Hierarchical Polymer–Nanotube Composites

COMMUNICATION

DOI: 10.1002/adma.200700765

Hierarchical Polymer–Nanotube Composites** By Tirtha Chatterjee, Cynthia A. Mitchell, Viktor G. Hadjiev, and Ramanan Krishnamoorti* The development of hierarchical structures in polymer nanocomposites remains a challenging issue.[1–4] In most cases, efforts have been expended to develop mesoscale and macroscale structures by controlling the assembly and interactions at the molecular level.[5,6] On the other hand, semi-crystalline polymers and their ability to form hierarchically templated structures resulting from the introduction of macro-fillers (such as glass fibers) and process them under non-equilibrium conditions has resulted in a wide array of engineering materials with structures that are correlated from the nanoscale to the macroscale.[7] In this work we report the ability to control the nano- and mesoscale arrangement of semi-crystalline polymers resulting from the ability to macroscopically align single walled carbon nanotubes (SWNTs) and the nucleation of polymer crystals from such highly oriented nanotube systems. Specifically we demonstrate that the polymer lamellae can be organized such that their layer normal is parallel to the nanotube axes due to the nucleating tendency of the SWNTs. We choose to investigate the hierarchical structure of nanocomposites of SWNTs in poly(e-caprolactone) (PCL) where the dispersion is assisted by the use of surfactants at low loadings (1 surfactant chain per 70 SWNT carbon atoms).[8,9] The unoriented nanocomposites have a geometrical and electrical percolation of the SWNTs at low loadings of nanotubes and indicate an effective aspect ratio for the dispersed nanotubes to be ∼ 750.[10] In these nanocomposites, the SWNTs act as nucleators for the crystallization of the polymer and demon-

– [*] Prof. R. Krishnamoorti, T. Chatterjee, Dr. C. A. Mitchell[+] Department of Chemical and Biomolecular Engineering University of Houston Houston, TX 77204-4004 (USA) E-mail: [email protected] Prof. R. Krishnamoorti Department of Chemistry, University of Houston Houston, TX 77204 (USA) Dr. V. G. Hadjiev Texas Center for Superconductivity, University of Houston Houston, TX 77204-5002 (USA) [+] Present Address: ExxonMobil Corporate Research, Annandale, NJ, USA. [**] We thank Dr. Igor Sics and the user program at the NSLS at Brookhaven National Laboratory. Financial support from the Air Force Office for Sponsored Research and the Texas Institute for Intelligent Bio-Nano Materials and Structures for Aerospace Vehicles, funded by NASA Cooperative Agreement No NCC-1-02038 is gratefully acknowledged. Supporting Information is available online from Wiley InterScience or from the authors.

3850

strate roughly a hundred fold increase in the rate of crystallization and a small (∼ 1–2 °C) increase in the melting temperature that do not change beyond 0.1 wt % SWNT.[10,11] The nanocomposite samples were subject to uniaxial extension in the melt state of the polymer by applying a symmetrical elongation that results in an extension of 10:1 or higher and subsequently cooled down to the crystalline state of the polymer under stretched but otherwise quiescent conditions. The alignment of the SWNTs is quantitatively determined using polarized Raman spectroscopy. Monitoring the dependence of the intensity of the tangential (G) mode of the SWNTs with the angle between the draw direction and the polarization direction of the incident light (w), as shown in Figure 1, allows us to quantitatively measure the orientation of SWNTs with respect to the draw direction. Specifically, the intensity scales as A + Bcos2(w) + Ccos4(w) and for the case shown in Figure 1 the intensity minimum and maximum are observed at w ∼ 90° and 180°, respectively, indicating orientation of the nanotubes in the draw direction.[12–14] The extent of alignment of the nanotubes, with their tube axis along the draw direction,[12,15–19] is quantified by a Herman’s orientation function whose values lie between 0.5 and 0.7 for all the samples studied. Further, repeated heating to the melt state followed by crystallization of the polymer does not lead to any significant change in the orientation state of the nanotubes as shown in Figure 1. On the basis of previous results for the dispersion of other anisotropic nanoparticles in a polymer matrix where attractive interactions exist between the polymer and the nanoparticle and considering the viscosity of the polymer matrix along with the large effective aspect ratio of the nanotubes in these polymers, these nanocomposites exhibit a very slow disorientation under quiescent conditions from their initially aligned state.[20] The arrangement of the crystalline structure of the polymer with respect to the oriented nanotubes is examined using small and wide angle X-ray scattering (SAXS and WAXS) on pre-oriented nanocomposite samples. Representative two-dimensional SAXS for the oriented PCL nanocomposites, shown in Figure 2, indicate that the crystalline lamellae are organized such that the layer normal points along the oriented nanotube cylinder axis (i.e., draw direction) suggesting a “shish-kebab” structure with the flow-aligned nanotubes represent the “shish” and the PCL crystalline lamellae are organized as the “kebabs”, similar to those observed for other nanotube or nanotube bundle nucleated polymer crystals.[21–24] Interestingly, the Herman’s orientation function for the polymer lamellae are remarkably consistent with a similar orientational order parameter for the nanotubes estimated

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2007, 19, 3850–3853

Adv. Mater. 2007, 19, 3850–3853

1000

As drawn

vv

I (a u)

Intensity ( a u )

ψ (º)

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advmat.de

COMMUNICATION

gests that the aligned SWNTs lead to a templated growth of the polymer crystals. 90 800 Further corroboration of this templating effect is After SAXS 120 800 150 obtained from the WAXS data for the flow-aligned 180 600 600 PCL nanocomposite and for the subsequently melted and recrystallized samples. We note that 400 400 PCL typically forms orthorhombic unit cells with 200 the polymer chains along the c-axis, and the (110) 200 and (200) planes subtend an angle of 56.4° for a 0 1280 1440 1600 1760 0 45 90 135 180 single crystal sample.[27] Further, when PCL is uni-1 ψ (deg) Raman shift (cm ) axially stretched during crystallization, it results in the c-axis of the unit cell to be along the stretching Figure 1. G band Raman intensity (with parallel incident and scattered (VV) polarizadirection. For the PCL nanocomposites described tion) for a 3 wt % SWNT dispersed in PCL nanocomposite. For each angular position (w, the angle subtended between the polarization state of the incident laser light and here, the (110) and (200) reflections are observed the draw direction applied to the sample), the spectra (shown on left) are baseline along the equator, and the (102) plane is observed –1 corrected and intensities obtained after resolving each G band peak, at 1593 cm , as a weak four point fiber pattern under all therusing a mixed Gaussian and Lorentzian function (right). The Ivv data for the sample mal/mechanical histories discussed here.[28] We as drawn and after SAXS measurements (after three successive exposures to the melt state for 10 minutes each followed by isothermal crystallization at 43, 45, and 50 °C) thus conclude that the c-axis is oriented parallel to are virtually identical. The line through the data for the Ivv versus w plot corresponds the nanotube axis and the “a” and “b” axes of the to a fit to the function Ivv = A + Bcos2(w) + Ccos4(w). unit cell are randomly oriented in a plane orthogonal to the nanotube-axis, resulting in the congruent observation of the (110) and (200) reflections along from the Raman data. Such a stacking of the polymer lamelthe equator.[28,29] lae has been observed following uniaxial extension flow and The results indicate a hierarchical structure for PCL-SWNT has been attributed to the chain stretching (in the flow direcnanocomposites, wherein the unit cell of the polymer crystal, tion).[25] Thus, the resulting “shish-kebab” structure reported the lamellar organization of the crystalline polymer and the here for the nanocomposites in response to uniaxial flow (folmacroscopic orientation of the nanotubes are interlinked, and lowed by rapid quenching) is expected.[23,24] in fact the macroscopic structure (i.e., orientation of the nanoThe oriented samples were subsequently heated (under tubes) dictates the nanoscale arrangement of the polymer chain quiescent conditions) to the melt state of the polymer and alunit. This shish-kebab structure results in a remarkable reinlowed to recrystallize isothermally in the absence of any exterforcement of the tensile mechanical properties of the PCL nal stresses with simultaneous monitoring of the structural denanocomposites along the draw direction (i.e., along the nanovelopment using SAXS and WAXS, with the X-ray beam tube axis) (Table 1) at room temperature. The linear tensile transmitting through the sample orthogonally to the draw dimodulus values obtained at room temperature show a ∼ 6 fold rection. The two-dimensional SAXS data indicated that the increase in the nanocomposites with 0.2 wt % SWNT, and sig“shish-kebab” morphology continues to persist even after renificantly higher than that anticipated based on conventional peated excursions well into the melt state of the polymer for mechanical models.[30] This significant improvement in the linetens of minutes. Similar exposure of the melt—drawn pure ar mechanical properties results from excellent stress transfer PCL to the melt state results in a complete loss of crystal oriby the strong ionic and hydrogen bonding interactions between entation. The extent of orientation, quantified by a lamellar the zwitterionic surfactant and the nanotubes and PCL and orientational order parameter S, for the nanocomposites renanotubes respectively as demonstrated recently using vibramains relatively unchanged during this process (Fig. 3). In tional spectroscopy measurements.[10] The templating of the fact, in some cases following rapid quenches from the melt polymer lamellae and the intrinsic anisotropic mechanical propstate (during or immediately following extension), reheating erties of the polymer lamellae probably add to the resulting reof the samples to the melt state followed by isothermal crysinforcement of the nanocomposites observed in this study. tallization results in a higher orientational order parameter We thus conclude that oriented SWNTs can render the deand indicative of a more perfected shish-kebab structure. We velopment of oriented semi-crystalline polymers, from the note that the terminal relaxation time of the parent PCL is < 0.1 s (at 75 °C) and is largely unaltered in the nanocomposites with SWNT loadings ranging from 0.1 to 3 wt %. This Table 1. Linear Tensile Testing of Drawn PCL Nanocomposite Fibers. templating of the polymer crystals suggests that the embedded Sample Young’s modulus [MPa] and largely unaltered template of the oriented nanotubes leads to the development of the oriented lamellae. The combiPCL (Drawn from Melt) 384 ± 30 0.05 wt % SWNT + PCL 1234 ± 65 nation of the sluggish disorientation of the nanotubes in the 0.1 wt % SWNT + PCL 1674 ± 103 polymer melt as indicated by the Raman data in Figure 1, and 0.2 wt % SWNT + PCL 2545 ± 75 [11] the known nucleation tendency of SWNTs for PCL, sug1000

3851

1200

1.1 1000

-1.1 -1.1

Intensity (a u)

0

φ

qy (nm-1)

COMMUNICATION

a)

q x (nm -1)

800 600 400 200

0

1.1

0

0

60

120

180

240

300

360

azimuthal angle φ (deg)

b)

(102)

(110) -40

(200) 0

2θ ( º )

40

c)

Figure 2. Hierarchical morphology of the polymer crystals for SWNT–PCL nanocomposites as determined by X-ray scattering (k = 1.371 Å). In part (a) the two-dimensional SAXS data for a 3 wt % SWNT nanocomposite and the azimuthal scan at a q-value corresponding to the lamellar reflection are shown. In part (b) the two-dimensional WAXS data for the same nanocomposite is shown and in part (c) the schematic of the morphology for the nanocomposite is shown.

molecular to the mesoscale, with significant improvements in reinforcement of the nanocomposites. Combining the ability to control the development of hierarchical structures with the wide variety of rod-like nanoparticles and the ability to manipulate their three dimensional structure using external fields, provides the possibility of developing materials with controlled multifunctional properties whose directionality can easily be manipulated. These along with the extension of the results to high concentrations of nanoparticles where directionally multi-functional materials can be prepared will enable a significant application of such polymer nanocomposites in a wide-range of applications.

3852

www.advmat.de

Experimental The SWNTs (supplied by Carbon Nanotechnologies Inc.) were produced by the HiPco process [31] and purified using standard procedure [32]. After purification the metal content, measured by energy dispersive X-ray spectroscopy, was below 1 wt %. The polymers and surfactants used in this study were purchased from Aldrich Chemical Co. and used as received. The number average (Mn) and weight average (Mw) molecular weights of poly(e-caprolactone) (PCL) used for this study were 42 500 Da and 65 000 Da, respectively. All nanocomposite samples were prepared using a surfactant assisted dispersion technique [8,9]. Nanotubes mixed with appropriate amounts of the surfactant were sonicated for 3 h (Fischer Scientific ultrasonic bath, 44 kHz) in toluene. The polymer was subsequently

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2007, 19, 3850–3853

[1] D. M. Lincoln, R. A. Vaia, R. Krishnamoorti, Macromolecules 2004, 37, 4554. 0.8 T =50º C T = 45º C T = 43º C crys [2] T. X. Dang, S. J. Farah, A. Gast, C. Rocrys 3.0 crys bertson, B. Carragher, E. Egelman, E. M. 0.6 Wilson-Kubalek, J. Struct. Biol. 2005, 150, 90. 0.4 [3] Z. Y. Tang, N. A. Kotov, Adv. Mater. 2005, 17, 951. 0.2 [4] R. Krishnamoorti, A. S. Silva, C. A. Mitchell, J. Chem. Phys. 2001, 115, 7175. As drawn 0 [5] R. A. Vaia, E. P. Giannelis, MRS Bull. 400 800 1200 0 0 400 800 1200 0 500 1000 1500 2000 2001, 26, 394. Time (s) [6] R. Krishnamoorti, MRS Bull. 2007, 32, 341. Figure 3. The anisotropy parameter S, obtained from the azimuthal intensity of the SAXS data [7] N. S. Enikolopyan, M. L. Fridman, I. O. for in-situ crystallization experiments (shown as insets) is conserved over crystallization temperaStalnova, V. L. Popov, Adv. Polym. Sci. ture and crystallization time. For all SAXS images, both qx and qy range from –1.1 to 1.1 nm–1. q 1990, 96, 1. 2 2 2 The anisotropy is calculated as S = …hu1 u1 i hu2 u2 i† ‡ 4…hu1 u2 i† , where 〈u1u1〉 = 〈cos f〉, [8] T. Chatterjee, K. Yurekli, V. Hadjiev, , Zp Zp

 R. Krishnamoorti, Adv. Funct. Mater. 2 2 2 〈u2u2〉 = 〈sin f〉, and 〈u1u2〉 = 〈cosf sinf〉 with cos f ˆ cos f I…f† df I…f† df being 2005, 15, 1832. 0 0 [9] K. Yurekli, C. A. Mitchell, R. Krishnathe typical relationship to the azimuthal intensities and f is defined as shown in Figure 2a [26]. moorti, J. Am. Chem. Soc. 2004, 126, 9902. Samples were heated to the melt state for 10 min and subsequently crystallized isothermally at the [10] C. A. Mitchell, R. Krishnamoorti, Macrotemperatures indicated. molecules 2007, 40, 1538. [11] C. A. Mitchell, R. Krishnamoorti, Polymer 2005, 46, 8796 added to this dispersion and the mixture stirred for 24 h. PCL nano[12] A. R. Bhattacharyya, T. V. Sreekumar, T. Liu, S. Kumar, L. M. Ericcomposites were compatibilized using a zwitterionic surfactant, son, R. H. Hauge, R. E. Smalley, Polymer 2003, 44, 2373. 12-aminododecanoic acid (ADA). The ratio of amine head group to [13] A. Jorio, A. G. Souza, V. W. Brar, A. K. Swan, M. S. Unlu, B. B. nanotube carbon was maintained at ∼ 1:70 for all the samples. The solGoldberg, A. Righi, J. H. Hafner, C. M. Lieber, R. Saito, G. Dresselvent was then removed by extensive drying under convective condihaus, M. S. Dresselhaus, Phys. Rev. B 2002, 65. tion followed by vacuum drying in the melt state (80 °C for at least [14] T. Liu, S. Kumar, Chem. Phys. Lett. 2003, 378, 257. 24 h). [15] H. H. Gommans, J. W. Alldredge, H. Tashiro, J. Park, J. Magnuson, Raman spectra were recorded on a Jobin Yvon S3000 spectromeA. G. Rinzler, J. Appl. Phys. 2000, 88, 2509. ter. A long-working distance microscope objective (50×) fitted on an [16] M. F. Islam, D. E. Milkie, C. L. Kane, A. G. Yodh, J. M. Kikkawa, Olympus 45 microscope was used to focus the incident laser beam Phys. Rev. Lett. 2004, 93, 037404. (514.5 nm laser line of an Ar+ laser) to a spot ∼ 2 lm in diameter on [17] L. Jin, C. Bower, O. Zhou, Appl. Phys. Lett. 1998, 73, 1197. the sample surface and to collect the scattered light. The laser power [18] B. S. Shim, N. A. Kotov, Langmuir 2005, 21, 9381. density was kept below 104 W cm–2 to prevent overheating of the sam[19] W. Zhou, K. I. Winey, J. E. Fischer, T. V. Sreekumar, S. Kumar, ple at the laser spot. The samples were placed on a microscope rotary H. Kataura, Appl. Phys. Lett. 2004, 84, 2172. table with draw direction matching table’s zero and the table (equiva[20] J. X. Ren, B. F. Casanueva, C. A. Mitchell, R. Krishnamoorti, Maclently the sample) was rotated (keeping the laser spot on the sample romolecules 2003, 36, 4188. and polarization fixed) to obtain the angular (W) dependence of the [21] Q. H. Zhang, D. R. Lippits, S. Rastogi, Macromolecules 2006, 39, Raman intensity. 658. Small and wide angle X-ray scattering (SAXS and WAXS respec[22] Q. H. Zhang, S. Rastogi, D. J. Chen, D. Lippits, P. J. Lemstra, Cartively) measurements were performed at the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory (beamline bon 2006, 44, 778. X27C) with a wavelength (k) of 1.371 Å and a beam diameter [23] L. Y. Li, C. Y. Li, C. Y. Ni, J. Am. Chem. Soc. 2006, 128, 1692. of ∼ 0.5 mm at the sample. X-ray samples were placed in brass [24] R. Haggenmueller, J. E. Fischer, K. I. Winey, Macromolecules 2006, holder with Kapton windows covered on both sides. For all scattering 39, 2964. experiments, samples were positioned so that the X-ray transmission [25] M. Muthukumar, in Advances in Polymer Science (Interphases and was orthogonal to the draw direction. A high resolution CCD X-ray Mesophases in Polymer Crystallization III), Vol. 191, Springer, Bercamera was used to record SAXS data and image plates were used lin, New York 2005, p. 241. to record WAXS data. For all samples, SAXS and WAXS data [26] F. E. Caputo, W. R. Burghardt, Macromolecules 2001, 34, 6684. were obtained for the ‘as drawn’ samples at room temperature. Then [27] H. L. Hu, D. L. Dorset, Macromolecules 1990, 23, 4604. samples were subsequently heated to 80 °C and held there for [28] M. Kakudo, N. Kasai, X-ray Diffraction by Polymers, Elsevier, Am∼ 10 min. Isothermal crystallization experiments were then consterdam 1972. ducted by rapidly swapping the samples into a pre-equilibrated oven [29] Y. B. Zhang, V. Leblanc-Boily, Y. Zhao, R. E. Prud’homme, Polymer at the crystallization temperature of interest. This heat to melt2005, 46, 8141. hold-isothermal crystallization cycle was repeated for different [30] T. D. Fornes, D. R. Paul, Polymer 2003, 44, 4993. isothermal crystallization temperatures. SAXS and WAXS data for [31] P. Nikolaev, M. J. Bronikowski, R. K. Bradley, F. Rohmund, D. T. the isotropic melt as well as during isothermal crystallization with Colbert, K. A. Smith, R. E. Smalley, Chem. Phys. Lett. 1999, 313, 91. periodic interval were collected and corrected for background [32] D. Chattopadhyay, L. Galeska, F. Papadimitrakopoulos, J. Am. and empty cell scattering using standard methods [11]. Chem. Soc. 2003, 125, 3370. wt% SWNT

Anisotropy (S)

1.5

COMMUNICATION

–

1

Received: March 29, 2007 Published online: October 31, 2007

Adv. Mater. 2007, 19, 3850–3853

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advmat.de

3853