Home
Search
Collections
Journals
About
Contact us
My IOPscience
Transport and thermoelectric properties of polyaniline/reduced graphene oxide nanocomposites
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Nanotechnology 25 475705 (http://iopscience.iop.org/0957-4484/25/47/475705) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 129.173.72.87 This content was downloaded on 11/11/2014 at 07:06
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 25 (2014) 475705 (9pp)
doi:10.1088/0957-4484/25/47/475705
Transport and thermoelectric properties of polyaniline/reduced graphene oxide nanocomposites Rakibul Islam1,2, Roch Chan-Yu-King3, Jean-François Brun4,5, Carole Gors1,2, Ahmed Addad2,6, Michael Depriester2,7, Abdelhak Hadj-Sahraoui2,7 and Frederick Roussel1,2 1
University of Lille, 1-Sciences and Technologies, UFR de Physique, Bat P5, 59655 Villeneuve d’Ascq, France 2 COMUE Lille Nord de France, 59044 Villeneuve d’Ascq, France 3 University of Science and Arts of Oklahoma, Chickasha, OK 73018, USA 4 CNRS UPR 3079 CEHMTI, F-45071 Orléans, France 5 UFR Collegium Sciences et Techniques, Université d’Orléans, F-45067 Orléans, France 6 University of Lille, Sciences and Technologies, UMET, CNRS UMR 8207, Bat C6, F-59655 Villeneuve d’Ascq, France 7 ULCO, UDSMM, MREID, F-59140 Dunkerque, France E-mail: [email protected] Received 24 July 2014, revised 9 October 2014 Accepted for publication 10 October 2014 Published 7 November 2014 Abstract
Polyanilines (PANI)/reduced graphene oxide (RGO) nanocomposites are chemically synthesized. Their structure and morphology are characterized by scanning and transmission electron microscopies, x-ray diffraction and Raman spectroscopy. In addition, the nanocomposites’ electrical, thermal and thermoelectric (TE) transport characteristics are investigated as a function of RGO content. The power factor and figure of merit (ZT) of PANI/ RGO hybrids are deduced from measurements of the electrical conductivity (σ), Seebeck coefficient (α) and thermal conductivity (κ). Experimental results reveal that the properties of PANI/RGO composites are inherently dependent on the volume fraction of RGO. It is observed that electrical percolation follows a 2D conduction process which takes place for samples having 0.099 vol% RGO content. Unlike electrical conductivity, the thermal conductivity of PANI/RGO increases only slightly with the RGO fraction and is successfully fitted using a modified MGEMA model which provides an interfacial (PANI/RGO nanoplatelets) resistance (Rk) of 4.9 × 10−10 m2 K W−1. This low Rk value is attributed to good interactions between the planar geometry of RGO platelets and PANI aromatic rings through π–π stackings as evidenced by Raman spectroscopy and x-ray studies. Compared to that of pure PANI, the TE performance of PANI/RGO composites exhibits a ZT enhancement of two orders of magnitude. Keywords: nanocomposites, graphene, transport properties, electrical conductivity, thermal conductivity, polyaniline (Some figures may appear in colour only in the online journal) 1. Introduction
fabrication of energy harvesting systems or heating/cooling devices. The gradual advances realized over the years and further improvement in TE technology could find widespread commercial applications in cost-effective and sustainable energy conversion and recovery. So far most high-
Solid state thermoelectrics (TE) possess the capability to capture heat from natural sources or waste heat and convert it into electricity. These materials are good candidates for the 0957-4484/14/475705+09$33.00
1
© 2014 IOP Publishing Ltd Printed in the UK
Nanotechnology 25 (2014) 475705
R Islam et al
performance TE units are based on inorganics/metalloids or their composites [1]. Since the discovery of TE properties of intrinsically conducting polymers (ICPs) such as polyanilines (PANI), polypyrroles, polythiophenes (PTH, PEDOT:PSS), ICPs or their derivatives are intensively investigated because they possess relatively low thermal conductivity, good electrical conductivity and are easy to prepare, air stable, light weight and much less costly in their production compared to their inorganic counterparts [2, 3]. Recently, it has been demonstrated that the introduction of low dimensional nanostructures in composites materials such as PANI/Bi2Te3 [4] or PEDOT:PSS/Te [5] has significantly improved TE characteristics due to inter-component junctions, thereby leading to phonon scattering and hoping of charge carriers [6, 7]. In addition, the use of carbonaceous nanomaterials like carbon nanotubes (CNTs) or graphene in polymer-based TE composites have attracted much attention due their exceptional electrical, thermal and mechanical characteristics [8– 14]. In particular, it has been recently demonstrated that the change in nanotube characteristics (single or multiple-walled CNTs, oxidized or unoxidized) allows a better tuning of the TE properties in PANI/CNT composites [15]. As an extension of this line of research, we report herein, the preparation and TE performance of composites made of reduced graphene oxide (RGO) and PANI. Compared to the 1D structure of CNTs, graphene exhibits a much higher aspect ratio, and is expected to offer a much larger 2D/templating surface area for interaction with PANI. This feature should, in principle, significantly influence the transport properties and TE characteristics of the composites. The PANI/RGO hybrids are in situ chemically synthesized with a volume fraction of RGO ranging from 0 to 14 vol% (0–21 wt%). Scanning electron microscopy (SEM), transmission electron microscopy (TEM) and x-ray diffraction are used for analyses of morphology and structure of the products while Raman spectroscopy is employed to investigate the characteristic mode structure of each component and its evolution into the composites. Thermal (κ) and electrical (σ) conductivities are determined by photothermal IR radiometry and four probes method, respectively. Experimental data are fitted with physical models in order to assess transport processes. These data (κ, σ) combined with Seebeck (α) coefficient allow the determination of the power factor (PF) and figure of merit (ZT) of the as-prepared PANI/RGO TE materials.
Figure 1. (a) Density of PANI/RGO pellets as a function of RGO
content. The solid line is the best fit curve using a classical mixing rule to estimate the pellet density of the composites. (b)–(d) SEM micrographs of the cross section of fractured PANI/RGO pellets prepared with 0, 12.6 and 20.9 wt% RGO, respectively; white scale bars: (b)100 μm, (c) 100 μm and (d) 200 μm.
beaker and a solution of ammonium peroxydisulfate (0.808 g; 3.54 mmol) in 15.0 mL of 1.00 M HCSA was added. The black suspension was stirred overnight at room temperature, it was then suction filtered, and then subsequently washed with de-ionized water (∼200 mL) and acetone (∼40.0 mL) until a colourless filtrate was obtained. The product was highvacuumed dried (50.0 °C; 12 h) and stored in a desiccator. For electrical, thermal and TE investigations, PANI/RGO raw materials were first subjected to grinding in a mortar with a pestle, followed by cold pressing (450 MPa) of the resulting powder which provided pellets with a diameter of 12.8 mm and a thickness of ca. 200 μm. In figure 1(a), the density of the pellets is plotted as a function of RGO content exhibiting a linear relationship. Experimental data were successfully fitted using a classical mixing rule: ρPANI/RGO = xρPANI + (1 − x ) ρRGO ,
(1)
providing ρPANI = 1.24 g cm−3 and ρRGO = 2.24 g cm−3. These values are in good agreement with those reported in the literature. The cross section of fractured PANI/RGO pellets was investigated by SEM and the corresponding micrographs are shown in figures 1(b)–(d) which indicate that highly dense composites are formed under high compression treatment of powders.
2. Experimental
2.1. Characterization techniques
All chemicals were obtained from Aldrich Chemical except for RGO which was purchased from TimesNano (China). A typical procedure is as follows for the production of RGO/ PANI composite (12.6% by mass in RGO): into a glass tube were placed RGO (44.0 mg) and 20.0 mL of 1.00 M (aq) camphorsulfonic acid (HCSA). This mixture was sonicated one hour in order to improve the dispersion of RGO nanoplatelets. It was then mixed with freshly distilled aniline (0.300 mL; 3.30 mmol). The resulting suspension is further sonicated one hour after which time it was transferred into a
A scanning electron microscope (Hitachi S4700 operating at 6 kV, 10 mA) and a transmission electron microscope (FEI Tecnai G2 operating at 20 kV) were used for morphological and structural investigations of PANI/RGO hybrids. Raman microspectrometry was carried out on a Horiba Jobin Yvon Labram HR operating with a He–Ne laser (λ = 632.8 nm, P = 0.1 mW). Spectra were fitted using the freely available software FOCUS [16] x-ray diffractograms were recorded on a curved multidetector INEL-CPS 120 set up. Samples were 2
Nanotechnology 25 (2014) 475705
R Islam et al
thickness of the resulting platelets is from about 35 to 50 nm (figure 2(d)) indicating that each side of the RGO sheet (0.5–3 nm thick according to the manufacturer) is covered by a PANI layer with a thickness of ca. 15–20 nm. In order to get further insights into the interactions between PANI and RGO, Raman spectroscopy was carried out and the vibrational modes were obtained from fitting the experimental data (figure 3). The spectrum of pure PANI shows the typical bands of the doped polymer including C–H bending of the quinoid and benzenoid ring at 1145 and 1169 cm−1, respectively; symmetric C–N stretching at 1222 cm−1; C–N+ stretching at 1340 cm−1; C=N stretching of the quinoid ring at 1460 cm−1 and 1500 cm−1, and C=C stretching of the quinoid rings at 1591 cm−1. The assignment of these peaks are in good agreement with previous reports confirming that the synthesized product contains polyaniline [15, 19]. The spectrum of RGO exhibits the two typical bands at 1347 and 1597 cm−1, which are assigned to the D band (disorder-induced band) and the G band (in-plane stretching mode), respectively [20]. All spectra of the RGO-reinforced PANI nanocomposites exhibit the characteristic bands of each constituent. It is interesting to note that the intensity of the vibration at 1460 cm−1 (C=N quinoid stretching) increases with increasing RGO content demonstrating that incorporation of RGO in the composite promotes the quinoid form of PANI, i.e., a conjugated planar geometry is favored. This is further supported by taking the ratio of the intensity of C–H bending vibration in quinoid unit (1145 cm−1) over that of the corresponding benzenoid (1169 cm−1) [15, 19]. That is with respect to pure PANI, the quinoid/benzenoid integral ratio (1145/1169) increases from ca. 0.24 (pure PANI) to 0.50 (12.6 wt% RGO) and in turn reaches 0.72 (20.9 wt% RGO) which clearly evidences an important increase of the quinoid fraction in all hybrids. The quinoid rings in the nanocomposites are obviously stabilized through their π–π interactions with the basal planes of RGO. Thus, the RGO templates induce a greater polymer chain ordering in the hybrids which improves physical interactions and charge transfer processes between RGO and PANI. This finding corroborates with similar results in which graphene can engage in charge transfer interactions with ICPs [21] or neutral molecules [22, 23]. The x-ray diffraction studies of PANI/RGO composites and their individual constituents are presented in figure 4. Pristine RGO exhibits a broad peak centered at q = 15.5 ± 0.5 nm−1, corresponding to a distance of d = 2π/q = 0.405 ± 0.008 nm. This value is typical of disordered carbon materials with large inter-layer spacing (0.39 nm) [24]. In general, pure PANI exhibits both amorphous and ordered structures. Here, a broad peak is observed from q ca. 10 to 25 nm−1 indicating mainly an amorphous PANI morphology along with a sharper peak at 17.9 nm−1 which is associated with the (110) reflection of crystallographic planes of the orthorhombic structure reported for semi-crystalline PANI [25–28]. With increasing RGO fraction, the diffraction peaks related to PANI domains at q = 10.8 nm−1 ((010) reflection) and q = 17.9 nm−1((110) reflection) become more intense in the PANI/RGO
placed into Lindeman glass capillary tubes (diameter 0.70 mm) and irradiated with a collimated monochromic beam (λ = 0.154056 nm); samples were rotated around the vertical axis of the goniometer head. A conventional collinear four-probe method using a Cascade Microtech EP6 probe station equipped with a Keithley 2635 sourcemeter was employed to determine the dc electrical conductivity of pelletized PANI/RGO samples at room temperature. Measurements of TE PF were conducted on pressed pellets (average thickness of ca. 300 μm) with an in-house made apparatus [15]. Thermal parameters of pelletized PANI/RGO samples were determined using a photothermal radiometry setup described elsewhere [17, 18]. In this work, multiple repeats were performed for each sample in all measurements and mean average values are reported for electrical conductivity, Seebeck constants, and thermal conductivity.
3. Results and discussion 3.1. Morphological and structural characterizations
The morphological investigations of pristine RGO, pure PANI and their hybrids, performed by FE-SEM, are presented in figures 2(a)–(d), respectively. Micrograph 2(a) shows the typical nanosheet morphology of the as-received RGO. The chemically synthesized pure PANI exhibits interconnected nanofibers with an average diameter of 79 ± 15 nm (figure 2(b)). In figure 2(c) the morphology of PANI/RGO (4.9 wt%) shows a combination of micrometer platelets with sizes ranging from 0.5 to 2 μm and nanofibers (diameter of ca. 84 ± 20 nm) confirming the presence of RGO- and PANI-rich domains. With increasing RGO contents the composites exhibit predominantly a platelet-like morphology. As shown in figures 2(e)–(g) TEM of PANI/RGO (20.9 wt%) at increasing magnifications reveals that the graphene sheets are covered with a PANI layer resulting in a lamellar morphology which is consistent with SEM investigations. In the inset of figure 2(h), the electron diffraction data show six fold symmetric diffraction points confirming the hexagonal crystalline structure of the graphitic planes of RGO. However, it should be pointed out that the structural disorder of composites is evident from the lattice fringe patterns as seen in the diffraction pattern. This feature can be associated with a combination of the structural signatures of semi-crystalline PANI and multilayered graphenes. To further address this point, detailed x-ray diffraction and Raman spectroscopy investigations are presented below. The above micrographs show that the RGO sheets have an average size of 0.5–3 μm which is in agreement with the manufacturer's specifications. This confirms that RGO structural integrity is essentially unaffected by the experimental protocol. These sheets act as 2D templates for aniline monomer adsorptions at the initial stage of polymerization. This physisorption of monomers and ensuing nucleation of oligomers are most likely favoured by π–π interactions between the basal planes of RGO and anilines, thereby promoting the polymerization of anilines along both sides of the 2D plane of RGO. For example, the 3
Nanotechnology 25 (2014) 475705
R Islam et al
Figure 2. (a)–(d) Morphologies observed by FE-SEM for (a) reduced graphene oxide (RGO), (b) Polyaniline (PANI), (c) PANI/RGO (4.9 wt% RGO) and (d) PANI/RGO (20.9 wt% RGO). Insets: enlarged views; the white scale bar corresponds to 200 nm. (e)–(h) high resolution TEM images of the nanohybrid including 20.9 wt% RGO (scale bars: (e) and (f) 200 nm, (g) 50 nm and (h) 100 nm); inset (h): electron diffraction pattern taken from the position of the red square (scale bar 5 nm−1).
in particular in the (010) direction, i.e., parallel to the benzene rings of PANI chains. Thus, the improved PANI crystallinity in the composites reflects a polymer chain growth mechanism taking place predominantly along the (010) direction which is parallel to the aromatic rings of PANI backbone. This added feature re-enforces the afore-mentioned strong quinoid (PANI)-basal plane (RGO) π–π interactions as evidenced by Raman work.
diffractograms. In order to quantitatively assess this result, the diffractograms were fitted with Lorentzian curves. Specifically, the peak at q = 10.8 nm−1, which does not overlap with that of RGO, was used to track the evolution of the crystallinity of PANI in the nanocomposites. Its relative intensity and width at half maximum (FWHM) are plotted versus volume fraction of RGO in figure 3(d) from which it can be seen that the relative intensity increases with RGO content while the FWHM decreases. Thus, the incorporation of RGO in the composites enhances the crystallinity of PANI leading to more organized domains and longer correlation lengths (2.3 < L < 3.5 nm) as calculated from the Scherrer formula [27, 29]. These results demonstrate that RGO acts as templates for PANI growth and improve the PANI crystallinity,
3.2. Transport properties 3.2.1. Electrical properties. Figure 5(a) shows the evolution
of the electrical conductivity (σ) of PANI/RGO nanohybrids as a function of RGO volume fraction. σ obeys over at least 4
Nanotechnology 25 (2014) 475705
R Islam et al
RGO composites which is obviously favored by the plateletlike morphology observed by SEM. In addition, it has been reported that the conductivity of carboneous basedcomposites at a given temperature can be described by the behavior of a single tunnel junction [32, 33]. According to these references, the dependence of the composite conductivity on RGO fraction is in the form of: ln σ ~ ϕ−1/3.
(4)
Figure 5(b), displays a linear relationship between ln σ and ϕ−1/3 at ϕ > ϕc which suggests tunneling between RGO platelets separated by thin weakly conducting PANI layers. 3.2.2. Thermal properties. The evolution of the thermal
conductivity κ as a function of RGO volume fraction is shown in figure 6(a). It is interesting to note that κ follows essentially a linear dependence on ϕ without any obvious thermal percolation threshold as theoretically predicted for graphenebased nano-composites [34, 35]. A thermal conductivity enhancement (TCE = (κ-κPANI)/κPANI) up to 0.5 is obtained for the composite with RGO content as high as 14.1 vol% (20.9 wt%) which is almost two orders of magnitude lower than that of epoxy/multilayer graphene composites with similar volume fraction of filler [36]. This finding may suggest that the thermal conductivity of RGO (multi-layered) platelets is much lower compared to that of single layer graphenes which results in a limited rise of the effective thermal conductivity of the composites. In order to get further insights into thermal transport properties, a Maxwell–Garnett type effective medium approximation (MG-EMA), (which is valid at low filler volume fractions in composites), is used to fit the dependence of κ on volume fraction of RGO [36, 37]:
Figure 3. Raman spectra of RGO, PANI and PANI/RGO
composites; symbols are experimental data whereas solid and dashed lines are retrieved from fitting.
three orders of magnitude the percolation critical power law given as:
(
)t
σ = σ0 ϕ − ϕc ,
(2)
where ϕc is the volume percolation threshold, t the critical exponent, and σ0 the electrical conductivity prefactor. Fitting the experimental data to equation (2) provides ϕc = 0.099 vol%, t = 0.98, and σ0 = 15.3 s m−1. According to the model proposed by Li and Kim [30], the percolation threshold for a composite made of graphitic disk platelets dispersed in a polymer matrix can be estimated from: 3
ϕc = 27πD 2t /4 ( D + D IP ) ,
⎡ ⎤ ⎢ ⎥ 3κ PANI + 2ϕ ( κ RGO − κ PANI ) ⎥ , (5) κ = κ RGO ⎢ R k κ RGO κ PANI ϕ ⎥ ⎢ ⎢⎣ (3 − ϕ) κ RGO + ϕκ PANI + ⎥⎦ th
(3)
where D is the diameter of the platelets, t their thickness and DIP the distance between filler platelets. In this study the estimated DIP from SEM measurements is ∼2 × 15 nm (PANI thickness between two RGO sheets). Knowing that D is in the micrometer range, it can be reasonably assumed that DIP ≪ D. Using the manufacturer’s RGO specifications, D/t ∼ 1000 and equation (3), the volume percolation threshold is found to be within 2.32 < ϕc < 2.63 vol%. In this work, the percolation in PANI/RGO composites occurs when the RGO fraction, ϕc, is very close to 0.1 vol% (figure 5(a)) which is at least twenty times lower than any value from the above ϕc range. Such a low percolation threshold is not only related to the extremely high aspect ratio of the graphene sheets (D/t ∼ 1000) but is also dependent on the shape of the platelet geometry, i.e. disk (oblate) versus ellipsoid (prolate). Indeed, randomly oriented oblate ellipsoids with an aspect ratio of 1000 are predicted to have a geometric percolation threshold of 0.1 vol% which is consistent with our results [31]. The inset of figure 5(a) shows the linear relationship between Log(σ) versus Log(ϕ − ϕc) providing a slope of 1.07 which is close to the value of t retrieved from fitting the experimental data to equation (2). Unlike 3D percolation whose t > 1.8, here, the lower t value supports a 2D, i.e. in plane, conduction process for the PANI/
where κRGO and κPANI are the thermal conductivity of RGO and PANI, respectively, ϕ is the volume fraction of RGO, Rk is the interfacial resistance at the graphene/matrix interface, also known as the Kapitza resistance, and th is the thickness of the graphene nanoplatelets. As shown in figure 6, a good agreement is obtained between experimental data points and the fit (solid line) providing values of κPANI = 0.44 W m−1 K−1 and κRGO = 1.96 W m−1 K−1 which are in accord with data reported in the literature for PANI and RGO platelets [38–40]. In addition, it should be noted that thermal conductivity of layered materials like graphenes or graphites is direction dependent. For example, in plane and through plane thermal conductivities in graphene-based papers, with a graphene sheet size of 1 μm and 7 nm thick, exhibit drastic differences, i.e. κIP(in plane) ∼0.6 W m−1 K−1 and κTP(through plane) ∼8.7 W m−1 K−1 [41]. The fact that our computed κRGO lies in between the above values suggests that both in plane and through plane thermal transports could take place in the composites. 5
Nanotechnology 25 (2014) 475705
R Islam et al
Figure 5. (a) Electrical conductivity (σ) of PANI/RGO composites as
a function of RGO volume fraction; the symbols represent experimental data whereas the solid line is the best fit curve to equation (2). Inset: log–log plot of σ versus reduced volume fraction of RGO showing a linear relationship (correlation coefficient R = 0.990). (b) Plot of ln σ versus ϕ−1/3. Figure 4. X-ray diffraction patterns of (a) RGO, (b) PANI/RGO
(12.6 wt% RGO), (c) PANI, (d) Evolution of the relative intensity and FWHM of the (010) reflection as a function of RGO content in the composite.
The value of interfacial resistance for PANI/RGO hybrids is found to be Rk = 4.9 × 10−10 m2 K W−1 which is at least two orders of magnitude lower than that of polymer/ CNTs composites [42, 43]. Such a low Rk value can be attributed to good interactions between the planar geometry of RGO platelets and PANI aromatic rings through π–π stackings (figure 4). This reduces the interfacial (Kapitza) resistance and suggests an improved coupling between some of the excited phonon modes in graphene sheets and the polymer matrix as predicted by Molecular Dynamics and Density Functional Theory simulations [44]. Recalling that the interlayer spacing between graphene sheets, estimated by x-ray measurements, is d ∼ 0.4 nm and that the thickness th of RGO platelets retrieved from fitting (equation (5)) is ∼1.6 nm, the number of layers n = th/d in RGO platelets is ∼4 which is consistent with the manufacturer’s specifications. Figure 6(b) shows that the ratio of the effective thermal conductivity over that of pure PANI versus vol% of RGO (κ/κPANI ) reaches 1.4 at ϕ = 14.1 vol% RGO. Compared to simulated κ/κmatrix ratios in ordered graphene composites whose values are as high as 360, that of the PANI/RGO composites is much lower suggesting that heat transport mostly occurs randomly through the samples [44]. An alternative explanation to the
Figure 6. (a) Thermal conductivity κ as a function of RGO volume fraction; symbols are experimental data point and the solid line is the best fit curve to equation (5) (correlation coefficient R2 = 0.996). (b) Plot of κ/κPANI versus vol% RGO. 6
Nanotechnology 25 (2014) 475705
R Islam et al
that of pure PANI. It is also interesting to note that the sharp rise in ZT occurs at low RGO loading (
Search
Collections
Journals
About
Contact us
My IOPscience
Transport and thermoelectric properties of polyaniline/reduced graphene oxide nanocomposites
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Nanotechnology 25 475705 (http://iopscience.iop.org/0957-4484/25/47/475705) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 129.173.72.87 This content was downloaded on 11/11/2014 at 07:06
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 25 (2014) 475705 (9pp)
doi:10.1088/0957-4484/25/47/475705
Transport and thermoelectric properties of polyaniline/reduced graphene oxide nanocomposites Rakibul Islam1,2, Roch Chan-Yu-King3, Jean-François Brun4,5, Carole Gors1,2, Ahmed Addad2,6, Michael Depriester2,7, Abdelhak Hadj-Sahraoui2,7 and Frederick Roussel1,2 1
University of Lille, 1-Sciences and Technologies, UFR de Physique, Bat P5, 59655 Villeneuve d’Ascq, France 2 COMUE Lille Nord de France, 59044 Villeneuve d’Ascq, France 3 University of Science and Arts of Oklahoma, Chickasha, OK 73018, USA 4 CNRS UPR 3079 CEHMTI, F-45071 Orléans, France 5 UFR Collegium Sciences et Techniques, Université d’Orléans, F-45067 Orléans, France 6 University of Lille, Sciences and Technologies, UMET, CNRS UMR 8207, Bat C6, F-59655 Villeneuve d’Ascq, France 7 ULCO, UDSMM, MREID, F-59140 Dunkerque, France E-mail: [email protected] Received 24 July 2014, revised 9 October 2014 Accepted for publication 10 October 2014 Published 7 November 2014 Abstract
Polyanilines (PANI)/reduced graphene oxide (RGO) nanocomposites are chemically synthesized. Their structure and morphology are characterized by scanning and transmission electron microscopies, x-ray diffraction and Raman spectroscopy. In addition, the nanocomposites’ electrical, thermal and thermoelectric (TE) transport characteristics are investigated as a function of RGO content. The power factor and figure of merit (ZT) of PANI/ RGO hybrids are deduced from measurements of the electrical conductivity (σ), Seebeck coefficient (α) and thermal conductivity (κ). Experimental results reveal that the properties of PANI/RGO composites are inherently dependent on the volume fraction of RGO. It is observed that electrical percolation follows a 2D conduction process which takes place for samples having 0.099 vol% RGO content. Unlike electrical conductivity, the thermal conductivity of PANI/RGO increases only slightly with the RGO fraction and is successfully fitted using a modified MGEMA model which provides an interfacial (PANI/RGO nanoplatelets) resistance (Rk) of 4.9 × 10−10 m2 K W−1. This low Rk value is attributed to good interactions between the planar geometry of RGO platelets and PANI aromatic rings through π–π stackings as evidenced by Raman spectroscopy and x-ray studies. Compared to that of pure PANI, the TE performance of PANI/RGO composites exhibits a ZT enhancement of two orders of magnitude. Keywords: nanocomposites, graphene, transport properties, electrical conductivity, thermal conductivity, polyaniline (Some figures may appear in colour only in the online journal) 1. Introduction
fabrication of energy harvesting systems or heating/cooling devices. The gradual advances realized over the years and further improvement in TE technology could find widespread commercial applications in cost-effective and sustainable energy conversion and recovery. So far most high-
Solid state thermoelectrics (TE) possess the capability to capture heat from natural sources or waste heat and convert it into electricity. These materials are good candidates for the 0957-4484/14/475705+09$33.00
1
© 2014 IOP Publishing Ltd Printed in the UK
Nanotechnology 25 (2014) 475705
R Islam et al
performance TE units are based on inorganics/metalloids or their composites [1]. Since the discovery of TE properties of intrinsically conducting polymers (ICPs) such as polyanilines (PANI), polypyrroles, polythiophenes (PTH, PEDOT:PSS), ICPs or their derivatives are intensively investigated because they possess relatively low thermal conductivity, good electrical conductivity and are easy to prepare, air stable, light weight and much less costly in their production compared to their inorganic counterparts [2, 3]. Recently, it has been demonstrated that the introduction of low dimensional nanostructures in composites materials such as PANI/Bi2Te3 [4] or PEDOT:PSS/Te [5] has significantly improved TE characteristics due to inter-component junctions, thereby leading to phonon scattering and hoping of charge carriers [6, 7]. In addition, the use of carbonaceous nanomaterials like carbon nanotubes (CNTs) or graphene in polymer-based TE composites have attracted much attention due their exceptional electrical, thermal and mechanical characteristics [8– 14]. In particular, it has been recently demonstrated that the change in nanotube characteristics (single or multiple-walled CNTs, oxidized or unoxidized) allows a better tuning of the TE properties in PANI/CNT composites [15]. As an extension of this line of research, we report herein, the preparation and TE performance of composites made of reduced graphene oxide (RGO) and PANI. Compared to the 1D structure of CNTs, graphene exhibits a much higher aspect ratio, and is expected to offer a much larger 2D/templating surface area for interaction with PANI. This feature should, in principle, significantly influence the transport properties and TE characteristics of the composites. The PANI/RGO hybrids are in situ chemically synthesized with a volume fraction of RGO ranging from 0 to 14 vol% (0–21 wt%). Scanning electron microscopy (SEM), transmission electron microscopy (TEM) and x-ray diffraction are used for analyses of morphology and structure of the products while Raman spectroscopy is employed to investigate the characteristic mode structure of each component and its evolution into the composites. Thermal (κ) and electrical (σ) conductivities are determined by photothermal IR radiometry and four probes method, respectively. Experimental data are fitted with physical models in order to assess transport processes. These data (κ, σ) combined with Seebeck (α) coefficient allow the determination of the power factor (PF) and figure of merit (ZT) of the as-prepared PANI/RGO TE materials.
Figure 1. (a) Density of PANI/RGO pellets as a function of RGO
content. The solid line is the best fit curve using a classical mixing rule to estimate the pellet density of the composites. (b)–(d) SEM micrographs of the cross section of fractured PANI/RGO pellets prepared with 0, 12.6 and 20.9 wt% RGO, respectively; white scale bars: (b)100 μm, (c) 100 μm and (d) 200 μm.
beaker and a solution of ammonium peroxydisulfate (0.808 g; 3.54 mmol) in 15.0 mL of 1.00 M HCSA was added. The black suspension was stirred overnight at room temperature, it was then suction filtered, and then subsequently washed with de-ionized water (∼200 mL) and acetone (∼40.0 mL) until a colourless filtrate was obtained. The product was highvacuumed dried (50.0 °C; 12 h) and stored in a desiccator. For electrical, thermal and TE investigations, PANI/RGO raw materials were first subjected to grinding in a mortar with a pestle, followed by cold pressing (450 MPa) of the resulting powder which provided pellets with a diameter of 12.8 mm and a thickness of ca. 200 μm. In figure 1(a), the density of the pellets is plotted as a function of RGO content exhibiting a linear relationship. Experimental data were successfully fitted using a classical mixing rule: ρPANI/RGO = xρPANI + (1 − x ) ρRGO ,
(1)
providing ρPANI = 1.24 g cm−3 and ρRGO = 2.24 g cm−3. These values are in good agreement with those reported in the literature. The cross section of fractured PANI/RGO pellets was investigated by SEM and the corresponding micrographs are shown in figures 1(b)–(d) which indicate that highly dense composites are formed under high compression treatment of powders.
2. Experimental
2.1. Characterization techniques
All chemicals were obtained from Aldrich Chemical except for RGO which was purchased from TimesNano (China). A typical procedure is as follows for the production of RGO/ PANI composite (12.6% by mass in RGO): into a glass tube were placed RGO (44.0 mg) and 20.0 mL of 1.00 M (aq) camphorsulfonic acid (HCSA). This mixture was sonicated one hour in order to improve the dispersion of RGO nanoplatelets. It was then mixed with freshly distilled aniline (0.300 mL; 3.30 mmol). The resulting suspension is further sonicated one hour after which time it was transferred into a
A scanning electron microscope (Hitachi S4700 operating at 6 kV, 10 mA) and a transmission electron microscope (FEI Tecnai G2 operating at 20 kV) were used for morphological and structural investigations of PANI/RGO hybrids. Raman microspectrometry was carried out on a Horiba Jobin Yvon Labram HR operating with a He–Ne laser (λ = 632.8 nm, P = 0.1 mW). Spectra were fitted using the freely available software FOCUS [16] x-ray diffractograms were recorded on a curved multidetector INEL-CPS 120 set up. Samples were 2
Nanotechnology 25 (2014) 475705
R Islam et al
thickness of the resulting platelets is from about 35 to 50 nm (figure 2(d)) indicating that each side of the RGO sheet (0.5–3 nm thick according to the manufacturer) is covered by a PANI layer with a thickness of ca. 15–20 nm. In order to get further insights into the interactions between PANI and RGO, Raman spectroscopy was carried out and the vibrational modes were obtained from fitting the experimental data (figure 3). The spectrum of pure PANI shows the typical bands of the doped polymer including C–H bending of the quinoid and benzenoid ring at 1145 and 1169 cm−1, respectively; symmetric C–N stretching at 1222 cm−1; C–N+ stretching at 1340 cm−1; C=N stretching of the quinoid ring at 1460 cm−1 and 1500 cm−1, and C=C stretching of the quinoid rings at 1591 cm−1. The assignment of these peaks are in good agreement with previous reports confirming that the synthesized product contains polyaniline [15, 19]. The spectrum of RGO exhibits the two typical bands at 1347 and 1597 cm−1, which are assigned to the D band (disorder-induced band) and the G band (in-plane stretching mode), respectively [20]. All spectra of the RGO-reinforced PANI nanocomposites exhibit the characteristic bands of each constituent. It is interesting to note that the intensity of the vibration at 1460 cm−1 (C=N quinoid stretching) increases with increasing RGO content demonstrating that incorporation of RGO in the composite promotes the quinoid form of PANI, i.e., a conjugated planar geometry is favored. This is further supported by taking the ratio of the intensity of C–H bending vibration in quinoid unit (1145 cm−1) over that of the corresponding benzenoid (1169 cm−1) [15, 19]. That is with respect to pure PANI, the quinoid/benzenoid integral ratio (1145/1169) increases from ca. 0.24 (pure PANI) to 0.50 (12.6 wt% RGO) and in turn reaches 0.72 (20.9 wt% RGO) which clearly evidences an important increase of the quinoid fraction in all hybrids. The quinoid rings in the nanocomposites are obviously stabilized through their π–π interactions with the basal planes of RGO. Thus, the RGO templates induce a greater polymer chain ordering in the hybrids which improves physical interactions and charge transfer processes between RGO and PANI. This finding corroborates with similar results in which graphene can engage in charge transfer interactions with ICPs [21] or neutral molecules [22, 23]. The x-ray diffraction studies of PANI/RGO composites and their individual constituents are presented in figure 4. Pristine RGO exhibits a broad peak centered at q = 15.5 ± 0.5 nm−1, corresponding to a distance of d = 2π/q = 0.405 ± 0.008 nm. This value is typical of disordered carbon materials with large inter-layer spacing (0.39 nm) [24]. In general, pure PANI exhibits both amorphous and ordered structures. Here, a broad peak is observed from q ca. 10 to 25 nm−1 indicating mainly an amorphous PANI morphology along with a sharper peak at 17.9 nm−1 which is associated with the (110) reflection of crystallographic planes of the orthorhombic structure reported for semi-crystalline PANI [25–28]. With increasing RGO fraction, the diffraction peaks related to PANI domains at q = 10.8 nm−1 ((010) reflection) and q = 17.9 nm−1((110) reflection) become more intense in the PANI/RGO
placed into Lindeman glass capillary tubes (diameter 0.70 mm) and irradiated with a collimated monochromic beam (λ = 0.154056 nm); samples were rotated around the vertical axis of the goniometer head. A conventional collinear four-probe method using a Cascade Microtech EP6 probe station equipped with a Keithley 2635 sourcemeter was employed to determine the dc electrical conductivity of pelletized PANI/RGO samples at room temperature. Measurements of TE PF were conducted on pressed pellets (average thickness of ca. 300 μm) with an in-house made apparatus [15]. Thermal parameters of pelletized PANI/RGO samples were determined using a photothermal radiometry setup described elsewhere [17, 18]. In this work, multiple repeats were performed for each sample in all measurements and mean average values are reported for electrical conductivity, Seebeck constants, and thermal conductivity.
3. Results and discussion 3.1. Morphological and structural characterizations
The morphological investigations of pristine RGO, pure PANI and their hybrids, performed by FE-SEM, are presented in figures 2(a)–(d), respectively. Micrograph 2(a) shows the typical nanosheet morphology of the as-received RGO. The chemically synthesized pure PANI exhibits interconnected nanofibers with an average diameter of 79 ± 15 nm (figure 2(b)). In figure 2(c) the morphology of PANI/RGO (4.9 wt%) shows a combination of micrometer platelets with sizes ranging from 0.5 to 2 μm and nanofibers (diameter of ca. 84 ± 20 nm) confirming the presence of RGO- and PANI-rich domains. With increasing RGO contents the composites exhibit predominantly a platelet-like morphology. As shown in figures 2(e)–(g) TEM of PANI/RGO (20.9 wt%) at increasing magnifications reveals that the graphene sheets are covered with a PANI layer resulting in a lamellar morphology which is consistent with SEM investigations. In the inset of figure 2(h), the electron diffraction data show six fold symmetric diffraction points confirming the hexagonal crystalline structure of the graphitic planes of RGO. However, it should be pointed out that the structural disorder of composites is evident from the lattice fringe patterns as seen in the diffraction pattern. This feature can be associated with a combination of the structural signatures of semi-crystalline PANI and multilayered graphenes. To further address this point, detailed x-ray diffraction and Raman spectroscopy investigations are presented below. The above micrographs show that the RGO sheets have an average size of 0.5–3 μm which is in agreement with the manufacturer's specifications. This confirms that RGO structural integrity is essentially unaffected by the experimental protocol. These sheets act as 2D templates for aniline monomer adsorptions at the initial stage of polymerization. This physisorption of monomers and ensuing nucleation of oligomers are most likely favoured by π–π interactions between the basal planes of RGO and anilines, thereby promoting the polymerization of anilines along both sides of the 2D plane of RGO. For example, the 3
Nanotechnology 25 (2014) 475705
R Islam et al
Figure 2. (a)–(d) Morphologies observed by FE-SEM for (a) reduced graphene oxide (RGO), (b) Polyaniline (PANI), (c) PANI/RGO (4.9 wt% RGO) and (d) PANI/RGO (20.9 wt% RGO). Insets: enlarged views; the white scale bar corresponds to 200 nm. (e)–(h) high resolution TEM images of the nanohybrid including 20.9 wt% RGO (scale bars: (e) and (f) 200 nm, (g) 50 nm and (h) 100 nm); inset (h): electron diffraction pattern taken from the position of the red square (scale bar 5 nm−1).
in particular in the (010) direction, i.e., parallel to the benzene rings of PANI chains. Thus, the improved PANI crystallinity in the composites reflects a polymer chain growth mechanism taking place predominantly along the (010) direction which is parallel to the aromatic rings of PANI backbone. This added feature re-enforces the afore-mentioned strong quinoid (PANI)-basal plane (RGO) π–π interactions as evidenced by Raman work.
diffractograms. In order to quantitatively assess this result, the diffractograms were fitted with Lorentzian curves. Specifically, the peak at q = 10.8 nm−1, which does not overlap with that of RGO, was used to track the evolution of the crystallinity of PANI in the nanocomposites. Its relative intensity and width at half maximum (FWHM) are plotted versus volume fraction of RGO in figure 3(d) from which it can be seen that the relative intensity increases with RGO content while the FWHM decreases. Thus, the incorporation of RGO in the composites enhances the crystallinity of PANI leading to more organized domains and longer correlation lengths (2.3 < L < 3.5 nm) as calculated from the Scherrer formula [27, 29]. These results demonstrate that RGO acts as templates for PANI growth and improve the PANI crystallinity,
3.2. Transport properties 3.2.1. Electrical properties. Figure 5(a) shows the evolution
of the electrical conductivity (σ) of PANI/RGO nanohybrids as a function of RGO volume fraction. σ obeys over at least 4
Nanotechnology 25 (2014) 475705
R Islam et al
RGO composites which is obviously favored by the plateletlike morphology observed by SEM. In addition, it has been reported that the conductivity of carboneous basedcomposites at a given temperature can be described by the behavior of a single tunnel junction [32, 33]. According to these references, the dependence of the composite conductivity on RGO fraction is in the form of: ln σ ~ ϕ−1/3.
(4)
Figure 5(b), displays a linear relationship between ln σ and ϕ−1/3 at ϕ > ϕc which suggests tunneling between RGO platelets separated by thin weakly conducting PANI layers. 3.2.2. Thermal properties. The evolution of the thermal
conductivity κ as a function of RGO volume fraction is shown in figure 6(a). It is interesting to note that κ follows essentially a linear dependence on ϕ without any obvious thermal percolation threshold as theoretically predicted for graphenebased nano-composites [34, 35]. A thermal conductivity enhancement (TCE = (κ-κPANI)/κPANI) up to 0.5 is obtained for the composite with RGO content as high as 14.1 vol% (20.9 wt%) which is almost two orders of magnitude lower than that of epoxy/multilayer graphene composites with similar volume fraction of filler [36]. This finding may suggest that the thermal conductivity of RGO (multi-layered) platelets is much lower compared to that of single layer graphenes which results in a limited rise of the effective thermal conductivity of the composites. In order to get further insights into thermal transport properties, a Maxwell–Garnett type effective medium approximation (MG-EMA), (which is valid at low filler volume fractions in composites), is used to fit the dependence of κ on volume fraction of RGO [36, 37]:
Figure 3. Raman spectra of RGO, PANI and PANI/RGO
composites; symbols are experimental data whereas solid and dashed lines are retrieved from fitting.
three orders of magnitude the percolation critical power law given as:
(
)t
σ = σ0 ϕ − ϕc ,
(2)
where ϕc is the volume percolation threshold, t the critical exponent, and σ0 the electrical conductivity prefactor. Fitting the experimental data to equation (2) provides ϕc = 0.099 vol%, t = 0.98, and σ0 = 15.3 s m−1. According to the model proposed by Li and Kim [30], the percolation threshold for a composite made of graphitic disk platelets dispersed in a polymer matrix can be estimated from: 3
ϕc = 27πD 2t /4 ( D + D IP ) ,
⎡ ⎤ ⎢ ⎥ 3κ PANI + 2ϕ ( κ RGO − κ PANI ) ⎥ , (5) κ = κ RGO ⎢ R k κ RGO κ PANI ϕ ⎥ ⎢ ⎢⎣ (3 − ϕ) κ RGO + ϕκ PANI + ⎥⎦ th
(3)
where D is the diameter of the platelets, t their thickness and DIP the distance between filler platelets. In this study the estimated DIP from SEM measurements is ∼2 × 15 nm (PANI thickness between two RGO sheets). Knowing that D is in the micrometer range, it can be reasonably assumed that DIP ≪ D. Using the manufacturer’s RGO specifications, D/t ∼ 1000 and equation (3), the volume percolation threshold is found to be within 2.32 < ϕc < 2.63 vol%. In this work, the percolation in PANI/RGO composites occurs when the RGO fraction, ϕc, is very close to 0.1 vol% (figure 5(a)) which is at least twenty times lower than any value from the above ϕc range. Such a low percolation threshold is not only related to the extremely high aspect ratio of the graphene sheets (D/t ∼ 1000) but is also dependent on the shape of the platelet geometry, i.e. disk (oblate) versus ellipsoid (prolate). Indeed, randomly oriented oblate ellipsoids with an aspect ratio of 1000 are predicted to have a geometric percolation threshold of 0.1 vol% which is consistent with our results [31]. The inset of figure 5(a) shows the linear relationship between Log(σ) versus Log(ϕ − ϕc) providing a slope of 1.07 which is close to the value of t retrieved from fitting the experimental data to equation (2). Unlike 3D percolation whose t > 1.8, here, the lower t value supports a 2D, i.e. in plane, conduction process for the PANI/
where κRGO and κPANI are the thermal conductivity of RGO and PANI, respectively, ϕ is the volume fraction of RGO, Rk is the interfacial resistance at the graphene/matrix interface, also known as the Kapitza resistance, and th is the thickness of the graphene nanoplatelets. As shown in figure 6, a good agreement is obtained between experimental data points and the fit (solid line) providing values of κPANI = 0.44 W m−1 K−1 and κRGO = 1.96 W m−1 K−1 which are in accord with data reported in the literature for PANI and RGO platelets [38–40]. In addition, it should be noted that thermal conductivity of layered materials like graphenes or graphites is direction dependent. For example, in plane and through plane thermal conductivities in graphene-based papers, with a graphene sheet size of 1 μm and 7 nm thick, exhibit drastic differences, i.e. κIP(in plane) ∼0.6 W m−1 K−1 and κTP(through plane) ∼8.7 W m−1 K−1 [41]. The fact that our computed κRGO lies in between the above values suggests that both in plane and through plane thermal transports could take place in the composites. 5
Nanotechnology 25 (2014) 475705
R Islam et al
Figure 5. (a) Electrical conductivity (σ) of PANI/RGO composites as
a function of RGO volume fraction; the symbols represent experimental data whereas the solid line is the best fit curve to equation (2). Inset: log–log plot of σ versus reduced volume fraction of RGO showing a linear relationship (correlation coefficient R = 0.990). (b) Plot of ln σ versus ϕ−1/3. Figure 4. X-ray diffraction patterns of (a) RGO, (b) PANI/RGO
(12.6 wt% RGO), (c) PANI, (d) Evolution of the relative intensity and FWHM of the (010) reflection as a function of RGO content in the composite.
The value of interfacial resistance for PANI/RGO hybrids is found to be Rk = 4.9 × 10−10 m2 K W−1 which is at least two orders of magnitude lower than that of polymer/ CNTs composites [42, 43]. Such a low Rk value can be attributed to good interactions between the planar geometry of RGO platelets and PANI aromatic rings through π–π stackings (figure 4). This reduces the interfacial (Kapitza) resistance and suggests an improved coupling between some of the excited phonon modes in graphene sheets and the polymer matrix as predicted by Molecular Dynamics and Density Functional Theory simulations [44]. Recalling that the interlayer spacing between graphene sheets, estimated by x-ray measurements, is d ∼ 0.4 nm and that the thickness th of RGO platelets retrieved from fitting (equation (5)) is ∼1.6 nm, the number of layers n = th/d in RGO platelets is ∼4 which is consistent with the manufacturer’s specifications. Figure 6(b) shows that the ratio of the effective thermal conductivity over that of pure PANI versus vol% of RGO (κ/κPANI ) reaches 1.4 at ϕ = 14.1 vol% RGO. Compared to simulated κ/κmatrix ratios in ordered graphene composites whose values are as high as 360, that of the PANI/RGO composites is much lower suggesting that heat transport mostly occurs randomly through the samples [44]. An alternative explanation to the
Figure 6. (a) Thermal conductivity κ as a function of RGO volume fraction; symbols are experimental data point and the solid line is the best fit curve to equation (5) (correlation coefficient R2 = 0.996). (b) Plot of κ/κPANI versus vol% RGO. 6
Nanotechnology 25 (2014) 475705
R Islam et al
that of pure PANI. It is also interesting to note that the sharp rise in ZT occurs at low RGO loading (
