Thermal energy storage and thermal conductivity properties of fatty acid/fatty acid-grafted-CNTs and fatty acid/CNTs as novel composite phase change materials

In recent year, fatty acids (FAs) are heavily studied for heat storage applications and they have shown promising advantages over other organic phase change materials (PCMs). Among the FAs; capric, palmitic and stearic acids are the most studied PCMs. Several researchers have investigated these FAs and tried to improve their thermal properties, mainly by adding different high conducting fillers, such as graphite, metal foams, CNTs, graphene etc. In most cases, these fillers improved the thermal conductivity and heat transfer property but reduce the heat storage capacity considerably. These composites also lose the mixing uniformity during the charging and discharging process. To overcome these issues, selected FAs were grafted on the functionalized CNT surfaces and used as conductive fillers to prepare FA based composite PCMs. This process significantly contributed to prevent the drastic reduction of the overall heat storage capacity and also showed better dispersion in both solid and liquid state. Thermal cycling test showed the variations in the thermal energy storage values of all composite PCMs, however, within the tolerable grade and they had appreciable phase change stability and good chemical stability even after 2,000 cycles.

Utilization of heat energy using phase change materials (PCMs) is an economical and environment friendly approach 1 . Among the different PCMs, there is a long list of organic compounds which have been studied for latent heat thermal energy storage (LHTES) 2,3 . These PCMs do not suffer from super cooling, they are chemically stable, and possess non-toxic/non-corrosive characteristics 2 . Among the organic PCMs, fatty acids (FAs) are extensively studied in recent years, due to their several superior properties over the other organic PCMs, such as, melting congruency, good chemical stability, nontoxicity, suitable melting temperature range and remarkable latent heat capacity (LHC) for solar passive heating applications 4,5 . Like all other organic PCMs, FAs also suffer from low thermal conductivity (TC) and leakage issue, if not properly housed. Easy and cheaper solution is the composite formation with suitable filler. A suitable filler based composite can enhance the heat conduction ability, and at the same time can provide stable form i.e. housing of the FAs. In a form-stable structure, we can avoid additional cost of encapsulation and can have wide dimensional flexibility. Different conductive Grafting of cnts on fA (cnts-g-fA). Reported literature based procedure 22,23 , was used to acid functionalize CNTs. Required amount of CNTs (5 g) were added to 350 mL of HNO 3 (60% v/v) solution in a round bottom flask and then stirred at room temperature for 3 h. The reaction mixture was refluxed at 120 °C for 12 h, and then cool to room temperature. It was filtrated via vacuum filtration, and the filtrate was washed with de-ionized water until a neutral pH was obtained. These functionalized CNTs (CNTs-COOH) were dried in vacuum oven at 60 °C until constant weight. In the second step of synthesis, CNTs-COOH (4 g) were mixed slowly with 60 mL SOCl 2 in a 250 mL round bottom flask at room temperature. After that, the reaction mixture was refluxed at 80 °C for 24 h. Finally, extra SOCl 2 was distilled off and the product was washed with de-ionized water to achieve acyl chloride-functionalized CNTs (CNTs-COCl).
In the final step, a specific amount of selected FA was melted and combined with the CNT-COCl at a ratio of 2:2 (w/w) into a round bottom flask. The mixture was refluxed at 75 °C for 48 h with stirring at 100 rpm. To

Scientific RepoRtS
| (2020) 10:15388 | https://doi.org/10.1038/s41598-020-71891-1 www.nature.com/scientificreports/ remove the unreacted FA, the product (CNTs-g-FA) was washed carefully, and finally dried in vacuum oven for 24 h. The above-mentioned procedure was repeated to synthesis CNTs-g-FA samples for the other weight ratios of CNTs:FA as 3:1, and 1:3 (w/w). The reaction scheme used for the synthesis of CNTs-COOH, CNTs-COCl and CNTs-g-FA products is indicated in Fig. 1. preparation of fA/cnts-g-fA and fA/cnts composites. To prepare FA/CNTs-g-FA composite samples, required amount of FA was melted and slowly mixed with CNTs-g-FA (grafting ratio of 2:2) in weight fraction of 95/5 (%) using a controlled shaker at 200 rpm for 30 min. After thorough mixing, the composite sample was cool to room temperature and stored in a glass bottle for further use. The same method was used to fabricate the other composites using CNTs-g-FA with grafting ratio of 3:1 and 1:3 (w/w). For comparison, composite samples with pure CNTs (FA/CNTs) were prepared using the same procedure, as described above. According to grafting ratio, 3:1, 2:2 and 1:3(w/w), the synthesized grafted products were called as CNTs-g-FA(3-1), CNTsg-FA  and CNTs-g-FA(1-3), whereas the fabricated composite PCMs were named as FA/CNTs-g-FA1, FA/ CNTs-g-FA2 and FA/CNTs-g-FA3, respectively. Moreover, the names of the composite samples doped with nongrafted CNTs were shortened as FA/CNTs. The selected FAs were doped with the same amount (5 wt%) of CNTsg-FA and CNTs additives, separately.

Dispersion stability experiment.
To observe the dispersion behavior of the grafted and non-grafted CNTs in FA, the melting/blending method was conducted. As a representative example, the dispersion of CNTsg-PA and CNTs in PA was carried out separately. The PA sample at specific amount was melted in two glass tubes at 70 °C. The CNTs-g-PA and CNTs powders were added in weight fractions of 5 wt% to the liquid PA samples. The obtained suspensions were centrifuged at 100 rpm for 20 min to achieve uniform dispersion. Afterwards, PA/CNTs-g-PA and PA/CNTs samples were kept at 70 °C for 30 min, and studied the dispersion stability.
characterization methods. The chemical and crystalline structures of the synthesized CNTs-g-FA, FA/ CNTs-g-FA and FA/CNTs samples were studied by Fourier-transform infrared (FTIR; Perkin Elmer, 16F PC) and X-ray powder diffraction (XRD) instrument (PANalytical X'-Pert3 model). The FTIR spectra were obtained in wavelength range of 4,000-400 cm −1 . The crystalline structures of the samples were identified using Cu (Kα = 1.5406 Å) irradiation at a step size of 0.0131°. The morphological structures of CNTs-g-FA, FA/CNTsg-FA, and FA/CNTs samples were investigated using a scanning electron microscope (SEM; LEO 440 model).  www.nature.com/scientificreports/ The TES properties of the synthesized CNTs-g-FA, FA/CNTs-g-FA and FA/CNTs samples were measured by differential scanning calorimetry (DSC; Hitachi-7020) at a heating/cooling rate of 3 °C min −1 . The analyses were repeated for three times to minimize the data error. The average deviation in phase change temperature and latent heat capacity (LHC) measurements was determined as ± 0.11 °C and ± 1.15%, respectively. Thermal degradation temperatures of the prepared CNTs-g-FA, FA/CNTs-g-FA and FA/CNTs samples were determined by thermogravimetric analysis (TGA) technique (Perkin-Elmer). The analysis was performed at heating ramp of 10 °C min −1 in temperature range of 25-580 °C. The degradation temperature data were determined with average discrepancy of ± 1.08 °C. To study the influence of the thermal cycling process on the chemical stability and phase change properties of the prepared FA/CNTs-g-FA and FA/CNTs composites, an accelerated heating-cooling process was adapted to the samples using a thermal cycler (Model: BIOER TC-25/H). This treatment was continued until the cycling number reached to 2000. After cycling procedure, FTIR, and DSC analyses were repeated on the studied samples. Thermal conductivity (TC) of CNTs-g-FA, FA/CNTs-g-FA and FA/CNTs samples as well as pure CA, PA and SA were measured at 20 ± 1 °C using a thermal property analyzer (Decagon KD2 pro Devices, Inc., USA). The measurements were repeated three times and the maximum discrepancy was determined as ± 0.02 W m −1 K.

Results and discussion
Microstructural and dispersion stability study. Surface morphology of the CNTs-g-FA, FA/CNTsg-FA and FA/CNTs composites were studied by SEM analysis. The obtained SEM microstructures are shown in Fig. 2. It is clear from the images (Fig. 2) that the CNTs-g-FA are uniformly distributed within the FA/CNTs-g-FA composites structure without agglomeration. Moreover, the nanotube phases are more visible on the surface of FA/CNTs composites. It can be assumed that this uniform distribution of the CNTs (both chemically grafted and pure) within the composites creates effective heat conduction path. Figure 3 exhibits the photographic image of the dispersion study of the grafted and non-grafted CNTs in melted PA as a representative example of other composites. This photograph was taken for PA/CNTs-g-PA (5 wt%) and PA/CNTs (5 wt%) composites samples at 70 °C after 30 min-standby period. As evidently perceived (Fig. 3), some CNTs precipitation was occurred in PA/CNTs sample while the PA/CNTs-g-SA (5 wt%) had a stable suspension. This result suggested that the grafted CNTs provide better dispersion stability, as compare to that of non-grafted CNTs. It can be directed to the comparatively stronger colloidal interaction between the CNTs-g-FA and FA. The similar observation was reported for acid modified CNTs/erythritol suspension 21 and CNTs-g-polyalcohol/paraffin suspensions 22 . chemical and crystalline characterization. The FTIR results of the carboxylic acid functionalized CNTs (CNTs-COOH), acyl chloride functionalized CNTs (CNTs-COCl) and fatty acid functionalized CNTs (CNTs-g-FA) are shown in Fig. 4. In the spectrum of CNTs-COOH, the wide peak at 3250-3650 cm −1 , signified the vibration band of -OH groups 24 . The peak at 1690 cm −1 is associated with the stretching band of C=O 25 as the peaks at 958 cm −1 and 829 cm −1 are linked with the bending vibration band of -OH groups 26 . In the spectrum of CNTs-COCl, the peak at 1776 cm −1 is associated with the stretching band of C=O group. The peaks at 1328 cm −1 and 639 cm −1 are due the C-O and C-Cl groups of acyl chloride functionalized CNTs, respectively.
On the other hand, in the spectra of CNTs-g-FA compounds synthesized at three different grafting ratios, the peaks observed in the range of about 2840-2918 cm −1 are characterized as the stretching band of -CH 2 and -CH 3 groups. The C=O and C-O stretching bands were shifted to the range of 1295-1296 cm −1 and 1690-1700 cm −1 after grafting reaction. The intensity of -OH absorption observed at 3400-3650 cm −1 was considerably reduced due to grafting reaction, although it was seen in spectrum of CNTs-g-SA3 and CNTs-g-PA3 synthesized at highest grafting mole ratio (FA/CNTs mole ratio: 3:1). The similar observations were reported for grafted lauric acid with graphene oxide 26 . Moreover, the disappearance of the C-Cl absorption band in all spectra of CNTs-g-FA compounds also confirmed that the grafting reaction between FA and CNTs-COCl was successfully carried out. As a whole, all of these results shown in Fig. 4 confirm that the FAs were successfully grafted to the CNTs. Figure 5 shows the FTIR spectra of the prepared FA/CNTs-g-FA and FA/CNTs composites. The peaks between 2943-2920 cm −1 and 2945-2918 cm −1 are associated with the stretching band of methyl and methylene groups of nine different FA/CNTs-g-FA composites and three different FA/CNTs composites, respectively. The peaks around 1687-1704 cm −1 and 1691-1697 cm −1 are related with their C=O stretching vibration bands. Moreover, the stretching and bending vibrations belong to C-O and C-H band are detected in the spectra of the composites. When compared to the spectra of CNTs-g-FA, the FTIR spectra of the composites does not show any significant shift of the respective key peak position. This result proved that the presence the good chemical compatibility among FAs and CNTs-g-FA or CNTs.
On the other hand, the XRD analysis of the prepared FA/CNTs-g-FA and FA/CNTs composites were carried out to examine the influence of added CNTs-g-FA and CNTs on the crystal structures of SA, PA and CA. Figure 6 shows the XRD diffraction patterns of CNTs-g-CA, CNTs-g-PA and CNTs-g-SA samples. While, the XRD results of CA/CNTs-g-CA, PA/CNTs-g-PA, SA/CNTs-g-SA and CA/CNTs, PA/CNTs and SA/CNTs composites were presented in Fig. 7. The CA, PA and SA grafted CNTs has typical crystalline peaks in 2θ range of 19.63°-43.67°, 7.31°-27.68° and 9.68°-43.57°, respectively. The characteristic peaks were detected at 7.58°-23.38° for CA/CNTs-g-CA, 7.39°-24.0° for PA/CNTs-g-PA, 6.59°-24.08° for SA/CNTs-g-SA while they are observed at 7.61°-23.42° for CA/CNTs, 7.37°-23.95° for PA/CNTs and 6.62°-23.94° for SA/CNTs composites. Moreover, it is hard to notice the distinctive crystalline peaks of grafted or non-grafted CNTs because their amounts are quite low in the composites 27   In addition, the melting temperatures of the composites including 5 wt% grafted compounds were measured in the range of 28.85 ± 0.10-29.43 ± 0.13 °C for CA/CNTs-g-CA, 59.42 ± 0.10-60.0 ± 0.10 °C for PA/CNTs-g-PA and 63.82 ± 0.11-64.90 ± 0.10 °C for SA/CNTs-g-SA composite. The significant decrease in the melting temperatures of the composite PCMs in comparison with those of pure FAs could be due to the dispersion of CNTs grafted with long-chain FA molecules among the FA crystals. Moreover, the melting temperature was determined as 28.07 ± 0.10 °C for CNTs/CA, 59.70 ± 0.12 °C for CNTs/PA and 66.5 ± 0.11 °C for CNTs/SA. The phase changes temperatures of the composites including non-grafted CNTs are relatively closer to those of the pure FAs given above. This means that the physical attractions (e.g., capillary and surface tension forces) between the FA and grafted CNTs are more effective compared to the attractions between FA and FAg-CNTs. Again, depending on the increase in the grafting ratio of FA, the melting LHCs of the composites were determined, which was in the range of 174-184 J g −1 for CA/CNTs-g-CA, 214-230 J g −1 for PA/CNTs-g-PA and 235-257 J g −1 for SA/CNTs-g-SA. Whereas the melting LHC of CNTs/CA, CNTs/PA and CNTs/SA was found to be about 171, 210 and 234 J g −1 , respectively. As seen from these data, especially the LHC values of the composites including CNTs-g-FA (grafting ratio: 1:3) were 6.5-8.9% higher than that of the composites including CNTs due to the LHC contribution of the CNTs-g-FA additive. On the other hand, when compared the LHC of the FA/CNTs-g-FA composite PCMs with those of different kinds of organic PCMs doped with carbon based fillers [14][15][16][17][18][19][20]22,26,29,36 , it was observed that the FA/CNTs-g-FA composite PCMs prepared in this work had relatively higher LHC than most of them. This boosted in the LHC value of the FA/CNTs-FA composite PCMs was due to the fact that the CNTs-g-FA used as doping agents had LHC of in the range of 25-93 J g −1 .
cycling phase change reliability and chemical stability of the composite pcMs. The cycling phase change reliability is one of the major characteristic preference of a PCM. Therefore, it is desired that the PCM should have stable phase change behavior after considerable number of heating-cooling cycles 24,28 . Based on this criterion, a cycling test including consecutive cycling process repeated for 2000 times was conducted with the FA/CNTs-g-FA (5 wt%) and CNTs/FA (5 wt%) composites. Figure 10 shows the DSC results of the samples after the 2,000 cycles, and all LHTES data are summarized in Table 3. By considering the phase change temperature data, it was realized that there was a slight reduction in phase change temperature. The changing quantity was calculated between − 1.22 °C and 0.40 °C for melting temperature and between − 1.20 °C and 0.93 °C for solidification temperature of FA/CNTs-g-FA composites. As for the CNTs/FA composites it was determined in the range of (− 0.11) to (− 0.70) °C and (− 0.90) to (+ 0.93) °C for melting and solidification period . On the   www.nature.com/scientificreports/ period and 7.7-13.6% for freezing period of the CNTs/FA composite. By overall evaluation of these results, it can be concluded that the fluctuations in the TES values of all composites were in tolerable range and they had appreciable phase change stability even after 2,000 cycles. Figure 11 compares the FTIR spectra of FA/CNTsg-FA (5 wt%) and CNTs/FA (5 wt%) composites before and after the 2,000 cycles. There was no significant shift of peaks were observed in the wavenumber characteristic absorption peaks of all the composites. Additionally, no extra band was observed after the cycling study. This indicates that all the prepared composite PCMs had adequate chemical stability.
thermal degradation stability of the composite pcMs. Thermal degradation stability is one of the major thermal characteristics considered for any PCM for its suitability in a TES system. With this sense, the thermal degradation stability of the prepared composites was studied and the TGA curves were presented in Fig. 12 and Table 4. As clearly observed from the thermal curves, the prepared CNTs-g-FA, FA/CNTs-g-FA and FA/CNTs composites showed one-step thermal degradation behavior. Moreover, the grafting ratio of FA was affected thermal degradation temperature range of the final product, CNTs-g-FA. These data indicated that the thermal decomposition temperatures of the FA/CNTs samples were extended slightly. It could be due to the covalent functionalization of CNTs with FA and more effective chemical interface of FA with CNTs-g-FA 22 . Consequently, the prepared composite PCMs had adequately thermal degradation stability because the decomposition temperature of each composite PCM is much higher than its own phase change temperature value.
tc of the prepared pcMs. Thermal conductivity (TC) is one of the essential properties for a PCM. The higher the thermal conductivity of the PCM, the faster will be the rate of heat loading or releasing from the TES system. Based on this impression, this study is also focused on the TC enhancement of the FAs by adding CNTsg-FA rather than only CNTs. Table 5 presents the measured TC values of CNTs-g-FA, FA/CNTs-g-FA and FA/ CNTs as well as pure FAs. As noticed from the tabulated data ( .4% compared to that of pure CA, PA and SA, respectively. In comparison to pure FAs or FA/CNTs composites, a higher degree of TC enhancement was observed from the FA/CNTs-g-FA composite PCMs, which was attributed to their increased interfacial compatibility/or interfacial contact area between the grafted CNTs and FA 23,29 . Relatively better miscibility and affinity between the covalent functionalized CNTs and FAs allowed improving interfacial thermal transport and thus remarkably enriching their TC values 22,26,30,31 . Accordingly,  www.nature.com/scientificreports/  www.nature.com/scientificreports/ CNTs-g-FA acted as good filler to increase TC of the FAs by reducing interfacial thermal resistance 32,33 . Similar phenomenon was observed for the erythritol as PCM by adding acid-functionalized CNTs 21 . Additionally, the effect of the thermal cycling operation on the TC property of the produced FA/CNTs-g-FA and FA/CNTs composites were also investigated. As seen from the tabulated data in Table 5, the TC values was not changed in the most of composite samples, while it was slightly boosted for few samples, mainly due to slight improvement in the dispersion stability after the thermal cycling. On the other hand, the measured TC values of FA/CNTs-g-FA and FA/CNTs composites were compared with those of some PCMs doped with different kind of carbon basedfillers in Table 6. As seen from the tabulated data, the CPCM prepared in this work had much superior thermal properties than the previously reported similar PCMs 14,20,22,26,[34][35][36] .

conclusions
In the present study, CNTs was functionalized and three FAs (CA, PA and SA) were grafted separately at three different ratios. FTIR, SEM and XRD techniques were used to investigate the grafting reactions, the surface modification of CNTs, compatibility and crystallites of the components. The synthesized materials were then used as additive in a mass fraction of 5 wt% to prepare respective composites to achieve good LHC and at the same time to boost the TC of the PCMs. The TES properties, thermal degradation temperatures and TC values of all prepared composite PCMs were measured. Moreover, their cycling phase change reliability and chemical structure stability were studied. The chemical structure and crystallites of the CNTs-g-FA, FA/CNTs-g-FA and FA/CNTs were confirmed by FTIR and XRD, and the surface morphologies of the different composite PCMs were evaluated by SEM. The melting temperatures of the composites containing 5 wt% non-grafted and FA grafted CNT additives were measured in the range of about 28-67 °C and 29-65 °C, respectively. While their corresponding LHC values of about 171-234 J g −1 and 183-257 J g −1 , respectively. After 2000, cycles, the melting temperature of FA/CNTs-g-FA composites was changed between − 1.22 °C and 0.40 °C, as it was determined in the range of (− 0.11) to (− 0.70) °C and CNTs/FA composites. The cycling results also revealed that all composite PCMs had appreciable phase change stability and good chemical stability even after 2,000 cycles. The composites including grafted CNTs and non-grafted CNTs additives were thermally degraded in temperature interval of around 102-280 °C and 108-309 °C, respectively. These data proved that the composites had satisfactory thermal stability.  www.nature.com/scientificreports/ www.nature.com/scientificreports/