Core–Shell Interactions in Coaxial Electrospinning and Impact on

ARTICLE pubs.acs.org/JPCC

Core
Shell Interactions in Coaxial Electrospinning and Impact on Electrospun Multiwall Carbon Nanotube Core, Poly(methyl methacrylate) Shell Fibers Timothy J. Longson,†,‡ Ranadeep Bhowmick,†,§ Claire Gu,‡ and Brett A. Cruden†,* †

Center for Nanotechology, NASA Ames Research Center, Mail Stop 230-3/Bldg. 230, Rm. 206, Moffett Field, California 94035, United States ‡ School of Engineering, University of California Santa Cruz, Santa Cruz, California, United States § Materials Science and Engineering, Stanford University, Stanford, California, United States ABSTRACT: The intricacies of the core
shell interaction in coaxial electrospinning are explored by way of the Hansen solubility parameters in an attempt to make micrometersized polymer-free multiwalled carbon nanotube (MWNT) core, poly(methyl methacrylate) shell fibers. Four solution regimes are explored in which the core solvent is either miscible or immiscible of the shell solvent and in which it is either a solvent or nonsolvent of the shell polymer. It is qualitatively found that the most well-defined MWNT bundle core is achieved using a core solvent that is semi-immiscible with the shell solution, yet still a solvent of the shell polymer. Hollow fibers are found to be produced best by using a core solvent that is immiscible with the shell solvent and also a nonsolvent of the shell polymer. The resulting MWNT-bundled cores are electrically characterized and found to have conductivities up to 2 orders of magnitude greater than homogeneously electrospun MWNT/polymer composite fibers.

1. INTRODUCTION Since its discovery, the carbon nanotube (CNT) has been proposed as being one of the ultimate materials for its electrical, thermal, and mechanical properties. Individual CNTs possess some remarkable characteristics, though the use of this material has been hindered by the ability to scale up these traits to the macroscopic world. Whereas significant improvements have been shown for polymer composites with the incorporation of only a slight fraction of CNTs,1 the ultimate potential of this material, in its pure form, has yet to be realized. Several research groups have made large strides toward this effort, particularly Richard Smalley’s group using a superacid, wet spinning approach and Ray Baughman’s group using a dry spinning technique.2,3 Whereas both of these methods are very promising, taking full advantage of the uniaxial properties remains a challenge. In particular, the alignment and packing density of CNTs within a fiber need to be greatly increased to obtain macroscopic fibers whose properties approach that of individual CNTs. The research presented here aims to tackle the problem using a coaxial electrospinning technique. Because of its small diameter, constricting nature, and electric/flow-field orientation, coaxial electrospinning may produce tightly packed, well-aligned CNT bundles. These bundles may then be woven into larger macroscopic twines and ropes. Whereas other groups have made progress with electrospinning homogeneous polymer-CNT composites, the electrical and mechanical properties of these fibers are still many orders of magnitude lower than that of individual CNTs.4
7 Liu et al. serendipitously made single-walled carbon r 2011 American Chemical Society

nanotube (SWNT) core-polymer shell nanofibers using a homogeneous electrospinning technique.8 However, their method of electrospinning a homogeneous CNT-polymer solution does not inherently lend itself to a true coaxial fiber, where the core is purely composed of CNTs. While it is not clear whether their results produce continuous and repeatable coaxial structures, their Raman results show an increase in the 236 cm
1 radial breathing mode (RBM) peak, which eludes to a compressive stress induced by the constricting shell. This, along with the generally superb alignment of CNTs within electrospun fibers,7,9 motivated the investigation of the coaxial electrospinning technique. Figure 1 shows a schematic of the coaxial setup along with the Taylor cone of the dual solution, coaxial needle. Further details of the setup can be found in the experimental section below. This work investigates the feasibility of creating continuous CNT core-polymer shell fibers using the coaxial electrospinning approach. To better understand and design this system, a detailed study of the polymer-free core solution interaction with the shell solution is presented by way of the Hansen solubility parameters. This interaction is a critical variable in coaxial electrospinning that had not previously been explored. A variety of morphologies were observed for the different core
shell solubility regimes and the challenges in their accurate characterization are addressed. Finally, electrical characterization of individual multiwalled carbon Received: February 1, 2011 Revised: May 24, 2011 Published: May 25, 2011 12742

dx.doi.org/10.1021/jp201077p | J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C

ARTICLE

Figure 1. Schematic of coaxial setup along with insertion of coaxial needle with Taylor cone composed of PMMA shell and MWNT core solutions.

nanotube (MWNT) cores were performed and their conductivities are compared to those of existing MWNT fibers.

2. EXPERIMENTAL DETAILS Electrospinning was performed in ambient room conditions using a homemade dual syringe pump based on the deposition tool from the Cornell University fab@home project.10 A custom coaxial needle was made for each experiment that consisted of a shell capillary (2.4 mm outer and 1.6 mm inner diameter) and a core capillary (0.84 mm outer diameter, ∼0.66 mm inner diameter). The core capillary protruded from the shell by 0.2
0.6 mm to facilitate entrainment of the core as demonstrated by Reznik et al.11 and seen in Figure 1. All fibers were spun with a spinneret to collector distance of 20 cm using a variable, high-voltage dc source (Spellman EPM15P30). The voltage source was limited to 15 kV dc, though the voltage was finely tuned, between 5 and 8 kV dc during spinning to match the extrusion rate from the syringe pump. The ratio of flow rates between the core and shell was maintained throughout the experiments with the shell flowing five times that of the core, though the absolute value of these flows could not be maintained due to variations in solution parameters. In particular, more volatile solutions had to be spun at higher flow rates to maintain a stable Taylor cone. The shell flow varied between 1 and 3 mL/h. Multiwalled carbon nanotubes (MWNT) were purchased from Cheap Tubes Inc. with an inner diameter of 5
10 nm, outer diameter of 20
40 nm and a length of 0.5
2 μm. Poly(methyl methacrylate) (MW = ∼996 000), sodium dodecyl sulfate (SDS), along with all of the organic solvents used in this research were purchased from Sigma-Adrich. None of the materials underwent further purification. The MWNT solutions were initially sonicated in a Cole-Palmer 8893 for one hour. Before spinning, 15 min of additional sonication were performed to redisperse agglomerated MWNTs. The polymer solutions were prepared and sonicated for one hour then left at room temperature for

three days to fully dissolve. For imaging and Raman spectroscopy, the samples were spun onto n-type Silicon, whereas for electrical characterization the samples were spun onto glass slides with thermally evaporated aluminum contact pads. Raman spectroscopy was performed on a Renishaw spectrometer with a HeNe 632.8 nm laser. Optical imaging was performed using a Leica DMLM microscope. Scanning electron micrographs were taken at Stanford nanocharacterization lab using an FEI SL30 Sirion SEM with an FEG source. Electrical measurements were performed on a Desert Cryogenics TTP4 probe station using custom-made electrochemically etched Tungsten probe tips, as described elsewhere,12 and an HP semiconductor parameter analyzer, 4156B. Aluminum contact pads were thermally evaporated using a BOC Edwards Auto 306 thermal evaporator through various sized TEM grids as hard masks.

3. RESULTS AND DISCUSSION 3.1. Experimental Space/Solution Properties/Morphology. The field of coaxial electrospinning is still fairly new and,

due to the large number of variables involved, is still poorly understood. Despite the proposed advantages of coaxially electrospinning a CNT core-polymer shell fiber, there are many obstacles in achieving such a morphology. There are two critical challenges. The first is the lack of viscoelasticity of the CNT core. Much of the past work with coaxial electrospinning has been done with pairs of solutions, which are both independently electrospinnable, facilitating the production of a clean core
shell morphology. Here, the shell acts as a guide for the nonelectrospinnable core. Many other research groups have used this technique to electrospin other non-electrospinnable polymers, though few have used completely polymer free core solutions. Song et al. were successful in coaxially electrospinning a core composed of FePt nanoparticles dispersed in hexane with a PCL/ 2,2,2-triuoroethanol shell.13 Whereas fairly well-defined core
shell 12743

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C

ARTICLE

where ΔH is the enthalpy of vaporization, RT is from the ideal gas law, and V is the molar volume. rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ΔH
RT ð1Þ δ¼ V

Figure 2. Hansen solubility parameters of core
shell solvent pairs in four miscibility
solvency regimes, data from Properties of Polymers.15

structures were presented in this work, the longest continuous core length was reported to be 3 μm, suggesting aspect ratios on the order of 100:1 for their 20
50 nm core diameters. A more detailed study of the core
shell interaction needs to be investigated to produce truly continuous cores from nonviscoeleastic solutions. The second challenge with the CNT core is the difficulty of effectively dispersing the CNTs. Even in an ideal solvent, only a small weight percent of CNTs can be dispersed. This poor dispersibility presents three problems. The first being the total mass of the solvent relative to the CNT. With such a small CNT fraction, it is difficult to produce a tightly bundled CNT core without voids left behind by the evaporating core solvent. The second problem is that CNTs may agglomerate on length scales comparable to core diameter during the course of electrospinning disrupting the continuity of the fiber structure. The third problem with the dispersibility is the restrictions it imposes upon the choice of core solvent. The interaction between the core and shell solvent plays a very important role in the final fiber morphology and ultimately governs whether or not the solutions can even be electrospun. Out of all of the variables involved in coaxial electrospinning, the most critical appear to be the mixing interactions of the two (core, shell) solutions. The core solution interaction with the shell’s polymer and solvent can be described by four fundamental solution combinations. These combinations are characterized by the interaction between the core’s solvent with the shell’s solvent and the shell’s polymer. The two solvents may be miscible or immiscible, whereas the shell polymer may or may not be soluble in the core solvent. In this work, we refer to these four possibilities as miscible-nonsolvent, immiscible-nonsolvent, miscible-solvent, or immiscible-solvent. These terms are applied qualitatively, considering that miscibility and solubility may occur to varying degrees depending on relative concentrations. These relationships are well described by the general Hildebrand solubility parameter and the more specific Hansen solubility parameters.14 Hildebrand is noted for much of the pioneering work in the study of quantifying solubility. This was done by using the square root of the cohesive energy, the amount of energy required to vaporize a quantity of solvent after the onset of boiling (eq 1),

While Hildebrand was able to fairly accurately predict solubility between solvents by their proximity in δ, this method fell short with particular pairs of solvents such as nitromethane and ethanol. Hansen and others aimed to solve this problem by breaking Hildebrand’s solubility parameter up into more representative components.14 Hansen did this by dividing it up into δd, the (atomic) London dispersion force, δp the (molecular) permanent dipole forces, and δh the (molecular) hydrogen bonding component. The Hilebrand parameter is equal to the square root of the sum of the squares of the Hansen components (eq 2). qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2Þ δ ¼ δd 2 þ δp 2 þ δh 2 The method for determining the components of the Hansen solubility parameter is out of the scope of this article, and more information can be found in Hansen’s book.14 Their use is straightforward, based on solubility proximity, as with Hildebrand’s solubility parameter. Similarly to the solvency between two solvents, the solvency between a solvent and a polymer can also be assessed using these parameters. The solubility parameters of polymers cannot be measured directly as in the case of volatile solvents but they must be dissolved in a variety of solvents to determine these values. After dissolving the polymer in a range of solvents and doubling the dispersion parameter scale, a solubility sphere can be assigned to the volume encompassing the three solubility components, with an interaction radius, R. If the Hansen parameters of a solvent or solvent mixture lie within the solubility sphere, it is likely that they will dissolve the polymer. This can be graphically assessed or calculated via eq 3, where R is the interaction radius and subscripts 1 and 2 represent the solvent and polymer respectively. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R g 4ðδd1 þ δd2 Þ2þ þ ðδp1 þ δp2 Þ2 þ ðδh1 þ δh2 Þ2 ð3Þ The solubility parameters and interaction radii of many common solvents and polymers have already been determined and are readily available.14,15 All of the Hansen values used for this research were taken from the book Properties of Polymers.15 Using the polar and hydrogen components of the Hansen solubility parameter, neglecting the relatively weak London dispersion component, the four core
shell interaction regimes can be graphically represented as in Figure 2. This graph plots the interaction of four specific pairs of solvents that were investigated. The lines connect the solvent pairs, whereas the arrow marks the core. The addition of MWNTs and the surfactant, sodium dodecyl sulfate (SDS), presumably alters the exact hydrogen and polar component of the solubility parameter but this was neglected for simplicity. Table 1 presents a more exhaustive list of the solutions explored and their representative morphologies. The core solvents available were limited due to the generally poor dispersibility of CNTs. DMF is known to be the most effective solvent for CNTs16 and was employed in most cases, except when exploring nonsolvent combinations that necessitated use of different core solvents. In this case, the SDS surfactant was added for its demonstrated capabilities in enabling CNT dispersion.17 12744

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C

ARTICLE

Table 1. Sumary of Coaxial Solution Pairs and Their Resulting Morphologiesa Regime

Shell conc. [wt %]

a

Core solvent

conc. [wt %]

Morphology solvent

miscible-nonsolvent

10

DMF

0.5

2%SDS/H2O

coagulated

immiscible-nonsolvent

14

CHCl3

0.5

2%SDS/H2O

HC, PS

14

DMF

0.5

2%SDS/IPA

HC, BC

14

DMF

0.2

(methanol/DMF) 3:1 by vol.

BC, few HC

miscible-solvent

6
16

DMF

0.1
0.75

DMF

HC, BC, A

immiscible-solvent

14

(CHCl3/DMF)

0.5

DMF

HC, BC, A

14

(toluene/DMF)

0.5

DMF

sparse BC

14

(MIBK/DMF)

0.5

DMF

HC, BC, PS, A

Morphology abbreviations: HC, hollow core (voids); BC, bundled core (containing CNTs); PS, porous shell; A, artifacts (noncylindrical fibers).

Figure 3. SEM images of various fiber morphologies, (A) collapsed hollow core with a small MWNT bundle protruding, (B) porous shell, (C) hollow core with MWNTs decorating side wall, (D) solid MWNT core in ribbon shaped fiber. Striations in the image are caused by charge accumulation induced by the electron beam.

To maintain bead-free and continuously spun fibers, the polymer solution concentrations were kept relatively high. Whereas the concentration regimes for the solutions listed in Table 1 were not determined via direct measurement, it is well established that for bead-free electrospun fibers, the polymer solution must be in the semidilute-entangled regime. This regime is satisfied for c/c* > 3, where c is the concentration and c* is the critical concentration for chain overlap.18 The shell concentrations were maintained throughout the experiments except for the miscible-solvent regime in which the affect of shell concentration was briefly explored, as seen in section 3.1.3. Because of these high viscosities, all of the fibers explored in this research had relatively large outer diameters, ranging from 1 to 20 μm. These high viscosities also

allowed for nearly continuous spinning of single fibers, with length scales estimated to be in the meter range, giving aspect ratios of millions to one. Making continuous fibers was a central goal of this research. Whereas the continuity of the shell is easily achieved, that of a polymer-free core is much more challenging. Investigating the production of a continuous polymer-free core led to the observation of a large variety of core and shell morphologies. These morphologies were generalized to a few basic sets as described in Table 1, though there was much variation within these sets. Figure 3 exhibits an example from each of the generalized sets of morphologies. Part A of Figure 3 shows how several samples exhibited a collapsed core or dumbbell crosssection. This is the result of a void forming from the evaporating 12745

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C core solvent, which then collapses due to a differential in atmospheric pressure. The Young’s modulus of such collapsed tube walls can be determined from their images, as demonstrated by Yarin et al.19 These morphologies, collapsed core and dumbbell, are generally described in Table 1 as fiber artifacts. Part B of Figure 3 shows a porous shell fiber, which is caused by evaporation of the highly volatile shell solvent, chloroform. Bognitzki et al. saw similar behavior in PLLA/dichloromethane fibers, which they attributed to the rapid phase separation during electrospinning, creating solvent rich regions that transformed into pores.20 Part C of Figure 3 shows a hollow core that has retained its cylindrical structure with a few scattered MWNT bundles decorating the inner side walls of the tube. Part D of Figure 3 has a fairly welldefined MWNT core protruding from the PMMA shell, with an elliptical cross-section, and in Table 1 this is described as bundled core (BC). The striations in the SEM image are a result of local charging on the fiber. These SEM images present the basic set of morphologies that were identified as in Table 1. Whereas many variations on these morphologies were observed, they can be generalized to these four basic groups: hollow core, bundled core, porous shell, and fiber artifacts. The wide variation in morphologies illustrates the impact of the core
shell solution interactions on the final fiber shape. A single electrospun sample, having on the order of meters of continuous length, presents dozens of fiber sections within one low-resolution image of the sample. Within one such image, it was not unusual to observe several of the morphologies described above. Variation in fiber diameter and within the core is seen in every sample. The core is never fully continuous, such that regions both with and without CNT cores are seen on one sample. Therefore, representative characterizations are presented in the images below, whereas Table 1 enumerates the entirety of these observations. The frequency of unfilled versus filled cores is discussed comparatively and qualitatively. SEM imaging of the cores proved challenging as the fibers do not cleave well, breaking in the weaker void sections between CNT-bundled cores. Additionally, attempts to section fibers by focused ion beam processing induced reflow in the fibers and would close off whatever core was present. Nevertheless, features in optical microscopy images were correlated against SEM images, such as those seen in Figure 3. Because of this difficulty, optical microscopy along with Raman spectroscopy is the primary approach used for evaluating the continuity of the core and presence of CNTs. 3.1.1. Miscible-Nonsolvent. The miscible-nonsolvent pair of H2O/N,N-dimethylformamide (DMF) could not be electrospun as the solvents mixed upon contact, causing the PMMA in the shell to precipitate out of the H2O/DMF mixture. This resulted in a coagulated, nonspinnable solution at the needle tip. This result was expected and similar effects are believed to occur for other miscible-nonsolvent pairs, hence no other combinations were investigated. 3.1.2. Immiscible-Nonsolvent. Figure 4 plots the three solution pairs tested in the immiscible-nonsolvent regime using the hydrogen and polar components of the Hansen solubility parameter. The inserts of this figure are optical micrographs showing the characteristic morphologies. The immiscible-nonsolvent combinations were expected to create the most well-defined core
shell structures as others have argued that the nonsolvency of the core solvent on the shell polymer will create a solid interface between the core and shell due to local precipitation of the shell polymer.21 For the most immiscible pair, H2O/chloroform, this was clearly evident as nearly all of the fibers had hollow cores, as

ARTICLE

Figure 4. Hansen solubility graph of immiscible-nonsolvent core
shell solution pairs, inserts: (A) shell 14%PMMA/CHCL3 core 0.5%MWNT/ 2%SDS/H2O, (B) shell 14%PMMA/DMF core 0.5%MWNT/2%SDS/ IPA, (C) shell 14%PMMA/DMF core 0.2%/(methanol/DMF).

seen in part A of Figure 4, though very few MWNT bundles were found within these hollow cores possibly due to a poor dispersion of MWNTs in the core solution. The other two pairs of solutions in this regime, DMF/IPA and DMF/(methanol/DMF 3:1 vol.), showed similar results but with fewer hollow cores and more apparent MWNT bundles, as seen in insets B and C respectively of Figure 4, though none produced the ideal morphology. 3.1.3. Miscible-Solvent. The simplest type of miscible-solvent pair is one where the shell and core share the same solvent. Given that DMF is the best solvent for dispersing CNTs16 and it is also a good solvent for PMMA, it was used for the tests in this regime. This pair of solutions produced reasonably well-defined coaxial structures with few voids (Figure 5). The production of core
shell morphologies in this regime may be attributed to the lower mobility and or immiscibility of the PMMA and MWNTs rather than due to the immiscibility of the solvents. MWNTs have similar aspect ratios to polymers and as such do not mix freely but rather tend to aggregate due to their high surface areas and van der Waals attraction. The general immiscibility of MWNTs with other materials is well-known.16 Because of the quality of these fibers, they were used for the subsequent electrical measurements. Although there were well-defined sections of MWNT cores in these fibers, there were also scattered and mixed MWNTs in other sections and regions with fiber artfacts. The fiber artifacts appear as a dark line under the optical microscope and may be mistaken as MWNT cores without closer examination. Figure 5 clearly demonstrates the utility of Raman spectroscopy in identifying the presence of MWNTs within the core and discarding artifacts. The four Raman spectra in this plot correspond to those taken in the center of the four fibers imaged to the left of the spectra. The presence of MWNTs in the top two samples is characterized by the first-order Raman graphitic, G, peak at ∼1582 cm
1, its disordered, D, peak at ∼1350 cm
1 caused by second-order single-phonon scattering, and its G0 or 2D band at ∼2700 cm
1 caused by second-order double-phonon scattering.22 The third sample confirms the presence of the core artifact by the absence of MWNTs in the Raman spectra. The artifacts observed in these fibers are attributed to the core collapse as in part A of Figure 3. The fourth spectrum of a pure PMMA fiber formed by ordinary (noncoaxial) electrospinning 12746

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C

ARTICLE

Figure 5. Raman spectra of MWNT core in PMMA fiber ref vs fiber artifact, 632 nm excitation. PMMA peak assignment from ref 23. The three top spectra were taken from different sections of the same sample, comprised of 14 wt % PMMA/DMF shell solution and 0.5 wt % MWNT/DMF core solution. The bottom spectrum was taken from a solid fiber made from a 14 wt % PMMA/DMF solution.

serves as a reference for the other peaks. These are all attributed to the PMMA,23 except for that between 900 and 1000 cm
1 caused by the silicon substrate.24 In this regime, variables such as flow rate, spinneret to collector distance, and shell viscosity were investigated to determine their role in core entrainment. No conclusive results were obtained from the flow rate or the collector to spinneret distance tests. Varying the shell concentration and hence its viscosity did however show notable influence on the fiber morphology. As others have noted, increasing the concentration of the polymer solution generally leads to increased fiber diameter.25,26 The outer diameter of the fibers roughly follows a power dependence upon concentration, where diameter ∼(c/c*)3.7, similar to solid fibers as seen by Gupta et al.14 Although the diameters of the cores were not as rigorously characterized due to their inconsistencies, the increased shell viscosity generally led to an increase in voids and MWNT clumping with shell concentrations above 14 wt %, part C of Figure 3, shows a fiber from the limit of this study, 16% PMMA/DMF shell, 0.5%MWNT/DMF core, that has a very large outer diameter and a well-defined hollow core with some MWNTs lining the inner wall of the shell. Varying the core concentration also clearly has an affect on the core morphology. Above 0.5 wt % MWNT/DMF, the core became significantly more clumpy leading to shorter continuous sections of core and voids adjacent to the MWNT bundles. This is due to increased MWNT agglomeration rather than a change in solution viscosity. 3.1.4. Immiscible-Solvent. The immiscible-solvent solution pair was speculated to be the optimal regime for producing densely packed MWNT cores. Whereas the core and shell will not mix because of the immiscibility, a solid interface between the two should not form because of the mutual solvency of the polymer, allowing the shell to continue to narrow, compressing

Figure 6. Hansen solubility graph of immiscible-solvent core
shell solution pairs, inserts: (A) shell 14%PMMA/(CHCl3:DMF) 6:4 core 0.5%MWNT/ DMF, (B) shell 14%PMMA/(CHCl3:DMF) 4:6 core 0.5%MWNT/DMF, (C) shell 14%PMMA/(MIBK:DMF) 6:4 core 0.5%MWNT/DMF.

the core as the core solvent evaporates. Though this approach seems promising, finding an immiscible pair of solvents that are both reasonable solvents of PMMA and where the core solvent suitably disperses the MWNTs is a challenge. Because DMF was the best candidate for the core solvent, it was used throughout this set of experiments. A variety of shell solvents and solvent mixtures were explored as depicted in Figure 6. Though the exact miscibility of these solvent pairs was not known, the assumption that like dissolves like guided the choice of solvents to opposite extremes of the PMMA solubility sphere. Unfortunately, this study did not show the trend that was postulated. Surprisingly, there was much variation within even a single solubility line such 12747

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C

ARTICLE

Figure 7. Optical micrographs of shell removal, (A) reflow of shell around MWNT bundle by heating entire sample at 200 °C in air for 30 min, (B) selective melting of shell around MWNT bundle via 632 nm HeNe laser pointed to with arrow.

as that for the chloroform/DMF shell solution mixtures. A slight trend observed in these samples was an increase in the quality of the core for shell solvent mixtures close to the center of the PMMA solubility sphere. Though there were many examples of hollow cores, the fraction of fibers with hollow cores was significantly less than in the immiscible-nonsolvent regime, which supports the theory that the formation of a solid interface facilitates the core
shell morphology. The insets in Figure 6 present a cross-section of the fiber morphologies observed, inset A shows well-defined core bundling with few voids, inset B shows hollow cores, and inset C presents a collapsed fiber with few MWNT bundles scattered approximately 20 μm apart. 3.2. Shell Removal. Selective shell removal is important both for electrical characterization and to estimate the MWNT alignment. The most obvious approach is to use traditional photolithography techniques, or in the case of a PMMA shell, electron beam lithography, where PMMA acts as a photoresist. Preliminary tests were performed using an industrial electron beam source and TEM grids as hard masks, though, due to the short lengths of the MWNT cores and the turbulent development bath, no exposed MWNT bundles were observed. A more straightforward approach to remove the shell, although not selectively, is to simply burn it off. Part A of Figure 7 shows a fiber in the middle of thermal decomposition where the shell has begun melting, whereas the MWNT bundle remains intact. Another method for selectively removing the shell was found by illuminating the fiber with the Raman laser at high power (∼1 mW/μm2). Absorption of the laser’s energy by the MWNTs and subsequent conversion to thermal energy melts the surrounding PMMA. This only worked in regions that contained well-defined MWNTbundled cores. Part B of Figure 7 shows a region in the fiber where the PMMA has been melted around a well-defined MWNT bundle. Similar optothermal transduction in SWNTs is wellknown.27,28 Whereas this technique for selectively exposing the core for electrical characterization looked promising, it often resulted in complete breaking of the fiber that prohibited direct probing. Subsequently, no electrical measurements were made using this method of exposing the core. 3.3. Electrical Characterization. Two point electrical measurements were performed on thermally decomposed 10% PMMA/DMF shell, 0.5%MWNT/DMF core fibers. The fibers were spun onto a glass slide substrate. The shell was completely removed by heating in air to 450 °C, far above the onset of thermal oxidative degradation of PMMA, for one hour.29 Aluminum contact pads (400 nm thick) were then thermally evaporated over

Figure 8. I
V curve of MWNT bundle, insert optical image of MWNT bundle bridging two Al contact pads.

the MWNT bundles using a TEM grid as a shadow mask. By chance, many of the MWNT bundles bridged adjacent contact pads allowing electrical measurements to be made. Because of the short length of the MWNT bundles, 10 to 40 μm, only two probe measurements were able to be performed. Without four probe measurements, the contributions to contact resistance were unable to be determined and the data thus represents a lower bound on the true conductivity. Figure 8 shows the linear behavior of a typical IV curve for one of the bundles, and the inset shows a bundle bridging two Al contact pads. Figure 9 plots the calculated conductivities for six bundles versus their diameter. The variation in conductivity is within the confidence of the measurement rather than indicating dependence on diameter. For relatively large bundles of MWNTs, such as those presented here, there should not be a strong conductivity dependence on diameter as the tubes will form a network of conduction pathways, emulating a bulk material. For the conductivity calculations the bundles were assumed to have semicircular cross sections and uniform edges, the radius of which was estimated via optical microscopy. Subsequent SEM images (e.g., part D of Figure 3) suggest that the assumption of a semicircular wire morphology may have been an overestimate of the actual mass of the bundles resulting in very conservative estimates of the bundle conductivities. This, along with the undetermined contact resistance, suggest much higher conductivities than reported. Even though these numbers represent an underestimate of the actual fiber conductivity, the conductivities of these short segments are on average 2 orders of magnitude greater than those reported for MWNT/polymer electrospun composite fibers.4 However, these conductivities are still many orders of magnitude lower than the theoretical values and even several orders below what other groups have achieved using alternate fiber making approaches.2,3 The obvious explanations for these lower conductivities are poor continuity, low density, and poor alignment. To gain a quantitative understanding of one of these factors, the MWNT alignment within the bundle was measured using image analysis of a high resolution SEM image of one of the thermally decomposed fibers. The fwhm of the Gaussian fit was quite large at 92.3°. This poor alignment could be a result of the polymer removal step, where polymer reflows before burning, which may reshape the MWNT bundle. To determine whether this occurs would require assessment of 12748

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C

ARTICLE

fringe of the PMMA solubility sphere) is expected to produce the best results. Although the ultimate fiber morphology, a continuous CNT core within a polymer shell, was not achieved, there is promise for increased nanotube alignment, continuity, and packing density with decreased fiber diameter. The limiting factor for this technique and other solution-based CNT fiber making techniques appears to be the quality of the CNT dispersion. Future work may include experimenting with functionalized CNTs for enhanced dispersion. It is speculated that the continuity and alignment of the MWNT core could be drastically increased if a suitable CNT dispersion in the ones to tens of a weight percent were achieved.

’ AUTHOR INFORMATION Corresponding Author Figure 9. Conductivity vs diameter of MWNT bundles. Each data point represents the measurement on a single bundle.

*E-mail: [email protected].

the alignment of the bundle while it is still encased in the polymer shell.

’ ACKNOWLEDGMENT A majority of this work was supported by the NASA Ames/ University of California, Santa Cruz, Aligned Research Program (ARP). Dr. Cruden was supported during this work on contract NAS2-03144 to the University Affiliated Research Center (UARC) operated by UC Santa Cruz.

4. CONCLUSIONS In summary, MWNT core-PMMA shell, micrometer-sized, composite fibers were prepared via a coaxial electrospinning technique. Two point electrical measurements were performed on short segments of MWNT cores following thermal degradation of the shell, showing conductivities up to 2 orders of magnitude higher than homogeneous MWNT/polymer composite fibers. The interaction of the polymer free core on the shell was investigated through the use of Hansen’s solubility parameters with the goal of promoting alignment, packing density, and continuity of the core. Four core shell solution interaction regimes were explored based on their miscibility and the core solutions solvency of the shell polymer. Miscible solutions with a PMMA nonsolvent core led to coagulation of the Taylor cone, preventing any electrospinning. Immiscible-nonsolvent solution pairs resulted in the most well-defined core
shell interface. Whereas the core shell structure is well-defined, this interface is formed early in the spinning process, before the core solvent has completely evaporated, causing many hollow regions. This regime may be useful for creating hollow tubes or well-defined polymer
polymer coaxial fibers but, for our purposes, the formation of this interface prevented further constriction, bundling, and alignment of the MWNT core. Surprisingly, miscible-solvent pairs created reasonably well-defined coaxial fibers with many bundled MWNT cores. This shows the importance of the MWNTs in altering the miscibility of the core solution. Whereas this regime did produce well bundled cores with few voids, it also produced many fiber artifacts that impeded accurate characterization of the MWNT core. Immiscible-solvent pairs produced the best bundled cores, though this regime had the most variation in fiber morphology, producing hollow cores, bundled cores, porous shells, and other fiber artifacts. The factors influencing the production of these undesirable morphologies in this regime are not clear, though there does appear to be a trend that forms the well bundled core. The number of bundled cores increased as the shell solvent approached the center of the PMMA solubility sphere. An ideally suited shell solvent for the polymer along with a semimiscible, semisolvent core solution (one that lies on the

’ REFERENCES (1) Breuer, O.; Sundararaj, U. Big Returns from Small Fibers: A Review of Polymer/Carbon Nanotube Composites. Polym. Compos. 2004, 25, 630–645. (2) Ericson, L.; et al. Macroscopic, Neat, Single-Walled Carbon Nanotube Fibers. Science 2004, 305, 1447–1450. (3) Zhang, M.; Atkinson, K.; Baughman, R. Multifunctional Carbon Nanotube Yarns by Down-sizing an Ancient Technology. Science 2004, 306, 1358–1361. (4) Sundaray, B.; Subramanian, V.; Natarajan, T.; Krishnamurthy, K. Electrical Conductivity of a Single Electrospun Fiber of Poly(Methyl Methacrylate) and Multiwalled Carbon Nanotube Nanocomposite. Appl. Phys. Lett. 2006, 88, 143114–143117. (5) Wang, G.; Tan, Z.; Liu, X.; Chawda, S.; Koo, J.-S.; Samuilov, V.; Dudley, M. Conducting MWNT/Poly(Vinyl Acetate) Composite Nanofibres by Electrospinning. Nanotechnology 2006, 17, 5829–5835. (6) Liu, L.; Tasis, D.; Prato, M.; Wagner, H. Tensile Mechanics of Electrospun Multiwalled Nanotube/Poly (methyl methacrylate) Nanofibers*. Adv. Mater. 2007, 19, 1228–1233. (7) Ko, F.; Gogotsi, Y.; Ali, A.; Naguib, N.; Ye, H.; Yang, G. Electrospinning of Continuous Carbon Nanotube-Filled Nanofiber Yarns*. J. Am. Chem. Soc. 2002, 15, 1161–1165. (8) Liu, J.; Wang, T.; Uchida, T.; Kumar, S. Carbon Nanotube CorePolymer Shell Nanofibers. J. Appl. Polym. Sci. 2005, 96, 1992–1995. (9) Hou, H.; Ge, J.; Zeng, J.; Li, Q.; Reneker, D.; Greiner, A. Electrospun Polyacrylonitrile Nanofibers Containing a High Concentration of Well-Aligned Multiwall Carbon Nanotubes. Chem. Mater. 2005, 17, 967–973. (10) Malone, E.; Lipson, H. Fab@Home: The Personal Desktop Fabricator Kit. Rapid Prototyping Journal 2007, 13, 245–255. (11) Reznik, S.; Yarin, A.; Zussman, E.; Bercovici, L. Evolution of a Compound Droplet Attached to a Core-Shell Nozzle under the Action of a Strong Electric Field. Phys. Fluids 2006, 18, 062101–062114. (12) Guise, O.; Ahner, J.; Jung, M.; Goughnour, P. Reproducible Electrochemical Etching of Tungsten Probe Tips. Nano Lett 2002, 2, 191–193. (13) Song, T.; Zhang, Y.; Zhou, T.; Lim, C.; Ramakrishna, S.; Liu, B. Encapsulation of Self-Assembled FePt Magnetic Nanoparticles in PCL 12749

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750

The Journal of Physical Chemistry C

ARTICLE

Nanofibers by Coaxial Electrospinning. Chem. Phys. Lett. 2005, 415, 317–322. (14) Hansen, M. Charles Hansen Solubility Parameters; Boca Raton: CRC Press, 2007. (15) Krevelen D. W. Properties of Polymers: Their Correlation with Chemical Structure, Their Numerical Estimation and Prediction from Additive Group Contributions; Elsevier, 1990. (16) Ham, H.; Choi, Y.; Chung, I. An Explanation of Dispersion States of Single-Walled Carbon Nanotubes in Solvents and Aqueous Surfactant Solutions Using Solubility Parameters. J. Colloid Interface Sci. 2005, 286, 216–223. (17) O’Connell, M; et al. Band Gap Fluorescence from Individual Single-Walled Carbon Nanotubes. Science 2002, 297, 593. (18) Gupta, P.; Elkins, C.; Long, T.; Wilkes, G. Electrospinning of Linear Homopolymers of Poly(Methyl Methacrylate): Exploring Relationships between Fiber Formation, Viscosity, Molecular Weight and Concentration in a Good Solvent. Polymer 2005, 46, 4799. (19) Yarin, A. L.; Zussman, E.; Wendorff, J. H.; Greiner, A. Material Encapsulation and Transport in Core-Shell Micro/Nanofibers, Polymer and Carbon Nanotubes and Micro/Nanochannels. J. Mater. Chem. 2007, 17, 2585–2599. (20) Bognitzki, M.; Czado, W.; Frese, T.; Schaper, A. Nanostructured Fibers via Electrospinning. Adv. Mater. 2001, 13, 70–72. (21) Zussman, E.; Yarin, A.; Bazilevsky, A.; Avrahami, R. Electrospun Polyacrylonitrile/Poly(methyl methacrylate)-Derived Turbostratic Carbon Micro-/Nanotubes**. Adv. Mater. 2006, 18, 348–353. (22) Dresselhaus, M.; Dresselhaus, G.; Saito, R.; Jorio, A. Raman Spectroscopy of Carbon Nanotubes. Phys. Rep. 2005, 409, 47–99. (23) Matsushita, A.; Ren, Y.; Matsukawa, K.; Inoue, H.; Minami, Y.; Noda, I.; Ozaki, Y. Two-Dimensional Fourier-Transform Raman and near-Infrared Correlation Spectroscopy Studies of Poly(Methyl Methacrylate Blends.... Vib. Spectrosc. 2000, 24, 171–180. (24) Temple, P.; Hathaway, C. Multiphonon Raman Spectrum of Silicon. Phys. Rev. B 1973, 7, 3685–3697. (25) Thompson, C.; Chase, G.; Yarin, A.; Reneker, D. Effects of Parameters on Nanofiber Diameter Determined from Electrospinning Model. Polymer 2007, 48, 6913–6922. (26) Li, D.; Xia, Y. Electrospinning of Nanofibers: Reinventing the Wheel? Adv Mater. 2004, 16, 1151–1170. (27) Ajayan, P.; Terrones, M.; de la Guardia, A.; Huc, V. Nanotubes in a Flash-Ignition and Reconstruction. Science 2002, 296, 705. (28) Panchapakesan, B.; Lu, S.; Sivakumar, K.; Taker, K.; Cesarone, G.; Wickstrom, E. Single-Wall Carbon Nanotube Nanobomb Agents for Killing Breast Cancer Cells. NanoBioTechnology 2005, 1, 133–139. (29) Song, J.; Fischer, C.; Schnabel, W. Thermal Oxidative Degradation of Poly(Methyl Methacrylate). Polym. Degrad. Stab. 1992, 36, 261.

12750

dx.doi.org/10.1021/jp201077p |J. Phys. Chem. C 2011, 115, 12742–12750