Nanoconfined Solid–Solid Transitions: Attempt To Separate the Size

Apr 16, 2015 - Department of Chemistry, University of Alabama at Birmingham, 901 S. 14th Street, Birmingham, Alabama 35294, United States. J. Phys. Ch...
0 downloads 0 Views 707KB Size
Article pubs.acs.org/JPCC

Nanoconfined Solid−Solid Transitions: Attempt To Separate the Size and Surface Effects Reza Farasat and Sergey Vyazovkin* Department of Chemistry, University of Alabama at Birmingham, 901 S. 14th Street, Birmingham, Alabama 35294, United States ABSTRACT: The solid−solid phase transitions in ammonium nitrate, ammonium perchlorate, and sodium nitrite confined to native and organically modified silica nanopores are analyzed by differential scanning calorimetry. The study reveals the effect of nanoconfinement on the transition temperature and links it to the kinetic parameters of the process. It is suggested that the effect of nanoconfinement is primarily the surface interaction effect in the native pores and the size effect in organically modified pores. It has been found that in organically modified pores the transition temperature is always depressed relative to the bulk, whereas in the native pores its behavior is generally unpredictable. Kinetic analysis of the transitions in terms of a nucleation model indicates that the experimentally observed shifts in temperature can be explained by a combination of changes in the free energy of nucleation and preexponential factor of the respective processes.

1. INTRODUCTION The effect of nanoconfinement on melting, crystallization, and glass transition has been studied widely. For example, two highly cited reviews1,2 published on this topic about a decade ago already contained several hundred references. Studies of nanoconfined solid−solid transitions are not nearly as plentiful despite the fact that the transitions affect a host of practically important properties, including magnetic, electrical, mechanical, and optical properties as well as chemical and biological activity. The earliest studies were on the solid−solid transitions in frozen oxygen3 and frozen cyclohexane4 in nanoporous silica glasses. Both substances are fluids at room temperature and enter the pores spontaneously. In order to load the nanopores with room temperature solids, the room temperature solids have to be either melted or dissolved. Loading from the melt has been utilized to study the solid−solid transitions in sodium nitrite,5−9 sodium nitrate,9 and potassium dihydrogen phosphate.10 Loading from a solution has been used more rarely, in particular in studies of solid−solid transition in sodium nitrate11 and ammonium chloride.12 All these studies have been conducted on substances loaded in native (i.e., not organically modified) silica pores. Silica has electrostatically active surface that includes SiO dipoles and surface silanol groups that can participate in various interactions with the guest molecules and ions.13 For example, molecular solids may interact with the surface via relatively weak van der Waals and dipole−dipole forces. On the other hand, ionic solids (salts) may be involved in much stronger interactions that may even result in the formation of ionic compounds (i.e., Si−O−NH4+)13 and coordination complexes with the Si−O−Si oxygen atoms.14 Strong interactions with the surface would tend to stabilize the guest solid decreasing the mobility of its constituents which would create an extra energy barrier to the phase transition and thus shifting it to higher temperature relative to the bulk. On the other hand, decreasing the size of nanocrystals increases the © 2015 American Chemical Society

contribution of the surface free energy relative to the free energy of the bulk phase that causes depression of the phase transition temperature. In other words, when a solid is confined to nanopores, the effect of nanoconfinement generally is the sum of the size and surface effects. If the size effect causes a depression of the phase transition temperature, the surface effect can diminish or even outweigh it, causing an increase in the phase transition temperature. If we consider the melting transition, there are plenty of examples1,2 of melting temperature depression due to reducing crystals to nanodimensions. Examples of an increase in the melting temperature are less common and include indium15 and lead16 in aluminum matrix and benzene17 and carbon tetrachloride18 in graphitic nanopores. As shown by simulations,19 an increase in the melting point should occur when the strength of the intermolecular interactions of confined molecules is smaller than the strength of interactions between the molecules and confining surface. If we review the available literature on the solid−solid transitions in the native silica-based pores, we will find conflicting reports of the effect of nanoconfinement on the phase transition temperature. The effect appears to be simple for molecular solids, i.e., compounds that should have relatively weak interaction with the silica surface. In solid oxygen, the temperature has been reported3 to drop relative to the bulk value (−229 °C) by respectively 2, 5, and 10 °C in 20, 5, and 2 nm pores. For solid cyclohexane confined to 30, 15, 7.5, and 4 nm pores the transition temperature has been found4 to decrease respectively by 2, 3, 8, and 10 °C relative to the bulk value −87 °C. It thus appears that the effect of nanoconfinement on molecular solids is primarily the size effect, i.e., Received: February 20, 2015 Revised: April 6, 2015 Published: April 16, 2015 9627

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C depression of the transition temperature, which makes sense because the surface effect is expected to be weak. For ionic solids the effect of nanoconfinement is not straightforward at all. In NaNO2 confined to 30−80 nm pores of synthetic opal the transition occurred at 1 °C lower on heating relative to the bulk transition temperature of ∼164− 165 °C.5 Another report8 indicates that this salt loaded in 7 nm pores demonstrates the transition at a temperature that is 22 °C lower than that for the bulk. NaNO2 loaded in 23 nm pores is also reported9 to have the transition temperature about 3−4 °C below the bulk value, whereas NaNO3 loaded in the same pores demonstrates the solid−solid transition only 1 °C below the bulk value (275 °C). Some other publications on nanoconfined transition in NaNO2 indicate that the temperature shift is either very small (7 nm pores)6 or entirely absent20 (5−20 nm pores or particles). In addition, there are at least two papers that report that nanoconfinement causes an increase in the temperature of the solid−solid phase transition. An increase over 7 °C has been detected6 for the transition in KH2PO4 loaded into 7−100 nm pores. NH4Cl confined to partially filled nanopores has demonstrated the phase transition at ∼4 °C higher than that in the bulk.12 The absence of a straightforward effect of nanoconfinement on the temperature of the solid−solid transition in ionic solids can be readily rationalized by an overlap of the size and surface effects whose combined outcome can be a decrease as well as an increase in the transition temperature. Following this hypothesis, it should be possible to probe separately the size and surface effect by changing the strength of interaction between an ionic solid and confining surface. Such change can be accomplished through organic modification of the silica surface, i.e., by replacing the silanol hydrogen with a hydrocarbon. Of course, organic modification of the surface would not eliminate entirely its interaction with the guest ions but it would drastically diminish the strength of the interaction. To be more precise, the energy of the interaction can be expected21 to drop by as much as 2 orders of magnitude from a few hundred kJ mol−1, which would be characteristic of the ion−ion interactions between the guest solid and the native silica surface, to a few kJ mol−1 that would be expected of the van der Waals interactions between the guest solid and the organically modified silica surface. Therefore, by comparing the solid−solid phase transitions taking place in organically modified and native pores, we should be able to see predominantly the size effect in the former case and combined size and surface effects in the latter one. In this paper, we perform such comparison for several ionic solids with the purpose of separating the size and surface effects on the solid−solid phase transitions. We also analyze the kinetics of the phase transition in terms of a nucleation model in order to understand the origins of the observed effects. To better highlight the surface effect, our study is conducted in partially filled pores that are naturally accomplished by filling a porous material from a solution.12,22 It should be noted that in fully filled pores only a small fraction of nanocrystals is in direct contact with the surface (Figure 1A). In this situation, a relative contribution of the surface effect is diminished as compared to the case of partially filled pores (Figure 1B), in which a much larger fraction of nanocrystals is affected by direct interaction with the surface. As experimental systems, we have selected the II → I transitions in ammonium nitrate and ammonium perchlorate and the III → II transition in sodium nitrite to be studied in three forms: bulk and loaded in native and

Figure 1. Schematic presentation of the fully (A) and partially (B) filled pores.

organically modified silica gel pores. To the best of our knowledge this is the first study of solid−solid transitions confined to organically modified nanopores. Basic thermodynamic parameters such as the enthalpy and temperature of the transitions have been reported in the literature.23−33 The choice of the experimental systems has been determined by two requirements. The first is that the transition should be accompanied by a reasonable thermal effect so that one can follow it by differential scanning calorimetry (DSC). The second is that the solid should be reasonably soluble in an organic solvent. This is because organically modified silica is hydrophobic and thus can be loaded only from a solution based on either entirely organic solvent or a mixture of an organic solvent with a small fraction of water. Both ammonium nitrate and ammonium perchlorate are sufficiently soluble in methanol. Sodium nitrite is not, but its solubility improves greatly on addition of water to methanol.

2. EXPERIMENTAL SECTION Ammonium nitrate and sodium nitrite were purchased from Fisher, and ammonium perchlorate was from Alfa Aesar. The salts were recrystallized before the use. The ammonium nitrate and ammonium perchlorate were recrystallized from methanolic solution. Then, the recrystallized salts were dissolved in methanol to make saturated solutions that were used for loading the salts into native and organically modified silica. In the case of the sodium nitrite, the salt was recrystallized from a mixture of water and methanol 1:2 by volume. The ratio was determined by adding methanol to water until the point when organically modified silica gel started to absorb the mixed liquid. The recrystallized sodium nitrite was then dissolved in the water−methanol mixture to make a saturated solution that was then employed for loading into native and organically modified silica. Native silica sample was SiliaFlash material generously gifted by SiliCycle Inc. (Quebec City, Canada). It is thin powder (average particle diameter of ∼50 μm) having nominal pore diameters of 15 nm, a pore volume of 1.19 cm3 g−1, and a surface area of 285 m2 g−1. Organically modified silica was prepared by reacting the native silica sample with hexamethyldisilazane as reported by Anwander et al.34 The reaction converts the surface silanol groups into trimethylsiloxane. This is quite thermally stable material similar to that used as the stationary phase in gas chromatographic columns capable of withstanding 350 °C.35 Our previous analysis of this organically modified silica indicated that the material did not degrade below 400−450 °C.36 9628

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C

typically denoted as k.41 It is clear from eq 1 that an increase in w0 as well as a decrease in ΔG* would cause the nucleation process to accelerate. When conducted under nonisothermal conditions, at any given temperature a faster process would progress to a greater extent than a slower one. This also means that the same extent would be reached by a faster process at lower temperature than by the slower one. In other words, an increase in w0 or a decrease in ΔG* would make the process shift to lower temperature as long as it has a positive temperature dependence, which is the case39 of the solid− solid transitions taking place on heating. Detailed derivations and expression for the free energy barrier can be found elsewhere.40,42 For the purpose of our analysis, it is sufficient to recognize that ΔG* can be estimated as follows:

The salts were loaded by immersing silica samples for 2−3 h into the saturated solutions of the salts in methanol (ammonium nitrate and ammonium perchlorate) or methanol and water (sodium nitrite). Afterward, the silica samples were separated by filtering the solutions. Excess of the solution was removed by rubbing the wet silica samples between the sheets of filter paper. The loaded samples were then dried in an oven at 90 °C for 2 h. Since ammonium nitrate and ammonium perchlorate decompose entirely into gas products37,38 when heated, the mass loads of both salts were evaluated by thermogravimetric analysis (TGA) as described earlier.22 The load of sodium nitrite was determined by extracting the guest solid from silica with a solvent and looking at the difference in the mass of silica before and after the extraction. The pore fullness22 did not show significant difference between the native and modified pores as well as between the salts and was estimated on average to be 15 ± 5%. All calorimetric measurements were taken with a heat flux DSC (Mettler-Toledo, model 823e). Indium and zinc standards were used to perform temperature, heat flow, and tau lag calibrations. The experiments were conducted under the nitrogen flow (80 mL min−1) at the heating rates 4.5, 6.75, 10.1, 14, and 20 °C min−1. The sample masses were around 5 mg for the bulk salts and 20 mg for the nanoconfined ones. The samples were run in closed 100 μL aluminum pans. The temperatures of the solid−solid transitions were estimated from the DSC peak temperatures, Tp, as the mean of three repetitive measurements. The Tp value was preferred over the onset temperature that is frequently used as an estimate for the equilibrium temperatures of the melting transitions. There were two reasons for that. First, evaluation of the Tp values involves far less uncertainty especially in the case of nanoconfined systems that are typically represented by broad and shallow DSC peaks. Second, unlike melting, the solid−solid transitions involve significant superheating39 so that the respective onset temperatures do not carry any special thermodynamic meaning. The average uncertainties for the Tp values were ±0.5 and ±1.3 °C for the bulk and nanoconfined systems, respectively. As a control, DSC runs were performed on mechanical mixtures of native or organically modified silica with the salts that were mixed in the same ratio as was determined for the respective solution loaded samples. DSC peaks for the mechanically mixed samples were practically identical regardless of whether the salts were mixed with native and organically modified silica. In addition, FTIR spectra of the nanoconfined samples of sodium nitrite have been taken by using a Bruker ALPHA FTIR spectrometer with a platinum ATR module.

ΔG* =

(2)

In eq 2, A is a constant and ΔT is superheating defined as T − T0, where T0 is the equilibrium temperature of the transition and T is the current temperature. By substituting eq 2 into eq 1 and taking the logarithmic derivative of the nucleation constant, one can derive the so-called effective activation energy as ⎛ d ln w ⎞ ⎟ E = −R ⎜ ⎝ dT −1 ⎠

(3)

The resulting activation energy is temperature dependent and has the following form:39,43 ⎛ 1 2T ⎞ E = A⎜ − ⎟ 2 (ΔT )3 ⎠ ⎝ (ΔT )

(4)

Since we study the kinetics of the transitions taking place on heating, ΔT in eq 4 is positive and so is the A value. Moreover, the value in the parentheses is positive and decreasing with increasing ΔT so that the effective activation energy decreases with increasing temperature. While eq 4 sets a theoretical dependence of E on T, an experimental dependence can be evaluated by the Kissinger method44 as was demonstrated earlier.45−47 The key equation of the Kissinger method is

E = −R

d ln(β /Tp2) dTp−1

(5)

where β is the heating rate and R is the gas constant. The experimental E vs T dependence is then derived by taking numerical derivative of the Kissinger plot (ln(β/Tp2) versus Tp−1) and replacing the reciprocal temperature with temperature. Continuous and smooth numerical derivative is obtained by fitting the discrete points of the Kissinger plot by an appropriate interpolating function. Once the experimental E vs T dependence is obtained, it can then be fitted by the theoretical one (eq 4) in order to estimate the parameters A and T0. Knowledge of the parameters can further be used to evaluate the temperature dependence of ΔG* in accord with eq 2. The overall rate of a phase transition can be presented in the following form:47

3. THEORETICAL AND COMPUTATIONAL BACKGROUND The kinetics of the solid−solid transition is generally treated in terms of a nucleation model.40 In its simplest form, the nucleation rate constant can be represented by the following equation: ⎛ −ΔG* ⎞ ⎟ w = w0 exp⎜ ⎝ RT ⎠

A (ΔT )2

(1)

where w0 is the preexponential factor, R is the gas constant, ΔG* is the free energy barrier for the formation of a nucleus. Here, the nucleation rate constant is denoted by w to emphasize the difference from the reaction rate constant

⎛ −ΔG* ⎞ dα n ⎟(1 − α) = w0 exp⎜ ⎝ RT ⎠ dt 9629

(6) DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C

close to the literature values. For the II → I (tetragonal to cubic) transition in ammonium nitrate the following temperatures and enthalpies have been reported: 120−121 °C and ΔH = 51 J g−1,23 126 °C and 55 J g−1,24 and 130 °C.25 Figure 3 presents a dependence of the DSC peak temperature on the heating rate. Each of the values presented

where α is the extent of conversion from one phase to another and n is the reaction order or simply an exponent that characterizes the strength of the rate dependence on the conversion. The rate is determined directly from the heat flow, dQ/dt, measured by DSC:

dα 1 dQ = dt Q 0 dt

(7)

where Q0 is the total heat of transition. The conversion is determined by integration of eq 7. A simple rearrangement of eq 6 leads to a linear equation: ⎡⎛ d α ⎞ ⎛ ΔG* ⎞⎤ ⎟ = ln w + n ln(1 − α) ln⎢⎜ ⎟ exp⎜ 0 ⎝ ⎠ ⎝ RT ⎠⎥⎦ ⎣ dt

(8)

Now the preexponential factor, ln w0, can be estimated as the intercept of linear plot of the left-hand side of eq 8 against ln(1 − α), whereas the slope would yield the value of n. The fits of eqs 4 and 8 were accomplished by using the Origin graphing software.

4. RESULTS AND DISCUSSION Ammonium Nitrate. Typical DSC peaks of the tetragonal to cubic transition of ammonium nitrate are shown in Figure 2.

Figure 3. DSC peak temperatures for the tetragonal-to-cubic transition of ammonium nitrate in bulk (half-filled circles), organically modified pores (open circles), and native pores (solid circles).

in Figure 3 is the mean of three measurements. At any individual heating rate, the transition temperature for ammonium nitrate nanoconfined to native silica pores is almost 3 °C higher than that for the bulk. An increase in the transition temperature means that the conversion of the tetragonal crystalline structure of the ammonium nitrate to the cubic phase has become slower (i.e., faces some extra difficulties in native nanopores). As mentioned earlier, in native silica the size effect is likely to be accompanied by a strong surface effect. The ionic surface of silica can interact strongly with nanoconfined ionic compounds and thus hinder the rearrangements of the crystalline lattice. On the other hand, the surface effect is expected to be rather small for ionic compounds nanoconfined in organically modified silica so that the main effect should be the size effect. As seen in Figure 3, the transition temperature of ammonium nitrate nanoconfined in modified silica decreases 6−8 °C relative to that for the bulk. That is, the size effect is depression of the transition temperature. Note that no such shifts in temperature were observed in mechanical mixtures of ammonium nitrate with either native or organically modified silica. For both types of mixtures the Tp values were nearly identical and similar to that for the bulk. Therefore, the effects observed in the loaded silica samples are specific to nanoconfined systems. The obtained DSC data were subjected to kinetic analysis in an attempt to reveal the kinetic origins of the observed effects. Figure 4 displays the Kissinger plots (eq 5) for the three ammonium nitrate systems. All the plots are of similar nonlinear shape so that the tangent line angle decreases with increasing temperature. Because the tangent of the plot is the activation energy, one should expect the activation energy to have a decreasing dependence on temperature. Recall that this type of experimental dependence is exactly what was predicted by the nucleation model (eq 4). This obviously confirms the validity of the nucleation model and justifies its use for estimating the nucleation parameters. The E versus T dependence results needed for this purpose were obtained

Figure 2. DSC curves measured at heating rate 14 °C min−1 for bulk ammonium nitrate, and ammonium nitrate nanoconfined in native and organically modified pores (solid line, bulk; dashed line, native pores; dotted line, organically modified pores).

The peaks for nanoconfined samples are relatively asymmetric and have a smaller height to width ratio than those for the bulk sample, thus signaling the slower process of transition under the nanoconfinement conditions. Also, the process in native silica appears to reveal a lower temperature shoulder that is a sign of the diversity of the nanostructures formed inside the pores. Similar effects have also been observed (vide infra) for ammonium perchlorate in native silica and sodium nitrite in organically modified silica. Considering that the shoulders are found at lower temperature than the main peak, it is reasonable to assume that they are associated with the transition of the smallest crystals that are formed sufficiently far away from the surface to be affected by the surface interaction. The endothermic heat of transition for the bulk sample was found to be 47 J g−1. The transition temperatures defined as Tp were between 129 and 134 °C for the range of the heating rates used. Both the enthalpy and temperature of the transition are 9630

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C

To delve further into the kinetics of the nanoconfinement effects on solid−solid phase transitions, we calculated the free energy barrier of nucleation (ΔG*) for both types of the nanoconfined samples of ammonium nitrate and compared them with the value for the bulk. To determine ΔG* values by eq 2, we estimated the value of A by fitting eq 4 to the E vs T plots shown in Figure 5. The fit produced the A and T0 values that were plugged into eq 2. The estimated values of these two parameters are given in Table 1. The fits are characterized by statistically significant values of the correlation coefficient, r. It should be noted that the estimated values of T0 do not follow the same order as the respective Tp values (Figure 3). The issue arises from the fundamental difficulty of accurately estimating the equilibrium temperature of solid−solid transitions because the forward process involves significant superheating, whereas the reverse process involves significant supercooling.39 Therefore, the true value of T0 usually lies outside the temperature range, within which the process is detectable experimentally. In this situation, the T0 value is necessarily evaluated by extrapolation and, thus, may not be quite accurate. Figure 6 shows the free energy barrier of nucleation as a function of superheating above the equilibrium temperature for

Figure 4. Kissinger plots for the tetragonal-to-cubic transition of ammonium nitrate in bulk (half-filled circles), organically modified pores (open circles), and native pores (solid circles).

from the Kissinger plots as explained earlier in the section dealing with theory and computations. Figure 5 demonstrates the effective activation energy for the three studied systems of ammonium nitrate as a function of

Figure 6. Temperature dependence of the free-energy barrier to the tetragonal-to-cubic transition in ammonium nitrate in bulk (solid line), native pores (dashed line), and organically modified pores (dotted line).

Figure 5. Temperature dependence of the activation energy for the tetragonal-to-cubic transition derived from the Kissinger plots for the bulk ammonium nitrate (solid line), ammonium nitrate nanoconfined in native pores (dashed line), and organically modified pores (dotted line).

bulk and nanoconfined ammonium nitrate. It is seen that for both nanoconfined systems ΔG* is lower than that for the bulk. Lowering of the free energy barrier suggests acceleration of the transition and therefore its shift to lower temperature. This can explain the experimentally observed behavior of ammonium nitrate nanoconfined in organically modified silica. However, it cannot explain the shift to higher temperature observed for this substance nanoconfined in native silica. Alternatively, a change in the rate of the transition can be caused by a change of the entropic component of the rate

temperature. For all three systems E decreases with increasing T as predicted by the nucleation model. An interesting feature of the E vs T dependencies shown in Figure 5 is the huge values of effective activation energy. This is readily rationalized in terms of the nucleation model (eq 4) that suggests that E should rise to infinity when the temperature approaches the equilibrium temperature of the transition (T0).

Table 1. Values of A, T0, r, ln w0, and n for the Tetragonal-to-Cubic Transition of Ammonium Nitrate in the Bulk and Nanoconfined Samples system

T0, K

A, kJ mol−1 K2

r

ln(w0, s−1)

n

bulk in modified silica in native silica

393.78 ± 0.08 394.33 ± 0.02 399.47 ± 0.04

711 ± 28 35 ± 1 394 ± 12

0.9957 0.9881 0.9959

0.77 ± 0.02 −0.12 ± 0.04 −0.63 ± 0.02

0.880 ± 0.004 0.80 ± 0.01 0.767 ± 0.008

9631

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C represented by the pre-exponential factor w0 in eq 1. That is, deceleration of the transition and, thus, its shift to higher temperature can be caused by a decrease in the w0. Fitting data to eq 8 resulted in the ln w0 and n values listed in Table 1. Indeed, the ln w0 for the transition of ammonium nitrate nanoconfined in native silica is about 1.4 units smaller than that for the bulk process. In other words, the preexponential factor for the former is about 4 times smaller than for the latter. Therefore, the temperature shift observed for the process in native silica appears to be caused by entropic reasons. As a side note, the preexponential factors estimated for the solid−solid phase transitions appear to be quite small, less than 10 s−1 (see Tables 1−3). It should be stressed that similar computations for the process of the coil to globule transition in an aqueous solution of PNIPAM yielded preexponential factors that were as large as 102−103 s−1.47 Apparently the lower values of the preexponential factor for the solid−solid phase transitions indicate that the relaxation times in the solid state are much longer than in the liquid. It is worth mentioning that the reaction order values found for the solid−solid transitions appear close to 1 that may indicate the monomolecular nature of the transition. This does not seem surprising considering that rearrangement of the crystalline lattice is somewhat similar to the process of isomerization, whose kinetics provides a common example41 of monomolecular processes. Note that the coil to globule transition in an aqueous solution of PNIPAM demonstrated47 reaction order close to 2 which would be expected of a bimolecular process that involves two different species, i.e., water and polymer. Ammonium Perchlorate. This compound undergoes the transition from orthorhombic to cubic phase. Figure 7 presents

the literature ones. The transition temperature has been reported to be 240 °C.24 A broader range 240−255 °C has been reported by Pai Verneker.26 The reported enthalpies are 85 ± 3 27 and 96 28 J g−1. Figure 8 demonstrates a dependence of the DSC peak temperatures on heating rate for the three systems of

Figure 8. DSC peak temperatures for the orthorhombic-to-cubic transition of ammonium perchlorate in bulk (half-filled circles), organically modified pores (open circles), and native pores (solid circles).

ammonium perchlorate. The data show that at each of the heating rates, the transition temperature of ammonium perchlorate nanoconfined in organically modified silica is consistently lower than for the bulk. The difference is ∼3 °C for all of the studied heating rates. On the other hand, the transition temperature of ammonium perchlorate nanoconfined in native silica reveals a very small shift to higher temperature at slower heating rates and a very small shift to lower temperature at faster heating rates. The shifts, however, are comparable to the errors in estimating the Tp values and, thus, can be neglected. To make sure that the observed effects are specific to nanoconfined systems, DSC runs were also performed on mechanical mixtures of ammonium perchlorate with native and organically modified silica. Just as in the case of ammonium nitrate, the resulting Tp values were practically identical for both mechanically mixed systems and the bulk. The obtained DSC data were subjected to kinetic analysis to determine the free energy barrier of nucleation and preexponential factors. The Kissinger plots and resulting E on T dependencies (not shown) looked very similar to those shown in Figures 4 and 5. That means that the Kissinger plots were strongly nonlinear and the E vs T dependencies were decreasing. Fitting the E vs T dependencies to eq 4 produced the A and T0 parameters shown in Table 2. Substitution of these parameters in eq 2 gave rise to the temperature dependencies of ΔG* depicted in Figure 9. Unlike in the case of ammonium nitrate, the ΔG* values for both nanoconfined systems of ammonium perchlorate are larger than the bulk values. An increase in ΔG* could explain slowing down the transition and, therefore, its shift to higher temperature. On the other hand, comparing the preexponential factors (Table 2) suggests that both nanoconfined systems have markedly larger ln w0 than the one for the bulk. This would mean acceleration of the process and its shift to lower

Figure 7. DSC curves measured at heating rate 14 °C min−1 for ammonium perchlorate in bulk and native and organically modified pores (solid line, bulk; dashed line, native pores; dotted line, organically modified pores).

DSC curves of bulk and nanoconfined ammonium perchlorate. The appearances of the DSC peaks for all three systems are quite similar to the respective peaks obtained for ammonium nitrate (Figure 2). Namely, the DSC peaks for nanoconfined systems become more asymmetrical and demonstrate a smaller height to width ratio. The endothermic heat of transition for the bulk sample of ammonium perchlorate was measured to be 90 J g−1. The transition temperature defined as Tp was found to be between 242 and 246 °C for the range of the heating rates used in our work. The obtained parameters compare well with 9632

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C

Table 2. Values of A, T0, r, ln w0, and n for the Orthorhombic-to-Cubic Transition of Ammonium Perchlorate in the Bulk and Nanoconfined Samples system

T0, K

A, kJ mol−1 K2

r

ln(w0, s−1)

n

bulk in modified silica in native silica

508.74 ± 0.04 501.5 ± 0.2 506.29 ± 0.06

529 ± 16 1996 ± 130 1636 ± 50

0.996 0.986 0.996

0.52 ± 0.04 1.06 ± 0.08 2.20 ± 0.02

0.96 ± 0.01 0.95 ± 0.02 1.20 ± 0.01

Figure 10. DSC curves measured at heating rate 14 °C min−1 for sodium nitrite in bulk, native, and organically modified pores (solid line, bulk; dashed line, native pores; dotted line, organically modified pores).

Figure 9. Temperature dependence of the free-energy barrier to nucleation for the orthorhombic-to-cubic transition in ammonium perchlorate in bulk (solid line), native pores (dashed line), and organically modified pores (dotted line).

that the respective literature values vary from 4−5 30−32 to 23 33and 26 24 J g−1. The Tp values presented in Figure 11 indicate that the transition temperature for sodium nitrite nanoconfined in

temperature. Although in both nanoconfined systems ΔG* and ln w0 change in a similar manner, we observe acceleration (decrease in Tp) for organically modified silica system but virtually no effect (no shift in Tp) for the native silica system. Apparently this difference is associated with the difference in the values of the reaction order (Table 2). The n value for the native silica system is markedly larger than for the organically modified one that makes the transition in the former slower, therefore canceling out the accelerating effect of an increase in the preexponential factor. This does not occur in the organically modified system so that the respective transition accelerates because of an increase in the ln w0 value. Sodium Nitrite. The transition in sodium nitrite is more complex than the two other transitions considered earlier. It combines two strongly overlapped transitions III → II and II → I which occur within approximately 1 °C from each other.29 The III → II transition is a first-order transition, during which the orthorhombic ferroelectric phase becomes antiferroelectric. It is immediately followed by a second-order30 transition of the antiferroelectric to paraelectric phase. The transitions merge into a single DSC peak at the heating rates that are at least 4 °C min−1.31 Figure 10 displays the DSC curves obtained for the transition in sodium nitrite in the bulk and nanoconfined states. The bulk sample gives rise to a sharp symmetrical peak, whereas the peaks for nanoconofined systems are asymmetrical with a large width to height ratio. In the bulk system, the transition temperature defined as Tp increases from 164 to 167 °C with increasing heating rates used in this study. The endothermic heat of transition was determined to be 18−20 J g−1. The literature values for the III → II transition temperatures are usually detected in the range29−32 160−165 °C. Because at slow heating rates the III → II and II → I transitions appear as two overlapped DSC peaks, the enthalpy of the III → II transition is difficult to evaluate accurately so

Figure 11. DSC peak temperatures for the ferroelectric-to-paraelectric transition of sodium nitrite in bulk (half-filled circles), organically modified pores (open circles), native pores (solid circles).

organically modified silica is about 3 °C lower than that for the bulk process. However, the transition in native silica demonstrated very unusual behavior. First of all, it revealed unusually large (12−15 °C) depression of the transition temperature. A shift to lower temperature for the transition in native silica pores has not been observed for ammonium nitrate and ammonium perchlorate as well as for ammonium chloride in our previous work.12 On the contrary, all these systems tend to demonstrate at least some increase in the transition 9633

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C

Table 3. Values of A, T0, r, ln w0, and n for the Ferroelectric-to-Paraelectric Transition of Sodium Nitrite in the Bulk and Nanoconfined Samples system

T0, K

A, kJ mol−1 K2

r

ln(w0, s−1)

n

bulk in modified silica

432.72 ± 0.04 431.54 ± 0.04

264 ± 20 95.9 ± 0.4

0.9890 0.9998

1.02 ± 0.02 −0.70 ± 0.04

0.88 ± 0.02 0.62 ± 0.04

pores faces a markedly lower energy barrier than that in bulk. This is obviously consistent with this process being faster and, thus, having lower transition temperature relative to the bulk. On the other hand, the preexponential factor for the transition in organically modified silica dropped slightly below the respective value for the bulk. Needless to say, such a change can only mean deceleration of the process and thus cannot explain the experimentally observed depression of the transition temperature. In other words, the effect seen for the transition in organically modified silica appears to be related to lowering the free energy barrier as it was in the case of the transition in ammonium nitrate nanoconfined to organically modified silica.

temperature relative to the bulk systems. In addition to that, unlike all other systems this one does not show a systematic increase in Tp with heating rate. The reasons for such unusual behavior are not clear. It was our guess that in native pores sodium nitrate may undergo very dramatic structural changes. These can be caused by exchange of H+ in the silanol groups for Na+. This process is known to occur in native silica.13 We attempted to find some evidence of this process with the help of IR spectroscopy. Unfortunately, the FTIR spectra of compounds loaded in silica are difficult to analyze because of the massive overlaps with the absorption peaks associated with silica itself. The only obvious change was related to the absorption band of silica at 975 cm−1. The peak is associated with the Si−O−H group in neat silica.48 Relative to neat silica, the intensity of this peak in the sample containing sodium nitrite was decreased significantly. This result is consistent with the possible exchange of H+ for Na+. Perhaps this process is a reason for the formation of strongly distorted crystalline lattice of sodium nitrite formed in the contact with native silica surface. At any rate, the absence of the heating rate dependence of the Tp values deemed the data unsuitable for kinetic analysis. Therefore, further study of the transition in the native silica was not pursued. The fact that depression of the transition temperature in organically modified silica is specific to nanoconfinement was confirmed by DSC runs on a mechanical mixture of organically modified silica with sodium nitrate. The mixed system did not demonstrate any depression of the transition temperature. The DSC data on the transition of the bulk and nanoconfined sodium nitrite were used for kinetic analysis that produced nonlinear Kissinger plots and decreasing dependencies of E vs T as expected from the nucleation model (eq 4). The E vs T data were employed to determine the parameters of eq 4 (Table 3) that were then substituted in eq 2 to evaluate the temperature dependence of ΔG* (Figure 12). According to Figure 12, the transition of organically modified

5. CONCLUSIONS The present study demonstrates that confining ionic solids to native and organically modified silica nanopores has significantly different effects on the temperature of the solid−solid transitions. Our interpretation of the difference is that in the native silica the effect of nanoconfinement is primarily the surface effect, i.e., the effect of strong interaction of the ionic solids with the charged silica surface. However, in organically modified silica the surface has very weak interaction with the ionic solids so that in this system the effect of nanoconfinement is primarily the size effect. The effect of nanoconfinement in organically modified silica consistently is a depression of the transition temperature. This effect is in agreement with other studies when the solid−solid transitions were probed in molecular solids nanoconfined to the native pores of silicate glasses. Thus, as regards the size effect, one can conclude that diminishing the size of a solid should result in a decrease of the temperature of the solid−solid transition. The surface effect appears to be far more complex because of the variety of possible interactions between ionic solids and charged surface of silica. In this study alone, we have found that the temperature of the solid−solid transitions taking place in native silica can be elevated (ammonium nitrate and ammonium chloride12), depressed (sodium nitrite), or nearly unchanged (ammonium perchlorate). This obviously suggests that the size effect should be studied only in the absence of strong surface interactions. Kinetic analysis demonstrates that the shifts in the transition temperature can be associated with a change in the free energy of nucleation as well as in the preexponential factor. In other words, the effect generally arises from a combination of changes in both parameters.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We gratefully acknowledge the partial financial support of the National Science Foundation under Grant CHE 1052828. We also thank SiliCycle Inc. (Canada) for the generous gift of

Figure 12. Temperature dependence of the free-energy barrier to nucleation for the ferroelectric-to-paraelectric transition in sodium nitrite in bulk (solid line) and organically modified pores (dotted line). 9634

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C

(23) Steiner, L. E.; Johnston, J. Development of a Method of Radiation Calorimetry, and the Heat of Fusion or Transition of Certain Substances. J. Phys. Chem. 1928, 32, 912−939. (24) Search Empirical Formula (in Russian). http://www.chem.msu. ru/cgi-bin/tkv.pl (accessed Dec 2014). (25) Erdey, L.; Gal, S.; Liptay, G. Thermoanalytical Properties of Analytical Grade Reagents. Ammonium Salts. Talanta 1964, 11, 913− 940. (26) Pai Verneker, V. R.; Rajeshwar, K. Crystal Structure Transformations in Ammonium and Alkali Metal Perchlorates. Thermochim. Acta 1975, 13, 305−314. (27) Kishore, K.; Pai Verneker, V. R.; Mohan, V. K. Differential Scanning Calorimetric Studies on Ammonium Perchlorate. Thermochim. Acta 1975, 13, 277−292. (28) Evans, M. W.; Beyer, R. B.; McCulley, L. Initiation of Deflagration Waves at Surfaces of Ammonium Perchlorate−Copper Chromite−Carbon Pellets. J. Chem. Phys. 1964, 40, 2431−2438. (29) Rao, C. N. R.; Prakash, B.; Natarajan, M. Crystal Structure Transformations in Inorganic Nitrites, Nitrates, and Carbonates; National Bureau of Standards: Washington, DC, 1975; http://www.nist.gov/ data/nsrds/NSRDS-NBS-53.pdf) (accessed Dec 2014). (30) Hatta, I.; Ichikawa, H.; Todoki, M. Application of Dynamic Differential Scanning Calorimetry To Study Phase Transitions. Thermochim. Acta 1995, 267, 83−94. (31) House, J. E.; Goerne, J. M., Jr. Thermal Studies on the Phase Transitions in Sodium Nitrite. Thermochim. Acta 1993, 215, 297−301. (32) Kourkova, L.; Svoboda, R.; Sadovska, G.; Podzemna, V.; Kohutova, A. Heat Capacity of NaNO2. Thermochim. Acta 2009, 491, 80−83. (33) Sakiyama, M.; Kimoto, A.; Seki, S. Heat Capacities and Volume Thermal Expansion of NaNO2 Crystal. J. Phys. Soc. Jpn. 1965, 20, 2180−2184. (34) Anwander, R.; Nagl, I.; Widenmeyer, M.; Engelhardt, G.; Groeger, O.; Palm, C.; Röser, T. Surface Characterization and Functionalization of MCM-41 Silicas via Silazane Silylation. J. Phys. Chem. B 2000, 104, 3532−3544. (35) Skoog, D. A.; Holler, F. J.; Crouch, S. R. Principles of Instrumental Analysis, 6th ed.; Thomson: Belmont, CA, 2007. (36) Yancey, B.; Vyazovkin, S. Venturing into Kinetics and Mechanism of Nanoconfined Solid-State Reactions: Trimerization of Sodium Dicyanamide in Nanopores. Phys. Chem. Chem. Phys. 2014, 16, 11409−11416. (37) Vyazovkin, S.; Wight, C. A. Kinetics of Thermal Decomposition of Cubic Ammonium Perchlorate. Chem. Mater. 1999, 11, 3386−3393. (38) Vyazovkin, S.; Clawson, J. S.; Wight, C. A. Thermal Dissociation Kinetics of Solid and Liquid Ammonium Nitrate. Chem. Mater. 2001, 13, 960−966. (39) Vyazovkin, S. Isoconversional Kinetics of Thermally Stimulated Processes; Springer: Heidelberg, Germany, 2015. (40) West, A. R. Solid State Chemistry and Its Applications; Wiley: Chichester, U.K., 1992. (41) Atkins, P.; de Paula, J. Physical Chemistry, 9th ed.; W. H. Freeman: New York, 2010. (42) Papon, P.; Leblond, J.; Meijer, P. H. E. The Physics of Phase Transitions; Springer: Berlin, 2002. (43) Chen, K.; Baker, A. N.; Vyazovkin, S. Concentration Effect on Temperature Dependence of Gelation Rate in Aqueous Solutions of Methylcellulose. Macromol. Chem. Phys. 2009, 210, 211−216. (44) Kissinger, H. E. Reaction Kinetics in Differential Thermal Analysis. Anal. Chem. 1957, 29, 1702−1706. (45) Vyazovkin, S.; Yancey, B.; Walker, K. Nucleation Driven Kinetics of Poly(ethylene terephthalate) melting. Macromol. Chem. Phys. 2013, 214, 2562−2566. (46) Vyazovkin, S.; Yancey, B.; Walker, K. Polymer Melting Kinetics Appears To Be Driven by Heterogeneous Nucleation. Macromol. Chem. Phys. 2014, 215, 205−209. (47) Farasat, R.; Vyazovkin, S. Coil-to-Globule Transition of Poly(Nisopropylacrylamide) in Aqueous Solution: Kinetics in Bulk and Nanopores. Macromol. Chem. Phys. 2014, 215, 2112−2118.

ultrapure silica gel samples and Mettler-Toledo for donation of the TGA instrument and loan of the DSC instrument used in this study.



REFERENCES

(1) Christenson, H. K. Confinement Effects on Freezing and Melting. J. Phys.: Condens. Matter. 2001, 13, R95−R133. (2) Alcoutlabi, M.; McKenna, G. B. Effects of Confinement on Material Behaviour at the Nanometre Size Scale. J. Phys.: Condens. Matter 2005, 17, R461−R524. (3) Awschalom, D. D.; Warnock. Supercooled Liquids and Solids in Porous Glass. J. Phys. Rev. B 1987, 35, 6779−6785. (4) Mu, R.; Malhotra, V. M. Effects of Surface and Physical Confinement on the Phase Transitions of Cyclohexane in Porous Silica. Phys. Rev. B 1991, 44, 4296−4303. (5) Pan’kova, S. V.; Poborchii, V. V.; Solov’ev, V. G. The Giant Dielectric Constant of Opal Containing Sodium Nitrate Nanoparticles. J. Phys.: Condens. Matter 1996, 8, L203−L206 (the article devoted to NaNO2 mistakenly called sodium nitrate in the title). (6) Colla, E. V.; Fokin, A. V.; Koroleva, E. Yu.; Kumzerov, Yu. A.; Vakhrushev, S. B.; Savenko, B. N. Ferroelectric Phase Transitions in Materials Embedded in Porous Media. Nanostruct. Mater. 1999, 12, 963−966. (7) Naberzhnov, A.; Fokin, A.; Kumzerov, Yu.; Sotnikov, A.; Vakhrushev, S.; Dorner, B. Structure and Properties of Confined Sodium Nitrate. Eur. Phys. J. E 2003, 12, S22−S24. (8) Kutnjak, Z.; Vodopivec, B.; Blinc, R.; Fokin, A. V.; Kumzerov, Y. A.; Vakhrushev, S. B. Calorimetric and Dielectric Studies of Ferroelectric Sodium Nitrite Confined in Nanoscale Porous Glass Matrix. J. Chem. Phys. 2005, 123, 084708-1−084708-5. (9) Rysiakiewicz-Pasek, E.; Komar, J.; Cizman, A.; Poprawski, R. Calorimetric Investigations of NaNO3 and NaNO2 Embedded into Porous Glasses. J. Non-Cryst. Solids 2010, 356, 661−663. (10) Colla, E. V.; Fokin, A. V.; Kumzerov, Yu. A. Ferroelectrics Properties of Nanosize KDP Particles. Solid State Commun. 1997, 103, 127−130. (11) Mu, R.; Jin, F.; Morgan, S. H.; Henderson, D. O.; Silberman, E. The Possible Crossover Effects of NaNO3 in Porous Media: From Bulk to Clusters. J. Chem. Phys. 1994, 100, 7749−7753. (12) Farasat, R.; Yancey, B.; Vyazovkin, S. High Temperature Solid− Solid Transition in Ammonium Chloride Confined to Nanopores. J. Phys. Chem. C 2013, 117, 13713−13721. (13) Iler, R. K. The Chemistry of Silica; Wiley: New York, 1979. (14) Hossain, D.; Pittman, C. U., Jr.; Saebo, S.; Hagelberg, F. Structures, Stabilities, and Electronic Properties of Endo- and Exohedral Complexes of T10-Polyhedral Oligomeric Silsesquioxane Cages. J. Phys. Chem. C 2007, 111, 6199−6209. (15) Saka, H.; Nishikawa, Y.; Imura, T. Melting Temperature of In Particles Embedded in an Al Matrix. Philos. Mag. A 1988, 57, 895− 906. (16) Grabaek, L.; Bohr, J. Superheating and Supercooling of Lead Precipitates in Aluminum. Phys. Rev. Lett. 1990, 64, 934−937. (17) Watanabe, A.; Iiyama, T.; Kaneko, K. Melting Temperature Elevation of Benzene Confined in Graphitic Micropores. Chem. Phys. Lett. 1999, 305, 71−74. (18) Kaneko, K.; Watanabe, A.; Iiyama, T.; Radhakrishnan, R.; Gubbins, K. E. A Remarkable Elevation of Freezing Temperature of CCl4 in Graphitic Micropores. J. Phys. Chem. B 1999, 103, 7061−7063. (19) Miyahara, M.; Gubbins, K. E. Freezing/Melting Phenomena for Lennard-Jones Methane in Slit Pores: A Monte Carlo Study. J. Chem. Phys. 1997, 106, 2865−2880. (20) Marquardt, P.; Gleiter, H. Ferroelectric Phase Transition in Microcrystals. Phys. Rev. Lett. 1982, 48, 1423−1425. (21) Israelachvili, J. Intermolcular & Surface Forces, 2nd ed.; Academic Press: Amsterdam, 1991. (22) Farasat, R.; Yancey, B.; Vyazovkin, S. Loading Salts from Solutions into Nanopores: Model and Its Test. Chem. Phys. Lett. 2013, 558, 72−76. 9635

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636

Article

The Journal of Physical Chemistry C (48) Vijayalakshmi, U.; Balamurugan, A.; Rajeswari, S. Synthesis and Characterization of Porous Silica Gels for Biomedical Applications. Trends Biomater. Artif. Organs 2005, 18, 101−105.

9636

DOI: 10.1021/acs.jpcc.5b01716 J. Phys. Chem. C 2015, 119, 9627−9636