Home
Search
Collections
Journals
About
Contact us
My IOPscience
Hydroxylamine-doping effect on the Tg of 160 K for water confined in silica-gel nanopores
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2013 J. Phys.: Condens. Matter 25 465110 (http://iopscience.iop.org/0953-8984/25/46/465110) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 128.6.218.72 This content was downloaded on 01/10/2015 at 14:26
Please note that terms and conditions apply.
IOP PUBLISHING
JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 25 (2013) 465110 (6pp)
doi:10.1088/0953-8984/25/46/465110
Hydroxylamine-doping effect on the Tg of 160 K for water confined in silica-gel nanopores A Nagoe and M Oguni Department of Chemistry, Graduate School of Science and Engineering, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8551, Japan E-mail: [email protected] (A Nagoe)
Received 30 August 2013, in final form 22 September 2013 Published 28 October 2013 Online at stacks.iop.org/JPhysCM/25/465110 Abstract The glass transition behavior of hydroxylamine (HA) aqueous solutions in bulk and confined in silica-gel nanopores with average width of 1.1 nm was studied by means of differential scanning calorimetry measurements and adiabatic calorimetry. The glass transition temperature (Tg ) of the confined solution with high HA mole-fraction (xHA ) was essentially the same as the value of the bulk. This suggests that the nano-size confinement affects the Tg of HA aqueous solution little. Meanwhile, the bulk solution with xHA < 0.3 revealed partial crystallization on cooling and, on the other hand, the confined solution with the same xHA did not crystallize. The Tg of the xHA = 0.076 confined solution was 174 K which is higher than the value of 160 K for pure water confined in the same silica-gel pores. This demonstrates that HA doping leads to no abrupt Tg -decrease, unlike doping of all the other second components reported so far, suggesting that HA is set neatly in a hydrogen-bond network formed by water molecules. We discuss the xHA dependence of Tg for the HA aqueous solutions from a viewpoint related to peculiar phase-behavior of pure water. Considering that the xHA = 0.076 aqueous solution revealed no anomaly compared with pure water, it was recognized as corresponding to the high-temperature phase of pure water.
1. Introduction
scanning calorimetry (DSC) heating curves [6]. Another method for producing glassy water is to dope a second component into water to suppress the formation of a crystal nucleus of ice; namely, highly concentrated aqueous solutions are used as samples. By extrapolating the Tg values of these glass-forming solutions with respect to the concentration, the Tg of pure water was estimated to be around 136 K [4, 7, 10, 12]. However, there remains a fatal issue: even slightly doping the second component removes water’s peculiar features of many properties such as a heat-capacity rise on cooling in the supercooled state [16, 17]. In other words, the value of 136 K corresponds to the Tg of water that shows no peculiar features and differs significantly from an aggregate of water molecules only. Angell suggested that for aqueous solutions an abrupt rise in Tg on approaching pure water might occur in a dilute range because the hydrogen-bond network which pure water makes develops only in the situation where water molecules
Water is the most abundant substance on the Earth. It is further well known by many peculiar behaviors revealed in the liquid state [1–3]. The peculiarity becomes remarkable at low temperatures in the supercooled state. No detail of the real behaviors has been, however, clarified because its crystallization to ice occurs [1–5]. The glass transition temperature (Tg ) is one such property. The Tg of supercooled water has been debated for a long time, mainly concerning whether it is 136 K or 165 K [3–15]. Amorphous solid water produced by vapor deposition reveals a thermal anomaly, at 136 K, resembling a glass transition [6]. However, some research groups have pointed out that the enthalpy-relaxation behavior differs from those observed in other glassy materials and insisted that a real glass transition occurs at around 165 K above the crystallization temperature of around 150 K on differential 0953-8984/13/465110+06$33.00
1
c 2013 IOP Publishing Ltd Printed in the UK & the USA
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
to 210 K and exhibits no continuous increase with increasing pore size. The appearance of this high Tg value suggests the existence of another structure with a more developed hydrogen-bond network. Therefore, the Tg value of 160 K was interpreted as being detected due to the pore size of 1.1 nm which is too small for water to form a fully developed hydrogen-bond network. However, there is no other work reporting this high Tg value of 210 K. Recently, we found that hydroxylamine (abbreviated as HA) aqueous solution confined in the silica MCM-41 pores shows a liquid–liquid first-order phase transition [21], which is connected to the phenomenon observed as a peculiar Cp -maximum for nanopore water [22, 23]. The HA molecule is considered to fit well to the hydrogen-bond network formed by water molecules. Therefore, doping HA into water is expected to yield a quite different mole-fraction (xHA ) dependence of Tg from other aqueous solutions such as methanol solutions. In this work, we studied the HA doping effect on the Tg of 160 K for water confined within silica-gel nanopores with average width of 1.1 nm. The Tg values of the nano-confined HA aqueous solutions were also compared with those obtained for bulk solutions with the same xHA values.
Figure 1. The mole-fraction dependence of Tg in various aqueous systems: open triangles, bulk hydrogen peroxide solution [17]; open squares, bulk ethylene glycol solution [10]; open circles, bulk methanol (ME) solution [10]; filled circles, ME solution confined in silica-gel pores with average width of 1.1 nm [14]. The dotted lines are drawn as guides for eyes.
2. Experimental details possess the majority fraction [7]. The development of the network causes the viscosity to increase and the Tg to shift up in aqueous solutions [7]. The problem is again that crystallization prevents one from discerning the Tg -rise behavior in the dilute concentration range close to pure water. We confined water into silica-gel nanopores with an average width of 1.1 nm to keep it from crystallizing and observed its glass transition at 160 K [8]. This observation supports Angell et al’s suggestion [6, 7]. Further, as methanol (abbreviated as ME) is doped into water little by little, the Tg of the confined aqueous solution decreases quickly from 160 K, bends rather at an ME mole-fraction xME = 0.3, and accords with that of glass-forming concentrated solutions in bulk above xME = 0.3 (figure 1) [14]. Extrapolation of the xME dependence of Tg above xME = 0.3 toward xME = 0 gives Tg = 136 K for pure water. This means that the development of a hydrogen-bond network gradually proceeds with increasing water fraction below xME = 0.3 and suggests that the environment surrounding the water molecule is quite different between the highly concentrated solutions (giving an extrapolated value of Tg = 136 K for pure water) and real pure water with a hydrogen-bond network (Tg = 160 K). Some research groups have tried to understand this difference as being caused by a liquid–liquid phase transition of supercooled water [18, 19]; namely, the values Tg = 160 K and 136 K (or lower than 136 K) correspond to those of low-density and high-density water respectively [7, 20]. Furthermore, we detected a glass transition at 210 K [15] for the water confined within silica MCM-41 pores with average diameters of 2.0 and 2.1 nm, which are larger than the width (1.1 nm) of the silica-gel nanopores stated above. It is intriguing that the Tg of water jumps discretely from 165
2.1. Sample preparation Aqueous solution with xHA = 0.35 was purchased from Wako Pure Chemical Industries, Ltd, and solution samples with different xHA values were prepared by mixing it with distilled water at specified ratios. The silica-gel CARiACT Q3 was supplied by Fuji Silysia Chemical Ltd, and had a particle size of a few mm and pore voids with average width of 1.1 nm as assessed from the results of nitrogen-gas absorption measurements by the MP method [8, 24]. The confinement into the pores was executed by directly pouring each degassed HA aqueous solution onto silica-gel particles under vacuum after the insides of the silica-gel pores had been exhausted at 473 K for 24 h. The silica-gel particles loaded with the solution were held, under an atmosphere of helium gas, within a quartz-glass container to prevent from reacting with the inner wall of a gold-plated copper cell during calorimetric measurements. 2.2. Method for determination of Tg The glass transition behaviors of HA aqueous solutions in bulk were examined by DSC measurements. The measurements were performed with a Perkin–Elmer DSC 8500 at a heating rate of 10 K min−1 after precooling the sample rapidly at −50 K min−1 . The Tg value was determined as the middle point of the heat-flow jump that accompanied the transition. The Tg s of the solutions confined in silica-gel pores were determined, with an adiabatic calorimeter, by following the enthalpy-relaxation rates as a function of temperature in the heating direction [25, 26]. In general, glasses are located 2
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
Figure 3. Mole-fraction dependence of Tg in HA (open symbols) and ME (filled symbols) aqueous solutions: triangles, bulk HA solution; circles, internal part of the confined aqueous solution; squares, interfacial part of the confined solution; all data for ME solutions were taken from [14].
Figure 2. DSC cooling curves for bulk hydroxylamine (HA) aqueous solutions at a rate of −10 K min−1 : dashed line, xHA = 0.30; solid line, xHA = 0.35. The inset represents the corresponding heating curves at a rate of 10 K min−1 after the samples were quenched at −50 K min−1 .
remaining liquid was left unclear in the dilute xHA range. Above xHA > 0.3, we detected no crystallization event on cooling. Figure 2 shows the DSC curves on cooling, at −10 K min−1 , the HA aqueous solutions in bulk with xHA = 0.30 and 0.35. In addition, the solutions showed no anomaly indicating the occurrence of a phase transition and vitrified at around 180 K. Considering that the high- to low-density phase transition is of first order [21], this fact potentially means that the HA aqueous solutions became frozen into nonequilibrium glassy states as was the case in the high-temperature phase corresponding to the high-density phase hypothesized for pure water’s phase diagram [18, 19]. The inset of figure 2 displays the DSC curves on heating the frozen, glassy solutions at 10 K min−1 . The Tg s of the bulk solutions were determined by using the heating curves and are plotted with open circles in figure 3. The Tg at xHA = 0.35 is higher by 35 K than that of the hydrogen peroxide aqueous solution with the same mole-fraction (xH2 O2 ) [16, 17]. Moreover, the Tg value derived by extrapolating, with a straight line, those of the bulk HA aqueous solutions toward xHA = 0 is around 165 K, which is higher than the extrapolated value of 136 K by using other solutions [4, 7, 10, 12]. A similar mole-fraction dependence giving Tg = 160 K at x = 0 has also been reported in aqueous solutions of propylene-glycol oligomer [11]. However, the validity of the extrapolation also remains questionable because the crystallization of ice occurred in the dilute x range [7]. Murthy reported a result that raised doubts about the extrapolation for the propylene-glycol oligomer solutions [12]. In our previous studies, doping methanol or ethylene glycol led to depression of Tg for confined water only in a dilute x range, as is also shown in figure 1, while the bulk solutions with the same xs easily crystallized [4, 10, 13, 14].
in different enthalpic states according to whether they were precooled at high or low speeds; namely, the temperature at which the molecular configuration of the solution is frozen in and brought to a nonequilibrium state depends on the precooling speed. On approaching Tg in heating measurements, the glasses relax toward their respective equilibrium states with release of heat for high-enthalpic glasses precooled at a high speed or absorption of heat for low-enthalpic glasses precooled at a low speed. In this work, the Tg s were determined as the temperatures giving the maximum heat-absorption effect, namely, minimum −dH/dt, for the samples precooled at a lower speed. This method enables us to detect the dynamic properties of the samples directly. Therefore, it might be efficient even if the jump in heat capacity at Tg is small or the heat-capacity behavior is quite complex as compared with those of normal glassy pure liquids. In addition, the high sensitivity for heat release/absorption and high resolution for thermometry of the adiabatic calorimetry help us to detect glass transition events precisely. Although thermodynamic properties such as bulk modulus and thermal expansion have been used besides calorimetry to determine the Tg [27], the presence of silica makes the determination rather difficult.
3. Results and discussion 3.1. The DSC curves of bulk HA aqueous solutions For the bulk solutions in the xHA range below 0.3, some of the water molecules crystallized in the cooling process. The xHA , and therefore the behavior, of the aqueous solution 3
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
is very small even in very small-in-size pores. This fact also confirms the effectiveness of the confinement method for exploring the glass transitions of these aqueous solutions. Two possible reasons are considered for this fact. One is that the size of the cooperative rearrangement region to determine the glass transition temperature is smaller than the width of the silica-gel nanopores. The other is that the rearrangement of molecules in the aqueous solutions involves essentially no cooperative nature in the unit to be activated. In view of the fact that the ME aqueous solutions with xME < 0.3 form a hydrogen-bond network inherent to water, the hydrogen-bond network formed at xHA = 0.076 is judged as the one inherent to water. Therefore, the higher-in-temperature Tg s would be linked as a function of xHA smoothly to the value of 160 K for pure water, as drawn with a solid line in figure 3. The xHA dependence of Tg in the very dilute range is completely opposite to the xME dependence; the Tg of water is increased and decreased by doping HA and ME respectively. This indicates that the presence of HA enhances while that of ME inhibits the formation of the water’s hydrogen-bond network. The lower-in-temperature glass transition due to the interfacial molecules was observed at 136 K for the solution at xHA = 0.35, although the heat-release and -absorption effects are small. In the xHA = 0.076 solution, the transition was observed clearly at 126 K. It is noticed that the Tg for the interfacial molecules shifts to higher temperatures with doping HA molecules while it shifts to lower temperatures with doping ME molecules. The Tg of pure HA is much higher than that of ME because of the capacity to form hydrogen bonds. Therefore, the interfacial water molecules close to HA molecules must be bound rather tightly and their Tg is expected to increase.
Figure 4. Enthalpy-relaxation rates as a function of temperature (−dH/dt) in the heating direction observed in the HA aqueous solutions within pores with average width of 1.1 nm: filled circles, sample cooled slowly over the temperature region; open circles, sample cooled rapidly over the region. Hysteresis between the open and the filled circles represents the existence of glass transitions. The data for xHA = 0 were taken from [8].
3.3. The xHA dependence of Tg in the dilute HA aqueous solutions and its relation to water’s anomaly
3.2. The glass transition behavior, observed by adiabatic calorimetry, of the aqueous solutions confined in silica-gel pores
As stated above, the glass transition of the internal solution within the pores can be recognized as exhibiting the same phenomenon as that of the solution in bulk. ME or ethylene glycol aqueous solutions showed a steep decrease in the Tg with doping the respective second components in the dilute x range [13, 14]. On the other hand, HA aqueous solution revealed rather an increase in the Tg . This means that HA doping maintains and even strengthens the hydrogen-bond network formed by the water molecules. This attribute of HA is considered to originate from the molecular structure of HA. HA has the same 1:1 ratio of hydrogen atoms to lone electron-pairs at 3–3 in number as the water molecule with 2–2; namely, HA can be recognized just as a water-molecular dimer. Therefore, HA can form a hydrogen-bond network fully in the same way as water. In reality, this is also supported by the fact that the Tg of the xHA = 0.35 solution is higher by 35 K than those of hydrogen peroxide or hydrazine aqueous solutions with x = 0.35 [16, 17]. Some research groups have predicted that pure water’s Tg is 160 K [6, 7, 11, 18]. Angell et al insisted that the high Tg value of 160 K for pure water originates from the
Figure 4 shows the results of enthalpy-relaxation rates for xHA = 0.076 and 0.35 aqueous solutions, as well as for pure water, confined in silica-gel pores with a width of 1.1 nm. The presence of heat-release and -absorption effects for the samples precooled rapidly and slowly respectively evidences the existence of a glass transition. The pure water revealed two glass transitions at 115 and 160 K, as shown in figure 4(a) [8]. The lower-in-temperature transition was assigned as due to the water molecules located on the pore walls, namely, the interfacial water molecules. The higher-in-temperature transition was assigned as due to the water molecules located in the central parts of the pore voids, namely, the internal water molecules. The Tg values derived are plotted in figure 3 with open circles and squares. The Tg at xHA = 0.35 is very close to the value for the bulk solution with the same xHA . The closeness is just the behavior found for ME aqueous solutions [14]. This means that the effect of nano-confinement on the Tg s of aqueous solutions of such simple hydrophilic compounds 4
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
internal water increased in the range xME < 0.3 and simultaneously an interfacial solution layer was formed to reveal a new glass transition around 115 K in the same range [14]. This was understood as the internal solution with the high water fraction making a hydrogen-bond network resembling the structure of the high-temperature phase of pure water and as the interfacial solution buffering the network structure from the pore wall mainly consisting of a rather hydrophobic siloxane-bond network. In view of the fact that the pore wall was formed by the same siloxane-bond network, the role of the buffer of interfacial water molecules is kept in the HA aqueous solution. It is noticed here that the Tg of the interfacial water in the HA aqueous solution revealed an obvious increase with increase of the fraction of dopant HA and that the glass transition due to the interfacial water was observed only with a very small effect in the xHA = 0.35 solution. This means that the interfacial water molecules were bound gradually more strongly by being surrounded by more HA molecules as well as the number of interfacial water molecules accordingly decreasing. It is not clear at present how the role of the buffer appears in the interfacial HA molecules. It would be reasonable to consider that the heat-release or -absorption effects of the glass transition are hidden in the wide higher-in-temperature range of over 140–190 K.
peculiar increase of Cp of water on decreasing the temperature below 0 ◦ C [7, 18]. The Cp -increase behavior below 0 ◦ C was connected with a liquid–liquid phase transition by many research groups [18, 19, 23, 28]. In this view, the phase exhibiting Tg = 160 K corresponds to the low-temperature low-density phase of water [7, 18]. A decrease in water’s Tg of 160 K and a reduction of the Cp singularity were observed in a parallel way for the aqueous solutions of ordinary second components [6, 7]. However, the xME dependence of Tg for ME aqueous solutions changes gradually for xME s below 0.3 [14], although DSC measurements of hydrogen peroxide aqueous solutions disclose remarkable reduction of the anomalous Cp component for supercooled water by doping hydrogen peroxide in a small amount of xH2 O2 = 0.1 [16, 17]. The two phenomena that the Cp anomaly disappears and the x dependence of Tg bends with increasing x occur at obviously different xs. Therefore, it is difficult to conclude that the two phenomena result from the same origin. In fact, the xHA dependence of Tg for HA aqueous solutions indicates that they maintain their hydrogen-bond networks even in the highly concentrated range. As reported elsewhere [21], HA aqueous solutions reveal a liquid–liquid phase transition the thermal behavior of which is linked smoothly with the heat-capacity maximum at 230 K for pure water [22]. However, the transition was found to be of first order, and the HA aqueous solutions could be easily supercooled and displayed no singularity, due to the phase transition, on the DSC curve as stated above. The glass transition observed presently in the HA aqueous solutions confined within the silica-gel pores is thus recognized as taking place in the high-temperature phase. It is concluded that the Tg of 160 K for pure water is attributed to the high-temperature phase. In other words, the Tg = 136 K, derived by extrapolating the composition dependence of the Tg s for the concentrated solutions without forming a hydrogen-bond network, is irrelevant to the Tg of pure water in the high-temperature phase. According to our consideration, the dynamic properties of water molecules in concentrated solutions with ordinary second components differ greatly from those of pure water in the high-temperature phase. If a low-temperature phase of pure water was formed through the liquid–liquid transition, a Tg jump would potentially occur [15, 18, 29]. This is expected in the case where the water’s Tg is much higher than 160 K. We actually reported that the pure water confined within mesoporous silica MCM-41 with an average diameter of 2.1 nm shows a glass transition at 210 K, while for 1.7 nm there is a transition at 165 K and for 2.0 nm there are two transitions at 165 and 210 K; namely, the Tg increased discretely with increase of the pore size [15]. It is reasonable to interpret that this discrete increase corresponds to the change of the water structure from the high- to the low-temperature phase. Further study is, of course, needed to confirm this interpretation.
4. Conclusions We successfully observed glass transitions in HA aqueous solutions in bulk at high xHA s by DSC and of dilute HA aqueous solutions confined in silica-gel pores with a width of 1.1 nm by adiabatic calorimetry. The Tg of 160 K for the confined pure water showed no steep decrease but rather an increase with increase in xHA , quite distinctly from other second components such as ME. The decrease of Tg is understood as reflecting collapse of the hydrogen-bond network. Therefore, it is concluded that the HA aqueous solutions maintain the hydrogen-bond network of pure water even in the high xHA range. Besides the fact that the HA aqueous solutions show a liquid–liquid phase transition in the state confined within silica MCM-41 pores [21], this peculiar doping effect on the Tg of water is very fascinating. It is suggested that the value of Tg (=160 K) for the confined water corresponds to that of pure water in the high-temperature phase. This leads us to predict that water displaying the glass transition at a temperature higher than 160 K is in the low-temperature phase with a well-developed hydrogen-bond network. This prediction is consistent with the assertion, reported previously [15], that the Tg of water is 210 K.
References [1] Rasmussen D H and MacKenzie A P 1973 J. Chem. Phys. 59 5003 [2] Kanno H and Angell C A 1979 J. Chem. Phys. 70 4008 [3] Debenedetti P G 2003 J. Phys.: Condens. Matter 15 R1669 [4] Rasmussen D H and Mackenzie A P 1971 J. Phys. Chem. 75 967 [5] Johari G P, Hallbrucker A and Mayer E 1987 Nature 330 552
3.4. The glass transition behavior of the interfacial water in confined HA aqueous solutions When the water fraction was increased in the ME aqueous solutions confined in the silica-gel pores, the Tg for the 5
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
[6] Velikov V, Borick S and Angell C A 2001 Science 294 2335 [7] Angell C A 2002 Chem. Rev. 102 2627 [8] Oguni M, Maruyama S, Wakabayashi K and Nagoe A 2007 Chem. Asian J. 2 514 [9] Capaccioli S, Ngai K L and Shinyashiki N 2007 J. Phys. Chem. B 111 8197 [10] Murthy S S N 1997 J. Phys. Chem. B 101 6043 [11] Cerveny S, Schwartz G A, Alegr´ıa A, Bergman R and Swenson J 2006 J. Chem. Phys. 124 194501 [12] Murthy S S N and Singh G 2008 Thermochim. Acta 469 116 [13] Nagoe A and Oguni M 2010 J. Phys.: Condens. Matter 22 325103 [14] Nagoe A, Kanke Y and Oguni M 2010 J. Phys.: Condens. Matter 22 365105 [15] Oguni M, Kanke Y, Nagoe A and Namba S 2011 J. Phys. Chem. B 115 14023 [16] Oguni M and Angell C A 1980 J. Chem. Phys. 73 1948 [17] Oguni M and Angell C A 1983 J. Chem. Phys. 78 7334
[18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28]
[29]
6
Mishima O and Stanley H E 1998 Nature 392 164 Mishima O and Stanley H E 1998 Nature 396 329 Angell C A 2007 J. Phys.: Condens. Matter 19 205112 Nagoe A and Oguni M 2013 J. Phys. Soc. Japan at press Nagoe A, Kanke Y, Oguni M and Namba S 2010 J. Phys. Chem. B 114 13940 Bertrand C E and Anisimov M A 2011 J. Phys. Chem. B 115 14099 Mikhail R Sh, Brunauer S and Bodor E E 1968 J. Colloid Interface 26 45 Suga H and Seki S 1980 Faraday Discuss. Chem. Soc. 69 221 Fujimori H and Oguni M 1993 J. Phys. Chem. Solids 54 271 Trachenko K and Brazhkin V V 2011 Phys. Rev. B 83 014201 Stanley H E, Kumar P, Franzese G, Xu L, Yan Z, Mazza M G, Buldyrev S V, Chen S-H and Mallamace F 2008 Eur. Phys. J.-Spec. Top. 161 1 Swenson J and Texeira J 2010 J. Chem. Phys. 132 014508
Search
Collections
Journals
About
Contact us
My IOPscience
Hydroxylamine-doping effect on the Tg of 160 K for water confined in silica-gel nanopores
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2013 J. Phys.: Condens. Matter 25 465110 (http://iopscience.iop.org/0953-8984/25/46/465110) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 128.6.218.72 This content was downloaded on 01/10/2015 at 14:26
Please note that terms and conditions apply.
IOP PUBLISHING
JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 25 (2013) 465110 (6pp)
doi:10.1088/0953-8984/25/46/465110
Hydroxylamine-doping effect on the Tg of 160 K for water confined in silica-gel nanopores A Nagoe and M Oguni Department of Chemistry, Graduate School of Science and Engineering, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8551, Japan E-mail: [email protected] (A Nagoe)
Received 30 August 2013, in final form 22 September 2013 Published 28 October 2013 Online at stacks.iop.org/JPhysCM/25/465110 Abstract The glass transition behavior of hydroxylamine (HA) aqueous solutions in bulk and confined in silica-gel nanopores with average width of 1.1 nm was studied by means of differential scanning calorimetry measurements and adiabatic calorimetry. The glass transition temperature (Tg ) of the confined solution with high HA mole-fraction (xHA ) was essentially the same as the value of the bulk. This suggests that the nano-size confinement affects the Tg of HA aqueous solution little. Meanwhile, the bulk solution with xHA < 0.3 revealed partial crystallization on cooling and, on the other hand, the confined solution with the same xHA did not crystallize. The Tg of the xHA = 0.076 confined solution was 174 K which is higher than the value of 160 K for pure water confined in the same silica-gel pores. This demonstrates that HA doping leads to no abrupt Tg -decrease, unlike doping of all the other second components reported so far, suggesting that HA is set neatly in a hydrogen-bond network formed by water molecules. We discuss the xHA dependence of Tg for the HA aqueous solutions from a viewpoint related to peculiar phase-behavior of pure water. Considering that the xHA = 0.076 aqueous solution revealed no anomaly compared with pure water, it was recognized as corresponding to the high-temperature phase of pure water.
1. Introduction
scanning calorimetry (DSC) heating curves [6]. Another method for producing glassy water is to dope a second component into water to suppress the formation of a crystal nucleus of ice; namely, highly concentrated aqueous solutions are used as samples. By extrapolating the Tg values of these glass-forming solutions with respect to the concentration, the Tg of pure water was estimated to be around 136 K [4, 7, 10, 12]. However, there remains a fatal issue: even slightly doping the second component removes water’s peculiar features of many properties such as a heat-capacity rise on cooling in the supercooled state [16, 17]. In other words, the value of 136 K corresponds to the Tg of water that shows no peculiar features and differs significantly from an aggregate of water molecules only. Angell suggested that for aqueous solutions an abrupt rise in Tg on approaching pure water might occur in a dilute range because the hydrogen-bond network which pure water makes develops only in the situation where water molecules
Water is the most abundant substance on the Earth. It is further well known by many peculiar behaviors revealed in the liquid state [1–3]. The peculiarity becomes remarkable at low temperatures in the supercooled state. No detail of the real behaviors has been, however, clarified because its crystallization to ice occurs [1–5]. The glass transition temperature (Tg ) is one such property. The Tg of supercooled water has been debated for a long time, mainly concerning whether it is 136 K or 165 K [3–15]. Amorphous solid water produced by vapor deposition reveals a thermal anomaly, at 136 K, resembling a glass transition [6]. However, some research groups have pointed out that the enthalpy-relaxation behavior differs from those observed in other glassy materials and insisted that a real glass transition occurs at around 165 K above the crystallization temperature of around 150 K on differential 0953-8984/13/465110+06$33.00
1
c 2013 IOP Publishing Ltd Printed in the UK & the USA
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
to 210 K and exhibits no continuous increase with increasing pore size. The appearance of this high Tg value suggests the existence of another structure with a more developed hydrogen-bond network. Therefore, the Tg value of 160 K was interpreted as being detected due to the pore size of 1.1 nm which is too small for water to form a fully developed hydrogen-bond network. However, there is no other work reporting this high Tg value of 210 K. Recently, we found that hydroxylamine (abbreviated as HA) aqueous solution confined in the silica MCM-41 pores shows a liquid–liquid first-order phase transition [21], which is connected to the phenomenon observed as a peculiar Cp -maximum for nanopore water [22, 23]. The HA molecule is considered to fit well to the hydrogen-bond network formed by water molecules. Therefore, doping HA into water is expected to yield a quite different mole-fraction (xHA ) dependence of Tg from other aqueous solutions such as methanol solutions. In this work, we studied the HA doping effect on the Tg of 160 K for water confined within silica-gel nanopores with average width of 1.1 nm. The Tg values of the nano-confined HA aqueous solutions were also compared with those obtained for bulk solutions with the same xHA values.
Figure 1. The mole-fraction dependence of Tg in various aqueous systems: open triangles, bulk hydrogen peroxide solution [17]; open squares, bulk ethylene glycol solution [10]; open circles, bulk methanol (ME) solution [10]; filled circles, ME solution confined in silica-gel pores with average width of 1.1 nm [14]. The dotted lines are drawn as guides for eyes.
2. Experimental details possess the majority fraction [7]. The development of the network causes the viscosity to increase and the Tg to shift up in aqueous solutions [7]. The problem is again that crystallization prevents one from discerning the Tg -rise behavior in the dilute concentration range close to pure water. We confined water into silica-gel nanopores with an average width of 1.1 nm to keep it from crystallizing and observed its glass transition at 160 K [8]. This observation supports Angell et al’s suggestion [6, 7]. Further, as methanol (abbreviated as ME) is doped into water little by little, the Tg of the confined aqueous solution decreases quickly from 160 K, bends rather at an ME mole-fraction xME = 0.3, and accords with that of glass-forming concentrated solutions in bulk above xME = 0.3 (figure 1) [14]. Extrapolation of the xME dependence of Tg above xME = 0.3 toward xME = 0 gives Tg = 136 K for pure water. This means that the development of a hydrogen-bond network gradually proceeds with increasing water fraction below xME = 0.3 and suggests that the environment surrounding the water molecule is quite different between the highly concentrated solutions (giving an extrapolated value of Tg = 136 K for pure water) and real pure water with a hydrogen-bond network (Tg = 160 K). Some research groups have tried to understand this difference as being caused by a liquid–liquid phase transition of supercooled water [18, 19]; namely, the values Tg = 160 K and 136 K (or lower than 136 K) correspond to those of low-density and high-density water respectively [7, 20]. Furthermore, we detected a glass transition at 210 K [15] for the water confined within silica MCM-41 pores with average diameters of 2.0 and 2.1 nm, which are larger than the width (1.1 nm) of the silica-gel nanopores stated above. It is intriguing that the Tg of water jumps discretely from 165
2.1. Sample preparation Aqueous solution with xHA = 0.35 was purchased from Wako Pure Chemical Industries, Ltd, and solution samples with different xHA values were prepared by mixing it with distilled water at specified ratios. The silica-gel CARiACT Q3 was supplied by Fuji Silysia Chemical Ltd, and had a particle size of a few mm and pore voids with average width of 1.1 nm as assessed from the results of nitrogen-gas absorption measurements by the MP method [8, 24]. The confinement into the pores was executed by directly pouring each degassed HA aqueous solution onto silica-gel particles under vacuum after the insides of the silica-gel pores had been exhausted at 473 K for 24 h. The silica-gel particles loaded with the solution were held, under an atmosphere of helium gas, within a quartz-glass container to prevent from reacting with the inner wall of a gold-plated copper cell during calorimetric measurements. 2.2. Method for determination of Tg The glass transition behaviors of HA aqueous solutions in bulk were examined by DSC measurements. The measurements were performed with a Perkin–Elmer DSC 8500 at a heating rate of 10 K min−1 after precooling the sample rapidly at −50 K min−1 . The Tg value was determined as the middle point of the heat-flow jump that accompanied the transition. The Tg s of the solutions confined in silica-gel pores were determined, with an adiabatic calorimeter, by following the enthalpy-relaxation rates as a function of temperature in the heating direction [25, 26]. In general, glasses are located 2
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
Figure 3. Mole-fraction dependence of Tg in HA (open symbols) and ME (filled symbols) aqueous solutions: triangles, bulk HA solution; circles, internal part of the confined aqueous solution; squares, interfacial part of the confined solution; all data for ME solutions were taken from [14].
Figure 2. DSC cooling curves for bulk hydroxylamine (HA) aqueous solutions at a rate of −10 K min−1 : dashed line, xHA = 0.30; solid line, xHA = 0.35. The inset represents the corresponding heating curves at a rate of 10 K min−1 after the samples were quenched at −50 K min−1 .
remaining liquid was left unclear in the dilute xHA range. Above xHA > 0.3, we detected no crystallization event on cooling. Figure 2 shows the DSC curves on cooling, at −10 K min−1 , the HA aqueous solutions in bulk with xHA = 0.30 and 0.35. In addition, the solutions showed no anomaly indicating the occurrence of a phase transition and vitrified at around 180 K. Considering that the high- to low-density phase transition is of first order [21], this fact potentially means that the HA aqueous solutions became frozen into nonequilibrium glassy states as was the case in the high-temperature phase corresponding to the high-density phase hypothesized for pure water’s phase diagram [18, 19]. The inset of figure 2 displays the DSC curves on heating the frozen, glassy solutions at 10 K min−1 . The Tg s of the bulk solutions were determined by using the heating curves and are plotted with open circles in figure 3. The Tg at xHA = 0.35 is higher by 35 K than that of the hydrogen peroxide aqueous solution with the same mole-fraction (xH2 O2 ) [16, 17]. Moreover, the Tg value derived by extrapolating, with a straight line, those of the bulk HA aqueous solutions toward xHA = 0 is around 165 K, which is higher than the extrapolated value of 136 K by using other solutions [4, 7, 10, 12]. A similar mole-fraction dependence giving Tg = 160 K at x = 0 has also been reported in aqueous solutions of propylene-glycol oligomer [11]. However, the validity of the extrapolation also remains questionable because the crystallization of ice occurred in the dilute x range [7]. Murthy reported a result that raised doubts about the extrapolation for the propylene-glycol oligomer solutions [12]. In our previous studies, doping methanol or ethylene glycol led to depression of Tg for confined water only in a dilute x range, as is also shown in figure 1, while the bulk solutions with the same xs easily crystallized [4, 10, 13, 14].
in different enthalpic states according to whether they were precooled at high or low speeds; namely, the temperature at which the molecular configuration of the solution is frozen in and brought to a nonequilibrium state depends on the precooling speed. On approaching Tg in heating measurements, the glasses relax toward their respective equilibrium states with release of heat for high-enthalpic glasses precooled at a high speed or absorption of heat for low-enthalpic glasses precooled at a low speed. In this work, the Tg s were determined as the temperatures giving the maximum heat-absorption effect, namely, minimum −dH/dt, for the samples precooled at a lower speed. This method enables us to detect the dynamic properties of the samples directly. Therefore, it might be efficient even if the jump in heat capacity at Tg is small or the heat-capacity behavior is quite complex as compared with those of normal glassy pure liquids. In addition, the high sensitivity for heat release/absorption and high resolution for thermometry of the adiabatic calorimetry help us to detect glass transition events precisely. Although thermodynamic properties such as bulk modulus and thermal expansion have been used besides calorimetry to determine the Tg [27], the presence of silica makes the determination rather difficult.
3. Results and discussion 3.1. The DSC curves of bulk HA aqueous solutions For the bulk solutions in the xHA range below 0.3, some of the water molecules crystallized in the cooling process. The xHA , and therefore the behavior, of the aqueous solution 3
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
is very small even in very small-in-size pores. This fact also confirms the effectiveness of the confinement method for exploring the glass transitions of these aqueous solutions. Two possible reasons are considered for this fact. One is that the size of the cooperative rearrangement region to determine the glass transition temperature is smaller than the width of the silica-gel nanopores. The other is that the rearrangement of molecules in the aqueous solutions involves essentially no cooperative nature in the unit to be activated. In view of the fact that the ME aqueous solutions with xME < 0.3 form a hydrogen-bond network inherent to water, the hydrogen-bond network formed at xHA = 0.076 is judged as the one inherent to water. Therefore, the higher-in-temperature Tg s would be linked as a function of xHA smoothly to the value of 160 K for pure water, as drawn with a solid line in figure 3. The xHA dependence of Tg in the very dilute range is completely opposite to the xME dependence; the Tg of water is increased and decreased by doping HA and ME respectively. This indicates that the presence of HA enhances while that of ME inhibits the formation of the water’s hydrogen-bond network. The lower-in-temperature glass transition due to the interfacial molecules was observed at 136 K for the solution at xHA = 0.35, although the heat-release and -absorption effects are small. In the xHA = 0.076 solution, the transition was observed clearly at 126 K. It is noticed that the Tg for the interfacial molecules shifts to higher temperatures with doping HA molecules while it shifts to lower temperatures with doping ME molecules. The Tg of pure HA is much higher than that of ME because of the capacity to form hydrogen bonds. Therefore, the interfacial water molecules close to HA molecules must be bound rather tightly and their Tg is expected to increase.
Figure 4. Enthalpy-relaxation rates as a function of temperature (−dH/dt) in the heating direction observed in the HA aqueous solutions within pores with average width of 1.1 nm: filled circles, sample cooled slowly over the temperature region; open circles, sample cooled rapidly over the region. Hysteresis between the open and the filled circles represents the existence of glass transitions. The data for xHA = 0 were taken from [8].
3.3. The xHA dependence of Tg in the dilute HA aqueous solutions and its relation to water’s anomaly
3.2. The glass transition behavior, observed by adiabatic calorimetry, of the aqueous solutions confined in silica-gel pores
As stated above, the glass transition of the internal solution within the pores can be recognized as exhibiting the same phenomenon as that of the solution in bulk. ME or ethylene glycol aqueous solutions showed a steep decrease in the Tg with doping the respective second components in the dilute x range [13, 14]. On the other hand, HA aqueous solution revealed rather an increase in the Tg . This means that HA doping maintains and even strengthens the hydrogen-bond network formed by the water molecules. This attribute of HA is considered to originate from the molecular structure of HA. HA has the same 1:1 ratio of hydrogen atoms to lone electron-pairs at 3–3 in number as the water molecule with 2–2; namely, HA can be recognized just as a water-molecular dimer. Therefore, HA can form a hydrogen-bond network fully in the same way as water. In reality, this is also supported by the fact that the Tg of the xHA = 0.35 solution is higher by 35 K than those of hydrogen peroxide or hydrazine aqueous solutions with x = 0.35 [16, 17]. Some research groups have predicted that pure water’s Tg is 160 K [6, 7, 11, 18]. Angell et al insisted that the high Tg value of 160 K for pure water originates from the
Figure 4 shows the results of enthalpy-relaxation rates for xHA = 0.076 and 0.35 aqueous solutions, as well as for pure water, confined in silica-gel pores with a width of 1.1 nm. The presence of heat-release and -absorption effects for the samples precooled rapidly and slowly respectively evidences the existence of a glass transition. The pure water revealed two glass transitions at 115 and 160 K, as shown in figure 4(a) [8]. The lower-in-temperature transition was assigned as due to the water molecules located on the pore walls, namely, the interfacial water molecules. The higher-in-temperature transition was assigned as due to the water molecules located in the central parts of the pore voids, namely, the internal water molecules. The Tg values derived are plotted in figure 3 with open circles and squares. The Tg at xHA = 0.35 is very close to the value for the bulk solution with the same xHA . The closeness is just the behavior found for ME aqueous solutions [14]. This means that the effect of nano-confinement on the Tg s of aqueous solutions of such simple hydrophilic compounds 4
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
internal water increased in the range xME < 0.3 and simultaneously an interfacial solution layer was formed to reveal a new glass transition around 115 K in the same range [14]. This was understood as the internal solution with the high water fraction making a hydrogen-bond network resembling the structure of the high-temperature phase of pure water and as the interfacial solution buffering the network structure from the pore wall mainly consisting of a rather hydrophobic siloxane-bond network. In view of the fact that the pore wall was formed by the same siloxane-bond network, the role of the buffer of interfacial water molecules is kept in the HA aqueous solution. It is noticed here that the Tg of the interfacial water in the HA aqueous solution revealed an obvious increase with increase of the fraction of dopant HA and that the glass transition due to the interfacial water was observed only with a very small effect in the xHA = 0.35 solution. This means that the interfacial water molecules were bound gradually more strongly by being surrounded by more HA molecules as well as the number of interfacial water molecules accordingly decreasing. It is not clear at present how the role of the buffer appears in the interfacial HA molecules. It would be reasonable to consider that the heat-release or -absorption effects of the glass transition are hidden in the wide higher-in-temperature range of over 140–190 K.
peculiar increase of Cp of water on decreasing the temperature below 0 ◦ C [7, 18]. The Cp -increase behavior below 0 ◦ C was connected with a liquid–liquid phase transition by many research groups [18, 19, 23, 28]. In this view, the phase exhibiting Tg = 160 K corresponds to the low-temperature low-density phase of water [7, 18]. A decrease in water’s Tg of 160 K and a reduction of the Cp singularity were observed in a parallel way for the aqueous solutions of ordinary second components [6, 7]. However, the xME dependence of Tg for ME aqueous solutions changes gradually for xME s below 0.3 [14], although DSC measurements of hydrogen peroxide aqueous solutions disclose remarkable reduction of the anomalous Cp component for supercooled water by doping hydrogen peroxide in a small amount of xH2 O2 = 0.1 [16, 17]. The two phenomena that the Cp anomaly disappears and the x dependence of Tg bends with increasing x occur at obviously different xs. Therefore, it is difficult to conclude that the two phenomena result from the same origin. In fact, the xHA dependence of Tg for HA aqueous solutions indicates that they maintain their hydrogen-bond networks even in the highly concentrated range. As reported elsewhere [21], HA aqueous solutions reveal a liquid–liquid phase transition the thermal behavior of which is linked smoothly with the heat-capacity maximum at 230 K for pure water [22]. However, the transition was found to be of first order, and the HA aqueous solutions could be easily supercooled and displayed no singularity, due to the phase transition, on the DSC curve as stated above. The glass transition observed presently in the HA aqueous solutions confined within the silica-gel pores is thus recognized as taking place in the high-temperature phase. It is concluded that the Tg of 160 K for pure water is attributed to the high-temperature phase. In other words, the Tg = 136 K, derived by extrapolating the composition dependence of the Tg s for the concentrated solutions without forming a hydrogen-bond network, is irrelevant to the Tg of pure water in the high-temperature phase. According to our consideration, the dynamic properties of water molecules in concentrated solutions with ordinary second components differ greatly from those of pure water in the high-temperature phase. If a low-temperature phase of pure water was formed through the liquid–liquid transition, a Tg jump would potentially occur [15, 18, 29]. This is expected in the case where the water’s Tg is much higher than 160 K. We actually reported that the pure water confined within mesoporous silica MCM-41 with an average diameter of 2.1 nm shows a glass transition at 210 K, while for 1.7 nm there is a transition at 165 K and for 2.0 nm there are two transitions at 165 and 210 K; namely, the Tg increased discretely with increase of the pore size [15]. It is reasonable to interpret that this discrete increase corresponds to the change of the water structure from the high- to the low-temperature phase. Further study is, of course, needed to confirm this interpretation.
4. Conclusions We successfully observed glass transitions in HA aqueous solutions in bulk at high xHA s by DSC and of dilute HA aqueous solutions confined in silica-gel pores with a width of 1.1 nm by adiabatic calorimetry. The Tg of 160 K for the confined pure water showed no steep decrease but rather an increase with increase in xHA , quite distinctly from other second components such as ME. The decrease of Tg is understood as reflecting collapse of the hydrogen-bond network. Therefore, it is concluded that the HA aqueous solutions maintain the hydrogen-bond network of pure water even in the high xHA range. Besides the fact that the HA aqueous solutions show a liquid–liquid phase transition in the state confined within silica MCM-41 pores [21], this peculiar doping effect on the Tg of water is very fascinating. It is suggested that the value of Tg (=160 K) for the confined water corresponds to that of pure water in the high-temperature phase. This leads us to predict that water displaying the glass transition at a temperature higher than 160 K is in the low-temperature phase with a well-developed hydrogen-bond network. This prediction is consistent with the assertion, reported previously [15], that the Tg of water is 210 K.
References [1] Rasmussen D H and MacKenzie A P 1973 J. Chem. Phys. 59 5003 [2] Kanno H and Angell C A 1979 J. Chem. Phys. 70 4008 [3] Debenedetti P G 2003 J. Phys.: Condens. Matter 15 R1669 [4] Rasmussen D H and Mackenzie A P 1971 J. Phys. Chem. 75 967 [5] Johari G P, Hallbrucker A and Mayer E 1987 Nature 330 552
3.4. The glass transition behavior of the interfacial water in confined HA aqueous solutions When the water fraction was increased in the ME aqueous solutions confined in the silica-gel pores, the Tg for the 5
J. Phys.: Condens. Matter 25 (2013) 465110
A Nagoe and M Oguni
[6] Velikov V, Borick S and Angell C A 2001 Science 294 2335 [7] Angell C A 2002 Chem. Rev. 102 2627 [8] Oguni M, Maruyama S, Wakabayashi K and Nagoe A 2007 Chem. Asian J. 2 514 [9] Capaccioli S, Ngai K L and Shinyashiki N 2007 J. Phys. Chem. B 111 8197 [10] Murthy S S N 1997 J. Phys. Chem. B 101 6043 [11] Cerveny S, Schwartz G A, Alegr´ıa A, Bergman R and Swenson J 2006 J. Chem. Phys. 124 194501 [12] Murthy S S N and Singh G 2008 Thermochim. Acta 469 116 [13] Nagoe A and Oguni M 2010 J. Phys.: Condens. Matter 22 325103 [14] Nagoe A, Kanke Y and Oguni M 2010 J. Phys.: Condens. Matter 22 365105 [15] Oguni M, Kanke Y, Nagoe A and Namba S 2011 J. Phys. Chem. B 115 14023 [16] Oguni M and Angell C A 1980 J. Chem. Phys. 73 1948 [17] Oguni M and Angell C A 1983 J. Chem. Phys. 78 7334
[18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28]
[29]
6
Mishima O and Stanley H E 1998 Nature 392 164 Mishima O and Stanley H E 1998 Nature 396 329 Angell C A 2007 J. Phys.: Condens. Matter 19 205112 Nagoe A and Oguni M 2013 J. Phys. Soc. Japan at press Nagoe A, Kanke Y, Oguni M and Namba S 2010 J. Phys. Chem. B 114 13940 Bertrand C E and Anisimov M A 2011 J. Phys. Chem. B 115 14099 Mikhail R Sh, Brunauer S and Bodor E E 1968 J. Colloid Interface 26 45 Suga H and Seki S 1980 Faraday Discuss. Chem. Soc. 69 221 Fujimori H and Oguni M 1993 J. Phys. Chem. Solids 54 271 Trachenko K and Brazhkin V V 2011 Phys. Rev. B 83 014201 Stanley H E, Kumar P, Franzese G, Xu L, Yan Z, Mazza M G, Buldyrev S V, Chen S-H and Mallamace F 2008 Eur. Phys. J.-Spec. Top. 161 1 Swenson J and Texeira J 2010 J. Chem. Phys. 132 014508
