Journal of Molecular Liquids 368 (2022) 120609

Contents lists available at ScienceDirect

Journal of Molecular Liquids

journal homepage: www.elsevier.com/locate/molliq

Insights into the liquid extraction mechanism of actual high-strength

phenolic wastewater by hydrophobic deep eutectic solvents
Haimei Cheng a,b, Ying Huang a,b, Haiqin Lv a,b, Ligang Li a,b, Qingguo Meng a,b,⇑, Mingzhe Yuan a,b,⇑,
Yexin Liang a,b, Mingliang Jin c
a
Guangzhou Institute of Industrial Intelligence (GI3), Guangzhou 511458, PR China
b
Guangdong Engineering and Technology Research Center for Environmental Purification and Functional Materials, Guangzhou 511458, PR China
c
Joint Laboratory of Optofluidic Technology and System, South China Academy of Advanced Optoelectronics, South China Normal University, 510006, PR China

a r t i c l e i n f o a b s t r a c t

Article history: Liquid-liquid extraction using recently developed hydrophobic deep eutectic solvents (HDESs) shows
Received 14 February 2022 great potential for the removal of phenol from wastewater. But most of the research limited themselves
Revised 11 October 2022 by being focused on the wastewater with phenol concentration lower than 1000 mg
L1 and most impor-
Accepted 12 October 2022
tantly, the extraction mechanism is still lacking in deep understanding. We pilot to remove phenol from
Available online 17 October 2022
actual industrial wastewater containing ultra-high-concentration phenol (54,000 mg
L1) using HDESs,
and also carried out a systematic investigation on their extraction mechanism. Eight HDESs composed of
Keywords:
DL-menthol (hydrogen bond acceptor, HBA) and various fatty acids (hydrogen bond donor, HBD) were
Liquid-liquid extraction
Actual phenolic wastewater
synthesized and the effects of various important conditions, such as HBA/HBD molar ratio, HDES/wastew-
Hydrophobic deep eutectic solvents ater mass ratio, and initial concentration of phenol in wastewater, on their extraction efficiencies were
Extraction efficiency evaluated. The experimental results showed that all the prepared HDESs possess excellent extraction
Extraction mechanism capacity for phenol with the best one being obtained from DL-menthol/nonanoic acid (C9) based HDES
(molar ratio 2:1) within 5 min at room temperature (97 %). Successive extraction was performed to min-
imize the amount of solvents used for economic consideration. Regeneration and reuse of HDESs after
phenol extraction was achieved by activated carbon recovery. The extraction mechanism was investi-
gated experimentally by Fourier transform infrared (FTIR) and 1H nuclear magnetic resonance (NMR)
characterization, together with the predicted activity coefficients of phenol in solvents by COSMO-RS
model and experimentally determined distribution coefficient (D) between HDES and aqueous phase.
The hydrogen bond formed between HDES and phenol during extraction was experimentally identified.
Moreover, quantum computation by DFT method was applied to model the extraction process, and the
extraction mechanism was investigated more in-depth by theoretical computation. The results of DFT
calculation provided good agreement with the experimental data and a new six-membered ring-like
structure formed during the extraction process between HDES and phenol molecule was suggested for
the first time to the best of our knowledge. We hope this work will provide guidance on the study of liq-
uid extraction process/mechanism using HDESs and promote their practical application in the future.
Ó 2022 Elsevier B.V. All rights reserved.

1. Introduction focused on the use of conventional organic solvents, such as

Liquid-liquid extraction is a generally widely adapted technique
for the removal of high-concentration phenolic compounds from
wastewater with minimal energy demand, which owns the advan-
tages of good extraction efficiency and easy operation in a contin-
uous mode. Many experimental studies of the extraction of
phenolic compounds from high-concentration wastewater were
⇑ Corresponding authors at: Guangzhou Institute of Industrial Intelligence (GI3),
Guangzhou 511458, PR China.
E-mail addresses: qgmeng@gmail.com (Q. Meng), mzyuan@sia.cn (M. Yuan).
methyl isobutyl ketone (MIBK), ethyl acetate, diisopropyl ether
(DIPE), tributyl phosphate (TBP) and trioctylamine (TOA) [1–5].
But further industrial applications lead to secondary pollution to
the environment because of their toxicity and volatility. In this
case, exploring environmental-friendly ‘‘green” solvents with
non-toxicity, lower volatility, and flame resistance is urgently
needed.
Deep eutectic solvents (DES) consist of one or more hydrogen
bond donors (HBD) and hydrogen bond acceptors (HBA), where
the eutectic mixtures formed by hydrogen bonding at certain
molar ratio have many advantages, e.g. non volatilization, wide

https://doi.org/10.1016/j.molliq.2022.120609
0167-7322/Ó 2022 Elsevier B.V. All rights reserved.
H. Cheng, Y. Huang, H. Lv et al.
liquid path, low melting point, strong thermal stability, easy avail-
ability of raw materials, simple green preparation, low price and no
further purification required [6]. Hydrophilic DES was developed at
the beginning, but its application was limited due to the stronger
hydrogen bond formation capability of water, resulting in the
instability of hydrophilic DES in the phase boundary within an
aqueous solution. Hydrophobic deep eutectic solvents (HDESs),
on the contrary, are free from this limitation, which have been
developed rapidly with promising application prospects in the field
of sustainable separation [7,8], such as radioactive contaminants
(99mTcO4 [9], [UO2]2+ [10]), metal ion (Co2+ [11]), etc. There have
been also a few reports on the extraction of organic compounds,
e.g. herbicide (2, 4-dichlo-rophenoxyacetic acid [12]), pharmaceu-
ticals [13–15] and phenolic compound [16] from aqueous solution
using HDESs. Marrucho et al. [17] prepared the HDES with fatty
acids as both hydrogen bond acceptor (HBA) and hydrogen bond
donor (HBD) for the extraction of bisphenol A from an aqueous
solution, and the extraction efficiency could reach 91 %. SAS
et al. [18] studied the HDESs based on organic acids (dodecanoic
acid, decanoic acid, and octanoic acid) and menthol as extraction
solvents for phenolic compounds. The effect of phenolic compound
structure, initial phenolic compound concentration and HDES
structures on the extraction efficiency was investigated. An et al.
[19] studied the extraction of 4-chlorophenol, 2,4-dichlorophenol
and 2,4,6-trichlorophenol by the HDES formed between quaternary
ammonium salts and octanoic/decanoic acid. Alnashef et al. [20]
prepared HDESs for the liquid extraction of chlorophenol from
wastewater, and the extraction efficiency was >94 % under the con-
ditions investigated. The alkanoic acid based HDESs exhibited gen-
erally higher extraction efficiencies as compared to those of thymol
based HDESs.
However, all of the above researches were focused on extracting
phenolic compounds from a lab-prepared aqueous solution with
low concentration (<1000 mg
L-1). There is very few exploration
of liquid extraction on actual industrial phenolic wastewater with
much higher concentration using HDESs. Due to the possible gap of
extraction process between lab samples and actual phenolic water,
it is urgent to investigate the liquid extraction from actual wastew-
ater with ultra-high phenol concentration using HDESs, taking into
account the fact that the extraction and separation efficiency is not
only related to their structures and compositions, but also to the
spatial structure and interactions between HDESs and solute
molecules.
Moreover, the extraction mechanism of deep eutectic solvents
has attracted the attentions of many researchers. Yanuar et al.
[21] studied the extraction mechanism of free fatty acids from
palm oil by DES. Using UV–vis, FT-IR and circular dichroism (CD)
spectra, Wang et al. [22] confirmed that the conformation of pro-
tein was not changed during the process of extraction. DES–protein
aggregates and embraces phenomenon play the considerable roles
in the separation process. To explore the absorption mechanism
between DESs and NH3, Deng et al. [23,24] compared the fresh
and NH3– saturated DESs using 1H NMR and IR spectroscopy. The
quantum calculation has also been gradually applied to simulate
the liquid extraction process. Banerjee et al. [25,26] studied the
mechanism of menthol based HDESs for the extraction of lower
alcohols from aqueous media by quantum computation, and it
was inferred that the HBA was playing a more vital role in the
extraction of alcohols than HBD. Yang et al. [27] elucidated the
mechanism of capturing NH3 by DESs via molecular dynamic sim-
ulation and spectroscopic analysis, and a supramolecular network
structure was formed due to hydrogen bonding, which provided
active sites for the absorption of NH3 molecules. Fu et al. [28]
observed the microstructure of menthol-based DES micelles during
extraction using a microscope and employed DFT calculations to
investigate hydrogen bond interactions between menthol and
2
Journal of Molecular Liquids 368 (2022) 120609
acetic acid. Analyses of the related mechanisms indicated that
micelle formation of menthol-based DES in water was the domi-
nant separating process, which provided large contact areas
between target compounds and extraction solvent, thereby pro-
moting extraction from G. biloba leaves. COSMOthermTR utilized
DFT calculations to estimate the electrostatic interactions between
HDESs and solvent extract molecules in liquids [29], e.g. phenolic
compounds and HDES [20,30].The interaction calculated by
COSMO-RS model can well predict the hydrogen bond formation
ability, activity coefficient and partition coefficient. However, there
is still in lack of comprehensive interpretation of the extraction
mechanism of phenolic compounds on the molecular scale, such
as the spatial structure between HDES and phenolic compounds
during extraction, or the possible formation of new bonds between
them.
So to this end, the application of HDES for the actual industrial
wastewater treatment with ultra-high phenol concentration was
systematically investigated in this work, together with the succes-
sive extraction, regeneration and recycling properties of the pre-
pared HDESs. The extraction mechanism was explored on the
molecular scale in terms of both experimental characterization
and DFT based quantum computation. It was strongly evidenced
the possible strong interactions between the prepared HDESs and
phenol molecules during extraction, and a new six-membered
ring-like structure between them was suggested.
2. Experimental
2.1. Materials
DL- menthol (purity  98 %), nonanoic acid (purity  98 %),
decanoic acid (purity  99 %), dodecanoic acid (purity  99 %), oleic
acid (purity  99 %), potassium ferricyanide (purity  99 %),
ammonium chloride (purity  99 %) and 4-aminoantipyrine
(purity  97 %) were purchased from Tansoole.
The actual industrial ultra-high-concentration phenolic
wastewater was collected from the phenolic resin production pro-
cess of a company in Guangzhou (pH = 0.6–2, phenol concentration
up to 54,000 mg
L1, COD up to 180,000–200,000 mg
L1). Phenol
is the major pollutant, together with some other minor organic
substances such as glyoxal and phenolic resin intermediate prod-
ucts, which has been proved to have negligible impact on our liq-
uid extraction analysis. The chemical name, density, molecular
formula and structure of all the reagents are tabulated in Table S1.
2.2. HDESs preparation
The DL-menthol (Mth) as the HBA and various fatty acids (C9 to
C18) as the HBD with the desired molar ratios were mixed and kept
warm in a water bath at 60 ℃ under vigorous magnetic stirring at
800 rpm. A transparent and homogeneous liquid sample was
obtained after 20 min, which was then left intact for 2 h to ensure
that no precipitation was formed during cooling. The obtained
HDESs were used as prepared for phenol extraction without fur-
ther treatment. The HDESs prepared are denoted as Mth-C9-y
(y = 1–5, corresponding to different menthol/C9 molar ratios) and
Mth-Cx-2 (x = 9, 10, 12 and 18, corresponding to different alkyl
chain lengths in fatty acids). Their compositions, physical appear-
ance and abbreviations are summarized in Table 1.
2.3. Physical properties of HDESs
The viscosity of the prepared HDESs was measured by a Vis-
cotester (LV-SSR, China) at 303.15 K. Stability test was performed
by mixing water and HDES with 50 % v/v (volume ratio 1:1) for
H. Cheng, Y. Huang, H. Lv et al. Journal of Molecular Liquids 368 (2022) 120609

Table 1
Composition and physical properties of the prepared HDESs.

Symbol HBA HBD HBA/HBD Molar ratio Viscosity (at 303.15 K, mPa
s) Appearance
Mth-C9-1 DL-Menthol C9 2:3 11.33 Colorless transparent Liquid
Mth-C9-2 DL-Menthol C9 1:1 13.28 Colorless transparent Liquid
Mth-C9-3 DL-Menthol C9 3:2 15.99 Colorless transparent Liquid
Mth-C9-4 DL-Menthol C9 2:1 17.38 Colorless transparent Liquid
Mth-C9-5 DL-Menthol C9 5:2 19.66 Colorless transparent Liquid
Mth-C10-2 DL-Menthol C10 1:1 13.84 Colorless transparent Liquid
Mth-C12-2 DL-Menthol C12 1:1 15.96 Colorless transparent solid–liquid phases
Mth-C18-2 DL-Menthol C18 1:1 28.83 Yellow transparent liquid

1.5 h to obtain full contact between these two phases. 1H NMR
spectra of the HDES and aqueous phases were measured and ana-
lyzed respectively to investigate the stability of HDES in water dur-
ing liquid–liquid extraction, using a Bruker Avance 400 NMR.
Deuterated CDCl3 and D2O were used as solvent for the HDES phase
and aqueous phase, respectively.
The formation of hydrogen bond between oxygen atom in DL-
menthol and hydrogen atom in organic fatty acid is the main driv-
ing force for the HDES preparation. In order to verify the formation
of hydrogen bonds and evaluate the extraction mode in our pre-
pared HDESs, Fourier Transform Infrared Spectrometer (FTIR,
Thermo Scientific Nicolet iS5, USA) was used to characterize the
bond vibrations of HDESs before and after extraction.
2.4. Liquid-liquid extraction
283 to 373 K (see Fig. S1a). So 303 K was applied as the working
extraction temperature. Varying the extraction time from 5 to
60 min didn’t benefit the extraction efficiency, either, indicating
the rapid extraction process with our prepared solvent (Fig. S1b).
So 5 min was employed in this work as the standard extraction
time.
The extraction efficiency (EE) was calculated based on Eq. (1):
C0  Ct
EE ¼  100% ð1Þ
C0
EE is the extraction efficiency of HDES for phenol, where C 0 and
C t are the concentration of phenol in water phase before and after
extraction, respectively.
Distribution coefficient (D) of phenol in HDES and water phase
during the liquid–liquid extraction of wastewater was estimated
by Eq. (2).

In a typical procedure, a certain amount of HDES and phenolic C sol

wastewater were weighed, mixed and magnetically stirred at a
stirring speed of 500 rpm for the desired amount of time, and then
left to stand for 30 min. The concentration of phenolic compounds
in the water phase before and after extraction was determined
using the spectrophotometric method with 4-amineantipyrine as
the testing reagent. This method is based on the formation of an
orange red antipyrine dye between the volatile phenolic com-
pounds and 4-amineantipyrine in the presence of potassium ferri-
cyanide at pH = 10.0 ± 0.2. The colour of dye formed is proportional
to the phenolic compound concentration, and the absorbance was
measured at 510 nm during 30 min after color development using
a UV–vis spectrophotometer following the standard operating pro-
cedure to quantify phenolic compound in water. Subsequently, the
extraction efficiency was evaluated and compared by varying HDES
compositions, HBA/HBD molar ratio, initial concentration of phe-
nol, and HDES/wastewater mass ratio, etc. The extraction of phenol
by a model HDES, Mth-C9-4 solvent, from wastewater at the tem-
peratures from 283 to 373 K indicates that the extraction process is
not quite sensitive to temperature because the extraction effi-
ciency fluctuates very slightly between 97.23 % and 97.61 % from
D¼ ð2Þ
Cw
where C sol and C w are the phenol concentration in HDES and
wastewater phase respectively after each extraction equilibrium.
2.5. Successive extraction and HDES recovery
To test the successive extraction capacity of the prepared HDES,
taking Mth-C9-4 as the example, after each phenol extraction pro-
cess the HDES-rich phase was reused without any treatment to
extract phenol from another batch of ‘‘fresh” phenolic wastewater
again in a new extraction process. This successive extraction pro-
cess was repeated 10 times.
To explore the recovery of HDES after liquid extraction of phe-
nol, the HDES phase was regenerated by removing phenol mole-
cules with activated carbon (AC). Typically, 10 ml of HDES after
liquid extraction of phenol were mixed with the desired amount
of AC under vigorous magnetic stirring at 800 rpm for 30 min, then
separated by centrifugation. After filtration, the AC-free HDES was
regenerated and reused for the next extraction process.

Fig. 1. FTIR spectra (a) and 1H NMR spectra (b) of the Mth-C9-2 solvent and its constituent components.

3
H. Cheng, Y. Huang, H. Lv et al.
2.6. Modeling and DFT calculations
The extraction models between HDESs and phenol molecules
were simulated by Gaussian 09. The present study employed the
B3LYP method [31,32], one of the popular functional in DFT, which
consists of Becke’s exchange functional [33] in conjunction with
Lee-Yang–Parr correlational functional [34]. The basis set is 6–
31 g. The extraction mechanism was analyzed by comparing the
changes in structures and energies. In the current work, extraction
energy was defined to compare the affinity between different
HDES models and phenol molecules, with the aim to model the
extraction process and explore the most probable structure
between HDES and phenol molecules during extraction. The
extraction energies to estimate the extraction capability of differ-
ent HDES to phenol. The extraction energies of phenol for HDES
samples were calculated as the following equation:
Eex ¼ Etot  EHDES  Eph
where Eex is the extraction energy, which is related to the extraction
capability of HDES to phenol. Etot is the total energy of the extrac-
tion model at the ground state. EHDES and Eph are the geometrically
optimized molecular energy of HDES (the mixture of HBA and
HBD compounds that have the same stoichiometric ratios as the
experimental samples) and phenol.
As to the single component extraction agent, such as nonanoic
acid, decanoic acid, dodecanoic acid, oleic acid and DL-menthol,
the extraction energies were calculated as:
Eex ¼ Etot  EEA  Eph
where Eex is the extraction energy, Etot is the total energy of the
extraction model at the ground state, EEA and Eph are the molecular
energy of the extraction agents (nonanoic acid, decanoic acid, dode-
canoic acid, oleic acid and DL-menthol) and phenol.
3. Results and discussion
3.1. HDES characterization
3.1.1. Viscosity
Table 1 summarizes the viscosities of our prepared HDESs, in
which the Mth-Cx-2 samples were prepared by 1:1 ratio between
DL-menthol and four different fatty acids with different alkyl chain
length (x = 9, 10, 12, and 18, respectively) and the Mth-C9-y sam-
ples represent different molar ratios (y = 1–5, corresponding to 2:3,
1:1, 3:2, 2:1, and 5:2, respectively) between DL-menthol and nona-
noic acid (C9). The results indicate that the viscosity of HDES
increased with alkyl chain length from 13.28 (Mth-C9-2) to
28.83 mPa
s (Mth-C18-2). In the meanwhile, increasing the DL-
menthol/C9 molar ratio resulted in the increased viscosity from
11.33 (Mth-C9-1, ratio 2:3) to 19.66 mPa
s (Mth-C9-5, ratio 5:2).
The DL-menthol was concluded to have more impact than fatty acid
on the viscosity of HDESs.
It is worth to mention that the viscosities of all the prepared
HDES samples are less than 30 mPa
s, which is much less than
the criteria of viscosity suggested by Osch et al. [1] for the sustain-
ability of HDES (＜100 mPa
s), meeting the requirement of practi-
cal applications.
3.1.2. HDES formation and stability
In order to confirm the successful formation of HDES, the FTIR
spectra of the Mth-Cx-2 samples are presented in Fig. S2a. Standard
FTIR spectra of each initial component are also given for compar-
ison (Fig. S2b). It was observed that the HDESs retain the features
of their initial components. Taking the Mth-C9-2 solvent as an
example (Fig. 1a), the COOH stretch vibration at 2923.6 cm1 in
4
Journal of Molecular Liquids 368 (2022) 120609
C9 could also be observed at 2923.1 cm1 in Mth-C9-2 and the nar-
row and sharp peak at 1710.5 cm1 corresponding to C@O group in
C9 was evidenced at 1706.8 cm1 in the Mth-C9-2 sample. The peak
for the –OH group stretching vibration in DL-menthol at
3325.6 cm1 was shifted to 3373.4 cm1 in the Mth-C9-2 sample
and a similar shift was also observed in other Mth-Cx-2 samples
(Fig. S2a), which could be due to the formation of new hydrogen
bonds near COOH group between DL-menthol and nonanoic acid
(C9). The redshift of the OAH peak also indicates an electron trans-
fer from an oxygen atom to a hydrogen bond, leading to decreased
force constants, proving the successful preparation of HDES [35].
The 1H NMR spectra of the Mth-C9-2 and its constituent compo-
nents are also shown in Fig. 1b. It is interesting to notice that the
broad peak at 11.4 ppm in C9 disappeared and a new peak at
5.9 ppm was observed in the 1H NMR spectrum of Mth-C9-2. The
chemical shift of other hydrogen atoms bonded with C in fatty acid
chain almost does not change, which further proves the formation
of hydrogen bond near COOH group between DL-menthol and C9.
In order to study the stability of our prepared HDESs in water,
the Mth-Cx-2 samples were mixed thoroughly with water. It was
found that the volume of the Mth-Cx-2 phase did not decrease at
all, indicating the significant hydrophobicity of our prepared
HDESs. The 1H NMR spectra of Mth-Cx-2-rich phase and water-
rich phase are given in Fig. S3. No new hydrogen peak appears after
mixing with water, which further indicates the absence of any
chemical change and proves the stability of the studied HDESs in
water.
3.2. Liquid-liquid extraction
3.2.1. The effect of HDES composition and HBA/HBD molar ratio
The dependence of extraction efficiency of HDES on their chain
length of fatty acids was studied. As shown in Fig. 2a, the extrac-
tion efficiency of the HDESs decreases slightly with the increase
of alkyl chain length of fatty acids from C9 to C18 (with the constant
molar ratio 1:1 between DL-menthol and fatty acids), but it is worth
noting that the extraction efficiencies of all the HDESs were higher
than 90 %.
The effect of the molar ratio between DL-menthol and C9 in the
preparation of HDESs on the extraction efficiency was also studied.
Using the Mth-C9-y samples as the examples, the moles of DL-
menthol were kept constant whilst the moles of nonanoic acid
were varied in order to obtain different HDESs (DL-menthol/
nonanoic acid molar ratios increasing from 2:3 to 5:2), and their
extraction efficiencies were tested (see Fig. 2b). It is obvious that
the extraction efficiency was increased with the HBA/HBD ratios
from 2:3 to 2:1, and then leveled off at 5:2. HBA is playing the pos-
itive role. The maximum extraction efficiency of 94.81 % was
achieved when the molar ratio was 2:1 (Mth-C9-4 solvent).
To understand the positive effect of DL-menthol as HBA on phe-
nol extraction efficiency, activity coefficients (c) were applied
where lower activity coefficient of phenolic compounds in a sol-
vent at infinite dilution indicates greater affinity of this solvent
molecule to phenolic compounds. From the predicted activity coef-
ficient at infinite dilution of phenolic compound in DL-menthol,
medium-chain fatty acids and eutectic solvents at 313.15 K using
a COSMO-RS method [30], DL-menthol owns a lower activity coef-
ficient (ln c -0.4) than fatty acids (ln c  0.01), making DL-
menthol possess greater affinity to phenolic compounds. There-
fore, this greater affinity leads to the higher extraction efficiency
of HDESs with a greater proportion of DL-menthol, which was also
verified experimentally by the phenol extraction by DL-menthol
and C9 fatty acid, respectively (see Fig. S4). The affinity of solvents
to phenol is concluded to play the important role in extraction
efficiency.
H. Cheng, Y. Huang, H. Lv et al.
Fig. 2. The dependence of extraction efficiency of HDES on the HBD chain lengths from C9 to C18 (a) and on the molar ratio between
3.2.2. The effect of initial phenol concentration and HDES/wastewater
mass ratio
Due to the optimal potential of our prepared HDES to extract
phenol from actual phenolic wastewater with ultra-high concen-
tration, it is necessary to study the effect of initial phenol concen-
tration on the extraction efficiency. Industrial phenol-containing
wastewater can be classified as high (>1000 mg
L-1) and low
(<1000 mg
L1) concentration phenol-containing wastewater.
Most of the researches [6,20,36] reported until now were
addressed on the treatment of low concentration phenol-
containing wastewater. So the extraction of phenol from wastew-
ater containing different initial phenol concentrations by the
Mth-C9-4 solvent was investigated, and the results are displayed
in Fig. 3a and 3b. It shows that when the initial concentration of
phenol in wastewater is less than 100 mg
L1, the extraction effi-
ciency increases with the initial concentration of phenol from
76.53 % (10 mg
L1) to 97.18 % (100 mg
L1). The lower extraction
efficiency for low concentration wastewater can be due to the fact
that phenol molecules at low concentration could be largely sur-
rounded by water molecules, hampering the extraction process
[37]. However, when it comes to the result at high and ultra-high
phenol concentration range (1000–54,000 mg
L1), it is interesting
to see that the extraction efficiency remains at a very high level of
97.35–97.97 % (Fig. 3b).
Due to the high cost of industrial wastewater treatment pro-
cess, it is necessary to obtain the greatest extraction efficiency with
the least amounts of extractant. The effect of the HDES/wastewater
mass ratio on the extraction efficiency was investigated, where the
Mth-C9-4 solvent was used as the model solvent to treat the
wastewater with ultra-high phenol concentration
(54,000 mg
L1) (Fig. 3c). It can be seen from Fig. 3c that with
the increase of HDES/wastewater mass ratio from 1:7 to 2:1 (with
the constant total mass), the extraction efficiency presents signifi-
cant increase from 70.35 % to 97.48 %. However, keeping increasing
this mass ratio didn’t benefit the extraction efficiency. So an opti-
mized mass ratio between HDES and wastewater investigated in
this study can be determined as 2:1. In addition, it is worth point-
ing out that even with a much lower mass ratio of 1:7, the remark-
able extraction efficiency was still achieved at 70.35 %, showing
good economic benefits of our prepared HDESs for future industrial
applications.
3.2.3. Successive extraction of phenol
As discussed above, the minimization of the amount of solvents
used is very important in the implementation of extraction pro-
5
Journal of Molecular Liquids 368 (2022) 120609
DL-menthol and nonanoic acid (C9) (b).
cess. So the successive extraction capability is also worth investi-
gating to save the cost on solvents. The successive extraction
results by the Mth-C9-4 solvent from wastewater with low and
ultra-high initial phenol concentration (100 and 54,000 mg.L1)
are displayed in Fig. 4a.
It is interesting to observe that the Mth-C9-4 solvent still main-
tained its capacity to extract phenol from wastewater even after 10
times of successive extraction. But the extraction efficiency shows
a downward trend with the repeated extraction process, which
could be governed by the distribution coefficient (D) of phenol
molecules between HDES and water phase. The distribution coeffi-
cients during the successive extraction of wastewater (phenol con-
centration 54,000 mg.L1) by the Mth-C9-4 solvent were measured
and the result is shown in Fig. S5. With the increase of the repeated
extraction process, the distribution coefficients show only a slight
fluctuation. So the gradually accumulated phenol in the Mth-C9-4
phase after each extraction process resulted in the decline of
extraction capacity. But even after being repeated 10 times, the
extraction efficiency was still >72 %, showing the promising extrac-
tion capacity of HDESs, which could be understood by the supple-
mentary solubility test below.
The solubility test of phenol in the Mth-C9-4 solvent was carried
out by the gradual addition of phenol (mass ratio from 1:1 to 5:1,
see Fig. S6) into Mth-C9-4. When the mass ratio is less than 4.5:1,
the phenol and Mth-C9-4 are completely miscible. But the mixture
was getting opaque with a mass ratio of 5:1, and a 5.5:1 mass ratio
resulted in complete solidification. From this observation, the max-
imum solubility of phenol in the Mth-C9-4 solvent was estimated
at around 4.5 g/gsol at room temperature, which is around 671
times higher than that in water (0.0067 g/gwater) at this tempera-
ture. The total amount of phenol extracted/accumulated in the
Mth-C9-4 phase after 10 times of successive extraction was esti-
mated as 0.0457 g/gsol, which is still far lower than its maximum
solubility in the Mth-C9-4 solvent. This comparison accounts for
the observed >72 % extraction efficiency even after 10 times of suc-
cessive extraction.
3.2.4. HDESs regeneration
To achieve full sustainability, the implementation of a circular
process, where HDESs can be regenerated and reused, is manda-
tory. Adsorption using activated carbon (AC) was proven to be suc-
cessful to recover HDES from extraction phase [13], in light of
which adsorption using AC was adopted in the current work to
regenerate our prepared HDES. The extraction efficiency of the
Mth-C9-4 solvent before and after AC regeneration was compared
H. Cheng, Y. Huang, H. Lv et al.
Fig. 3. The effect of initial phenol concentration in the low range (a), the high range
(b), and HDES/wastewater mass ratio (c) on the extraction efficiency. (The Mth-C9-4
was selected as the testing solvent.).
in Fig. 4b. The results show that the extraction efficiency of the
Mth-C9-4 solvent after 1.0 g of AC regeneration was recovered from
86.5 % to 90.5 %, corresponding to a 93.01 % of recovery percentage.
After repeating five circular processes, the extraction efficiency
was still >90 %. Increasing the amount of AC to 2.0 g recovered
the extraction efficiency by 95 % (with the recovery percentage
of 97.74 %). However, the adsorptive loss of solvents by AC turned
more obvious with the amount of AC, which adversely resulted in
the loss of HDES available for reuse, and in turn dramatically
increasing the extraction cost. So there exists an optimized amount
6
Journal of Molecular Liquids 368 (2022) 120609
of AC used for the regeneration of HDES. In general, the study on
the regeneration and reuse of HDES is still in lacking of extensive
research and more lights should be shed on this concern in the liq-
uid extraction research on deep eutectic solvents.
There might be an argument that why we don’t treat the origi-
nal phenolic wastewater directly with AC. AC-based technologies
have been widely applied in the treatment of phenolic wastewater
as a common solution to many industrial applications, where the
wastewater concentration is usually 1000 mg/L. So AC direct
adsorption was tested to treat our ultra-high concentration
phenol-containing wastewater. But unfortunately, besides the
solid AC contamination in the water phase (Fig. S7a), reusing the
AC extractant for the next phenol wastewater treatment presented
much lower extraction efficiency, i.e. 20 % as tested and can
hardly be reused again, as opposed to our reusable HDES for >10
times without any additional treatment. This serves as the advan-
tage of our HDES extractant. If a similar extraction efficiency is
desired, much more AC samples will be required, which dramati-
cally increasing the cost of water treatment. Plus the
contamination-free (Fig. S7c) feature of our HDES treatment strat-
egy, we believe that it is more appropriate to apply AC for the
extraction in HDES phase for HDES regeneration than direct extrac-
tion from the wastewater phase.
3.3. Extraction mechanism investigation
3.3.1. FTIR spectrum
To investigate the extraction mechanism of phenol from
wastewater by HDES, the Mth-C9-4 solvent dissolved with phenol
molecules (Mth-C9-4 + phenol) was prepared and FTIR analysis
was carried out. For comparison, the FTIR spectra of the authentic
Mth-C9-4 solvent, Mth + phenol and C9 + phenol mixtures are also
given (Fig. 5a). The peak for the –OH stretching can be observed at
3300–3400 cm1 in the spectra of all the tested samples. But a red-
shift of this peak was observed from 3378 cm1 in the Mth-C9-4
solvent to 3308 cm1 in the Mth-C9-4 + phenol mixture, and a dif-
ferent redshift was observed in the IR spectra of Mth + phenol and
C9 + phenol mixtures, compared to that of Mth and C9 respectively
(see Fig. S1a), which suggests that the new interaction during phe-
nol extraction, most likely hydrogen bonds, was formed between
phenol molecules and the Mth-C9-4 solvent. It is the first experi-
mental evidence, to the best of our knowledge, of hydrogen bond
formation between HDES and phenol molecules during liquid
extraction, which has been theoretically predicted previously in
the literature [20] by COSMO-RS calculation between chlorophenol
and HDES.
3.3.2. 1H NMR spectrum
1
H NMR spectra of the Mth-C9-4 solvent and Mth-C9-4 + phenol
were also measured to further confirm the formation of the new
hydrogen bond between HDES and phenol during extraction
(Fig. 5b). The results show that the broad peak at 4.65 ppm, which
has been assigned above to hydrogen bond between HBD and HDA
in the Mth-C9-4 solvent, shifts to 6.46 ppm after extracting phenol
molecules. This confirms the formation of a new hydrogen bond
between Mth-C9-4 and phenol molecules, as evidenced above in
the IR spectra. This strong interaction accounts for the excellent
extraction capability of our prepared HDESs to phenol from high-
strength wastewater with ultra-high concentration.
Then one drop of the phenol extracted Mth-C9-4 solvent mix-
ture was dispersed in 10 ml of water. The broad peak in the 1H
NMR spectrum taking responsible for the new hydrogen bond dis-
appeared, with only phenyl hydrogen peaks from phenol detected,
as shown in Fig. 5c. These observations indicate that the new
hydrogen bond most likely plays an important role in the forma-
tion of a relatively stable structure between phenol and HDES,
H. Cheng, Y. Huang, H. Lv et al. Journal of Molecular Liquids 368 (2022) 120609

Fig. 4. Extraction efficiency of the Mth-C9-4 solvent as a function of the number of successive extraction from wastewater with low and high initial phenol concentrations
(100 and 54,000 mg
L1) (a), and the extraction cycles repeated with and without AC regeneration (b).

Fig. 5. The FTIR spectra of the Mth-C9-4 solvent, Mth-C9-4 + phenol, C9 + phenol and DL-menthol + phenol mixtures (a). The 1H NMR spectra of the Mth-C9-4 before and after
being saturated with phenol (b) and water dissolved by one drop of the phenol saturated Mth-C9-4 solvent mixture (c). The structures of the model (DL-menthol: nonanoic
acid: phenol = 2:1:1) after quantum geometry optimization (d). (The red, grey and white balls are corresponding to oxygen, carbon and hydrogen atoms, respectively. Atoms
No. 1 and No. 2 are from phenol molecules. Atoms No. 3 and No. 4, No. 9 and No. 10 are from the two menthol molecules. Atoms No. 5, No. 6 and No. 7 belong to the nonanoic
acid molecule). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

which could be destroyed by a large amount of water molecules.
This also explains the low extraction efficiency of HDES from
wastewater with a low initial concentration of phenol, as observed
above. Because the relatively large amount of water breaks down
the new structure, part of the extracted phenol molecules might
return to the aqueous phase, resulting in the reduction of extrac-
tion efficiency.
3.3.3. DFT quantum computation
In order to further investigate the extraction mechanism theo-
retically and identify the possible new structure formed between
phenol molecules and HDES, quantum computation based on the
7
first-principle DFT method was carried out. The computed extrac-
tion energies for phenol by the tested HDESs (including the
authentic DL-menthol and fatty acids) were tabulated in Table 2.
When DL-menthol or fatty acids are used as solvents for phenol,
all the computed extraction energies are negative, ranging from
39.1 to 158.2 kJ/mol. This suggests the spontaneous extraction
of phenol from water by DL-menthol or fatty acids. Moreover, when
HDESs are modeled as the solvents, most of the computed extrac-
tion energies are more negative (e.g. 203.4 kJ/mol for the Mth-C9-
4 solvent). This indicates the enhanced extraction capacity of
HDES. The Mth-C9-4 solvent displays the most negative extraction
energy among all the tested solvents, suggesting its best extraction
H. Cheng, Y. Huang, H. Lv et al. Journal of Molecular Liquids 368 (2022) 120609

Table 2 extraction efficiency (72 %) from wastewater with low phenol

The computed extraction energies of the tested solvents for phenol. concentration. Quantum computation results indicate the strong
Solvents Molar ratio between Mth, Extraction energies binding capability between HDES and phenol molecules. The pos-
fatty acid and phenol (kJ/mol) sible most stable new structure was suggested as a six-
C9 1:1 66.3 membered ring structure formed by two Mth, one C9 and one phe-
C10 1:1 158.2 nol molecules. The experimental result where the Mth-C9-4 sol-
C12 1:1 71.5 vent showed the best extraction efficiency qualitatively matches
C18 1:1 71.6
1:1 39.1
the computational results where the most negative extraction
DL-menthol (Mth)
Mth-C9-1 2:3:1 102.2
energy was obtained on this solvent. This suggests the DFT quan-
Mth-C9-2 1:1:1 198.5 tum computation is the strong tool for the investigation of extrac-
Mth-C9-3 3:2:1 119.4 tion mechanism of HDES.
Mth-C9-4 2:1:1 203.4
Mth-C9-5 5:2:1 133.2
Mth-C10-2 1:1:1 135.6 CRediT authorship contribution statement
Mth-C12-2 1:1:1 582.6
Mth-C18-2 1:1:1 141.3 Haimei Cheng: Investigation, Methodology, Writing – original
draft. Ying Huang: . Haiqin Lv: Visualization. Ligang Li: Funding
acquisition, Supervision. Qingguo Meng: Supervision, Writing –
review & editing. Mingzhe Yuan: Project administration, Supervi-
sion. Yexin Liang: . Mingliang Jin: Software.
performance. This qualitatively agrees with our experimental
results observed above where the Mth-C9-4 solvent showed the
Data availability
best extraction efficiency.
The geometry optimized extraction structure model between
Data will be made available on request.
Mth-C9-4 and phenol molecule is present in Fig. 5d. It is very inter-
esting to observe that the O and H atoms in one phenol molecule
Declaration of Competing Interest
and two DL-menthol molecules, plus the C and two O atoms in C9
molecule are almost in the same plane, forming a six-membered
The authors declare that they have no known competing finan-
ring-like structure. This reveals the possible new stable structure
cial interests or personal relationships that could have appeared
between HDES and phenol molecule during extraction process, as
to influence the work reported in this paper.
suggested from the experimental results above. The computation
results reported in this work provide guidance for the research of
extraction mechanism of HDES using quantum modeling and Acknowledgments
computation.
This work is supported by Guangzhou Science, Technology and
Innovation Commission through project number: 202002030406.
4. Conclusion
Appendix A. Supplementary material
In this work, eight HDESs based on DL-menthol and four fatty
acids (nonanoic acid, decanoic acid, dodecanoic acid and oleic acid) Supplementary data to this article can be found online at
were prepared as HDESs to extract phenol from actual industrial https://doi.org/10.1016/j.molliq.2022.120609.
wastewater with ultra-high concentration. The formation of the
hydrophobic deep eutectic solvents via the hydrogen bonds References
between DL-menthol and fatty acids was confirmed by 1H NMR
and FTIR results. The HDES prepared by DL-menthol and nonanoic [1] C. Yao, D. Yanhui, Z. Shengli, Quaternary liquid-liquid equilibria for (water +
ethanol + diisopropyl ether + n-heptane) at 298.15 K, Chem. J. Int. 8 (10) (2006)
acid owns the best extraction efficiency in this work, and the
66–72.
extraction equilibrium was reached within 5 min. The maximum [2] N. Messikh, M.H. Samar, L. Messikh, Neural network analysis of liquid–liquid
extraction efficiency of 97.48 % was achieved at the DL-menthol/ extraction of phenol from wastewater using TBP solvent, Desalination 208
nonanoic acid molar ratio of 2:1 and the HDES/wastewater mass (2007) 42–48.
[3] J. Saien, M. Razi Asrami, Comparative studies on the performance of ionic
ratio of 2:1. We summarized the phenolic compounds extraction liquid and conventional solvent drops in extraction of phenol from water,
efficiencies of previously reported solvents, together with this Chem. Eng. Res. Des. 166 (2021) 259–266.
study with HDES in Table S2. Activity coefficient comparison [4] H. Uslu, D. Datta, H.S. Bamufleh, Reactive extraction of phenol from aqueous
solution using tri-octylamine dissolved in alkanes and alcohols, J. Mol. Liq. 212
explains the higher extraction efficiency of HDESs with a higher (2015) 430–435.
proportion of DL-menthol, which was predicted theoretically from [5] P. Jiao, C. Yang, Y. Lei, Z. Deng, Y. Zhou, The recovery of gallic acid from
COSMO-RS method. This is superior to the viscosity of HDESs on wastewater by extraction with tributyl phosphate/4-methyl-2-pentanone/n-
hexane, tributyl phosphate/n-octanol/n-hexane and n-hexanol, Rsc Adv. 6
the determination of extraction efficiency to phenol. Successive (2016) 93626–93639.
extraction results reveal our prepared HDES still maintained its [6] Q. Zhang, D. Karine, S. Royer, F. Jér?Me, Deep eutectic solvents: syntheses,
capacity (>72 %) to extract phenol from wastewater after 10 times properties and applications, Chem. Soc. Rev. 41 (2012) 7108–7146.
[7] D. Osch, L.F. Zubeir, A. Bruinhorst, M. Rocha, M.C. Kroon, Hydrophobic deep
of successive extraction. Regeneration of HDESs was achieved by eutectic solvents: water-immiscible extractants, Green Chem. 17 (2015) 4518–
AC adsorption, after which they can be reused with very little loss 4521.
of extraction ability. [8] B.D. Ribeiro, C. Florindo, L.C. Iff, M.A.Z. Coelho, I.M. Marrucho, Novel menthol-
based eutectic mixtures: hydrophobic low viscosity solvents, Acs Sust. Chem.
During solvent extraction, the possible new hydrogen bond
Eng. 3 (10) (2015) 2469–2477.
between HDES and phenol molecules was experimentally con- [9] T.E. Phelps, N. Bhawawet, S.S. Jurisson, G.A. Baker, Efficient and selective
firmed by the results of FTIR and 1H NMR measurements in this extraction of 99mTcO4- from aqueous media using hydrophobic deep eutectic
work, which accounts for the excellent extraction capability of solvents, ACS Sust. Chem. Eng. 6 (11) (2018) 13656–13661.
[10] M. Gilmore, N.E. McCourt, F. Connolly, et al., Hydrophobic deep eutectic
our prepared HDESs to phenol. This hydrogen bond is destroyed solvents incorporating trioctylphosphine oxide: advanced liquid extractants,
when dispersed in a large amount of water. This explains the low ACS Sustainable, Chem. Eng 6 (12) (2018) 17323–17332.

8
H. Cheng, Y. Huang, H. Lv et al.
[11] D.J.G.P. van Osch, D. Parmentier, C.H.J.T. Dietz, A. van den Bruinhorst, R.
Tuinier, M.C. Kroon, Removal of alkali and transition metal ions from water
with hydrophobic deep eutectic solvents, Chem. Commun. 52 (80) (2016)
11987–11990.
[12] R. Toprak, A.G. Pekel, E. Kurtulba, S. Ahin, Special designed menthol-based
deep eutectic liquid for the removal of herbicide 2,4-dichlorophenoxyacetic
acid through reactive liquid–liquid extraction, Chem. Pap. 74 (2020) 3995–
4002.
[13] C. Florindo, F. Lima, L.S.C. Branco, I. M. Marrucho, Hydrophobic deep eutectic
solvents: a circular approach to purify water contaminated with ciprofloxacin,
ACS Sust. Chem. Eng. 7 (2019) 14739–14746.
_ Toprakçı, G. Özçelik, M. Bilgin, S. Sßahin, Hydrophobic
[14] E. Kurtulbasß, A.G. Pekel, I.
carboxylic acid based deep eutectic solvent for the removal of diclofenac,
Biomass Convers. Bioref. 12 (6) (2022) 2219–2227.
[15] B. Cfa, A. Nvm, C. Bdr, D. Lcb, B. Imma, Hydrophobic deep eutectic solvents
for purification of water contaminated with Bisphenol-A, J. Mole. Liq. 297
(2020) 111841.
[16] R. Cañadas, M. González-Miquel, E.J. González, I. Díaz, M. Rodríguez,
Hydrophobic eutectic solvents for extraction of natural phenolic
antioxidants from winery wastewater, Sep. Purif. Technol. 254 (2021) 117590.
[17] C. Florindo, L. Romero, I. Rintoul, L.C. Branco, I.M. Marrucho, From phase
change materials to green solvents: hydrophobic low viscous fatty acid-based
deep eutectic solvents, ACS Sust. Chem. Eng. 6 (3) (2018) 3888–3895.
[18] O.G. Sas, M. Castro, N. Domínguez, B. González, Removing phenolic pollutants
using deep eutectic solvents, Sep. Purif. Technol. 227 (2019) 115703-.
[19] Y. An, W. Ma, K.H. Row, Preconcentration and determination of chlorophenols
in wastewater with dispersive liquid-liquid microextraction using
hydrophobic deep eutectic solvents, Anal. Lett. 53 (2) (2020) 262–272.
[20] I. Adeyemi, R. Sulaiman, M. Almazroui, A. Al-Hammadi, I.M. Alnashef, Removal
of chlorophenols from aqueous media with hydrophobic deep eutectic
solvents: experimental study and COSMO RS evaluation, J. Mol. Liq. 311
(2020) 113180.
[21] I. Zahrina, M. Nasikin, K. Mulia, M. Prajanto, A. Yanuar, Molecular interactions
between betaine monohydrate-glycerol deep eutectic solvents and palmitic
acid: computational and experimental studies, J. Mol. Liq. 251 (2018) 28–34.
[22] L. Na, Y. Wang, K. Xu, Y. Huang, X. Ding, Development of green Betaine-based
deep eutectic solvent aqueous two-phase system for the extraction of protein,
Talanta 152 (2016) 23–32.
[23] D. Deng, B. Gao, C. Zhang, X. Duan, Y. Cui, J. Ning, Investigation of protic
NH4SCN-based deep eutectic solvents as highly efficient and reversible NH3
absorbents, Chem. Eng. J. 358 (2019) 936–943.
[24] K.e. Li, H. Fang, X. Duan, D. Deng, Efficient uptake of NH3 by dual active sites
NH4SCN-imidazole deep eutectic solvents with low viscosity, J. Mol. Liq. 339
(2021) 116724.
Journal of Molecular Liquids 368 (2022) 120609
[25] R. Verma, T. Banerjee, Liquid–liquid extraction of lower alcohols using
menthol-based hydrophobic deep eutectic solvent: experiments and
COSMO-SAC predictions, Ind. Eng. Chem. Res. 57 (2018) 3371–3381.
[26] R. Verma, M. Mohan, V.V. Goud, T. Banerjee, Operational strategies and
comprehensive evaluation of menthol based deep eutectic solvent for the
extraction of lower alcohols from aqueous media, ACS Sust. Chem. Eng. 6
(2018) 16920–16932.
[27] Y. Li, M. C. Ali, Q. Yang, Z. Zhang, Z. Bao, B. Su, et al., Front cover: hybrid deep
eutectic solvents with flexible hydrogen-bonded supramolecular networks for
highly efficient uptake of NH3 (ChemSusChem 17/2017), ChemSusChem 10
(2017) 3279–3279.
[28] L.-T. Wang, Q. Yang, Q.i. Cui, X.-H. Fan, M.-Z. Dong, M.-Z. Gao, M.-J. Lv, J.-Y. An,
D. Meng, X.-H. Zhao, Y.-J. Fu, Recyclable menthol-based deep eutectic solvent
micellar system for extracting phytochemicals from Ginkgo biloba leaves, J.
Clean. Prod. 244 (2020) 118648.
[29] F. Eckert, A. Klamt, Fast solvent screening via quantum chemistry: COSMO-RS
approach, AIChE J. 48 (2002) 369–385.
[30] D. Rodríguez-Llorente, A. Cañada-Barcala, C. Muñoz, G. Pascual-Muñoz, P.
Navarro, R. Santiago, V.I. Águeda, S. Álvarez-Torrellas, J. García, M. Larriba,
Separation of phenols from aqueous streams using terpenoids and
hydrophobic eutectic solvents, Sep. Purif. Technol. 251 (2020) 117379.
[31] R. Jose, N.U. Zhanpeisov, H. Fukumura, Y. Baba, M. Ishikawa, Structure-
property correlation of CdSe clusters using experimental results and first-
principles DFT calculations, JACS 128 (2006) 629–636.
[32] W. Quapp, Finding the transition state without initial guess: the growing
string method for Newton trajectory to isomerization and enantiomerization
reaction of alanine dipeptide and poly(15)alanine, J. Comput. Chem. 28 (11)
(2007) 1834–1847.
[33] Becke, D. Axel, Density-functional thermochemistry. III. The role of exact
exchange, J. Chem. Phys. 98 (1998) 5648–5652.
[34] C. Lee, W. Yang, R.G. Parr, Development of the Colle-Salvetti correlation-energy
into a function of the electron density, Phys. Rev. B Condens. Matter 37 (1988)
785–789.
[35] P. Makoś, A. Przyjazny, G. Boczkaj, Hydrophobic deep eutectic solvents as
‘‘green” extraction media for polycyclic aromatic hydrocarbons in aqueous
samples, J. Chromatogr. A 1570 (2018) 28–37.
[36] M. Tiecco, F. Cappellini, F. Nicoletti, T. Del Giacco, R. Germani, P. Di Profio, Role
of the hydrogen bond donor component for a proper development of novel
hydrophobic deep eutectic solvents, J. Mol. Liq. 281 (2019) 423–430.
[37] O.S. Hammond, D.T. Bowron, K.J. Edler, The effect of water upon deep eutectic
solvent nanostructure: an unusual transition from ionic mixture to aqueous
solution, Angew. Chem. 56 (2017) 9782–9785.