Anomalous and Not-So-Common Behavior in Common Ionic Liquids and Ionic Liquid-Containing Systems

This work highlights unexpected, not so well known responses of ionic liquids and ionic liquid-containing systems, which are reported in a collective manner, as a short review. Examples include: (i) Minima in the temperature dependence of the isobaric thermal expansion coefficient of some ILs; (ii) Viscosity Minima in binary mixtures of IL + Molecular solvents; (iii) Anomalies in the surface tension within a family of ILs; (iv) The constancy among IL substitution of Cp/Vm at and around room temperature; (v) ILs as glass forming liquids; (vi) Alternate odd-even side alkyl chain length effects; (vii) Absolute negative pressures in ILs and IL-containing systems; (viii) Reversed-charged ionic liquid pairs; (ix) LCST immiscibility behavior in IL + solvent systems.


INTRODUCTION
Ionic Liquids (ILs) are constituted quasi-exclusively by anions and cations, melting at temperatures that are much lower than those of their conventional, inorganic salts counterparts. For instance, "table salt" (NaCl) melts at about 800 • C. In contrast, many ILs present melting points lower than room temperature. Generally, ILs have a large liquid range and, some of them do not easily crystallize on cooling, instead they supercool and undergo a glass transition.
Several high quality reviews (Welton, 1999(Welton, , 2018Plechkova and Seddon, 2008;Armand et al., 2009;Hallett and Welton, 2011;Niedermeyer et al., 2012;Tariq et al., 2012;Chatel et al., 2014;Hayes et al., 2015;Hunt et al., 2015;Podgoršek et al., 2016;Kar et al., 2019) are available. No review focused on the unexpected behavior of ILs has been published, where a scrutiny of their not so well known physical chemistry responses is made. This work constitutes a first attempt in this direction.
With the upsurge in the research activity around these novel salts during the last two decades, peculiar, unique and interesting behavior of these complex materials have been revealed. ILs are considered a link between molten salts and molecular solvents (Leal et al., 2009).  using MD simulations and Triolo et al. (2007) using experimental X-ray data have demonstrated that ILs, even in their pure state, contain nanostructured organization at a molecular level. In addition, this very distinct feature has been confirmed using experimental thermodynamic approaches (Pereiro et al., 2013;Rocha et al., 2013). This phenomenon is responsible for many of their peculiar behavior. Also, how a particular IL interacts with the co-solvent is very unique and depends on the interactions between the IL and the chemical nature of the other component: polar/apolar/associated fluids Pádua et al., 2007).
The topics tackled in this contribution are not comprehensive. Examples not herein presented include: their ability to form halogen-bonds (Bernardes and Canongia Lopes, 2017;Saccone et al., 2017;Cavallo et al., 2018;Lodeiro et al., 2018), the formation of liquid crystals (Alvarez Fernandez and Kouwer, 2016;Goossens et al., 2016), the total miscibility in water of fluorinated ILs (Pereiro et al., 2015;de Ferro et al., 2018), and other unexpected behavior in respect to their physical properties (Singh et al., 2015(Singh et al., , 2017Dzida et al., 2018;Rahman and Senapati, 2018). Zwitterionic liquids (Blesic et al., 2017;Ohno et al., 2018;Wu et al., 2018), which fill the gap between small-ion ILs and ILs with strong H-bonds, as well as hydrated ILs (Haberler et al., 2012;Fujita et al., 2016), where H-bonds between one or both of the ions and a small number of water molecules is sufficient to produce new materials with superior properties, constitute other examples of not so-well-known behavior.

MINIMA IN THE TEMPERATURE DEPENDENCE OF THE ISOBARIC THERMAL EXPANSION COEFFICIENT OF SOME ILs
For most liquids the isobaric thermal expansion coefficient, α p , is positive and increases with increasing temperature as it has to diverge to a positive infinite value at the critical point. The best well-known example of a distinct behavior is that of water, in which α p at about atmospheric pressure is negative at temperatures between 0 and 3.98 • C, meaning a contraction on heating. At the temperature of maximum density, 3.98 • C, α p reaches a null value, becoming positive for higher temperatures. Despite this "anomaly, " the α p of water always increases with increasing temperature. In this section, we highlight examples of the anomalous behavior of α p with temperature increase for some ionic liquids. The results discussed in this section are the first examples of substances that show a minimum in α p as temperature increases.
Conflicting results were reported earlier about the temperature dependence of α p of ILs (Rebelo et al., 2004). Later some reports for a small set of ILs have shown that it is possible to obtain negative values for (∂α p /∂T) p (Rodríguez and Brennecke, 2006;Sanmamed et al., 2007). Accurate determinations of this property may be difficult since the determination of α p from the temperature dependence of density, ρ(T), can produce numerical artifacts (Cerdeirina et al., 2001;Troncoso et al., 2010). Most often the density of ILs is measured using a vibrating tube densimeter. Their high viscosity may influence the determination of the density and, if viscosity corrections are not taken into consideration, erroneous α p (T) values may be obtained (Sanmamed et al., 2007).
Systematic studies on the measurements of density along a homologous series and wide temperature and pressure conditions are not commonly reported. In order to fill this gap, the densities of a series of ILs in a wide temperature and pressure range were reported. Tariq et al. (2010) have measured the densities of imidazolium bistriflimide ILs, [C n C 1 im][Ntf 2 ], where n = 2-14 in the temperature range of 283-483 K. This dataset was the largest one both in terms of an extended homologous series of ILs, as well as in a large temperature interval, enabling the authors to study subtle specificities of their thermal response. Nieto de Castro et al. (2010) have carried out high-precision density measurements over broad temperature (293-473 K)  It has been demonstrated  that in the [C n C 1 im][Ntf 2 ] series, an increase in temperature results on a small decrease in the values of α p . This inverse proportionality between α p and T constitutes an anomalous behavior and is observed at "low-temperature." Since thermodynamics impose that at the liquid-vapor critical point, a divergence to +∞ in α p ought to occur, at some lower temperature a minimum in α p = f(T) is obtained ( Figure 1A).
Similarly, densities of four ILs in wide pressure and temperature ranges were measured (Nieto de Castro et al., 2010). Again, α p decreases as temperature increases in the low-T range ( Figure 1B). Navia et al. (2010a), Navia et al. (2010b) also obtained data for a large set of ILs confirming negative (∂α p /∂T) p values at low temperatures.
These works have shown that the temperature derivative of the thermal expansion coefficient, (∂α p /∂T) p , of some ILs changes sign at a temperature that depends both on pressure and IL nature.

VISCOSITY MINIMA IN BINARY MIXTURES OF IL + MOLECULAR SOLVENTS (MSs)
Most models for the viscosity of a mixture predict that the mixture's viscosity of two components with identical viscosities is invariant along the whole composition range.
A minimum in viscosity-composition plots of binary mixtures is an unusual phenomenon, which has been observed for some non-polar + polar systems (Kouris and Panaylotou, 1989;Papanastaaiou and Ziogas, 1991;Laesecke et al., 2007). However, the molecular reasoning behind this phenomenon is not very well understood (Srinivas et al., 2001;Abraham et al., 2007) and thereby it is hard to model the viscosity values of such mixtures using existing mixing rules and predictive methods (Qunfang and Yu-Chun, 1999). Such uncommon phenomenon has been shown for the first time in systems containing ionic liquids (ILs) + molecular solvents (MSs) (Tariq et al., 2015). Tariq et al. (2015) selected four binary IL+MS systems composed of a molecular solvent (2-amino-ethanol (2AE) or 3-amino-1-propanol (3AP)) and an IL (from the 1-alkyl-3-methylimidazolium family ([C n mim] + ) using dicyanamide ([DCA]) or bistriflimide ([Ntf 2 ]) as the anion). All the components forming the four binary systems are completely miscible in the entire composition range and show crossover temperatures where the IL and MS viscosity values are identical (Figure 2).
The η(T,x) plots presented in Figure 3 reveal that (i) there is a viscosity minimum at low temperatures for IL-rich mixtures and that (ii) the viscosity minimum is centered around the  equimolar composition as one approaches the temperature at which both components present identical viscosities. The overall viscosity trend for these systems is similar along the composition/temperature surface.
The origin of the viscosity minima is 2-fold: (i) it comes from the different cohesive energies of both pure MSs and ILs and (ii) from changes in the structure and interactions of the mixtures compared to the pure components.
The molecular interactions between the molecules can be revealed by comparison of the cohesive energies of the two classes of components (obtained through vaporization enthalpies at room temperature). The values of the vaporization enthalpies of the MSs are in the range of 60-70 kJ mol −1 (Marsh et al., 2004;Yaws, 2009;Acree and Chickos, 2010). In contrast, ILs present vaporization enthalpies above 135 kJ mol −1 (Marsh et al., 2004;Esperança et al., 2010). In the case of these MSs, the cohesive energy is largely related to the intra-hydrogen bonding between the functional groups of each molecule. In the case of ILs, MD simulations (Santos et al., 2007;Shimizu et al., 2010) have revealed that Coulomb forces play an important role, contributing substantially for their enhanced cohesive energy.
H-bonds between the different ILs and MSs and the structural differences between the mixture and the pure molecular components originates a reduction on the viscosity of the mixtures. Other type of uncommon behavior of IL containing binary mixtures has been reported by Andrzejewska et al. (2009) and Trivedi and Pandey (2011), where a maximum in the viscosity has been found in mixtures of IL + polymers. It should be noted that systems containing polymers are known to show non-Newtonian behavior.

ANOMALIES IN THE SURFACE TENSION WITHIN A FAMILY OF ILs
The values of the surface tension of most ILs fall in between those of water and molecular solvents. One interesting trend has been found in the 1-alkyl-3-methylimidazolium bistriflimide, [C n C 1 im][Ntf 2 ] family (Carvalho et al., 2008;Kolbeck et al., 2010;Tariq et al., 2010;Haddad et al., 2018;Shimizu et al., 2018). Carvalho et al. (2008) were the first to report that the surface tension values of [C n C 1 im][Ntf 2 ] series do not decrease linearly with the alkyl chain length increase (n = 2-10), but rather reach a plateau for long cation side alkyl chains. Later, Tariq et al. (2010) measured the surface tension for C 2 -C 14 within a broad temperature range of 303-493 K and Kolbeck et al. (2010) measured it at room temperature for C 1 -C 12 . Both confirmed these trends ( Figure 4A).
The trend is not regular, and was checked out by three distinct works: there is a substantial decrease in the surface tension value from [C 2 C 1 im][Ntf 2 ] to [C 5 C 1 im][Ntf 2 ] and a relatively lower decrease from [C 6 C 1 im][Ntf 2 ] to [C 12 C 1 im][Ntf 2 ]. Changes in the ratio of prominence of non-polar to polar moieties of the ILs at the surface are a consequence of the length of the alkyl side chains. MD simulations and X-ray diffraction studies Triolo et al., 2007) have shown the creation of a second nanostructured domain (formation of non-polar continuous domains) for ILs with alkyl chain length equal or >6 carbon atoms.
Recently, Shimizu et al. (2018) have been able to predict the surface tension of the 1-alkyl-3-methylimidazolium bistriflimide family by combining angle-resolved X-ray photoelectron spectroscopy data and MD simulations results using the Langmuir principle. Haddad et al. (2018) have also used angstrom-resolution X-ray methods to understand the reason behind the peculiar behavior of the surface tension within the [C n C 1 im][Ntf 2 ] homologous series. They also found a distinct behavior for ILs with alkyl chain length equal or longer than six carbon atoms due to the formation of nanosegregated domains (polar/apolar) which create alternating layers at the surface. This study clarifies the liquid-air interface structure for a common homologous series of ILs. By varying the cation's alkyl chain length one can tune the interactions's importance, from long-range coulombic forces to short-range van der Waals interactions. Such variation causes the interface structure to turn from simple, to layered, to liquid crystalline. The quantitative results obtained from this work may constitute a reference for validating simulations and theory.
An even more peculiar behavior of the surface tension within a homologous series has been reported by Luís et al. (2016) for [C n C 1 im] cation based ILs (where n = 2-12) combined with perfluorobutanesulfonate anions. Instead of a plateau, a minimum in the surface tension is observed for a cation alkyl side chain of 8 ( Figure 4B). The existence of three nanosegregated domains (polar, apolar, and fluorinated) for the long cation's alkyl chain length is the main reason for this distinctive trend. More specifically, it results from the competition between the diverse domains for the gas-liquid interface. Their surface entropy is the lowest when compared to conventional ILs.

THE CONSTANCY AMONG IL SUBSTITUTION OF C P /V M AT AND AROUND ROOM TEMPERATURE
The temperature dependence of the enthalpy is known as heat capacity. Some heat capacity changes may indicate the occurrence of a phase transition and enable to understand variations in the structure of the compounds. Zabransky et al. (1990) and Domalski and Hearing (1996) compiled and analyzed, independently, the heat capacity data for a huge number of substances. To the best of our knowledge, group contribution methods and/or corresponding state principal based methods are the approaches generally used to estimate liquid heat capacities.
For ionic liquids, Strechan et al. (2008) and Gardas and Coutinho (2008) demonstrated independently that there is a linear relationship between the heat capacities and the molar volumes of imidazolum, pyridinium, and pyrrolidinium based-ILs (Figure 5). They have used the dataset of limited number of ILs to establish this correlation and propose that the volumetric heat capacity of ILs is almost invariant at 298.15 K.
Later, Paulechka et al. (2010) have refined this relation by carefully selecting dataset of 19 ILs (for which the precise density data was available) and measured their heat capacity with an uncertainty of (±1%). They have also witnessed that there is constancy among the volumetric heat capacities of ILs. The average value at 298.15 K is: This average value agrees well with those of Gardas and Coutinho (2008). The difference between the experimental volumetric heat capacities and this C p /V m value was found to be < ±5%.
Contrarily to molecular solvents, a simple rule exists to predict the heat capacity of distinct ILs. It has been demonstrated that C p /V m is basically independent of the structure of the IL. Therefore, Paulechka et al. (2010) concluded that ILs used for application as heat accumulators  Frontiers in Chemistry | www.frontiersin.org should be chosen on the basis of other parameters such as thermal stability, viscosity, thermal conductivity, to name a few.

ILs AS GLASS FORMING MATERIALS
ILs have low melting points (Tm) when compared to conventional inorganic salts. This fact gives them the status of a unique class of ionic materials composed almost solely of cations and anions that exist in the liquid state at or near room-temperature. Many ILs that do exist in the liquid state never crystallize and thus do not show any melting point, instead, on cooling they show a glass transition (Tg). Most ILs (Valderrama et al., 2017) show a glass transition temperature in the 150-250 K range.
It is easier to accurately determine melting points (typically for ILs a good (±5 K) uncertainty is obtained) than those of vitrification-large discrepancies (up to 20%) have been found in reported Tg values. Glass transitions do not occur at a specified fixed temperature as they are not first-order changes (Brandrup et al., 1999), and are kinetically dependent. The dependence of Tg of ILs on the scan rate has thoroughly been investigated (Goḿez et al., 2013;Tao et al., 2017). Blokhin et al. (2006) and Rodrigues and Santos (2016) have continuously presented high quality thermal properties data of ILs and the latter demonstrated how the nanostructuration of [C n C 1 im][NTf 2 ] and [C n C n im][NTf 2 ] family influences their glass and melting temperatures. The work demonstrates a variation of the Tg behavior at the critical alkyl size, when n = 6 ( Figure 6). T g increases as the alkyl side chain of the imidazolium cation grows, mainly due to the enhancement of the van der Waals interactions. For alkyl chains longer than n = 6, a plateau in the Tg values after nano-structuration is observed.
Some works have tackled the prediction of the Tg of ILs (Mirkhani et al., 2012;Valderrama et al., 2017). Other studies have identified difficulties behind observing IL's crystallization (Serra et al., 2017;Ferreira et al., 2019). Lima et al. (2018) have measured the Tg and Tm of a pyrrolodinium based-IL, [C 4 C 1 Pyrr][Ntf 2 ], from atmospheric pressure up to an extremely high pressure of 2 GPa using X-ray diffraction and Raman scattering techniques. They have found that Tg and Tm both follow similar pressure dependences.
At atmospheric pressure, most compounds seem to follow the well-known "2/3 golden rule" meaning that the ratio of glass FIGURE 7 | (A) Melting points vs. glass transition temperatures for PILs reported by Belieres and Angell (2007). (Reprinted with permission from Belieres andAngell, 2007. Copyright (2007) American Chemical Society) (B) T g values as a function of molar volume (V m ) for PILs reported by Xu and Angell (2003). The line through the points is a guide to the eye. [From Xu and Angell (2003). Reprinted with permission from AAAS]. transition temperature (Tg) to that of melting (Tm), Tg/Tm, for all compounds should be around 0.66. Belieres and Angell (2007) have collected data for several protic ionic liquids (PILs) and showed that similar to other materials, most of the PILs fall in the 2/3 line ( Figure 7A). However, there are outliers that can reach a ratio as high as 3/4. Xu and Angell (2003) have discussed the Tg dependence on the molar volume ( Figure 7B). Recently, Ferreira et al. (2019) and Serra et al. (2017) have shown that many of the aprotic ILs show high Tg/Tm values that are close to the 3 4 value.

ALTERNATE ODD-EVEN SIDE ALKYL CHAIN LENGTH EFFECTS
Whenever the thermophysical characterization of the members of a family of ILs within a homologous series is performed, odd members are often neglected (Tariq et al., 2009). The main difficulty in the synthesis of ILs with alkyl chains, C n , where n is an odd number, is the high cost of their chemical precursors. Adamová et al. (2011) have measured the densities of a series of alkyltrioctylphosphonium chloride, [P8 8 8 n]Cl based ILs and found that the density values show a clear odd-even chain length alternation effect ( Figure 8A). There are two independent trends: one for the odd and another one for the even-numbered compounds. This was observed up to n = 9. Data analyzed in terms of their molar volume, Vm = M/ρ, presents remarkably this see-saw effect ( Figure 8B). Adamova et al. (2014) have carried out Ab initio calculations and MD simulations to understand the molecular reasoning behind these effects. Simulation runs revealed that the type of conformation/packing of the liquid (transoid conformations of the cation and head-to-head packing) are responsible for the observed alternation effects. The unexpected results first seen for the [P8 8 8 n]Cl series, enabled the authors to reveal a similar trend for another IL family, [C n C 1 im][NTf 2 ]. Moreover, this seesaw effect, already known for the solid phase of linear alkanes and alkanols, was also seen in their liquid phase molar volume.
Very recently, it has been shown (Belchior et al., 2018) that diluted solutions of [C n C 1 im]Cl (n = 2-14) used to form aqueous biphasic systems (ABS) with salts also exhibit such odd/even effects. The odd/even effect was observed in the Setschenow salting-out coefficients (ks) for systems containing water, [C n C 1 im]Cl and K 2 CO 3 ( Figure 9A). The salting-out ability depends on the molar volume of the IL, and therefore the alternation odd-even volume pattern is reflected in the ks values. This is more prominently seen for ILs with alkyl side chain length up to n = 6. As for the CMC values of the [C n C 1 im]Cl series  this effect was not observed. In contrast, it was also shown that an odd-even effect occurs for properties, such as, the degree of ionization (Figure 9B), the molar conductivity, and the molar conductivity at infinite dilution.
FIGURE 10 | The Berthelot cycle used to obtain tensioning in the IL samples inside a glass capillary: liquid at L-V equilibrium (A); the liquid fills in the entire internal volume at Tfill (B); the liquid is forced along an isochore (constant volume) creating higher, positive pressures (C); if the liquid adheres extremely well to the glass capillary walls, at point (D) the liquid is experiencing an isotropic, bulk tension (absolute negative pressure metastable state). At Tcav (cavitation temperature) it will collapse at point (E), relaxing back to its stable condition located along its vapor pressure curve at (F). (Reprinted from Silva et al., 2018).
In addition to the above-discussed cases, odd/even effects have also been encountered in several other IL properties, such as, viscosities (Rocha et al., 2013), entropy and enthalpy of vaporization (Rocha et al., 2012(Rocha et al., , 2014, glass forming behavior (Leys et al., 2014), and diffusion coefficients (Yang et al., 2016).

ABSOLUTE NEGATIVE PRESSURES IN ILs AND IL-CONTAINING SYSTEMS
Liquids can be mechanically stretched. If this is performed in an isotropic fashion it is possible to obtain absolute negative pressures. Most of the focus of negative pressure experiments was placed on water samples and only very recently some results for pure ionic liquids appeared. The results show that a variety of commonly used ionic liquids (ILs) can be stretched successfully to tensions of about −100 MPa in Pyrex glass capillaries of internal volume of ∼0.05 cm 3 (Silva et al., 2018). These results contrast with the ones obtained for water, in which the maximum stretching of samples of similar size was around −35 MPa (Visak et al., 2002(Visak et al., , 2003.
The main reasons for achieving such enormous absolute negative pressures in ionic liquids appears to be a consequence of distinct properties, namely almost null volatility, increased viscosity compared to common liquids, low surface tension, enhanced wettability toward pyrex glass, and easiness to supercool.
Recent work by Silva et al. (2018) has shown that it is possible to use pulsed field gradient NMR spectroscopy to describe the change on the fluid molecular dynamics (transport response FIGURE 11 | Relative cation's self-diffusion coefficients vs. temperature in positive, negative, and null pressure regimes. (Reprinted from Silva et al., 2018).
Frontiers in Chemistry | www.frontiersin.org functions) of liquid salts under homogeneous negative pressure regimes (down to about −20 MPa). The experiments followed the thermodynamic Berthelot cycle (Figure 10) in order to prepressurize the samples and create the conditions to further enter, by cooling the sample, into negative pressure regimes. Negative pressures have been accurately estimated by using experimental values of thermal-pressure coefficients of each liquid at the FIGURE 12 | The conceptual transfer of two neutrons and two protons to obtain charge-inverted ionic pairs-the 1-alkyl-3-methylimidazolium chloride and its isoelectronic and isostructural counterpart, the potassium 1-alkyl-3-methylcyclopentadienyl.
relevant-temperatures of the corresponding isochore. A marked augment in both anions' and cations' self-diffusion coefficients is observed as one enters the metastable negative pressure region as compared to the saturated liquid conditions. Figure 11 shows the ratio between the cation's diffusion under isochoric conditions and the cation's diffusion under isobaric conditions. The vertical dotted line indicates the filling temperature of the samples, and consequently, data at higher temperatures represent the positive pressure regime while data below this temperature characterize the negative pressure region. It is important to note that for temperatures below 25.5 • C (p ∼ −150 bar), the liquid salt has already cavitated to the L-V condition and, therefore, the ratio in D's was restored to 1.0. The results show a markedly decrease of the self-diffusion coefficient upon applied positive pressure and a steep increase in it for regimes of negative pressure. It is worth to note that a mild volumetric expansion of about 0.6-0.7% ( V/V%) per each p = −100 bar of applied negative pressure can be estimated. This contrasts with the increase in the self-diffusion coefficients of about 5% per each p = −100 bar, representing an almost 10-fold increase as compared to the volume expansion of the ionic liquid.

REVERSED-CHARGED IONIC LIQUID PAIRS
It is well-known that salts based on the halides' anions show much lower melting temperatures than those in which their isoelectronic alkyl metals counterparts are present as cations. For instance, already-known salts based on potassium are not members of the ionic liquids family, whereas many with chloride are.
It was recently shown (Cruz et al., 2018(Cruz et al., , 2019) that using alkali metal cations with distinctive 1-alkyl-3-methylcyclopentadienyl anions it is possible to generate ILs. More specifically, if one uses the same isoelectronic concept as we do for alkali metals vs. halides (e.g., K + vs. Cl − ), it is possible to recognize the 1alkyl-3-methylcyclopentadienyl anion as the isoelectronic and isostructural counterpart of the 1-alkyl-3-methylimidazolium cation (Figure 12). Conceptually, one simply has to transfer one proton and one neutron from each of the two nitrogen atoms of the imidazolium ring to the nucleus of the chloride anion. Differential scanning calorimetry has been used to show that K[C 4 C 1 Cp] and K[C 6 C 1 Cp] melt without decomposition at around 90 • C. Molecular Dynamics (MD) simulations were used to understand the structural differences between these two IL families. The marked structural differences between the K[C n C 1 Cp] and [C n C 1 im]Cl series is a consequence of the charge-reversion among ion pairs. In the case of these alkali metal based ILs, peculiarities of metal coordination chemistry enables the creation of new structural features. In conclusion, charge-inverted salts can also present low melting temperatures. This fact opens the door for the synthesis of new families of ILs.

LCST Immiscibility Behavior in IL + Solvent Systems
The liquid-liquid lower critical solution temperature (LCST) type of phase diagram rarely occurs. However, it is an important type of demixing that is the basis of several key applications (Albertsson, 1986). It means that phase separation occurs upon temperature increase, with the system attaining a higher order (as compared to the separated components). The closed-loop phase diagram (a temperature-composition island of immiscibility) is even rarer and appears as a result of a very subtle balance between enthalpic and entropic contributions to the Gibbs energy of a system as temperature is changed. It is characterized by a LCST at a given temperature followed by an upper critical solution temperature (UCST) at a higher temperature.
LCST-type of immiscibility was typically only found in some aqueous or (polymer + solvent) solutions. For the first time, Lachwa et al. (2005) have encountered both LCST and closed-loop type of behavior in binary and quasibinary liquid solutions of alkylmethylimidazolium bistriflimide, [C n mim][Ntf 2 ], with chloroform or with (chloroform + carbon tetrachloride) mixtures. This study revealed the tunable character of the liquid-liquid phase diagrams involving ILs. Two variables were taken into account, first, in the solutions with chloroform, the number of the carbon atoms in the IL's cation alkyl side chain were varied and secondly the chain length was kept constant at n = 5, but the composition of the mixed solvent was altered by adding carbon tetrachloride to chloroform. This work has demonstrated the extreme sensitivity of the phase diagrams upon small changes of two variables: long IL alkyl chain lengths promote better solubility; whereas addition of more CCl 4 to the CHCl 3 worsens the solubility. Fukumoto and Ohno (2007) have presented another example of LCST behavior, which was related to the solutions of amino acid-based ILs and water ( Figure 13A). The LCST temperatures of the solutions were tuned by changing the length of the alkyl side chains of either the phosphonium cation or the triflateaminoacid anion. Longer alkyl chains lower the temperature of the phase separation due to enhanced hydrophobicity. The insertion of four methylene groups to one cation's alkyl side chain brings the LCST down from room temperature to the freezing temperature of water. The anion chain addition of only one -CH 2 -group lowers the LCST by about 15 • C. Insertion of a phenyl group with a CH 2 group attached to it also reduces the LCST by about 15 • C. The tuning of the LCST behavior has immense potential in extraction and separation processes.
Other cases of LCST behavior within solutions of ILs and polymer as two component systems were demonstrated independently by Ueki and Watanabe (2007) and Lee and Lodge (2011). Alkyl-methylimidazolium bistriflimide ILs, [C n mim][NTf 2 ], mixed with poly(benzyl-methacrylate) (Ueki and Watanabe, 2007) or poly(n-butyl methacrylate) (Lee and Lodge, 2011) show that the increase of the cation chain length, either in the pure IL or through a mixture of cations with different chains, provokes an opposite effect as compared to similar aqueous solutions. Thereby, longer alkyl chains exhibit better solubility and higher LCST. The authors related the improved solubility to the enhancement of the dispersive forces between the (longer) cation alkyl chains and the polymer chains.
Recently, Costa et al. (2013) have reported for the first time functionalized IL + ether systems exhibiting a LCST behavior. Investigating the phase behavior of binary mixtures of three N-alkyl-N,N-dimethyl-N-hydroxyethylammonium bis(trifluoromethane)sulfonylimide ILs, [N 11 n2OH ][Ntf 2 ], with nine distinct ethers at atmospheric pressure, they have found that most systems exhibit unusual LCST-type of phase separation ( Figure 13B). By increasing the alkyl side chain of the cholinium-derived cation, an enhancement of the mutual solubilities of ILs and ethers is attained. The availability of the oxygen atom of the ether molecule to perform hydrogen bonding with the IL cation and the size/ramification of the alkyl groups of the ether molecule are the key factors that control the solubility of these systems. This LCST behavior is a consequence of the disruption of the H-bond network between the hydroxyl group of the cholinium based cation and the oxygen atom of the ether molecule.
Other works also report systems showing LCST behavior containing mixtures of ILs with polymers, water, supramolecular compounds and other solvents and are summarized in a recent review (Qiao et al., 2017) which highlights their potential applications.

CONCLUDING REMARK
We have highlighted some unexpected features of ionic liquids and ionic liquid-containing systems which are not so well-known of the scientific community.

AUTHOR CONTRIBUTIONS
All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

FUNDING
This work was supported by the Associate Laboratory for Green Chemistry-LAQV which is financed by national funds from FCT/MCTES (UID/QUI/50006/2019).