Impact of Water on the Melting Temperature of Urea + Choline Chloride Deep Eutectic Solvent

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains. Accepted Manuscript NJC

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content.The journal's standard Terms & Conditions and the Ethical guidelines still apply.In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.According to the 5th principle of green chemistry, safer solvents should be preferred.

Accepted Manuscript
The research for alternative liquids that could be used in synthesis or separation, has been very active for several years.An acceptable solvent should be not only nontoxic but also biodegradable or easily recycled.In addition it should be cheap and fulfill the technical requirements of the process.
Among the possible alternatives, the so-called 'eutectic mixtures' are of great interest.In the isobaric solid-liquid phase diagram of a binary system, a eutectic is a particular point displaying an equilibrium between two crystalline phases and a liquid phase.
[1] It corresponds to the minimum in the liquidus curve (curve above which the system is liquid) or in other words, at the eutectic composition the system can be used as liquid solvent at the lowest temperature.In 2003, Abbott et al. [2] noticed that different mixtures composed of quaternary ammonium salts and urea (two solids at ambient temperature), were liquid at ambient temperature.Even if the concept of eutectic was well known for years, it was relatively new to find eutectic points near ambient temperatures.
Abbott et al. introduced the concept of 'deep eutectic solvent' as a system composed of a hydrogen bond acceptor and a hydrogen bond donor, liquid at ambient temperature.In these mixtures, the decrease of the melting point can be explained by strong H-bonds that limit the crystallization of the salt.Since this work from Abbott, many papers have been published on similar mixtures referred to as deep eutectic solvents (DES), eutectic-based ionic liquids or low transition temperature mixtures (LTTM).Most of the publications concern the preparation of the eutectic liquid or its possible applications in various domains: electrochemistry, synthesis and catalysis, separation and material preparation.[3,4,5,6,7] Eutectic solvents are sometimes presented as an alternative to ionic liquids or considered as a whole new class of ionic liquids themselves.One reason for this comparison comes from their similar low melting points, typically below 100°C.Moreover, they can be considered as tunable solvents since it is possible to combine numerous H-bond donors and acceptors.When compared to ionic liquids, they are generally far less expensive, easy to prepare, non-toxic, biodegradable [6] and thus more attractive as green media.However, high viscosities and limited thermal and chemical stabilities are the main drawbacks of these liquids.[4] At the molecular level, ionic liquids are purely ionic whereas the H-bond donor of eutectic liquids is generally not ionic.
The decreased melting point of DES is explained by strong H-bond interactions between the donor and the acceptor [7].These interactions were studied by molecular dynamics simulations and experimentally by spectroscopy.Using molecular dynamics simulations, Perkin et al. [8] and Shah et al. [9] observed, in the case of urea:choline chloride, strong interactions between NH 2 moiety of urea and the chloride anion.IR and NMR spectroscopies have also confirmed the presence of these H-bonds.[8,10,11] For urea:choline chloride, the maximum number of H-bonds are obtained with the 2:1 molar ratio leading to the lowest melting point.At this particular composition the mixture urea:choline chloride is named reline.
To optimize the use of eutectic solvants, physico chemical characterizations of these systems would be helpful.A careful study of the literature shows a lack of thermodynamic studies on DES systems.For example, only a few references describe the phase behavior of the most studied eutectic system urea:choline chloride.Abbott et al. [2] established the first phase diagram for this system in 2002.In this pioneer paper, the liquidus curve is obtained from freezing temperatures.The eutectic composition and temperature were found at 2:1 molar ratio and 12°C.Later on, using Differential Scanning Calorimetry, Shah et al. have observed a melting point of the same 2:1 sample at 12°C.[9] With the same experimental technique, Morrisson et al. [12] have also identified 2:1 as the mole fraction of the eutectic composition melting point at 17°C.
Confusion appears in the literature, as the freezing points measured by Abbott et al. are referred to as melting points by some authors.[4,6] In many systems, supercooled liquids can exist, leading to differences between the freezing and melting temperatures.In order to establish the phase diagram, melting should be preferred to freezing, as it is independent of the experimental conditions.[13] Both choline chloride and urea are hygroscopic.The presence of water may impact their use in further applications and will modify the interactions between the two components.Using molecular simulation, Shah et al. [9] has shown a decrease of urea-anion interactions in the presence of water.With NMR spectroscopy, a similar tendency was observed in other eutectic mixtures.[11] Using solvatochromic probes on urea:choline chloride mixtures with water Pandey et al. [14] has shown an increased dipolarity/polarizability and a decrease of the H-bond acceptor basicity.Furthermore, the strong organization of DES (partly explained by the H-bond network) can be advantageous for selective catalysis or nanoparticles synthesis.[5,15,16] The addition of water will disturb the molecular structure and thus limit these applications.This effect was investigated for gold nanoparticles in reline DES.[17] At 30°C, the density of pure reline was measured at 1.1945, 1.1952 and 1.216 g.cm -3 , respectively, by Yadav et al. [18], Xie et al. [19] and Shah et al. [9].With the addition of water a systematic decrease of the density (1% for the addition of 5 wt% of water) was observed.[18] Large discrepancies were also observed for the viscosities of reline: at 30°C, 527 mPas, 552 mPas and 954 mPas were measured respectively by Yadav et al. [18], Shah et al. [9] and Xie et al. [19].These deviations may be due to different quantities of water in the neat eutectic mixtures, which have not been quantified in any of these works.When water is added, there is a dramatic decrease of the viscosity.For example, an addition of 5 wt% of water leads to a viscosity divided by a factor of 6. [18] The presence of water modifies volumetric and transport properties of reline whereas its influence on the melting point of reline has not been yet considered.
In this work, we determined the solid-liquid phase diagram for the system urea:choline chloride.For that purpose, three complementary techniques were implemented: (i) visual determination of the melting point on samples of a few grams using a thermostated bath, (ii) optical microscopy on smaller samples of typically 20-200 mg with polarized light that allows the detection of amorphous and crystalline particles and, (iii) DSC (differential scanning calorimetry) that can also provide the heat associated to the phase transitions.Furthermore, this work includes a study of the effect of naturally present water, in these highly hygroscopic mixtures, on the melting point using the same three techniques.

Materials and methods : Materials
Choline chloride (ChCl) and urea were purchased from Sigma-Aldrich with purities higher than 98%.Both powders were dried under vacuum (0.1 mbar) for at least 24 hours at room temperature.Urea:ChCl mixtures were prepared by weight.The samples at any composition, under mechanical stirring, became liquid and perfectly transparent within few minutes at 80°C.
When heating a dry urea sample at 5 °C/min in the DSC, a decomposition temperature which is very close to its melting point (130-135 °C) was observed as reported in the literature [20].If the heating rate is reduced to 1°C/min, the decomposition peak is shifted to 105°C.To avoid possible decomposition of the samples, only mixtures with liquidus temperature below 80 °C were studied.Eight compositions between x urea = 0.50 and 0.80 have been investigated (urea:ChCl molar ratio = 50:50, 55:45, 60:40, 65:35, 67:33, 70:30, 75:25 and 80:20).Given the precision of the balance and the presence of residual water in the powders, the uncertainty of the mole fraction is estimated to 0.2%.
The water contents of all mixtures were measured using a Coulometric Karl-Fischer titrator (Mettler Toledo, DL31).Quantities were around 1000 ppm and systematically below 2000 ppm for dried mixtures.
Thermostated bath 1 gram of urea:ChCl mixture is inserted in a vial together with a magnetic stirring bar under nitrogen atmosphere.The vial is heated up to 80°C and kept at this temperature during 30 minutes for homogenizing the sample.It is then quickly cooled down with liquid nitrogen to obtain a solid transparent sample.The vial is then immerged in the thermostated bath equipped with two insulated glass windows (Julabo 18V).The bath was previously cooled down to 5°C with an immersion cooler (Julabo FT200).While heating to 5°C, a transition is observed for all samples from transparent to white solid.Samples are left at 5°C for 30 minutes for equilibration before use.In a typical experiment, the bath is heated from 5°C to 80°C at a scanning rate of 0.25 or 0.025°C/min using the controller on the thermostated bath.The temperature is measured with a 100 Ω platinum resistance thermometer immerged in the bath between the different samples.A webcam (Logitec C920) is fixed in front of one of the windows of the thermostated bath.
Images of the samples are taken automatically every 0.5°C using Yawcam software.The phase transitions of the samples were studied by analyzing the brightness of each image using a home-made software.

Microscope
A few drops of the liquid mixture are placed on a glass slide and immediately covered by a glass coverslip to avoid water absorption.The glass slide is placed in the hot stage (Linkam LTS420) of the thermostated optical microscope (Leica DM2500M).
A typical temperature program analysis of the sample is: i) heat from room temperature to 80°C at 20°C/min; ii) hold for 1 minute at 80°C; iii) cool down to -60°C at 10°C/min; iv) hold for 10 minutes at -60°C and v) heat up to 80°C at either 0.10 or 0.25°C/min.During the experiment, the hot stage containing the sample is continuously purged with nitrogen to keep the sample dry.To follow the phase transitions of the sample, images were recorded every 0.5 °C and the brightness of each image was analyzed using a home-made software.There was no evolution of the samples during our experiments.This was verified by performing two successive cycles on the same sample.

Differential Scanning Calorimetry
A differential scanning calorimeter mDSC 2920 from TA-instrument was used.It was calibrated in temperature and energy using standard indium and lead.The validation of the calibration was realized using fusion of water and naphthalene.The maximum deviation from literature in temperature obtained for those two compounds was below 0.2 °C.The maximum deviation in enthalpy was below 1%.About 5 to 10 mg of the liquid mixture is weighted in a hermetic sealed crucible.The sample and reference crucibles are placed inside the calorimeter.The atmosphere in the calorimeter oven is continuously flushed with nitrogen to increase the homogeneity of the temperature and to evacuate any volatile compound that may be produced during heating.Experiments were conducted as followed: i) fast increase of the temperature up to 80°C; ii) isotherm at this temperature for 10 minutes; iii) decrease of the temperature to -55°C at a scanning rate of 10°C/min; iv) isotherm at this temperature for 10 minutes; v) increase the temperature with a scanning rate of 0.1, 0.25 or 1°C /min, up to 80°C; vi) isotherm at this temperature for 10 minutes; vii) return to room temperature at a scanning rate of 10°C/min.The first and the last steps are not recorded as they are used to initialize the sample and finish the experiment, respectively.

Thermostated bath
The transition temperatures of all urea:ChCl mixtures were first determined using a thermostated bath as illustrated in Figure 1 in the case of 55:45 and 67:33 mixtures.While heating the reline sample (67:33), a rapid transition from a white solid to a transparent liquid is observed between 30 and 35 °C.In the case of the 55:45 sample, the mixture is crystalline up to 30°C.Above 60°C, it is a transparent liquid.Between 30 and 60°C, a white solid-liquid biphasic system is observed.This temperature range corresponds to the domain between the solidus (temperature below which the mixture is a solid) and the liquidus temperatures.
To precisely build the phase diagram of the urea:ChCl system, the brightness of all samples was automatically analyzed as illustrated in Figure 2 for the reline and the 55:45 samples.At the lowest temperatures, below 30°C, the mixtures are crystalline, the brightness is the most stable.The variations in the signals are similar for all samples.They correspond to changes in the overall luminosity and not to a physical modification of the sample.Then, around 30 °C, the brightness quickly decreases.The beginning of this transition corresponds to the solidus temperature.At higher temperatures, strong oscillations of the brightness are observed which corresponds to the movement of the stir bar.Thus, it was not possible to precisely determine the liquidus temperature with this technique.This part of the diagram will be obtained using the optical microscope and DSC experiments.
The solidus temperatures obtained after the brightness analysis of the eight samples at two different heating rates (0.25 and 0.025 °C/min) are presented in Table 1.An average value of 30 °C was obtained.We cannot exclude a small effect of the scanning rate as the values obtained at 0.25°C/min are slightly higher than those obtained at 0.025°C/min.The average deviation of all the results is 1.5 °C that will be considered as the global uncertainty of this method.A crystallization experiment on a dried reline was also conducted by decreasing the temperature from 35 to 5°C at a scanning rate of -0.025 °C/min under stirring.The beginning of the crystallization occurs at 20 °C.This value is lower than the melting temperature of 30°C.This is different from the data measured by Abbott et al. [2] (12°C).This deviation can be partially explained by different crystallization conditions.

Polarized microscopy
The phase transitions of the same eight samples were analyzed by optical microscopy using polarized light to enhance the sensitivity to crystalline structures.Experiments were conducted using scanning rates of 0.10 and 0.25 °C/min.Examples of the images of the 67:33 and 50:50 mixtures are given in Figure 3.At 10 °C, the reline sample presents a spherulitic crystalline structure (Figure 3a) characteristic of an eutectic mixture.All the eutectic crystals melt between 24 and 30 °C, leading to a liquid above this temperature (Figure 3c).In contrast, the 50:50 mixture presents several types of crystals (Figure 3b).The whiter crystals melt at the same temperature as the solid particles in the reline sample.The elongated needle shape crystals are still visible at 30 °C (Figure 3d), and melt continuously when increasing the temperature up to the liquidus transition.
Using brightness analysis it is difficult to determine the initial point of melting.Thus, the solidus temperature cannot be precisely determined with this technique.An analysis of all the images permits the estimation of the liquidus temperature.This is defined as the temperature of the first image without any particles.The results are presented in Table 2 for the two scanning rates studied.Similar temperatures (average deviation of 2 °C) are obtained with the two sets of experiments.This average deviation is considered as the global uncertainty of this technique.The composition corresponding to the minimum of the liquidus curve (29°C) is 67:33 molar ratio.

Differential scanning calorimetry
The solidus and liquidus temperatures were also determined using differential scanning calorimetry (DSC).Experiments were conducted at different scanning rates (0.10, 0.25 and 1 °C/min).Sample thermograms obtained for two mixtures at 3 scanning rates are presented in Figure 4.The general behavior of the thermograms at reline composition (left figure 4) can be described as following: -Exothermic peaks associated to the crystallization of the sample.The crystallization does not happen during the cooling step but during the heating of the sample.This phenomenon is named cold crystallization.It is generally observed with strongly metastable systems such as ionic liquids or polymers.The temperature associated to this phenomenon is drastically dependent on the scanning rate.
[21] -An endothermic peak associated to the melting of the eutectic crystals.The solidus temperature is the onset temperature of this peak, which is defined as the intersection between the baseline just before the peak and the tangent of the peak at the point with the maximum slope as illustrated in Figure 4.This temperature is characteristic of the beginning of the melting of the eutectic crystal.In the case of the 60:40 mixture (right graphs in Figure 4), as well as for all the mixtures with compositions different from the reline, a last event can be identified.Its related temperature is determined from the intersection of the straight lines obtained from the thermal flux after the solidus peak and the line to return to the baseline (see Figure 4).This temperature corresponds to the end of the melting of non-eutectic crystals (liquidus temperature).With fast scanning rates (1°C/min), melting begins before the end of the crystallization peak which prevents the precise determination of the solidus temperature (Figure 4).When the scanning rate is slow (0.1 °C/min) the crystallisation and the melting point are well separated but the effect associated to the liquidus temperature is difficult to distinguish from the baseline.The solidus temperatures are then obtained from the 0.1°C and 0.25°C/min experiments.An average value of 25 °C is obtained with a global uncertainty (calculated as the average deviation) of +/-2°C.The liquidus temperatures are obtained from the 0.25 and 1.0 °C/min experiments.All characteristic temperatures determined by DSC are reported in Table 3.In addition to the melting temperatures, it is possible to calculate the enthalpy of melting of the eutectic crystals, by integration of the thermal flux versus time.This calculation is done at the reline composition.Using the experiments conducted with scanning rates of 0.1 °C/min and 0.25 °C/min, the enthalpy of melting is 88±2 J/g.

Phase diagram
The phase diagram obtained is presented in Figure 5.For the solidus line, a discrepancy was observed between the data obtained from the DSC and the thermostated bath.The temperatures measured with the thermostated bath are systematically higher than those obtained with the DSC.The sizes of the samples are different in the two apparatuses (5-10 mg in the DSC, 1 g in the thermostated bath).
In the thermostated bath, the melting temperature obtained from the two scanning rates (0.25 and 0.025 °C/min) are not significantly different.These experiments are not sufficient to confirm possible heat transfer limitations in the thermostated bath.Another possible explanation for the deviations between the two techniques could

New Journal of Chemistry Accepted Manuscript
be the sensitivity to detect the start of the melting with the thermostated bath.Due to the thickness of the sample, a difference in the brightness cannot be detected before a sufficient quantity has melt.It can be observed in the thermostated bath experiments (for example, see Figure 1, 55:45 mixture, at 30°C) that small solid particles are present on the glass surface of the vial, they melt a few degrees before the bulk phase.This is coherent with the results of the DSC analysis.All these reasons could explain why a higher solidus temperature is found with the thermostated bath than with the DSC experiments.For the solidus, only the DSC data are then presented on the phase diagram in Figure 5.The liquidus temperatures obtained with the microscope and the DSC are similar as observed in Figure 5, except in the particular case of the 75:25 sample.At this composition, using the microscope, most of the non-eutectic crystals melts quickly but a small quantity of them remains dispersed in the liquid up to 60°C.The energy associated to thee melting of the last crystals cannot be detected by DSC due to sensitivity limitation.The lowest melting temperatures were measured for urea mole fractions between 0.65 and 0.70, in agreement with literature (eutectic composition of 67:33 [2,12]).Following the discussion on the possible overestimation of the solidus temperatures in the thermostated bath, we finally estimate the eutectic temperature to be 25°C using the DSC experiments.This is a higher value compared to the results of Morrison et al. [12] and Shah et al. [9].This difference could be explained by different water quantities.
The Tammann's graph is also given in Figure 5.It shows the enthalpy of melting of the eutectic crystals (in kJ.mol -1 ) versus the composition of the mixture.As expected, this representation shows that the enthalpy is maximum at the eutectic composition and decreases linearly when moving far from this particular composition.With this representation, the eutectic composition can be precisely obtained at 0.67 and with an enthalpy of melting of 93 J/g or 8.1 kJ/mol (24.3 kJ/mol of reline).Morrison et al. found a lower value (71.09J/g) for their reline mixture melting at 17 °C.Finally, the thermal effect associated to the melting of the eutectic temperature disappears at urea mole fractions of 0.53 and 0.81.

Effect of water of the melting temperature of reline
The eutectic mixture is highly hygroscopic, so, inevitability water will always be present in the sample.As already reported in literature, its presence will modify the interactions between the two components of the eutectic system and impact its physico-chemical properties.[9,18,19] In this work, we propose (i) to evaluate the typical water quantity that can be absorbed from atmosphere and (ii) to quantify the impact of this residual water on the melting point.
A dry 67:33 eutectic mixture was exposed to atmosphere (21°C, 50% relative humidity) under stirring for three weeks.An aliquot was regularly taken from the sample.The water content was quantified by Karl-Fisher titration and its evolution as function of time is presented in Figure 6.A first linear increase of the water quantity is observed.An absorption of 5.5 wt% of water is measured after 48 hours corresponding to 1150 ppm per hour.After a week under our experimental conditions, a plateau is obtained at 18-20 wt% of water.Water can easily be present in a reline sample, if powders are not initially dried or if the mixing is not performed under dried atmosphere.To understand the impact of this residual water on melting, different water quantities were systematically added to dry reline samples, up to 10 wt%.The eutectic temperatures of the wet samples are measured with the same techniques used to build the solid-liquid phase diagram.
A linear decrease of the melting temperature to 15°C was observed for a mixture containing 5 wt% of water, as shown on Figure 7. Similar values were obtained with the three techniques.Given the melting point of water, this result was expected.However this effect was never quantified.This could explain some discrepancies observed in literature in which the water quantity is not always controlled.The higher temperatures found in the present study may be due to smaller water quantities than in previous published data.

Conclusions
The phase diagram of the binary system urea:choline chloride was established using complementary tools.The thermostated bath permits working with samples on the gram scale, which are representative of practical applications.The sensitivity to observe the last particles (determination of the liquidus temperature) is much higher when using optical microscopy compared to the thermostated bath.Over the determination of the phase transition, DSC experiments also give access to the energy associated to the phase transition.The melting temperature at eutectic composition (two molecules of urea per choline chloride) is 25°C, a value higher than the previously reported data.This difference is mainly due to the presence of water.Its impact on the melting temperature was studied in details for the first time.After 48 hours in contact with atmosphere, the sample absorbed 5.5 wt % of water.This quantity leads to a decrease of more than 15 °C of the melting temperature.This result could explain the discrepancies observed in literature on the melting temperatures of eutectic systems.Quantification of water content in DES should be done systematically for any physico chemical characterization as well as for applications.
The laboratory thanks the Auvergne region, France, for its financial support (Project CPER Environment).

Figure 1 :
Figure 1: Photographs of 55:45 (top) and 67:33 (bottom) urea:ChCl mixtures taken in the thermostated bath during an experiment at a scanning rate of 0.25 °C/min.After the melting, the stirrer bar is clearly visible in both liquid mixtures.

Figure 2 :
Figure 2: Evolution of the brightness of the 55:45 and 67:33 samples during an experiment in the thermostated bath at a heating rate of 0.25 °C/min.For readability reasons, the brightness of the two samples has been shifted from each other.

Figure 3 :
Figure 3: Microscope images of an eutectic composition (67:33, left) and a 50:50 mixture (right) at 10 °C (a, b) and 30 °C (c, d) at a scanning rate of 0.10 °C/min.The contrast of the pictures taken at 30°C has been improved to help the reader to see the particles still present at this temperature in the non-eutectic composition (d).The size of all pictures is around 1 mm 2 .

Figure 4 :
Figure 4 : Thermograms of 67:33 (left) and 60:40 (right) samples obtained with scanning rates of 0.1, 0.25 and 1 °C/min.Exothermic peaks (signal up) correspond to crystallization and endothermic ones (signal down) to melting.For readability reasons, the thermograms have been shifted from each other.Construction lines used to obtain the solidus (in red) and liquidus (blue) temperatures are added to the thermograms.

Figure 5 :
Figure 5: Solid-liquid phase diagram of the system urea:ChCl as function of the urea mole fraction.For clarity reason, only the average temperature is shown when experiments were repeated with two scanning rates.The red lines are only added to help the reader.On the bottom, the Tammann's plot of the enthalpy of the eutectic transition is also given.

Figure 6 :
Figure 6: Evolution of the water quantity (expressed in weight percent) in a reline sample in contact with atmosphere (under mixing), as function of time.The average temperature and the atmospheric relative humidity are respectively 21°C and 50%.

Figure 7 :
Figure 7: Evolution of the melting point of a 67:33 mixture as a function of the water quantity expressed in wt%.In the thermostated bath, the temperature determined is obtained from the beginning of the decrease of the brightness.With the microscope, it corresponds to the first image without any crystals.With the DSC, the temperature is obtained at the end of the melting determined similarly to the liquidus temperature of the dried samples.

Table 1 :
Solidus temperatures (in °C) obtained in the thermostated bath at two heating rates.

Table 2 :
Liquidus temperatures (in °C) obtained with the microscope experiments at two scanning rates.

Table 3 :
Transition temperatures obtained with the DSC experiments.