Low-thermal-conductivity nitrogen-doped graphene aerogels for thermal insulation

Abstract

Aerogels such as SiO₂ aerogels, Al₂O₃ aerogels and carbon aerogels have been widely used in thermal insulation. However, graphene aerogels (or reduced graphene oxide aerogels), which have similar structures, have never been used in this field. In this paper, the concept of suppressing the thermal conductivities of graphene aerogels by introducing defects or doping atoms in graphene was introduced. Nitrogen-doped (N-doped) graphene aerogels with low thermal conductivity were prepared with paraphenylene diamine as a bridging and doping agent by CO₂ supercritical drying. With the introduction of doping atoms and the bridging agent, the solid thermal conductivity is depressed. Also, with CO₂ supercritical drying, the pore size is reduced, and the gaseous thermal conductivity is suppressed. The lowest thermal conductivity of N-doped graphene aerogels is 0.023 W (m⁻¹ K⁻¹), which is nearly 1/2 of the lowest reported value and even lower than that of static air. Meanwhile, the thermal insulation mechanisms were also studied. The low thermal conductivity and low bulk density make N-doped graphene aerogels a potentially useful thermal insulation material that may significantly lighten thermal insulation systems.

---

1. Introduction

Due to their nanoporous structures, high porosities and low densities,¹ aerogels have been widely used in thermal insulation.² Their nanoporous structures suppress gaseous thermal conductivity,³ and their low densities depress solid thermal conductivity.⁴ Examples include SiO₂ aerogels,² Al₂O₃ aerogels⁵ and carbon aerogels.⁶ Compared with these aerogels, graphene aerogels have similar structure, lower density, higher strength and perfect opacity. The opacity can lower radiant thermal conductivity and hence may further reduce thermal conductivity. Despite these promising characteristics, graphene aerogels have never been used in thermal insulation. The most critical reason for this is likely the super-high thermal conductivity of graphene.

Graphene is a single atomic layer of sp² carbon atoms.^7,8 Since it was first obtained by mechanical exfoliation,⁶ it has received considerable attention duce to its excellent electrical,⁹ thermal,¹⁰ optical¹¹ and mechanical¹² properties. Its measured values of thermal conductivity at room temperature for suspended samples are as high as 2500–5300 W (m⁻¹ K⁻¹).^13,14

However, defects and doping atoms in graphene may significantly impact its thermal conductivity.^15,16 Several groups have studied the effects of different kinds of defects (vacancies^17–19 and doping^20,21) on its thermal conductivity by computer simulation using the force-constant method,²² molecular dynamics²³ and Green's function method.¹⁷ These studies revealed that the thermal conductivity of graphene sharply decreases with increasing amount of defects or doping atoms. Taking the fact that the thermal conductivity of graphene oxide (3.19 W (m⁻¹ K⁻¹)) is much lower than that of graphene (5300 W (m⁻¹ K⁻¹)) into consideration, the thermal conductivity of graphene can be decreased by introducing a large amount of defects or doping atoms.

Zhong²⁴ reported the thermal conductivities of graphene aerogels (reduced graphene oxide (rGO) aerogels) prepared by hydrothermal reduction and freeze-drying for the first time. The thermal conductivity measured using the laser flash technique is as high as 2.183 W (m⁻¹ K⁻¹). The bulk density and surface area are 227 mg cm⁻³ and 43 m² g⁻¹, respectively. Fan^25,26 studied the effects of thermal treatment on the thermal conductivities of graphene aerogels (rGO aerogels) using infrared microscopy. The aerogels were prepared by CO₂ supercritical drying, which reduces their pore size and hence decreases the gaseous thermal conductivity. However, the thermal conductivities of the graphene aerogels before and after thermal treatment at 450 °C for 5 h are still 0.12–0.36 W (m⁻¹ K⁻¹) and 0.18–0.31 W (m⁻¹ K⁻¹), respectively, and the corresponding bulk densities are 14.1–52.4 mg cm⁻³ and 16.4–49.0 mg cm⁻³, respectively. Tang²⁷ synthesized graphene aerogels (rGO aerogels) with paraphenylene diamine (PPD) as a reducing and functionalizing agent in the presence of ammonia (NH₃·H₂O) by freeze-drying and characterized their thermal conductivities using a Hot Disk Techmax TPS1500 thermal meter. The thermal conductivity ranges from 0.040–0.053 W (m⁻¹ K⁻¹) for bulk densities in the range of 1.8 to 27.2 mg cm⁻³. Due to the use of freeze-drying, the pore size of the aerogels is relatively large. Thus, the thermal conductivity can be further depressed by reducing the pore size.

In this paper, the concept of decreasing the thermal conductivities of graphene aerogels by introducing defects and/or doping atoms to graphene was introduced. By introducing doping atoms together with reducing the pore size by drying with supercritical CO₂, N-doped graphene aerogels with low thermal conductivity were obtained. These aerogels are potentially useful in thermal insulation due to their low thermal conductivity and low bulk density.

2. Experimental section

2.1 Materials

Graphene oxide (GO) aqueous suspension (15 mg ml⁻¹) was purchased from Nanjing Jicang nanotechnology Co. Ltd. A GO suspension with another concentration was obtained by diluting the aqueous suspension with deionized water. PPD was purchased from Sinopharm Chemical Reagent Co. Ltd. Ethanol and concentrated NH₃·H₂O were obtained from Hunan Hengyang Kaixin Chemical Co. Ltd. All the chemicals were used as received without further purification.

2.2 Synthesis of graphene aerogels

A 100 ml GO suspension (3, 6, 9, 12, 15 mg ml⁻¹; 12 mg ml⁻¹ for a typical procedure) was commixed with X g PPD and 2.7 ml concentrated NH₃·H₂O (X = 1.2, 2.4, 3.6, 4.8, and 6.0, respectively; for a typical procedure, X = 4.8). Subsequently, the mixture was heated at 90 °C for 8 h to obtain N-doped graphene hydrogels. The solvent in the hydrogel was then exchanged five times with water and ethanol. Finally, the gel was dried with supercritical CO₂ to form N-doped graphene aerogels.

2.3 Characterization

The bulk density was calculated from the mass and corresponding volume. The microstructure was observed with field-emission scanning electron microscopy (SEM; Hitachi S4800). The surface composition was characterized using X-ray photoelectron spectroscopy (XPS; TESCALAB 250Xi). Nitrogen adsorption–desorption tests were carried out at 77 K in an autosorb-1 physical adsorption instrument after 16 h of sample outgassing in vacuum, and the specific surface area and pore size distribution were calculated using the Brunauer–Emmett–Teller (BET) and Barrett–Joyner–Halenda (BJH) methods, respectively. The thermal conductivity was measured with samples with dimensions of Φ 39 ∼ 42 × 15 mm using a Hot Disk TPS2500 apparatus with a 5465 sensor.

3. Results and discussion

During heat preservation, N-doped graphene hydrogels are obtained by PPD grafting with GO sheets and π–π stacking between reduced GO sheets.²⁷ The GO sheets are simultaneously reduced and functionalized by PPD. From the XPS spectra (Fig. 1), N atoms and O atoms account for 13.34 at% and 9.98 at%, respectively, in the sample (Table 1), illustrating the introduction of PPD into the as-prepared N-doped graphene aerogels. According to the C 1s XPS spectra (Fig. 1(b)) of the as-prepared N-doped graphene aerogels, the peaks at 285.4 eV corresponding to C–N and 286.1 eV corresponding to C(O)N indicate the covalent bonding of PPD to the graphene sheets. This can be further demonstrated by the peaks at 399.1 eV and 400.4 eV in the N 1s XPS spectra (Fig. 1(c)) and the peak at 532.2 eV in the O 1s XPS spectra (Fig. 1(d)).

Fig. 1 (a) Broad XPS spectra, (b) C 1s spectra, (c) N 1s spectra and (d) O 1s spectra of as-prepared N-doped graphene aerogels.

Table 1 Elemental percentages of as-prepared N-doped graphene aerogels

Element	C	O	N
Content (at%)	76.68	9.98	13.34

---

However, in contrast to freeze drying, the porous network is fully preserved after CO₂ supercritical drying. Fig. 2 shows SEM micrographs of the as-prepared N-doped graphene aerogels. As shown in the images, the N-doped graphene sheets are randomly interconnected with each other, forming pores with sizes of several dozens or hundreds of nanometers. The pore size is much smaller than that of N-doped graphene aerogels prepared by freeze drying (Fig. 3). The microstructures of the as-prepared N-doped graphene aerogels are homogeneous to some extent.

Fig. 2 SEM micrographs of as-prepared N-doped graphene aerogels.

Fig. 3 SEM micrographs of as-prepared N-doped graphene aerogels prepared by freeze drying.

The N₂ adsorption–desorption isotherm and BJH pore size distribution of the as-prepared N-doped graphene aerogels with a bulk density of 34.5 mg cm⁻³ are shown in Fig. 4. The isotherm is a typical type III adsorption–desorption isotherm,²⁸ revealing plenty of pores with sizes of more than 5 nm and a wide range of pore sizes. This result is consistent with the SEM results. Also, the adsorption capacity is large, reflecting the large pore volume. According to the BJH pore size distribution, most of the pore diameters are larger than 10 nm, with a maximum diameter of 49 nm. This further verifies the SEM results. The pore volume decreases as the pore diameter increases to 140 nm, but the number of pore with sizes exceeding 140 nm is still large. These pores cannot be studied by N₂ adsorption–desorption but can be observed by SEM. According to the bulk densities of the as-prepared N-doped graphene aerogels (11.1–35.0 mg cm⁻³, Table 2), the pore volume may be 28–89 cm³ g⁻¹, compared with 3.26–5.88 cm³ g⁻¹ determined by the BJH method based on N₂ adsorption–desorption isotherms.

Fig. 4 N₂ adsorption–desorption isotherms (left) and BJH pore size distribution (right) of as-prepared N-doped graphene aerogels.

Table 2 Textural properties of the as-prepared N-doped graphene aerogels with different bulk densities

GO concentration (mg ml^−1)	Bulk density (mg cm^−3)	BET surface area (m^2 g^−1)	Pore volume (cm^3 g^−1)
3	11.1 ± 0.7	568.7	3.26
6	19.9 ± 0.6	705.5	4.93
9	24.7 ± 1.1	802.4	4.90
12	34.5 ± 1.0	891.7	5.88
15	35.0 ± 1.1	751.7	4.15

---

At the same time, a shrinkage of 1–8% in sample diameter was observed with the variation of GO concentration during CO₂ supercritical drying. This may be attributed to the surface tension during drying, although it is relatively weak due to most of the resolve in the gels have been replaced by the supercritical CO₂. However, the shrinkage does not impact the structures and appearances of the as-prepared N-doped graphene aerogels, and cracks are not observed in the photograph (Fig. 5) or micrograph (Fig. 2).

Fig. 5 Photograph of as-prepared N-doped graphene aerogel.

Fig. 6 shows the thermal conductivity of the as-prepared N-doped graphene aerogels with different GO concentrations. As the GO concentration increases from 3 to 12 mg ml⁻¹, the thermal conductivity of the as-prepared N-doped graphene aerogels at room temperature (25 °C) and atmospheric pressure (10⁵ Pa) decreases. However, when the GO concentration increases further to 15 mg ml⁻¹, the thermal conductivity increases slightly. At a GO concentration of 12 mg ml⁻¹, the thermal conductivity is only 0.023 W (m⁻¹ K⁻¹). The thermal conductivity at room temperature and 5 Pa shows a different trend. The thermal conductivity increases almost linearly with increasing GO concentration. Increasing the GO concentration increases the bulk density of the as-prepared N-doped graphene aerogels. This reinforces the junctures of graphene sheets and magnifies the solid content in the unit volume. Since graphene sheets are the carriers of heat, the solid thermal conductivity increases. As the radiant thermal conductivity at this temperature and the gaseous thermal conductivity at 5 Pa are both negligible, the thermal conductivity at room temperature and 5 Pa primarily consists of solid thermal conductivity. Thus, the thermal conductivity of the as-prepared N-doped graphene aerogels at 5 Pa increases with increasing GO concentration. In contrast, the difference between the thermal conductivity at atmospheric pressure and that at 5 Pa decreases. The difference matches very well with the inverse S-curve. The reason for this is that the difference is mainly attributed to gaseous thermal conductivity. The enhancement of bulk density caused by the increase in GO concentration decreases the pore size of as-prepared N-doped graphene aerogels (Fig. S11† and 2). Hence, the gaseous thermal conductivity was restrained to a greater extent. When the GO concentration is relatively low (<6 mg ml⁻¹), the pore size is relatively large, and the gaseous thermal conductivity is not significantly restrained. Meanwhile, when the GO concentration is relatively high (>12 mg ml⁻¹), the effect of the GO concentration on bulk density diminishes. Hence, the relationship between GO concentration and gaseous thermal conductivity is weaker. Thus, as the GO concentration increases, the difference (namely, gaseous thermal conductivity) shows an inverse S-curve. Logically, the linear increase in the thermal conductivity at 5 Pa and the nonlinear decrease in the difference (the gaseous thermal conductivity) induces the variation in the thermal conductivity of the as-prepared N-doped graphene aerogels at atmospheric pressure.

Fig. 6 Thermal conductivity of as-prepared N-doped graphene aerogels.

Compared with the values reported in previous works (Table 3), the thermal conductivity of the as-prepared N-doped graphene aerogels is significantly lower (almost half of the lowest value ever reported). This result may be attributed to four factors.

Table 3 Thermal conductivity of rGO aerogels

Author	Bulk density (mg cm^−3)	BET surface area (m^2 g^−1)	Thermal conductivity (W (m^−1 K^−1))	Reference
Yajuan Zhong	227	43	2.183	24
Zeng Fan	14.1–52.4	—	0.12–0.36	25 and 26
Gongqing Tang	1.8–27.2	—	0.040–0.053	27
Chenwu Yue	11.1–35.0	568–892	0.023–0.026	This work

---

First, as reported by Fan et al.,²⁶ the defects in graphene and the relatively small sizes of graphene sheets limit the transmission of heat and thus reduce the thermal conductivity of the as-prepared N-doped graphene aerogels to some extent.

Second, drying with supercritical CO₂ sharply diminishes the pore size in as-prepared N-doped graphene aerogels. According to the computation using the BJH method based on the N₂ adsorption–desorption curve, the pore size distribution centers on 49 nm. This is shorter than the mean free path of the molecule (∼70 nm³) at room temperature and atmospheric pressure and hence can obviously depress the gaseous thermal conductivity. The work of Feng et al.²⁹ showed that the gaseous thermal conductivity will be critically suppressed only when the pore diameter is smaller than two-times of the molecule's mean free path (namely, 140 nm at room temperature and atmospheric pressure). Thus, the gaseous thermal conductivity is reduced. In addition, the two-dimensional structure of graphene sheets may effectively block the transfer of gas molecules in wall-like fashion compared to other aerogels (e.g., SiO₂ aerogels, Al₂O₃ aerogels, and carbon aerogels) with three-dimensional open-cell network structures (Fig. 7). Hence, graphene aerogels can greatly suppress the gaseous heat transmission through gas molecular collisions. Through these two mechanisms, the gaseous thermal conductivity of the as-prepared N-doped graphene aerogels is significantly depressed.

Fig. 7 Schematic diagram of an aerogel with an open-cell network structure (left) and the as-prepared N-doped graphene aerogels (right).

Third, the introduction of doping atoms transforms sp² carbon atoms into sp³ carbon atoms, destroying the perfect π-electron conjugated structure and crystal structure, which makes up the main path of heat transmission through graphene. Phonons, the main carrier of heat delivery in graphene, transfer heat by lattice vibration. With the introduction of doping atoms, the crystal structure of doped graphene is divided into smaller pieces by sp³ carbon atoms (namely, carbon chemically linked with doping atoms (Fig. 1(b), the peaks at 285.4 and 286.1 eV)). The scattering of phonons by these obstacles together with the defects and edges can significantly diminish the amount of heat passing through the doped graphene (Fig. 8). Meanwhile, electrons also transfer heat in graphene via electron travelling. With the introduction of doping atoms, the intactness of the π-electron conjugated structure is broken, boosting the resistance of electron motion. As a result, the heat transferred through electron travelling is reduced. In this way, the solid thermal conductivity of as-prepared N-doped graphene aerogels is further decreased. Fig. 9 shows the thermal conductivity (5 Pa) of the as-prepared N-doped graphene aerogels (GO concentration = 12 mg ml⁻¹) with different masses of PPD. When the mass of PPD is no more than 4.8 g, the thermal conductivity at 5 Pa decreases with increasing PPD mass. Since increasing PPD mass increases the amount of N doped into the sample, the as-prepared N-doped graphene aerogels with more PPD show better thermal insulation performances at 5 Pa.

Fig. 8 Schematic diagram of heat passing through N-doped graphene.

Fig. 9 Thermal conductivity (5 Pa) of as-prepared N-doped graphene aerogels with different masses of PPD.

Finally, PPD also acts like a bridging agent. It links doped graphene sheets with each other. With the introduction of a bridging agent, the heat transferred between the graphene is significantly limited. The bridging agent crucially restricts the heat flux in a bottleneck-like fashion. Therefore, the heat passing through the as-prepared N-doped graphene aerogels is reduced, and the solid thermal conductivity decreases. According to Fig. 9, the increase in thermal conductivity (5 Pa) at a PPD mass of 6.0 g illustrates the increase in the number of ‘bridges’ accelerating the heat transfer to some extent.

Due to the above four reasons, the thermal conductivity of the as-prepared N-doped graphene aerogels is low, even lower than that of static air (0.026 W (m⁻¹ K⁻¹)).³⁰ The low thermal conductivity (0.023–0.026 W (m⁻¹ K⁻¹)) and low bulk density (11.1–35.0 mg cm⁻³) are crucially important for the use of N-doped graphene aerogels in thermal insulation. The use of this material may significantly lighten the weights of thermal insulation systems. Meanwhile, the gaseous thermal conductivity of N-doped graphene aerogels still makes up a large proportion of the thermal conductivity. This may be because the percentage of pores with relatively large sizes (>140 nm) is still considerable. Thus, the thermal conductivity of N-doped graphene aerogels may be further depressed by decreasing the pore size.

4. Conclusion

We introduced the concept of lowering the thermal conductivity of graphene aerogels by introducing defects or doping atoms in graphene and prepared low-thermal-conductivity N-doped graphene aerogels using PPD as a bridging and doping agent by CO₂ supercritical drying. The PPD is anchored in the graphene sheets during reaction. By CO₂ supercritical drying, the pore sizes of the as-prepared N-doped graphene aerogels are much smaller than those obtained by freeze drying. Remarkably, the lowest thermal conductivity of the as-prepared N-doped graphene aerogels is only 0.023 W (m⁻¹ K⁻¹), even lower than that of static air, demonstrating its good thermal insulation characteristics. These results are explained as follows. (1) The transmission of heat is limited by the defects in graphene and the relatively small sizes of the graphene sheets; (2) the relatively small pore sizes and two-dimensional structures of the graphene sheets suppress the gaseous thermal conductivity; (3) the introduction of doping atoms boosts the scattering of phonons, greatly depressing the solid thermal conductivity; and (4) the bridging agent restricts the heat flux in a bottleneck-like manner, further lowering the thermal conductivity. With a low bulk density (11.1–35.0 mg cm⁻³) and low thermal conductivity (0.023–0.026 W (m⁻¹ K⁻¹)), the as-prepared N-doped graphene aerogels are potentially useful in thermal insulation and may significantly lighten the weights of thermal insulation systems.

Acknowledgements

The authors are thankful for the support from the National Natural Science Foundation (51172279, 51302317) of China, Aid Program for Science and Technology Innovative Research Team in Higher Educational Institutions of Hunan Province and Aid Program for Innovative Group of National University of Defense Technology.

References

1. N. Hüsing and U. Schubert, Angew. Chem., Int. Ed., 1998, 37, 22–45.
2. J. Feng, Q. Gao, J. Feng and Y. Jiang, J. Natl. Univ. Def. Technol., 2010, 32, 40–44.
3. X. Lu, M. C. Arduini-Schuster and J. Kuhn, Science, 1992, 255, 971–972.
4. S. N. Schiffres, K. H. Kim, L. Hu, A. J. H. McGaughey, M. F. Islam and J. A. Malen, Adv. Funct. Mater., 2012, 22, 5251–5258.
5. Q. Gao, C. Zhang, J. Feng, W. Wu, J. Feng and Y. Jiang, Chin. J. Inorg. Chem., 2008, 24, 1456–1460.
6. J. Feng, J. Feng and C. Zhang, J. Porous Mater., 2012, 19, 551–556.
7. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666–669.
8. D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Sun, A. Slesarev, B. A. Lawrence, L. Wei and M. T. James, ACS Nano, 2010, 8, 4806–4814.
9. D. Basu, M. Gilbert, L. Register, S. Banerjee and A. MacDonald, Appl. Phys. Lett., 2008, 92, 042114.
10. A. A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao and C. N. Lau, Nano Lett., 2008, 8, 902–907.
11. M. Liu, X. Yin, E. Ulin-Avila, B. Geng, T. Zentgraf, L. Ju, F. Wang and X. Zhang, Nature, 2011, 474, 64–67.
12. A. K. Geim, Science, 2009, 324, 1530–1534.
13. S. Ghosh, W. Bao, D. L. Nika, S. Subrina, E. P. Pokatilov, C. N. Lau and A. A. Balandin, Nat. Mater., 2010, 9, 555–558.
14. W. Cai, A. L. Moore, Y. Zhu, X. Li, S. Chen, L. Shi and R. S. Ruoff, Nano Lett., 2010, 10, 1645–1651.
15. P. Yang, X. Wang, P. Li, H. Wang, L. Zhang and F. Xie, Acta Phys. Sin., 2012, 61, 076501.
16. W. Yu, H. Xie and X. Wang, J. Eng. Thermophys., 2012, 33, 1609–1611.
17. H. Zhang, L. Geunsik and C. Kyeongjae, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84, 115460.
18. H. Justin, K. Alper, S. Cem, S. H^aldun, C. Gianaurelio and C. Tahir, ACS Nano, 2011, 5, 3779–3787.
19. Y. Y. Zhang, Y. Cheng, Q. X. Pei, C. M. Wang and Y. Xiang, Phys. Lett. A, 2012, 376, 3668–3672.
20. L. Wang and H. Sun, J. Mol. Model., 2012, 18, 4811–4818.
21. Z. Hui, P. He, Y. Dai and A. Wu, Acta Phys. Sin., 2014, 63, 074401.
22. K. Hossein, N. Neophytos, P. Mahdi and K. Hans, J. Comput. Electron., 2012, 11, 14–21.
23. N. Khosravian, M. K. Samani, G. C. Loh, G. C. K. Chen, D. Baillargeat and B. K. Tay, Comput. Mater. Sci., 2013, 79, 132–135.
24. Y. Zhong, M. Zhou, F. Huang, T. Lin and D. Wan, Sol. Energy Mater. Sol. Cells, 2013, 113, 195–200.
25. Z. Fan, D. Z. Y. Tng, C. X. T. Lim, P. Liu, S. T. Nguyen, P. Xiao, A. Marconnet, C. Y. H. Lim and H. M. Duong, Colloids Surf., A, 2014, 445, 48–53.
26. Z. Fan, A. Marconnet, S. T. Nguyen, C. Y. H. Lim and H. M. Duong, Int. J. Heat Mass Transfer, 2014, 76, 122–127.
27. G. Tang, Z. Jiang, X. Li, H. Zhang, A. Dasari and Z. Yu, Carbon, 2014, 77, 592–599.
28. F. Rouquerol, J. Rouquerol and K. Sing, Adsorption by powders and porous solid, principles, methodology and applications, Academic Press, 1999.
29. J. Feng, J. Feng, Y. Jiang and C. Zhang, Aerosp. Mater. Technol., 2012, 42, 1–6.
30. R. Singh and H. S. Kasana, Appl. Therm. Eng., 2004, 24, 1841–1849.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra23236h

---