Diverting the phase transition behaviour of adipic acid via mesoporous silica confinement

Abstract

Manipulating the phase transition behaviour of phase change materials (PCMs) to a favorable range is an urgent issue which poses a considerable challenge to their widespread application, since PCMs with desirable operational temperatures and high efficiency are very few in practice. An attempt to address this puzzle is presented by confining adipic acid in silica scaffolds through an impregnation method. Three different nanoscale chambers – 2D-hexagonal silicas MCM-41 and SBA-15, and 3D-silica mesoporous foam (MCF) – were employed. In this way, novel form-stable PCMs were developed, which acquired size-dependent thermal behaviour. The evolution of thermal behaviour as a function of pore size was investigated. Phase transition temperature (T_m) deviations presented a strong linear relationship with inverse pore sizes, conforming to the modified Gibbs–Thomson equation. Enthalpy changes (ΔH_m) shared the same trend with T_m, as pore sizes decreased, while the variation of phase transition temperatures (ΔT_m) was in the opposite direction. Unlike the bulk state, the phase transition temperature of adipic acid in a confined space could be regulated over 90 °C. The maximum thermal storage efficiency was 72% with respect to the confined acid content. Thus, the solid–liquid phase transition behaviour of adipic acid in a confined space could be altered, and the manipulation of thermal behaviour was achieved.

---

1 Introduction

Thermal energy conversion is both economically and environmentally vital.¹ To relieve at one stroke the growing concerns about environmental problems and the imminent energy shortage, investigations into latent heat storage have been on the agenda of researchers for a while. As one of the most efficient ways of energy conversion, latent heat storage presents many benefits, such as high energy storage density, renewability, low cost and so on. Therefore, latent heat storage has regained much attention in the past few decades, and much research has been done since it was pioneered by Telkes and Raymond in 1940s.^2–8

However, the large-scale utilization of latent heat storage is possible only if consideration is given to comprehensive factors including operational temperature, thermal conductivity, corrosion and so on. In particular, there are few phase change materials (PCMs) with the desired operational temperature in latent heat storage systems. Hence, diverting the phase transition of PCMs is of great significance.

To manipulate the phase transition over a large regulation window, the eutectoids of both inorganic and organic compounds have been widely investigated on the basis of their corresponding binary phase diagrams.⁹ Nevertheless, investigations concerning PCMs with high thermal storage capacity are scarce with respect to the wide applied temperature range in practical use.

One other approach to tackle this puzzle is to incorporate PCMs into porous materials to form nominal shape-stabilized PCMs. In this case, PCMs can be absorbed into various pores by capillary forces and surface tension. It has been reported that a variety of characteristics of the inclusions, such as magnetic, electrical, mechanical, and optical properties, can be markedly affected by spatial confinement.^10,11 For example, alterations of crystallization state,^12–14 crystal habits,^15–17 and polymorphism phenomena^18,19 have been observed when compared with the bulk. It is reasonable to hypothesize that phase transitions, on the basis of crystallization and fusion processes, may be altered as well. Meanwhile, the spatial confinement effect may play a prominent role in the crystallization processes of inclusions, even leading to extraordinarily high heat storage efficiency.^20–22 Collectively, the thermal behaviour of systems dimensionally reduced to nanometer sizes is a topic of great fundamental significance and practical relevance.

As a favorable class of scaffolds in the fields of catalysis,^23–25 separation,^26–29 and drug delivery,^30–35 mesoporous silica nanoparticles (MSNs) present versatile characteristics, such as high specific area, good stability, periodic structure, and nanopores with a narrow size distribution and variable size. These nanoparticles have aroused wide interest in both theoretical and experimental fields since they were pioneered by Mobil Co., Ltd in 1992.³⁶ Furthermore MSNs can afford composite PCMs. Thus, latent heat storage can offer nominal solid–solid phase transitions rather than their solid–liquid counterparts, which are even more popular in practice.

The aim of the present work was two-fold: firstly, to investigate the thermal behaviour of a dicarboxylic acid imbibed in nanochambers; and secondly, to attempt to validate the feasibility of manipulating the phase transitions of dicarboxylic acid-based PCMs. To this end, in this study, adipic acid was imbibed into mesoporous silica chambers by the molten-impregnation method. Adipic acid underwent a solid–liquid transition at 152 °C with a latent heat of ca. 250 J g⁻¹; this was selected as a representative dicarboxylic acid PCM. Mesoporous silica particles with different nominal pore sizes, namely MCM-41 (3.3 nm), SBA-15 (9.4 nm), and mesoporous foam (MCF; 15.0, 18.4, and 24.6 nm), served as the corresponding scaffolds.

2 Experimental section

2.1 Materials

Adipic acid (99.5%), hydrochloric acid (36.0–38.0%), aqueous ammonia (25.0–28.0%), hexadecyl trimethyl ammonium bromide (CTAB, 99%), 1,3,5-trimethylbenzene (TMB, 99.5%) and tetraethyl orthosilicate (TEOS, 99.99%) were obtained from Aladdin. Triblock copolymer surfactant P123 (EO20PO70EO20, M_w = 5800) was obtained from Sigma-Aldrich. All chemicals were used as received.

2.2 Preparation of mesoporous silica

MCM-41 was synthesized by the surfactant templating method as reported elsewhere.^36,37 Typically, 2.5 g CTAB was dissolved in 120 g deionized water, and then 9.5 g aqueous ammonia was added in solution. After 10 g TEOS was added with stirring, a gel was obtained. The mixture was stirred for another hour, the precipitate was filtered, washed with water, and dried in an oven at 90 °C for 12 h.

SBA-15 was prepared as reported in the literature.^38,39 In a typical procedure, 4 g P123 was dissolved in 150 ml of 1.6 M HCl aqueous solution, and then 8.5 g TEOS was added dropwise with vigorous stirring. After being stirred for another 5 minutes, the mixture was transferred into a Teflon-lined autoclave at 35 °C for 20 h, and then aged at 100 °C for 24 h. The white precipitate was filtered, washed with deionized water, dried and conserved.

The synthesis procedure of MCF was similar to that of SBA-15, except that TMB was added. MCF samples with different pore sizes were obtained by changing the weight ratio of TMB : P123. For convenience of description, the as-prepared MCFs are denoted as MCFx, in which x represents the weight ratio of TMB : P123, and the MCFs containing PCMs are denoted as yMCFx, in which y represents the ratio of PCM : MCF. To remove the organic template, the obtained MSNs were calcined at 550 °C (1 °C min⁻¹) for 6 h.

2.3 Impregnation experiments

PCM/MSN composites were prepared by a conventional molten impregnation method. MSN powders were preheated overnight at 110 °C, then cooled to room temperature and collected before use. The impregnation procedure was as follows: briefly, a given weight of adipic acid and predried MSNs were placed in a flask at 90 °C under vacuum for 3 h to remove any possible absorbed water. Afterwards, the vacuum pump was turned off and the temperature was elevated to 160 °C for 2 h while stirring. At the end of the plateau temperature, samples were quenched to room temperature and conserved.

2.4 Characterization

The calcined samples were characterized using nitrogen sorption experiments, scanning electron microscopy (SEM), transmission electron microscopy (TEM) and small-angle X-ray diffraction (SAXRD). The thermal behaviour of adipic acid/mesoporous silica composite materials was characterized using differential scanning calorimetry (DSC). Nitrogen sorption experiments were performed with a Quanta chrome NOVA 4000e analyser at −196 °C. Samples were degassed at 200 °C under vacuum for 6 h before characterization. The pore sizes of MCM-41 and SBA-15 were determined from adsorption branches of the isotherm employing the Barrett–Joyner–Halenda (BJH) method. Considering the ink-bottle pore structure of MCF, the size of the cells and windows of MCF were calculated using the BJH method from adsorption and desorption branches of the isotherm, respectively. Total pore volumes were estimated from the amount adsorbed at a relative pressure (P/P₀) of 0.99. The specific surface areas were determined using the Brunauer–Emmett–Teller (BET) method. SAXRD patterns of pristine mesoporous silica materials and wide-angle X-ray diffraction (WAXRD) patterns of composite materials were collected from a PANalytical X'Pert PRO X-ray diffractometer with Cu Kα (λ = 0.15418 nm) incident radiation. The chemical structure analysis of composite materials was performed using a Nicolet Nexus Fourier transform infrared (FT-IR) spectrophotometer, with KBr pellets. FT-IR spectra were recorded in the range of 4000–400 cm⁻¹ at room temperature. The morphologies of calcined samples were characterized using SEM and TEM. The external surface morphologies and microstructures of mesoporous materials were observed using a JEOL JSM-5610LV High Vacuum Tungsten Filament emission SEM. TEM was performed on a JEOL JEM-2100F field emission TEM with an accelerating voltage of 200 kV. The thermal behaviour characterization was recorded on a Mettler-Toledo DSC2 under a static nitrogen flow of 100 ml min⁻¹ at a scanning rate of 10 °C min⁻¹. Thermogravimetric experiments (TG) were performed using a Nietzsche STA 449F3 simultaneous thermal analyser under a nitrogen atmosphere to investigate the exact amounts of adipic acid in the composite materials. The heating rate in the range of 30–500 °C was 5 °C min⁻¹. Certified indium (supplied by the instrument manufacturer) was used for temperature and heat flow calibration. Unless specified otherwise, the temperature and enthalpy change reported here are the peak temperatures in the DSC curves and the average value of triplicate runs from the second runs onwards.

3 Results and discussion

3.1 Mesoporous silica structure analysis

To investigate the structure of the as-prepared MSNs, SAXRD analysis was carried out, and the results are presented in Fig. 1.

Fig. 1 (a) SAXRD patterns of MCM-41 and SBA-15. (b) SAXRD patterns of MCFs.

Typically, four well-resolved Bragg peaks in the low-angle 2θ region, indexed to the (100), (110), (200) and (210) planes, were observed for MCM-41, indicating the long-range order of the synthesized material.³⁶ At 2θ below 2°, the diffraction patterns of SBA-15 exhibited three well-resolved characteristic reflections, which can be indexed as the (100), (110) and (200) planes, typical of p6mm symmetry. This suggests that pristine MCM-41 and SBA-15 are well ordered at a long range, and contain mesopores organized in a hexagonal array, as expected.³⁸ When TMB was introduced into the synthesis procedure, mesocellular silica foam was obtained, while the p6mm phase disappeared.⁴⁰ As can be seen in the SAXRD patterns of the MCFs, the diffraction peaks indexed to the (110) and (200) planes decreased until they disappeared, and the strongest peak gradually moved to lower angles.

Both SEM and TEM were employed to verify the existence of ordered arrays in MCM-41 and SBA-15, and the characteristic pore structure in the MCFs (Fig. 2). It can be concluded from the micrographs that both MCM-41 and SBA-15 present the typical “rod like” morphology. The ultra large primary mesopores of the “ink-bottle” MCF structure could be observed in the TEM micrographs. To obtain the comprehensive pore structure parameters of the pristine mesoporous materials, a nitrogen sorption experiment was carried out, and the results are summarized in Table 1.

Fig. 2 (I) SEM micrographs of (a) MCM-41 and (b) SBA-15. (II) TEM micrographs of pristine (a) MCF0.5, (b) MCF1, and (c) MCF2 at the same magnification.

Table 1 Nitrogen sorption results for MSNs

Samples	Pore diameter (d, nm)		Pore volume^c (V, cm^3 g^−1)	Specific surface area^d (SBET, m^2 g^−1)
	Primary pore size^a	Window pore size^b
a Pore sizes were derived by the BJH method from the adsorption branches of the isotherm.b Pore sizes were derived by the BJH method from the desorption branches of the isotherm.c Total pore volumes were estimated by the amount adsorbed at a relative pressure of about 0.99.d Specific surface areas were evaluated using the multipoint BET algorithm using adsorption data in a relative pressure range of 0.04–0.2.
MCM-41	3.3	n. a.	0.8	1043
SBA-15	9.4	n. a.	1.0	586
MCF0.5	15.0	6.2	1.3	425
MCF1	18.4	4.8	1.7	521
MCF2	24.6	8.1	2.0	449

---

Thus, the characterization results mentioned above indicate that the pore sizes of the pristine mesoporous silica samples were in the range of 3.3–24.6 nm, and that all the pristine mesoporous silica scaffolds were well-defined.

3.2 Thermal behaviour analysis of PCMs

The thermal behaviour of adipic acid in various confined mesoporous chambers was investigated using DSC, and the corresponding results are summarized in Table 2 and Fig. 3. All these data were obtained using a heating/cooling rate of 10 °C min⁻¹. Because of the broadening phenomenon of peaks in DSC curves, in these experiments, the peak temperature was denoted as T_m.

Table 2 Thermal behaviour parameters of samples^a

Samples	(O/U)^b	Tmc^c (°C)	Tmb^d (°C)
a All data were obtained from DSC under a nitrogen atmosphere at a scanning rate of 10 °C min^−1 in the operational temperature range of 0–170 °C.b O and U represent the pore filling condition, in which O represents over-filled and U represents under-filled pores.c Tmc values represent the melting temperatures of adipic acid in confined spaces.d Tmb values represent the melting temperatures of adipic acid in its bulk state.
Bulk AA			152
1MCM-41	O	61.6	147.0
1.5MCM-41	O	60.6	148.1
2MCM-41	O	60.0	148.5
1SBA-15	O	122.8	148.1
1.5SBA-15	O	120.1	147.5
2SBA-15	O	118.8	147.9
1MCF0.5	U	132.4	n. a
1.5MCF0.5	O	133.6	147.9
2MCF0.5	O	134.6	149.3
1MCF1	U	136.4	n. a
1.5MCF1	U	138.5	n. a
2MCF1	O	139.3	147.6
1MCF2	U	138.0	n. a
1.5MCF2	U	138.1	n. a
2MCF2	U	140.5	148.6
2.5MCF2	O	141.2	149.6

---

Fig. 3 DSC profiles of samples (composite materials with an adipic acid mass ratio of 0.67 are taken as the representative samples).

3.2.1 Analysis of the phase transition temperature in confined chambers. As displayed in Table 2 and Fig. 3, all the DSC curves show similar thermal events. When the confined chambers in the porous framework were under-filled, there was only one endothermic event in the temperature range of 0–170 °C, related to the adipic acid phase transition in its confined state. When over-filled pores were present, two endothermic events with high resolution could be observed, ascribing to the adipic acid phase transition in the confined chambers and the spillover outside. In these two situations, all the thermal energy could be retrieved in the sequential temperature ramp. As presented in Table 2, for each set of composite materials, the confined portion of the curves shows similar endothermic events, nearly at a constant temperature. At the same time, the spillover counterparts in all sets of samples show similar behaviour at 146–150 °C, slightly below the melting temperature of adipic acid in its bulk state (T_mBulk = 152 °C).

Some nanocrystals adsorbed on the outer surface of the mesoporous silica particles or formed among the mesoporous silica particle aggregates may account for the second peaks in the DSC curves. It is well known that an increased proportion of surface atoms lead to a depressed fusion point.²¹ However, in general, the second peak temperature (T_mb) is in good agreement with that of adipic acid in its bulk state (T_mBulk). This was further verified by WAXRD patterns of composite materials as well (Fig. S2†). In addition, the peak breadth on the DSC curves tends to be broader when the composite samples are confined in smaller mesoporous pores due to the smear effect.⁴¹

An evolution of the adipic acid melting temperature in confined chambers was noticed. A similar phenomenon has frequently been observed in confined systems.^{9–11,18,21,22,41–48} For samples with different confined pore spaces, the smaller the pore size, the greater the depression of the fusion point.

It is interesting to note that the T_mc variations in the different sets of composite samples show distinctive mass ratio dependencies. In the case of the MCM-41 and SBA-15 based composite samples, T_mc is reduced as the mass ratio of adipic acid increases. However, in the MCF-based composite materials, T_mc deviates significantly upward (Fig. 4).

Fig. 4 T_mc evolution dependent on the mass ratio of adipic acid in (a) MCM-41 and SBA-15, and (b) MCF-based materials.

To theoretically analyze the evolution of T_m in confined mesoporous chambers, three factors should be taken into consideration.

(1) Gibbs–Thomson equation.

The Gibbs–Thomson equation, pioneered by Gibbs and Thomson,^41,49,50 has often been used to thermodynamically describe the melting point depression in confined geometries. The equation is based on a thermodynamic equivalence assumption of the free energies between crystal and liquid at T_m. The melting point of the confined crystal can be calculated using the following equation:

where ΔT_m denotes the difference between the melting point of the confined crystal in a pore of a certain size and its bulk counterpart, σ is the solid–liquid interface energy, and θ is the contact angle. ΔH_m and ρ are the bulk melting enthalpy and bulk solid density respectively.

However, it is important to notice that eqn (1a) does not take into consideration the existence of a nonfreezing interlayer with thickness h. If d is replaced by (d − 2h), one can get the modified Gibbs–Thomson equation.⁵¹

here, β represents the Thomson length. According to eqn (1b), the nonfreezing interlayer and Thomson length can be obtained from a plot of d versus t_m. In our case, from the d versus t_m plot (Fig. 5a), h and β were estimated as 1.13 nm, and 1.23 nm respectively.

Fig. 5 (a) Plot of d versus t_m. (b) Plots of T_mc and ΔT_m versus (d − 2h)⁻¹.

To clarify the correlation between T_mc and d, we plotted T_mc in confined pores as a function of (d − 2h)⁻¹ (Fig. 5b). To obtain the adjustable window threshold value of T_m, ΔT_m was also plotted against (d − 2h)⁻¹ (Fig. 5b). In Fig. 3 and 5b, for all sets of DSC profiles, T_m values shifted linearly with (d − 2h)⁻¹ as predicted by the modified Gibbs–Thomson equation.

To describe the experimental data better, two parameters A and B were introduced. Coefficients A and B can be obtained from the ΔT_m versus plot by least-squares fitting.

Thus, the variation in melting temperatures of adipic acid in confined spaces depends on the pore sizes. The final relation can be expressed as ΔT_m = 202.81/(d − 2.263) + 2.51.

The dependencies of T_mc and ΔT_m evolution behaviour on pore diameters are displayed in Fig. 6.

Fig. 6 Plots of T_mc and ΔT_m evolution versus d.

As displayed in Fig. 6, both T_mc and ΔT_m present remarkable evolution behaviour dependent on the confined chamber size. This plot should be of great use in manipulating the phase transition of adipic acid to the desired operational temperature. However, the presence of a threshold value should be noted. For confined pore sizes down at 3.3 nm, a distinct deviation is presented between the experimental data and the value derived from the Gibbs–Thomson equation, and this cannot be ignored. In this situation, the explicit assumptions in the equation are questionable or no longer apply.

(2) The driving parameters for phase changes in confined pores are not restricted to the confined pore sizes, but also include the strength between the adsorbate and the pore wall. The latent heat can contribute to intermolecular forces (especially van der Waals forces). If the interaction between the adsorbate and wall is strong, the potential energy barrier will be increased in the melting process. Thus, T_m will shift to a higher temperature. Otherwise, a reduced T_m could be expected.

It is known that adipic acid exists as dimers in the solid state. This dimer configuration is formed by an intermolecular hydrogen bond through the carboxyl groups. Due to the cooperative interaction and long-range delocalization of the hydrogen bond electronic charges, these hydrogen bonded molecules are more stable than uncoupled ones. Carboxylic acid groups can be linked with silanol groups through weak hydrogen bonds. Physisorbed dimer molecules can be found at silica surfaces as well.^31,52

To investigate the chemical interactions between adipic acid and MSNs, FT-IR measurements were carried out (Fig. 7). As can be seen in the spectrum of adipic acid, the peak at 3200–2500 cm⁻¹ with a broad band and a series of minor satellite bands represents associated –OH stretching vibrations in adipic acid dimers.⁵³ The peak at 1694 cm⁻¹ which represents C=O groups coupled with the peak at 928 cm⁻¹ confirms the presence of carboxylic acid. The peak at 1462 cm⁻¹ can be ascribed to the asymmetric bending of C–C in –CH₂ groups. The peaks at 1428 cm⁻¹ and 1280 cm⁻¹ signify the in-plane deformation vibration of –OH and stretching vibration of C–O in dimers, respectively. The peak at 1412 cm⁻¹ is assigned to the bending vibration of –CH₂. The peak at 736 cm⁻¹ represents the in-plane deformation vibration of –CH₂. In the FT-IR spectra of the composite materials, all samples present similar peaks. The broad peaks at 920–960 cm⁻¹ can be ascribed to the superposition of the Si–OH stretching vibrations in MSNs and the out-of-plane deformation vibration of –OH groups in adipic acid. Typical asymmetric and symmetric Si–O–Si stretching vibrations are centered at 1082 cm⁻¹ and 804 cm⁻¹, respectively. The peak at around 465 cm⁻¹ can be ascribed to the bending vibration of Si–O–Si groups.

Fig. 7 FT-IR spectra of adipic acid and the composite materials (composite materials with an adipic acid mass ratio of 0.67 are taken as the representative samples).

The detailed data relating to the FT-IR spectrum of pristine MCM-41 is given in Fig. S1.† The silica presents typical bands at around 3440 cm⁻¹, 1082 cm⁻¹, 962 cm⁻¹, 801 cm⁻¹, and 461 cm⁻¹. The intense peak at 3200–3500 cm⁻¹ can be assigned to silanol “nets” with cross-linking hydrogen bonding.²⁹ The peaks at 1082 cm⁻¹ and 801 cm⁻¹ represent asymmetric and symmetric stretching vibrations of Si–O–Si. The bands at 962 cm⁻¹ and 461 cm⁻¹ can be ascribed to stretching vibrations of Si–OH groups and bending vibrations of Si–O–Si respectively.

Interestingly, the FT-IR spectra of MCM-41 based composite PCMs are not the result of the superposition of data from pure PCMs and the silica matrixes (Table S1†).

As can be seen in the spectrum of 2MCM-41, there is a new distinct peak centered at 1238 cm⁻¹, which can be ascribed to the bending vibration of –OH in isolated adipic acid molecules. Meanwhile, the characteristic peak ascribed to stretching vibrations of associated Si–OH groups shifts to a lower wavenumber. This indicates that the characteristic peak is averaged in the composite materials. Besides these changes, no other evident changes in the characteristic peaks related to adipic acid dimers (1694 cm⁻¹, 1428 cm⁻¹, 1280 cm⁻¹, 928 cm⁻¹) are observed. Considering the existence of a quasi-liquid layer, it is presumed that the associated adipic acid dimers break into isolated adipic acid. As is well known, weak interactions can be enhanced by shortening the intermolecular spacing due to the high pressure and electric charge density caused by a defined environment.^21,48 Thus, the –C=O⋯H–O–Si hydrogen bonds can be generated. Accordingly, there are two forms of adipic acid in composite materials. It should be noted that not all of the adipic acid can be in contact with the silica surface even in MCM-41 with its high specific surface area. Thus, it is speculated that adipic acid in its dimer form will be located in the pore interior, while isolated adipic acid molecules will be adsorbed on the silica pore surface.

Based on the discussion above, there would be a negative correlation between the fraction of the contact layer and the diameter of the pores. During the melting–cooling process, the adipic acid existing in a dimer configuration should experience a phase transition like that in the bulk state. However, the melting temperature is expected to be reduced due to the nano-size effect of the corresponding crystals. Owing to the existence of the contact layer, a depression of the enthalpy change in composite materials is reasonable.

In the other sets of composite materials, the characteristic peaks overlapped with peaks at 1194 cm⁻¹, and were not as evident as those in the MCM-41 based composite materials. Considering the nature of the weak hydrogen bonds between acid and silanol groups, the relatively large curvature, and the dehydroxylation of the silica surface in the preparation process, the effect of hydrogen bonds in large pores could be marginal.⁴⁵

(3) The Clausius–Clapeyron equation is often used to interpret phase transitions in closed systems.⁴⁴ The specific form of this equation is as follows:

where P and T are pressure and melting temperature, respectively, and ΔV_m and ΔH_m are molar volume and the enthalpy change in the phase transition process. According to this equation, during a solid–liquid transition process, the volume change leads to a pressure change, and eventually to T_m evolution.

In our case, during a heating ramp, ΔH_m is negative and ΔV_m is experimentally positive, thus dT has a negative correlation with dP. In nominal “over-filled pores”, the melting temperature will decrease with increasing mass ratio of adipic acid until the open pores in the composite material are entirely crammed. This is in accordance with our sample sets of MCM-41 and SBA-15. With an increasing mass ratio of adipic acid, the mesopores in MCM-41 and SBA-15 will be enclosed gradually, then the pressure will be elevated, and the melting temperature in these two sample sets will eventually be slightly depressed. The mass ratio dependent evolution of T_mc in MCF-based materials is somewhat different. The 3D well-developed interconnected pore structure endows MCF with good mass transfer characteristics, thus pressure caused by volume changes during solid–liquid transitions can be easily unloaded through the interconnected pore system. What is more, pores in the MCF samples are almost always under-filled, so eqn (2) is not applicable any more. However, the origin of this abnormal phenomenon in MCF composite materials is still perplexing, and currently under investigation.

In terms of ΔT_m, as presented in Fig. 6, the deviation of melting temperature is size-dependent. The phase transition temperature can be regulated over a wide temperature range, and the maximum variable threshold value is as high as 92 °C in our experiments.

3.2.2 Analysis of enthalpy changes in confined chambers. Fig. 8 shows the enthalpy change evolution behaviour in relation to pore size in confined spaces. As presented, the experimental enthalpy change ΔH_mc decreases with decreasing pore size. The narrower the pore is, the greater the depression in the enthalpy change. This phenomenon is widespread in nanoconfined systems.^{9,13,41,43,44,46,47,54}

Fig. 8 Relationship between pore diameter and experimental enthalpy changes, ΔH_mc; calculated ideal enthalpy changes with respect to the crystal phase in pores, ΔH_mci; and ideal enthalpy changes with respect to the total weight of PCMs in confined spaces, ΔH_mi.

The reduced latent heat capacity can be attributed to the physical presence of a “nonfreezing layer” as part of the pore wall.⁵⁵ Thus, in confined pores, there will be an intermediate phase between the silica wall and the well-organized crystal phase. During the heating/cooling process, the crystal phase undergoes a solid–liquid transition, which does not involve the intermediate phase. Therefore, in confined pores, partial melting or freezing frequently occur, leading to a reduced enthalpy change. The different ΔH_m values in the different sets of composite PCMs could be ascribed to a variation in the “nonfreezing layer”. The interfacial inclusions as a proportion of the total inclusions should increase with a decrease in pore diameter. Therefore, it would be reasonable to understand that the crystallized phase of adipic acid formed during the cooling process should be less structured with pores of decreasing size.

It should be noted that the exact amount of adipic acid in a composite material is another significant parameter, which markedly affects the enthalpy change. The corresponding curves and the data of remaining mass fractions are given in Fig. S1 and Table S2† respectively. The result of TG experiments is consistent with that expected from the fabrication procedure. However, the continuous weight loss curves in Fig. S1† make it difficult to determine the exact amounts lost in different stages. It is interesting to note that the TG profile of MCM-41 based materials is more distinguishable. Two stages can be observed in Fig. S1 and S2.† These can be ascribed to the separate weight loss of acid inside and outside of the pores. This phenomenon is consistent with two thermal events with a remarkable temperature difference in the DSC profiles. Evidently, the stability of the acid is affected by confinement. It is presumed that the enhanced interaction between the silica surface and the acid may play a prominent role in this respect.

To calculate the thermal efficiency in a confined space, the ideal thermal enthalpy change and the thermal efficiency with respect to the weight of PCMs in confined pores can be expressed as follows:

ΔH_mi = ΔH_mW_c (3a)

Taking into account the existence of the nonfreezing layer volume, one can estimate the ideal enthalpy change in pores by:

here, V(d − 2h), V(d), W_c, M_pcm, and ΔH_mci represent interior pore volume, total pore volume, weight ratio of confined PCMs in composite materials, weight of PCMs in composite materials, and the ideal enthalpy change excluding the nonfreezing layer volume, respectively. W_c was calculated under the assumption that the specific latent heat of the PCMs outside the pores is identical to that of the bulk state. N is a shape exponent of pores, with a value of 2 for cylindrical pores, and 3 for spherical pores. In our case, with typical cylindrical pores, the shape exponent of MCM-41 and SBA-15 is 2. In the set of MCFs with tangent spherical pores connected with cylindrical windows, the shape exponent value should lie between 2 and 3, and the optimum fitted value is 2.97. The calculated thermal enthalpy changes and the corresponding efficiencies are presented in Fig. 8 and the data are summarized in Table 3.

Table 3 Calculated enthalpy changes and thermal efficiency data

Samples	ΔHmc^a (J g^−1)	ΔHmi^a (J g^−1)	ΔHmci^a (J g^−1)	Φc1^b (100%)	Φc2^c (100%)
a Enthalpy changes presented here are average values derived from DSC profiles related to over-filled pores.b Φc1 are calculated efficiencies with 2 as the shape exponent.c Φc2 are calculated efficiencies with 2.97 as the shape exponent.
MCM-41	21.5	82.2	7.9	26.1	271
SBA-15	28.2	94.8	54.7	29.8	51.7
MCF0.5	83.3	137.3	84.5	60.7	98.6
MCF1	107.7	158.8	107.5	67.8	100.2
MCF2	117.2	162.7	119.6	72.0	98.0

---

As presented in Fig. 8 and Table 3, with the optimum value of 2.97, the calculated enthalpy changes in MCF-based composite materials agree well with those from experimental data, and the efficiency is almost 100%. However, in the set of MCM-41 composite materials, a significant error is present. In this set, the calculated ΔH_mci is 7.9 J g⁻¹, far below that of the experimental data at 21.5 J g⁻¹. It should be noted that the Gibbs–Thomson equation cannot hold when the confined pore size reaches a critical value.^56,57 Thus, this error should be ascribed to the nonfreezing thickness derived from the Gibbs–Thomson equation. In small pores, the effect of the intermediate phase on the interior crystal phase is expected to increase as the pore size decreases, resulting in the break-down of the empirical Gibbs–Thomson equation (Table 4).

Table 4 Extent of supercooling in confined nanochambers

Samples	ΔTc^a (°C)	ΔTb^a (°C)
a Average value in over-filled pores.
MCM-41	27.5	7.8
SBA-15	21.3	12.4
M0.5	32.3	11.2
M1	28.6	9.9
M2	24.6	14.4

---

In addition, the thermal efficiencies with respect to the total weights of PCMs imbibed in nanochambers are sizable. In the set of MCFs, all the efficiency values are above 60%, and the maximum value for Φ_c1 is as high as 72%, while the adjustable melting temperature is 14 °C. Even in MCM-41, with an adjustable melting temperature of 92 °C, an evident phase change is attained, and the corresponding efficiency is still 26.1%.

3.2.3 Analysis of the extent of supercooling in composite materials. During the solid–liquid phase transition process, the hysteresis in confined systems is much greater than that in the bulk state (Table 4 and Fig. 9). The peak breadth in DSC curves tends to be wider and continuous. A simulation in structured pores presents the same characteristic behaviour, indicating that the kinetic effect due to bottle-necks and the pore network, and polydispersity in pores, do not account for this phenomenon.⁵⁸

Fig. 9 Relationship between pore diameter and the extent of supercooling of inclusions, ΔT_c; interfacial PCMs outside pores ΔT_b; and bulk adipic acid ΔT_Bulk.

The increased extent of supercooling is explained by the fact that the pore size can be smaller than the critical nucleus size. To be accommodated in the confined space, the inclusions have to cool further to reduce the critical nucleus size. Besides, the nucleation from the melt is easily accomplished by surface melting.⁵⁸ Thus, the greater extent of hysteresis associated with melting/cooling in a confined system is presented.

3.3 X-ray diffraction analysis

It has been suggested that the crystallographic structure of inclusions could be different from that in the bulk state.^{10,12–14,18} In this case, the thermodynamic parameters in the Gibbs–Thomson equation would change.⁵⁹ To investigate the crystallographic structure of adipic acid in confined chambers, WAXRD patterns were obtained. Two MCM-41 based composite samples (S1 and S2) were prepared as reference samples to simulate cases of acid in and acid outside of the pores, respectively. Sample S1 was prepared by the impregnation method in the same way as the other composite materials. The mass ratio of adipic acid in S1 was calculated according to the equilibrium adsorption capacity in MCM-41. Sample S2 was prepared by just mixing together MCM-41 powder with adipic acid in the same mass ratio as used in 1MCM-41.

In the DSC profiles of the reference samples (Fig. S3†), only one thermal event could be found in each curve, ascribed to adipic acid inside (sample S1) and outside (sample S2) of the pores. XRD spectra of sample S1 present typical patterns of adipic acid and a characteristic amorphous silica matrix. Specially, the Bragg peaks around 6° could be ascribed to the (210) planes in MCM-41. That is, adipic acid was imbibed into MCM-41 successfully. Compared with the bulk, the XRD patterns of composite materials present somewhat weaker diffraction intensities, while all characteristic diffraction peaks of normal adipic acid are maintained (Fig. S2†). Besides, a shift in the reflection angle and broadened peaks could also be found, which are the common phenomena in nano-sized systems.⁴⁷ Consequently, there is no significant difference between the crystallographic structures of adipic acid in nanochambers and in the bulk state.

4 Conclusions

In this work, novel composite PCMs with large tunable operational temperature windows have been fabricated. The melting temperature in confined nanochambers presents remarkable size dependent behaviour over a large temperature range, conforming to the modified Gibbs–Thomson equation. The maximum experimental variation of melting temperature is as high as 92 °C. Excluding the nonfreezing layer volume, the calculated ideal enthalpy change in confined pores agrees well with the experimental data, and the corresponding experimental efficiency (thermal efficiency with respect to the total weight of PCMs in the confined space) is as high as 98% for MCF-based composite materials. Furthermore, 72% of the enthalpy change still could be attained with respect to the total weight of PCMs in the confined space.

Due to an increment for nonfreezing layer volume, the enthalpy change in confined chambers presents a depression behaviour related to pore size, but generally is acceptable. The extent of supercooling in confined systems is larger than that of the bulk state, but is not too severe. It should be noted that the modified Gibbs–Thomson equation breaks down at pore sizes down at 3.3 nm. To estimate the phase change behaviour in small pores, further investigation is needed. Nevertheless, as a crude estimation, the methodology presented here paves the way for manipulation of operational temperatures of PCMs, especially dicarboxylic acid.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (U1407205) and the Youth Innovation Promotion Association of CAS (2015351).

Notes and references

1. D. Haillot, T. Bauer, U. Kröner and R. Tamme, Thermochim. Acta, 2011, 513, 49–59.
2. M. Telkes and E. Raymond, Heat. Vent., 1949, 46, 80–86.
3. F. Agyenim, N. Hewitt, P. Eames and M. Smyth, Renewable Sustainable Energy Rev., 2010, 14, 615–628.
4. M. Delgado, A. Lazaro, J. Mazo and B. Zalba, Renewable Sustainable Energy Rev., 2012, 16, 253–273.
5. D. C. Hyun, N. S. Levinson, U. Jeong and Y. Xia, Angew. Chem., Int. Ed. Engl., 2014, 53, 3780–3795.
6. M. Kenisarin and K. Mahkamov, Renewable Sustainable Energy Rev., 2007, 11, 1913–1965.
7. E. Oró, A. de Gracia, A. Castell, M. M. Farid and L. F. Cabeza, Appl. Energy, 2012, 99, 513–533.
8. B. Zalba, J. M. Marin, L. F. Cabeza and H. Mehling, Appl. Therm. Eng., 2003, 23, 251–283.
9. C.-F. Gao, L.-P. Wang, Q.-F. Li, C. Wang, Z.-D. Nan and X.-Z. Lan, Sol. Energy Mater. Sol. Cells, 2014, 128, 221–230.
10. R. Farasat, B. Yancey and S. Vyazovkin, J. Phys. Chem. C, 2013, 117, 13713–13721.
11. P. Huber, J. Phys.: Condens. Matter, 2015, 27, 103102.
12. M. Beiner, G. T. Rengarajan, S. Pankaj, D. Enke and M. Steinhart, Nano Lett., 2007, 7, 1381–1385.
13. L. P. Wang, J. Sui, M. Zhai, F. Tian and X. Z. Lan, J. Phys. Chem. C, 2015, 119, 18697–18706.
14. D. Li and Y. Huang, J. Raman Spectrosc., 2011, 42, 1396–1404.
15. J. Zhang, Z. Zhang, Q. Ji, Y. Jiang, S. Zhang and Z. Wang, Cryst. Growth Des., 2014, 14, 1520–1525.
16. J.-M. Ha, B. D. Hamilton, M. A. Hillmyer and M. D. Ward, Cryst. Growth Des., 2012, 12, 4494–4504.
17. D. I. Trauffer, A. K. Maassel and R. C. Snyder, Cryst. Growth Des., 2016, 16, 1917–1922.
18. J.-M. Ha, B. D. Hamilton, M. A. Hillmyer and M. D. Ward, Cryst. Growth Des., 2009, 9, 4766–4777.
19. B. D. Hamilton, M. A. Hillmyer and M. D. Ward, Cryst. Growth Des., 2008, 8, 3368–3375.
20. J. Zhang, S. S. Wang, S. D. Zhang, Q. H. Tao, L. Pan, Z. Y. Wang, Z. P. Zhang, Y. Lei, S. K. Yang and H. P. Zhao, J. Phys. Chem. C, 2011, 115, 20061–20066.
21. S. Zhang, S. Wang, J. Zhang, Y. Jiang, Q. Ji, Z. Zhang and Z. Wang, J. Phys. Chem. C, 2013, 117, 23412–23417.
22. L. Pan, Q. Ji, Y. Qin, Y. Jiang, Z. Zhang, S. Zhang and Z. Wang, RSC Adv., 2013, 3, 22326–22331.
23. J. Wisniewska, M. Ziolek, N. Artioli and M. Daturi, J. Catal., 2016, 336, 58–74.
24. B. Dragoi, A. Ungureanu, C. Ciotonea, A. Chirieac, S. Petit, S. Royer and E. Dumitriu, Microporous Mesoporous Mater., 2016, 224, 176–189.
25. S. H. Joo, J. Y. Park, C.-K. Tsung, Y. Yamada, P. Yang and G. A. Somorjai, Nat. Mater., 2009, 8, 126–131.
26. Y.-S. Jun, Y. S. Huh, H. S. Park, A. Thomas, S. J. Jeon, E. Z. Lee, H. J. Won, W. H. Hong, S. Y. Lee and Y. K. Hong, J. Phys. Chem. C, 2007, 111, 13076–13086.
27. A. M. Liu, K. Hidajat, S. Kawi and D. Y. Zhao, Chem. Commun., 2000, 1145–1146,  10.1039/b002661l.
28. X. Wang, M. Lu, H. Wang, Y. Pei, H. Rao and X. Du, Sep. Purif. Technol., 2015, 153, 7–13.
29. L. Giraldo, M. Bastidas-Barranco and J. C. Moreno-Piraján, Colloids Surf., A, 2016, 496, 100–113.
30. Y. Hu, Z. Zhi, Q. Zhao, C. Wu, P. Zhao, H. Jiang, T. Jiang and S. Wang, Microporous Mesoporous Mater., 2012, 147, 94–101.
31. M. Vallet-Regi, F. Balas and D. Arcos, Angew. Chem., Int. Ed., 2007, 46, 7548–7558.
32. F. Balas, M. Manzano, P. Horcajada and M. Vallet-Regí, J. Am. Chem. Soc., 2006, 128, 8116–8117.
33. C. Barbe, J. Bartlett, L. Kong, K. Finnie, H. Q. Lin, M. Larkin, S. Calleja, A. Bush and G. Calleja, Adv. Mater., 2004, 16, 1959–1966.
34. J. M. Rosenholm, C. Sahlgren and M. Linden, Nanoscale, 2010, 2, 1870–1883.
35. D. Wu, S. J. Hwang, S. I. Zones and A. Navrotsky, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 1720–1725.
36. C. T. Kresge, M. E. Leonowicz, W. J. Roth, J. C. Vartuli and J. S. Beck, Nature, 1992, 359, 710–712.
37. M. Grün, K. K. Unger, A. Matsumoto and K. Tsutsumi, Microporous Mesoporous Mater., 1999, 27, 207–216.
38. D. Y. Zhao, J. L. Feng, Q. S. Huo, N. Melosh, G. H. Fredrickson, B. F. Chmelka and G. D. Stucky, Science, 1998, 279, 548–552.
39. D. Y. Zhao, Q. S. Huo, J. L. Feng, B. F. Chmelka and G. D. Stucky, J. Am. Chem. Soc., 1998, 120, 6024–6036.
40. J. S. Lettow, Y. J. Han, P. Schmidt-Winkel, P. D. Yang, D. Y. Zhao, G. D. Stucky and J. Y. Ying, Langmuir, 2000, 16, 8291–8295.
41. C. L. Jackson and G. B. McKenna, J. Chem. Phys., 1990, 93, 9002.
42. T. Kadoono and M. Ogura, Phys. Chem. Chem. Phys., 2014, 16, 5495–5498.
43. S. Kittaka, S. Ishimaru, M. Kuranishi, T. Matsuda and T. Yamaguchi, Phys. Chem. Chem. Phys., 2006, 8, 3223–3231.
44. R. A. Mitran, D. Berger, C. Munteanu and C. Matei, J. Phys. Chem. C, 2015, 119, 15177–15184.
45. K. Nakano, Y. Masuda and H. Daiguji, J. Phys. Chem. C, 2015, 119, 4769–4777.
46. J. Rault, R. Neffati and P. Judeinstein, Eur. Phys. J. B, 2003, 36, 627–637.
47. C. Wang, Q. Li, L. Wang and X. Lan, Thermochim. Acta, 2016, 632, 10–17.
48. Y. C. Wang, L. Y. Zhang, S. Y. Tao, Y. L. An, C. G. Meng and T. Hu, Microporous Mesoporous Mater., 2014, 193, 69–76.
49. J. W. Gibbs, The collected works of J. Willard Gibbs, Longmans, Green, New York, 1928.
50. W. Thomson, Philos. Mag., 1871, 42, 448.
51. P. R. Couchman and W. A. Jesser, Nature, 1977, 269, 481–483.
52. H. T. Flakus, A. Tyl and M. Jablońska, Chem. Phys., 2009, 364, 76–89.
53. L. J. Bellamy, in The Infra-red Spectra of Complex Molecules, Springer, Netherlands, Dordrecht, 1975, pp. 183–202,  DOI:10.1007/978-94-011-6017-9_10.
54. D. Zhang, S. Tian and D. Xiao, Sol. Energy, 2007, 81, 653–660.
55. R. Schmidt, E. W. Hansen, M. Stoecker, D. Akporiaye and O. H. Ellestad, J. Am. Chem. Soc., 1995, 117, 4049–4056.
56. G. Dosseh, Y. Xia and C. Alba-Simionesco, J. Phys. Chem. B, 2003, 107, 6445–6453.
57. R. Radhakrishnan, K. E. Gubbins, A. Watanabe and K. Kaneko, J. Chem. Phys., 1999, 111, 9058–9067.
58. R. Radhakrishnan and K. E. Gubbins, Mol. Phys., 1999, 96, 1249–1267.
59. J. A. Lee, H. Rösner, J. F. Corrigan and Y. Huang, J. Phys. Chem. C, 2011, 115, 4738–4748.

---

Footnote

† Electronic supplementary information (ESI) available: TGA curves, XRD patterns of composite materials, and DSC profiles of reference samples. See DOI: 10.1039/c6ra23498d

---