epoxy composites for stimuli-responsive sensors and Joule heating deicers

Carbon 124 (2017) 296e307

Contents lists available at ScienceDirect

Carbon journal homepage: www.elsevier.com/locate/carbon

Mechanically robust and electrically conductive graphene-paper/ glass-fibers/epoxy composites for stimuli-responsive sensors and Joule heating deicers Qiangqiang Zhang a, b, *, Yikang Yu a, Kaichun Yang a, Baoqiang Zhang a, b, Keren Zhao a, b, Guoping Xiong c, Xingyi Zhang a, b, ** a

College of Civil Engineering and Mechanics, Lanzhou University, Lanzhou, 730000, PR China Key Laboratory of Mechanics on Disaster and Environment in Western China (Lanzhou University), The Ministry of Education of China, Lanzhou, 730000, PR China c Department of Mechanical Engineering, University of Nevada, Reno, NV 89557, United States b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 6 May 2017 Received in revised form 11 July 2017 Accepted 1 September 2017 Available online 4 September 2017

In this article, by composing three functional constituents into a multilayered structure with wellbonded interfaces, a conductive graphene-papers (GPs)/glass-fibers (GFs) reinforced epoxy composite (GPs-GE) with outstanding mechanical robustness, high electrical conductivity and sensitive stimuliresponsive performance is highlighted. Thereinto, GPs serves as a stimuli-responsive and Joule heating chip due to its superiorities on both of mechanical and electrical properties, while GFs possessing mechanically protective and strengthening functionality. The stimuli-responsive characterizations by electrical resistance change reveal the synchronous sensing properties of GPs-GE to different stimuli inputs such as mechanical deformation, temperature fluctuation and humidity. Subsequently, Joule heating and de-icing/anti-icing properties of GPs-GE are investigated symmetrically, indicating its large heating rate, high efficiency of energy conversion and low cost. Such superior performance of GPs-GE confirms that the design of multilayer microstructure to achieve multi-functionalization paves a novel way to scale-up fabrication of advanced graphene composites, indicating promising applications as mechanically reinforcing elements, stimuli-responsive sensors and electrical Joule heating chips in intelligent engineering monitoring and icing-induced disaster prevention at low-temperature environment. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction With today’s increasing demands to enable many multifunctional applications, such as smart sensing [1,2] and Joule heating effect [3e6] for intelligent monitoring as well as electro-thermal performance (e.g., anti-icing, fast clean-up of crude-oil and engineering mass heating) [7e9], there is an intriguing interest to seek smart and Joule heating nanomaterials with superiorities on mechanical, stimuli-responsive and electrical properties. Effective decreasing of the electrical resistance has been verified as one of

* Corresponding author. College of Civil Engineering and Mechanics, Lanzhou University, Lanzhou, 730000, PR China. ** Corresponding author. College of Civil Engineering and Mechanics, Lanzhou University, Lanzhou, 730000, PR China. E-mail addresses: [email protected] (Q. Zhang), [email protected] (X. Zhang). http://dx.doi.org/10.1016/j.carbon.2017.09.001 0008-6223/© 2017 Elsevier Ltd. All rights reserved.

the most critical routes to maximize power performance for electrical Joule heating and intelligent sensing capacity [2,5,7,8,10,11]. The state-of-the-art carbon nanostructures have been tremendously propagated in this field due to their unique physical properties, including high electrical conductivity, excellent mechanical robustness, and well manipulability for scalable fabrication, to name a few [12,13]. In general, nanocarbon assembled derivations offer novel kinds of smart and heating materials on the macroscopic scale through scale-up constructions. Particularly, paper-like graphene films (e.g., 2D scale-up monoliths of graphene sheets) are endowed with decent properties, such as mechanical flexibility, large electrical conductivity, sensitive stimuli-responsive performance and high Joule heating efficiency, suggesting great potential for intelligent sensor and electro-thermal applications. Currently, few studies on graphene monoliths or their related composites have been reported focusing on Joule heating performance, most of which attain large heating rate and maximum

Q. Zhang et al. / Carbon 124 (2017) 296e307

temperature with high power inputs. For instance, Shaffer et al. reported Joule heating studies of emulsion-templated graphene aerogel with a heating rate up to 0.08
C/s under a heating power of 1 W/cm3 [14]. Zhang et al. fabricated graphene aerogel-based conductive polymer composite with a 3D bulky configuration and demonstrated heating rate as high as 0.796
C/sec for power input of 12 W/cm3 by graphene content over 1 wt% [2]. Hu et al. designed micro-sized 3D printable graphene heaters (~200 mm) with fast rates up to ~20 000
C/sec under a power input of 5 W, and adopted carbon nanofiber-based micro-heaters to weld their microjunctions with temperature up to 2000
C at a heating rate of 200
C/min [15,16]. Yu et al. employed graphene-wrapped sponge for crude-oil clean-up due to Joule heating promoted adjustment of viscosity and fast absorption under applied power reaching up to 0.58 W/cm3 and corresponding temperature of 148
C within 250 s [7]. However, the Joule heating and good electrical conductivity derived deicing/anti-icing and stimuli-responsive applications have not been demonstrated extensively, which are essential for these graphene materials serving in engineering applications that require intelligent monitoring and disaster preventions. These reported Joule heating 3D graphene bulky materials have their inherent shortages in practical applications on some structural surfaces under extremely harsh serving conditions because of their poor resistance to shear stress, low strength, high brittleness, non-protective encapsulation and the difficulty of being malleable for surface covering. Therefore, 2D buckypaper-like graphene film can easily satisfy the proposed requirements for surface engineering with the coupling compliance, structural compatibility, mechanical robustness and flexibility after mechanical encapsulated protection and reinforcement by multilayered structures. Currently, 2D graphene films can be directly fabricated by chemical vapour deposition process [17], oxide paper reduction [18,19], directly aqueous dispersion [20], electrophoretic deposition [8], and roll-toroll producing strategy [21], to name a few. The structures of graphene films are designed by controlling stacking processes and assembling orientations with a multilayered micromorphology; therefore, these materials can recover from relatively large deformations with outstanding flexibility and mechanical robustness [22]. However, the intrinsic brittleness for anti-shear force among graphene sheet in nanoscale and thin thickness induced low strength make them difficult to defeat large tensile strain, which severely restrains their widespread applications as conductive Joule heating and intelligent sensing candidates unless the structural encapsulation and mechanical reinforcement are sufficiently implemented. Therefore, in order to not only enable multifunctionalizing combinations of high Joule heating performance, intelligent stimuli-responsive behavior, excellent mechanical properties and facile electrical/structural encapsulation, but also to ensure each unit functions independently without neighboring interference, new designs of graphene composites with a multilayered microstructure is warranted. In this article, a conductive graphene-papers (GPs)/glass-fibers (GFs) reinforced epoxy composite (GPs-GE) was fabricated by GPs content as low as 1 wt%. Due to the intrinsically excellent electrical properties of free-standing GPs, GPs-GE inherits the similarly high electrical conductivity (3.56  103 U/cm) from GPs and favorably shows a significant reinforcement on mechanical properties (e.g., Young’s modulus, flexure strength and fatigue resistance). The stimuli-responsive behaviors of this GPs-GE were evaluated under various stimuli-inputs, including static and dynamic periodic excitations, temperature fluctuation and humidity variation. And its outstanding Joule heating performance was verified by electrothermal characterizations under the influence of various factors such as heating power, thermal convection and surrounding temperature. Additionally, uniform distributions of temperature fields

297

were observed by isotropic heating mappings using infrared thermography. The Joule heating derived deicing/anti-icing performances of GPs-GE reveal large heating rate, high efficiency of energy conversion and competitive economic cost. 2. Experiment 2.1. Materials Graphene nanoplates with average lateral diameters ranging from 0.5 to 2 mm were commercially purchased from Nanjing Xianfeng Nanomaterials Tech. Co., LTD (China) to prepare graphene ink precursor. Both epoxy resin (Tyfo® S-T Epoxy A, 1.16 g/cm3) and amine type curing agent (Tyfo® S-T Epoxy B, 0.93 g/cm3) were supplied by Fyfe Co. LLC (USA) with the mixture production of 1.1 g/ cm3. Polyethylene glycol octylphenol ether (Triton X-100, Average Mw ~ 80,000) was supplied by Sigma-Aldrich-China to use as a nonionic surfactant in production of GPs, and highly conductive silver paint (~5  103 S/cm) was purchased from Beijing Emerging Tech. Co., LTD (China) to prepare electrodes. Commercially available glass fiber tapes (GFT) were purchase from Fyfe Co. LLC (USA) with an average area density of 920 g/m2. N-hexane and deionized water were all obtained from local suppliers (Lanzhou, China) and used as received. 2.2. Preparations of GPs and GPs-GE 2.2.1. GPs fabrication Graphene nanoplates, obtained by a physical method [23], are employed as basic building units to prepare free-standing GPs via a vacuum filtration assisted in situ assembling process [24]. And the strong p-p interaction among graphene nanoplates provides GPs with superior electrical properties, structural robustness and fatigue resistance. In detail, as schematically illustrated in Fig. 1a, 1 g graphene nanoplates were added in 5 g Triton X-100 and mechanically stirred for 30 min. The obtained ink-like mixture was diluted in 1 L deionized water (DI water) and then dispersed to be homogenous by ultrasonic treatment (JY96 IIN, Scientz, Ningbo, China) for 30 min with a power of 100 W. Subsequently, the asprepared graphene/Triton X-100 ink was filtrated through a microporous membrane (pore diameter ¼ 150 nm) to facilitate graphene nanoplates self-assembling in situ into a free-standing thin film by a thickness of 20 mm under vacuum conditions. To remove residual impurities (e.g., DI water and Triton X-100 solvent) and further strengthen stacking interfaces among graphene nanoplates, the as-prepared wet GPs were transferred away from the filter membrane to anneal at 120
C for 6 h. Finally, the obtained GPs exhibits the expected properties of high electrical conductivity, mechanical flexibility and structural robustness. Commercially purchased silver paste was then painted to fabricate electrodes with a size of 100  5  0.5 mm3 and then make electrical contact by attaching copper wires to electrodes. 2.2.2. GPs-GE fabrication As demonstrated in Fig. 1b, the GPs-GE composite was composed of GPs, GFT and epoxy matrix. The as-fabricated GPs as an embedded conductive chip was firstly sandwiched with GFT to form a multilayered structure. To aid the infiltration process, epoxy mixture was prepared by mixing epoxy base agent, curing agent and n-hexane solvent in the ratio of 10:0.266:1 by weight and then magnetically stirred for 30 min to be uniform and clear. The multilayered GPs/GFT composite slab was then immersed into the epoxy mixture under vacuum and ice-bath conditions for 6 h to facilitate the infiltration of epoxy with trapped bubbles releasing simultaneously. Because of the good fluidic characteristics of epoxy

298

Q. Zhang et al. / Carbon 124 (2017) 296e307

Fig. 1. Schematic illustration of the fabrication processes of GPs-GE composite. (a) In situ assembling fabrication of GPs under vacuum filtration of graphene nanoplates/Trion X-100 mixed ink, and the electrodes painted by silver paste; (b) Multilayer slab of GFT sandwiched GPs immersed in epoxy/n-hexane mixture; (c) Epoxy resin curing under 120
C for 6 h. (A colour version of this figure can be viewed online.)

resin after sufficiently diluted, the epoxy matrix can be fully filled in the interspace of GF/GFT composite slab under vacuum conditions, and form strong coupling interfaces among GPs, epoxy matrix and GFT. Subsequently, Fig. 1c elucidates that the wetted slab was transferred to a vacuum oven for cure of epoxy matrix at 85
C for 8 h and simultaneously remove n-hexane. Eventually, the obtained GPs-GE presents a sandwiched multilayer structure with a bulk density of 1.35 g/cm3, and a composition of 1 wt% GPs, 40 wt% glass fibers and 59 wt% epoxy matrix. The final GPs-GE samples were tailored into specific slabs with a size of 20  100  3 mm3 for stimuli-responsive characterizations and 100  100  3 mm3 for Joule-heating investigations. 2.3. Characterization and measurements The microstructural morphologies of graphene nanoplates, GPs, GPs-GE were characterized by a FIB/dual electron beam scanning electron microscopy (SEM) (Helios Nanolab 600i, FEI, US) with an operating voltage of 20 kV. X-ray diffraction (XRD) analysis was carried out by a X-ray diffractometer (X’Pert-Pro MPD, PANalytical, Netherlands) using Cu-Ka radiation (1.540598 Å) with a 2q ranging from 5 to 50
. Raman spectra were recorded by a high-solution Raman spectrometer (inVia, Renishaw, UK) with Raman shifts in the range of 750e3250 cm1. Fourier transform infrared (FT-IR) spectroscopy (Nicolet iS10, Thermo Fisher Scientific, US) data were recorded over the range of 500e4000 cm1. X-ray photoelectron spectroscopy (XPS, K-Alpha 1063, UK Thermo Fisher) was employed to investigate the chemical composition of GPs. A material test machine (MTS-810, MTS, US) was employed to assess mechanical robustness and deformation induced stimuliresponsive performance. Electrical conductivities of GPs and GPsGE were measured by a typical two-probe configuration using a digital multimeter (Victor 8245, Victor, China). The electrodes were painted with silver paste on two sides of the samples to reduce impacts of contact resistance. K-type digital thermometer (Center 309, Taiwan) combined with infrared imaging camera (A310, Flir, US) was used to record temperature variation and mapping distribution. Humidity was tested by hygrothermograph (635-1, Testo, Germany). 3. Results and discussion 3.1. Characterization of materials 3.1.1. Microstructural characterization As shown in Fig. 2a, graphene nanoplates used for assembling

GPs present typical wrinkled SEM morphologies. Fig. 2b contains an optical image of paper-like thin free-standing GPs with the little uneven surface due to the capillary stress induced slight shrinkage during water evaporating at annealing stage after it transferred from filter membrane. The microstructures of as-prepared GPs, as shown in Fig. 2ced, have been observed by SEM at different magnifications (50 and 2 mm scale bar, respectively), displaying uniformly assembled graphene tissues and layer-by-layer stacked cross-section micrographs. Strong p-p interactions among graphene nanoplates efficaciously facilitate the well-connection of ordered inter-junctions and portend a surpassing combination of mechanical robustness and electrical conductivity. Selected area electronic diffraction (SAED) with a typical pattern of hexagonal spots validates high graphitization of graphene nanoplates with the few-layered microstructures (Fig. 2d, inset). This appreciable nature of GPs renders applications in many promising fields, including flexible electrodes, stimuli-responsive sensors and Joule heating skins. As shown in Fig. 2e, GPs, GFT and epoxy resin are composited together as a sandwiched multilayer structure at cross-section, while the inset illustrates the distribution and orientation of glass fibers in plane. GPs and GFT show direct side-to-side connection bonded by epoxy resin. Particularly, the ultrathin interlayered GPs stacked approx. ten thousands of the graphene layers has no disruption occurred during composing process with the other two constituents (Fig. 2f), greatly maintaining the electrical conductivity and stimuli-responsive sensitivity. And potent contacts between GPs and glass fibers are achieved through the well-combined interface with homogenous inner networks (Fig. 2geh), which significantly improves the mechanical properties by coupling two brittle components into a strengthened entirety. Uniformly distributed glass fibers and compacted multilayers of GPs-GE further facilitate the reinforcement of mechanical strength and toughness for the reason of tightly combined interfaces, which, of course, also enhances the deformation compatibility and thermal transfer efficiency in practical Joule-heating and anti-icing applications. Moreover, Fig. 2i exhibits densely stick-packing-like alignment of glass fibers with epoxy matrix sufficiently infiltrated and encapsulated, forming a seamless interface system. 3.1.2. Chemical compositon characterization XRD, Raman spectra and XPS were combined with FT-IR to investigate the chemical compositon of GPs and GPs-GE. As shown in Fig. 3a, a sharp peak at 2q ¼ 25.78
in the XRD spectra exemplifies the typical characteristic of graphitic strutures in GPs and GPs-GE. GPs demonstrates well-orderly stacked layer-by-layer

Q. Zhang et al. / Carbon 124 (2017) 296e307

299

Fig. 2. Microstructural characterizations of graphene nanoplates, GPs and GPs-GE. (a) SEM micrograph of graphene nanoplates; (b) Optical image of GPs; (c)e(d) SEM morphologies of GPs with different magnifications. The inset is SAED pattern of graphene nanoplates; (e) The sandwiched structure of GPs-GE slab at cross-section; (f) The SEM image of GPs-GE micro-compositions; (g) The layer-by-layer stacked multilayer structure of GPs chip; (h) SEM micrograph of stick-packing-like alignment of densely arrayed glass fibers; (i) Wellbonded interface between micro-glass fibers and epoxy matrix in a cross-sectional SEM image. (A colour version of this figure can be viewed online.)

(a)

(b)

Edge

G-band

2D-band

Intensity (a.u.)

Intensity (a.u.)

GPs-GE GPs Natural graphene flake

D-band

Center G-band 2D-band D-band

(c) 81

15

20

25

30 35 o 2 Theta ( )

40

45

50

750 1000 1250 1500 1750 2000 2250 2500 2750 3000 3250 Raman shift  cm-1

(d)

Baseline

C-C/C=C

C-H) C=C)

Intesnity (a.u.)

Absorbance (a.u.)

80 79

C-C)

78 COOH)

C-O-C)

77 C-OH, COOH, H2O)

C1s

Intensity (a.u.)

10

O1s

0 C-OH

200 400 600 Binding energy (eV)

76 75 4000

3500

3000

2500 2000 1500 Wavenumber (cm-1)

1000

500

280.0

282.5

285.0 287.5 290.0 292.5 Binding energy (eV)

295.0

Fig. 3. (a) Comparative XRD pattern of GPs, GPs-GE and natural graphite flake; (b) Raman spectra at edge and in plane locations of GPs; (c) FT-IR spectra of GPs-GE; (d) XPS C1s spectrum of GPs. The inset is wide scan spectrum. (A colour version of this figure can be viewed online.)

microstructures and a high graphitization with the interlayer spacing of 3.5 Å (between (002) planes in GPs) compared to that of prinstine sapcing 3.4 Å for the natural graphite flakes. Comparatively, the XRD characteristic peak for GPs-GE presents a slight shift

from GPs’ 25.78
e25.69
, indicating negligible impact of expoy cure on GPs intrinsic attributes, and facilitating the great inheritance of GPs’ excellent properties passing onto GPs-GE, such as mechanical robustness, electrical conductivity and stimuli-

300

Q. Zhang et al. / Carbon 124 (2017) 296e307

responsive performance. As displayed in Fig. 3b, Raman spectra analysis shows typical weak D-band at 1330 cm1 and intensive Gband at 1585 cm1 from edge to center, revealing the predominant features with high populations of sp2 hybridization and conjunction domains on graphene nanoplates. The sharp 2D-band peaked at 2660 cm1 in all parts of GPs and intensity ratio of 2D to G -band (I2D/IG) larger than 0.6 jointly demonstrate that the original graphene nanoplates have the high monolayer or bilayer rates, which, in turn, guarantes the pristine excellent mechanical and electrical conductive properties of GPs propagating in the GPs-GE composite [25e27]. Comparing to the other characteristic peaks, the weak Dband means graphene nanoplates used for GPs construction possesses much few defect populations or oxygenic functional groups than those of chemical converted graphene precusors [2,7,28,29]. FT-IR was conducted to further characterize chemical groups of GPs-GE, as shown in Fig. 3c. The spectra of FT-IR presents the absorption peaks of CeOH and COOH at 3440 cm1 and CeOeC at 1220 cm1, while C]C locates at 1660 cm1 as an identification of conjuncted domains of sp2 hybridization on graphene nanoplates. Comparing with the natural graphite flakes, a few of oxygenic functional groups still remain winthin the GPs, which are coincident with weak D-band in Fig. 3b as a characterization of limited defect populations on pristine graphene nanoplates. Such defects on GPs are known to reduce the electron mean free path, leading to the decrease of its electrical conductivity (2.8  104 S/m) by resonant scattering phenomena in contrast with that of perfect graphene lattice (~106 S/m) [30]. In addition, XPS was employed to investigate the chemical composition on GPs. As shown in the inset

of Fig. 3d, GPs surface contains carbon (C1s) sharply peaked at 288 eV and weak peak of oxygen (O1s) at 534 eV. The C1s spectrum in Fig. 3d can be fitted into two typical peaks such as pristine graphitic CeC/C]C at 284.6 eV and carbon in alcohol (CeOH) at 286.2 eV. The dominant peak at 284.6 eV further confirms the GPs mostly consists of nonfunctional graphitic structures but a few retained oxygenic groups with a high carbon to oxygen ratio of 24:1 [2]. Such low oxygenic derivatives enable to address the high electrical conductivity and mechnical robustness of GPs originated from graphene nanoplates, which suggests its promising applications (e.g., stimuli-responsive sensor and Joule heating skin). 3.2. Mechanical properties and deformation derived stimuliresponsive performances 3.2.1. Mechanical robustness and static deformation Here, we report the static deformation derived stimuliresponsive performance of as-prepared GPs-GE composite with outstanding mechanical robustness by a series of electromechanical measurements. As shown in Fig. 4a, the slab sample of GPs-GE by size of 20  100  3 mm3 was stretched by MTS with a loading rate of 0.4 mm/s. Simultaneously, two electrodes were connected with a digital multimeter to monitor the electrical resistance variations induced by mechanical deformations. A higher Young’s modulus (~4.0 GPa) and elastic stress (~80 MPa) associated with a larger elastic strain (over 2%) of GPs-GE slab are depicted from the strain-stress curve (see Fig. 4b), revealing a significant reinforcement on both tensile strength and toughness than

(b) 90

Stress (MPa)

70 60

80 Stress (MPa)

80

50

re St

40

tch

ing

le Re

20 0 0.0

40

2000 cycles

Stretching Releasing

60

as

ing

0.4 0.8 1.2 1.6 Tension strain (%)

2.0

30 20 10 0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

(c) 40 35

(d) 50 45 40 35 30 25 20 15 10 5 0

Experimental result Linear fitting result

| R/R| (%)

30

R/R (%)

25 20 15 10 5

R/R=2.25+12.97

0 -5 0.0

F

0.2

0.4

0.6

0.8

1.0

1.2

Tension Strain (%)

1.4

1.6

1.8

2.0

T

T

80mm

Resistivity ( *cm)

Tension strain (%)

0.0060 0.0055 g 0.0050 hin etc Str 0.0045 Re 0.0040 lea sin g 0.0035 0 1 2 3 4 5 6 7 8 Bending deflection (mm)

Flat

0

2

4

6

8mm bending deflection

8 10 12 14 Bending Cycle

16

18

20

Fig. 4. Mechanical properties and static deformation derived stimuli-responsive performance of GPs-GE. (a) The snapshot of mechanical tensile test; (b) The strain-stress curve at the maximum tensile strain of 2% for 2000 stretching-releasing cycles. The inset is a single stretching-releasing cycle; (c) Resistance-strain dependent curves at maximum strain of 2%; (d) Resistance change depending on deflection under three-point-bending deformation. (A colour version of this figure can be viewed online.)

Q. Zhang et al. / Carbon 124 (2017) 296e307

that of three individual constituted components. Furthermore, the stretching-releasing cyclic test was conducted to investigate mechanical stability and fatigue resistance. GPs-GE presents an excellent recoverability and robust linear elastic behavior without any large geometric fracture, mechanical strength degeneration and microstructural delamination among those of composed three constitutes (GPs, glass fibers and epoxy resin) during 2000 cycles, as shown in Fig. 4b. As a result, the severe brittleness and poor resistance to shear/tensile force of pristine GPs have been highly improved by the mechanically protective components (GFT and epoxy resin), which is attributed to the strong interface-derived synergistic effects. Firstly, we evaluated the stimuli-responsive properties of GPsGE responding to static tensile deformation. A linear fitting was extracted to manifest the change of electrical resistance depending on different tensile strains. As demonstrated in Fig. 4c, experimental data statistically coincide with a linear fitting function (),

2

0 -1 10

20

30

40 50 60 Time (sec)

70

80

90

30 20 10

Strain (%)

-1

2

3

4 5 6 Tension cycle

7

8

9

10

0

14

16

18

20

1

2

3

4 5 6 Tension cycle

7

8

9

10

(d)35

25

0.4

0.6

0.8 1.0 1.2 Time (sec)

1.4

1.6

1.8

2.0

R/R (%)

0.2

R/R (%)

-1

R/R=+28.9%

20 15 10 5 0 R/R=-1.8%

-5 1

2

3

(e) 3

4 5 6 Tension cycle

7

8

9

0

10

(f) 3

Strain (%)

2 1 0 10

20

30

40

40 50 60 Time (sec)

70

80

90

100

200 400 600 800 1000 1200 1400 1600 1800 2000 Tension cycle

2 1 0 0

10

20

30

40 50 60 Time (sec)

70

80

90

100

1

2

3

4 5 6 Tension cycle

7

8

9

10

40

R/R (%)

30 20 10 0 0

8 10 12 Time (sec)

10

30

0

6

20

0

40 30 20 10 0 -10 0

4

30

1

-2 0.0

2

0 1

(c) 2

Strain (%)

0

40

0

R/R (%)

1

-2 0

100

R/R (%)

R/R (%)

40

0

2

(b)

1

-2 0

indicating that resistance variation positively following the change of deformation with a high tensile strain based sensitivity. Theoretically, the resistance change is linearly dependent on the change of elongation as a function of . Therefore, the increase of the resistivity under the mechanical deformation can be ascribed to internal tensile strain induced changes in the electrical properties of the individual graphene nanoplates within the GPs or resulted from the shear strain triggered enlargement of interfacial distance between nanoplates. In addition, as shown in Fig. 4d, the three-pointbending test of GPs-GE slab sample by size of 20  100  3 mm3 was implemented to study the responsive properties towards bending impacts. The related |DR/R| stabilizes at 15% with a deflection in the mid-span up to 8 mm (10% of the testing slabs’ span of 80 mm between two support points) through 20 bendingreleasing cycles, which reveals excellent structure robustness and stable stimuli-responsive sensitivity. Interestingly, the inset of Fig. 4d exhibits that the electrical resistivity firstly increases from

Strain (%)

Strain (%)

(a)

301

30 20 10 0

1

2

3

4 5 6 Tension cycle

7

8

9

10

0

Fig. 5. The stimuli-responsive performance of GPs-GE derived by dynamic excitation. (a)e(c) sine wave for 0.1, 0.5 and 5 Hz, respectively; (d) 5 Hz sine wave with 2000 repeated cycles; (e)e(f) 0.1 Hz square and triangle waves, respectively. (A colour version of this figure can be viewed online.)

Q. Zhang et al. / Carbon 124 (2017) 296e307

original 3.56  103 to 5.8  103 U/cm with corresponding bending deflection ranging from 0 to 4 mm, and then dramatically decreases to 0.00325 U/cm following the continuous increase of deflection up to 8 mm. The releasing curve demonstrates the similar tendency as the bending process but significant hysteresis responsive effect is presented (see the inset in Fig. 4d). Because of the further compaction induced well-touch conditions of multilayer interfaces during bending deformation, the resistivity change of GPs-GE at finial flat state approaches up to 5% with resistivity lowering to 0.00325 U/cm after first loading-releasing cycle, Overall, GPs-GE suggests outstanding mechanical robustness, significant strength reinforcement and sensitive stimuli-responsive performance. 3.2.2. Dynamic deformation derived stimuli-responsive performances The stimuli-responsive properties of GPs-GE, derived by stimuliinputs of dynamic deformations by different frequencies and waveforms, were further evaluated using MTS by strain amplitude up to 2%, as shown in Fig. 5. Fig. 5aec demonstrates that the variations of the electrical resistance stabilize around 30, 25 and 20% as sine-wave-like responses corresponding to stimuli-inputs of 0.1, 0.5 and 5 Hz sine wave, respectively, revealing sensitive synchronous responses and frequency-dependent behaviors. Obviously, the amplitude change of resistivity tends to stable after initial several cycles, while the responsive sensitivity and coupling synchronicity between resistivity and dynamic stimuli-inputs are fading dependently with the increase of inputting frequencies, indicating well selectivity to low-frequency signal but weaker identification for higher frequency inputs. Moreover, stimuli-responsive properties of GPs-GE composites were confirmed over 2000 stretching cycles, as shown in Fig. 5d, which presents relative electromechanical stability with maximum resistance change of DR/R approx. 28.9%. Comparatively, as shown in Fig. 5eef, the 0.1 Hz square and triangle wave inputs lead to well-identified square/triangle-like responses with resistance changes of approx. 15 and 30%, respectively, implying a well coincidence of GPs-GE to other type stimuliinputs. Particularly, GPs-GE illustrates superior sensitivity and adaptability for the gently implemented stimuli-inputs (sine and triangle waves) to those of sharply applied signals (square wave). Consequently, the variations in electrical resistance of GPs-GE composites under dynamic excitations are confirmed to be stable, selectable and reliable, suggesting their promising potential in applications such as highly sensitive stimuli-responsive sensors.

Resistivity ( *cm)

0.009 0.008 0.007

300 250

0.006 200

0.005

150

0.004 0.003 0.002

100

-40

-20

0

20

40

60 o

Temp. ( C)

80

100

120

(b)

350

24

1.6 Water ratio Resistivity

1.4

22 20

1.2 18 1.0

16

0.8

14

0.6

12

R/R (%)

Experiment of Resistivity Experiment of Conductivity Fitting curve of Resistivity Fitting curve of Conductivity

Conductivity (Scm-1)

(a) 0.010

3.2.3. Temperature and humidity variations induced stimuliresponsive performance The impacts of temperature stimuli on electrical resistance and conductivity is illustrated in Fig. 6a, which presents Gauss functionlike dependent relationships with temperature ranging from 40 to 120
C and reveals a dual negative/positive temperature coefficient of expansion (NTCE/PTCE). It is generally recognized that the conduction mechanism under thermal activation in non-crystalline materials is governed by expansion strain induced bandgap changes and hopping conductions of graphene nanoplates within GPs [2,31], both of which is a trade-off on different temperature regimes, as shown in Fig. 6a. For temperature ranging from 40 to 80
C, the transport mechanism of electrons is suppressed by bandgap increasing under local thermal strains [32], leading to the resistance increase with temperature elevating. In contrast, once the temperature is higher than 80
C, the hopping conduction mechanism of graphene nanoplates gives rise to enhancement of the electrical conductivity of monolithic GPs [2,31], which demonstrates a dual 2D/3D effect in microscale (2D in basal plane of graphene nanoplates: ln (r)fT[1/(2þ1)], 3D in stacked interfaces among graphene nanoplates: ln (r)fT[1/(3þ1)], where r is resistivity and T is temperature), comparable to that of state-of-theart carbon nanostructures (e.g., carbon nanotubes sheets, graphene aerogels and related composites) [2,31,33]. In general, GPsGE composite possesses eminent electrical properties with original conductivity over 280 S/cm at room temperature (RT), which is much higher than those of reported conductive composites (0.01e2 S/cm) under similar content of graphene fillers (~1 wt%) [2,7,34]. Moreover, the sensitivity of GPs-GE’s electrical property to ambient humidity was investigated under different water absorption amounts, which is considered for stimuli-monitoring applications under the humid environment. As shown in Fig. 6b, the increase of water ratio in the GPs-GE triggers monotonic enlarging of resistivity in primary regimes. The reason for such tendency is that the water absorption in epoxy matrix would lead to the microstructural expansion of GPs-GE composite, and induce the enlargement of internal swelling strains in epoxy matrix, glass fibers and GPs, which eventually results in the significant increase of GPs’ resistance within GPs-GE composite. Within 24 h’ water adsorption process, the electrical resistivity realizes 22% increase corresponding to approx. 1.3 wt% water ratio (absorbed water weight to GPs-GE slab weight). After that, the growth of water ratio immigrates to a stabilized balance, which, in turn, leading to a synchronous equilibrium of electrical resistivity (slightly fluctuated

Water ratio (%)

302

10 0.4 8 0.2 0.0

6 0

4

8

12 16 20 24 28 32 36 40 44 48 Water adsorption time (h)

4

Fig. 6. The stimuli-responsive characteristics of GPs-GE under different environment impacts. (a) Variations of the electrical resistivity and conductivity as a function of the temperature; (b) Electrical resistivity variations in respect to water ratio under 25
C absorption procedure. (A colour version of this figure can be viewed online.)

Q. Zhang et al. / Carbon 124 (2017) 296e307

around 22%). In this regard, the long-time serving of this GPs-GE composite as a smart sensor under the humid condition is highly favorable. On the other hand, those environmental impacts induced sensing responses of GPs-GE electrical resistance can be synchronously monitored to feedback and prevent icing disasters with timely deicing/anti-icing measurements in practices. 3.3. Joule heating performance of GPs-GE composite Fig. 7 describes the electro-thermal properties of this GPs-GE composite used as Joule heating element by size of 100  100  3 mm3 with different heat flux densities under varying wind induced thermal convection conditions. The temperature elevates monotonously from original RT under breezeless conditions, as depicted in Fig. 7a. And the Joule heating rates as high as 0.06, 0.09, 0.14, 0.18, 0.20, 0.25, 0.29 and 0.35
C/sec are corresponding to

(a) 200

180

Temp. (oC)

160 140

inputted heat flux densities of 400, 800, 1200, 1600, 2000, 2400, 2800 and 3200 W/m2, respectively. The related maximum records of temperature are up to 38, 55, 85, 105, 120, 148, 175 and 210
C within 10 min. Comparatively, as shown in Fig. 7b, the curve trends to be stable for 5 min after a dramatic rising procedure under inplane wind fluctuation of 3 m/s, the related equilibrium temperatures approximately stabilize at 35.3, 50.05, 64.76, 87.07, 105.57 and 123.61
C corresponding to inputted heat flux densities of 1000, 2000, 3000, 4000, 5000 and 6000 W/m2 (Table 1). The maximum records of Joule heating rates decay to 0.06, 0.12, 0.18, 0.27, 0.34 and 0.41
C/sec correspondingly compared to those of breezeless case (Table 2). Obviously, the forced thermal convection under the higher wind speed accelerates the progress achieving equilibrium state, which ranges from 5 min to 100 s corresponding to the wind speed of 3e13.5 m/s. Deicing efficiencies are jointly affected by multi-parameters

(b) 140

800W/m2 1600W/m2 2400W/m2 3200W/m2

120 100 80 60

2000W/m2 5000W/m2

3000W/m2 6000W/m2

100 80 60 40

40

20

20

(c) 0.45

0.40 0.35

100

200

300 400 Time (sec)

500

0

600

0.30 0.25 0.20 0.15 0.10 0.05 0.00

1000

2000

3000 4000 5000 Heating power (W/m2)

50 100 150 200 250 300 350 400 450 500 550 600 Time (sec)

(d) 120

Heating ratio-3m/s Heating ratio-6m/s Heating ratio-9m/s Heating ratio-13.5m/s

Equilibrium Temp. (oC)

0

Heating ratio (oC/sec)

1000W/m2 4000W/m2 Reference

120

Temp. (oC)

400W/m2 1200W/m2 2000W/m2 2800W/m2 Reference

303

EquilibriumTemp.-3m/s EquilibriumTemp.-6m/s EquilibriumTemp.-9m/s EquilibriumTemp.-13.5m/s

100 80 60 40 20 0

6000

1000

2000

3000 4000 5000 Heating power (W/m2)

6000

40 20

ng ooli lly C

Temp. (oC)

60

ra Natu

80

Jou l e-h eat ing

(f) 100

0 -20 0

250 500 750 1000 1250 1500 1750 2000 2250 2500 Time (sec)

Fig. 7. (a)-(b) The relationship between time and temperature at different heat flux densities under windless condition and wind speed of 3 m/s, respectively; (c) The heating power-rate curves with wind speeds ranged from 3 to 13.5 m/s. (d) The comparison chart for the relationship between heat flux densities and equilibrium temperatures at wind speeds from 3 to 13.5 m/s; (e) The infrared image of GPs-GE slab during Joule heating process; (f) The stability of Joule heating process under heating-cooling cyclies for 20
C ambient temperature at heat flux density of 3000 W/m2. (A colour version of this figure can be viewed online.)

304

Q. Zhang et al. / Carbon 124 (2017) 296e307

Table 1 Equilibrium temperatures under different wind speeds and heat flux densities (oC). Heat flux density (W/m2)

Wind speed (m/s)

3 6 9 13.5

1000

2000

3000

4000

5000

6000

35.3 29.18 24.5 22.43

50.05 39.37 30.56 26.24

64.76 51.53 38.99 30.83

87.07 61.38 45.24 36.45

105.57 72.38 53.56 41.41

123.61 86.35 61.56 47.69

Table 2 Joule heating rates under different wind speeds and heat flux densities (oC/sec). Heat flux density (W/m2)

Wind speed (m/s)

3 6 9 13.5

1000

2000

3000

4000

5000

6000

0.06 0.05 0.03 0.02

0.12 0.10 0.07 0.04

0.18 0.16 0.13 0.07

0.27 0.20 0.17 0.11

0.34 0.26 0.22 0.14

0.41 0.33 0.28 0.18

such as heating power input, weed speed, environment temperature etc., for this reason, it is considerable to solely study them under manual controlling. First of all, several check tests were conducted to understand the heating performance with different heating power inputs and wind speeds. As shown in Fig. 7ced, heating rates and equilibrium temperature rise along with increments of heat flux densities. However, both of them decline monotonically with wind speed increasing. The stable heatingcooling performance, satisfactory mechanical property and acceptable resistance variation in humid working environment judiciously back up the practical requirements for deicing/antiicing applications of this GPs-GE composite. Fig. 7c presents the heating rate is approximately linear-dependent on heating flux density, the higher wind speed results in the lower heating rate, such as 0.02e0.18
C/sec for 13.5 m/s wind speed while 0.06e0.41
C/sec for 3 m/s. Comparatively, the Joule heating rates under breezeless cases are 2e5 times higher than those of wind turbulence conditions (see Table 2). It is, therefore, verified that the thermal convection negatively influences the Joule heating efficiency due to the forced thermal energy dissipating. Similarly, the equilibrium temperatures reveal a positive dependence on heat flux densities but as negative correlation for wind speeds (see Fig. 7d). Theoretically, the temperature of GPs-GE heating element responds to the supply-dissipation relationship during its heating procedure. Normally, for a certain heat flux input under the

fluctuation of forced thermal convection, the supply energy far outweighs its heat dissipation in the beginning, resulting in the rapid rises of GPs-GE slab’s temperature. However, in turn, it could lead to the larger heat dissipation rate due to the increscent temperature gradient between GPs-GE surface and the ambient environment. The equilibrium state means the balance trade-offs between energy consumes and the related inputting powers. In this regard, larger heating power brings about much higher equilibrium temperature distinctively. A homogenous temperature distribution was underlined by a uniform infrared mapping of GPs-GE slab during heating procedure (Fig. 7e). As show in Fig. 7f, the cyclic curves of Joule heating and naturally cooling processes nearly overlap each other for same heat flux density of 3000 W/m2 under windless and 20
C ambient conditions, indicating constant and stable Joule heating property of GPs-GE. Moreover, Fig. 8 characterizes the stability of both electrical and structural properties of GPs-GE slab during 200 J heating cycles with temperature ranging from 20 to 90
C. An outstanding stability of electrical resistivity for GPs-GE heater undergoing long term heating/cooling cycles has been confirmed with a slight fluctuation less than 2%, as shown in Fig. 8a. The microstructural evolution of GPs-GE was further detected by SEM images corresponding to the original state, 100th and 200th cycles (see Fig. 8b). Being coincident with resistance change, no visible structural degenerations and microstructural delamination are appeared during heating/cooling processes, which benefits from the strongly bonded interfaces among GPs, glass fibers and epoxy resin to endure cyclic thermal stress-induced structure fatigue. Both the uniformly distributed temperature field and high stability of Joule heating performance may propel a suite of electrothermal applications of GPs-GE composite such as indoors self-heater or aircraft de-icing/anti-icing skin. 3.4. The deicing/anti-icing applications of GPs-GE’s joule heating behavior It is greatly necessary to investigate GPs-GE’s Joule heating performance in practical cases with a view to evaluate the possibility of its applications such as pavement infrastructural snow thawing, aircrafts and wind turbine blades anti/de-icing. Therefore, we have monitored the entire heating and deicing procedures for two validating cases of pre-heating derived anti-icing and direct deicing performances, respectively. The experimental setups are schematically illustrated in Fig. 9a, which contains a thermally insulated substrate to create one-dimensional thermal conduction, DC electric power supplier, and K-type thermocouples for

(a) 0.006

cm)

0.004

Resistivity (

0.005

0.003 0.002 0.001 0.000

0

20

40

60

80

100 120 140 160 180 200

Cycling times Fig. 8. (a)-(b) Electrical resistivity change and microstructural evolution of GPs-GE during cyclic Joule heating processes. (A colour version of this figure can be viewed online.)

Q. Zhang et al. / Carbon 124 (2017) 296e307

305

(b) 15

Deicing curve Reference

5 0

Freezing

Temp. (oC)

10 Pre-heating

Anti-icing

Clean o

0 C

-5 -10 -15 -20

0

500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Time (sec)

(c) 30

(d) 50

Deicing curve Reference

25

40

20

5 0

Fluid

Solid

Phase change

Temp. (oC)

Freezing

Ice formation

-5

10 0

Deicing

-10

-10

250

250

200

200

150

150

100

100

50

50

0

1000

2000 3000 4000 Heat flux density (W/m2)

5000

0

350 300 250 200 150 100 50

0

50

(f) 0.16

Phase change stage (sec-mm-1-m-2)

300

Total deicing time (sec-mm-1-m-2) 350 Heating stage (sec-mm-1-m-2) -1 -2 Phase change stage (sec-mm -m ) 300

350

-20

0

100

150

200 250 300 Time (sec/mm)

350 0.16

0.14

Total deicing energy (kWh-mm-1-m-2) Heating stage (kWh-mm-1-m-2) 0.14 Phase change stage (kWh-mm-1-m-2)

0.12

0.12

0.10

0.10

0.08

0.08

0.06

0.06

0.04

0.04

0.02

0.02

0.00

1000

2000 3000 4000 Heat flux density (W/m2)

5000

400

0.00

450

0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02

Phase change stage (kWh-mm-1-m-2)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Time (sec)

Heating stage (sec-mm-1-m-2)

0

Heating stage (kWh-mm-1-m-2)

-15

Total deicing time (sec-mm-1-m-2)

20

Total deicing energy (kWh-mm-1-m-2)

Temp. (oC)

10

(e)

2000W/m2 4000W/m2 Reference

30

15

-20

1000W/m2 3000W/m2 5000W/m2

0.00

Fig. 9. - The deicing/anti-icing applications of GPs-GE. (a) Schematic illustration of deicing system setup; (b) Anti-icing with pre-heating strategy under 15
C ambient temperature and 300 W/m2 heat flux density; (c) The ice preformation under 15
C and followed deicing with heat flux density of 3000 W/m2; (d) Dicing performance of unit thickness ice accretion under 15
C ambient temperature and different heat flux density input ranging from 1000 to 5000 W/m2; (e)e(f) Curves of time and energy consumption vs. heat flux density under 15
C ambient temperature with the comparison chart of total component, the phase change and heating stages, respectively. (A colour version of this figure can be viewed online.)

temperature monitoring at the interface between GPs-GE slab and deposited ice layer. Firstly, the GPs-slab was placed in a low-temperature chamber with an environmental temperature of 15
C, and supplied 300 W/m2 heat flux density to per-heat slab up to freezing point (0
C) within 950 s, as shown in Fig. 9b. Then the continuous heating power was constantly transferred to thermal energy maintaining the temperature over slab surface around 0
C. After that, the freezing spray by initial temperature of ~0
C was shed on slab within 1 h by the simulated precipitation of 10 mm/h. The data in Fig. 9(b), obtained using as-fixed thermocouple probes over the surface of GPs-GE self-heating slab, reveals that the anti-icing purpose was realized by the detected temperature rising over

1e3
C with surface free of icing accretion. However, when the freeze water spray falls over the surface of GPs-GE, the thermocouple probes were disturbed slightly, leading to the fluctuation phenomena of curve for anti-icing stage. It makes feasible for antiicing purpose under 15
C harsh conditions using GPs-GE with only 300 W/m2 heating power applied in advance. Therefore, the slab is capable to keep clean surface in high humidity cold conditions and prevent possible ice-induced hazards with a quite low serving power and economic consumption. On the other hand, the ice was created on slab in advance to simulate the real situation of icing accretion, which was used to conduct practical deicing purpose of GPs-GE. As shown in Fig. 9c, the 10 mm thick ice accretion was gradually formed after 1500 s

306

Q. Zhang et al. / Carbon 124 (2017) 296e307

using freezing spray (~0
C) over slab under the same low temperature (15
C) as above anti-icing case. Then, the temperature of as-prepared ice dropped back to an original environment temperature of 15
C within 2400 s as an initial condition for followed deicing experiment. 3000 W/m2 heat flux density was then supplied at moment of 4200 s to melt the accreted ice which would completely cleaned after 365 s, indicting the rapider deicing performance of GPs-GE slab in practice under the larger power input. An entire deicing process can be divided into three stages: system heating, phase change and continuous heating stage. At the beginning, temperature rises quickly from 15
C to freezing point (0
C) for 80 s. Subsequently, 10 mm thick ice accretion melts from solid to water completely within 285 s with the temperature remaining around freezing point (0 ± 1
C) at interface between ice layer and GPs-GE slab. Finally, the melting water is evaporated and the temperature over GPs-GE slab surface continuously enlarges over 0
C. Comparatively, both anti-icing and deicing realize 10 mm thick ice clean with almost similar energy consumption of approx. 0.03 kW h/(mm-m2) but are suitable for different schemes according to specific practice requirements, such as anti-icing option used for taking precautions against icing in humidly low temperature weather, deicing operation pertinently taking account for situation of ice clean after it have been formed. As shown in Fig. 9d, deicing time consumption can be optimized by using different heat flux densities. When the inputting power ranges from 1000 to 5000 W/m2, the melting processes totally cost approx. 330 to 35 s for per millimeter thick ice under an ambient temperature of 15
C with the phase change state dominating a majority of energy and time consumptions due to large latent heat of ice (335 kJ/kg). Particularly, the maximum power input of 5000 W/m2 realizes less than 1 min time consumption, which is the twentieth part of the state-of-the-art time cost formerly [3,4,24], showing an outstanding performance of Joule heating effect on deicing applications. The feasibility of GPs-GE in deicing/anti-icing applications was further evaluated, as shown in Fig. 9eef, time and energy efficacies have been meditated together corresponding to different heat flux densities ranging from 1000 to 5000 W/m2. The total components of time and energy consumptions are decomposed to diverse costs for specific heating and phase change stages, respectively. Comparatively, 5000 W/m2 power input presents the shortest time and most effective energy utilization by 37.5 s and 0.048 kW h for per millimeter thick and per square meter covered ice under 15
C, while the 2000 W/m2 reveals the second higher time cost approx. 225 sec and the highest energy demand of 0.125 kW h. Entirely, the time cost monotonously decreases from 325 to 37.5 s with the related heat flux densities increased from 1000 to 5000 W/ m2, but ‘olive-styled’ tendency for energy consumption. In addition, the phase change stage consumes over 85% of total energy and time due to larger latent heat of 335 kJ/kg compared to its specific heat of 63 kJ/kg. Using these evaluated indices during deicing process, the deicing setups of GPs-GE-based Joule heating systems can be further optimized. Accordingly, we can deduce that 5000 W/m2 is the most cost-effective heat flux density, since it consumes the shortest de-icing time and the lowest energy in contrast to the other four cases. The performance of GPs-GE composites suggests their promising applibility in deicing applications because of higher enegy utilizing efficiency and larger Joule heating rate than those of state-of-the-art reportes [3,4,24]. 4. Conclusion By composing of three constituents as a multilayered structure, a conductive graphene-papers (GPs)/glass-fibers (GFs)

reinforced epoxy composite (GPs-GE) with sensitive stimuliresponsive behavior and outstanding Joule heating performance is highlighted in this work. Due to the well-bonded interface induced multi-functionalization effect of the multilayers in microstructures, GPs-GE demonstrates a low electrical resistivity of 3.56  103 U/cm and remarkably enhanced mechanical properties (tensile strength ~ 80 MPa, Young’s modulus ~ 4 GPa). The stimuli-responsive characterization of electrical resistance change reveals the sensitively synchronous responses of GPs-GE to different stimuli inputs such as mechanical deformation, temperature fluctuation and humidity variation. Meanwhile, the structural stability of GPs-GE composites has been demonstrated in cyclic stretching tests with a maximum tensile strain up to 2% over 2000 cycles. Joule heating and de-icing/anti-icing investigations of GPs-GE indicate large heating rate, high energy efficiency and competitive cost. Such superior performance of GPs-GE confirms that the design of multilayered microstructure to achieve multi-functionalization paves a novel way to scale-up fabrication of advanced graphene composites, suggesting promising applications as stimuliresponsive sensors and electric heating units in engineering intelligent monitoring and icing-induced disaster prevention at low temperature environment. Acknowledgements The authors would like to greatly appreciate financial support from Natural Science Foundation of China (Grant No.11622217, Grant No. 81571829) and National Program for Special Support of Top-Notch Young Professionals (Grant No. 11622217). The Fundamental Research Funds for the Central Universities (Grant No. lzujbky-2017-k17, Grant No. lzujbky-2017-kb03). References [1] Y. Qin, Q. Peng, Y. Ding, Z. Lin, C. Wang, Y. Li, et al., Lightweight, superelastic, and mechanically flexible graphene/polyimide nanocomposite foam for strain sensor application, ACS Nano 9 (9) (2015) 8933e8941. [2] Q. Zhang, X. Xu, H. Li, G. Xiong, H. Hu, T.S. Fisher, Mechanically robust honeycomb graphene aerogel multifunctional polymer composites, Carbon 93 (2015) 659e670. [3] H. Li, Q. Zhang, H. Xiao, Self-deicing road system with a CNFP high-efficiency thermal source and MWCNT/cement-based high-thermal conductive composites, Cold Regions Sci. Technol. 86 (2013) 22e35. [4] Q. Zhang, Y. Yu, W. Chen, T. Chen, Y. Zhou, H. Li, Outdoor experiment of flexible sandwiched graphite-PET sheets based self-snow-thawing pavement, Cold Regions Sci. Technol. 122 (2016) 10e17. [5] J.J. Bae, S.C. Lim, G.H. Han, Y.W. Jo, D.L. Doung, E.S. Kim, Heat dissipation of transparent graphene defoggers, Adv. Funct. Mater. 22 (22) (2012) 4819e4826. [6] D. Janas, K.K. Koziol, A review of production methods of carbon nanotube and graphene thin films for electrothermal applications, Nanoscale 6 (6) (2014) 3037e3045. [7] J. Ge, L.A. Shi, Y.C. Wang, H.Y. Zhao, H.B. Yao, Y.B. Zhu, et al., Joule-heated graphene-wrapped sponge enables fast clean-up of viscous crude-oil spill, Nat. Nanotechnol. 12 (2017) 434e440. [8] Y. Yao, K.K. Fu, C. Yan, J. Dai, Y. Chen, Y. Wang, et al., Three-dimensional printable high-temperature and high-rate heaters, ACS Nano 10 (5) (2016) 5272e5279. [9] O. Parent, A. Ilinca, Anti-icing and de-icing techniques for wind turbines: critical review, Cold Regions Sci. Technol. 65 (1) (2011) 88e96. [10] C. Mayer, A. Ilinca, G. Fortin, J. Perron, Wind tunnel study of electro-thermal de-icing of wind turbine blades, Int. J. Offshore Polar Eng. 17 (03) (2007). [11] H.S. Jang, S.K. Jeon, S.H. Nahm, The manufacture of a transparent film heater by spinning multi-walled carbon nanotubes, Carbon 49 (1) (2011) 111e116. [12] T.J. Kang, T. Kim, S.M. Seo, Y.J. Park, Y.H. Kim, Thickness-dependent thermal resistance of a transparent glass heater with a single-walled carbon nanotube coating, Carbon 49 (4) (2011) 1087e1093. [13] D. Janas, K.K. Koziol, Rapid electrothermal response of high-temperature carbon nanotube film heaters, Carbon 59 (2013) 457e463. [14] R. Menzel, S. Barg, M. Miranda, D.B. Anthony, S.M. Bawaked, M. Mokhtar, et al., Joule heating characteristics of emulsion-templated graphene aerogels, Adv. Funct. Mater. 25 (1) (2015) 28e35. [15] Y. Yao, K.K. Fu, C. Yan, J. Dai, Y. Chen, Y. Wang, et al., Three-dimensional

Q. Zhang et al. / Carbon 124 (2017) 296e307

[16] [17]

[18]

[19]

[20]

[21]

[22] [23]

[24]

printable high-temperature and high-rate heaters, ACS Nano 10 (5) (2016) 5272e5279. Y. Yao, K.K. Fu, S. Zhu, J. Dai, Y. Wang, G. Pastel, et al., Carbon welding by ultrafast Joule heating, Nano Lett. 16 (11) (2016) 7282e7289. K.S. Kim, Y. Zhao, H. Jang, S.Y. Lee, J.M. Kim, K.S. Kim, et al., Large-scale pattern growth of graphene films for stretchable transparent electrodes, Nature 457 (7230) (2009) 706e710. D.A. Dikin, S. Stankovich, E.J. Zimney, R.D. Piner, G.H. Dommett, G. Evmenenko, et al., Preparation and characterization of graphene oxide paper, Nature 448 (7152) (2007) 457e460.  P. Simek, Z. Sofer, O. Jankovský, D. Sedmidubský, M. Pumera, Oxygen-free highly conductive graphene papers, Adv. Funct. Mater. 24 (31) (2014) 4878e4885. H. Chen, M.B. Müller, K.J. Gilmore, G.G. Wallace, D. Li, Mechanically strong, electrically conductive, and biocompatible graphene paper, Adv. Mater. 20 (18) (2008) 3557e3561. S. Bae, H. Kim, Y. Lee, X. Xu, J.S. Park, Y. Zheng, et al., Roll-to-roll production of 30-inch graphene films for transparent electrodes, Nat. Nanotechnol. 5 (8) (2010) 574e578. K. Davami, L. Zhao, E. Lu, J. Cortes, C. Lin, D.E. Lilley, et al., Ultralight shaperecovering plate mechanical metamaterials, Nat. Commun. (2015) 6. M.J. McAllister, J.L. Li, D.H. Adamson, H.C. Schniepp, R. Car, R.K. Prud’homme, et al., Single sheet functionalized graphene by oxidation and thermal expansion of graphite, Chem. Mater. 19 (18) (2007) 4396e4404. H. Chu, Z. Zhang, Y. Liu, J. Leng, Self-heating fiber reinforced polymer composite using meso/macropore carbon nanotube paper and its application in deicing, Carbon 66 (2014) 154e163.

307

[25] Z. Chen, W. Ren, L. Gao, B. Liu, S. Pei, H.M. Cheng, Three-dimensional flexible and conductive interconnected graphene networks grown by chemical vapour deposition, Nat. Mater. 10 (6) (2011) 424e428. [26] A.C. Ferrari, J.C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, et al., Raman spectrum of graphene and graphene layers, Phys. Rev. Lett. 97 (18) (2006), 187401. [27] E.H.M. Ferreira, M.V.O. Moutinho, F. Stavale, M.M. Lucchese, R.B. Capaz, C.A. Achete, et al., Evolution of the Raman spectra from single-, few-, and many-layer graphene with increasing disorder, Phys. Rev. B 82 (12) (2010), 125429. [28] Q. Zhang, F. Zhang, S.P. Medarametla, H. Li, C. Zhou, Lin D. 3D printing of graphene aerogels, Small 12 (13) (2016) 1702. [29] Q. Zhang, X. Xu, D. Lin, W. Chen, G. Xiong, Y. Yu, et al., Hyperbolically patterned 3D graphene metamaterial with negative Poisson’s ratio and superelasticity, Adv. Mater. 28 (11) (2016) 2229. [30] F. Giannazzo, S. Sonde, R.L. Nigro, E. Rimini, V. Raineri, Mapping the density of scattering centers limiting the electron mean free path in graphene, Nano Lett. 11 (11) (2011) 4612e4618. [31] S.N. Mott, Conduction in Non-crystalline Materials, Clarendon, Oxford, 1987. [32] J. Li, T.F. Chung, Y.P. Chen, G.J. Cheng, Nanoscale strainability of graphene by laser shock-induced three-dimensional shaping, Nano Lett. 12 (9) (2012) 4577e4583. [33] Z. Xu, Y. Zhang, P. Li, C. Gao, Strong, conductive, lightweight, neat graphene aerogel fibers with aligned pores, ACS Nano 6 (8) (2012) 7103e7113. kalova , Electronic conduction in polymers, carbon [34] A.B. [34] Kaiser, V. Ska nanotubes and graphene, Chem. Soc. Rev. 40 (7) (2011) 3786e3801.