U (5.22 × 1014 ) as demonstrated in Table II. For both urea
are related to each other, k is the number of factors included in the and thiourea, the H-bond donor atom count is generally two com-
model, and E is the matrix of residuals, which contains the infor- prising the two NH2 groups50,51 (Table II). These two NH2 groups
mation not retained by the model. PCA was applied in the study to show a very strong N–H⋅ ⋅ ⋅Cl− hydrogen bond with the shortest
correlate and distinguish the spectral data of the HBD molecules, H⋅ ⋅ ⋅Cl− distance [Figs. 2(a) and 2(b)]. In addition, C=O⋅ ⋅ ⋅H–C and
HBA molecule (ChCl), and DESs. C=O⋅ ⋅ ⋅H–O hydrogen bonds between the C=O group of urea and
the –CH3 and –OH groups of Ch+ , respectively, help to stabilize the
DES [Fig. 2(a)].51 Experimentally, ChCl:U is found to form more
B. Materials, preparation, and experimental spontaneously than ChCl:TU at room temperature since the eutec-
IR spectroscopy tic point of ChCl:U is 12 ○ C and that of ChCl:TU is 69 ○ C (Table I).
Choline chloride (ChCl), urea (U), and oxalic acid (Ox) were Both urea and thiourea can often form hydrogen bonded polymeric
purchased from Acros Organics (NJ, USA), EMD chemicals (NJ, networks. Such a type of network can have two types of arrange-
USA), and Spectrum (NJ, USA), respectively, and all the chemi- ments; the α-tape, which is linear and corrugated in shape, and the
cals have purity over 99.0%. Two of nine studied DESs, namely, β-tape, which is zigzag in shape.52 For urea, both the α-tape and
1:1 ChCl:U and 1:2 ChCl:Ox, were experimentally formulated by β-tape commonly occur, but for thiourea, only the β-tape is predom-
constant stirring (≤600 rpm) at a temperature in the range of inantly observed.52–55 In the α-tape, two hydrogen atoms are on the
40–50 ○ C and under atmospheric pressure for 3 h using a cov- opposite side of the tape, and in the β-tape, they are on the same
ered beaker until clear liquids were obtained. In this study, the IR side.52 This makes the zigzag β-tape of thiourea form more H-bonds
spectra of all the studied DESs and their pure constituents were than urea, which shows both α and β-tape arrangements. This can be
computed using the optimized structures to get insights into the further seen in Table II, where the total number of H-bond counts
structures of DESs in addition to their thermochemical proper- in the static structure of ChCl:U is 19 and that of ChCl:TU is 21. The
ties as computed IR can produce comparable spectra to experi- existence of the β-tape arrangement of thiourea molecules among
mental IR spectra.37 However, the experimental spectra of only themselves in the MD simulation structures might help to produce a
these two experimentally formulated DESs (one solvent was selected more stable static structure of ChCl:TU compared to ChCl:U. Being
from the 1:1 DES group and another was chosen from the 1:2 the control DES of this study, the interactions and thermochemical
DES group) were collected to prove the good degree of agreement properties of 1:2 ChCl:U obtained can be compared to the interac-
between the experimental and computed IR spectral data of this tions explained in several other journals.14,51,56–58 One well explained
study. A Shimadzu IRAffinity-1S Fourier Transform Infrared Spec- work by the group of Ashworth56 showed that urea forms the most
trophotometer was used for the analyses of the IR spectra of the stable DES with ChCl at a ratio of 1:2, where the urea–Cl non-
DES samples. All the spectra were collected using freshly prepared bonding interaction added a maximum value (−59 kJ/mol) to the

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-4


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

TABLE II. Thermochemical properties, such as the change in Gibbs free energy (ΔG), the change in enthalpy (ΔH), the change in entropy (ΔS), and equilibrium constant (Keq )
of the nine studied DESs. In addition, the hydrogen bond donor (HBD) and acceptor (HBA) atom count and the HBD-acidity and HBA-basicity values of them are collected from
the literature, which are crucial values for a DES. The total H-bond count calculated from the static cluster structure of the DES is also shown. NA = not available.14,16,47–49

H-bond H-bond
donor atom acceptor H-bond count
ΔG ΔH ΔS count of atom count HBD-acidity HBA-basicity of the static
DES Ratio (kJ/mol) (kJ/mol) (kJ/mol) Keq the HBD of the HBD value value structure

ChCl:Ox (1:1) −6.758 04 −48.107 04 −0.138 69 1.53 × 101 2 4 1.19 0.27 5


ChCl:Ma (1:1) −49.228 13 −103.783 40 −0.182 98 4.22 × 108 3 5 1.49 0.42 9
ChCl:Glu (1:1) −71.560 63 −127.194 98 −0.186 60 3.45 × 1012 5 6 NA NA 11
ChCl:U (1:2) −84.005 50 −173.661 09 −0.301 93 5.22 × 1014 2 1 1.42 0.5 19
ChCl:TU (1:2) −98.747 69 −196.319 15 −0.328 19 1.99 × 1017 2 1 1.41 0.497 21
ChCl:Pro (1:2) −29.896 57 −124.519 60 −0.317 69 1.73 × 105 2 3 NA NA 15
ChCl:Gly (1:2) −57.199 15 −169.363 14 −0.375 45 1.05 × 1010 3 3 1.49 0.52 16
ChCl:Phe (1:2) −31.624 15 −126.811 66 −0.319 26 3.48 × 105 1 1 NA NA 8
ChCl:Ace (1:2) −48.608 51 −132.968 46 −0.283 55 3.28 × 1008 1 1 1.79 1.06 16

total ΔG value of the cluster. Although in that study, the method H-bond acceptor counts (five –OH groups and one hetero O group)
of calculation was different, the final conclusion matches our of which five –OH groups also act as donors (Table II), which might
findings. have some effect on the H-bond donating efficiency of glucose. Due
ChCl:Glu comes next in order of stability of complexation with to this, despite having five H-bond donor counts, ChCl:Glu appears
a ΔG value of −71.56 kJ/mol and a Keq value of 3.45 × 1012 at a to be less stable than either ChCl:TU or ChCl:U (which have two
ratio of 1:1 (Table II). Glucose having a total of five –OH groups H-bond donor counts) according to their ΔG value and Keq value
as H-bond donor atoms can form a much stable DES with the help differences.
of forming several H-bonds with the Cl− group of ChCl, such as The other two 1:1 DESs are ChCl:Ox and ChCl:Ma with the ΔG
O–H⋅ ⋅ ⋅Cl− [Fig. 1(c)]. In addition, the O of the –OH group of glu- values of −6.76 and −49.23 kJ/mol, respectively. Their Keq values are
cose and the hetero O of the cyclic glucose chain can also form 1.53 × 101 and 4.22 × 108 , respectively (Table II). For both cases,
several hydrogen bonds with Ch+ as the acceptor atom, such as the cluster formation is stabilized via strong O–H⋅ ⋅ ⋅Cl− hydrogen
O–HCh+ ⋅ ⋅ ⋅O–Hglucose , C–HCh+ ⋅ ⋅ ⋅O–Hglucose , and C–HCh+ ⋅ ⋅ ⋅Oglucose bonding [Figs. 1(a) and 1(b)]. ChCl:Ox appears to be the least stable
[Fig. 1(c)]. A total of 11 H-bonds were calculated in the static struc- of all the nine DESs we have analyzed. The reason behind it could be
ture of ChCl:Glu, which is the maximum number of H-bonds among attributed to the fact that the two H-bond donor groups of oxalic
all the reported 1:1 clusters. It is further observed that glucose has six acid (–OH) are intramolecularly H-bonded with its two H-bond

FIG. 1. Optimized structures of three representative simple cluster structures of 1:1 DESs, which are (a) ChCl:Ox, (b) ChCl:Ma, and (c) ChCl:Glu retrieved from MD
simulations. All the structures are computed at the ωB97XD/6-31++G(d,p) level of theory. Colors of the atoms are as follows: hydrogen atoms = silver, carbon atoms =
gray, nitrogen atoms = blue, oxygen atoms = red, and chlorine atoms = green.

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-5


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

acceptor groups (C=O), such as O–H⋅ ⋅ ⋅O=C [Fig. 1(a)]. Similar to [Fig. 2(d)]. This –OH group may also act as the H-bond accep-
oxalic acid, malic acid also has intramolecular H-bonds among its tor group, such as H–Ophenol ⋅ ⋅ ⋅H–OCh+ and H–Ophenol ⋅ ⋅ ⋅H–CCh+
donor and acceptor groups but of weaker strength due to the longer [Fig. 2(d)]. Following the same probable explanation about the sta-
carbon chain [Fig. 1(b)]. The intramolecular hydrogen bond might bility of ChCl:Glu and also the total H-bond counts of static struc-
weaken the H-bond acidity of the donor groups.59 Further obser- tures of the respective clusters being 16 for ChCl:Gly and eight for
vation of the static structures of the clusters reveals that a total of ChCl:Phe, it can be predicted that ChCl:Gly might form a more sta-
five H-bonds in ChCl:Ox and a total of nine H-bonds in ChCl:Ma ble cluster than ChCl:Phe, which is also in accordance with their
are formed. Probably, due to these reasons, ChCl:Ma forms a more ΔG and Keq values. ChCl:Ace has a ΔG value of −48.61 kJ/mol and
stable DES. a Keq value of 3.28 × 108 while forming a DES at a ratio of 1:2
ChCl:Gly and ChCl:Phe form stable 1:2 clusters with the ΔG (Table II). These values fall in between the ΔG and Keq values of
values of −57.19 and −31.62 kJ/mol, respectively. Their Keq val- ChCl:Gly and ChCl:Phe. A similar explanation for ChCl:Glu stabil-
ues are 1.05 × 1010 and 3.48 × 105 , respectively, which show that ity could be applied here as well. Acetamide has one H-bond donor
ChCl:Gly is formed more spontaneously than ChCl:Phe (Table II). group forming a N–H⋅ ⋅ ⋅Cl− hydrogen bond with ChCl [Fig. 2(f)]
Glycerol has three –OH groups, which show strong hydrogen and one H-bond acceptor group (C=O). The structural features
bond interactions with ChCl via O–H⋅ ⋅ ⋅Cl− hydrogen bonding thus can be related to phenol. However, in phenol, the –OH group
[Fig. 2(c)]. These three –OH groups might also work as acceptor may act as both a donor and an acceptor, whereas in acetamide,
groups via several H-bonding, such as H–Oglycerol ⋅ ⋅ ⋅H–OCh+ and –NH2 acts as a donor and C=O acts as an acceptor. In addition, the
H–Oglycerol ⋅ ⋅ ⋅H–CCh+ [Fig. 2(c)]. Again, phenol has only one –OH total number of H-bonds in the static structure of ChCl:Ace is 16,
group showing strong O–H⋅ ⋅ ⋅Cl− hydrogen bonding with ChCl which is more than that of ChCl:Phe (eight) but similar to that of

FIG. 2. Optimized structures of six 1:2 representative simple cluster structures of 1:2 DESs, namely, (a) ChCl:U, (b) ChCl:TU, (c) ChCl:Gly, (d) ChCl:Phe, (e) ChCl:Pro, and
(f) ChCl:Ace retrieved from MD simulations. All the structures are computed at the ωB97XD/6-31++G(d,p) level of theory. Colors of the atoms are as follows: hydrogen
atoms = silver, carbon atoms = gray, nitrogen atoms = blue, oxygen atoms = red, sulfur atoms = yellow, and chlorine atoms = green.

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-6


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

ChCl:Gly (16). These could be the probable reasons due to which


ChCl:Ace forms a more stable DES than ChCl:Phe but less than
ChCl:Gly.
The pattern that we obtained here can also be correlated with
one of our recent works where the stability of 1:1 ChCl:Aspirin was
analyzed.37 In aspirin, the H-bond donor count is one (–OH) and
the H-bond acceptor count is four (–OH, two C=O, and C–O).
With aspirin being an aromatic compound, this cluster can be
related to ChCl:Phenol of our study. Similar to phenol, aspirin also
has one –OH group as the H-bond donor, which forms strong
O–H⋅ ⋅ ⋅Cl− interaction. However, unlike phenol, the –OH group
can function as an acceptor group as well, and aspirin has four
H-bond acceptor groups, of which only –OH can work as both
acceptor and donor groups and the other three work as accep-
tors only. These three acceptor groups form nonbonding interac-
tions with Ch+ , such as C=Oaspirin ⋅ ⋅ ⋅H–OCh+ , C=Oaspirin ⋅ ⋅ ⋅H–CCh+ ,
C–Oaspirin ⋅ ⋅ ⋅O–HCh+ , and C–Oaspirin ⋅ ⋅ ⋅H–CCh+ . Therefore, the ΔG
value of the most stable cluster of ChCl:Aspirin (−47.23 kJ/mol)37
is slightly higher than that of ChCl:Phenol (−31.62 kJ/mol)
(Table II).
ChCl:Pro is the only amino acid-based DES of our study, where FIG. 3. Score plots from the principal component analysis (PCA) of the IR spec-
the donor molecule is L-proline, a chiral molecule. It forms at a tral data (400–4000 cm−1 ) of ChCl (black), ChCl:Ox (pink), ChCl:Ma (aqua),
molar ratio of 1:2 with a ΔG value of −29.9 kJ/mol and a Keq value of ChCl:Glu (green), ChCl:U (purple), ChCl:TU (navy), ChCl:Pro (olive), ChCl:Gly
1.73 × 105 (Table II). Here, the H-bond donor count is two compris- (blue), ChCl:Phe (yellow), and ChCl:Ace (red).
ing one –OH group and one –NH group, and the total number of
H-bonds in its static structure is 15. They form strong intermolecu-
lar interactions with ChCl via O–H⋅ ⋅ ⋅Cl− and N–H⋅ ⋅ ⋅Cl− hydrogen
bonding [Fig. 2(e)]. The acceptor count here is three, which com- For further analysis, two more PCAs are conducted on the spec-
prises –OH, –NH, and C=O groups. It is the second least stable DES tra of 1:2 [Fig. 4(b)] and 1:1 [Fig. 5(b)] DESs separately, keeping
among all. However, being a completely different donor molecule ChCl as a relative control for both cases. Comparing the IR spectral
(chemically and physically) from the rest eight HBDs, its stability data of 1:2 [Fig. 4(a)] and 1:1 [Fig. 5(a)] clusters with their respective
could not be explained by any of the patterns predicted for the other PCA [Figs. 4(b) and 5(b)], it could be further correlated that which
eight clusters. components are responsible for the strong H-bond interactions and
thereby the stable formation of the DES clusters.
B. FTIR spectroscopy and PCA Of the six 1:2 cluster structures, the relative shifts of the
Computations of all the spectra (Fig. S2) of ChCl, HBDs, and component’s peaks in the spectra of the DESs indicate which HBD
DESs are performed, and PCA is conducted on the spectral data contributes more strongly for the formation of the hydrogen bonds.
to get a pattern and correlation (Fig. 3). PCA is performed on the For this analysis, some prominent peaks of ChCl and its corre-
400–4000 cm−1 spectral data of all the IR spectra of the DES sam- sponding DESs on its IR spectra are pointed out (Fig. 4 and Fig.
ples and ChCl. The first principal component, denoted by PC1, S2). The PCA loading plot is also shown in the supplementary
explains the highest possible variability. In Fig. 3, the PC1 of IR material to simplify the comparison (Fig. S1). These peaks are at
spectra explains 24.2% of the total spectral data. The second prin- 3746 cm−1 (O–H stretching), 1075 cm−1 (C–O stretching), and
cipal component, PC2, is accounted for the second largest variance 946 cm−1 (C–N stretching). Then, by analyzing the IR data of the
of the spectral data, and it explains 15.2% of the total spectral data. 1:2 DES clusters, the approximate relative percent shift of these
Both PC1 and PC2 are obtained from the eigenvalues of the scree prominent peaks is calculated. The more the shift of the func-
plot, which help to determine the highest variance of the dataset. tional group vibration of ChCl, the stronger the H-bond with the
Therefore, PC1 and PC2 together illustrate the distribution of the donor. From the PCA data [Fig. 4(b)], it is observed that the
IR spectra of the pure component (ChCl) and the nine DES clus- scores of ChCl:TU and ChCl:Pro reside separately from others. For
ters over 39% on the two PC score axes. Donor components were ChCl:TU, a prominent peak for the O–H stretching of ChCl at
omitted to simplify the analysis. 3319 cm−1 denotes strong interaction with a shift of −11%. For
Comparing the score plots in Fig. 3, it could be assumed that L-proline, O–H stretching at 3605 cm−1 shows a shift of −4%.
1:1 and 1:2 clusters behave quite differently from each other. How- Though O–H vibration has great disparity in relative shift values in
ever, one 1:1 cluster (ChCl:Glu) behaves somewhat like 1:2 clusters the two clusters, the shifts in C–O (stretching) and C–N (stretch-
as its score in the plot is located around the scores of 1:2 clusters ing) appear closer in magnitude. For ChCl:TU, the C–O (stretching)
(Fig. 3). This might happen because of a large number of H-bond and C–N (stretching) of ChCl are found at 1103 and 920 cm−1 ,
donor counts (five –OH groups) present in the glucose structure and showing shifts of 2.6% and −2.8%, respectively. Similar peaks of
the total number of H-bonds in its static structure is found to be 11 ChCl:Pro appear at 1109 and 988 cm−1 with shifts of 3.2% and 4.4%,
(Table II). respectively.

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-7


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

FIG. 4. (a) Computed IR spectra of ChCl, ChCl:U, ChCl:TU, ChCl:Pro, ChCl:Gly, ChCl:Phe, and ChCl:Ace (all the 1:2 DESs) and (b) score plots from the principal component
analysis (PCA) of the IR spectral data (400–4000 cm−1 ) of ChCl (black), ChCl:U (yellow), ChCl:Pro (aqua), ChCl:Gly (green), ChCl:Phe (blue), and ChCl:Ace (red).

Figure 4(b) shows that the scores of the spectra of ChCl:Gly and 944 cm−1 , showing a shift of −0.21%, and for ChCl:Ace, it appears at
ChCl:U DESs behave quite similarly with some very good H-bond 924 cm−1 , showing a shift of −2.4%. The C–O (stretching) of ChCl
interactions. The total number of H-bonds in their static structures for ChCl:Phe appears at 1069 cm−1 , showing a shift of −0.56%, and
also appear to be very close: 19 H-bonds for ChCl:U and 16 H-bonds for ChCl:Ace, it appears at 1099 cm−1 , showing a shift of 2.21%.
for ChCl:Gly (Table II). The IR data corroborates this information Thus, it can be predicted that the overall H-bond interaction in
as well. For ChCl:U, the prominent O–H (stretching) vibration of ChCl:Ace is stronger than in ChCl:Phe, which is mirrored in the
ChCl is found at 3476 cm−1 with a shift of −7%, and for ChCl:Gly, total count of H-bonds in their static structures as well: 16 H-bonds
it is found at 3510 cm−1 with a shift of −6.3%. The peaks of C–O in ChCl:Ace and eight H-bonds in ChCl:Phe (Table II).
(stretching) for ChCl:U and ChCl:Gly are found at 1118 cm−1 with Of the three 1:1 DES cluster structures, ChCl:Ox and ChCl:Ma
a shift of 3.9% and at 1097 cm−1 with a shift of 2.04%. The peaks of appear to behave quite differently from each other according to the
C–N (stretching) for ChCl:U and ChCl:Gly are found at 905 cm−1 PCA score plot shown in Fig. 5(b). However, from the PCA of all
with a shift of −4.3% and at 922 cm−1 with a shift of −2.6% shift nine DESs (Fig. 3), it is observed that ChCl:Glu behaves somewhat
compared to the C–N peak position of ChCl. like the 1:2 clusters. The probable reason behind it is the five –OH
The PCA [Fig. 4(b)] further shows that the scores of ChCl:Ace groups available for the formation of H-bonds with ChCl. A total of
and ChCl:Phe clusters are located around the score of ChCl along 11 H-bonds can be counted from the static structure of ChCl:Glu,
with ChCl:Gly and ChCl:U, inferring that most of the 1:2 clusters which is the maximum number of H-bonds counted among all the
behave in similar ways. From the IR data, it is found that there is three 1:1 DESs; for ChCl:Ox, the total number of H-bonds in the
only a shift of −0.03% of O–H (stretching) of ChCl for ChCl:Ace static structure is five, and for ChCl:Ma, it is nine (Table II). From
appearing at 3745 cm−1 , and for ChCl:Phe, the respective shift is the IR analysis, this can be further confirmed. For ChCl:Glu, the
−0.3% appearing at 3735 cm−1 . It might appear that phenol binds O–H (stretching) peak of ChCl appears at 3743 cm−1 , showing a
strongly with ChCl than acetamide. However, by observing their shift of −0.08%. The C–N (stretching) and C–O (stretching) peaks
shifts at the C–O (stretching) and C–N (stretching) of ChCl, a dif- of ChCl in ChCl:Glu are found at 922 and 1080 cm−1 , showing shifts
ferent scenario appears. For ChCl:Phe, C–N (stretching) appears at of −2.54% and 0.42%, respectively.

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-8


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

FIG. 5. (a) Computed IR spectra of ChCl and all the 1:1 DESs and (b) score plots from the principal component analysis (PCA) of the IR spectral data (400–4000 cm−1 ) of
the three 1:1 DESs and the pure component ChCl.

FIG. 6. (a) Experimental FTIR spectrum of 1:2 ChCl:U and (b) computed IR spectrum of 1:2 ChCl:U (scaled by a factor of 0.952, NIST database37 ) optimized at the
ωB97XD/6-31++G (d,p) level of theory.

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-9


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

FIG. 7. (a) Experimental FTIR spectrum of 1:1 ChCl:Ox and (b) computed FTIR spectrum of 1:1 ChCl:Ox (scaled by a factor of 0.952, NIST database37 ) optimized at the
ωB97XD/6-31++G (d,p) level of theory.

The O–H (stretching) vibration of ChCl for ChCl:Ox is found computational data to reality is necessary. IR spectral analysis is
at 3590 cm−1 , and for ChCl:Ma, it is found at 3630 cm−1 ; both an excellent approach to understand how and to what extend the
shift by −4.2% and −3.1%, respectively. For ChCl in ChCl:Ma, the intermolecular interactions within the DESs affect their stability. So,
stretching vibration of C–N is observed at 901 cm−1 with a shift the computed IR of one 1:1 cluster (ChCl:Ox) and one 1:2 cluster
of −4.8%. However, for ChCl:Ox, the same vibration (C–N stretch- (ChCl:U) has been compared to their corresponding experimental
ing) is observed at 933 cm−1 , showing a shift of −1.4%. Finally, the IR to analyze their degree of agreement.
C–O stretching vibration of ChCl is found to be slightly greater The computed IR spectrum of 1:2 ChCl:U in Fig. 6(b) shows
in ChCl:Ox at 1110 cm−1 with a shift of 3.3% and in ChCl:Ma at two peaks at 1434 cm−1 for the C–N stretching vibration of ChCl
1097 cm−1 with a shift of 2.1%. and 1599 cm−1 for the N–H bending vibration of urea. In addi-
The IR spectral analysis of the pure components vs the clus- tion, a strong peak at 1682 cm−1 for the C=O stretching vibration
ters of DESs is an excellent approach to understand how and to of urea is also observed. These are similar to the peaks observed
what extent the intermolecular interactions affect the formation of in the experimental IR spectrum of 1:2 ChCl:U in Fig. 6(a), where
DESs, which could help to explain the changes in the free energies strong peaks are observed at 1445, 1610, and 1666 cm−1 , respectively.
of complexation. Since the HBA is kept the same in all the stud- The computed spectrum also shows strong peaks in the range of
ied DESs, this approach has provided a comparative overview of the 3120–3500 cm−1 , which are generally separated from the rest of the
contribution of the studied HBDs. peaks on the spectrum. These peaks are highly likely due to the N–H
stretching vibration (3120 cm−1 ) of urea and O–H stretching vibra-
C. Comparison of computed IR with experimental IR tions (3334, 3470, and 3568 cm−1 ) of ChCl. The experimental IR
As the study focuses on extracting a pattern describing spectrum shows broad peaks in a similar range of 3100–3500 cm−1 .
the relative stability of the nine DESs, the resemblance of the The reason behind these broadened peaks could be explained by the

TABLE III. Experimental spectral frequencies of IR of some of the major functional groups of pure ChCl, frequencies for the
same bonds in ChCl:U and ChCl:Ox, and their peak shifts calculated in percent (%) change.

Frequency (cm−1 ) Percent (%) change


Type ChCl ChCl:U ChCl:Ox ChCl:U ChCl:Ox
C–N stretching vibration 945–980 925–960 868–1033 −2.08 −1.25
C–O stretching vibration 1056–1088 1050–1165 1083–1153 3.31 4.29
O–H stretching vibration 3100–3750 3100–3550 3100–3600 −2.92 −2.19

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-10


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

ChCl:Ace ChCl:U ChCl:TU ChCl:Pro ChCl:Gly ChCl:Phe ChCl:Ace


TABLE IV. Computed spectral frequencies of IR of the major functional groups of pure ChCl, frequencies for the same bonds in the 1:2 DESs, and their peak shifts calculated in percent (%) change. St. means
long-range interactions and formation of the hydrogen bonds with

−2.36

2.21

1.64

1.55

0.71

−0.03
the O–H functional group of ChCl. In addition, a large number of
other interactions in different stable clusters, such as ChCl–ChCl,
ChCl-urea, ChCl-2urea, and ChCl-3urea, could exist in the same
DES system.20 Since the computed IR has demonstrated only the

−0.21

−0.56

−0.72

1.50

0.50

−0.29
interactions among the components of one 1:2 ChCl:U cluster struc-
ture, it generates specific peaks of specific functional groups with-
out any bias. However, this comparison proves a good degree of

Percent change (%)


agreement between the spectra of 1:2 ChCl:U generated from the-

−2.57

2.04

−0.84

2.23

0.90

−6.29
oretical and experimental IR. Similar comparison has been explored
for a 1:1 DES selected from the studied 1:1 DESs in the following
paragraph.
The computed IR spectrum of 1:1 ChCl:Ox in Fig. 7(b) shows

4.38

3.19

2.86

2.21

0.90

−3.76
strong peaks at 1150 and 1293 cm−1 . These peaks get merged in
the experimental spectra and show a strong and broadened peak in
the range of 1150–1270 cm−1 in Fig. 7(a). These peaks are highly

−2.82

2.57

1.18

1.45

−0.10

−11.39
likely due to the stretching vibrations of the two C–O groups in
oxalic acid, and peak broadening occurs because of the formation of
long-range hydrogen bonds with the C–O group. In the computed
spectra, a number of weak peaks of hydrocarbons are observed

−4.33

3.94

0.53

0.79

2990–3054 2955–3069 2962–3075 3019–3078 3003–3094 3008–3065 3012–3073 −0.30

−7.22
around 1500 cm−1 , while these peaks get merged and appear as
a strong peak at 1480 cm−1 in the experimental spectrum. The
peak for the C=O functionality appears in the similar region

1049–1102 1097–1139 1088–1118 1091–1128 1075–1120 1020–1119 1077–1121

1230–1269 1187–1325 1219–1310 1253–1318 1216–1262 1219–1262 1214–1326

2821–2952 2871–2947 2911–2945 2949–2951 2941–2961 2913–2946 2913–2949


856–992
(1730–1810 cm−1 ) in both the computed and experimental spectra

3745
of 1:1 ChCl:Ox, and as usual, due to the long-range interactions, the
experimental peak for C=O gets broadened. Peaks corresponding to
the stretching vibrations of hydrocarbon and –OH are also observed
ChCl:Phe

941–947
in the range of 2700–3600 cm−1 . Overall, the computed spectra of

3735
the 1:1 DES show a good degree of agreement.
From Table III, the changes in the peak position (in percent)
of some major functional groups of ChCl (C–N stretching vibra-

3471–3550
tion, C–O stretching vibration, and O–H stretching vibration) from
ChCl:Gly

850–994

a pure component to a DES can be observed. The stack plots of the


pure components, ChCl, urea, and oxalic acid and their correspond-
Frequency (cm−1 )

ing DESs (e.g., ChCl:U and ChCl:Ox) show the peak changes (Fig.
S3). These peak shift values can be compared to the respective com-
942–1034
ChCl:Pro

puted values in Table IV (ChCl: U) and Table V (ChCl:Ox). The

3605
computed peak shift for the C–N stretching vibration of ChCl in
ChCl:U is −4.33%, and the experimental peak shift is −2.08%. For
ChCl:Ox, the computed and experimental peak shifts for the same
3471–3481 3280–3359
ChCl:TU

spectra are −1.38% and −1.25%, respectively. The computed and


853–986

experimental peak shifts for the C–O stretching vibration of ChCl


in ChCl:U are 3.94% and 3.31%, respectively, and for ChCl:Ox, they
are 3.25% and 4.29%, respectively. The computed and experimen-
tal peak shifts for the O–H stretching vibration of ChCl in ChCl:U
855–956
ChCl:U

are −7.22% and −2.92%, and for ChCl:Ox, they are −4.16% and
−2.19%, respectively. Overall, it has been observed that not only the
computed spectra resemble the reality but also the peak shifts of
the prominent functional groups are comparable to the peak shifts
917–975

calculated from the experimental spectra.


ChCl

3746

D. Radial distribution function analysis by molecular


dynamics simulation
asymmetric st.
symmetric st.

The analysis of the radial distribution function (RDF) is a very


common way to study DESs to get insights into their microstruc-
C–H (sp3 )

C–H (sp3 )
vibration

vibration

vibration

vibration
stretching.

O–H st.
C–O st.
C–N st.

C–N st.

tures. It is an example of a pair correlation function, which describes


Type

how density varies as a function of distance from a reference


particle.27 The probability of finding any atom or moiety at a certain

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-11


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

TABLE V. Computed spectral frequencies of IR of the major functional groups of pure ChCl, frequencies for the same bonds in the 1:1 DESs, and their peak shifts calculated in
percent change.

Frequency (cm−1 ) Percent (%) change


Type ChCl ChCl:Ox ChCl:Ma ChCl:Glu ChCl:Ox ChCl:Ma ChCl:Glu
C–N stretching vibration 917–975 864–1002 851–952 862–982 −1.374 −4.704 −2.537
C–O stretching vibration 1049–1102 1093–1128 1073–1122 1078–1082 3.254 2.045 0.418
C–N stretching vibration 1230–1269 1250–1261 1223–1267 1237–1264 0.480 −0.360 0.080
C–H (sp3 ) symmetric stretching 2821–2952 2862–2947 2905–2948 2916–2951 0.624 1.385 1.628
C–H (sp3 ) asymmetric stretching 2990–3054 2969–3077 2965–3086 2964–3074 0.033 0.116 −0.099
O–H stretching vibration 3746 3562–3618 3601–3660 3741–3745 −4.164 −3.083 −0.080

distance from the reference particle compared to the statistical distri- 2.4 and 2.6 Å, which is in the ideal bonding distance in the case of
bution can be investigated by employing RDF analysis on the snap- conventional or non-conventional hydrogen bonding (H-bond).
shots of MD simulations.60 To elicit the pattern of the microstruc- However, there is a significant difference in the peak intensity, indi-
tures forming in nine DES systems, the analysis of atom–atom RDFs cating the probability of this specific interaction. In the case of
and the center of mass RDFs was carried out for all the systems. The thiourea, sulfur shows remarkably the highest peak intensity among
atom–atom RDFs between the most electronegative oxygen atoms all other structures at 2.6 Å, signifying the strong occurrences of H-
(specifically hydroxylic oxygen atoms or a statistical mean of their bonds between the sulfur and hydrogen atoms of Ch+ . The IR analy-
identical cases) or sulfur atoms (in the case of thiourea), and the sis also shows the highest percentage of changes in the O–H stretch-
hydrogen atoms of Ch+ are analyzed and represented in Fig. 8(a). ing vibrational frequency of ChCl from a pure component to a DES
According to the spectral evidences retrieved from RDF analysis, it in ChCl:TU. The RDF spectral pattern for urea and acetamide shows
is observed that the highest peak in all the cases remain in between almost the same kind of distribution, which apparently marks that

FIG. 8. (a) Atom–atom RDFs showing the probability of hydrogen atoms of Ch+ around oxygen atoms (hydroxylic) or sulfur (thiourea) atoms of HBD molecules and (b) the
center of mass RDFs of Cl− and different HBD molecules.

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-12


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

they have similarities in atomic level structural distribution. This is a 1:2 DES. By monitoring their structures, vibrational spectra, and
probably because of their structural similarities. The spectral behav- interactions and the strength of the interactions, a coherent explana-
ior has similarities among L-proline, oxalic acid, and malic acid, tion of their stability was tried to be established. A particular pattern
which is obvious as they all have oxygen atoms in carboxylic acid has been obtained, where (1) intramolecular H-bonding within the
form. The scenario gets totally changed when it comes to glucose. It donor molecule weakens nonbonding interactions with ChCl and
shows almost flattened peaks, which could be for the formation of thus decreases stability,59 (2) an increased number of H-bond donat-
intra-molecular H-bonds. ing groups in a donor molecule creates more viable clusters, and (3)
To observe the structural differences, the analyses of the center in a particular donor molecule, the hydrogen bond with different
of mass RDFs between Cl− and the donor molecules were carried donating and accepting groups provides more viable stability. Spec-
out and are illustrated in Fig. 8(b). These analyses show a totally tral analysis further revealed that the –OH functional group of Ch+
different pattern from the earlier analysis. The highest peaks for participates in the DES formation actively by hydrogen bonding with
all cases are in the range of 2.2–2.4 Å, indicating the contribution the acceptor groups of the donor molecules, such as O–H⋅ ⋅ ⋅O=C,
of Cl− as a H-bond acceptor. The peak intensity differs from each O–H⋅ ⋅ ⋅H–O, and O–H⋅ ⋅ ⋅N–H. Because in most cases, its spectral
other. Oxalic acid shows the highest peak among the three cases data showed a maximum shift after cluster formation (a maximum
of 1:1 DESs, whereas the HBD, thiourea, shows the highest peak of 11.39% shift was found in the case of ChCl:TU). This information
among the 1:2 DESs. The probable reason lies in their structures. is also supported by RDF analysis carried out by MD simulation. The
All the hydrogen atoms of oxalic acid molecules are directly con- RDF analysis further revealed that several hydrogen bonds could
nected to the highly negative oxygen atoms and make the hydrogen be formed among ChCl and the donor molecules, specially between
atoms highly electropositive. In the case of thiourea, all the hydro- –OH or –NH2 groups of several donor molecules and the Cl− group
gen atoms are covalently bonded to nitrogen and the presence of of ChCl. The chloride (Cl− ) ion can act as a charge transfer bridge
a sulfur atom holds the two nitrogen atoms tightly. In contrast, between Ch+ and the donor. The insights into the study would pro-
L-proline and phenol show low peak intensity, which signifies a vide a broader perspective and deeper understanding in the field of
lower number of interactions with the Cl− atoms. The other HBDs DES research. It will also help to explain nonbonding interactions of
of 1:2 composition DESs, such as urea, glycerol, and acetamide, show other similar structured DESs.
similar spectral patterns, which means that they maintain a sim-
ilar type of microstructure interacting with Cl− atoms. The three
RDFs for three HBDs from the 1:1 ratio DES family differ from each SUPPLEMENTARY MATERIAL
other in intensity, indicating different patterns of atomic level inter- See the supplementary material for the single point energy val-
actions with Cl− ions. Glucose shows the lowest intensity, which ues of different structures in DESs, the IR frequency and peak shift of
may be due to the presence of six oxygen atoms in each glucose donor and acceptor molecules of ChCl, the thermodynamic data for
molecule. explaining their physical property, the IR plot from DFT calculation,
Analyzing two different sets of RDFs, the bonding features sug- and PCA and its corresponding loading plot.
gest that thiourea could be the most thermodynamically stable DES
among all of them. Because it forms the highest number of H-bonds
with Cl− , which are basically the noncovalent interactions originat- AUTHORS’ CONTRIBUTIONS
ing from the charge transfer process. Therefore, it promotes easier M.A.H. supervised the whole study. M.S.R. designed the meth-
formation of DESs, although this DES is formed at higher tem- ods and processes. He also generated the PCA plot, contributed to
perature (69 ○ C) and the rest of the studied DESs are room tem- the writing part, and revised the paper. M.S. completed the RDF data
perature DESs.16 The reason behind it can be speculated from the analysis and write-up. M.A.A. conducted the molecular dynamics of
fact that thiourea only forms stable DESs with ChCl in its β-tape all the DESs. M.I.R. and H.I. have contributed equally. They con-
conformation, unlike other donors that do not have such conforma- ducted the structure optimization of all donors, ChCl, and DESs,
tional restriction. Probably, at that temperature (69 ○ C), the β-tape calculated the single point energy, generated the IR spectra, com-
conformation stabilizes and promotes the formation of ChCl:TU pleted the thermodynamic calculations, and wrote the paper. H.C.
DESs. synthesized DESs and performed IR experiments.

IV. CONCLUSION ACKNOWLEDGMENTS


In this study, we employed a computational and statistical We are grateful to our donors who supported to build a com-
approach to unravel the chemical insights that are responsible for putational platform in Bangladesh (http://grc-bd.org/donate/). All
the stability of three 1:1 and six 1:2 ChCl-based DESs. A higher authors would like to acknowledge The World Academy of Science
number of nine different ChCl-based DESs were thoroughly ana- (TWAS) for providing funding (Grant No. 17-479 RG/CHE/AS_I)
lyzed in this study. Thus, we opted to work with only the most to purchase a high performance computer for performing molecular
stable cluster structures from each DES sample, which were detected dynamics simulation.
using MD simulation followed by DFT calculation. From the DFT
results, thermodynamic properties, such as ΔG, ΔH, ΔS, and Keq ,
DATA AVAILABILITY
were calculated for all DES systems. PCA on the spectral data of all
the DESs revealed more insights into their behavioral differences. The data that support the findings of this study are available
PCA also showed that ChCl:Glu is indeed a 1:1 DES but behaves like within the article and its supplementary material.

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-13


Published under an exclusive license by AIP Publishing
The Journal
ARTICLE scitation.org/journal/jcp
of Chemical Physics

32
REFERENCES A. P. Abbott, C. D’Agostino, S. J. Davis, L. F. Gladden, and M. D. Mantle, Phys.
1
Chem. Chem. Phys. 18, 25528 (2016).
Y. Liu, J. B. Friesen, J. B. McAlpine, D. C. Lankin, S.-N. Chen, and G. F. Pauli, 33
M. J. Frisch et al., Gaussian 09, Revision A.02, 2009.
J. Nat. Prod. 81, 679 (2018). 34
2 E. Krieger and G. Vriend, J. Comput. Chem. 36, 996 (2015).
A. P. Abbott, D. Boothby, G. Capper, D. L. Davies, and R. K. Rasheed, J. Am. 35
D. A. Case, T. E. Cheatham, T. Darden, H. Gohlke, R. Luo, K. M. Merz,
Chem. Soc. 126, 9142 (2004).
3 A. Onufriev, C. Simmerling, B. Wang, and R. J. Woods, J. Comput. Chem. 26,
M. S. Rahman, R. Roy, B. Jadhav, M. N. Hossain, M. A. Halim, and D. E. Raynie,
1668 (2005).
J. Mol. Liq. 321, 114745 (2020). 36
4 T. Darden, D. York, and L. Pedersen, J. Chem. Phys. 98, 10089 (1993).
M. Armand, F. Endres, D. R. MacFarlane, H. Ohno, and B. Scrosati, Nat. Mater. 37
M. Saha, M. S. Rahman, M. N. Hossain, D. E.. Raynie, and M. A. Halim, J. Phys.
8, 621 (2009).
5 Chem. A 124, 4690 (2020).
N. V. Plechkova and K. R. Seddon, Chem. Soc. Rev. 37, 123 (2008). 38
6 Origin Lab, 2021.
K. N. Marsh, J. A. Boxall, and R. Lichtenthaler, Fluid Phase Equilib. 219, 93 39
H. Martens and P. Geladi, Encycl. Stat. Sci. 48, 139 (2004).
(2004). 40
7
A. P. Abbott and K. J. McKenzie, Phys. Chem. Chem. Phys. 8, 4265 (2006). V. T. Andrew, P. Abbott, D. L. Davies, G. Capper, R. K. Rasheed, U.S. Patent
8 No. US7183433B2 (2002).
R. Roy, M. S. Rahman, and D. E. Raynie, Curr. Res. Green Sustainable Chem. 3, 41
100035 (2020). M. Faggian, S. Sut, B. Perissutti, V. Baldan, I. Grabnar, and S. Dall’Acqua,
9 Molecules 21, 1531 (2016).
A. P. Abbott, G. Capper, D. L. Davies, H. L. Munro, R. K. Rasheed, and 42
G. García, S. Aparicio, R. Ullah, and M. Atilhan, Energy Fuels 29, 2616 (2015).
V. Tambyrajah, Chem. Commun. 1, 2010 (2001). 43
10 J. Zhu, K. Yu, Y. Zhu, R. Zhu, F. Ye, N. Song, and Y. Xu, J. Mol. Liq. 232, 182
Q. Zhang, K. De Oliveira Vigier, S. Royer, and F. Jérôme, Chem. Soc. Rev. 41,
(2017).
7108 (2012). 44
11 M. Hayyan, F. S. Mjalli, M. A. Hashim, and I. M. AlNashef, Fuel Process.
B. Tang and K. H. Row, Monatsh. Chem. 144, 1427 (2013).
12 Technol. 91, 116 (2010).
M. S. Rahman and D. E. Raynie, J. Mol. Liq. 324, 114779 (2020). 45
13 A. Hayyan, F. S. Mjalli, I. M. Alnashef, Y. M. Al-Wahaibi, T. Al-Wahaibi, and
X. Li and K. H. Row, J. Sep. Sci. 39, 3505 (2016).
14
M. A. Hashim, J. Mol. Liq. 178, 137 (2013).
A. P. Abbott, G. Capper, D. L. Davies, R. K. Rasheed, and V. Tambyrajah, Chem. 46
J. W. Gooch, “Gibbs’ free energy,” in Encyclopedic Dictionary of Polymers, edited
Commun. 9, 70 (2003).
15
by J. W. Gooch (Springer, New York, NY, 2011), https://doi.org/10.1007/978-1-
B. Tang, H. Zhang, and K. H. Row, J. Sep. Sci. 38, 1053 (2015). 4419-6247-8_5492, Online ISBN 978-1-4419-6246-1.
16
E. L. Smith, A. P. Abbott, and K. S. Ryder, Chem. Rev. 114, 11060 (2014). 47
C. Florindo, A. J. S. McIntosh, T. Welton, L. C. Branco, and I. M. Marrucho,
17
S. L. Perkins, P. Painter, and C. M. Colina, J. Phys. Chem. B 117, 10250 (2013). Phys. Chem. Chem. Phys. 20, 206 (2017).
18
S. L. Perkins, P. Painter, and C. M. Colina, J. Chem. Eng. Data 59, 3652 (2014). 48
D. Kundu, P. S. Rao, and T. Banerjee, Ind. Eng. Chem. Res. 59, 11329 (2020).
19
Q. Chemistry, 2020. 49
S. Kim, J. Chen, T. Cheng, A. Gindulyte, J. He, S. He, Q. Li, B. A. Shoemaker,
20
A. Ali, M. S. Rahman, R. Roy, P. Gambill, D. E. Raynie, and M. A. Halim, J. Phys. P. A. Thiessen, B. Yu, L. Zaslavsky, J. Zhang, and E. E. Bolton, Nucleic Acids Res.
Chem. A 125, 2402 (2021). 47, D1102 (2019).
21 50
A. J. Cohen, P. Mori-Sánchez, and W. Yang, Chem. Rev. 112, 289 (2012). A. F. Arkawaz, A. A. Barzinjy, and S. M. Hamad, Singapore J. Sci. Res. 10, 417
22
M. Orio, D. A. Pantazis, and F. Neese, Photosynth. Res. 102, 443 (2009). (2020).
23 51
S. Zhu, H. Li, W. Zhu, W. Jiang, C. Wang, P. Wu, Q. Zhang, and H. Li, J. Mol. D. V. Wagle, H. Zhao, C. A. Deakyne, and G. A. Baker, ACS Sustainable Chem.
Graphics Modell. 68, 158 (2016). Eng. 6, 7525 (2018).
24 52
D. Ballabio, Chemom. Intell. Lab. Syst. 149, 1 (2015). G. K. Kole and M. Kumar, J. Mol. Struct. 1163, 18 (2018).
25 53
S. Wold, K. Esbensen, and P. Geladi, Chemom. Intell. Lab. Syst. 2, 37 (1987). F. Topić and K. Rissanen, Acta Crystallogr., Sect. A: Found. Crystallogr. 68, s73
26 (2012).
S. Mainberger, M. Kindlein, F. Bezold, E. Elts, M. Minceva, and H. Briesen, Mol.
54
Phys. 115, 1309 (2017). C. Taouss, C. Döring, P. G. Jones, L. Pinkert, and M. Strey, CrystEngComm 18,
27
K. Gotoh, “Radial distribution function,” Particulate Morphology (Else- 1842 (2016).
55
vier, 2012), pp. 37-59. ISBN 9780123969743, https://doi.org/10.1016/B978-0-12- A. Marczak, P. Czarnecki, and S. Mielcarek, Z. Naturforsch., A 59, 857 (2004).
56
396974-3.00004-7. C. R. Ashworth, R. P. Matthews, T. Welton, and P. A. Hunt, Phys. Chem. Chem.
28
O. S. Hammond, D. T. Bowron, and K. J. Edler, Green Chem. 18, 2736 (2016). Phys. 15, 11566 (2016).
29 57
L. I. N. Tomé, M. Jorge, J. R. B. Gomes, and J. A. P. Coutinho, J. Phys. Chem. B D. Z. Troter, Z. B. Todorović, D. R. ćokić-Stojanović, B. S. ćordević,
116, 1831 (2012). V. M. Todorović, S. S. Konstantinović, and V. B. Veljković, J. Serb. Chem. Soc.
30
C. D’Agostino, L. F. Gladden, M. D. Mantle, A. P. Abbott, E. I. Ahmed, 82, 1039 (2017).
58
A. Y. M. Al-Murshedi, and R. C. Harris, Phys. Chem. Chem. Phys. 17, 15297 C. Ma, Y. Guo, D. Li, J. Zong, X. Ji, C. Liu, and X. Lu, J. Chem. Eng. Data 61,
(2015). 4172 (2016).
31 59
C. D’Agostino, M. D. Mantle, C. L. Mullan, C. Hardacre, and L. F. Gladden, F. T. T. Huque and J. A. Platts, Org. Biomol. Chem. 1, 1419 (2003).
60
ChemPhysChem 19, 1081 (2018). B. G. Levine, J. E. Stone, and A. Kohlmeyer, J. Comput. Phys. 230, 3556 (2011).

J. Chem. Phys. 155, 044308 (2021); doi: 10.1063/5.0052569 155, 044308-14


Published under an exclusive license by AIP Publishing

You might also like