By a simple differential thermal analysis (DTA) system, the concentration dependence of the glass transition temperatures ( 𝑇 𝑔 s) for the quaternary ammonium-type ionic liquid, N,N-diethyl-N-methyl-N-(2-methoxyethyl)ammonium bromide [DEME][Br] and H2O mixtures, after quick precooling was measured as a function of water concentration π‘₯ (mol% H2O). We compared the results with the previous results of [DEME][I]-H2O and [DEME][BF4]-H2O mixtures in which a double-glass transition behavior was observed. Remarkably, the [DEME][Br]-H2O mixtures basically show one- 𝑇 𝑔 behavior and the 𝑇 𝑔 decreases monotonically with increasing H2O content up to around π‘₯ = 9 1 . 5 . But it suddenly jumps to higher 𝑇 𝑔 value at a specific π‘₯ = ∼ 9 2 . At this very limited point, two 𝑇 𝑔 s ( 𝑇 𝑔 1 , 𝑇 𝑔 2 ) which we might consider as a transition state from the structure belonging to the 𝑇 𝑔 1 group to another one due to the 𝑇 𝑔 2 group were observed. These results clearly reflect the difference in the anionic effects among Brβˆ’, Iβˆ’, and B F 4 βˆ’ . The end of the glass-formation region of [DEME][Br]-H2O mixtures is around π‘₯ = 9 8 . 9 and moves to more water-rich region as compared to those of [DEME][BF4]-H2O ( π‘₯ = 9 6 . 0 ) and [DEME][I]-H2O ( π‘₯ = 9 5 . 0 ) mixtures.

1. Introduction

Room temperature ionic liquids (RTILs) are molten salts with melting temperature below (<373 K) and well known to have many attractive properties as solvents, for example, almost zero vapor pressure, wide electrochemical window, high recyclability, nonflammability, and so forth [1, 2], so that RTILs stimulate the interests of a wide range of applications [3, 4]. Many of the attractive features require a thorough knowledge of their thermophysical properties. Interactions in RTILs are characterized by the presence of Coulomb interactions among the constituent ions. Hydrogen-bonding, van der Waals, and Ο€-Ο€ interactions also take place in RTILs and affect their nature and the liquid structures. Thus, the liquid structures of RTILs are determined by a balance between long-range electrostatic forces and local geometric factors. As a result, many RTILs can be easily supercooled and form the glassy state [5–7]. Changes in the cation and anion combinations allow the physical and chemical properties of ionic liquids to be effectively tuned, for example, for manipulating the solvent properties of RTILs in order to achieve different purposes [7–9].

Although thermodynamic properties including 𝑇 𝑔 have been so far investigated mainly on the imidazolium-based series of pure RTILs [10–12], we point out that investigations of glass transition behavior of the RTILs-H2O mixtures are still scarce. In previous work, we reported the glass transition behavior of the quaternary ammonium-type ionic liquid, N,N-diethyl-N-methyl-N-(2-methoxyethyl)ammonium tetrafluoroborate [DEME][BF4]-H2O [13] and iodide [DEME][I]-H2O [14] mixtures, after quick precooling. Remarkably, double-glass transitions were observed in both mixtures. But concentrations of the two- 𝑇 𝑔 s region are completely different from each other; the double-glass transitions for the [DEME][BF4]-H2O mixtures were observed in the RTIL-rich region of π‘₯ = 1 6 . 5 ∼ 3 0 . 0  (mol% H2O), whereas the region moves to a water-rich side of π‘₯ = 7 7 . 5 ∼ 8 5 . 0 (mol% H2O) for the [DEME][I]-H2O mixtures. These clearly reflect the difference in the anionic effect between B F 4 βˆ’ and I βˆ’ on the water structure. It is interesting to quote that the ionic radius of B F 4 βˆ’ anion (229 pm [15, 16]) is slightly larger than that of I βˆ’ (216 pm [17]) anion. We suspect that the subtle difference in the ionic radii between B F 4 βˆ’ and I βˆ’ anions together with the anionic nature plays an important role in determining the regions where double-glass transitions occur.

Then, what happens to the phase behavior of other quaternary ammonium-type ionic liquid, [DEME][Br]-H2O mixtures, in which the anionic radius of B r βˆ’ (195 pm [18]) is smaller than that of I βˆ’ anion? Does the double-glass transition phenomenon occur or not? The aim of the present paper is to show the glass transition behavior of [DEME][Br]-H2O mixtures and to compare the results with those of [DEME][BF4]-H2O [13] and [DEME][I]-H2O [14] mixtures. Here we have measured the 𝑇 𝑔 variation of [DEME][Br]-H2O mixtures as a function of H2O concentration π‘₯ (mol% H2O). A limit of the glass-formation region is also determined.

2. Experiments

2.1. Material

As an RTIL in this study, we used N,N-diethyl-N-methyl-N-(2-methoxyethyl)ammonium bromide, [DEME][Br] which was synthesized in our laboratory. [DEME][Br] was synthesized following procedures reported in the literature [19]. The synthetic route of this material is shown in Scheme 1. The schematic chemical structures of [DEME][Br] are shown in Figure 1.


5.37 g (61.6 mmol) of N,N-diethylmethylamine and 8.03 g (57.8 mmol) of 1-bromo-2-methoxyethane were dissolved in about 7 mL of anhydrous acetone. The mixture was stirred overnight at 𝑇 = 3 7 3  K in an autoclave. After cooling, the reaction mixture was poured into ice-cold diethyl ether. The precipitated crude product was filtered off and was repeatedly recrystallized from anhydrous acetone. The purified product, [DEME][Br] were a colorless crystal. The molecular structures of [DEME][Br] was identified by 1H NMR spectra (JEOL JEM-ECS400). The spectral data (400 Hz, CDCl3) of [DEME][Br] are as follows: Ξ΄ 3.91 (t, 2H, 𝐽 = 4 . 3  Hz, –CH2–OCH3), Ξ΄ 3.87 (br s, 2H, –CH2CH2–OCH3), 3.66 (m, 4H, –CH2–CH3), 3.37 (s, 3H, –OCH3), 3.32 (s, 3H, –CH3), 1.38 (t, 6H, 𝐽 = 7 . 2  Hz, –CH2–CH3). No other peaks were observed except for the identified peaks from the objective compound mentioned above. This result suggests that organic impurities, for example, the starting materials, solvents, and byproducts, were fully removed. [DEME][Br] is a crystalline state at room temperature as described above. The melting temperature was reported to be 358.2 K in the available literature [19].

We prepared the sample solutions of [DEME][Br]-H2O mixtures at different concentrations, π‘₯ = 6 0 . 0 ∼ 9 9 . 8 Β± 0 . 0 5 (mol% H2O). There was a limitation of the solubility of [DEME][Br] in water ( π‘₯ < 6 0 ). Special care was taken with the sample preparations in a dry box to avoid atmospheric H2O and CO2.

2.2. DTA Measurements

To detect the glass transition temperature, 𝑇 𝑔 , a simple differential thermal analysis (DTA) system designed for quick cooling experiments was used. As a reference material for the measurements, we used benzene (Wako Pure Chemical Co.). A sample cell (about 35 mm long and 2 mm i.d. glass tube with one side sealed) was filled with a mixture, and then a thermocouple junction was placed 25~30 mm from the mouth of the sample cell. As a precooling procedure, a vitrification was done by putting the whole sample solution directly into liquid nitrogen where the quenching rate was estimated to be about 500 K/min. After taking out the sample from the liquid nitrogen, the DTA traces were recorded. The detailed procedure is basically the same as the previously reported ones [13, 14]. 𝑇 𝑔 values were found to be reproducible in this study to within 0.5 K, and the accuracy of temperature reading is estimated to be 1 K from a determination of melting temperatures of several guaranteed grade reagents (absolute ethanol, acetone and chloroform).

2.3. Raman Spectral Measurements

Raman spectra were typically measured by a JASCO NRS-1000DT Raman spectrophotometer equipped with a single monochromator and a CCD detector at room temperature (298 K). The 533 nm from green laser (Showa Optronics Co., Ltd.) with a power of 100 mW was used as an excitation source.

3. Results and Discussion

3.1. Glass Transition Behavior of [DEME][Br]-H2O Mixed Solutions

Schematic DTA thermograms of [DEME][Br]-H2O mixed solutions at three typical concentrations ( π‘₯ = 6 5 . 0 , 92.0, 95.0) are shown in Figure 2. At π‘₯ = 6 5 . 0 , the 𝑇 𝑔 was observed at 182 K. After the glass transition, we could not see the cold crystallization [20] with a large exothermal peak, which was observed in the case of [DEME][I]-H2O mixed solutions [14]. Just at π‘₯ = 9 2 . 0 , we captured the intriguing result: the DTA trace gives a first glass transition followed by an exothermic crystallization-like peak at around 161 K, then another set of an endothermic inflection at 172 K and an exothermic peak at 177 K in the DTA trace, respectively. To our knowledge, this is a typical DTA trace showing two glass transitions [21–23]. To ensure the second glass transition, we performed supplementary experiments using a requenching method in the following way: just after recording the first 𝑇 𝑔 ( 𝑇 𝑔 1 ) and the exothermic inflection as usual in the first run, we requenched the sample to the liquid nitrogen at 77 K quickly and recorded a DTA trace repeatedly. If the exothermic peak just after the 𝑇 𝑔 1 is due to the crystallization, we should have a clear downward 𝐢 𝑝 shift as the second 𝑇 𝑔 ( 𝑇 𝑔 2 ) in the second run. The schematic DTA traces obtained in this way are shown in Figure 3. On the other hand, the DTA trace at π‘₯ = 9 5 . 0 in Figure 2 shows one- 𝑇 𝑔 behavior, but interestingly the DTA trace keeps to show a cold crystallization peak. Taken together, glass transition behavior of the [DEME][Br]-H2O mixed solutions basically shows one glass transition against the water content. But at only limited concentration region of π‘₯ = 9 1 . 5 ∼ 9 2 . 5 , two glass transitions ( 𝑇 𝑔 1 , 𝑇 𝑔 2 ) were observed which we will describe in detail below.

Summarized results of 𝑇 𝑔 variations (⬀) as a function of π‘₯ are shown in Figure 4. For comparisons, 𝑇 𝑔 data of [DEME][BF4]-H2O [13] and [DEME][I]-H2O [14] are also shown. The 𝑇 𝑔 value monotonically decreases with increasing π‘₯ up to ∼ 9 1 . 5  mol%. But it suddenly jumps to higher 𝑇 𝑔 value at π‘₯ β‰₯ 9 1 . 5 . At this very limited point ( π‘₯ = 9 1 . 5 ∼ 9 2 . 5 ), two 𝑇 𝑔 s (β–‘: 𝑇 𝑔 1 , βšͺ: 𝑇 𝑔 2 ) were observed. As to an explanation for the mechanism of a double-glass transition phenomenon, the following was proposed [21–23]. When we lower a temperature of the sample, the homogeneous solution splits into two phases due to its thermodynamic instability where a metastable liquid-liquid immiscibility occurs. Based on the idea, we consider that the [DEME][Br]-H2O mixed solution may separate into two phases from the original solution at π‘₯ = 9 1 . 5 ∼ 9 2 . 5 . It is important to quote that the 𝑇 𝑔 of a glass-forming liquid has a correlation with its viscosity [24]. Thus, the two- 𝑇 𝑔 s behavior implies that the viscosities (and thus the structures) of two glassy phases are very different from each other. Looking into the results more closely, we may consider this as a transition state from the structure belonging to the 𝑇 𝑔 1 group ( π‘₯ < 9 1 . 5 ) to another one due to the 𝑇 𝑔 2 group ( π‘₯ > 9 2 . 5 ). At π‘₯ β‰₯ 9 2 . 5 some inclusions of H2O ice crystals in the quenched samples were visually confirmed. However, the solutions keep showing the clear but small 𝑇 𝑔 value up to around 98.9 mol% meaning that the system still holds the glassy state. We find that the edge of a glass-forming composition range extends to more water-rich region as compared to those of the [DEME][I]-H2O ( π‘₯ = 9 5 . 0 ) and/or [DEME][BF4]-H2O ( π‘₯ = 9 6 . 0 ) mixtures.

3.2. Comparison with the Glass Transition Behaviors of [DEME][I]-H2O and [DEME][BF4]-H2O Mixed Solutions

Firstly, we compare the 𝑇 𝑔 behavior with the results of [DEME][I]-H2O solutions. The double-glass transition phenomenon was obsevbed in the [DEME][I]-H2O mixtures, but there is a major difference in the region where the double glass transitions appeared. One is that the double- 𝑇 𝑔 s range is relatively wide ( π‘₯ = 7 7 . 5 to 85.0). Another one is that the 𝑇 𝑔 2 value changes little with π‘₯ in the double-glass transition region, though the 𝑇 𝑔 1 value decreases with increasing π‘₯ . This behavior is similar to that in the aqueous solutions of β€œnormal salt” such as symmetrical tetraalkyl ammonium halide (R4NX; R  = Ethyl (Et), n-Propyl (n-Pr), and X  = C l βˆ’ , B r βˆ’ ) [21–23].

On the other hand, the two- 𝑇 𝑔 s region moves to a very water-poor side in the case of [DEME][BF4]-H2O solution and the concentration range of the double glass transitions is much wider (13.5 mol%). These clearly reflect the difference in the anionic effects on the water structure. Previously, we explained that the differences in the results between the [DEME][BF4] mixtures and the [DEME][I] mixtures come from the variation in the solvation abilities and also the relative positions of the respective anions ( B F 4 βˆ’ and I βˆ’ ) and [ D E M E ] + cation in the RTILs [14]. The interactions between water molecules and the anions are significantly different in the two RTILs. Considering the smaller ionic radius of the I βˆ’ anion rather than B F 4 βˆ’ anion, the anion-water interaction in the [DEME][I] solution should be stronger than that in the [DEME][BF4] solution. This was partly evidenced by the existence of nearly free hydrogen-bonded Raman band (NFHB) of water molecules in the [DEME][BF4]-H2O system, which will be described in the next section. NFHB is assigned to the water molecules which exist as single molecules (not self-associated state) without forming the hydrogen-bonding network among themselves as in pure liquid H2O [14]. Unfortunately nor the detailed liquid structure of [DEME][BF4] or [DEME][I] has apparently not been repeted as yet. It is interesting to point out that the double-glass transition behavior is systematically changing from [DEME][BF4]-H2O (two- 𝑇 𝑔 s behavior in water-poor region, π‘₯ = 1 6 . 5 ∼ 3 0 . 0 ), [DEME][I]-H2O (two- 𝑇 𝑔 s behavior in water-rich region, π‘₯ = 7 7 . 5 ~85.0), and finally to [DEME][Br]-H2O (two 𝑇 𝑔 s at very limited concentrations, π‘₯ = 9 1 . 5 ~92.5) with decreasing the anionic radii. Clearly the solvation abilities of the respective anions provide significant effect on the structures of the RTILs-H2O mixtures which was reflected in the glass transition behavior.

3.3. Raman Spectra

To look further into states of the water structure in the mixtures, we measured Raman spectra. A vibrational spectroscopy is highly sensitive to the intermolecular interactions including hydrogen bonding among water molecules. Firstly, Figure 5(a) shows a comparison of the Raman spectra in the CH and OH stretching regions among [DEME][BF4]-H2O, [DEME][I]-H2O, and [DEME][Br]-H2O at π‘₯ = 6 5 in the liquid state. The signals arising from the CH stretching modes of the [DEME] cation appear ranging from 2800 to 3100 cmβˆ’1, though unfortunately the precise assignments of the respective peaks are not available at present. The overall CH frequency of the solutions shifts to a lower frequency side on going from [DEME][BF4]-H2O to [DEME][Br]-H2O with slight changes in the spectral shapes. This indicates that the environment around the alkyl chains of [DEME] cation is perturbed with the change of anion.

On the contrary, peaks in the range from 3200 to 3800 cmβˆ’1 belong to the OH stretching vibrational mode of water molecules. Figure 5(b) shows the enlarged spectra in this region. In viewing the results, the difference in the Raman spectra are very distinct. As pointed out in a previous paper [25], the spectrum of [DEME][BF4]-H2O typically displays a small peak at around 3565 cmβˆ’1 with a shoulder of 3620 cmβˆ’1. This peak is assigned to the nearly free hydrogen-bonded band (NFHB) of water molecules [26], as mentioned in the previous section. Here water molecules of the NFHB are probably very weakly interacting via H-bonding with the B F 4 βˆ’ anions. There is no water molecules due from NFHB in the case where the anion is iodide. Instead, an intenisty of the Raman peak centered at ~3440 cmβˆ’1 is appreciable. This band is mainly due to the OH stretching vibrations of water molecules weakly hydrogen bonded to halide ions [27], suggesting that the interactions between water molecules and the anions are significantly different in the two RTILs, [DEME][BF4] and [DEME][I] solutions; the anion-water interaction in the [DEME][I] solution is much stronger than that in the [DEME][BF4] solution.

In viewing the results, the results of [DEME][Br]-H2O show basically similar behavior to that of [DEME][I]-H2O. The important result obtained from the comparison of the two spectra is that the OH stretching spectrum of the [DEME][Br]-H2O solution lies as a whole in a lower-frequency region than that of the [DEME][I]-H2O solution, although the shift is only a marginal one at room temperature. This shows that average strength of hydrogen bonds in [DEME][Br]-H2O solution is slightly stronger than that in [DEME][I]-H2O solution.

In summary, we have investigated the glass transition behavior of the quaternary ammonium-type ionic liquid, [DEME][Br]-H2O mixtures by a simple DTA method. The 𝑇 𝑔 of [DEME][Br]-H2O mixtures decreases with increasing π‘₯ up to around π‘₯ = 9 1 . 5 and basically show one- 𝑇 𝑔 behavior. But at a specific concentration of around π‘₯ = 9 2 two 𝑇 𝑔 s were observed. Additionally, we have compared the anionic effect on the glass transition behavior of the series of [DEME][X]-H2O (X  = B r βˆ’ , I βˆ’ , B F 4 βˆ’ ) mixtures. Similar studies on other different RTILs with, for example, imidazolium-based cation-H2O systems, will deepen our understanding of the property of mixed solutions. Actually, we reported the glass transition temperatures of 1-butyl-3-methylimidazolium tetrafluoroborate-H2O mixed solutions as a function of H2O concentration [28]. Interestingly, in contrary to the results of quaternary ammonium-type ionic liquids, the multiple glass transition behavior was not observed and the system shows only one 𝑇 𝑔 throughout the whole concentration range.


The authors appreciate Dr. M. Aono and Dr. T. Takekiyo of National Defense Academy and Professor M. Kato and Mr. R. Wada of Ritsumeikan University for experimental supports and the fruitful discussion.