TGA analysis of polypropylene–carbon nanofibers composites

Available online at www.sciencedirect.com Polymer Degradation and Stability 93 (2008) 871e876 www.elsevier.com/locate/polydegstab TGA analysis of polypropyleneecarbon nanofibers composites Magdalena Chipara a, Karen Lozano b, Anna Hernandez b, Mircea Chipara a,* a Department of Physics and Geology, University of Texas Pan American, 1201 West University Drive, Edinburg, 78541 TX, USA b Department of Mechanical Engineering, University of Texas Pan American, Edinburg, 78541 TX, USA Received 22 August 2007; received in revised form 26 December 2007; accepted 7 January 2008 Available online 12 January 2008 Abstract TGA investigations on the thermal degradation of isotactic polypropyleneevapor grown carbon nanofibers composites in nitrogen are reported. The mass evolution as a function of temperature is a single sigmoid for both polypropylene and polypropylene loaded with carbon nanofibers. The inflection temperature of these sigmoids increases as the concentration of carbon nanofibers is increased. The width of the degradation process narrows as the concentration of carbon nanofibers is increased due to a better homogenization of the local temperature provided by the high thermal conductivity of carbon nanofibers. Thermogravimetric analysis data indicate the formation of polymerecarbon nanofiber interface. Based on TGA data, a two-layer structure is proposed for carbon nanofibersepolypropylene interface. The external layer is soft and has a thickness of about 102 nm that confines most polymer molecules in interaction with nanofibers. The core layer is rigid and has a thickness of the order of few nanometers. Ó 2008 Elsevier Ltd. All rights reserved. Keywords: Polymer; Carbon nanofibers; Composite; Thermogravimetric analysis; Polymerenanofiber interaction; Polypropylene 1. Introduction Isotactic polypropylene (IPP) has three crystalline phases (a, b, g) [1e5] and a mesomorphic smectic phase [6]. The chain conformation of each crystalline phase is a 31 helix. The polymorphism of polypropylene (PP) is surfacing from different packing of the helix into the unit cell, as observed by wide angle X-ray scattering (WAXS). The a phase always occurs in the regular processed PP as reported by Natta and Corradini [7]. Special crystallization conditions are required [8] to obtain higher amounts of b phase. The g phase is not enough explored; only recently, its face-centered orthorhombic cell has been clarified [9,10]. The thermal degradation of isotactic polypropyleneecarbon nanofibers (IPPeCNFs) is expected to be the result of the superposition of two sigmoids, one representing the thermal degradation of the polymer and the other assigned to the thermal degradation of nanofibers. Due to the absence of * Corresponding author. Tel.: þ1 956 381 2152; fax: þ1 956 381 2423. E-mail address: [email protected] (M. Chipara). 0141-3910/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.polymdegradstab.2008.01.001 oxygen and owing to the high thermal stability of carbon nanofibers in inert atmosphere, only the degradation of the polymer would be sensed by TGA (in the temperature range 50  C to 1000  C) [11,12]. 2. Experimental techniques IPP type Marlex HLN-120-01 (Philips Sumika Polypropylene Company) with density 0.906 g/cm3 and melt flow rate at 230  C of 12 g/10 min has been utilized as polymeric matrix. Vapor grown carbon nanofibers VGCNFs (PR24AG) with diameters ranging between 60 and 100 nm and lengths between 30,000 and 100,000 nm have been supplied by Pyrograf Products, Inc and used to reinforce the polymeric matrix, according to the process developed by Lozano and Barrera [13]. In this study purified vapor grown carbon fibers (VGCF) were mixed into the IPP matrix to form nanofiber composites. The purification of VGCNFs implied refluxing VGCNTs in dichloromethane and deionized water followed by vacuum filtering (for 24 h) and drying at 110  C for at least 24 h. Lozano et al. [14] showed 872 M. Chipara et al. / Polymer Degradation and Stability 93 (2008) 871e876 that the purification process does not affect the length of VGCNFs. High-shear mixing has been used to disperse the VGCNFs homogenously throughout the IPP matrix. The mixing has been performed by a HAAKE Rheomix at 180  C for 9 min with a speed of 65 rpm followed by an additional mixing at 90 rpm for 5 min. Composites loaded with various amounts of VGCNFs (0%, 1%, 2.5%, 5%, 7.5%, 10%, 15%, and 20% wt.) have been prepared. The as obtained samples have been hot pressed into sheets with a thickness of about 0.6 mm at about 180  C and at a weight of 9000 kg for 100 s. TGA investigations on the thermal stability of IPPeVGCNF composites in nitrogen, have been performed using a TA Instrument (TGA Q500). Additional data regarding the dispersion of VGCNFs within the polymeric matrix (IPP) have been obtained by using a Philips Transmission Electron Microscope operating at 200 kV. 3. Experimental results and discussions Fig. 1 shows the temperature dependence of the sample mass for IPPeVGCNFs composites, for different loading concentrations of VGCNFs (ranging from 0% wt. up to 20% wt. VGCNFs). In order to ensure a good reproducibility of experimental data all samples had a weight of about 10.0  1.5 mg. The experimental errors for the as recorded TGA data sets were found to be better than 1%. To observe easily the general trends and the differences between different TGA data sets, all data sets (excepting the one for pristine polymer) have been shifted upwards along the vertical axis by 4 mg relative to the previous data set. The experimental data were not shifted along the horizontal axis. The temperature dependence of the composite’s mass obeys the usual sigmoid like shape [11,12]. The mass evolution as a function of temperature for both the pristine IPP and the IPPeVGCNFs composites is described by single asymmetric sigmoid. From Fig. 1 it is noticed that the thermal degradation of IPP shifts towards higher temperatures as the concentration of VGNCFs is increased. This substantiates the interactions between macromolecular chains and VGCNFs revealing the formation of an IPPeVGCNF interface with enhanced thermal stability. Within the experimental errors, no additional mass loss has been noticed as the temperature of the IPPeVGCNF composites was raised from 700  C to 1000  C. In order to increase the resolution of TGA analysis, the first derivative of the mass loss versus the degradation temperature has been analyzed. The as obtained dependencies are collected in Fig. 2. As in the previous figure, all data sets shown in Fig. 2 were shifted upwards by 0.1 g/ C relative to the previous data set. The data for pristine polymer were not shifted, without any horizontal shift. The dependence of the mass loss derivative versus temperature should present a bell like shape (Lorentzian or Gaussian) as the derivation process converts the inflection point into an extreme point. From Fig. 2 it is Fig. 1. The temperature dependence of the weight of IPPeVGCNFs composites. The single sigmoidal shape is noticed and the shift of the temperature at which the speed of thermal degradation is maximum as the loading with VGCNFs is increased is emphasized by the dotted line. The inset shows the TGA data for the whole range of temperatures (50  Ce1000  C). Please notice a vertical offset of 4 mg and no horizontal offset. M. Chipara et al. / Polymer Degradation and Stability 93 (2008) 871e876 873 Fig. 2. The dependence of the first derivative of the TGA signal (versus the temperature) on temperature in the degradation region. The inset shows the temperature dependence of the TGA signal for the whole temperature range (50  Ce1000  C). Please notice a vertical offset of 0.1 g/ C and no horizontal offset. observed that for all spectra, the derivative of the residual mass (versus the temperature of the sample) as a function of temperature, has a single maximum, is asymmetric, and shifts towards higher temperatures as the concentration of VGCNF is increased. The derivative of the residual mass of the sample as a function of sample temperature has been fitted by an extended Breit Wigner Fano Lorentz line shape: IðeÞ ¼ ðP3 þ P4 eÞ2 P5 32 þ P6 at which the mass loss rate is maximum (named also inflection temperature e TI), has been estimated. Fig. 4 depicts the dependence of TI on the concentration of VGCNFs. It is noticed from Fig. 4 that TI rises as the concentration of VGCNFs dispersed within the polymeric matrix is increased. The dependence of the inflection temperature on the concentration of VGCNFs is well described by the equation: ð1Þ with: e¼ x  P1 P2 ð2Þ where x is the temperature, P1 defines the temperature at which the mass loss rate is maximum, P2 identifies the width of the derivative, P3, P4, P5 and P6 are constants. The simple Lorentzian shape is symmetric and cannot fit accurately the experimental data. The proposed lineshape degenerates into a Lorentzian like line shape for P4 ¼ 0 and P5/P6 ¼ 1. The Breit Wigner Fano has been frequently used [15e17] in various spectroscopic techniques (such as Raman) to simulate asymmetric resonance lines. As it is noticed from Fig. 3, the experimental data are fairly well fitted by the proposed line shape. From this equation the maximum of dm/dT, which actually represents the temperature Fig. 3. The first derivative of the sample’s mass relative to the sample’s temperature as a function of temperature. The black narrow line represents the as recorded data and the gray line the best fit (obtained by using Eq. (1)). 874 M. Chipara et al. / Polymer Degradation and Stability 93 (2008) 871e876 Fig. 4. The effect of the loading with VGCNFs on the temperature at which the thermal degradation is maximum (left axis, open circles). The effect of the loading with VGCNFs on the width of the degradation process (right axis, stars). Fig. 5. The dependence of the fraction of polymer chains captured in the elastic layer of the interface on the polymer loading with VGCNFs (left axis, filled squares). The dependence of the fraction of polymer chains (actually mostly C atoms) captured in the hard layer of the interface on the polymer loading with VGCNFs (right axis, stars). TI ¼ T1 þ T2 expðCxÞ of the polymer located in the interface, mIx, by using the expression: ð3Þ where T1, T2, and C are fitting constants and x is the weight fraction of VGCNFs. It is noticed that T1 þ T2 ¼ T0I , where T0I is the inflection temperature for the pristine polymer. The experimental data are well fitted by this equation (see the dotted line in Fig. 4). The parameters corresponding to the best fit are T1 ¼ 21,616  1 K, T2 ¼ 22,078  1 K, and C ¼ 0.00003  0.00001. This corresponds to T0I ¼ 462  1 K, which is an acceptable value and confirms the proposed equation (see Fig. 4). The shift of TI as the concentration of VGCNFs is increased demonstrates the enhancement of the thermal stability of IPP upon loading with VGCNFs and confirms the formation of IPPeVGCNFs interface. It is this interface that is responsible for the overall increase of the thermal stability of IPPeVGCNFs composites. The width (calculated between the inflection points of the derivative) of the first derivative of the mass loss (versus temperature) depends on the concentration of VGCNFs (see Fig. 4). It is noticed that the width of the thermal degradation process (W ) is narrowed by the loading of the polymeric matrix with CNFs. The dependence of W on the loading with VGCNFs (x) has been fitted by: W ¼ W1 þ W2 expðDxÞ mI0 ¼ mðxÞI0 mI0  100ð%Þ mI0 ð5Þ where mI0 is the mass of pristine polymer at the temperature at which the mass loss rate of the pristine temperature is maximum ð4Þ where W1, W2, and D are fitting constants and W1 þ W2 represents the width of the thermal degradation for the pristine polymer. The dependence of the thermal degradation width on the concentration of VGCNFs is very well described by this equation (see the bold line in Fig. 4). The parameters associated to the best fit are: W1 ¼ 19.5  0.5 K, W2 ¼ 3.5  0.5 K, and D ¼ 0.10  0.04. The negative value of D reflects the narrowing of the thermal degradation process as the concentration of VGCNFS dispersed within IPP is increased. TGA data have been utilized to estimate of the mass fraction Fig. 6. TEM micrograph of IPPeVGCNF composites loaded with 20% wt. VGCNFs. M. Chipara et al. / Polymer Degradation and Stability 93 (2008) 871e876 875 Fig. 7. A model of the adhesion of polypropylene chains to VGCNFs. From the top to bottom, first panel shows the IPPeVGCNFs morphology at room temperature. Second panel shows the morphology of IPPeVGCNFs at temperature lower than the temperature at which the mass loss is maximum (TI0). Some macromolecular chains interacting with VGCNFs have been removed due to the enhancement of molecular motions (note that mostly chains that have no molecules in the hard layer were evaporated). Third panel depicts the IPPeVGCNFs immediately above the temperature at which the maximum mass loss is reached. The decrease of the density of macromolecular chains in interaction with VGCNFs is noticed. Last panel shows the VGCNFs above 700  C. In the temperature range 700  Ce1000  C the polymer is fully vaporized; only the molecules trapped by strong van der Waals interactions are left on VGCNFs. For such molecules the energy of van der Waals interaction (with the nearest molecules of VGCNFs) is stronger than the energy of the CeC bond. The hard layer is represented by the strong gray color while the light gray color is associated to the elastic layer. The ratio between the light gray and white indicates qualitatively the fraction of polymer confined within the soft layer at a given temperature (relative to RT). (TI0), m(x)I0 is the mass of the composite containing x% wt. VGCNF that was not degraded at TI0. As it is inferred from Eq. (5), the mass fraction (mIx) defines the stability of IPPe VGCNFs. From Fig. 5 it is noticed that mIx increases as the amount of VGCNF dispersed within the IPP is increased. Even more, the dependence of mIx (see Eq. (5)) on the concentration of VGCNFs is almost linear. As the concentration of VGCNFs is proportional to the total contact area between IPP and VGCNFs, Fig. 5 suggests that the thickness of the interface IPPeVGCNFs is not significantly affected by the concentration of VGCNFs. Assuming that the density of VGCNF is about 2 kg/m3, the polymer density is about 1 kg/m3, the average diameter of VGCNFs is about 80 nm and the average length 65,000 nm, the thickness of the soft IPPeVGCNFs interface has been estimated to be of the order of 102 nm. This value is consistent with the radius of gyration of IPP [18] and indicates that solely the chains that have molecules in contact with the nanofibers have an enhanced thermal stability. This interface is elastic because these segments are not in the near proximity of the CNFs. Hence, these macromolecular chains still preserve a certain degree of flexibility. A detailed analysis of the residual mass of these composites at fairly large temperatures (above 700  C) reveals that weight of PPeVGCNFs residues exceeds the weight of VGCNFs. Hence, it is speculated that some carbon atoms originating from the polymer chains wrapped around the nanofibers are 876 M. Chipara et al. / Polymer Degradation and Stability 93 (2008) 871e876 not affected by the thermal treatment. To explain this increase it was assumed that some polymer molecules are captured by VGCNFs via van der Waals interaction. As it is observed from Fig. 5, the fraction of polymer captured by such van der Waals interactions is fairly small and depends linearly on the concentration of VGCNFs. This suggests that these extra carbon atoms scale as the total surface of nanofibers indicating that the thickness of this core layer does not depend significantly on the concentration of VGCNFS in IPP. Actually, the estimated thickness of this extra layer assuming that the density of VGCNFs is 2 kg/m3 and the polymer density is about 1 kg/m3 is of the order of 100 nm. In order to fully exploit the benefits derived from the nanometer scale size of VGCNFs it is mandatory to achieve a good dispersion of VGCNFs within IPP. Transmission electron microscopy (TEM) investigations were utilized to check for the dispersion of the nanofiller within the polymeric matrix. As may be noticed from Fig. 6, even for the composite loaded with 20% VGCNFs the filler is well dispersed within the polymeric matrix. This degree of dispersion results in a huge surface area and explains the effect of the interface on the macroscopic thermal stability of the polymeric matrix. 4. Conclusions The detailed analysis of the thermal degradation of IPPe VGCNFs composites revealed unique and amazing features regarding the morphology and the structure of this nanoreinforced composite (Fig. 7). The observed enhancement of the thermal stability of IPP by the loading with VGCNFs is a direct proof for the existence of an IPPeVGCNFs interface. Mathematical modeling provided quantitative estimations on the inflection temperature shift and suggested a better homogenization of the temperature distribution within the nanocomposite as the loading of VGCNFs is increased. The experimental data suggested that the IPPeVGCNFs interface has a two-layer structure. The external or soft layer has a thickness comparable to the radius of gyration of the polymer and contributes to the enhanced thermal stability of the composite. For IPPeVGCNFs this thickness has been estimated to be of the order of 102 nm. TGA analysis revealed also a new inner or core layer, with a thickness of the order of 100 nm. This layer is tentatively assigned to those molecules belonging to VGCNFs that are located around VGCNFs at distances smaller or equal to 100 nm. Such molecules are trapped around VGCNFs molecules by strong van der Waals interactions. These interactions are stronger than the energy of CeC bonds and accordingly during the thermal degradation process the molecules are confined within the core or rigid layer while the polymer degrades and volatilizes. The excellent dispersion of VGCNFs within the polymeric matrix (IPP), confirmed by TEM data, supports the proposed interpretation. Acknowledgements This research has been supported by DMR 0606224 and by the US Army STTR grant #A2-1299. References [1] Padden FJ, Keith HD. Spherulitic crystallization in polypropylene. J Appl Phys 1959;30(10):1479e84. [2] Pae KD. gea Solidesolid transition of isotactic polypropylene. J Polym Sci Pol Phys 1968;6(4):657e63. [3] Isasi JR, Mandelkern L, Galante MJ, Alamo RG. The degree of crystallinity of monoclinic isotactic poly(propylene). J Polym Sci Part B Polym Phys 1999;37(4):323e34. [4] Menyhard A, Varga J, Liber A, Belina G. Polymer blends based on the bmodification of polypropylene. Eur Polym J 2005;4:669e77. [5] Varga J. Crystallization, melting and supermolecular structure of isotactic polypropylene. In: Karger-Kocsis J, editor. Polypropylene: structure, blends and composites. 1st ed., vol. 1. London: Chapman and Hall; 1995. p. 56e67. [6] Corradini P, Petraccone C, Rosa CDe, Guerra G. On the structure of the quenched mesomorphic phase of isotactic polypropylene. Macromolecules 1986;19(11):2699e703. [7] Natta G, Corradini P. Structure and properties of isotactic polypropylene. Nuovo Cimento 1960;15(Suppl):40e51. [8] Varga J, Ehrenstein GW. Formation of b-modification of isotactic polypropylene in its late stage of crystallization. Polymer 1996; 37(26):5959e63. [9] Bruckner S, Meille SV. Non-parallel chains in crystalline g-isotactic polypropylene. Nature (London) 1989;340:455e7. [10] Lotz B, Wittmann JC, Lovinger AJ. Structure and morphology of poly(propylenes): a molecular analysis. Polymer 1996;37(22):4979e92. [11] Shen J, Huang W, Wu L, Hu Y, Ye M. Thermo-physical properties of epoxy nanocomposites reinforced with amino-functionalized multi-walled carbon nanotubes. Composites A 2007;38:1331e6. [12] Zou Y, Feng Y, Wang L, Liu X. Processing and properties of MWNT/ HDPE composites. Carbon 2004;42(2):271e7. [13] Lozano K, Barrera EV. Nanofiber-reinforced thermoplastic composites: thermoanalytical and mechanical analysis. J Appl Polym Sci 2000;79:125e33. [14] Lozano K, Files B, Rodriguez-Macias F, Barrera EV. Purification and functionalization of vapor grown carbon nanotubes and single wall nanotubes. Symposium powder materials: current research and industrial practices, TSM Fall meeting 1999. p. 333e40. [15] Brown SDM, Jorio A, Corio P, Dresselhaus MS, Dresselhaus G, Saito R, et al. Origin of the Breit-Wigner-Fano lineshape of the tangential G-band feature of metallic carbon nanotubes. Phys Rev B 2001;63:155414. [16] Nguyen KT, Gaur A, Shim M. Fano lineshape and phonon softening in single isolated metallic carbon nanotubes. Phys Rev Lett 2007;98:145504. [17] Banerjee S, Kim DI, Robinson RD, Herman IP, Mao Y, Wong SS. Observation of Fano asymmetry in Raman spectra of SrTiO3 and CaxSr1xTiO3 perovskite nanocubes. Appl Phys Lett 2006;89:223130. [18] Krause B, Stephan M, Volkland S, Voigt D, Haußler L, Dorschner H. Long-chain branching of polypropylene by electron-beam irradiation in the molten state. J Appl Polym Sci 2006;99:260e5.