You are on page 1of 17

Chemical Engineering Research and Design 184 (2022) 402–418

Available online at www.sciencedirect.com

Chemical Engineering Research and Design

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

Deep eutectic solvents as entrainers in extractive


distillation – A review
]]
]]]]]]
]]


Maximilian Neubauer, Thomas Wallek, Susanne Lux
Graz University of Technology, Institute of Chemical Engineering and Environmental Technology, NAWI Graz,
Inffeldgasse 25C, 8010 Graz, Austria

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

Article history: One of the most common techniques for separating azeotropes and close-boiling mixtures
Received 13 April 2022 is extractive distillation, where the relative volatility of the components to be separated is
Received in revised form 31 May altered by adding an entrainer. In recent years, deep eutectic solvents have emerged as a
2022 new class of entrainers in extractive distillation. Similar to the related class of ionic li­
Accepted 10 June 2022 quids, deep eutectic solvents combine the high separation capability of solid salts with the
Available online 14 June 2022 simple handling of liquids, additionally exhibiting low to negligible vapour pressures and
non-flammability. Compared to ionic liquids, deep eutectic solvents offer advantages in
Keywords: terms of toxicity issues but also solvent costs. In this review, the current state of research
Extractive distillation regarding deep eutectic solvents in extractive distillation spanning from vapour-liquid-
Azeotropes equilibrium measurements and thermodynamic modelling of the corresponding systems
Deep eutectic solvents to general entrainer feasibility considerations and process simulations is presented and
Vapour-liquid-equilibrium critically evaluated. Additionally, future prospects and comments on unresolved issues
measurements are provided.
Process simulation © 2022 The Authors. Published by Elsevier Ltd on behalf of Institution of Chemical
Engineers.
CC_BY_4.0

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
1.1. Extractive process feasibility assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404
1.2. Process synthesis, design, optimization and control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
2. History and current state of research. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
2.1. VLE measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
2.1.1. Ethanol – water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
2.1.2. IPA – water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
2.1.3. Other systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
2.2. Process simulations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
3. Evaluation of the current state of research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
3.1. Screening and modelling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
3.2. Process simulations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
3.3. Entrainer feasibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
3.3.1. Miscibility gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
3.3.2. Thermal stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
3.3.3. Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413


Corresponding author.
E-mail address: susanne.lux@tugraz.at (S. Lux).
https://doi.org/10.1016/j.cherd.2022.06.019
0263-8762/© 2022 The Authors. Published by Elsevier Ltd on behalf of Institution of Chemical Engineers.
CC_BY_4.0
Chemical Engineering Research and Design 184 (2022) 402–418 403

3.4. Comparison with other solvent types. . . . . . 413


Nomenclature
3.4.1. Minimum amounts . . . . . . . . . . . . . . 414
General 3.4.2. Price . . . . . . . . . . . . . . . . . . . . . . . . . . 414
BOD biological oxygen demand 4. Conclusion and future prospects . . . . . . . . . . . . . . 415
DES deep eutectic solvents Declaration of Competing Interest . . . . . . . . 415
HBA hydrogen-bond acceptor Acknowledgement . . . . . . . . . . . . . . . . . . . . . 415
HBD hydrogen-bond donor Appendix A Supporting information. . . . . . . 415
ILs ionic liquids References . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
NADES natural deep eutectic solvents
NMR nuclear magnetic resonance
NRTL Non-Random-Two-Liquid
TAC total annualized cost 1. Introduction
UNIFAC Universal Quasichemical Functional Group
Activity Coefficients “It may form an azeotrope” – a sentence chemical engineers
UNIQUAC Universal Quasichemical normally do not like to hear. And it is hardly surprising, since
VLE vapour-liquid-equilibrium the separation of azeotropes is technically more challenging
and energy intensive as opposed to the zeotropic case.
Deep eutectic solvent components Numerous techniques have been developed to separate
CC choline chloride azeotropic mixtures, including membrane and special dis­
EG ethylene glycol tillation processes like pressure swing distillation, hetero­
G glycerol azeotropic rectification and extractive distillation (Chen
GA glycolic acid et al., 2005; Mahdi et al., 2015). In extractive distillation, an
IPA 2-propanol additional component, the so-called entrainer, is added to
LA lactic acid the mixture to be separated. The task of the entrainer is to
LVA levulinic acid alter the relative volatilities of the mixture components by
MA malic acid different interactions with them, in order to eliminate the
MCA malonic acid azeotropic point and make separation feasible. Since an op­
TBAB tetrabutylammonium bromide timized process design utilizing an average entrainer may be
TEG triethylene glycol much more expensive than an average design utilizing the
TMAC tetramethylammonium chloride optimal entrainer, the choice of the entrainer is a critical step
Ur Urea (Laroche et al., 1991). Up to now, four classes of entrainers are
known: conventional liquids, solid salts, mixtures of con­
Ionic liquids ventional liquids with solid salts, and ionic liquids (ILs). A
[Bmim] 1-butyl-3-methylimidazolium short description listing advantages and disadvantages ac­
[Emim] 1-ethyl-3-methylimidazolium cording to Lei et al. (2003) is shown in Table 1.
[EMpy] 1-ethyl-3-methylpyridinium When using conventional liquids as entrainers, large
[MEA] 2-hydroxy-ethylammonium amounts are needed (entrainer/feed mass ratios up to 5 – 8).
[BF4] tetrafluoroborate Still, conventional liquid entrainers are widely utilized in
[Br] bromide industry, simply owing to the easy handling and operation of
[Cl] chloride liquids when compared to solid salts or solids in general (Lei
[EtSO4] ethylsulphate et al., 2003). Extractive distillation with ionic liquids (ILs) is
[hexanoate] hexanoate also feasible (Lei et al., 2003, 2014). ILs have received con­
[N(CN)2] dicyanamide siderable attention in recent years, seen by many as solvents
[OAc] acetate of the future (Earle and Seddon, 2000; Rogers and Seddon,
[OTf] trifluoromethanesulfonate 2003). However, many aspects of ILs have been brought into
question lately, such as the costly production as well as
Symbols possible toxicity and biodegradability issues, latter especially
∞ infinite dilution concerning imidazolium-based and fluorinated ILs (Kunz and
C capacity Häckl, 2016; Chen and Mu, 2021; Kudłak et al., 2015; Zhao
K vapour-liquid distribution ratio et al., 2007). Deep eutectic solvents (DES) are a relatively new
P pressure class of entrainers in extractive distillation. The connection
P0 pure component vapour pressure between DES and extractive distillation was first made in
S selectivity 2014 by Rodríguez et al. (2015), even though DES are known
T temperature since 2003.
w mass fraction In 2003, Abbott et al. (2003) first came up with the term
x mole fraction liquid phase “deep eutectic solvent”, when discovering unusually deep
x' mole fraction liquid phase on entrainer free melting point depressions in choline chloride/urea mixtures.
basis Since then, substantial research has been devoted to this
y mole fraction vapour phase topic, with an increasing trend observable, as shown in Fig. 1.
α relative volatility Yet, the term DES is often vaguely defined and no gen­
γ activity coefficient erally accepted definition exists. One common definition for
η dynamic viscosity DES is that they result from the combination of a hydrogen-
bond donor (HBD) and hydrogen-bond acceptor (HBA) to form
an eutectic mixture with unusually deep melting points
404 Chemical Engineering Research and Design 184 (2022) 402–418

Table 1 – Advantages and disadvantages of different entrainer classes (Lei et al., 2003)
Entrainer Advantages Disadvantages

Liquids (“traditional + easy handling and operation - low separation capability, large amounts needed
organic solvents”) - possible contamination of overhead product
Solid salts + high separation capability, small amounts - solids handling: dissolution and transport issues
needed - corrosion
+ non-volatile, no contamination of overhead
product
Liquid + solid salt + easy handling and operation of liquids - solubility of solid salts in liquid solvents often limited,
combined with high separation capability of solid low number of systems where application possible
salts - possible contamination of overhead product
Ionic liquids + easy handling and operation of liquids - often complex synthesis
combined with high separation capability of solid - high price
salts - often toxicity issues
+ non-volatile, no contamination of overhead
product

flammability is also a beneficial property of DES (Ramón and


Guillena, 2019; Marcus, 2019). Compared to ILs, DES have
several advantages. For one, the preparation is easy by
simply combining the starting materials with no further
purification required. This is usually achieved by mixing the
starting materials at elevated temperatures until a homo­
genous liquid is obtained. With the starting materials in
general being relatively cheap this keeps the price of DES
lower in comparison to ILs (Płotka-Wasylka et al., 2020;
Florindo et al., 2014). Concerning toxicity and biodegrad­
ability, DES are thought to be more environmentally friendly
than ILs (Kudłak et al., 2015; Płotka-Wasylka et al., 2020),
especially in the case of DES composed of primary metabo­
lites like amino acids, organic acids, sugars, or choline deri­
vatives, which are called natural deep eutectic solvents
Fig. 1 – Number of publications over the years, data (NADES) (Paiva et al., 2014).
generated with results count for "deep eutectic solvent" in The question is now: How can DES be integrated into the
Web of Science. framework of extractive distillation?
For an answer as coherent as possible it is beneficial to
(Hansen et al., 2021; Smith et al., 2014; El Achkar et al., 2021). give a short outline on this framework as described by
A prominent example is the combination of choline chloride Gerbaud et al. (2019).
and urea, both solids at room temperature, as reported by
Abbott et al. (2003). When brought into contact, a liquid at
room temperature is obtained. However, this definition of 1.1. Extractive process feasibility assessment
DES is not satisfactory from a scientific point of view, and
some researchers have called for a stricter definition of DES, The entrainer selection is a critical step in extractive dis­
especially concerning the term “deep”. Martins et al. (2019) tillation. One of the most widely used selection criteria is the
propose to define a “deep eutectic” as a mixture of pure selectivity SAB which is derived from the relative volatility
components where the eutectic temperature is lower than in yA
0
an ideal mixture. Accordingly, the melting point depression KA xA A PA
AB = = yB = 0 (1)
should then be defined as the difference between real and KB B PB
xB
ideal eutectic point. For further information on the history,
definition and properties of DES, the reader is referred to the presuming the validity of the simplified equilibrium condi­
abundant literature available (Hansen et al., 2021; Smith tion yi P = x i i Pi0 . Under the assumption that the ratio of the
pure component vapour pressures for components A and B is
et al., 2014; El Achkar et al., 2021; Martins et al., 2019; Ramón
roughly constant for small changes in temperature, the re­
and Guillena, 2019; Marcus, 2019; Liu et al., 2015; Warrag and
lative volatility is only influenced by introduction of an en­
Kroon, 2019; Francisco et al., 2013; Płotka-Wasylka
trainer E which changes the ratio of the activity coefficients.
et al., 2020).
This ratio is defined as the selectivity SAB
Coming back to extractive distillation - what makes DES
interesting for this application? Similar to ILs, DES combine AE
SAB = (2)
the easy handling and operation of liquids with the high BE
separation capability of solid salts due to their partly ionic
nature while exhibiting low to negligible vapour pressures For easier comparison and screening purposes often the
(Shahbaz et al., 2016). Latter is desirable in the entrainer re­ selectivity at infinite dilution is used
generation step of the extractive distillation process in terms
A
of energy demand, furthermore solvent loss and emissions SAB = (3)
B
are minimized. In contrast to volatile organic solvents, non-
Chemical Engineering Research and Design 184 (2022) 402–418 405

Another ranking criterion is the product of selectivity and 2. History and current state of research
capacity at infinite dilution
Even though DES were first mentioned in 2003, the connec­
P = SAB CBE (4) tion to extractive distillation was not made until 2014, when
Rodríguez et al. (2015) were the first to evaluate DES as po­
where the capacity is given by
tential entrainers in extractive distillation for the separation
1 of the ethanol – water azeotrope. Four DES were tested with
CBE = (5)
B
regard to the elimination of the azeotropic point. While lactic
acid (LA) – choline chloride (CC) (molar ratio 2:1) shifted the
This product, also termed solvent power, correlates better azeotropic point towards the pure ethanol side, malic acid
with the total annualized cost (TAC) than the selectivity (MA) – CC (1:1), glycolic acid (GA) – CC (3:1) and GA – CC (1:1)
alone as suggested by Kossack et al. (2007). were able to break the azeotrope (Rodríguez et al., 2015).
The minimum amount of entrainer needed to break the Since then a number of papers has been published, mainly
azeotrope is also of interest. It is calculated as the amount focusing on vapour-liquid-equilibrium (VLE) measurements
necessary above which the uni-volatility condition AB = 1 and thermodynamic modelling. Concerning process simula­
is eliminated on an entrainer-free base. Residue curve maps tion, very few papers exist to date.
are also a valuable tool in the feasibility assessment, where This chapter is divided into two parts, VLE measurements
the intersection point of the uni-volatility curve AB = 1 with and process simulations. Although an experimental section
the triangle edge corresponds to the minimum amount of involving actual distillation with DES in columns would be of
entrainer needed (Gerbaud et al., 2019). interest, there exists only one paper who dealt with batch
extractive distillation in addition to a process simulation
investigation. Therefore, the only experiment of this kind to
1.2. Process synthesis, design, optimization and control date will be included in the process simulation section.

After the feasibility assessment and entrainer selection, 2.1. VLE measurements
process synthesis, design and optimization are carried out.
This includes column configuration and operation modes, To give an overview, a summary of azeotropic systems ex­
finding the optimal operating parameters and an appropriate amined via VLE measurements is presented in Table 2.
control structure as well as economic optimization. Since it Some general observations can be made. Firstly, it is
would be beyond the scope of this review, this will not be striking that the overwhelming majority of papers deals with
covered in detail here and the reader is referred to the lit­ choline chloride-based DES. And secondly, about 60% of pa­
erature (Gerbaud et al., 2019). So as already asked above: How pers have been focusing on two systems: ethanol – water and
can DES be integrated into the framework of extractive dis­ IPA – water. Therefore, these two systems will be discussed
tillation? DES follow the same feasibility criteria and con­ separately while research regarding the other systems will be
siderations as for conventional extractive distillation with summarized.
heavy-boiling entrainers and therefore can be handled with
the literature available (Gerbaud et al., 2019). This means that 2.1.1. Ethanol – water
process flowsheets for extractive distillation with DES are The industrially important ethanol – water system exhibits a
similar to flowsheets when using conventional solvents. low-boiling azeotrope (351.3 K, 1 atm) at a composition of
The differences lie in the availability and generation of 89.7 mol% (95.7 wt%) ethanol (CRC Handbook of Chemistry
fundamental thermodynamic properties and, most im­ and Physics, 2016).
portantly, the application of predictive models (e.g. for the It was also the first system to be examined in extractive
calculation of activity coefficients at infinite dilution for distillation with DES by Rodríguez et al. (2015). They in­
screening purposes). For most common organic solvents vestigated four DES: LA – CC (2:1), MA – CC (1:1), GA – CC (3:1)
these properties are readily available in databanks and and GA - CC (1:1). LA – CC was not able to break the azeo­
literature, or can be computed with relatively low effort trope, while the other DES successfully eliminated the
with existing predictive models, whereas for many DES the azeotropic point. A pseudo-pure component approach was
data concerning basic properties is scarce and often not used for thermodynamic modelling, where each DES was
available. However, this data is a prerequisite for feasibility treated as one inseparable component. The non-random-
assessment and process design/simulation. Additionally, two-liquid (NRTL) model was used to correlate the VLE data
taking into account the multitude of possible combinations and obtain binary interaction parameters for pseudo-binary
between DES constituents, which makes the experimental water – DES respectively ethanol – DES systems. The pseudo-
determination of various parameters a tedious and time- ternary systems were then modelled with these parameters.
consuming procedure, the need for accurate and reliable However, systematic errors were found for all systems,
predictive models becomes obvious. These issues will be which the authors ascribed to the non-consideration of
discussed in detail later. First, in the following chapter the ternary interactions when using parameters obtained from
history and current state of research concerning DES in pseudo-binary systems (Rodríguez et al., 2015).
extractive distillation will summarized. Subsequently, Gjineci et al. (2016) studied the separation of the ethanol –
the current state of research will be critically evaluated, water azeotrope with ILs and DES. The DES studied were urea
followed by a conclusion and comments on the future (Ur) – CC (2:1) and triethylene glycol (TEG) – CC (3:1), both
prospects of this topic. The aim of this paper is to provide a were found to eliminate the azeotropic point. VLE measure­
comprehensive review on all aspects of DES as entrainers in ments and biodegradability tests for biological oxygen de­
extractive distillation. mand (BOD) via a VELP BOD manometric apparatus were
406

Table 2 – Overview of azeotropic system - entrainer combinations investigated via VLE measurements, ratios in brackets represent molar ratios.
Azeotropic system Entrainer Subsystem VLE Entrainer Temperature [K] Pressure [kPa] Reference
Type fractions

Polar
ethanol - water lactic acid (LA) - choline chloride (CC) (2:1) ethanol - LA:CC (2:1) PT 10 – 20 mol% 292.84–351.40 5.01–101.45 Rodríguez et al., 2015
water - LA:CC (2:1) PT 10 – 20 mol% 313.82–372.60 6.29–99.95 (Rodríguez et al., 2015)
ethanol - water - LA:CC (2:1) Txy 20 mol% 351.11–371.47 100 (Rodríguez et al., 2015)
malic acid (MA) – choline chloride (1:1) ethanol - MA:CC (1:1) PT 10 – 20 mol% 291.72–351.00 4.90–99.71
water - MA:CC (1:1) PT 10 – 20 mol% 312.53–373.52 6.20–93.75 (Rodríguez et al., 2015)
ethanol - water - MA:CC (1:1) Txy 20 mol% 351.09–372.39 100 (Rodríguez et al., 2015)
glycolic acid (GA) – choline chloride (1:1) ethanol - GA:CC (1:1) PT 10 – 20 mol% 292.23–350.97 5.01–99.96 (Rodríguez et al., 2015)
water - GA:CC (1:1) PT 10 – 20 mol% 308.87–370.86 4.96–101.31 (Rodríguez et al., 2015)
ethanol - water - GA:CC (1:1) Txy 20 mol% 351.09–369.39 100 (Rodríguez et al., 2015)
glycolic acid – choline chloride (3:1) ethanol - GA:CC (3:1) PT 10 – 20 mol% 292.06–351.40 4.93–101.27 (Rodríguez et al., 2015)
water - GA:CC (3:1) PT 10 – 20 mol% 308.01–372.62 4.98–99.95 (Rodríguez et al., 2015)
ethanol - water - GA:CC (3:1) Txy 20 mol% 350.92–370.07 100 (Rodríguez et al., 2015)
urea (Ur) - choline chloride (2:1) ethanol - Ur:CC (2:1) Tx 0 – 60 wt% 351.45–357.55 101.3 (Peng et al., 2017)
water - Ur:CC (2:1) Tx 0 – 83 wt% 373.15–395.33 101.3 (Peng et al., 2017)
ethanol - water - Ur:CC (2:1) Txy 5–10 wt% 351.58–352.03 101.3 (Gjineci et al., 2016)
ethanol - water - Ur:CC (2:1) Txy 10–30 wt% 352.51–375.77 101.3 (Peng et al., 2017)
triethylene glycol (TEG) - choline chloride (3:1) ethanol - water - TEG:CC (3:1) Txy 5–15 wt% 351.73–352.28 101.3 (Gjineci et al., 2016)
ethylene glycol (EG) - choline chloride (2:1) ethanol - water - EG:CC (2:1) Txy 10–80 wt% 356.61–371.34 101.3 (Zhang et al., 2019)
methanol - ethanol - water - EG:CC (2:1) Txy 5 – 65 mol% 349.07–369.11 101.3 (Shu et al., 2020)
glycerol (G) - choline chloride (2:1) ethanol - G:CC (2:1) Txy 0 – 91 mol% 305.2–348.5 11.3 (Souza et al., 2021)
water - G:CC (2:1) Txy 0 – 79 mol% 321.5–348.5 11.3 (Souza et al., 2021)
ethanol - water - G:CC (2:1) Txy 50 mol% 306.0–337.3 11.3 (Souza et al., 2021)
ethanol - water - G:CC (2:1) Txy 10–80 wt% 352.10–378.50 101.3 (Li et al., 2021)
2-propanol (IPA) - water glycolic acid – choline chloride (3:1) IPA - GA:CC (3:1) PT 5 – 10 mol% 306.8–355.3 10–100 (Rodriguez and Kroon, 2015)
water - GA:CC (3:1) PT 5 – 10 mol% 319.7–372.8 10–100 (Rodriguez and Kroon, 2015)
IPA -water - GA:CC (3:1) Txy 5 – 10 mol% 354.3–372.0 100 (Rodriguez and Kroon, 2015)
lactic acid - choline chloride (2:1) IPA - LA:CC (2:1) PT 5 – 10 mol% 307.2–354.5 10–100 (Rodriguez and Kroon, 2015)
water - LA:CC (2:1) PT 5 – 10 mol% 319.5–371.6 10–100 (Rodriguez and Kroon, 2015)
IPA -water - LA:CC (2:1) Txy 5 – 10 mol% 353.6–371.5 100 (Rodriguez and Kroon, 2015)
Chemical Engineering Research and Design 184 (2022) 402–418

glycerol - choline chloride (2:1) IPA -water - G:CC (2:1) Txy 10–80 wt% 353.86–370.05 100 (Zhang et al., 2017)
ethylene glycol - choline chloride (2:1) IPA -water - EG:CC (2:1) Txy 10–80 wt% 355.43–373.29 101.3 (Zhang et al., 2018)
triethylene glycol - choline chloride (3:1) IPA -water - TEG:CC (3:1) Txy 3–15 wt% 354.77–376.25 101.3 (Jiang et al., 2019a)
acetonitrile - water glycolic acid – choline chloride (3:1) acetonitrile - GA:CC (3:1) Tx 0 – 60 mol% 354.7–376.5 101.3 (Sharma et al., 2018)
water - GA:CC (3:1) Tx 0 – 39 mol% 373.1–388.0 101.3 (Sharma et al., 2018)
acetonitrile - water - GA:CC (3:1) Txy 5–15 mol% 356.0–376.6 101.3 (Sharma et al., 2018)
glycolic acid – tetramethylammonium chloride acetonitrile - GA:TMAC (3:1) Tx 0 – 60 mol% 354.7–381.4 101.3 (Sharma et al., 2019)
(TMAC) (3:1)
water - GA:TMAC (3:1) Tx 0 – 39 mol% 373.1–398.8 101.3 (Sharma et al., 2019)
acetonitrile - water - GA:TMAC (3:1) Txy 5–15 mol% 357.3–383.4 101.3 (Sharma et al., 2019)
malic acid – choline chloride (1:1) acetonitrile - MA:CC (1:1) Tx 0 – 48 mol% 354.7–374.3 101.3 (Sharma et al., 2020)
water - MA:CC (1:1) Tx 0 – 29 mol% 373.1–385.6 101.3 (Sharma et al., 2020)
acetonitrile - water - MA:CC (1:1) Txy 5–15 mol% 356.4–378.7 101.3 (Sharma et al., 2020)
(continued on next page)
Chemical Engineering Research and Design 184 (2022) 402–418 407

conducted. Further, 1H‐NMR measurements to study the


reusability of the DES were carried out. Universal-quasi­
chemical (UNIQUAC) and NRTL models were used to corre­
late the ternary VLE data directly, with the results reported to
(Jiang et al., 2019b)
(Jiang et al., 2019b)

2021b)
2021b)
2021b)
2021b)
2021b)

(Bai et al., 2021b)

(Bai et al., 2021b)


(Bai et al., 2021a)

(Bai et al., 2021a)

(Bai et al., 2021a)


be in good agreement with experimental results. Concerning
(Qiu et al., 2021)

(Qiu et al., 2021)

biodegradability, values of organic carbon degraded ranged

al.,
al.,
al.,
al.,
al.,
from 85.6% in case of Ur - CC (2:1) to 42.8% for the IL 2-hy­
droxy ethylammonium acetate. After five boiling cycles
et
et
et
et
et
(Bai
(Bai
(Bai
(Bai
(Bai
at 110 °C for 1.5 – 2 h, no change in chemical structure of the
DES was found, which was confirmed by 1H‐NMR measure­
ments (Gjineci et al., 2016).
The influence of Ur - CC (2:1) on the ethanol – water
azeotrope was also studied by Peng et al. (2017), however
101.3
101.3
101.3

101.3

101.3

101.3
101.3
101.3
101.3
101.3

101.3

101.3

101.3

101.3
over a broader concentration range than Gjineci et al. (2016).
with the ranges depicted in Table 2. Pseudo-binary interac­
tion parameters for water – DES respectively ethanol – DES
362.94–373.43
362.55–372.53
418.55–470.75

402.85–414.45

353.25–353.85

353.25–353.85
353.25–353.85
353.25–353.85
353.25–353.85
353.25–353.85

353.25–353.85

353.25–353.85

353.25–353.85

353.25–353.85
systems were reported. As the calculation of pseudo-ternary
VLE with parameters obtained from pseudo-binary VLE
measurements is in general not satisfactory, as also reported
by Rodriguez and Kroon (2015), additionally the pseudo-
ternary systems were correlated directly with the NRTL
model. As a result, two NRTL parameter sets were reported.
5–15 mol%
5–15 mol%
10–80 wt%

10–80 wt%

10 mol%

10 mol%
15 mol%
20 mol%
10 mol%

10 mol%

10 mol%

10 mol%

10 mol%

One set for the pseudo-binary systems, and one for the
5 mol%

pseudo-ternary system (Peng et al., 2017).


Zhang et al. (2019) studied the mixed entrainer ethylene
glycol (EG) - CC at different molar ratios, where the mixed
Txy
Txy
isoamylalcohol - ethylenglycol - water- Txy

isoamylalcohol - ethylenglycol - water- Txy

benzene - cyclohexane - LVA:TBAB (2:1) Txy

Txy
Txy
Txy
Txy
Txy

benzene - cyclohexane - LVA:TBAB (3:1) Txy

benzene - cyclohexane - LVA:TBAB (4:1) Txy

Txy

Txy

entrainer EG – CC (2:1) was also a DES. The addition of choline


chloride increased the relative volatilities leading to im­
benzene - cyclohexane - LVA:TBAB (2:1)
benzene - cyclohexane - LVA:TBAB (2:1)
benzene - cyclohexane - LVA:TBAB (2:1)
benzene - cyclohexane - LVA:TBAB (2:1)
benzene - cyclohexane - LVA:TBAB (1:1)

benzene - cyclohexane - EG:TBAB (2:1)

proved entrainer performance comparable or better than ILs


benzene - cyclohexane - LVA:CC(2:1)

for the same separation task. In this work, for the first time in
allyl alcohol - urea - Ur:CC (2:1)
allyl alcohol - water- G:CC (1:1)

extractive distillation entrainer studies with DES, the qua­


ternary approach for correlation of VLE data with the NRTL
model was used, with the modelling results found to be in
good agreement with experimental data (Zhang et al., 2019).
The influence of EG – CC besides other mixed entrainers on
the ternary system water – methanol – ethanol was studied
MCA:CC (2:1)

by Shu et al. (2020).


G:CC (2:1)

A DES consisting of glycerol (G) – CC (2:1) was tested for


applicability in the ethanol-water separation by Souza et al.
(2021). VLE data was correlated by NRTL and UNIQUAC
models to obtain binary interaction parameters for the
pseudo-binary systems water - DES respectively ethanol -
malonic acid (MCA) – choline chloride (2:1)

tetrabutylammoniumbromide (TBAB) (2:1)

DES. Similar to the work of Rodriguez and Kroon (2015), it


was reported that the representation of the pseudo-ternary
levulinic acid - choline chloride (2:1)
tetrabutylammoniumbromide (1:1)

tetrabutylammoniumbromide (3:1)

tetrabutylammoniumbromide (4:1)

tetrabutylammoniumbromide (2:1)

systems with the parameters obtained by correlation of


glycerol - choline chloride (1:1)

glycerol - choline chloride (2:1)

pseudo-binary systems was not satisfactory. This was at­


urea - choline chloride (2:1)

tributed to neglecting ternary interactions as well as the


application of the pseudo-pure component approach (Souza
levulinic acid (LVA) -

et al., 2021).
Li et al. (2021) also studied the system G – CC (2:1) –
ethylene glycol -
levulinic acid -

levulinic acid -

levulinic acid -

ethanol – water. However, they used a quaternary approach


for correlation of VLE data. A reduction of the minimum
entrainer mole fraction by 40% as compared to glycerol alone
was reported (Li et al., 2021), as graphically depicted in Fig. 2.

2.1.2. IPA – water


Table 2 – (Continued)

Next to the ethanol – water system, the IPA – water system is


ethylenglycol - water

benzene - cyclohexane

also of industrial relevance, exhibiting a low-boiling azeo­


allyl alcohol - water

trope (353.7 K, 1 atm) at a composition of 67.4 mol% (87.3 wt%)


isoamylalcohol -

IPA (CRC Handbook of Chemistry and Physics, 2016).


Non-polar

The first investigation of a DES as entrainer for the se­


paration of the IPA – water azeotrope was conducted by
Rodriguez and Kroon (2015). LA – CC (2:1) and GA – CC (3:1)
were investigated as entrainers. It was found that these DES
408 Chemical Engineering Research and Design 184 (2022) 402–418

Fig. 2 – Relative volatility of ethanol (1) to water (2) in


Fig. 3 – Relative volatility of IPA (1) to water (2) in relation to
relation to entrainer mole fraction xs at ethanol mole
entrainer weight fraction ws at IPA mole fraction x1 = 1
fraction x1 = 1 (entrainer free basis) and P = 101.3 kPa (Li
(entrainer free basis) and P = 100 kPa (Zhang et al., 2017),
et al., 2021), solid line: ethanol (1) + water (2) + glycerol (3)
solid line: IPA (1) + water (2) + glycerol (3) + choline chloride
+ choline chloride (4), dashed line: ethanol (1) + water (2)
(4), dashed line: IPA (1) + water (2) + glycerol (3).
+ glycerol (3).

were not able to break the azeotrope at the concentrations Sharma et al. investigated the acetonitrile – water azeo­
studied, however a shift of the azeotropic point towards the trope. The following DES were tested: GA – CC (3:1) (Sharma
pure IPA side was observed. The VLE data was correlated by et al., 2018), GA – tetramethylammonium chloride (TMAC)
the NRTL model, with parameters obtained from pseudo- (3:1) (Sharma et al., 2019) and MA – CC (1:1) (Sharma et al.,
binary systems not being able to accurately describe the 2020). Pseudo-binary and pseudo-ternary VLE measurements
pseudo-ternary systems. Pseudo-ternary VLE data was also were conducted and the data of the pseudo-ternary systems
fitted directly, with adequate results reported. Here, for each were directly correlated by the NRTL model. All three DES
entrainer mole fraction investigated (0.05 and 0.1) a different were found to break the azeotrope. The recoverability of the
set of binary interaction parameters was reported (Rodriguez DES was also studied by FT-IR, comparing the spectra before
and Kroon, 2015). and after several boiling cycles in a rotary evaporator, with
Zhang et al. (2018) studied the mixed solvent EG – CC at no significant changes in the spectra reported. However, a
different ratios as well as the DES G – CC (2:1) (Zhang et al., drop in performance for MA – CC (1:1) was observed, which
2017) in regard to the IPA – water azeotrope. Again their mod­ was explained by possible esterification of malic acid with
elling approach was to correlate the quaternary system with choline chloride during the process.
the NRTL model in both works, with satisfactory agreement The allyl alcohol – water system was studied by Jiang et al.
with the experimental data. The azeotrope was eliminated at a (2019b). Two DES, G – CC (1:1) and Ur – CC (2:1) were tested for
minimum entrainer mass fraction of 0.142 for G – CC (2:1) and their applicability as entrainer. Pseudo-ternary VLE data was
0.157 for EG – CC (2:1). Concerning the glycerol-based DES, this measured and correlated directly by the NRTL model. Both
corresponds to a reduction of 38% compared to pure glycerol as DES were reported to eliminate the azeotropic point (Jiang
entrainer, which is also graphically depicted in Fig. 3. et al., 2019b).
Jiang et al. (2019a) applied the COSMO-RS (conductor like The only non-polar azeotropic system investigated until
screening model for real solvents) for the screening of 19 cho­ now is the benzene – cyclohexane system. Bai et al. (2021a)
line chloride-based DES for the separation of the IPA – water studied the influence of three hydrophobic DES: levulinic
azeotrope. Based on the product of selectivity and capacity as acid – tetrabutylammonium bromide (TBAB) (2:1), EG – TBAB
selection criterion, which they termed performance index (PI), (2:1) and levulinic acid – CC (2:1). Pseudo-ternary VLE data
TEG – CC (3:1) was chosen as entrainer to be investigated. VLE was measured for all three DES – azeotrope systems and
measurements and pseudo-ternary correlation with the NRTL correlated by the NRTL model, where levulinic acid – TBAB
model revealed that the azeotropic point is eliminated at an (2:1) was found to break the azeotrope (Bai et al., 2021a).
entrainer mass fraction of 10 wt% (Jiang et al., 2019a). In a follow up paper, Bai et al. (2021b) studied the influ­
ence of levulinic acid – TBAB (1:1, 2:1, 3:1, 4:1) on the benzene
– cyclohexane azeotrope at different molar ratios. The DES
2.1.3. Other systems with the ratio 2:1 was found to break the azeotrope at the
Relatively few other systems than ethanol – water and IPA – concentrations studied. Subsequently, pseudo-ternary VLE
water have been studied up to now. The research concerning data for different entrainer mole fractions (0.05; 0.1; 0.15; 0.2)
these systems is summarized in this section. was reported and correlated with the NRTL model, with four
Chemical Engineering Research and Design 184 (2022) 402–418 409

different parameter sets given for the four investigated en­ simulation. Here for the first time a novel approach con­
trainer mole ratios (Bai et al., 2021b). cerning the thermodynamic modelling was used, where two
The ternary azeotrope isoamylalcohol – ethylene glycol – different sets of NRTL binary interaction parameters were
water was investigated by Qiu et al. (2021) with regard to used for the extractive column and the entrainer regenera­
extractive distillation by means of quantum-chemical cal­ tion column (Fontana et al., 2021). The parameters were
culations. The interactions of the system with different en­ taken from Peng et al. (2017).
trainers were examined, including the deep eutectic solvents A comprehensive study on the extractive distillation of
G – CC (2:1) and malonic acid – CC (2:1). VLE data was reported ethanol – water with Ur – CC (2:1) was conducted by Shang
for the corresponding systems isoamylalcohol – ethylene et al. (2019). The NRTL model was used for modelling, with
glycol – water – DES, however no correlation with thermo­ binary interaction parameters obtained by Peng et al. (2017).
dynamic models was attempted and no statements about The COSMO-RS model was used to compare and explain
entrainer performance based on experimental results were differences in separation performance between glycerol, the
made (Qiu et al., 2021). DES Ur – CC (2:1) and the IL [EMIM][BF4]. Optimization of the
extractive distillation process via a multi-objective genetic
2.2. Process simulations algorithm showed a better separation performance of the
DES compared to glycerol and [EMIM][BF4] (Shang et al., 2019).
Seven research papers dealing with simulation of the ex­ A detailed investigation on control structures for the ex­
tractive distillation process with DES were found; five con­ tractive distillation of ethanol – water with Ur – CC (2:1) was
cerning the separation of ethanol – water and two carried out by Pan et al. (2019a). A multi-objective genetic
concerning the IPA – water azeotrope. algorithm was used for optimization of the process, with
Ma et al. (2017) investigated the system ethanol – water heat intensification incorporated for further savings in en­
with GA – CC (3:1) as entrainer through simulation in Aspen ergy. Five control schemes were evaluated in terms of con­
Plus and a batch distillation experiment. Entrainer recovery trollability and energy efficiency (Pan et al., 2019a).
was carried out in a flash tank at 0.004 kPa and 423 K. A De et al. (2019) investigated the separation of the IPA –
sensitivity analysis was carried out for the determination of water azeotrope with three different entrainers: glycerol,
the main design variables and the process was finally opti­ glycerol + MgCl2 and glycerol – CC (2:1). The symmetric
mized via a heat exchanger network. The NRTL model was eNRTL (electrolyte non-random-two-liquid) model was
used, with binary interaction parameters for DES – ethanol chosen for thermodynamic modelling with parameters re­
and DES – water taken from Rodríguez et al. (2015). A re­ gressed partly from measured VLE data and partly from data
duction in specific energy consumption per kilogram ethanol available in literature. A reduction in TAC by 32.5% and 21.4%
produced by 46.6% and 15.9% compared to the traditional for the saline entrainer glycerol + MgCl2 respectively glycerol
entrainer ethylene glycol – potassium acetate respectively – CC (2:1) as compared to glycerol alone as well as a decline in
the ionic liquid [EMIM][BF4] was reported. The batch experi­ specific energy consumption (SEC) by 33.9% and 24.1% was
ments were carried out in a packed column with a length of reported (De et al., 2019).
1.5 m. As feed, an ethanol – water mixture with an ethanol Cui et al. (2020) evaluated the extractive distillation pro­
molar fraction of 0.85 was used. High purity ethanol in the cess for IPA dehydration. The traditional process with the
distillate with mole fractions ranging from 0.970 to 0.995 conventional solvent ethylene glycol was compared with
depending on reflux ratio and entrainer/feed ratio was ob­ DES-based processes utilizing G – CC (2:1) and EG – CC (2:1) as
tained (Ma et al., 2017). entrainers. For modelling, the NRTL model was used, with
Simulation of ethanol - water separation with the DES Ur – parameters taken from Zhang et al., (2017, 2018). So far, this
CC (2:1) was studied by Han and Chen (2018). Like Ma et al. is the only simulation utilizing the quaternary approach,
(2017) they used the NRTL model, with parameters from Peng treating the system as consisting of four separate compo­
et al. (2017). For optimization of the simulation a sequential nents. Savings in TAC by 30.3% and 30.9% for G – CC (2:1) and
iterative strategy was used. In contrast to Ma et al. (2017) an EG – CC (2:1) compared to ethylene glycol were reported. The
entrainer recovery column was used instead of a flash tank, environmental impact was also compared, with the results
since the proposed pressure for the flash vessel (0.004 kPa) by in favour of the DES-based processes (Cui et al., 2020).
Ma et al. was deemed to be difficult to reach in an industrial
setting. The process was simulated for both the DES and the
traditional solvent ethylene glycol, with savings in TAC by 3. Evaluation of the current state of research
14.61% for the DES-based process compared to EG. Ad­
ditionally, through combination of the preconcentration 3.1. Screening and modelling
section with the entrainer recovery in a single column, fur­
ther savings in TAC by 29.48% for the EG-based process and Screening of entrainers in regard to separation performance
17.42% for the DES-based process could be realised (Han and is usually the first step in the development of an extractive
Chen, 2018). distillation process. It becomes even more important when
The separation of the ethanol – water azeotrope with Ur – using DES as entrainers, since the combination of different
CC (2:1) was also studied by Fontana et al. (2021). A reduction HBAs and HBDs yields a very large number of possible can­
in the specific energy consumption by 27 – 43% for the DES- didates, similar to ILs. Screening of DES is not trivial since
based process depending on solvent flow rates between 28 there is a lack of predictive models. In contrast to DES, pre­
and 31.2 kmol/h was reported when compared to the con­ dictive models have been developed specifically for ILs, as for
ventional entrainer ethylene glycol. This study is the only instance the UNIFAC-Lei model (Lei et al., 2009) which is
one so far dealing with rate-based simulation for extractive continuously revised with an extending parameter matrix
distillation with DES, with 62% more stages required and an (Lei et al., 2012; Dong et al., 2020). For many DES components,
increase in TAC by 8.4% compared to the equilibrium stage especially the ionic part (e.g. choline chloride), group
410 Chemical Engineering Research and Design 184 (2022) 402–418

parameters are missing. Therefore, UNIFAC has not been


used for predicting VLE in extractive distillation with DES up
to now.
Predictive models for conventional molecules like UNIFAC
with its various modifications are well established. Also,
experimental data like activity coefficients at infinite dilu­
tion, which are the basis for most screening procedures, are
often readily available in databanks for conventional com­
ponents. Even integrated software packages exist for
screening, utilizing databanks and/or predictive models, e.g.
the software Entrainer Selection by the Dortmund Databank.
For DES-based systems, few experimental data concerning
activity coefficients at infinite dilution is available to date.
To the best of our knowledge, the only predictive model
available for an at least qualitative screening of DES is the
COSMO-RS model via estimation of activity coefficients at
infinite dilution, as applied by Jiang et al. (2019a) for the se­
paration of the IPA – water azeotrope. Only the molecular
structure is needed as input, with no experimental data re­
quired. A detailed description of this model can be found in
literature (Klamt, 1995, 2011). In the case of extractive dis­
Fig. 4 – Pseudo-ternary data points for the ethanol (1) –
tillation with DES, few screening procedures have been uti­
lized by now, with the choice of the entrainer made rather water (2) – [CC – Ur (1:2)] (3) system from (Peng et al., 2017),
randomly. In general, the COSMO-RS model is widely used 30 wt% CC – Ur ( ), 20 wt% CC – Ur ( ), 10 wt% CC – Ur ( );
for screening procedures in various DES applications lines: NRTL calculations with parameters regressed from
(Hizaddin et al., 2016; Liu and Zhang, 2022; Hadj-Kali et al., pseudo-binary measurement data; NRTL parameters from
2017; Jeliński and Cysewski, 2018; Salleh et al., 2017; Gouveia (Peng et al., 2017); calculation conducted in Aspen Plus V11
et al., 2016; Cheng et al., 2018; Wojeicchowski et al., 2020). with NRTL set 2 (see Table S2 supporting information).
For VLE modelling of systems containing DES, various
approaches can be found in literature (González De Castilla
et al., 2020; Alkhatib et al., 2020). These range from activity
coefficient models and classical equations of state (EoS) to
more sophisticated approaches derived from the statistical
associating fluid theory (SAFT). Since the topic of this review
is extractive distillation with DES and not thermodynamic
modelling of DES in general, the discussion in this section is
restricted to models which were actually used in extractive
distillation with DES. For more detailed information on the
thermodynamic modelling of DES, the reader is referred to
the literature (González De Castilla et al., 2020; Alkhatib et al.,
2020). For modelling VLE of systems containing DES, the
standard NRTL model is widely used. The majority of papers
follows a pseudo-pure component approach, treating the
DES as one single molecular entity. Binary interaction para­
meters obtained from fitting VLE data of the pseudo-binary
systems (component 1 – DES, component 2 – DES) do not
yield satisfying results for the description of the pseudo-
ternary systems (Rodríguez et al., 2015; Peng et al., 2017;
Souza et al., 2021), very likely due to neglection of ternary
interactions, with the binary interactions not dependent on
other mixture components. Exemplary this is shown for two Fig. 5 – Pseudo-ternary data points for the ethanol (1) –
systems in Figs. 4 and 5. water (2) – [CC – G (1:2)] (3) system, 50 mol% CC – G (◆);
Therefore, in literature, the pseudo-ternary systems are dashed line: UNIQUAC calculation, dashed-dotted line:
sometimes correlated directly for constant entrainer frac­ NRTL calculation; calculations with parameters regressed
tions (Rodríguez et al., 2015; Rodriguez and Kroon, 2015; Bai from pseudo-binary measurement data (Souza et al., 2021).
et al., 2021a, 2021b), which has the drawback that the para­
meter set is only valid for this single point. However, there is
also the approach to simultaneously correlate ternary VLE the measured range and also reproducing the azeotropic
data for several constant entrainer fractions (Peng et al., point of ethanol – water with sufficient accuracy as depicted
2017; Jiang et al., 2019a; Sharma et al., 2018, 2019, 2020; Jiang in Figs. 6 and 7.
et al., 2019b). For instance, Peng et al (Peng et al., 2017). cor­ However, the question of validity for these parameters
related pseudo-ternary VLE data for three constant entrainer arises beyond the measured and thus fitted entrainer frac­
fractions (10, 20 and 30 wt%) with the NRTL model, with the tions. Indeed, this is a challenge for simulation of columns,
parameters depicting the measured VLE data quite well in as poor parameters which do not describe experimental VLE
Chemical Engineering Research and Design 184 (2022) 402–418 411

data accurately can lead to questionable results in regard to


process design (Gmehling et al., 2019). The quaternary
modelling approach or individual constituent approach
seems to yield better results in terms of depicting experi­
mental VLE data. However, the applicability of this approach
for process simulation is questionable. The special properties
of a DES arise from the complex molecular interactions of
HBA and HBD (Hansen et al., 2021); if the components are
added separately into the database of a process simulator, it
is doubtful that this depicts reality sufficiently, since the
physical and transport properties of the DES like viscosity are
not represented, even if the VLE is projected more accurately
in comparison to the pseudo-pure component approach,
which will be discussed in more detail later. This was also
pointed out by other authors; H. Zhang et al. investigated
aqueous solutions of CC – Ur (1:2) regarding excess molar
volumes and viscosity and suggested treating the DES in this
case as a pseudo-component to be the more appropriate
approach (Zhang et al., 2021).

3.2. Process simulations


Fig. 6 – Pseudo-ternary data points for the ethanol (1) –
water (2) – [CC – Ur (1:2)] (3) system from (Peng et al., 2017), Few process simulations of extractive distillation with DES
30 wt% CC – Ur ( ), 20 wt% CC – Ur ( ), 10 wt% CC – Ur ( ); are available at the moment, with only two systems (ethanol
lines: NRTL calculations with parameters regressed from – water, IPA – water) studied. Since DES are not available in
pseudo-ternary measurement data; NRTL parameters from commercial process simulator databases, they have to be
(Peng et al., 2017); calculation conducted in Aspen Plus V11 entered as user-defined components. Scalar thermodynamic
with NRTL set 1 (see Table S2 supporting information). properties like critical temperature, pressure and volume, as
well as normal boiling point and acentric factor, are either
taken from literature or estimated, mostly with a combina­
tion of the modified Lydersen-Joback-Reid method and Lee-
Kesler mixing rules as described by Mirza et al. (2015).
Temperature dependent properties like heat capacities, va­
pour pressures or viscosity are either regressed from ex­
perimental data or also estimated. Regarding the choice of
the thermodynamic model, the NRTL model is used ex­
clusively until now, owing to the availability of binary in­
teraction parameters for this model in the literature.
Since the thermodynamic model chosen and the para­
meters used are the core part of any simulation, a closer look
has to be taken on this topic. In their work regarding the
ethanol – water azeotrope separation with GA – CC (3:1), Ma
et al. (2017) used parameters from Rodríguez et al. (2015).
Here, the NRTL model was fitted to pseudo-binary VLE data
at a constant entrainer mole fraction (20 mol%). From a
modelling perspective, this means that the parameters are
only valid for this exact entrainer mole fraction. However, in
the simulation of Ma et al. (2017), the entrainer mole fraction
inside the separation column varies between 0 and ap­
proximately 65 mol%, which should be considered, ad­
ditionally taking into account the fact that parameters
Fig. 7 – Enlarged section of Fig. 6; pseudo-ternary data obtained from fitting pseudo-binary data do not yield sa­
points for the ethanol (1) – water (2) – [CC – Ur (1:2)] (3) tisfying results for the description of ternary VLE (Rodríguez
system from (Peng et al., 2017), 30 wt% CC – Ur ( ), 20 wt% CC et al., 2015; Peng et al., 2017; Souza et al., 2021).
– Ur ( ), 10 wt% CC – Ur ( ); lines: NRTL calculations with Regarding the simulation of ethanol – water separation
parameters regressed from pseudo-ternary measurement with Ur – CC (2:1), Han and Chen (2018), Shang et al. (2019)
data; NRTL parameters from (Peng et al., 2017); calculation and Pan et al. (2019b) used NRTL parameters from Peng et al.
conducted in Aspen Plus V11 with NRTL set 1 (see Table S2 (2017), which were fitted to pseudo-ternary VLE data for en­
Supporting information); azeotropic point prediction: trainer fractions of 10, 20 and 30 wt% corresponding to a
T = 351.3 K; xEthanol = 0.894 (literature (CRC Handbook of maximum mole fraction of 19 mol% for a DES – ethanol
Chemistry and Physics, 2016): T = 351.3 K, xEthanol = 0.897). mixture. In this range the data is depicted quite well by the
model, however the validity outside this range must be
412 Chemical Engineering Research and Design 184 (2022) 402–418

questioned. In the work of Han and Chen (2018), entrainer


mole fractions inside the separation column up to 40 mol%
and in the simulation of Shang et al. (2019) and Pan et al.
(2019b) up to 65 mol% are reported, which is a significant
deviation from the range the parameters were correlated to.
Fontana et al. (2021) also used parameters from Peng et al.
(2017), but as already described in Section 2, two different
NRTL sets were used. In the separation column, parameters
from fitting pseudo-ternary VLE data were used, and in the
entrainer regeneration column, parameters from pseudo-
binary measurements were used. Since in the regeneration
column, only water and DES are present, this approach may
be justified and is further supported by comparing the ex­
perimental data with calculations from different parameter
sets, which shows good agreement, as shown in Fig. 8. Fig. 9 – Experimental (◇) (Abbott et al., 2011) vs calculated
However, the maximum mole fraction in the separation mixture viscosity (calculated in Aspen Plus V11) (▲) for CC –
column simulated by Fontana et al. (2021) is still above G (1:2).
50 mol%, thus the same considerations concerning the
parameter validity for this concentration range should be (1:2) are compared. The calculation of the mixture viscosity
taken into account. was conducted in Aspen Plus V11, with choline chloride en­
The only research group up to now using the quaternary tered as user-defined component and the same viscosity data
or individual constituent approach in process simulation of (see Supporting information Table S1) used as by Cui
extractive distillation with DES was the team of Cui et al. et al. (2020).
(2020). The extractive distillation of the IPA – water azeotrope There is a significant deviation between the calculated
with conventional entrainer EG and DES CC – G (1:2) as well and the experimental values, with the experimental values
as CC – EG (1:2) was investigated (Cui et al., 2020). For the being higher up to a factor of more than 3. Issues like this
simulation, choline chloride was entered as user-defined should be kept in mind when conducting but also evaluating
component and the thermodynamic model chosen was the process simulations with DES. Viscosity is an important
NRTL model, with parameters from Zhang et al. (2017, 2018). transport quantity in distillation, influencing mass transfer
As mentioned in Section 3.1., caution should be taken when and thus also the separation efficiency (Bradtmöller and
treating the DES as “conventional mixture” of two compo­ Scholl, 2015; Mahiout and Vogelpohl, 1987; Böcker and Ronge,
nents, especially considering transport properties, for in­ 2005). Especially for rate-based modelling approaches, using
stance viscosity. In Fig. 9, experimental and calculated viscosities that depict the real system as well as possible are
(mixture of two separate components) viscosities of CC – G needed in order to achieve meaningful results. Nevertheless,
the influence on the simulation results of Cui et al. (2020) is
expected to be low, since an equilibrium stage approach was
used, not taking into account mass transfer phenomena,
which may be affected by higher viscosities.

3.3. Entrainer feasibility

While most research groups up to now have focused on the


general ability of a DES entrainer to break the azeotrope,
which is undoubtedly the key point, less information is
available about general feasibility criteria like miscibility
gaps or thermal stability. Nevertheless, knowledge about
these properties is essential for process design, strongly in­
fluencing the decision of the final entrainer choice.

3.3.1. Miscibility gaps


One important aspect in extractive distillation is the mis­
cibility of the entrainer with the system of interest, as mis­
cibility gaps may set the boundaries for the design of feasible
process ranges. In extractive distillation, miscibility of the
entrainer with the azeotropic system is generally preferred,
Fig. 8 – Boiling curve of water in relation to DES CC – Ur (1:2) whereas in heterogeneous azeotropic distillation liquid-li­
content, experimental points (◆) from (Peng et al., 2017); quid splitting is encountered on some column trays at least
solid line: NRTL calculation with parameters from pseudo- (Springer et al., 2003).
ternary measurements (set 1), dashed line: NRTL In the case of deep eutectic solvents, little information is
calculation with parameters from pseudo-binary available up to now even concerning the most investigated
measurements (set 2); parameters from (Peng et al., 2017); systems ethanol – DES and IPA – DES, for instance in form of
plotting conducted in Aspen Plus V11 with NRTL sets 1 and ternary diagrams. Zhang et al. (2018) observed liquid-liquid
2 (see Table S2 Supporting information). splitting during VLE measurements of IPA – water with EG –
Chemical Engineering Research and Design 184 (2022) 402–418 413

CC (4:3) in the IPA-rich region (IPA fraction on entrainer-free


basis 95 mol%) of the VLE, due to salting out effects by cho­
line chloride (Zhang et al., 2018). Miscibility gaps may be
expected in more systems, since choline chloride, which is
the HBA in the majority of examined DES until now, has only
limited solubility in aliphatic alcohols like IPA (Zhang et al.,
2018) and also acts as salting out agent (Zafarani-Moattar
et al., 2019). Additionally, considering that in many papers
only small total entrainer fractions up to a certain limit were
considered (see Table 2), there is lack of information about
the whole concentration range.

3.3.2. Thermal stability


Attention should also be paid to the thermal stability of the
entrainer, in order to avoid decomposition during the pro­ Fig. 10 – Viscosity of glycerol (•) (Segur and Oderstar, 1951)
cess. Operating columns at lower pressure to decrease and CC – G (1:2) (○) (Abbott et al., 2011).
boiling points can help to mitigate this issue. Nevertheless,
accurate and reliable experimental data is needed for process
design.
In the case of Ur – CC (2:1), there is some discrepancy
between the results of different research groups. In the
process simulations that cover this type of DES, maximum
temperatures in the sump of the extractive distillation
column respectively the entrainer regeneration column of
493 K (Han and Chen, 2018), 418 K (Shang et al., 2019; Pan
et al., 2019b), and 451 K (Fontana et al., 2021) are reported.
Thermal decomposition was not considered by Han and
Chen (2018), while Shang et al. (2019), Pan et al. (2019b) and
Fontana et al. (2021) designed the process based on a de­
composition temperature of 484 K, as reported by Chemat
et al. (2016) in 2016. However, in a more recent study by
Gilmore et al. (2019) from 2019, thermal decomposition of Ur Fig. 11 – Viscosity of ethylene glycol (▲) (Gajardo-Parra
– CC (2:1) at temperatures as low as 363 K is reported. It was et al., 2020) and CC – EG (1:2) (△) (Gajardo-Parra et al., 2020).
also found that samples of Ur – CC (2:1) exposed to elevated
temperatures (353.15 − 363.15 K) for extended periods of time ordered. This is also backed up by a decrease in density,
had a strong ammoniacal odor and crystalline solids were while for diol-based systems (e.g. ethylene glycol) this is not
deposited on the walls of the sample containers. Information the case (Gajardo-Parra et al., 2020; Abbott et al., 2011).
like this is critical and must be considered prior the design of This again shows the complex behaviour of DES, which
the process. Some authors did report on the reusability of may have beneficial or adverse implications as shown for
DES after solvent recycling (Gjineci et al., 2016; Sharma et al., glycerol and ethylene glycol, and the need for reliable ex­
2018, 2019, 2020). However, the number of cycles and thus perimental data for process design. Another example, which
the time periods of exposing the DES to higher temperatures supports the argument for tedious inquiry of transport
were rather low, as for instance in the experiment of Gjineci properties, which should in fact be part of any good process
et al. (2016), where Ur – CC (2:1) and TEG – CC (3:1) were design routine but more so in the case of DES-based pro­
subjected to five boiling cycles with a duration of 1.5 - 2 h. cesses, is the CC – Ur system, where large differences in re­
However, in a real application, the cycling solvent would be ported viscosities are observed as depicted in Fig. 12. Gilmore
subjected to elevated temperatures for weeks or months. et al. (2019) attributed this observation to samples being not
fully dried, which results in lower viscosities.
3.3.3. Viscosity A final remark concluding viscosities of DES in extractive
In extractive distillation, viscosity is an important transport distillation is, that despite rather high viscosities especially
quantity, directly influencing mass transfer and thus the at room temperature, the decrease with rising temperatures
number of stages in a column, with further consequences on is significant. For thermally sensitive processes this may be a
process cost. In general, DES are solvents with rather high problem, in distillation, however, higher temperatures are
viscosity (AlOmar et al., 2016; Liu et al., 2019; Gygli et al., 2020; common anyway and thus mitigate the viscosity issue of DES
Gajardo-Parra et al., 2020), which is a disadvantage in ex­ somewhat. Still, the viscosity issue should be kept in mind,
tractive distillation. However, each DES has to be evaluated in view of mass transfer but also design of column internals.
specifically since surprising phenomena have been observed
for choline chloride-based DES with glycerol and diols as
HBD. In the case of glycerol, the viscosity is reduced sig­ 3.4. Comparison with other solvent types
nificantly by adding choline chloride, while in the case of
ethylene glycol the viscosity increases, which is depicted in From an engineering perspective, another essential question
Figs. 10 and 11. This effect is attributed to the break-up of the is: How do DES in extractive distillation compare to other
three-dimensional interaction of H-bonds in glycerol by ad­ entrainers like ILs or conventional solvents with regard to
dition of choline chloride, resulting in a system which is less price and amounts needed for a specific separation task?
414 Chemical Engineering Research and Design 184 (2022) 402–418

Fig. 12 – Viscosity of CC – Ur (1:2); Abbott et al. ( ) (Abbott Fig. 14 – Minimum entrainer amount (weight fractions)
et al., 2003), Yadav et al. ( ) (Yadav and Pandey, 2014), needed for breaking the IPA – water azeotrope, a (Zhang
Gilmore et al. ( ) (Gilmore et al., 2019), Xie et al. ( ) (Xie et al., 2007), b (Zhang et al., 2018), c (Zhang et al., 2014),
et al., 2014). d (Zhang et al., 2018), e (Zhang et al., 2017); ▨ ILs, ■
conventional solvents, □ DES.

3.4.1. Minimum amounts volatility curves in ternary diagrams (Luyben and Chien,
The amounts needed can be compared on the basis of direct 2010). Further, information from uni-and equi-volatility
results from VLE measurements and thermodynamic models curves is mainly relevant for the separation column and not
on a “standalone” basis, where the minimum amount of the entrainer regeneration/recycle column, which also con­
entrainer needed for breaking the azeotrope is calculated tributes to the total cost of the process.
with a mole fraction of unity (entrainer free basis) and the Therefore, the most conclusive results regarding en­
condition AB = 1, which corresponds to the intersection trainer performance are achieved by process simulation.
point of the uni-volatility curve with the triangle edge in the However, until now, few process simulations with DES as
ternary diagram. For the case of ethanol – water, which re­ entrainers are available, which is also followed by scarcity in
presents the most investigated system, a comparison of en­ direct comparison between different solvents. For the se­
trainers in regard to minimum amounts needed to break the paration of the ethanol – water azeotrope with GA – CC (3:1),
azeotrope is provided in Fig. 13. Ma et al. (2017) reported a reduction in specific energy con­
As can be seen, the investigated DES outperform con­ sumption (MJ/kgEthanol) of 28.2% and 15.9% compared to the
ventional solvents with regard to minimum amounts and are conventional entrainer ethylene glycol respectively the ionic
comparable or even better than ILs. Also for the IPA – water liquid [EMIM][BF4]. Han and Chen (2018) studied the separa­
system, DES show remarkable entrainer performance as tion of ethanol – water with Ur – CC (2:1) and found a re­
shown in Fig. 14. However, the minimum amount of en­ duction in TAC of 14.6% compared to the conventional
trainer deduced from the isovolatility criterion does not ne­ solvent ethylene glycol. The same system was studied by
cessarily correlate with the effectiveness of the entrainer. Fontana et al. (2021) with a rate-based setup, with savings of
The influence of increasing entrainer amounts on the re­ 27 – 43% in specific energy consumption reported. Con­
lative volatility also needs to be taken into account. A vi­ cerning the IPA – water azeotrope, De et al. (2019) reported a
sualization of this phenomenon can be done with equi- reduction in TAC by 21.4% for the entrainer G – CC (2:1)
compared to glycerol. Cui et al. (2020) also investigated the
IPA – water azeotrope, comparing the conventional solvent
ethylene glycol with EG – CC (2:1) and G – CC (2:1), with
savings in TAC by 30.9% respectively 30.3% for the DES-based
process. The findings of the process simulation investiga­
tions concur with the deductions from the calculated
minimum amounts. Compared to conventional solvents, DES
seem to show a better entrainer performance based on TAC
and specific energy demand. Only one process simulation
study directly compared a DES with an IL, where the DES
showed a better performance. However, since this is the only
direct comparison available until now, no reliable deduction
can be drawn from this.

3.4.2. Price
Fig. 13 – Minimum entrainer amount (weight fractions) In Table 3, prices of common ILs and DES components are
needed for breaking the ethanol – water azeotrope, a listed. For the sake of comparison, the cost on a small scale
(Orchillés et al., 2010), b (Ge et al., 2008), c (Tsanas et al., basis from a single vendor (MilliporeSigma, 2021) is com­
2014), d (Zhang et al., 2016), e (Zhang et al., 2015), f (Gjineci pared, under the assumption that economics of scale is valid
et al., 2016), g (Li et al., 2021), h (Zhang et al., 2019); ▨ ILs, ■ for all components.
conventional solvents, □ DES. Although this list is comprised of a limited selection of
solvents, it can be deduced that the price of DES is
Chemical Engineering Research and Design 184 (2022) 402–418 415

Table 3 – Prices of common ILs, conventional solvents and DES constituents, prices retrieved from86.
IL CAS number Price [$] DES component CAS number Price [$]

[Emim][OTf] > 98% (1 kg) 145022–44–2 2440 choline chloride > 98% (1 kg) 67–48–1 111
[Emim][BF4] > 98% (1 kg) 143314–16–3 1480 ethylene glycol > 99% (1 l) 107–21–1 95
[Bmim][OTf] > 97% (1 kg) 174899–66–2 1440 glycerol > 99% (1 l) 56–81–5 65
[Bmim][OAc] > 95% (1 kg) 284049–75–8 1280 urea > 99.5% (1 kg) 57–13–6 53
[Emim][OAc] > 95% (1 kg) 143314–17–4 1150 triethylene glycol 99% (1 kg) 112–27–6 50
[Bmim][BF4] > 98% (1 kg) 174501–65–6 1000
[Emim][Cl] > 95% (1 kg) 65039–09–0 473
[EMpy][EtSO4] > 95% (1 kg) 872672–50–9 418

considerably lower than that of ILs and lies in the range of parameters which then can be used for simulations yielding
conventional solvents, since for preparation of the DES only a more reliable results. The development of predictive models
simple mixing of the starting materials is required. for DES similar to ILs is also an open topic, which would not
only contribute greatly to entrainer screening procedures,
but also to the field of DES in general. Furthermore, the
4. Conclusion and future prospects generation and validation of fundamental thermodynamic
and transport data is needed, especially in view of entrainer
In this first review of its kind, the current state of research miscibility with azeotropic systems over the whole compo­
regarding DES as entrainers in extractive distillation was sition range and long term thermal stability of the DES en­
summarized and evaluated. Most research up to now has trainers. For conclusive comparisons of DES with
been focusing on experimental VLE data procurement and conventional solvents or ILs, process simulations are needed
subsequent modelling of the obtained data. Most DES in­ which still are limited in number, especially those directly
vestigated up to now are based on choline chloride as HBA, comparing economics of DES and ILs. Finally, actual lab scale
with the focus on azeotropic alcohol – water systems. For extractive distillation experiments are encouraged, with only
modelling of the VLE data, gE-models are exclusively used, one reported experiment up to date, offering room for pio­
with the NRTL model being the most used. As DES are a neering work in a novel research field.
special class of solvents the standard procedure for para­ Deep eutectic solvents as entrainers in extractive dis­
meter regression, where model parameters obtained from tillation are a promising new technology, combining high
binary subsystems are used to calculate ternary VLE, does separation capability and low volatility similar to the related
not yield satisfying results. Therefore, two approaches are class of ionic liquids, with possible advantages concerning
found in literature: Direct correlation of pseudo-ternary price and toxicity issues over the latter. However, substantial
systems for fixed entrainer fractions and the individual research has still to be made for their use in an industrial
constituent approach, where model parameters are re­ setting.
gressed for the quarternary systems treating the DES as two
separate components. Regardless of the modelling approach Declaration of Competing Interest
used, DES show a higher separation performance than con­
ventional solvents, comparable to ILs. These findings are also The authors declare that they have no known competing fi­
supported by process simulations, with savings in TAC up to nancial interests or personal relationships that could have
30% reported. However, direct comparisons with ionic liquids appeared to influence the work reported in this paper.
are scarce, so no reliable conclusion can be drawn yet.
Concerning general entrainer feasibility, DES behave like Acknowledgement
conventional heavy boiling entrainers, and the same con­
siderations regarding properties like miscibility, viscosities or Financial support by the Austrian Research Promotion
thermal stability are valid. Even though the decrease of Agency (FFG) is gratefully acknowledged (FFG project
viscosity with rising temperature is significant, in general number: 879587).
viscosities of DES are rather high, similar to ionic liquids,
which may affect mass transfer in columns. Information Appendix A. Supporting information
about miscibility with azeotropic systems over the full
composition range is mostly lacking. Concerning thermal Supplementary data associated with this article can be found
stability, there is also lack on consistent data, with sig­ in the online version at doi:10.1016/j.cherd.2022.06.019.
nificant differences reported by different research groups.
There is great potential for future research on this topic. References
The number of azeotropic systems investigated is still lim­
ited and may be expanded. Especially non-polar systems in Abbott, A.P., Capper, G., Davies, D.L., Rasheed, R.K., Tambyrajah,
combination with hydrophobic DES are hardly covered until V., 2003. Novel solvent properties of choline chloride/urea
now. Concerning the DES themselves, the vast majority used mixtures. Chem. Commun. 1, 70–71. https://doi.org/10.1039/
in extractive distillation research is based on choline chloride b210714g
Abbott, A.P., Harris, R.C., Ryder, K.S., D’Agostino, C., Gladden, L.F.,
as HBA, however, many different HBA are possible. However,
Mantle, M.D., 2011. Glycerol eutectics as sustainable solvent
before advancing towards new systems and component
systems. Green. Chem. 13 (1), 82–90. https://doi.org/10.1039/
combinations, the existing data should be validated and ex­ c0gc00395f
panded over larger concentration ranges. Especially in view Alkhatib, I.I.I., Bahamon, D., Llovell, F., Abu-Zahra, M.R.M., Vega,
of model parameter validity, in order to get more robust L.F., 2020. Perspectives and guidelines on thermodynamic
416 Chemical Engineering Research and Design 184 (2022) 402–418

modelling of deep eutectic solvents. J. Mol. Liq. 298, 112183. Gajardo-Parra, N.F., Cotroneo-Figueroa, V.P., Aravena, P., Vesovic,
https://doi.org/10.1016/j.molliq.2019.112183 V., Canales, R.I., 2020. Viscosity of choline chloride-based deep
AlOmar, M.K., Hayyan, M., Alsaadi, M.A., Akib, S., Hayyan, A., eutectic solvents: experiments and Modeling. J. Chem. Eng.
Hashim, M.A., 2016. Glycerol-based deep eutectic solvents: Data 65 (11), 5581–5592. https://doi.org/10.1021/acs.jced.
physical properties. J. Mol. Liq. 215, 98–103. https://doi.org/10. 0c00715
1016/j.molliq.2015.11.032 Ge, Y., Zhang, L., Yuan, X., Geng, W., Ji, J., 2008. Selection of ionic
Bai, F., Hua, C., Li, J., 2021a. Separation of benzene‐cyclohexane liquids as entrainers for separation of (water + ethanol). J.
azeotropes via extractive distillation using deep eutectic sol­ Chem. Thermodyn. 40 (8), 1248–1252. https://doi.org/10.1016/j.
vents as entrainers. Processes 9 (2), 1–13. https://doi.org/10. jct.2008.03.016
3390/pr9020336 Gerbaud, V., Rodriguez-Donis, I., Hegely, L., Lang, P., Denes, F.,
Bai, F., Hua, C., Bai, Y., Ma, M., 2021b. Design optimization of deep You, X.Q., 2019. Review of extractive distillation. process de­
eutectic solvent composition and separation performance of sign, operation, optimization and control. Chem. Eng. Res.
cyclohexane and benzene mixtures with extractive distilla­ Des. 141, 229–271. https://doi.org/10.1016/j.cherd.2018.09.020
tion. Processes 9 (10). https://doi.org/10.3390/pr9101706 Gilmore, M., Swadzba-Kwasny, M., Holbrey, J.D., 2019. Thermal
Böcker, S., Ronge, G., 2005. Distillation of viscous systems. Chem. properties of choline chloride/urea system studied under
Eng. Technol. 28 (1), 25–28. https://doi.org/10.1002/ceat. moisture-free atmosphere. J. Chem. Eng. Data 64 (12),
200407050 5248–5255. https://doi.org/10.1021/acs.jced.9b00474
Bradtmöller, C., Scholl, S., 2015. Geometry and viscosity effects CRC Handbook of Chemistry and Physics, 2016. In: Haynes, W.M.,
on separation efficiency in distillation. Chem. Eng. Res. Des. Lide, D.R., Bruno, T.J. (Eds.), CRC Press. https://doi.org/10.1201/
99, 75–86. https://doi.org/10.1016/j.cherd.2015.03.013 9781315380476
Chemat, F., Anjum, H., Shariff, A.M., Kumar, P., Murugesan, T., Gjineci, N., Boli, E., Tzani, A., Detsi, A., Voutsas, E., 2016.
2016. Thermal and physical properties of (Choline Chloride + Separation of the ethanol/water azeotropic mixture using
Urea +l-Arginine) deep eutectic solvents. J. Mol. Liq. 218, ionic liquids and deep eutectic solvents. Fluid Ph. Equilib. 424,
301–308. https://doi.org/10.1016/j.molliq.2016.02.062 1–7. https://doi.org/10.1016/j.fluid.2015.07.048
Chen, B., Lei, Z., Zhongwei, Ding, 2005. Special Distillation Gmehling, J., Kleiber, M., Kolbe, B., Rarey, J., 2019. Chemical
Processes. Elsevierhttps://doi.org/10.1016/B978-0-444-51648-0. Thermodynamics for Process Simulation. Wiley https://doi.
X5000-9 org/10.1002/9783527809479
Chen, Y., Mu, T., 2021. Revisiting greenness of ionic liquids and González De Castilla, A., Bittner, J.P., Müller, S., Jakobtorweihen,
deep eutectic solvents. Green. Chem. Eng. 2 (2), 174–186. S., Smirnova, I., 2020. Thermodynamic and transport proper­
https://doi.org/10.1016/j.gce.2021.01.004 ties modeling of deep eutectic solvents: a review on GE-
Cheng, H., Liu, C., Zhang, J., Chen, L., Zhang, B., Qi, Z., 2018. models, equations of state, and molecular dynamics. J. Chem.
Screening deep eutectic solvents for extractive desulfuriza­ Eng. Data 65 (3), 943–967. https://doi.org/10.1021/acs.jced.
tion of fuel based on COSMO-RS model. Chem. Eng. Process. 9b00548
Process. Intensif. 125, 246–252. https://doi.org/10.1016/j.cep. Gouveia, A.S.L., Oliveira, F.S., Kurnia, K.A., Marrucho, I.M., 2016.
2018.02.006 Deep eutectic solvents as azeotrope breakers: liquid–liquid
Cui, P., Liu, X., Zhao, F., Zhu, Z., Wang, L., Wang, Y., 2020. extraction and COSMO-RS prediction. ACS Sustain. Chem.
Molecular mechanism, thermoeconomic, and environmental Eng. 4 (10), 5640–5650. https://doi.org/10.1021/acssuschemeng.
impact for separation of isopropanol and water using the 6b01542
choline-based DESs as extractants. Ind. Eng. Chem. Res. 59 Gygli, G., Xu, X., Pleiss, J., 2020. Meta-analysis of viscosity of
(36), 16077–16087. https://doi.org/10.1021/acs.iecr.0c02794 aqueous deep eutectic solvents and their components. Sci.
De, D., Siva Naga Sai, M., Aniya, V., Naga Jyothi, K., Satyavathi, B., Rep. 10 (1), 21395. https://doi.org/10.1038/s41598-020-78101-y
2019. Economic and environmental impact assessment of Hadj-Kali, M.K., Hizaddin, H.F., Wazeer, I., El blidi, L., Mulyono, S.,
extractive distillation with renewable entrainers for re­ Hashim, M.A., 2017. Liquid-liquid separation of azeotropic
processing aqueous 2-propanol. Chem. Eng. Process. Process. mixtures of ethanol/alkanes using deep eutectic solvents:
Intensif. 143 (July), 107616. https://doi.org/10.1016/j.cep.2019. COSMO-RS prediction and experimental validation. Fluid Ph.
107616 Equilib. 448, 105–115. https://doi.org/10.1016/j.fluid.2017.05.
Dong, Y., Guo, Y., Zhu, R., Zhang, J., Lei, Z., 2020. UNIFAC model 021
for ionic liquids. 2. revision and extension. Ind. Eng. Chem. Han, D., Chen, Y., 2018. Combining the preconcentration column
Res. 59 (21), 10172–10184. https://doi.org/10.1021/acs.iecr. and recovery column for the extractive distillation of ethanol
0c00113 dehydration with low transition temperature mixtures as
Earle, M.J., Seddon, K.R., 2000. Ionic liquids. green solvents for the entrainers. Chem. Eng. Process. Process. Intensif. 131 (July),
future. Pure Appl. Chem. 72 (7), 1391–1398. https://doi.org/10. 203–214. https://doi.org/10.1016/j.cep.2018.08.005
1351/pac200072071391 Hansen, B.B., Spittle, S., Chen, B., Poe, D., Zhang, Y., Klein, J.M.,
El Achkar, T., Greige-Gerges, H., Fourmentin, S., 2021. Basics and Horton, A., Adhikari, L., Zelovich, T., Doherty, B.W., Gurkan, B.,
properties of deep eutectic solvents: a review. Environ. Chem. Maginn, E.J., Ragauskas, A., Dadmun, M., Zawodzinski, T.A.,
Lett. 19 (4), 3397–3408. https://doi.org/10.1007/s10311-021- Baker, G.A., Tuckerman, M.E., Savinell, R.F., Sangoro, J.R., 2021.
01225-8 Deep eutectic solvents: a review of fundamentals and appli­
Florindo, C., Oliveira, F.S., Rebelo, L.P.N., Fernandes, A.M., cations. Chem. Rev. 121 (3), 1232–1285. https://doi.org/10.1021/
Marrucho, I.M., 2014. Insights into the synthesis and proper­ acs.chemrev.0c00385
ties of deep eutectic solvents based on cholinium chloride and Hizaddin, H.F., Hadj-Kali, M.K., Ramalingam, A., Ali Hashim, M.,
carboxylic acids. ACS Sustain. Chem. Eng. 2 (10), 2416–2425. 2016. Extractive denitrogenation of diesel fuel using ammo­
https://doi.org/10.1021/sc500439w nium- and phosphonium-based deep eutectic solvents. J.
Fontana, M., Marchesan, A.N., Maciel Filho, R., Maciel, M.R.W., Chem. Thermodyn. 95, 164–173. https://doi.org/10.1016/j.jct.
2021. Extractive distillation to produce anhydrous bioethanol 2015.12.009
with choline chloride with urea (1:2) as a solvent: a com­ Jeliński, T., Cysewski, P., 2018. Application of a computational
parative evaluation of the equilibrium and the rate-based model of natural deep eutectic solvents utilizing the COSMO-
models. Chem. Eng. Process. - Process. Intensif. 168 (June), RS approach for screening of solvents with high solubility of
108580. https://doi.org/10.1016/j.cep.2021.108580 rutin. J. Mol. Model. 24 (7), 180. https://doi.org/10.1007/s00894-
Francisco, M., Van Den Bruinhorst, A., Kroon, M.C., 2013. Low- 018-3700-1
transition-temperature mixtures (LTTMs): a new generation Jiang, H., Xu, D., Zhang, L., Ma, Y., Gao, J., Wang, Y., 2019a. Vapor-
of designer solvents. Angew. Chem. - Int. Ed. 52 (11), liquid phase equilibrium for separation of isopropanol from
3074–3085. https://doi.org/10.1002/anie.201207548 its aqueous solution by choline chloride-based deep eutectic
Chemical Engineering Research and Design 184 (2022) 402–418 417

solvent selected by COSMO-SAC model. J. Chem. Eng. Data 64 Gemische Auf Siebböden. Chem. Ing. Tech. 59 (9), 746–747.
(4), 1338–1348. https://doi.org/10.1021/acs.jced.8b00895 https://doi.org/10.1002/cite.330590915
Jiang, H., Diao, B., Xu, D., Zhang, L., Ma, Y., Gao, J., Wang, Y., Marcus, Y., 2019. Deep Eutectic Solvents. Springer International
2019b. Deep eutectic solvents effect on vapor-liquid phase Publishing, Cham. https://doi.org/10.1007/978-3-030-00608-2
equilibrium for separation of allyl alcohol from its aqueous Martins, M.A.R., Pinho, S.P., Coutinho, J.A.P., 2019. Insights into
solution. J. Mol. Liq. 279, 524–529. https://doi.org/10.1016/j. the nature of eutectic and deep eutectic mixtures. J. Solut.
molliq.2019.01.163 Chem. 48 (7), 962–982. https://doi.org/10.1007/s10953-018-
Klamt, A., 1995. Conductor-like screening model for real solvents: 0793-1
a new approach to the quantitative calculation of solvation MilliporeSigma 〈https://www.sigmaaldrich.com/US/en〉
phenomena. J. Phys. Chem. 99 (7), 2224–2235. https://doi.org/ (Accessed Sep 20, 2021).
10.1021/j100007a062 Mirza, N.R., Nicholas, N.J., Wu, Y., Kentish, S., Stevens, G.W.,
Klamt, A., 2011. The COSMO and COSMO‐RS solvation models. 2015. Estimation of normal boiling temperatures, critical
WIREs Comput. Mol. Sci. 1 (5), 699–709. https://doi.org/10.1002/ properties, and acentric factors of deep eutectic solvents. J.
wcms.56 Chem. Eng. Data 60 (6), 1844–1854. https://doi.org/10.1021/acs.
Kossack, S.; Kraemer, K.; Gani, R.; Marquardt, W. Generating and jced.5b00046
Evaluating Entrainers for Extractive Distillation Processes in a Orchillés, A.V., Miguel, P.J., Vercher, E., Martínez-Andreu, A.,
Systematic Synthesis Framework. 2007 AIChE Annu. Meet. 2010. Using 1-Ethyl-3-methylimidazolium tri­
2007, No. September, 16–20. fluoromethanesulfonate as an entrainer for the extractive
Kudłak, B., Owczarek, K., Namieśnik, J., 2015. Selected issues re­ distillation of ethanol + water mixtures. J. Chem. Eng. Data 55
lated to the toxicity of ionic liquids and deep eutectic sol­ (4), 1669–1674. https://doi.org/10.1021/je900719z
vents—a review. Environ. Sci. Pollut. Res. 22 (16), 11975–11992. Paiva, A., Craveiro, R., Aroso, I., Martins, M., Reis, R.L., Duarte,
https://doi.org/10.1007/s11356-015-4794-y A.R.C., 2014. Natural deep eutectic solvents - solvents for the
Kunz, W., Häckl, K., 2016. The hype with ionic liquids as solvents. 21st century. ACS Sustain. Chem. Eng. 2 (5), 1063–1071. https://
Chem. Phys. Lett. 661, 6–12. https://doi.org/10.1016/j.cplett. doi.org/10.1021/sc500096j
2016.07.044 Pan, Q., Shang, X., Li, J., Ma, S., Li, L., Sun, L., 2019a. Energy-effi­
Laroche, L., Andersen, H.W., Morari, M., Bekiaris, N., 1991. cient separation process and control scheme for extractive
Homogeneous azeotropic distillation: comparing entrainers. distillation of ethanol-water using deep eutectic solvent. Sep.
Can. J. Chem. Eng. 69 (6), 1302–1319. https://doi.org/10.1002/ Purif. Technol. 219 (March), 113–126. https://doi.org/10.1016/j.
cjce.5450690611 seppur.2019.03.022
Lei, Z., Li, C., Chen, B., 2003. Extractive distillation: a review. Sep. Pan, Q., Shang, X., Li, J., Ma, S., Li, L., Sun, L., 2019b. Energy-effi­
Purif. Rev. 32 (2), 121–213. https://doi.org/10.1081/SPM- cient separation process and control scheme for extractive
120026627 distillation of ethanol-water using deep eutectic solvent. Sep.
Lei, Z., Zhang, J., Li, Q., Chen, B., 2009. UNIFAC model for ionic Purif. Technol. 219 (December 2018), 113–126. https://doi.org/
liquids. Ind. Eng. Chem. Res. 48 (5), 2697–2704. https://doi.org/ 10.1016/j.seppur.2019.03.022
10.1021/ie801496e Peng, Y., Lu, X., Liu, B., Zhu, J., 2017. Separation of azeotropic
Lei, Z., Dai, C., Liu, X., Xiao, L., Chen, B., 2012. Extension of the mixtures (Ethanol and Water) enhanced by deep eutectic
UNIFAC model for ionic liquids. Ind. Eng. Chem. Res. 51 (37), solvents. Fluid Ph. Equilib. 448, 128–134. https://doi.org/10.
12135–12144. https://doi.org/10.1021/ie301159v 1016/j.fluid.2017.03.010
Lei, Z., Dai, C., Zhu, J., Chen, B., 2014. Extractive distillation with Płotka-Wasylka, J., de la Guardia, M., Andruch, V., Vilková, M.,
ionic liquids: a review. AIChE J. 60 (9), 3312–3329. https://doi. 2020. Deep eutectic solvents vs ionic liquids: similarities and
org/10.1002/aic.14537 differences. Microchem. J. July, 159. https://doi.org/10.1016/j.
Li, P., Wu, Y., Hao, X., Zhang, L., 2021. Performance of the gly­ microc.2020.105539
cerol-choline chloride deep eutectic solvent as an entrainer Qiu, X., Qu, Y., Zhou, M., Liu, Y., Zhu, Z., Wang, Y., Yang, J., 2021.
for separation of ethanol and water. J. Chem. Eng. Data 66 (8), Comparison of deep eutectic solvents and organic solvent
3101–3106. https://doi.org/10.1021/acs.jced.1c00207 effects on the separation of ternary azeotropes by the ex­
Liu, P., Hao, J.W., Mo, L.P., Zhang, Z.H., 2015. Recent advances in perimental study and molecular simulation. ACS Sustain.
the application of deep eutectic solvents as sustainable media Chem. Eng. 9 (48), 16424–16436. https://doi.org/10.1021/
as well as catalysts in organic reactions. RSC Adv. 5 (60), acssuschemeng.1c06379
48675–48704. https://doi.org/10.1039/c5ra05746a Ramón, D.J., Guillena, G., 2019. Deep Eutectic Solvents. Wiley
Liu, Q., Zhang, X., 2022. Systematic method of screening deep https://doi.org/10.1002/9783527818488
eutectic solvents as extractive solvents for m-cresol/cumene Rodriguez, N.R., Kroon, M.C., 2015. Isopropanol dehydration via
separation. Sep. Purif. Technol. 291, 120853. https://doi.org/10. extractive distillation using low transition temperature mix­
1016/j.seppur.2022.120853 tures as entrainers. J. Chem. Thermodyn. 85, 216–221. https://
Liu, X., Fu, N., Zhang, Q., Cai, S., Wang, Q., Han, D., Tang, B., 2019. doi.org/10.1016/j.jct.2015.02.003
Green tailoring with water of choline chloride deep eutectic Rodríguez, N.R., González, A.S.B., Tijssen, P.M.A., Kroon, M.C.,
solvents for the extraction of polyphenols from palm samples. 2015. Low transition temperature mixtures (LTTMs) as novel
J. Chromatogr. Sci. 57 (3), 272–278. https://doi.org/10.1093/ entrainers in extractive distillation. Fluid Ph. Equilib. 385,
chromsci/bmy099 72–78. https://doi.org/10.1016/j.fluid.2014.10.044
Luyben, W.L., Chien, I., 2010. Design and Control of Distillation Rogers, R.D., Seddon, K.R., 2003. Ionic liquids–solvents of the fu­
Systems for Separating Azeotropes. Wiley https://doi.org/10. ture? Sciences 302 (5646), 792–793. https://doi.org/10.1126/
1002/9780470575802 science.1090313
Ma, S., Hou, Y., Sun, Y., Li, J., Li, Y., Sun, L., 2017. Simulation and Salleh, Z., Wazeer, I., Mulyono, S., El-blidi, L., Hashim, M.A., Hadj-
experiment for ethanol dehydration using low transition Kali, M.K., 2017. Efficient removal of benzene from cyclo­
temperature mixtures (LTTMs) as entrainers. Chem. Eng. hexane-benzene mixtures using deep eutectic solvents –
Process. Process. Intensif. 121, 71–80. https://doi.org/10.1016/j. COSMO-RS screening and experimental validation. J. Chem.
cep.2017.08.009 Thermodyn. 104, 33–44. https://doi.org/10.1016/j.jct.2016.09.
Mahdi, T., Ahmad, A., Nasef, M.M., Ripin, A., 2015. State-of-the- 002
art technologies for separation of azeotropic mixtures. Sep. Segur, J.B., Oderstar, H.E., 1951. Viscosity of glycerol and its
Purif. Rev. 44 (4), 308–330. https://doi.org/10.1080/15422119. aqueous solutions. Ind. Eng. Chem. 43 (9), 2117–2120. https://
2014.963607 doi.org/10.1021/ie50501a040
Mahiout, S., Vogelpohl, A., 1987. Zur fluiddynamik Und Zum Shahbaz, K., Mjalli, F.S., Vakili-Nezhaad, G., AlNashef, I.M.,
Stoffübergang Newtonscher Und Nicht-Newtonscher Asadov, A., Farid, M.M., 2016. Thermogravimetric
418 Chemical Engineering Research and Design 184 (2022) 402–418

measurement of deep eutectic solvents vapor pressure. J. Mol. Yadav, A., Pandey, S., 2014. Densities and viscosities of (choline
Liq. 222, 61–66. https://doi.org/10.1016/j.molliq.2016.06.106 chloride + urea) deep eutectic solvent and its aqueous mix­
Shang, X., Ma, S., Pan, Q., Li, J., Sun, Y., Ji, K., Sun, L., 2019. Process tures in the temperature range 293.15 K to 363.15 K. J. Chem.
analysis of extractive distillation for the separation of etha­ Eng. Data 59 (7), 2221–2229. https://doi.org/10.1021/je5001796
nol–water using deep eutectic solvent as entrainer. Chem. Zafarani-Moattar, M.T., Shekaari, H., Jafari, P., 2019. Liquid-liquid
Eng. Res. Des. 148, 298–311. https://doi.org/10.1016/j.cherd. equilibria of choline chloride + 1-propanol or 2-propanol +
2019.06.014 water ternary systems at Different temperatures: study of
Sharma, B., Singh, N., Jain, T., Kushwaha, J.P., Singh, P., 2018. choline chloride ability for recovering of these alcohols from
Acetonitrile dehydration via extractive distillation using low water mixtures. J. Mol. Liq. 273, 463–475. https://doi.org/10.
transition temperature mixtures as entrainers. J. Chem. Eng. 1016/j.molliq.2018.10.050
Data 63 (8), 2921–2930. https://doi.org/10.1021/acs.jced. Zhang, H., Lu, X., González-Aguilera, L., Ferrer, M.L., Del Monte,
8b00228 F., Gutiérrez, M.C., 2021. Should deep eutectic solvents be
Sharma, B., Singh, N., Kushwaha, J.P., 2019. Ammonium-based treated as a mixture of two components or as a pseudo-
deep eutectic solvent as entrainer for separation of acetoni­ component? J. Chem. Phys. 154 (18). https://doi.org/10.1063/5.
trile–water mixture by extractive distillation. J. Mol. Liq. 285, 0049162
185–193. https://doi.org/10.1016/j.molliq.2019.04.089 Zhang, L., Han, J., Deng, D., Ji, J., 2007. Selection of ionic liquids as
Sharma, B., Singh, N., Kushwaha, J.P., 2020. Natural deep eutectic entrainers for separation of water and 2-propanol. Fluid Phase
solvent-mediated extractive distillation for separation of Equilib. 255 (2), 179–185. https://doi.org/10.1016/j.fluid.2007.04.
acetonitrile + water azeotropic mixture. J. Chem. Eng. Data 65 016
(4), 1497–1505. https://doi.org/10.1021/acs.jced.9b00932 Zhang, L., Zhang, W., Yang, B., 2014. Experimental measurement
Shu, G., Tan, Y., Cui, L., Zhang, Y., Zhang, L., 2020. Effect of mixed and modeling of ternary vapor-liquid equilibrium for water +
solvents containing ethylene glycol and various salts on the 2-propanol + glycerol. J. Chem. Eng. Data 59 (11), 3825–3830.
vapor-liquid equilibrium of water + methanol + ethanol. J. https://doi.org/10.1021/je500724p
Chem. Eng. Data 65 (6), 3029–3036. https://doi.org/10.1021/acs. Zhang, L., Yang, B., Zhang, W., 2015. Vapor-liquid equilibrium of
jced.0c00057 water + ethanol + glycerol: experimental measurement and
Smith, E.L., Abbott, A.P., Ryder, K.S., 2014. Deep eutectic solvents modeling for ethanol dehydration by extractive distillation. J.
(DESs) and their applications. Chem. Rev. 114 (21), Chem. Eng. Data 60 (6), 1892–1899. https://doi.org/10.1021/acs.
11060–11082. https://doi.org/10.1021/cr300162p jced.5b00116
Souza, G.A.L., Silva, L.Y.A., Martinez, P.F.M., 2021. Vapour-liquid Zhang, L., Wang, X., Zhu, X., Shen, D., 2016. Experimental mea­
equilibria of systems containing deep eutectic solvent based surement and modeling of vapor-liquid equilibrium for the
on choline chloride and glycerol. J. Chem. Thermodyn. 158, ternary systems water + ethanol + ethylene glycol, water + 2-
106444. https://doi.org/10.1016/j.jct.2021.106444 propanol + ethylene glycol, and water + 1-propanol + ethylene
Springer, P.A.M., Baur, R., Krishna, R., 2003. Composition trajec­ glycol. J. Chem. Eng. Data 61 (7), 2596–2604. https://doi.org/10.
tories for heterogeneous azeotropic distillation in a bubble- 1021/acs.jced.6b00264
cap tray column: influence of mass transfer. Chem. Eng. Res. Zhang, L., Zhang, Z., Shen, D., Lan, M., 2017. 2-propanol dehy­
Des. 81 (4), 413–426. https://doi.org/10.1205/ dration via extractive distillation using a renewable glycerol-
026387603765173682 choline chloride deep eutectic solvent: vapor-liquid equili­
Tsanas, C., Tzani, A., Papadopoulos, A., Detsi, A., Voutsas, E., brium. J. Chem. Eng. Data 62 (2), 872–877. https://doi.org/10.
2014. Ionic liquids as entrainers for the separation of the 1021/acs.jced.6b00912
ethanol/water system. Fluid Phase Equilib. 379, 148–156. Zhang, L., Lan, M., Wu, X., Zhang, Y., 2018. Vapor-liquid equilibria
https://doi.org/10.1016/j.fluid.2014.07.022 for 2-propanol dehydration through extractive distillation
Warrag, S.E.E., Kroon, M.C., 2019. Hydrophobic deep eutectic using mixed solvent of ethylene glycol and choline chloride. J.
solvents. Deep Eutectic Solvents Synth. Prop. Appl. 83–93. Chem. Eng. Data 63 (8), 2825–2832. https://doi.org/10.1021/acs.
https://doi.org/10.1002/9783527818488.ch5 jced.8b00162
Wojeicchowski, J.P., Ferreira, A.M., Abranches, D.O., Mafra, M.R., Zhang, Y., Fang, J., Zhang, L., 2019. Isobaric vapor-liquid equilibria
Coutinho, J.A.P., 2020. Using COSMO-RS in the design of deep for the quaternary system water + ethanol + ethylene glycol +
eutectic solvents for the extraction of antioxidants from ro­ choline chloride and the ternary system water + ethanol +
semary. ACS Sustain. Chem. Eng. 8 (32), 12132–12141. https:// choline chloride at 101.3 KPa. J. Chem. Eng. Data 64 (6),
doi.org/10.1021/acssuschemeng.0c03553 2894–2903. https://doi.org/10.1021/acs.jced.9b00254
Xie, Y., Dong, H., Zhang, S., Lu, X., Ji, X., 2014. Effect of water on Zhao, D., Liao, Y., Zhang, Z.D., 2007. Toxicity of ionic liquids.
the density, viscosity, and CO2 solubility in choline chloride/ Clean. Soil Air Water 35 (1), 42–48. https://doi.org/10.1002/clen.
urea. J. Chem. Eng. Data 59 (11), 3344–3352. https://doi.org/10. 200600015
1021/je500320c

You might also like