Why do ammonium salt/phenol-based deep eutectic solvents show low viscosity?

2.1 Preparation of deep eutectic solvent

Ethylammonium chloride (Eth, 98%), Diethylammonium chloride (Die, 99%), Triethylammonium chloride (Tri, 99%), Glycolic acid (GA, 98%), Phenol (Phe, ≥99.5%) and 4-Methylphenol (Met, 99%) were bought from Shanghai Macklin Biochemical Technology Co., Ltd. DESs were prepared by mixing above staring materials at certain molar ratios. The mixtures were constantly stirred at 70 °C until the formation of homogeneous transparent liquids.

2.2 Characterization

All of DESs and their components were measured by Attenuated Total Reflection Fourier Transform Infrared Spectroscopy (ATR-FTIR, IRTracer-100, SHIMADAU, Japan). The functional groups spectra were recorded in the range of 4000–400 cm⁻¹ using absorption mode with 50 scans at 25 °C.

The solid–liquid phase equilibria can be described by Schröder equation (1) (Prausnitz et al., 1999):

2.3 Viscosity assessment and modeling

Viscosity values of the DESs were measured under atmospheric pressure over the temperature range from 298.15 to 353.15 K (temperature uncertainty: ±0.2 K) using a NDJ-8S digital rotational viscometer (relative uncertainty of the dynamic viscosity ≤±2%). The viscosity values at different temperature for all DES were correlated using the Vogel-Fulcher-Tammann (VFT) model as described by equation (2) (Elhamarnah et al., 2019). The activation energy (E) was calculated based on the viscosity dependence with temperature described by equation (3) (Elhamarnah et al., 2019):

2.4 Computational details

Geometries of molecules and DES complexes were fully optimized at the B3LYP/def2-TZVPD level with Grimme’s DFT-D3(BJ) empirical dispersion correction using BIOVIA Tmolex 2021 (version 21.0.0). The symmetry-adapted perturbation theory (SAPT) was used to decompose binding energy (BE) into various physical parts for deeply understanding the nature of hydrogen bonds. SAPT analyses were performed at the SAPT2+/aug-cc-pVDZ level by PSI4 code without the basis set superposition error (BSSE) (Riley et al., 2011). The thermodynamic property of DES systems was calculated by the software BIOVIA COSMOtherm 2020 (version 20.0.0). For convenient and direct investigation of hydrogen bonds between HBA and HBD, corresponding quantum-chemistry calculations were done in the ratio 1:1. Molecular dynamics (MD) simulations were conducted according to the published work (Fan et al., 2021a). The related electron density, atomic charge and atoms in molecules (AIM) analysis were done by Multiwfn package 3.7 (Lu and Chen, 2012).

3.1 Characterization

Three ammonium salts were chosen as HBAs including ethylammonium chloride (Eth), diethylammonium chloride (Die) and triethylammonium chloride (Tri). Phenol (Phe) and 4-methylphenol (Met) were applied as HBDs. Glycolic acid, one of prototypical HBDs, is frequently used in the model molecules studied. It was employed for comparison with the phenolics. FTIR spectra for GA-Eth, Phe-Eth, Met-Eth, and their pure components are presented in Fig. 1. The results of the other six systems were provided in supplementary material (Fig. S1). The characteristic bands of glycolic acid are observed at 3249 and 1715 cm⁻¹ corresponding to OH stretching vibrations and C⚌O in agreement with reported study (Ahokas et al., 2018). The absorption bands of OH stretching in phenol and 4-methylphenol are located around 3310 and 3306 cm⁻¹, respectively. The OH band shifts to 2923 cm⁻¹ in glycolic acid and the same trend is also founded in phenolics (shifts to 3043 and 3021 cm⁻¹). The red shifts confirm the formation of associated complex in these DES systems.

Close

Fig. 1

FTIR spectra for obtained eutectic solvent systems and their pure sources: (A) GA-Eth; (B) Phe-Eth; (C) Met-Eth. HBA/HBD molar ratio: 1:3.

Export to PPT

Their melting properties are given in Table 1. The conductor like screening model for real solvent model (COSMO-RS) was used to estimate the solid–liquid equilibrium conditions because of its accurate description of non-ideality in DES (Alhadid et al., 2021). The sigma profiles of the individual components were given in the Fig. S2. The predicted phase diagrams with the ideal liquidus curve are showed in Fig. S3. It can be found that temperature depression happens in all systems with a general view. The reduction of melting point for all the investigated mixtures is considerably large and they are liquids at room temperature in a mole ratio of 1:3. Additionally, the mole ratio (1:3) is close to the deep eutectic point as the predicted result. Since the pure sources are all solids at room temperature, it can be qualitatively inferred that the melting point values of the obtained DESs are much lower than those of their individual components.

3.2 Temperature dependence of viscosity

The variation of viscosity of the obtained DESs (mole ratio 1:3) with temperature is depicted in Fig. 2. Phe-Eth and GA-Eth have been studied in reported literatures and the other solvent systems prepared have not been reported as far as we known. The viscosity data obtained herein are almost consistent with previous works (Jiang et al., 2020; Jiang et al., 2019). The viscosity values for the eutectic mixtures increase in the following order: phenol-based systems < 4-methylphenol-based systems < glycolic acid-based systems. With the temperature rising, the difference between phenol and 4-methylphenol is reduced. The viscosity values of DESs containing triethylammonium chloride are slightly higher than the other two salts among the three groups. Despite the share of the same HBAs and proportion in the mixtures, systems based on phenolic compounds and glycolic acid display very distinct viscosity values. In fact, the ammonium salt/glycolic acid-based system’s viscosity is much higher than that of studied ammonium salt/phenolics-based system in the whole temperature range.

Close

Fig. 2

Experimental viscosity (η) values of the eutectic mixtures at different temperature. HBA/HBD molar ratio: 1:3.

Export to PPT

For all eutectic mixtures, adjustable parameters of VFT model are shown in Table 2 with good correlation. The activation energy (E) can be utilized for the description of viscosity behavior depending on temperature, which allows investigating the energy barrier of a system (Table 3). It can be observed that glycolic acid-based systems exhibit higher activation energy values compared with phenolics-based systems. This means that it is much difficult for the monomers/conjugates to move. DESs based on phenol have the smallest E values, in agreement with their small viscosity. The energy is related to the entanglement of molecules or their size or interactions of HBA-HBD (Biernacki et al., 2020; Mirza et al., 2016; Ribeiro et al., 2015). The molar volume (V_m) values of different systems are shown in Table S2. The low viscosity values of liquids are linked to their low V_m values. Obviously, the values of 4-methylphenol-based systems are largest among three groups. Therefore, low viscosity of the studied phenolics-based system does not appear to be a consequence of small V_m.

3.3 Non-bonded interaction

The optimized structures of DES systems were shown in the Fig. 3. SAPT was used to evaluate binding energy; moreover, it is capable of decomposing binding energy into different physical components, which facilitate a deep understanding of the underlying nature of non-bonded interactions. The binding energy can be decomposed into five physically meaningful components including electrostatic, exchange-repulsion, induction, dispersion and charge transfer interaction. Fig. 4A represents values of binding energy and different components in the SAPT-derived energy. The ionic DES systems usually show very strong hydrogen bonding interactions, just like the systems studied in this work. They can be classified as charges assisted hydrogen bonds. From these data in Fig. 4A, it is quite obvious that the sum of the charge transfer, dispersion, induction, and electrostatic terms outweighs the exchange-repulsion part. This leads to negative values of interaction energy in the systems. The total binding energy of them range from −20.99 to −30.71 kcal mol⁻¹.

Close

Fig. 3

The optimized structures of hydrogen bond donor-hydrogen bond acceptor. C, green; Cl, yellow; N, blue; O, red; H, white.

Export to PPT

Close

Fig. 4

(A) Binding energy (BE) and physical components of BE calculated at SAPT2+/aug-cc-pVDZ level; (B) Contribution percentage plot of the SAPT-derived induction, dispersion and electrostatic interaction; (C) Contribution percentage plot of the SAPT-derived charge transfer interaction.

Export to PPT

Furthermore, it can be found that attractive electrostatics always plays a major role, all above 50% in total energy (Fig. 4B). Additionally, the dispersion component of phenolics-based system is larger than that of corresponding glycolic acid-based system. The dispersion interaction is the reflection of the London dispersion force, and it arises because of instantaneous charge fluctuation (An et al., 2019). More importantly, the high viscosity observed in DES is usually linked with the presence of a stronger intermolecular hydrogen bond. However, it can be seen that the only GA-Eth and GA-Die possess slightly higher binding energy than the other systems. More specifically, GA-Tri and Met-Tri show almost same energy (−20.99 and −23.04 kcal mol⁻¹, respectively). While the viscosity value of GA-Tri is 23 times over than that of Met-Tri. When charge transfer is considered into the total energy, it can be interestingly found that the percentage of ammonium salt/glycolic acid-based system is noticeably higher than that of ammonium salt/phenolics-based system (Fig. 4C). Therefore, charge transfer interaction plays a vital role in the fluidity difference of studied DES systems.

3.4 Electron density and atomic charge analysis

Non-bonded interactions are accompanied by charge transfer and polarization, therefore studying intermolecular electron transfers between components in DES, as well as how the electron density is polarized due to the presence of another DES molecule are important. In order to investigate the charge transfer phenomenon in these mixtures, electron density and atomic charge of them were analysed, taking GA-Tri and Met-Tri as cases. The difference of electron density can be revealed by subtracting electron density of HBA (glycolic acid and 4-methylphenol) and triethylammonium chloride in their isolated states from the associated complex (referred to as DESs). As can be seen from Fig. 5A-B, the variation of electron density in GA-Tri is more remarkable than that in Met-Tri. Visualization of the isosurface of the electron density difference is illustrated in Fig. 5C-D. The green and blue parts represent the regions where electron density is increased and decreased after the formation of DESs, respectively. It can be obviously seen that although green and blue one interlace with each other, which confirms a certain amount of charge rearrangement in both compositions with the electron density of HBDs being polarized by the negative donor. The total trend is that electrons are transferred from chloride ion of ammonium salt to glycolic acid or 4-methylphenol.

Close

Fig. 5

Difference map of electron density for eutectic mixtures with respect to its constituents: GA-Tri (A) and Met-Tri (B); Isosurface map of the electron density difference: GA-Tri (C) and Met-Tri (D), setting the isovalue to 0.003, The green (blue) part represents the region where electron density is increased (decreased) after the formation of eutectic mixtures; Variation of atomic charges of monomers in their isolated states and in complex state: GA-Tri (E) and Met-Tri (F), red (blue) part corresponds to the region where atomic charge is increased (decreased) after the formation of eutectic mixtures.

Export to PPT

Variation of atomic charges of used DES sources in their isolated states and in complex state can quantitatively demonstrate how the electrons are transferred between different atoms because of the interactions (Fig. 5E-F). Atomic dipole moment corrected Hirshfeld atomic charges (ADCH) were utilized (Table S3). The positive (negative) part of atomic charge difference corresponds to the region where the atomic charge is increased (decreased) after the formation of DESs. The red and blue colors reflect that the atom has positive and negative charge difference, respectively. The deeper red (blue) the more positive (negative) the charge difference. The charge transferred amount of chloride ion in GA-Tri is obviously larger than that in Met-Tri. Furthermore, as expected, the total transfer value of GA-Tri is higher than that of Met-Tri. The replacement of glycolic acid with phenolic compound results in a decrease of the charge transfer (up to four times). It is indicated that a certain degree of covalency difference is presented between the formed eutectic systems.

3.5 Atoms in molecules (AIM) analysis

For the confirmation of the hydrogen bonds, topology analysis of the gradient vector field is necessary. Orange and yellow spheres correspond to (3, −1) and (3, +1) bond critical point (BCP), brown lines denote bond paths. Fig. 6A-B demonstrates the existence of BCPs for the proposed hydrogen bonds between HBAs and HBDs. It can be seen that marked parts correspond to the critical points of Cl⁻···HO_carboxyl and Cl⁻···HO_{phenolic hydroxyl} hydrogen bond, respectively. The covalent character of strong hydrogen bonding interactions (such as charge assisted hydrogen bond) is significantly greater than that of other hydrogen bonding interactions (such as neutral complexes). Quantum theory of atoms-in-molecules (QTAIM) descriptors is shown in Table 4. It has been pointed out that bonds with covalency must have a BCP with negative H(r) (electronic energy density) (Pakiari et al., 2006). The more negative the value, the stronger will be the interaction (Jenkins and Morrison, 2000). In addition, |V(r)| < G(r), G(r) < |V(r)| < 2G(r) and |V(r)| > 2G(r) correspond to closed-shell (electrostatic) interaction, intermediate interaction, and covalent interaction respectively, where G(r) is the electronic kinetic energy density and V(r) is the electronic potential energy density (Bianchi et al., 2000). The values of electron density (ρ(r)) of normal hydrogen bonds at BCPs generally lie within the range of 0.002–0.035 au (Sarkar et al., 2020). For Met-Tri, the value of ρ(r) classifies its bond as ordinary (closed-shell) hydrogen bond; meanwhile its H(r) value are small negative which distinguish it from ordinary hydrogen bond. For GA-Tri, Δ²ρ(r) and H(r) indicate that it is partially covalent, with ρ(r) larger than that of ordinary hydrogen bond. According to the results of AIM and SAPT, the studied hydrogen bonds can be termed as partially covalent and partially electrostatic. Furthermore, Cl⁻···HO_carboxyl is expected to be more covalent than Cl⁻···HO_{phenolic hydroxyl}, since its H(r) is more negative.

Close

Fig. 6

Atoms in molecules (AIM) analysis of GA-Tri (A) and Met-Tri (B); AIM basin analysis of GA-Tri (C) and Met-Tri (D).

Export to PPT

In order to quantitatively measure the number of electrons delocalized or shared between two atoms, delocalization index (DI) in the AIM basins was studied. As can be seen from Fig. 6C-D, the DI value is 0.1502 between AIM atomic space of Cl and H for GA-Tri, exhibiting that there are 0.1502 electrons shared by the two atoms in average. However, this value is 0.0108 for Met-Tri, about 14 times lower than the former one. To some extent, this observation also reflects that covalent nature of Cl⁻···HO_carboxyl is stronger than that of Cl⁻···HO_{phenolic hydroxyl}. The averaged reduced density gradient (aRDG) method, which is developed based on Bader’s atoms in AIM theory, can be used to deeply investigate various non-bonded interactions in a fluctuating environment (Wu et al., 2013). Different type of interaction is intuitively shown with a different color: highly attractive interaction is blue such as hydrogen bond; weak interaction is green such as Van der Waals interaction; repulsive interaction is red such as steric clashes. As can be seen from Fig. 7A-B, darker blue reveals that the hydrogen-bond interactions of Cl⁻···HO_carboxyl and Cl⁻···HO_{phenolic hydroxyl} are strong. The thermal fluctuation index (TFI) is a valuable new term defined in aRDG method, which can reflect the stability of non-covalent interactions network. More blue or red means the corresponding region is more stable or unstable. From Fig. 7C-D, it can be apparently seen that the hydrogen bond formed by phenolic compound is more flexible than that formed by glycolic acid. According to the published works (Fan et al., 2021b; Smith et al., 2014), the stability of hydrogen bonding interactions in the fluctuating systems is associated with the viscosity of DESs. Therefore, due to the strong covalency of Cl⁻···HO_carboxyl, interaction network stability of ammonium salt/glycolic acid-based DES is more robust than that of ammonium salt/phenolics-based one. This is the main reason why ammonium salt/phenol-based DES show lower viscosity than ammonium salt/acid-based one.

Close

Fig. 7

Visualization and analysis of averaged non-covalent interaction (aNCI) at 298 K: Averaged non-covalent interaction analysis of GA-Tri (A) and Met-Tri (B); Fluctuation index (TFI) of GA-Tri (C) and Met-Tri (D). HBA/HBD molar ratio: 1:3.

Export to PPT