- Email: [email protected]
Dynamics of water clusters in solution with LiCl
Physica A 442 (2016) 261–267
Contents lists available at ScienceDirect
Physica A journal homepage: www.elsevier.com/locate/physa
Dynamics of water clusters in solution with LiCl Carmelo Corsaro a,b,∗ , Domenico Mallamace c , Nicola Cicero c,d , Sebastiano Vasi b , Giacomo Dugo c,d , Francesco Mallamace a,b,e a
CNR-IPCF, Istituto per i Processi Chimico-Fisici, Viale F. Stagno D’Alcontres 37, 98158 Messina, Italy
b
Dipartimento di Fisica e Scienze della Terra, Università di Messina, Viale F. Stagno D’Alcontres 31, 98166 Messina, Italy
c
Dipartimento SASTAS, Università di Messina, Viale F. Stagno d’Alcontres 31, 98166 Messina, Italy
d
Science4Life, spin-off Università di Messina, Viale F. Stagno D’Alcontres 31, 98166 Messina, Italy
e
NSE Department, Massachusetts Institute of Technology, Cambridge MA 02139, USA
highlights • We observe three different water clusters in a solution with LiCl at eutectic point. • Two important dynamical changes occur at two relevant temperatures for water. • The driving force is the tendency of water to develop its characteristic HB network.
article
info
Article history: Received 7 May 2015 Received in revised form 30 July 2015 Available online 11 September 2015 Keywords: Lithium chloride Dynamical crossover Water solution
abstract In this work we study by means of Nuclear Magnetic Resonance spectroscopy the dynamics of the different water clusters that form within a solution with LiCl at eutectic concentration in the temperature range 320–205 K. This solution is considered a model system allowing the investigation of water properties in the deep supercooled regime in its bulk phase. Our data reveal two important dynamical changes occurring at two relevant temperatures for water: the highest temperature coincides with that of the water density maximum (277 K) and the lowest with that of the so-called dynamical crossover (≃225 K). We interpret our data in terms of the different influence that the ions exert on water by lowering the temperature and of the tendency that water displays to develop its characteristic hydrogen bond network. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The dynamics of supercooled glass forming materials can be highly nonlinear, especially for those glass-formers that are defined fragile because their structure changes rapidly when temperature changes [1]. In fact, Angell in 1995 has classified glass-forming liquids into two different classes by using the concepts of fragility and glass transition [1]. In the last years, some authors suggested a different and universal scenario by invoking the concept related to the socalled dynamical crossover or fragile-to-strong transition [2–5]. It was observed that all supercooled glass-forming materials show a dynamical transition, that can be more or less evident depending on the system, before intervening the dynamical arrest or glass transition [2,4,6]. The higher is the temperature the higher is the number of degrees of freedom that the system can explore. Therefore, the corresponding dynamics of the system is characterized by multiple relaxations that
∗
Corresponding author at: Dipartimento di Fisica e Scienze della Terra, Università di Messina, Viale F. Stagno D’Alcontres 31, 98166 Messina, Italy. E-mail address: [email protected] (C. Corsaro).
http://dx.doi.org/10.1016/j.physa.2015.09.008 0378-4371/© 2015 Elsevier B.V. All rights reserved.
262
C. Corsaro et al. / Physica A 442 (2016) 261–267
Fig. 1. The thermal evolution of normalized SE ratio. It is noteworthy the sharp increase below 240 K.
involve the exploration of the local minima of the energy landscape [7]. By lowering the temperature, and thus the number of allowed energy states, the system can only experience a dynamics characterized by hopping processes between minima of uniform height [8]. This scenario is consistent with the theoretical predictions of the extended version of the Mode Coupling Theory (MCT) for which the critical temperature of the ideal version of the theory corresponds to the temperature of a dynamical crossover [3,9,10]. In other words, the fluids dynamics is characterized by two different behaviors above and below a precise temperature. Theory agrees with experiments that the dynamics of the system can be considered as the sum of two contributions: that of the ideal MCT i.e., the cage dynamics and that of the EMCT i.e., the hopping process. The cage dynamics shows a temperature dependence that can be represented by a power law function diverging at the MCT critical temperature. Below it the hopping process is the only possible dynamics that shows an Arrhenius dependence with temperature [9,10]. The detailed analysis of the dynamical behavior of water is relevant for pure science and technological applications. In particular water dynamics is more intriguing in the supercooled regime and there is a big and actual debate on the existence and on the significance of the dynamical crossover [11–19]. This liquid–liquid transition was observed experimentally for water in particular environments and in MD simulations. However, it is hard to supercool bulk water below the homogeneous nucleation temperature. It is well known that an eutectic solution of water in LiCl can be cooled to 200 K without the occurrence of any crystallization process [20,21]. For that reason, aqueous solutions of LiCl were used to investigate the occurrence of the dynamical crossover and contrasting scenarios have emerged [5,22–27]. The existence of different results seems, at least in part, due to the usage of different techniques that cover (and probe) different dynamical regimes [4,22,24,27]. The coexistence of different local structures and of the associated relaxations does not allow a simple data interpretation. For example NMR and OKE experiments did not observe any abrupt change in the water dynamics by means of the self-diffusion and time constant, respectively. However, by considering the stretching exponent of the time constant distribution, a dramatic change at the temperature of about 240 K was observed [28]. Moreover, by taking into account the Stokes–Einstein (SE) relation, it breaks just about that temperature signaling the decoupling between transport properties [29]. In Fig. 1 we report the Stokes–Einstein ratio, Dη/T, obtained by the data reported in Ref. [29], as a function of T and normalized with respect to its value at the highest temperature. As one can see, for T & 240 K this quantity is nearly constant reflecting the validity of the SE relation that instead breaks below 240 K where the reported quantity shows a sharp increase. The related decoupling between translational and rotational dynamics has been explained by invoking the existence of dynamic heterogeneities, which refers to the presence of transient spatially separated regions with vastly different relaxation times [30]. Optical and dielectric spectroscopies have detected the presence of extra signals such as an excess wing for temperatures below 225 K but the interpretations are somehow contradictory [27,31]. Furthermore, Molecular Dynamics (MD) simulations showed that the water tendency to develop its characteristic tetrahedral network provokes a segregation of the system. The ions are not included in the network of water molecules and contribute to the formation of solute rich regions within the system [32,33]. Previous studies have shown that water properties are not too altered by the presence of ions suggesting that it can be considered a model system for studying deeply supercooled water in its bulk phase [21,23–26,28,34,35]. At the same time, it was shown that the local tetrahedral structure of water is partially distorted by the presence of ions [36]. In particular, according to the Collins scenario, Li cation is an enhancer of the water tetrahedral structure (structure maker) whereas the
C. Corsaro et al. / Physica A 442 (2016) 261–267
263
Fig. 2. Snapshot of the different tetrahedral structures that water can adopt within the considered solution. From the left to the right the tetrahedral structure is increasingly open and less rigid [36].
Cl anion tends to disrupt water structure (structure breaker) [37]. As a consequence the lithium solvation shell is more structured and less mobile with respect to that of chlorine that is less structured and more mobile. As a matter of fact, ab-initio calculations show that in a tetrahedron of water molecules coordinated by a lithium cation, the Li-O radial distance is about 0.19 nm whereas in a tetrahedron of water molecules coordinated by a chlorine anion the Cl-O radial distance is about 0.31 nm (Fig. 2). Note that the O-O distance in pure water and in solution with LiCl is about 0.27 nm [36]. In Fig. 2 we report a snapshot of the different tetrahedral structures that water can adopt within the solution. From the left to the right the tetrahedral structure is increasingly open and less rigid [36]. The presence of tetrahedral structures makes the water/LiCl solution suitable for studying the bulk properties of water in the deep supercooled regime. In this paper we study by means of proton NMR spectroscopy the dynamics of the different water clusters that coexist within the solution. In fact, many studies demonstrated that in aqueous solution of LiCL, especially at high concentration, transient local structures exist around the two types of ions. In particular, the water solvation shell around Li cation is more rigid with respect to that found in pure water and even more with respect to that around Cl anion [33,36]. Our aim is to characterize the water dynamics in such a bulk environment in a region of the water phase diagram that is not accessible for pure water. In fact, for T < 235 K (the homogeneous nucleation temperature) the crystallization rate is too fast for any experimental observation. Therefore, in order to study liquid water within the so-called ‘‘no man’s land’’, confining systems or proper aqueous solutions have to be used [11,12,38]. Understanding the anomalous behavior of water has been the subject of many experimental and theoretical works that are still under consideration (see e.g. [39,40]). Starting from the stability limit hypothesis of Speedy [41], the scenario that is just receiving many independent confirmations is that of the liquid–liquid transition associated with a second critical point for water [17,42]. This liquid–liquid transition, occurring from a high density liquid at higher temperatures to a low density liquid at lower temperatures, seems to be connected with the so-called Widom line where thermodynamic response functions take on extrema values [39,43,44]. Even if the water/LiCl system is actually very studied, there is no complete consensus about the occurrence of the dynamical crossover in the temperature interval 210–230 K and about its origin [24–29,31]. As above mentioned, the dynamical crossover has been observed for water in different environments [12,13] and it has been suggested that can also be generalized for every liquid approaching the dynamical arrest before intervening the glass transition [2,8]. In our study we use NMR spectroscopy because it is a local probe and, even if we look at averaged relaxational dynamics, we are able to discriminate and follow separately the dynamics of the different local structures of water within the considered solution. We want to probe the hydrogen-bond relaxation dynamics because it is crucial for having a deep insight into water properties and anomalies [45]. In particular, we have acquired the NMR spectra by cooling from 320 K to 205 K each 5 K and analyzed the line-shape of the water peak. By extracting the relevant observables such as the apparent spin–spin relaxation time, T2∗ , for each of the detected contributions, we were able to follow and characterize the peculiar dynamics of the different water clusters that form in the solution. 2. Materials and methods The water/LiCl solution at the eutectic concentration of 6.76 M was prepared starting by anhydrous LiCl and deionized water (Sigma-Aldrich) by using Mohr’s method. NMR experiments were performed at atmospheric pressure with a Bruker Avance spectrometer operating at 700 MHz (proton Larmor frequency) equipped by a probehead with inner coil optimized for proton observation. The investigated temperature range was 205 K < T < 320 K with an accuracy of ±0.1 K and the temperature was calibrated by means of the Bruker standard sample of 4% of CH3 OH in CD3 OD. The acquisition parameters are the following: duration of the hard pulse 20 µs and relative attenuation 0.3 dB; spectral width 15 kHz; 64 k points in the time domain; 8 transients and 5 s of relaxation time. After the Fourier transformation and phase and baseline correction, we observed that the best fit of the water peak was obtained by using a Gaussian deconvolution with three components, reported for the highest and lowest measured temperatures in Fig. 3. For what concerns the fitting parameters we have an uncertainty below 1% and correlations between 0.5 and 0.9.
264
C. Corsaro et al. / Physica A 442 (2016) 261–267
Fig. 3. The Gaussian deconvolution of the proton NMR spectra of water/LiCl solution at eutectic point at the highest (left) and lowest (right) measured temperatures.
3. Results and discussions The measured 1 H NMR spectra of water/LiCl solution at eutectic concentration display only one signal and it belongs to water. From an inspection of Fig. 3, it is possible to notice that the peak position is lower than that in pure water even at the lowest measured temperature. In fact, the proton NMR peak position of water increases with temperature due to the increasing shielding effect that each water molecule feels during the development of its hydrogen bonded network [46]. Its chemical shift value within LiCl solution is lower because of the presence of electric charges in solution that, being in motion, influence the local magnetic field experienced by water protons. Moreover, the peak at 205 K is more intense and broader than at 320 K due to the competing effects of increasing polarization and decreasing mobility of water molecules on lowering the temperature. For all the investigated temperature ranges, the best fit of the water peak, reported in Fig. 3, was obtained by using three different Gaussian components. As above mentioned, theoretical studies showed that in water/LiCl solutions, there are different clusters of water molecules structured by lithium and chlorine ions [32,33]. Furthermore, at the eutectic concentration, there are 7 water molecules per molecule of LiCl, hence some water molecules make a dynamic bridge between the two ions pointing the oxygen towards lithium and the hydrogens towards chlorine. Therefore, we ascribed the three Gaussian components to the three different local water structures that can be found in the solution. In particular we named the different water structures as: bulk-like, around Li and around Cl. Note that these different local structures are transient in nature and can be explained in terms of nanosegregation. By means of our experimental technique we can discriminate between the solvation shells of water around lithium cation and chlorine anion. In Fig. 4 we report the intensity of the three Gaussian components, that is the magnetization of the different water clusters, corrected by the Curie law, as a function of the inverse temperature. We ascribe the most intense Gaussian component (circles) to bulk-like water (water molecules interacting with each other), the broader one (triangles) to water molecules around Li cation and the sharper one (squares) to water molecules around chlorine anion. Our choice is motivated by the results of the cited theoretical studies about the higher rigidity of the water solvation shell around Li and of the more mobile structure that water forms around Cl [32,33,36]. Note that, the unusual increase of the magnetization of all the three Gaussian components of the water peak on lowering the temperature from 320 K to T ≃ 277 K is due to the fact that water molecules within salt solutions are highly polarized and their polarization increases on decreasing the thermal energy. For T < 277 K, the magnetization of water molecules interacting with lithium and chlorine ions follows two different thermal behaviors that cross each other at about 227 K. This trend can be explained if we assume that water coordination around the two ions changes with the temperature. In particular, our data suggest that for 320 K > T > 277 K water molecules make part of the coordination spheres of both lithium and chlorine with the same weight or probability. By lowering the temperature, water molecules are more localized around chlorine ions down to about 260 K where the amplitude of the Gaussian component of the chlorine shell displays a maximum and that of the lithium shell shows a minimum suggesting an inversion of the trend. At about 227 K the magnetization corresponding to the two solvation shells assumes the same value and decreases on lowering the temperature.
C. Corsaro et al. / Physica A 442 (2016) 261–267
265
Fig. 4. The thermal evolution of NMR magnetization of the different water components for water/LiCl solution at eutectic concentration. The value is corrected by the Curie law. Vertical lines refer to the temperatures of 277 K (solid) and 227 K (dotted).
For what concerns the magnetization of the bulk-like water component, it shows a monotonic increase on decreasing the temperature until about 227 K where it assumes its maximum value. At this particular temperature, the correlation length of water molecules is maximum [39] and water fully develops its hydrogen bonded network. Li and Cl ions constitute only local defects of this extended network because they ‘‘prefer’’ (or have) to segregate from water, as happens in ice formation [47], forming a solute-rich water nanophase [32]. Therefore, below 225 K water molecules tendency to hydrogen bond with each others is the strongest interaction that determines the thermodynamic properties of the system. This is confirmed by the decreasing of the magnetization of all Gaussian components for T < 225 K (Fig. 4). Finally, we have evaluated the apparent spin–spin relaxation time, T2∗ , of the proton nuclear magnetization by the width of the three Gaussian components as:
√ ∗
T2 =
2 ln 2
π ∆ν
(1)
where ∆ν is the full width at half maximum (in Hz) of each Gaussian component. The spin–spin relaxation time, is a measure of the strength of dipolar interaction between spins of the same species. Since different local structures of water manifest different dipolar interactions, it is a good probe to study their separate dynamical behavior. T2∗ represents the time needed by the transverse component of the macroscopic magnetization to vanish, due to the spin dephasing mechanism, in the plane orthogonal to the direction of the static magnetic field. In Fig. 5 we report T2∗ for each of the three Gaussian components as a function of the inverse temperature. Note that the T2∗ of all the three components has a very similar thermal behavior: starting from the highest temperature, they slowly increase reaching a maximum value at about 280 K. Moreover, all show a minimum at about 260 K and a maximum close to 225 K. Thus, our data reveal that even if different local water clusters exist within the solution and their relative concentrations (reflected by the corresponding magnetization value reported in Fig. 4) change with temperature, they have the same dynamics in the considered thermal range that includes the supercooled regime. This justifies why other experimental techniques such as the previous dynamical measurements of self-diffusion coefficient or relaxation time [26,28,29] have observed only one contribution. Besides, transient grating experiments proved how the clusters dynamics is not affected by the temperature until about 210 K, below which an additional signal appears [24]. Furthermore, large scale Molecular Dynamics studies suggested that below the liquid–liquid transformation temperature water becomes a four-coordinated low-density liquid (LDL) [32]. However, the existence of three temporal scales confirms the different rigidity (and thus mobility) of the local water structures when coordinated by the different ions. The maximum in T2∗ at about 225 K reinforces the suggestions that at this temperature water shows a liquid–liquid transformation becoming a four-coordinated low-density liquid characterized by an extended hydrogen bond network that dominates the thermodynamic properties of the system.
266
C. Corsaro et al. / Physica A 442 (2016) 261–267
Fig. 5. Plot of the apparent spin–spin relaxation time, T2∗ , of the proton nuclear magnetization for each of the three Gaussian components as a function of the inverse temperature for water/LiCl solution at eutectic point.
4. Conclusions Water is simultaneously the simplest and the most complex liquid. Although the water molecule consists of only three atoms, its high tendency to form an extended network of hydrogen bonds determines the existence of anomalous properties that indeed are responsible for life. Above the temperature of about 320 K, water can be considered a simple liquid (water molecules display essentially a local high density structure) because the strength and lifetime of hydrogen bonds are not enough to form the characteristic tetrahedral structure [48]. Below 320 K, when the high directionality of the hydrogen bond begins to dominate over the thermal disorder and the van der Waals interactions, the thermodynamic properties of water begin to be anomalous. In particular, many thermodynamic response functions of water show a critical-like behavior with an apparent divergence at about 228 K [39,43]. In this paper, we have studied by means of 1 H NMR spectroscopy the dynamics of the different water local structures that exist in an aqueous solution of lithium chloride at the eutectic concentration (6.76 M). This kind of system is considered a prototype that allows to study the thermodynamic properties of water in the deep supercooled regime without the need of confining environments. The temperature interval that we have considered, includes the supercooled regime and extends from 320 K to 205 K. In particular, we have used a Gaussian deconvolution with three components to best reproduce the obtained NMR peak of water. The three Gaussian components reflect the different local environments that water molecules can experience within the solution. We ascribed the most intense Gaussian component to bulk-like water, the broader to the water shell solvating the lithium cation and the sharper to the water in the chlorine coordination sphere. In fact the water local structure around the lithium is more rigid with respect to that found in bulk water. This is why lithium has the property to enhance the water structure (structure maker). On the contrary, the chlorine anion is a structure breaker and the water local structure surrounding it is more mobile than that of bulk water. Our data, in terms of the macroscopic magnetization and the apparent spin–spin relaxation time, T2∗ , reveal the existence of two characteristic temperatures at which the clustering organization of water molecules within the solution shows relevant dynamic changes. In detail we found that water coordination around the two ions changes with the temperature. For T < 277 K, water molecules are more localized around chlorine ions down to about 260 K where the amplitude of the Gaussian component of the chlorine shell displays a maximum and that of the lithium shell shows a minimum. The magnetization of the bulk-like water component shows a monotonic increases on decreasing the temperature until about 227 K where it assumes its maximum value. Finally, the maximum in T2∗ for all the three Gaussian components at about 225 K signals the onset of the water liquid–liquid transformation into a four-coordinated low-density liquid characterized by an extended hydrogen bond network that dominates the thermodynamic properties of the system. References [1] C.A. Angell, Science 267 (1995) 1924. [2] F. Mallamace, C. Branca, C. Corsaro, N. Leone, J. Spooren, S.-H. Chen, H.E. Stanley, Proc. Natl. Acad. Sci. USA 107 (2010) 22457. [3] F. Mallamace, C. Corsaro, N. Leone, V. Villari, N. Micali, S.-H. Chen, Sci. Rep. 4 (2014) 3747.
C. Corsaro et al. / Physica A 442 (2016) 261–267 [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46]
267
Q. Sun, C. Zhou, Y. Yue, L. Hu, J. Phys. Chem. Lett. 5 (2014) 1170. E. Mamontov, J. Phys. Chem. B 113 (2009) 14073. F. Mallamace, C. Corsaro, H.E. Stanley, D. Mallamace, S.-H. Chen, J. Chem. Phys. 139 (2013) 214502. C.P. Royall, S.R. Williams, Phys. Rep. 560 (2015) 1. F. Mallamace, C. Corsaro, H.E. Stanley, S.-H. Chen, Eur. Phys. J. E 34 (2011) 94. S.-H. Chong, S.-H. Chen, F. Mallamace, J. Phys.: Condens. Matter. 21 (2009) 504101. F. Mallamace, C. Corsaro, N. Micali, V. Villari, N. Leone, S.-H. Chen, AIP Conf. Proc. 1518 (2013) 67. F. Mallamace, C. Corsaro, D. Mallamace, S. Vasi, C. Vasi, H.E. Stanley, J. Chem. Phys. 141 (2014) 18C504. F. Mallamace, C. Branca, C. Corsaro, N. Leone, J. Spooren, H.E. Stanley, S.-H. Chen, J. Phys. Chem. B 114 (2010) 1870–1878. F. Mallamace, C. Corsaro, P. Baglioni, E. Fratini, S.-H. Chen, J. Phys.: Condens. Matter. 24 (2012) 064103. P. Gallo, M. Rovere, S.-H. Chen, J. Phys. Chem. Lett. 1 (2010) 729–733. P. Gallo, D. Corradini, M. Rovere, J. Chem. Phys. 139 (2013) 204503. D.T. Limmer, D. Chandler, J. Chem. Phys. 135 (2011) 134503. J.C. Palmer, F. Martelli, Y. Liu, R. Car, A.Z. Panagiotopoulos, P.G. Debenedetti, Nature 510 (2014) 385. A. Cupane, M. Fomina, I. Piazza, J. Peters, G. Schiró, Phys. Rev. Lett. 113 (2014) 215701. A. Cupane, M. Fomina, G. Schiró, J. Chem. Phys. 141 (2014) 18C510. A. Elarby-Aouizerat, J.-F. Jal, P. Chieux, J.M. Letoffe, P. Claudy, J. Dupuy, J. Non-Cryst. Solids 104 (1988) 203. P.O. Maurin, J. Dupuy-Phillon, J.-F. Jal, N. Asahi, T. Kamiyama, J. Kawamura, Y. Nakamura, Progr. Theoret. Phys. Suppl. 126 (1997) 141. F. Aliotta, M. Pochylski, R.C. Ponterio, F. Saija, G. Salvato, C.S. Vasi, Phys. Rev. B 86 (2012) 134301. M.E. Gallina, L. Bove, C. Dreyfus, A. Polian, B. Bonello, R. Cucini, A. Taschin, R. Torre, R.M. Pick, J. Chem. Phys. 131 (2009) 124504. L. Bove, C. Dreyfus, R. Torre, R.M. Pick, J. Chem. Phys. 139 (2013) 044501. S.C. Santucci, L. Comez, F. Scarponi, G. Monaco, R. Verbeni, J.-F. Legrand, C. Masciovecchio, A. Gessini, D. Fioretto, J. Chem. Phys. 131 (2009) 154507. E. Mamontov, A. Faraone, E.W. Hagaman, K.S. Han, E. Fratini, J. Phys. Chem. B 114 (2010) 16737. M. Nakanishi, P. Griffin, E. Mamontov, A.P. Sokolov, J. Chem. Phys. 136 (2012) 124512. D.A Turton, C. Corsaro, D.F. Martin, F. Mallamace, K. Wynne, Phys. Chem. Chem. Phys. 14 (2012) 8067. D.A. Turton, C. Corsaro, M. Candelaresi, A. Brownlie, K.R. Seddon, F. Mallamace, K. Wynne, Faraday Discuss. 150 (2011) 493. Z. Shi, P.G. Debenedetti, F.H. Stillinger, J. Chem. Phys. 138 (2013) 12A526. L. Bove, C. Dreyfus, A. Polian, B. Bonello, I. Malfanti, A. Taschin, R. Torre, R.M. Pick, J. Chem. Phys. 134 (2011) 034514. L. Le, V. Molinero, J. Phys. Chem. A 115 (2011) 5900. L. Petit, R. Vuilleumier, P. Maldivi, C. Adamo, J. Chem. Theory Comput. 4 (2008) 1040. D.G. Archer, R.W. Carter, J. Phys. Chem. B 104 (2000) 8563. D. Corradini, M. Rovere, P. Gallo, J. Chem. Phys. 132 (2010) 134508. J.L. Aragones, M. Rovere, C. Vega, P. Gallo, J. Phys. Chem. B 118 (2014) 7680. K.D. Collins, Biophys. J. 72 (1997) 65. F. Mallamace, M. Broccio, C. Corsaro, A. Faraone, L. Liu, C.-Y. Mou, S.-H. Chen, J. Phys.: Condens. Matter. 18 (2006) S2285–S2297. F. Mallamace, C. Corsaro, H.E. Stanley, Proc. Natl. Acad. Sci. USA 110 (2013) 4899. J. Russo, H. Tanaka, Nature Comm. 5 (2014) 3556. R.J. Speedy, J. Phys. Chem. 86 (1982) 982. P.H. Poole, F. Sciortino, U. Essmann, H.E. Stanley, Nature 360 (1992) 324. F. Mallamace, C. Corsaro, D. Mallamace, C. Vasi, H.E. Stanley, Faraday Discuss. 167 (2013) 95. L. Xu, P. Kumar, S.V. Buldyrev, S.-H. Chen, P.H. Poole, F. Sciortino, H.E. Stanley, Proc. Natl. Acad. Sci. USA 102 (2005) 16558. Y. Huang, X. Zhang, Z. Ma, Y. Zhou, W. Zheng, J. Zhou, C.Q. Sun, Coord. Chem. Rev. 285 (2015) 109. F. Mallamace, C. Corsaro, M. Broccio, C. Branca, N. González-Segredo, J. Spooren, S.-H. Chen, H.E. Stanley, Proc. Natl. Acad. Sci. USA 105 (2008) 12725–12729. [47] S. Klotz, L. Bove, T. Straessle, T.C. Hansen, A.M. Saitta, Nature Mater. 8 (2009) 405. [48] F. Mallamace, C. Corsaro, H.E. Stanley, Sci. Rep. 2 (2012) 993.
Contents lists available at ScienceDirect
Physica A journal homepage: www.elsevier.com/locate/physa
Dynamics of water clusters in solution with LiCl Carmelo Corsaro a,b,∗ , Domenico Mallamace c , Nicola Cicero c,d , Sebastiano Vasi b , Giacomo Dugo c,d , Francesco Mallamace a,b,e a
CNR-IPCF, Istituto per i Processi Chimico-Fisici, Viale F. Stagno D’Alcontres 37, 98158 Messina, Italy
b
Dipartimento di Fisica e Scienze della Terra, Università di Messina, Viale F. Stagno D’Alcontres 31, 98166 Messina, Italy
c
Dipartimento SASTAS, Università di Messina, Viale F. Stagno d’Alcontres 31, 98166 Messina, Italy
d
Science4Life, spin-off Università di Messina, Viale F. Stagno D’Alcontres 31, 98166 Messina, Italy
e
NSE Department, Massachusetts Institute of Technology, Cambridge MA 02139, USA
highlights • We observe three different water clusters in a solution with LiCl at eutectic point. • Two important dynamical changes occur at two relevant temperatures for water. • The driving force is the tendency of water to develop its characteristic HB network.
article
info
Article history: Received 7 May 2015 Received in revised form 30 July 2015 Available online 11 September 2015 Keywords: Lithium chloride Dynamical crossover Water solution
abstract In this work we study by means of Nuclear Magnetic Resonance spectroscopy the dynamics of the different water clusters that form within a solution with LiCl at eutectic concentration in the temperature range 320–205 K. This solution is considered a model system allowing the investigation of water properties in the deep supercooled regime in its bulk phase. Our data reveal two important dynamical changes occurring at two relevant temperatures for water: the highest temperature coincides with that of the water density maximum (277 K) and the lowest with that of the so-called dynamical crossover (≃225 K). We interpret our data in terms of the different influence that the ions exert on water by lowering the temperature and of the tendency that water displays to develop its characteristic hydrogen bond network. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The dynamics of supercooled glass forming materials can be highly nonlinear, especially for those glass-formers that are defined fragile because their structure changes rapidly when temperature changes [1]. In fact, Angell in 1995 has classified glass-forming liquids into two different classes by using the concepts of fragility and glass transition [1]. In the last years, some authors suggested a different and universal scenario by invoking the concept related to the socalled dynamical crossover or fragile-to-strong transition [2–5]. It was observed that all supercooled glass-forming materials show a dynamical transition, that can be more or less evident depending on the system, before intervening the dynamical arrest or glass transition [2,4,6]. The higher is the temperature the higher is the number of degrees of freedom that the system can explore. Therefore, the corresponding dynamics of the system is characterized by multiple relaxations that
∗
Corresponding author at: Dipartimento di Fisica e Scienze della Terra, Università di Messina, Viale F. Stagno D’Alcontres 31, 98166 Messina, Italy. E-mail address: [email protected] (C. Corsaro).
http://dx.doi.org/10.1016/j.physa.2015.09.008 0378-4371/© 2015 Elsevier B.V. All rights reserved.
262
C. Corsaro et al. / Physica A 442 (2016) 261–267
Fig. 1. The thermal evolution of normalized SE ratio. It is noteworthy the sharp increase below 240 K.
involve the exploration of the local minima of the energy landscape [7]. By lowering the temperature, and thus the number of allowed energy states, the system can only experience a dynamics characterized by hopping processes between minima of uniform height [8]. This scenario is consistent with the theoretical predictions of the extended version of the Mode Coupling Theory (MCT) for which the critical temperature of the ideal version of the theory corresponds to the temperature of a dynamical crossover [3,9,10]. In other words, the fluids dynamics is characterized by two different behaviors above and below a precise temperature. Theory agrees with experiments that the dynamics of the system can be considered as the sum of two contributions: that of the ideal MCT i.e., the cage dynamics and that of the EMCT i.e., the hopping process. The cage dynamics shows a temperature dependence that can be represented by a power law function diverging at the MCT critical temperature. Below it the hopping process is the only possible dynamics that shows an Arrhenius dependence with temperature [9,10]. The detailed analysis of the dynamical behavior of water is relevant for pure science and technological applications. In particular water dynamics is more intriguing in the supercooled regime and there is a big and actual debate on the existence and on the significance of the dynamical crossover [11–19]. This liquid–liquid transition was observed experimentally for water in particular environments and in MD simulations. However, it is hard to supercool bulk water below the homogeneous nucleation temperature. It is well known that an eutectic solution of water in LiCl can be cooled to 200 K without the occurrence of any crystallization process [20,21]. For that reason, aqueous solutions of LiCl were used to investigate the occurrence of the dynamical crossover and contrasting scenarios have emerged [5,22–27]. The existence of different results seems, at least in part, due to the usage of different techniques that cover (and probe) different dynamical regimes [4,22,24,27]. The coexistence of different local structures and of the associated relaxations does not allow a simple data interpretation. For example NMR and OKE experiments did not observe any abrupt change in the water dynamics by means of the self-diffusion and time constant, respectively. However, by considering the stretching exponent of the time constant distribution, a dramatic change at the temperature of about 240 K was observed [28]. Moreover, by taking into account the Stokes–Einstein (SE) relation, it breaks just about that temperature signaling the decoupling between transport properties [29]. In Fig. 1 we report the Stokes–Einstein ratio, Dη/T, obtained by the data reported in Ref. [29], as a function of T and normalized with respect to its value at the highest temperature. As one can see, for T & 240 K this quantity is nearly constant reflecting the validity of the SE relation that instead breaks below 240 K where the reported quantity shows a sharp increase. The related decoupling between translational and rotational dynamics has been explained by invoking the existence of dynamic heterogeneities, which refers to the presence of transient spatially separated regions with vastly different relaxation times [30]. Optical and dielectric spectroscopies have detected the presence of extra signals such as an excess wing for temperatures below 225 K but the interpretations are somehow contradictory [27,31]. Furthermore, Molecular Dynamics (MD) simulations showed that the water tendency to develop its characteristic tetrahedral network provokes a segregation of the system. The ions are not included in the network of water molecules and contribute to the formation of solute rich regions within the system [32,33]. Previous studies have shown that water properties are not too altered by the presence of ions suggesting that it can be considered a model system for studying deeply supercooled water in its bulk phase [21,23–26,28,34,35]. At the same time, it was shown that the local tetrahedral structure of water is partially distorted by the presence of ions [36]. In particular, according to the Collins scenario, Li cation is an enhancer of the water tetrahedral structure (structure maker) whereas the
C. Corsaro et al. / Physica A 442 (2016) 261–267
263
Fig. 2. Snapshot of the different tetrahedral structures that water can adopt within the considered solution. From the left to the right the tetrahedral structure is increasingly open and less rigid [36].
Cl anion tends to disrupt water structure (structure breaker) [37]. As a consequence the lithium solvation shell is more structured and less mobile with respect to that of chlorine that is less structured and more mobile. As a matter of fact, ab-initio calculations show that in a tetrahedron of water molecules coordinated by a lithium cation, the Li-O radial distance is about 0.19 nm whereas in a tetrahedron of water molecules coordinated by a chlorine anion the Cl-O radial distance is about 0.31 nm (Fig. 2). Note that the O-O distance in pure water and in solution with LiCl is about 0.27 nm [36]. In Fig. 2 we report a snapshot of the different tetrahedral structures that water can adopt within the solution. From the left to the right the tetrahedral structure is increasingly open and less rigid [36]. The presence of tetrahedral structures makes the water/LiCl solution suitable for studying the bulk properties of water in the deep supercooled regime. In this paper we study by means of proton NMR spectroscopy the dynamics of the different water clusters that coexist within the solution. In fact, many studies demonstrated that in aqueous solution of LiCL, especially at high concentration, transient local structures exist around the two types of ions. In particular, the water solvation shell around Li cation is more rigid with respect to that found in pure water and even more with respect to that around Cl anion [33,36]. Our aim is to characterize the water dynamics in such a bulk environment in a region of the water phase diagram that is not accessible for pure water. In fact, for T < 235 K (the homogeneous nucleation temperature) the crystallization rate is too fast for any experimental observation. Therefore, in order to study liquid water within the so-called ‘‘no man’s land’’, confining systems or proper aqueous solutions have to be used [11,12,38]. Understanding the anomalous behavior of water has been the subject of many experimental and theoretical works that are still under consideration (see e.g. [39,40]). Starting from the stability limit hypothesis of Speedy [41], the scenario that is just receiving many independent confirmations is that of the liquid–liquid transition associated with a second critical point for water [17,42]. This liquid–liquid transition, occurring from a high density liquid at higher temperatures to a low density liquid at lower temperatures, seems to be connected with the so-called Widom line where thermodynamic response functions take on extrema values [39,43,44]. Even if the water/LiCl system is actually very studied, there is no complete consensus about the occurrence of the dynamical crossover in the temperature interval 210–230 K and about its origin [24–29,31]. As above mentioned, the dynamical crossover has been observed for water in different environments [12,13] and it has been suggested that can also be generalized for every liquid approaching the dynamical arrest before intervening the glass transition [2,8]. In our study we use NMR spectroscopy because it is a local probe and, even if we look at averaged relaxational dynamics, we are able to discriminate and follow separately the dynamics of the different local structures of water within the considered solution. We want to probe the hydrogen-bond relaxation dynamics because it is crucial for having a deep insight into water properties and anomalies [45]. In particular, we have acquired the NMR spectra by cooling from 320 K to 205 K each 5 K and analyzed the line-shape of the water peak. By extracting the relevant observables such as the apparent spin–spin relaxation time, T2∗ , for each of the detected contributions, we were able to follow and characterize the peculiar dynamics of the different water clusters that form in the solution. 2. Materials and methods The water/LiCl solution at the eutectic concentration of 6.76 M was prepared starting by anhydrous LiCl and deionized water (Sigma-Aldrich) by using Mohr’s method. NMR experiments were performed at atmospheric pressure with a Bruker Avance spectrometer operating at 700 MHz (proton Larmor frequency) equipped by a probehead with inner coil optimized for proton observation. The investigated temperature range was 205 K < T < 320 K with an accuracy of ±0.1 K and the temperature was calibrated by means of the Bruker standard sample of 4% of CH3 OH in CD3 OD. The acquisition parameters are the following: duration of the hard pulse 20 µs and relative attenuation 0.3 dB; spectral width 15 kHz; 64 k points in the time domain; 8 transients and 5 s of relaxation time. After the Fourier transformation and phase and baseline correction, we observed that the best fit of the water peak was obtained by using a Gaussian deconvolution with three components, reported for the highest and lowest measured temperatures in Fig. 3. For what concerns the fitting parameters we have an uncertainty below 1% and correlations between 0.5 and 0.9.
264
C. Corsaro et al. / Physica A 442 (2016) 261–267
Fig. 3. The Gaussian deconvolution of the proton NMR spectra of water/LiCl solution at eutectic point at the highest (left) and lowest (right) measured temperatures.
3. Results and discussions The measured 1 H NMR spectra of water/LiCl solution at eutectic concentration display only one signal and it belongs to water. From an inspection of Fig. 3, it is possible to notice that the peak position is lower than that in pure water even at the lowest measured temperature. In fact, the proton NMR peak position of water increases with temperature due to the increasing shielding effect that each water molecule feels during the development of its hydrogen bonded network [46]. Its chemical shift value within LiCl solution is lower because of the presence of electric charges in solution that, being in motion, influence the local magnetic field experienced by water protons. Moreover, the peak at 205 K is more intense and broader than at 320 K due to the competing effects of increasing polarization and decreasing mobility of water molecules on lowering the temperature. For all the investigated temperature ranges, the best fit of the water peak, reported in Fig. 3, was obtained by using three different Gaussian components. As above mentioned, theoretical studies showed that in water/LiCl solutions, there are different clusters of water molecules structured by lithium and chlorine ions [32,33]. Furthermore, at the eutectic concentration, there are 7 water molecules per molecule of LiCl, hence some water molecules make a dynamic bridge between the two ions pointing the oxygen towards lithium and the hydrogens towards chlorine. Therefore, we ascribed the three Gaussian components to the three different local water structures that can be found in the solution. In particular we named the different water structures as: bulk-like, around Li and around Cl. Note that these different local structures are transient in nature and can be explained in terms of nanosegregation. By means of our experimental technique we can discriminate between the solvation shells of water around lithium cation and chlorine anion. In Fig. 4 we report the intensity of the three Gaussian components, that is the magnetization of the different water clusters, corrected by the Curie law, as a function of the inverse temperature. We ascribe the most intense Gaussian component (circles) to bulk-like water (water molecules interacting with each other), the broader one (triangles) to water molecules around Li cation and the sharper one (squares) to water molecules around chlorine anion. Our choice is motivated by the results of the cited theoretical studies about the higher rigidity of the water solvation shell around Li and of the more mobile structure that water forms around Cl [32,33,36]. Note that, the unusual increase of the magnetization of all the three Gaussian components of the water peak on lowering the temperature from 320 K to T ≃ 277 K is due to the fact that water molecules within salt solutions are highly polarized and their polarization increases on decreasing the thermal energy. For T < 277 K, the magnetization of water molecules interacting with lithium and chlorine ions follows two different thermal behaviors that cross each other at about 227 K. This trend can be explained if we assume that water coordination around the two ions changes with the temperature. In particular, our data suggest that for 320 K > T > 277 K water molecules make part of the coordination spheres of both lithium and chlorine with the same weight or probability. By lowering the temperature, water molecules are more localized around chlorine ions down to about 260 K where the amplitude of the Gaussian component of the chlorine shell displays a maximum and that of the lithium shell shows a minimum suggesting an inversion of the trend. At about 227 K the magnetization corresponding to the two solvation shells assumes the same value and decreases on lowering the temperature.
C. Corsaro et al. / Physica A 442 (2016) 261–267
265
Fig. 4. The thermal evolution of NMR magnetization of the different water components for water/LiCl solution at eutectic concentration. The value is corrected by the Curie law. Vertical lines refer to the temperatures of 277 K (solid) and 227 K (dotted).
For what concerns the magnetization of the bulk-like water component, it shows a monotonic increase on decreasing the temperature until about 227 K where it assumes its maximum value. At this particular temperature, the correlation length of water molecules is maximum [39] and water fully develops its hydrogen bonded network. Li and Cl ions constitute only local defects of this extended network because they ‘‘prefer’’ (or have) to segregate from water, as happens in ice formation [47], forming a solute-rich water nanophase [32]. Therefore, below 225 K water molecules tendency to hydrogen bond with each others is the strongest interaction that determines the thermodynamic properties of the system. This is confirmed by the decreasing of the magnetization of all Gaussian components for T < 225 K (Fig. 4). Finally, we have evaluated the apparent spin–spin relaxation time, T2∗ , of the proton nuclear magnetization by the width of the three Gaussian components as:
√ ∗
T2 =
2 ln 2
π ∆ν
(1)
where ∆ν is the full width at half maximum (in Hz) of each Gaussian component. The spin–spin relaxation time, is a measure of the strength of dipolar interaction between spins of the same species. Since different local structures of water manifest different dipolar interactions, it is a good probe to study their separate dynamical behavior. T2∗ represents the time needed by the transverse component of the macroscopic magnetization to vanish, due to the spin dephasing mechanism, in the plane orthogonal to the direction of the static magnetic field. In Fig. 5 we report T2∗ for each of the three Gaussian components as a function of the inverse temperature. Note that the T2∗ of all the three components has a very similar thermal behavior: starting from the highest temperature, they slowly increase reaching a maximum value at about 280 K. Moreover, all show a minimum at about 260 K and a maximum close to 225 K. Thus, our data reveal that even if different local water clusters exist within the solution and their relative concentrations (reflected by the corresponding magnetization value reported in Fig. 4) change with temperature, they have the same dynamics in the considered thermal range that includes the supercooled regime. This justifies why other experimental techniques such as the previous dynamical measurements of self-diffusion coefficient or relaxation time [26,28,29] have observed only one contribution. Besides, transient grating experiments proved how the clusters dynamics is not affected by the temperature until about 210 K, below which an additional signal appears [24]. Furthermore, large scale Molecular Dynamics studies suggested that below the liquid–liquid transformation temperature water becomes a four-coordinated low-density liquid (LDL) [32]. However, the existence of three temporal scales confirms the different rigidity (and thus mobility) of the local water structures when coordinated by the different ions. The maximum in T2∗ at about 225 K reinforces the suggestions that at this temperature water shows a liquid–liquid transformation becoming a four-coordinated low-density liquid characterized by an extended hydrogen bond network that dominates the thermodynamic properties of the system.
266
C. Corsaro et al. / Physica A 442 (2016) 261–267
Fig. 5. Plot of the apparent spin–spin relaxation time, T2∗ , of the proton nuclear magnetization for each of the three Gaussian components as a function of the inverse temperature for water/LiCl solution at eutectic point.
4. Conclusions Water is simultaneously the simplest and the most complex liquid. Although the water molecule consists of only three atoms, its high tendency to form an extended network of hydrogen bonds determines the existence of anomalous properties that indeed are responsible for life. Above the temperature of about 320 K, water can be considered a simple liquid (water molecules display essentially a local high density structure) because the strength and lifetime of hydrogen bonds are not enough to form the characteristic tetrahedral structure [48]. Below 320 K, when the high directionality of the hydrogen bond begins to dominate over the thermal disorder and the van der Waals interactions, the thermodynamic properties of water begin to be anomalous. In particular, many thermodynamic response functions of water show a critical-like behavior with an apparent divergence at about 228 K [39,43]. In this paper, we have studied by means of 1 H NMR spectroscopy the dynamics of the different water local structures that exist in an aqueous solution of lithium chloride at the eutectic concentration (6.76 M). This kind of system is considered a prototype that allows to study the thermodynamic properties of water in the deep supercooled regime without the need of confining environments. The temperature interval that we have considered, includes the supercooled regime and extends from 320 K to 205 K. In particular, we have used a Gaussian deconvolution with three components to best reproduce the obtained NMR peak of water. The three Gaussian components reflect the different local environments that water molecules can experience within the solution. We ascribed the most intense Gaussian component to bulk-like water, the broader to the water shell solvating the lithium cation and the sharper to the water in the chlorine coordination sphere. In fact the water local structure around the lithium is more rigid with respect to that found in bulk water. This is why lithium has the property to enhance the water structure (structure maker). On the contrary, the chlorine anion is a structure breaker and the water local structure surrounding it is more mobile than that of bulk water. Our data, in terms of the macroscopic magnetization and the apparent spin–spin relaxation time, T2∗ , reveal the existence of two characteristic temperatures at which the clustering organization of water molecules within the solution shows relevant dynamic changes. In detail we found that water coordination around the two ions changes with the temperature. For T < 277 K, water molecules are more localized around chlorine ions down to about 260 K where the amplitude of the Gaussian component of the chlorine shell displays a maximum and that of the lithium shell shows a minimum. The magnetization of the bulk-like water component shows a monotonic increases on decreasing the temperature until about 227 K where it assumes its maximum value. Finally, the maximum in T2∗ for all the three Gaussian components at about 225 K signals the onset of the water liquid–liquid transformation into a four-coordinated low-density liquid characterized by an extended hydrogen bond network that dominates the thermodynamic properties of the system. References [1] C.A. Angell, Science 267 (1995) 1924. [2] F. Mallamace, C. Branca, C. Corsaro, N. Leone, J. Spooren, S.-H. Chen, H.E. Stanley, Proc. Natl. Acad. Sci. USA 107 (2010) 22457. [3] F. Mallamace, C. Corsaro, N. Leone, V. Villari, N. Micali, S.-H. Chen, Sci. Rep. 4 (2014) 3747.
C. Corsaro et al. / Physica A 442 (2016) 261–267 [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46]
267
Q. Sun, C. Zhou, Y. Yue, L. Hu, J. Phys. Chem. Lett. 5 (2014) 1170. E. Mamontov, J. Phys. Chem. B 113 (2009) 14073. F. Mallamace, C. Corsaro, H.E. Stanley, D. Mallamace, S.-H. Chen, J. Chem. Phys. 139 (2013) 214502. C.P. Royall, S.R. Williams, Phys. Rep. 560 (2015) 1. F. Mallamace, C. Corsaro, H.E. Stanley, S.-H. Chen, Eur. Phys. J. E 34 (2011) 94. S.-H. Chong, S.-H. Chen, F. Mallamace, J. Phys.: Condens. Matter. 21 (2009) 504101. F. Mallamace, C. Corsaro, N. Micali, V. Villari, N. Leone, S.-H. Chen, AIP Conf. Proc. 1518 (2013) 67. F. Mallamace, C. Corsaro, D. Mallamace, S. Vasi, C. Vasi, H.E. Stanley, J. Chem. Phys. 141 (2014) 18C504. F. Mallamace, C. Branca, C. Corsaro, N. Leone, J. Spooren, H.E. Stanley, S.-H. Chen, J. Phys. Chem. B 114 (2010) 1870–1878. F. Mallamace, C. Corsaro, P. Baglioni, E. Fratini, S.-H. Chen, J. Phys.: Condens. Matter. 24 (2012) 064103. P. Gallo, M. Rovere, S.-H. Chen, J. Phys. Chem. Lett. 1 (2010) 729–733. P. Gallo, D. Corradini, M. Rovere, J. Chem. Phys. 139 (2013) 204503. D.T. Limmer, D. Chandler, J. Chem. Phys. 135 (2011) 134503. J.C. Palmer, F. Martelli, Y. Liu, R. Car, A.Z. Panagiotopoulos, P.G. Debenedetti, Nature 510 (2014) 385. A. Cupane, M. Fomina, I. Piazza, J. Peters, G. Schiró, Phys. Rev. Lett. 113 (2014) 215701. A. Cupane, M. Fomina, G. Schiró, J. Chem. Phys. 141 (2014) 18C510. A. Elarby-Aouizerat, J.-F. Jal, P. Chieux, J.M. Letoffe, P. Claudy, J. Dupuy, J. Non-Cryst. Solids 104 (1988) 203. P.O. Maurin, J. Dupuy-Phillon, J.-F. Jal, N. Asahi, T. Kamiyama, J. Kawamura, Y. Nakamura, Progr. Theoret. Phys. Suppl. 126 (1997) 141. F. Aliotta, M. Pochylski, R.C. Ponterio, F. Saija, G. Salvato, C.S. Vasi, Phys. Rev. B 86 (2012) 134301. M.E. Gallina, L. Bove, C. Dreyfus, A. Polian, B. Bonello, R. Cucini, A. Taschin, R. Torre, R.M. Pick, J. Chem. Phys. 131 (2009) 124504. L. Bove, C. Dreyfus, R. Torre, R.M. Pick, J. Chem. Phys. 139 (2013) 044501. S.C. Santucci, L. Comez, F. Scarponi, G. Monaco, R. Verbeni, J.-F. Legrand, C. Masciovecchio, A. Gessini, D. Fioretto, J. Chem. Phys. 131 (2009) 154507. E. Mamontov, A. Faraone, E.W. Hagaman, K.S. Han, E. Fratini, J. Phys. Chem. B 114 (2010) 16737. M. Nakanishi, P. Griffin, E. Mamontov, A.P. Sokolov, J. Chem. Phys. 136 (2012) 124512. D.A Turton, C. Corsaro, D.F. Martin, F. Mallamace, K. Wynne, Phys. Chem. Chem. Phys. 14 (2012) 8067. D.A. Turton, C. Corsaro, M. Candelaresi, A. Brownlie, K.R. Seddon, F. Mallamace, K. Wynne, Faraday Discuss. 150 (2011) 493. Z. Shi, P.G. Debenedetti, F.H. Stillinger, J. Chem. Phys. 138 (2013) 12A526. L. Bove, C. Dreyfus, A. Polian, B. Bonello, I. Malfanti, A. Taschin, R. Torre, R.M. Pick, J. Chem. Phys. 134 (2011) 034514. L. Le, V. Molinero, J. Phys. Chem. A 115 (2011) 5900. L. Petit, R. Vuilleumier, P. Maldivi, C. Adamo, J. Chem. Theory Comput. 4 (2008) 1040. D.G. Archer, R.W. Carter, J. Phys. Chem. B 104 (2000) 8563. D. Corradini, M. Rovere, P. Gallo, J. Chem. Phys. 132 (2010) 134508. J.L. Aragones, M. Rovere, C. Vega, P. Gallo, J. Phys. Chem. B 118 (2014) 7680. K.D. Collins, Biophys. J. 72 (1997) 65. F. Mallamace, M. Broccio, C. Corsaro, A. Faraone, L. Liu, C.-Y. Mou, S.-H. Chen, J. Phys.: Condens. Matter. 18 (2006) S2285–S2297. F. Mallamace, C. Corsaro, H.E. Stanley, Proc. Natl. Acad. Sci. USA 110 (2013) 4899. J. Russo, H. Tanaka, Nature Comm. 5 (2014) 3556. R.J. Speedy, J. Phys. Chem. 86 (1982) 982. P.H. Poole, F. Sciortino, U. Essmann, H.E. Stanley, Nature 360 (1992) 324. F. Mallamace, C. Corsaro, D. Mallamace, C. Vasi, H.E. Stanley, Faraday Discuss. 167 (2013) 95. L. Xu, P. Kumar, S.V. Buldyrev, S.-H. Chen, P.H. Poole, F. Sciortino, H.E. Stanley, Proc. Natl. Acad. Sci. USA 102 (2005) 16558. Y. Huang, X. Zhang, Z. Ma, Y. Zhou, W. Zheng, J. Zhou, C.Q. Sun, Coord. Chem. Rev. 285 (2015) 109. F. Mallamace, C. Corsaro, M. Broccio, C. Branca, N. González-Segredo, J. Spooren, S.-H. Chen, H.E. Stanley, Proc. Natl. Acad. Sci. USA 105 (2008) 12725–12729. [47] S. Klotz, L. Bove, T. Straessle, T.C. Hansen, A.M. Saitta, Nature Mater. 8 (2009) 405. [48] F. Mallamace, C. Corsaro, H.E. Stanley, Sci. Rep. 2 (2012) 993.