Phase transitions in higher-melting imidazolium-based ionic liquids: Experiments and advanced data analysis

Article abstract of DOI:10.1016/j.molliq.2019.111222

As thermal energy storage becomes an increasingly important topic, good knowledge of properties of phase-change materials (PCM) is essential. Among other properties, a good PCM should show a large enthalpy of melting, reproducible melting/solidification cycles, and long-term thermal stability. Temperatures and enthalpies of fusion should be determined at a possibly large range of conditions to allow for a critical evaluation of the experimental data and assessment of the material application potential. In this work, imidazolium-based ionic liquids (ILs) with long alkyl chain substituents 1-hexadecyl-3-methylimidazolium chloride and 1-hexadecyl-3-methyl-imidazolium saccharinate were studied in view of their possible use as phase-change materials. Differential scanning calorimetry (DSC) and the heat-leak modulus (HLM) methods were used to determine the temperatures and the enthalpies of phase transitions in the studied ILs, enabling us to study the influence of the heating and cooling rates on the measured properties. Enthalpies of fusion near to or larger than 100 J·g^?1 were found in the studied ionic liquids, making them promising candidates for thermal energy storage. Peaks corresponding to possible liquid crystalline phases in the DSC traces of 1-hexadecyl-3-methylimidazolium saccharinate were observed, pointing to more complex phase behaviour of the studied ionic liquids. Finally, to critically evaluate the experimental data measured in this work, methods based on mathematical gnostics were used. Repeatability of measurements and the degree of mutual agreement between the methods used in this work could thus be determined.

Full text of DOI:10.1016/j.molliq.2019.111222

Accepted Manuscript
Phase transitions in higher-melting imidazolium-based ionic
liquids: Experiments and advanced data analysis
Maja Čanji, Magdalena Bendová, Milen G. Bogdanov, Zdeněk
Wagner, Nikola Zdolšek, François Quirion, Věra Jandová, Pavel
Vrbka
PII:
DOI:
Article Number:
Reference:
S0167-7322(19)30876-1
https://doi.org/10.1016/j.molliq.2019.111222
111222
MOLLIQ 111222
To appear in:
Journal of Molecular Liquids
Received date:
Revised date:
Accepted date:
14 February 2019
4 June 2019
20 June 2019
Please cite this article as: M. Čanji, M. Bendová, M.G. Bogdanov, et al., Phase transitions
in higher-melting imidazolium-based ionic liquids: Experiments and advanced data
analysis, Journal of Molecular Liquids, https://doi.org/10.1016/j.molliq.2019.111222
This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.

ACCEPTED MANUSCRIPT
Phase Transitions in Higher-Melting Imidazolium-Based
Ionic Liquids: Experiments and Advanced Data Analysis
Maja Čanjia,, Magdalena Bendováa, Milen G. Bogdanovb, Zdeněk Wagnera,
Nikola Zdolšekc, François Quiriond, Věra Jandováe, Pavel Vrbkaf
aDepartment of Aerosol Chemistry and Physics, Institute of Chemical Process
Fundamentals CAS, v. v. i., Rozvojova 135/1, 165 02 Prague 6 – Lysolaje, Czech Republic
bFaculty of Chemistry and Pharmacy, Sofia University St Kliment Ohridski, 1 J. Bourchier
blvd., 1164 Sofia, Bulgaria
cCentre of Excellence for Hydrogen and Renewable Energy-CONVINCE, Vinča Institue of
Nuclear Sciences, University of Belgrade, POB 552, 11000, Belgrade, Serbia
dConsultant R&D Physicochimie, 261 Boulevard Pereire, 75017 Paris, France
eDepartment of Laser Chemistry, Institute of Chemical Process Fundamentals CAS, v. v.
i., Rozvojova 135/1, 165 02 Prague 6 – Lysolaje, Czech Republic
fDepartment of Physical Chemistry, University of Chemistry and Technology, Technická 5,
166 28 Prague 6, Czech Republic
Abstract
As thermal energy storage becomes an increasingly important topic, good
knowledge of properties of phase-change materials (PCM) is essential. Among
other properties, a good PCM should show a large enthalpy of melting, repro-
ducible melting/solidification cycles, and long-term thermal stability. Temper-
atures and enthalpies of fusion should be determined at a possibly large range
of conditions to allow for a critical evaluation of the experimental data and
assessment of the material application potential.
In this work, imidazolium-based ionic liquids (ILs) with long alkyl chain sub-
stituents 1-hexadecyl-3-methylimidazolium chloride and 1-hexadecyl-3-methyl-
imidazolium saccharinate were studied in view of their possible use as phase-
change materials. Differential scanning calorimetry (DSC) and the heat-leak
modulus (HLM) methods were used to determine the temperatures and the
enthalpies of phase transitions in the studied ILs, enabling us to study the
∗Corresponding author
Email addresses: canji@icpf.cas.cz (Maja Čanji), bendova@icpf.cas.cz (Magdalena
Bendová)
Preprint submitted to Journal of Molecular Liquids
June 25, 2019

ACCEPTED MANUSCRIPT
influence of the heating and cooling rates on the measured properties. En-
thalpies of fusion near to or larger than 100 J·g−1 were found in the studied
ionic liquids, making them promising candidates for thermal energy storage.
Peaks corresponding to possible liquid crystalline phases in the DSC traces of
1-hexadecyl-3-methylimidazolium saccharinate were observed, pointing to more
complex phase behaviour of the studied ionic liquids.
Finally, to critically evaluate the experimental data measured in this work,
methods based on mathematical gnostics were used. Repeatability of measure-
ments and the degree of mutual agreement between the methods used in this
work could thus be determined.
Keywords: ionic liquids, phase change materials, liquid crystals,
thermodynamic properties, phase transitions, differential scanning calorimetry,
heat-leak modulus, mathematical gnostics
1. Introduction
Limited supplies of fossil fuels along with the efforts towards decreasing the
emissions of greenhouse gases lead to an increasing demand for economically
viable sustainable sources of energy. However, one of the disadvantages of re-
newable energies is their intermittent supply. Therefore, meaningful ways of
storing energy have to be investigated. In thermal energy storage, utilization
of phase-change materials (PCM) is one of the currently explored options [1].
These materials allow for storing heat thanks to their sufficiently high enthalpy
of phase change (>100 J·g−1 ); the material melts during the charging period,
absorbing the heat at a higher temperature, releasing it on crystallization during
the discharging period at a lower temperature. Conventional paraffins, inorganic
salts, or inorganic salt hydrates are the most often used materials [2], ionic liq-
uids are however also considered in energy storage in general for their rather
multipurpose character [3].
Ionic liquids (IL) are broadly defined by a melting temperature lower than
◦
100 C. Even though ILs showing a large liquid range seem to be of the great-
2

ACCEPTED MANUSCRIPT
est interest to both fundamental and applied research, there is a number of
higher-melting ILs solid at room temperature that show a wealth of interesting
thermodynamic and phase transition behavior, for instance liquid crystalline
phases (see e.g. [4, 5]). Some of these ILs even show sufficiently large enthalpies
of melting to be considered for use as phase-change materials (PCMs) [6]. All
this suggests that they could find their use as even more versatile task-specific
materials.
Thermal properties of higher-melting ionic liquids with a common 1-hexadecyl-
3-methylimidazolium cation are therefore studied in this work by means of the
differential scanning calorimetry (DSC) and heat-leak modulus method (HLM)
[7] to determine the influence of the anion on their phase transitions. As a re-
sult of the presence of a long alkyl chain substituent on the cation, the studied
ionic liquids are solids at room temperature, typically showing melting tem-
◦
peratures higher than 50 C and enthalpies of melting close to or larger than
100 J·g−1. Also, as expected, the DSC experiments point towards the occur-
rence of a liquid crystalline phase. In addition, heat capacity as a function of
temperature was measured by means of the DSC in the liquid and solid phase
for the studied ionic liquids. Indeed, previous works addressing phase behaviour
of 1-methyl-3-hexadecylimidazolium chloride demonstrate that the structure of
the crystalline and liquid phases are far from being simple, the phase transi-
tions depending significantly on the conditions of melting and/or crystallization
[8–13]. While the liquid crystalline structures may be very interesting from the
point of view of many specialized applications [14–16], irreproducible melting
may be a hindrance to the utilization of such a material as a phase-change ma-
terial. Understanding their behaviour is therefore of key importance to their
potential application in thermal energy storage.
To be able to assess the application potential of the studied ionic liquids,
methods based on mathematical gnostics (MG) [17] were used in this work to
analyze the experimental data. Indeed, the variety of the experimental methods
used and the different data uncertainty that result from those make it sometimes
difficult to evaluate the data quality. Moreover, when ionic liquids are concerned,
3

ACCEPTED MANUSCRIPT
different levels of impurities, either from synthesis precursors or water that
may be absorbed on sample handling also lead to discrepancies between data
[18, 19]. It was shown in our previous work that mathematical gnostics is a
truly robust method of critical data evaluation [19, 20] especially when only a
small number of data is available (which, truth be told, is most of the times
in physico-chemical measurements). In this work, MG will be used to compare
different measurement methods as well as different experimental conditions in
the same method.
2. Experimental section
2.1. Chemicals used
To synthesize the studied ionic liquids, 1-methylimidazole (99% ReagentPlus,
Sigma Aldrich), 1-hexadecylchloride (95%, Sigma Aldrich), sodium saccharinate
dihydrate (>99.0%, Merck), N,N-dimethylformamide anhydrous (99.8%, Sigma
Aldrich), acetone anhydrous (≥99.9%, Merck), ethylacetate (p.a., LachNer),
and in-house doubly-distilled water were used. Biphenyl (99 %, Alfa Aesar) was
used as calibration standard for the HLM measurements, whereas a naphthalene
standard provided by Setaram and tap water were used to calibrate the Setaram
μDSC III evo device. Distilled water treated by a Milli-Q Water Purification
System (Millipore, USA), gallium, naphthalene, indium, and tin provided by
GEFTA c/o Merck, Darmstadt, Germany, were used as calibration standards
for the TA Q1000 device.
Synthesis of 1-hexadecyl-3-methylimidazolium chloride (abbrev.
[C16C1im]Cl)
For the preparation of [C16C1im]Cl [18], 23.81 g (0.29 mol) of 1-methyl-
imidazole and 91.31 g (0.35 mol) of 1-hexadecylchloride were dissolved in 50 mL
N,N-dimethylformamide in a round-bottomed flask equipped with a reflux con-
denser. The mixture was continuously stirred and heated at 90 ◦C for 70 h
under inert argon atmosphere. The resulting brownish viscous liquid readily
4

ACCEPTED MANUSCRIPT
crystallized upon cooling giving pale yellow crystals. To obtain IL of higher
purity, the solvents were evaporated under reduced pressure and the resulting
residue was crystallized several times from ethyl acetate. The final product was
obtained as white crystals in 86% yield and was further dried in vacuo (3 mbar)
for 48 h. The water content after drying was determined by the Karl-Fischer
titration and was typically 1% (mass). Its structure was confirmed by means of
NMR spectroscopy [10].
Synthesis of 1-hexadecyl-3-methylimidazolium saccharinate (ab-
brev. [C16C1im][Sac])
[C16C1im][Sac] was prepared by ion exchange reaction from [C16C1im]Cl and
sodium saccharinate (NaSac) as precursors. A methanol solution of [C16C1im]Cl
(20 g, 0.058 mol, in 109 mL) was slowly added (dropwise) to a stirred solution
of NaSac (22 g, 0.099 mol) in 20 g of methanol at 60 ◦C. The flask was equipped
with a magnetic stirrer, a dropping funnel, a reflux condenser and a calcium
chloride tube. The reaction mixture was stirred for 24 h under reflux. The
solvent was then evaporated under reduced pressure giving a suspension. This
residue was treated with acetone in order to precipitate the NaCl and unre-
acted Na[Sac], while keeping [C16C1im][Sac] in solution. The inorganic solids
were further filtered off and the organic solvent was removed under reduced
pressure. The above procedure was repeated several times until no inorganic
precipitates in the acetone were observed. The IL obtained, [C16C1im][Sac],
was subsequently crystallized from ethyl acetate (yield 90%) and dried in vacuo
(approx. 3 mbar) for 48 h. The water content after drying was determined
by the Karl-Fischer titration and was was typically 1% (mass). Its structure
was confirmed by NMR spectroscopy: 1H NMR (400.13 MHz, CD3OD): δ =
0.88 (3H, t, J = 7.3, CH2CH3), 1.23–1.34 (26H, m, CH2), 1.82–1.90 (2H, m,
CH2), 3.91 (3H, s, N–CH3), 4.18 (2H, t, J = 7.4, N–CH2), 7.54 (1H, t, J = 1.8,
CH/im), 7.61 (1H, t, J = 1.8, CH/im), 7.64–7.69 (2H, m, CH/Sac), 7.72–7.82
(2H, m, CH/ Sac), 8.91 (1H, bs, N–CH=N).
5

ACCEPTED MANUSCRIPT
2.2. Experimental methods used
Differential scanning calorimetry
Phase transition temperatures and enthalpies were determined by means of
a Setaram μDSCIII evo microcalorimeter under N2 atmosphere. Temperature
calibration of the calorimeter was performed using the melting points of water
and naphthalene, and the energetic calibration was carried out using the Joule
effect. Tap water was used as calibration standard, according to the recommen-
dation of the calorimeter manufacturer, to avoid subcooling that often occurs
in distilled water. The naphthalene standard was provided by the manufac-
turer. Cooling was carried out at 0.6 ◦C·min−1. Heating was carried out at
0.2, 0.5, and 1 ◦C·min−1, respectively. Masses of both calibration and ionic
liquid samples were of 5 to 35 mg. The temperature ranges in which the exper-
iments were performed were determined for each sample individually according
to the temperatures at which crystallization, fusion, or other phase transitions
occurred. The experiments were carried out in the temperature range 25–90 ◦C
for [C16C1im]Cl, whereas for [C16C1im][Sac] the temperature ranges used were
◦
◦
◦
◦
(25–115) C, (40–115) C, (25–80) C, and (60–115) C, respectively. The un-
certainty in determination of temperature of phase transition was found to be
◦
better than 0.3 C and in enthalpy of phase transition better than 1.7 J·g−1
.
To measure phase transitions in [C16C1im][Sac] at higher temperature rates,
the differential scanning calorimeter, Model Q1000 (TA Instruments, USA) was
used. The measurements were carried out in the temperature range from 183
◦
◦
to 383 C at heating rates of 1, 2, 5, and 10 C· min−1. The samples of 5 to
10 mg were placed in hermetic aluminum pans. The calorimeter was periodi-
cally calibrated with respect to temperature and enthalpy using water, gallium,
naphthalene, indium, and tin. The estimated combined expanded uncertainty
◦
of temperature measurements is uc(T) = 0.3 C and the relative combined un-
certainty of enthalpy determination is ur(∆H) = 0.01 %.
Heat-leak modulus method
To further investigate the influence of the heating and cooling rates on
6

ACCEPTED MANUSCRIPT
the measured phase transitions, the temperatures and enthalpies of fusion for
[C16C1im]Cl were also determined by the heat-leak modulus (HLM) method [7]
using an AT-HLM prototype provided by François Quirion. The HLM is a mea-
sure of the heat transfer between the environment and the sample. The sample
is mounted on a temperature probe (a Pt-100 Omega PR11 thermometer, di-
◦
ameter 1/8 in (3.18 mm) with a precision of ±0.01 C, calibrated according to
ITS-90) and inserted into a reservoir kept at a constant temperature TR. The
temperature of the sample, TS , is then determined as a function of time. From
this dependence the HLM is calculated from Eq.1:
dTS/dt
HLM =
(1)
TR − TS
The HLM can be determined both on heating (TR>TS ) or cooling (TS >TR).
In the absence of an enthalpic transition, the HLM is a monotonous function
of TS . An enthalpic transition results in a peak from which the temperature
and enthalpy of the phase transition can be determined, the temperature of the
transition corresponding to the peak maximum and the enthalpy to the area of
the peak.
An assembly of aluminium foil and fiberglass grid was used to mount the
sample to the temperature probe. This assembly was calibrated with biphenyl
as standard. To obtain high heating rates, samples of approx. (70–80) mg were
exposed to an oven temperature around 150 ◦C. They were then taken out of the
◦
oven at temperatures around 90 C and transferred into a cold chamber. The
cold chamber was cooled either by a mixture of dry ice and ethanol, in which
case its temperature was around –70 ◦C, or by crushed ice i.e. to a temperature
of approx. 0 ◦C. This enabled us to determine the influence of the cooling
rate on the phase transitions of the sample. The sample was removed as it
reached about 0 ◦C and the cycle was repeated. Under the conditions described
◦
above, the heating rates were typically of (14–26) C·min−1 when the sample
was at 66 ◦C, whereas the cooling rates were about 13 ◦C·min−1 for cooling
with crushed ice and (30–40) ◦C·min−1 for cooling with the dry ice/ethanol
7

ACCEPTED MANUSCRIPT
mixture, respectively, when the sample was at 50 ◦C. These rates are much
higher than the rates investigated by DSC (heating at 0.2–10 ◦C·min−1 and
◦
cooling at 0.6 C·min−1). It should, however, be noted, that as opposed to the
DSC the heating and cooling rates on the HLM experiments are not constant,
they are continuously decreasing as the sample approaches the temperature of
the reservoir. To enable us to compare the influence of the heating and cooling
rates as precisely as possible for the DSC and the HLM measurements, for the
latter method these rates are reported at the temperature of the transition.
Thermal gravimetric analysis
Finally, to determine the full liquid range of the studied ionic liquids, de-
composition temperatures were determined by the thermal gravimetric analysis,
a sensitive technique for determination of thermal stability of the analysed mate-
rials. Thermal decomposition of measured materials was conducted by heating
◦
◦
samples of mass of several miligrams to 600 C at the rate of 2 C·min−1, using
a NETZSCH TG 209 F1 Libra recording microbalances in a stream of inert
gas (Ar) at the flow rate 50 mL·min−1. The sample was heated in the oven
whose temperature was regulated by programmable digital temperature con-
troller. The flow of carrier gas was controlled by electronic mass flowmeters. In
addition to the TGA and DTG (first derivative of the TGA) curves, endother-
mic and exothermic effects determined by means of the calculated DTA signal
(c-DTA) can also be obtained.
3. Data Analysis
Mathematical gnostics (MG) [17] provides us with a non-statistical approach
towards the uncertainty treatment. In statistics, the measurement uncertainty
is estimated from a series of repeated measurements using an a priori assump-
tion of the error distribution function. Since the underlying theory of MG was
derived considering the uncertainty of every single measurement, the properties
of a data sample are then evaluated by aggregation of properties of each mea-
surement as opposed to statistics. Small data samples can thus be analyzed
8

ACCEPTED MANUSCRIPT
and a sound estimate of the experimental uncertainty obtained. In this work,
marginal analysis was used to determine the repeatability of the experimental
data. As in statistics, the task of gnostic data analysis is to find a single value
representing the whole data sample and to estimate its uncertainty. First, the
existence of the estimating global distribution function (EGDF) is verified. If it
exists, the data sample can be replaced with a single value and its uncertainty
and the data are assumed homogeneous. If the EGDF does not exists, the data
sample is split into clusters. The data sample is nonhomogeneous and the data
outside the main cluster, considered as outliers, are discarded. Further analysis
can only be performed for a homogeneous subsample. Presence of outliers means
that these data represent a different quantity (in our case a value corresponding
to a different crystalline phase), or are affected by a hystheresis effect, or an
unusual source of measurement error occurred, or any combination of them.
To assess the characteristic value and its uncertainty we construct the esti-
mating local distribution function (ELDF). The mode of the ELDF’s density
(Z0), i.e. the location of the density’s maximum, is the most probable value
of the data sample and is considered as an estimate of the true value of the
measured quantity. The upper and lower bounds of the tolerance interval Z0U
and Z0L, respectively, then delimit the tolerance interval meaning the measure-
ment repeatability, while the upper and lower bounds of the interval of typical
data ZU and ZL then delimit the interval in which the data can typically be
expected (see e.g. [19] for greater details on the use of marginal analysis in
assessing thermodynamic data). An experimental value that falls outside the
interval of typical data is then considered as dubious.
If the tolerance intervals are symmetric or only slightly asymmetric, the
repeatability of the measurements can be evaluated as
Z0U − Z0L
u =
(2)
2
where Z0U and Z0L are the upper and lower bounds of the tolerance interval,
respectively.
9

ACCEPTED MANUSCRIPT
4. Results and Discussion
1-hexadecyl-3-methylimidazolium chloride
Table 1 lists the temperatures and enthalpies of transitions determined by
means of DSC (Setaram) and HLM for [C16C1im]Cl. The listed values are the
ELDF’s modes and the repeatability intervals were obtained from Eq. 2. Raw
data are then listed in Table SI1 of the Electronic Supporting Material. Figure 1
shows examples of the DSC traces obtained on heating at a rate of 0.2 ◦C·min−1
◦
and on cooling at a rate of 0.6 C·min−1. The ionic liquid [C16C1im]Cl shows a
◦
single endothermic peak in the temperature range of temperatures (18–90) C.
10

Table 1: Phase transitions for the ionic liquid [C16C1im]Cl. Temperatures of fusion, tf ; enthalpies of fusion, ∆Hf ; temperatures of cold crystallization;
tcc, enthalpies of cold crystallization; ∆Hcc, temperatures of solidification; ts, enthalpies of solidification; ∆Hs
method
HLMa
HLMb
DSCc
tf /◦C
∆Hf /J·g−1
150±1
149.6±0.9
159±1
ts/◦C
∆Hs/J·g−1 tcc/◦C ∆Hcc/J·g−1
65.96±0.04
63.99±0.06
63.44±0.05
53.9±0.7
51.09±0.09 –131.6±0.5
52.1±0.6 –151.2±0.6
–112±5
44±1
–19±1
.
a
b
c
heating rates 21–24 ◦C·min−1, cooling rates 33–108 ◦C·min−1
heating rates 14–26 ◦C·min−1, cooling rates 12–13 ◦C·min−1
device used: Setaram μDSC III evo, heating rates 0.2–1 ◦C·min−1, cooling rate 0.6 ◦C·min−1
11

(a)
(b)
Figure 1: Examples of DSC scans for [C16C1im]Cl. a) Heating at 0.2 ◦C·min−1. b) Cooling at 0.6 ◦C·min−1
.
12

(a)
(b)
Figure 2: Examples of HLM heating scans for [C16C1im]Cl. Black lines, melting of the native salt. Red lines, melting of the solidified melt. a)
Cooling rate preceding melting: 12 ◦C·min−1. b) Cooling rate preceding melting 38 ◦C·min−1
13

ACCEPTED MANUSCRIPT
Table 1 shows that the temperature of fusion of [C16C1im]Cl determined by
◦
DSC is of 63.44±0.05 C while its enthalpy of fusion is of 159±1 J·g−1. The
◦
temperature and enthalpy of solidification were then found to be 52.0±0.5 C
and –151.2±0.6 J·g−1, respectively. HLM experiments then provide the melting
◦
temperature of 65.96±0.04 C for experiments carried out at heating rates 21–
24 ◦C·min−1 and cooling rates 33–108 ◦C·min−1, respectively, and of 63.99±0.06 ◦C
◦
for experiments carried out at heating rates 14–26 C·min−1 and cooling rates
12–13 ◦C·min−1, respectively. Cooling rates seem to have very little influence on
the enthalpy of melting determined by the HLM method, since for both modali-
ties of measurements it was found to be 150±1 J·g−1. On solidification, the HLM
experiments carried out at larger cooling rates of 38 ◦C·min−1 yield the temper-
◦
ature and enthalpy of solidification of 53.9±0.8 C and –112±5 J·g−1, showing
in addition a cold crystallization at 44±1 ◦C of an enthalpy of –19±2 J·g−1
.
Figure 1 represents examples of DSC scans on heating at 0.2 ◦C·min−1 and
cooling at 0.6 ◦C·min−1. Figure 2 then shows examples of HLM heating scans
that show melting following two different cooling rates, 12 and 38 ◦C·min−1
,
respectively. It appears that the cold crystallization observed on HLM heating
experiments following a higher cooling rate (see Figure 2b) was neither observed
on HLM nor DSC heating experiments carried out following lower cooling rates
(Figures 2a and 1, respectively). Indeed, higher cooling rates often lead to more
substantial subcooling and incomplete crystallization. It is also interesting to
note that the first heating leads to a significantly higher enthalpy of melting
than that obtained from the solidified melt, e.g. 186 J·g−1 on the first heating
scan vs. 154–159 J·g−1 on subsequent heating scans in the DSC experiment,
whereas the enthalpy of the first crystallization appears to be lower than the
values obtained in the subsequent scans (see Table SI1 of the Electronic Sup-
porting Material). The same pattern was observed on HLM experiments carried
out at lower cooling rates, but not on higher cooling rates.
To better understand the patterns in the observed phenomena and compare
the individual datasets obtained by different measurement methods, marginal
analysis was carried out. Figure 3 shows the results of the marginal analysis
14

ACCEPTED MANUSCRIPT
of temperatures and enthalpies of fusion for [C16C1im]Cl, the error bars repre-
senting the tolerance interval and the interval of typical data, respectively. It
is evident that the agreement between the enthalpies of fusion measured by the
HLM method at different heating rates are in very good agreement, the mea-
sured values mutually falling into the respective tolerance intervals, intervals of
typical data overlapping. The enthalpies measured by the DSC are in relative
good agreement with the HLM, the overlap occurring only between the inter-
vals of typical data. Good agreement was found for the temperatures of fusion
measured at very low heating rates by means of the DSC and those measured
by the HLM at intermediate heating and cooling rates, whereas temperatures of
fusion measured at high heating rates by means of the HLM are more than 2 ◦C
higher. Heating and cooling rates indeed do influence the values of the phase
transition characteristics measured by the DSC. As the rate of the temperature
change increases, the measured temperature of phase change may depart from
the true equilibrium temperature, due to slower kinetics of the phase transition
[21]. The temperatures of melting thus significantly increase with increasing
heating rate.
It also appears from the marginal analysis that the enthalpies and temper-
atures, respectively, of the first and second melting/solidification were found to
lie outside the interval of typical data, showing that their values correspond to
a different type of transition than those observed on subsequent scans. This is
actually common in DSC measurements. Firstly, during the first heating scan
equilibrium in heat transfer between the sample and the measurement cell mate-
rial is still not established. Temperatures measured on the first heating are then
usually higher than those observed on the subsequent scans which was indeed
observed in this work. In addition, since a fresh sample of an ionic liquid re-
crystallized from ethylacetate was always used, the first heating run corresponds
to melting of the stable crystalline phase. On subsequent cooling, the samples
may not recrystallize completely into the most stable solid phase, depending on
the kinetics of the phase transfer between the individual crystalline modifica-
tion. Consequently, the enthalpies of fusion shift as a function of the crystalline
15

ACCEPTED MANUSCRIPT
Figure 3: Results of marginal analysis for temperatures and enthalpies of melting of
[C16C1im]Cl. Red and magenta error bars are tolerance intervals, blue and brown error
bars are intervals of typical data. HLM_1, heating rates 21–24 ◦C·min−1, cooling rates 33–
108 ◦C·min−1; HLM_2, heating rates 14–26 ◦C·min−1, cooling rates 12–13 ◦C·min−1; DSC,
heating rates 0.2–1 ◦C·min−1, cooling rate 0.6 ◦C·min−1
.
16

ACCEPTED MANUSCRIPT
phase formed. The enthalpies of fusion and solidification measured by the DSC
on these subsequent scans were well balanced for [C16C1im]Cl (∆H f =159±1
J·g−1 vs. ∆H s =–151.2±0.06 J·g−1), suggesting that at very low cooling rates
the solid that formed is the one that melts and a formation of a phase at equilib-
rium. This can actually be a mixture of stable and metastable crystalline phase
◦
as shown in [8]. At intermediate cooling rates (i.e. 12–13 C·min−1) used on
HLM experiments, the solidification enthalpy of −131.6±0.5 J·g−1 corresponds
to the sum of the solidification and solid-solid transition enthalpy measured at
higher cooling rates.
As shown in [9–13], [C16C1im]Cl also shows a transition from smectic liquid
◦
crystalline phase to isotropic phase at about 220 C, however, the temperature
range of the Setaram µDSC III evo device used in this work did not allow
for the determination of this transition. Li et al. [8] further demonstrated
that the apart from the stable triclinic crystalline phase, the solid phase may
show other transient crystalline modifications, their occurrence depending on
the state of the ionic liquid before the solidification process. For example, on
slow cooling at 1 ◦C·min−1, i.e. at conditions used in the DSC experiments,
the triclinic phase was found to appear only after the sample undergoes the
transient phase. Fast quenching from a temperature below the isotropization
temperature, i.e. at conditions similar to the HLM experiment, then leads to
the occurrence of the stable triclinic phase, but also partially of an unstable
perpendicular double bilayer. In addition, the type of the crystalline phase was
shown to depend not only on the cooling rate, but also on the initial state of the
studied IL. The growth of the stable triclinic phase was thus found to be difficult
when crystallization started from the ordered smectic phase. The difference of
the measured enthalpies seems to corroborate this. For instance, on the HLM
experiment carried out on 7 September 2018 (see Table SI1 of the Electronic
◦
Supporting Information) the first cooling rate was of 108 C·min−1 and led to
a significantly lower enthalpy of melting than on the subsequent cooling scan
carried out at 38 ◦C·min−1. This value was then detected by the marginal
analysis as far outside the interval of typical data. The arrangement of ions of
17

ACCEPTED MANUSCRIPT
the native ionic liquid on the first melting thus appears to be different than that
obtained on the subsequent heating scans from the solidified melt.
Both the temperature and enthalpy of fusion found in this work are larger
than those in [10] who report the temperature of fusion of 56.9 ◦C and the
enthalpy of fusion of 61.8 J·g−1 for an ionic liquid of water content of up to
10.6 wt.% as obtained from TGA, however, no information is given in their
study on the way the samples were treated prior to the DSC experiments and
consequently the water content of their IL. Goujon et al. [11] give a higher
◦
melting temperature for [C16C1im]Cl of 69.4 C, however, the water content of
their IL is of 4.8 wt.%. It has been noted [12–14] that hydrated salts show higher
melting temperatures and melting enthalpies than their anhydrous counterparts,
the discrepancy between our and literature data can therefore be attributed to
differences in water contents in the samples studied in this work. For instance
Goujon et al. [11] have only dried their sample under vacuum for several hours,
whereas in this work, drying was carried out for at least 48 hours. In addition,
the influence of the water content on the specific phase behaviour of [C16C1im]Cl
has not yet been quantified.
1-hexadecyl-3-methylimidazolium saccharinate
Table 2 lists the experimental temperatures and enthalpies of phase transi-
tions for [C16C1im][Sac] determined by the DSC. The listed values and the re-
peatability intervals were obtained by means of the marginal analysis, whereas
raw data are listed in Table SI3 of the Electronic Supporting Information. Two
heat-flux DSC devices were used to measure phase transitions in this ionic liq-
uid. The Setaram µDSC III evo device is a Calvet-type microcalorimeter with
a 3D temperature temperature control and sensor whereas the TA Q1000 DSC
is a 2D plate-like instrument in which samples are heated from below. The
former device was used to obtain data denoted as DSC_1 and DSC_2, while
the latter was used to measure data denoted as DSC_3 and DSC_4, in Table
SI3 respectively.
Figure 4 shows examples of a heating and cooling scan, respectively, for
18

ACCEPTED MANUSCRIPT
[C16C1im][Sac] measured by Setaram µDSC III evo (experiments DSC_1 and
DSC_2 in Table SI3 of the Electronic Supporting Information). In these exper-
iments, the scans were divided into several shorter temperature ranges. In the
first experiment, the characteristics of the phase transition were measured in the
range of (40–115) ◦C and (25–115) ◦C, respectively. The second experiment was
◦
◦
carried out at (25–80) C and (60–115) C, respectively. The latter experiment
enabled us to separate the measurements of the temperatures and enthalpies of
fusion from measurements of the isotropization temperature. Indeed, apart from
a large peak corresponding to melting/solidification, both experiments showed
a small endothermic peak of a small enthalpy of transition in the latter tem-
perature region. Moreover, this transition was found to be perfectly reversible,
with the same temperature of transition on heating as well as cooling runs.
This is an indication of a presence of a liquid crystalline phase [21]. The tem-
perature of the endothermic transition from the liquid crystalline phase to the
◦
isotropic phase observed on the first set of experiments changed from 87.4 C
◦
to 95.5 C, showing the enthalpy of transition of about 1 J·g−1. Based on the
above discussion, the presence of a liquid crystalline phase is thus concluded on,
the temperature at which the phenomenon is observed being the temperature
of isotropization, i.e. the transition from a liquid crystal to the isotropic phase.
Similarly to the temperature of fusion, the isotropization temperature seems to
depend on the phase from which the sample undergoes change. Indeed, for the
◦
first experiment starting at 40 C, the first temperature of isotropization shows
a slightly higher value of 92.30 ◦C than the subsequent ones, finally coming
down to approx. 86 ◦C for the same sample measured at the temperature range
◦
◦
of (25–115) C. First, the sample heated from and re-cooled to 40 C may have
not been fully solidified, which explains a larger scatter between the measured
data. Then, the sample may have gradually equilibrated in the liquid crystalline
sample to an ordered state that remains the same after several scans. This is
corroborated by the fact that the first value of the isotropization temperature
lies just outside the interval of typical data and could be thus considered as
slightly dubious. For the second experiment, in which the sample was heated
19

ACCEPTED MANUSCRIPT
◦
from 60 C the sample was never resolidified, the liquid phase thus remaining
in the same type of liquid crystalline ordering throughout the experiment.
20

Table 2: Temperatures and enthalpies of phase transitions for the ionic liquid [C16C1im][Sac] measured by the DSC. Temperatures of fusion, tf ;
enthalpies of fusion, ∆Hf ; isotropization temperatures, tLC ; enthalpies of isotropization, ∆HLC ; temperatures of solidification, ts; enthalpies of
solidification, ∆Hs
.
◦
◦
method
DSCa
DSCb
DSCc
tf /◦C
∆Hf / J · g−1 tLC / C ∆HLC / J · g−1
ts / C
∆Hs / J · g−1
–91.2±0.6
–105.2±0.1
–102.2±0.5
50.99±0.07
53.2±0.2
54.71±0.03
98±2
87.4±0.3
95.5±0.1
96.2±0.1
0.930±0.004
1.005±0.003
1.18±0.06
38.88±0.06
41.01±0.03
39.4±0.1
108.3±0.7
109.6±0.2
a
device used: Setaram μDSC III evo, heating rates 0.2–1 ◦C·min−1, cooling rate 0.6 ◦C·min−1, temperature range (40–115) ◦C and (25–115) ◦C
device used: Setaram μDSC III evo, heating rates 0.2–1 ◦C·min−1, cooling rate 0.6 ◦C·min−1, temperature range (25–80) ◦C and (60–115) ◦C
device used: TA Q1000 DSC, heating rates 1–10 ◦C·min−1, cooling rate undefined, temperature range (–90–110) ◦C
b
c
21

(a)
(b)
Figure 4: Examples of DSC scans for [C16C1im][Sac]. a) Heating at 0.2 ◦C·min−1. b) Cooling at 0.6 ◦C·min−1
.
22

ACCEPTED MANUSCRIPT
It appears that the data obtained by the μDSC III evo show that the sac-
charinate ionic liquid phase transitions depend on the range of temperatures
in which the measurements were performed. The samples heated above the
isotropization temperature show melting temperatures higher than those mea-
sured for the samples solidified from the temperature below the isotropization
temperature. Figure 5 shows the results of the marginal analysis of tempera-
tures and enthalpies of fusion. It appears that the tolerance intervals of the
enthalpies and temperatures of melting for these two types of measurements do
not overlap, confirming that the datasets correspond indeed to different phe-
nomena. Similarly to [C16C1im]Cl the degree of the liquid phase ordering seems
to influence the crystalline modification obtained. In the absence of relevant
diffraction data it can only be assumed that the most stable phase has the
highest melting temperature.
On the experiment carried out in the ranges of temperatures (40–115) and
(25–115) ◦C two endothermic phenomena were observed, a large peak at 50.99 ◦C
of 98 J·g−1 corresponding to the melting of the ionic liquid into a liquid crystal
mesophase. Similarly to [C16C1im]Cl, the first enthalpy of fusion is significantly
larger than the following, ones, however, the lower temperature of the scan
(40 ◦C) is too close to the temperature of fusion, not allowing for a proper reso-
lidification of the sample. Consequently, both the temperatures and enthalpies
of fusion obtained on this first experiment all lie outside the tolerance interval
and almost all of them lie also outside the interval of typical data. The dataset
of temperatures of fusion is in addition found to be non-homogeneous. In com-
parison with [C16C1im]Cl, the bulkier saccharinate anion contributes to lower
temperatures and enthalpies of fusion. The transition from the liquid crystalline
phase to isotropic liquid also occurs at a lower temperature for the IL with the
saccharinate anion than for the chloride.
Similarly as for the [C16C1im]Cl and in keeping with the usually observed
kinetics of melting, the temperature of fusion of [C16C1im][Sac] increases with
the increasing heating rate. This is obvious from the data acquired by means
of the TA Q1000 DSC device at rates ranging from (1–10) ◦C·min−1 listed
23

ACCEPTED MANUSCRIPT
Table 3: Temperatures and enthalpies of phase transitions for the ionic liquid [C16C1im][Sac]
measured by HLM. Temperatures of fusion, tf , enthalpies of fusion; ∆Hf ; temperatures of
solidification, ts; enthalpies of solidification, ∆Hs
◦
method
HLMa 58.3±0.8
a
tf /◦C
∆Hf / J · g−1
111.7±0.6
ts / C
∆Hs / J · g−1
-103±3
40.9±0.4
heating rates 21–24 ◦C·min−1, cooling rates 30–40 ◦C·min−1
for experiments DSC_3 and DSC_4 in Table SI3 of the Electronic Supporting
Information and from the HLM experiments (21–24) ◦C·min−1 listed in Ta-
ble SI2 of the Electronic Supporting Information. The temperature of fusion
◦
found using the TA Q1000 DSC was found to be of 54.7±0.03 C. The samples
◦
were in this case mostly cooled down to –90 C (except for one case when the
◦
sample was cooled down to 10 C on the last repeated measurement run), pro-
viding sufficient conditions for the complete crystallization of the sample. The
isotropization temperature was then found to be higher than that measured by
the Setaram μDSC III evo, but closer to that obtained in samples cooled to
◦
◦
25 C than in those cooled to 45 C only, corroborating our assumption that
it depends largely on the phase from which the sample undergoes the change.
Marginal analysis showed that the datasets of the temperature of fusion and
isotropization, respectively, were homogeneous, pointing to a good repeatability
of the measurements.
On HLM experiments heating at (21–24) ◦C·min−1 and cooling at 30–77 ◦C·min−1
was used. Table 3 then lists the enthalpies and temperatures of melting and
solidification. The listed values and the repeatability intervals were obtained
by means of the marginal analysis, whereas raw data are listed in Table SI2
of the Electronic Supporting Information. The measured data are in relative
good agreement with the DSC scans carried out at larger heating rates using
the TA Q1000 instrument, the intervals of typical data overlapping for both
measurements.
,
It appears from Figure 5 that regardless of the method used, the intervals
of typical data overlap for all the enthalpies of melting. The influence of the
24

ACCEPTED MANUSCRIPT
Figure 5: Results of marginal analysis for temperatures and enthalpies of melting of
[C16C1im][Sac]. Red and magenta error bars are tolerance intervals, blue and brown error
bars are intervals of typical data. HLM_1, heating rates 21–24 ◦C·min−1, cooling rates 30–
40 ◦C·min−1; DSC_1, heating rates 0.2–1 ◦C·min−1, cooling rate 0.6 ◦C·min−1, temperature
range (10–80) ◦C; DSC_2, heating rates 0.2–1 ◦C·min−1, cooling rate 0.6 ◦C·min−1, tem-
perature range (60–115) ◦C; DSC_3, heating rates 1–10 ◦C·min−1, cooling rate undefined,
temperature range (–90–110) ◦C, DSC_12, marginal analysis of all data from experiments
DSC_1 and DSC_2.
25

ACCEPTED MANUSCRIPT
temperature range in which the scans were performed is also made evident
for the data measured at low heating rates; neither the tolerance interval nor
the intervals of typical data overlap for the temperatures of fusion obtained in
DSC experiments in which sample was solidified from temperatures below the
isotropization temperature tLC, i.e. from a highly ordered liquid, and those
that were solidified from the melt above tLC. However, the cooling rate may
have a strong influence on the solidification of the sample. For instance, in the
HLM experiment carried out at the cooling rate of 77 ◦C·min−1, the enthalpy of
solidification is again much lower than that obtained at the subsequent cooling
of 38 ◦C·min−1 (see Table SI2 of the Electronic Supporting Material). The value
of the enthalpy was therefore detected by the marginal analysis as lying outside
the interval of typical data. Some rearrangement of the liquid phase may thus
occur, similarly as for the [C16C1im]Cl discussed above.
To determine the full liquid range of the studied ionic liquids, thermal gravi-
metric analysis was used to measure the decomposition temperatures. The ther-
◦
mogrammes obtained at a heating rate of 2 C·min−1 for native salts without
any previous drying are shown in Figures SI1 and SI2 of the Electronic Sup-
porting Information. It appears that at temperatures below 100 ◦C [C16C1im]Cl
and [C16C1im][Sac] show a mass loss of 5.65 wt.% and 1.13 wt.%, respectively.
This mass loss corresponds to the loss of water contained in samples that had
only been kept in the dessicator prior to the measurements. Considering that
drying the studied ionic liquids under reduced pressure for 48 hours leads to the
water content in both samples lower than 1 wt.% it can be concluded from these
results that the saccharinate IL is less prone to absorbing air moisture than chlo-
◦
ride. The chloride IL then decomposes at 262.7 C, whereas the saccharinate
◦
IL decomposes at 285.5 C. The data acquired for [C16C1im]Cl correspond to
those by Thomas et al. [10] who found a similar mass loss of 5.1 % below 100 ◦C
and Goujon et al. [11] who state that it was very difficult to dry the IL to water
levels lower than 4.8 wt. %.
26

ACCEPTED MANUSCRIPT
5. Conclusion
Thermal properties of two higher melting ionic liquids with a common 1-
hexyl-3-methylimidazolium cation and a chloride or saccharinate anion were
investigated by means of the DSC and HLM methods in view of their possible
use as phase-change materials. The enthalpies of melting of 159 J·g−1 and
(98 and 108) J·g−1 for [C16C1im]Cl and [C16C1im][Sac], respectively, seem to
indicate that both ILs could be considered as promising PCMs. This, however,
remains to be confirmed by further feasibility studies and life-cycle analysis.
The influence of the water content on the phase behaviour of the studied ILs
also needs to be quantified, to evaluate changes to the material after a possible
air moisture uptake.
The studied ILs also appear show a more complex phase behavior than simple
melting/crystallization. Evidence of a liquid crystalline mesophase has already
been given by other authors for [C16C1im]Cl. The [C16C1im][Sac] ionic liquid
also seems to form liquid crystals. Moreover, the mechanisms of melting and
crystallization appear to be rather complex, showing that not only the rates at
which the samples are heated or cooled but also that the phase from which the
sample is solidified seem to influence rather significantly their phase transition
characteristics. These phenomena should be thoroughly understood before these
materials are considered as PCMs in a real application.
Acknowledgments
The authors gratefully acknowledge the Czech Science Foundation under
Grant No. 17-08218S (MB) and the Ministry of Education, Science and Tech-
nological Development of Republic of Serbia within project No. III45006 (NZ).
The authors are also grateful to Dr. Lucie Červenková-Šťastná for the measure-
ments of the NMR spectra.
27

ACCEPTED MANUSCRIPT
References
References
[1] H. Zhang, J. Baeyens, G. Cáceres, J. Degrève, Y. Lv,
Thermal energy storage: Recent developments and practical aspects,
Progress in Energy and Combustion Science 53 (2016) 1–40.
doi:10.1016/j.pecs.2015.10.003.
URL http://dx.doi.org/10.1016/j.pecs.2015.10.003
[2] K. Pielichowska, K. Pielichowski, Phase change materials for thermal energy storage,
Progress
doi:10.1016/j.pmatsci.2014.03.005.
URL http://dx.doi.org/10.1016/j.pmatsci.2014.03.005
in
Materials
Science
65
(2014)
67–123.
[3] D. R. MacFarlane, M. Forsyth, P. C. Howlett, M. Kar, S. Passerini, J. M.
Pringle, H. Ohno, M. Watanabe, F. Yan, W. Zheng, S. Zhang, J. Zhang,
Ionic liquids and their solid-state analogues as materials for energy generation and storage,
Nature
doi:10.1038/natrevmats.2015.5.
URL http://dx.doi.org/10.1038/natrevmats.2015.5
Reviews
Materials
1
(2)
(2016)
15005+.
[4] P. Zurner, H. Schmidt, S. Bette, J. Wagler, G. Frisch,
Ionic liquid, glass or crystalline solid? Structures and thermal behaviour of (C4mim)2CuCl3,
Dalton Trans. 45 (8) (2016) 3327–3333. doi:10.1039/c5dt03772g.
URL http://dx.doi.org/10.1039/c5dt03772g
[5] K.
Lava,
K.
Binnemans,
T.
Cardinaels,
Piperidinium, Piperazinium and Morpholinium Ionic Liquid Crystals,
J. Phys. Chem. B 113 (28) (2009) 9506–9511. doi:10.1021/jp903667e.
URL http://dx.doi.org/10.1021/jp903667e
[6] J.
Zhu,
L.
Bai,
B.
Chen,
W.
Fei,
Thermodynamical properties of phase change materials based on ionic liquids,
Chemical
Engineering
Journal
28
147
(1)
(2009)
58–62.

ACCEPTED MANUSCRIPT
doi:10.1016/j.cej.2008.11.016.
URL http://dx.doi.org/10.1016/j.cej.2008.11.016
[7] F.
Quirion,
D.
Lambert,
G.
Perron,
The HLM method: a simple way to get the solid–liquid phase diagrams and enthalpies of transition of pu
Canadian Journal of Chemistry 70 (11) (1992) 2745–2750.
doi:10.1139/v92-348.
URL http://dx.doi.org/10.1139/v92-348
[8] L.
Li,
J.
Groenewold,
S.
J.
Picken,
Transient Phase-Induced Nucleation in Ionic Liquid Crystals and Size-Frustrated Thickening,
Chemistry of Materials 17 (2) (2005) 250–257. doi:10.1021/cm048811f.
URL http://dx.doi.org/10.1021/cm048811f
[9] Y. Zhao, X. Chen, X. Wang, Liquid Crystalline Phases Self-Organized from a Surfactant-like Ionic Liquid
The Journal of Physical Chemistry B 113 (7) (2009) 2024–2030.
doi:10.1021/jp810613c.
URL http://dx.doi.org/10.1021/jp810613c
[10] E. Thomas, D. Thomas, S. Bhuvaneswari, K. P. Vijayalakshmi, B. K.
George, 1-Hexadecyl-3-methylimidazolium chloride: Structure, thermal stability and decomposition mech
Journal
doi:10.1016/j.molliq.2017.11.029.
URL http://dx.doi.org/10.1016/j.molliq.2017.11.029
of
Molecular
Liquids
249
(2018)
404–411.
[11] N. Goujon, M. Forsyth, L. F. Dumée, G. Bryant, N. Byrne,
Characterization of the phase behaviour of a novel polymerizable lyotropic ionic liquid crystal,
Physical Chemistry Chemical Physics 17 (35) (2015) 23059–23068.
doi:10.1039/c5cp03797b.
URL http://dx.doi.org/10.1039/c5cp03797b
[12] C. J. Bowlas, D. W. Bruce, K. R. Seddon, Liquid-crystalline ionic liquids,
Chem. Commun. (14) (1996) 1625–1626. doi:10.1039/cc9960001625.
URL http://dx.doi.org/10.1039/cc9960001625
29