Polymer Nanocomposite Processing , Characterization , and Applications 2012

1 Department of Mechanical Engineering, Stevens Institute of Technology, Hoboken, NJ 07030, USA 2 Department of Polymer Engineering, The University of Akron, 250 South Forge Street, Akron, OH 44325, USA 3 DST/CSIR National Centre for Nano-Structured Materials, Council for Scientific and Industrial Research, Pretoria 0001, South Africa 4 School of Mechanical and Aerospace Engineering, Queen’s University, Belfast, UK

Polymer nanocomposites have attracted interest in the last few decades providing scope for improvement of various functional properties, such as mechanical, thermal, optical, rheological, magnetic, and electrical. The improvements in functional properties of polymers are achieved at very low loading of nanoparticles. While significant work has been done on preparation and properties of polymer nanocomposites, effort is still needed to elucidate the interrelationship between processing, morphology, and functional properties of nanocomposites. The properties of nanocomposites are affected by a large number of factors including microstructural distributions that are generated during nanocomposite processing as well as the state of nanoparticle distribution in polymer systems. It is believed that understanding of the relationship between processing, morphology and functional properties of nanocomposites will be very helpful in optimizing the ultimate properties of nanocomposites as well as improving the models for predicting properties of nanocomposite systems.
Considering the challenges in the area of processing of nanocomposites, we invited research articles for this special issue with a particular focus on the interrelationships between polymer nanocomposite processing, characterization, and novel applications. This special issue with a total of seven papers covers a wide range of areas related to fabrication of electrically conductive compounds of poly(carbonate), liquid crystalline polymer, and multiwalled carbon nanotubes (CNTs); gas permeability and morphological properties of nanocomposites prepared by the mixing of poly(lactic acid) (PLA), poly(butylene succinate) (PBS), and clay; synthesis and evaluation of gold/chitosan nanocomposites as a therapeutic agent for the photothermal therapy applications; novel direct-write drawing of CNT/poly(methyl methacrylate) (PMMA) composite microfibers; influence of laurolactam content on clay intercalation in poly(amide) 6, 12/clay nanocomposites synthesized by open ring anionic polymerization; fabrication and application of iron(III)-oxide nanoparticle/poly(dimethylsiloxane) composite cone in microfluidic channels; and finally layer-by-layer (LbL) nanoassembly of copper indium gallium selenium (CIGS) nanoparticle films for solar cell applications.
It is expected that this special issue will help readers with a wide range of backgrounds to understand the impact of various processing methods as well as nanoparticles on nanocomposite properties, and applications.

Introduction
Microfluidic device applied in chemical and biological process has drawn more and more attention in research due to versatile advantages such as low fabrication cost, short reaction time, and low reagent consumption. However, the laminar characteristics of flow in microchannel make it difficult to mix two separate flow streams together. Many mixing methodologies have been reported to enhance the mixing efficiency. One of the popular mixing methods is to generate chaotic advection utilizing particular geometric structures in microchannel, such as staggered herringbone pattern on channel surface [1,2], triangular-wave structures as obstacles along channel side walls [3], and spiral microchannels [4]. Furthermore, PDMS multilayer microfluidic chips with 3D configuration have been fabricated simply to carry out mixing work [5].
Magnetic beads are popular carriers for biological manipulation in microfluidic system under the application of magnetic field. Particularly, bioanalytic processes such as separation, capture, and recognition [6][7][8][9] could be conducted and integrated utilizing magnetic beads. In previous studies [10][11][12][13], the magnetic fields were applied from the outside of device with permanent magnet. If the magnetic fields could be generated in specified locations in a microchannel, the external magnet would not be needed and hence the device implementation would be made easier.
In this paper, we fabricate a magnetic nanoparticle/polydimethylsiloxane (PDMS) microcone for mixing fluids due to its three-dimensional asymmetric shape in a microfluidic chip. The fabricated microcone could also be used to capture magnetic beads with its intrinsic magnetism. two different fluids to enter the microchannel and then to encounter the micro-cone. The fluids can be collected in the buffer tank and then flow out through the outlet hole. There are four alignment holes on each PMMA layer for accurately assembling the microfluidic chip. Figure 2 illustrates the functionalities of the proposed micro-cone in the microfluidic chip. Let the first fluid in the microchannel contain particles A, which do not possess magnetism, and particles C attached on magnetic beads. The second fluid contains only particles B. The micro-cone is proposed to separate particles A from particles C and then to mix particles A with particles B together. Before the fluids pass through the micro-cone, they are laminar flows in the microchannel. The particles C would be captured by the micro-cone due to the magnetic force when the first fluid passes through the micro-cone. Beyond the micro-cone, the first and the second fluids could be mixed due to the turbulence caused by the cone shape. Therefore, particles A and B could interact. The particles C captured on the micro-cone and the mixture of particles A and B can then be analyzed for versatile applications of lab on a chip. It should be noted that the position of the proposed micro-cone has to be deviated from the central line of the microchannel for enhancing the mixing effect. In our design, the deviated distance is 10% of the microchannel width. Figure 3 details the process for fabricating the microfluidic chip. It is composed of three PMMA layers. All the layers are machined using a CO 2 laser. Each layer is made of a bulk PMMA plate of 1 mm in thickness. To assemble the microfluidic chip, both sides of the middle layer are coated with adhesive of 50 μm in thickness (Figure 3(a)). Therefore, the total depth of the microchannel is 1.1 mm, and the width is 2 mm.

Fabrication
To fabricate the micro-cone, the PDMS prepolymer and curing agent with 10 : 1 weight ratio are prepared, then the iron(III)-oxide nanoparticles are dispersed and mixed thoroughly in the PDMS for 10 minutes using regular stirring machine. The mixture is then put in a vacuum chamber for degassing. The iron(III)-oxide nanoparticles (nanopowder of 544884-5G Iron (III)-oxide, SIGMA-ALDRICH) are comprised of primarily the gamma-form Fe 2 O 3 which exhibits superparamagnetic behavior [14][15][16][17]. The average diameter of the nanoparticles is 50 nm. After degassing, the nanoparticle/PDMS composite is poured into a syringe and then degassing process is conducted again. An automatic dispenser is used to extrude the nanoparticle/PDMS composite into the microchannel. In the beginning of the dispensing process, the initial portion of the nanoparticle/PDMS composite is disregarded in order to obtain constant volume of every single drop. A reference plate with a defined area is put onto the sample stage for aligning the dispenser (Figures 4 and 5). Therefore, the syringe needle can be precisely placed above the designated area. After finishing the alignment, the reference plate is replaced by the bottom layer of the microfluidic chip. Then a drop of nanoparticle/PDMS composite is dispensed onto the bottom layer (see Figure 3(b)). The dispensing pressure is 0.25 MPa. After dispensing the nanoparticle/PDMS composite, a permanent magnet is placed above the composite (Figure 3(c)). The composite becomes a cone shape due to the external magnetic field. Then the cone is cured at 70 • C for 30 minutes. The fabrication environment is at 23 • C and 71% relative humidity. After the cone is cured, the three layers of the microfluidic chip are assembled together (Figure 3(d)). There are two setups to facilitate the fabrication process. One is the automatic dispensing system (Figure 4), and the other is the magnetic platform for generating the cone shape of the composite ( Figure 5). The automatic dispensing system consists of a dispenser (SR-330D), two CCD cameras, and a sample stage. The dispenser is used to extrude iron(III)-oxide nanoparticle/PDMS composites. It features three-axis movement with 50 μm resolution and a controller for adjusting the dispensing pressure and duration. The two CCD cameras assist the alignment of the syringe needle on the substrate and monitor the dispensing process. The sample stage is used to displace the substrate. The magnetic platform includes a vertical manipulation stage, a permanent magnet, a CCD camera, a light source, a sample stage, and a hot plate. The permanent magnet is connected to the vertical manipulation stage with 10 nm resolution. The high resolution of the manipulation stage is necessary to finely adjust the distance between the bottom of the magnet and surface of the substrate. The CCD camera is used to monitor the deformation of the uncured nanoparticle/PDMS composite. There is also a sample alignment stages which is placed on the hot plate so that the alignment would not be deviated in the curing process. The composites with different weight ratios of the nanoparticles are added into the PDMS for evaluating the fabrication parameters. The weight ratios are prepolymer: curing agent: iron(III)-oxide nanoparticle = 40 : 4 : 1, 40 : 4 : 3, 40 : 4 : 5, respectively. Therefore, the weight ratios of the nanoparticles to the composite are 2.22%, 6.38%, and 10.20%, respectively. Different magnetic fields which are controlled by the distance (D) from the bottom of the magnet to the substrate surface are applied. The corresponding magnetic fields of variant distances between substrate and magnet are measured by tesla meter (TM-701, KANETEC CO., LTD). The results are summarized in Table 1.

Results
The size of drops could be controlled by the diameter of syringe needle, dispensing pressure and dispensing duration. The stainless steel syringe needle which its inner diameter is 0.15 mm and outer diameter is 0.3 mm was used. The dispensing pressure is fixed at 0.25 MPa. Figure 6 shows the images of PDMS droplets without nanoparticles at different dispensing durations t = 2.5, 5, 7.5, and 10 s. The variation of the droplet size at the same dispensing parameters is not obvious before and after the nanoparticles adding into the PDMS. This parametric study is conducted for obtaining the desirable droplet diameter (d) and height (h) by controlling the dispensing duration. Since the height of the microchannel is 1.1 mm, the initial height of the PDMS droplet should be relatively smaller. All the initial heights of the PDMS droplets using different dispensing durations in our tests are smaller than 1.1 mm, as shown in Figure 7(a). The average height is h = 0.09 (±0.007), 0.14 (±0.005), 0.17 (±0.009), 0.19 (±0.018) mm for t = 2.5, 5, 7.5, 10 s, respectively. Since, in this design, the channel width is 2 mm and the deviated distance from the cone center to the central line of the microchannel is 0.2 mm, the diameter of the PDMS droplet should be smaller than 1.6 mm. The obtained cone diameter is shown in Figure 7(b). The average diameter is d = 1.28 (±0.029), 1.51 (±0.032), 1.64 (±0.061), and 1.79 (±0.067) mm for t = 2.5, 5, 7.5, and 10 s, respectively. Therefore, the optimum dispensing duration is 5 s in this test.
Journal of Nanomaterials  For each parameter in Table 1, the images of the cone formation are captured every 5 s in 60 s. For the 2.22% weight ratio with 1.0 mm magnetic height, the 6.38% weight ratio with 1.25 mm magnetic height, and the 10.2% weight ratio with 1.5 mm magnetic height, the nanoparticle/PDMS droplets are pulled up and touch the magnet immediately. Therefore, the images of the cone formation for these three particular samples are captured every 1 s. Figure 8(a) shows the height increase (Δh) of the droplet deformation as a function of process time. All the data in Figure 8(a) are from the samples of the 10.20% nanoparticles with the magnetic height of 1.5 mm, 1.75 mm, and 2.0 mm, respectively. For the same process time, the height increase is larger for the lower magnetic height. The error bars in Figure 8(a) stand for the standard deviation from five samples. The deviations are attributed to the vertical and horizontal alignment between the magnet and the center of the droplet. The nondispersed nanoparticles in the uncured PDMS might also affect the repeatability of the cone formation.
The inset of Figure 8(a) shows the height increase of the droplet deformation for the magnetic height of 1.5 mm. The five samples are pulled up and reach the magnet at the process time of 11,12,19,20, and 23 s, respectively. Because the magnet is very close to the droplet, the magnetic force is strong and sensitive to the magnetic height. The variation of the initial droplet might significantly deviate the process time for the droplet to reach the magnet. The height increase of the droplet with 1.5 mm magnetic height and the nanoparticle weight ratio of 2.22%, 6.38%, and 10.20%, respectively, are shown in Figure 8(b). The height increase is larger for the higher weight ratio of nanoparticles. Table 2 shows the height increase at the process time of 30, 40, 50, and 60 s, respectively. The evolution of the height increase with the increasing process time for the samples with 1.5 mm magnetic height and 2.22% nanoparticles is close to that with 1.75 mm magnetic height and 10.2% nanoparticles.
For the 1.75 mm magnetic height in Figure 8(a) and the 6.83% nanoparticles in Figure 8(b), the slope of the curve changes in a piecewise manner with the process time. In the beginning of applying the magnetic field, the slope is steep for the process time from 0 s to 5 s in Figure 8(a), while the slope becomes smaller for the process time of 5∼25 s. Figure 9(a) shows the fabricated microfluidic chip, and Figure 9(b) shows the cross-sectional image of the microchannel. The nanoparticle/PDMS cone is successfully fabricated in the microchannel. The height of the micro-cone is 300 μm, and the deviated distance from the central line is 0.3 mm.

Discussions
The method to form the cone shape of the nanoparticle/ PDMS composite is to exploit the uncured PDMS deformation as result of being dragged by the nanoparticles. The iron(III)-oxide nanoparticles can be attracted by the permanent magnet. They are mixed in the PDMS and enveloped by the entangling polymer chains. As the magnetic nanoparticles are attracted by the magnetic field above, they would move upward and carry the PDMS together. Because the distribution of the magnetic is hyperbolic (Figure 10(a)), the center of the nanoparticle/PDMS composite is subjected to a relatively stronger magnetic force. Conversely, the rim of the composite is subjected to a weaker magnetic force. As a result, the nanoparticle/PDMS cone can be obtained. When the magnetic field is applied, the nanoparticles in the composite start to aggregate and to form radiated lines on the cone surface ( Figure 10(b)) [18]. There are more nanoparticles aggregating on the tip of the cone in    Figure 11: Illustration of the mechanism of the composite deformation.
comparison to the rim region. Therefore, a darker color around the cone tip is observed. Before the nanoparticle/PDMS composite is cured, the nanoparticles are allowed to move in the PDMS. As soon as the PDMS is cured, the nanoparticles cannot move anymore. This results in the gradient magnetism due to the nonuniform distribution of the nanoparticles. Originally, the nanoparticles distribute uniformly in the PDMS. Then the magnetic field on the droplet attracts the nanoparticles which are closer to the droplet surface. These nanoparticles move upward, causing the PDMS to deform as illustrated in Figure 11(a). The momentum of the nanoparticles causes the PDMS to deform rapidly, and hence, the initial slope in Figure 8(a) is steep. When the magnetic force and the restoration force of the polymer chains reach transient equilibrium, the speed of deformation becomes slower, as illustrated in Figure 11(b). Although it is in equilibrium at 5 s in Figure 8(a), the height increase reduces the distance between the magnet and the top of the droplet. Therefore, the magnetic force exerting on the nanoparticles which are close to the droplet surface increases. Nevertheless, the nanoparticles are insufficient to cause rapid deformation of the PDMS. Hence the slope becomes smaller for the process time during 5∼25 s. Since the top of the droplet is subjected to a relatively stronger magnetic force, the nanoparticles in the internal portion of the droplet will move to the surface gradually and then accumulate on top of the droplet, in the center of the drop top, as illustrated in Figure 11(c). When the nanoparticles are sufficient and the droplet reaches the appropriate height, the droplet deformation becomes more rapid for the process time during 25∼30 s in Figure 8(a). The nanoparticles keep moving to the droplet top and hence the PDMS deformation concentrates at the top of the droplet until the cone shape is obtained. The more nanoparticles in the PDMS, the stronger magnetism the cured cone has. However, the shorter magnetic height would cause the cone to touch the magnet. The 6.38% nanoparticles with the 1.5 mm magnetic height are the optimum parameters in this study and hence are used in the following fabrication of the microfluidic chip. Figure 12 shows the deformation process for the 10.2% nanoparticles at 0, 10, and 20 s in Figure 8(a). The first  Figure 8(a), the micro-cone will touch the magnet and then break into two parts, as shown in (Figure 12(j)). The smaller portion of the composite is attached to the magnet while the large portion remains on the substrate, as shown in Figure 12(k). The process images for the magnetic height of 1.75 mm are shown in Figures 12(d)-12(f). The cone is formed gradually but cannot be clearly observed for the process time shorter than 20 s. The process images for the magnetic height of 2.0 mm are shown in Figures 12(g)-12(i). The composite barely deforms at this magnetic height because the magnetic force is not strong enough to overcome the restoration force of the polymer chains.

Micorchannel Mixing Test
In order to enhance the mixing efficiency, three nanoparticle/PDMS micro-cones are fabricated in the microchannel. Two kinds of the fluids, deionized water and glycerin, are used for microchannel mixing test. The width of the channel is 2 mm, the height of channel is 1.1 mm, the hydraulic diameter is calculated as 1.05 mm, and the flow rate is set as 10 mL/h. The Reynolds numbers, Re = ρvD H /μ, of these two fluids are listed in Table 3. In the same condition of the diameters of the channel and the flow rate, however, the Reynolds numbers are different mainly due to variations in the dynamic viscosity, μ. the fluidic test of deionized water without and with microcone, respectively. Because the Reynolds number is 2.96, which is not low enough, the laminar phenomenon is not obvious. Different diffusion results across the interface between the two fluids are also observed under different Reynolds numbers. In Figure 13(a) without the micro-cone, the distribution of the red-ink water crosses the whole channel width at a relatively downstream location as marked by the line A. In Figure 13(b) with the micro-cones, the distribution of the red-inked water crosses the whole channel width at a relatively upstream location as marked by the line B. The result demonstrates that the micro-cones effectively enhanced the mixing efficiency. Figures 13(c) and 13(d) show the fluidic test of glycerin and the Reynolds number is 3.45 × 10 −3 (Re 1). The laminar phenomenon could be seen clearly in Figure 13(c). In Figure 13(d), although the red ink glycerin does not distribute across the channel width through the whole channel length, the flow field is redistributed as passing through the micro-cones and the flow of the red-inked glycerin becomes wider.

Magnetism of the Micro-Cones Test
Streptavidin has extremely high affinity for biotin, so the streptavidin-coated magnetic particles and fluorescently labeled biotin are combined to carry out the magnetism Journal of Nanomaterials of the micro-cones test. The 50 μL of streptavidin-coated magnetic particles (Dynabeads M-270 streptavidin, Invitrogen, 2.8 μm in diameter, 10 mg/mL) are mixed with 1 mL of biotin which were labeled with fluorescence (Biotin-4fluorescein, Molecular Probes, Invitrogen) at 20 μg/mL in concentration for 30 min. In the beginning, the deionized water is injected in and full of the microchannel to expel air. The reagent is continued to flow into the microchannel at 5 mL/h flow rate. Then, the deionized water is used to wash the reagent out of the channel. The chip is mounted on a fluorescence microscope (BX51, Olympus, Japan), and the images are taken by CCD camera. Figure 14(a) shows the microchannel full of reagent, which is observed under the microscope without the blue light filter. Figure 14(b) is the same location of the microchannel as Figure 14(a) but is observed by microscope with blue light filter. The fluorescent particles captured on the micro-cone are observed. In order to confirm that the particles are truly attracted on the microcone, the deionized water is used to wash the channel again. The particles still remain on the micro-cone after the wash, which proves that the iron(III)-oxide nanoparticle/PDMS cone has the magnetism and could be used in capturing magnetic particles.

Conclusion
An approach to fabricate the 3D iron(III)-oxide nanoparticle/PDMS cone in a microchannel has been demonstrated. The 6.38% weight ratio of the nanoparticles in the PDMS was used. An automatic dispenser was used to align the composite cone in the microchannel and to control the diameter of the composite droplet. The dispensing pressure is 0.25 MPa, and the dispensing time is 5 s. A permanent magnet connected to the vertical manipulation stage was used to attract the droplet to form the cone shape. The distance between the magnet and the PMMA substrate is 1.5 mm. The cone is cured by heating at 70 • C for 30 mins. The micro-cone has been successfully fabricated in the microchannel. The height of the cone is 300 μm, and the deviated location of the cone from the central line of the microchannel is 0.3 mm. The cone has the magnetism due to the arrangement of the nanoparticles and can enhance the mixing efficiency. The fabricated cone in the microchannel could be applied in lab on a chip in the future.

Introduction
Polymer/organically modified layered silicate nanocomposites have attracted much attention in the past few decades because of their demonstrated significant enhancement of a large number of the properties including barrier, thermal, mechanical, rheological, flammability resistance, and other physical properties relative to an unmodified silicate or neat polymer resin. These improvements are mainly achieved at lower silicate loadings (≤5 wt%), which result in a high aspect ratio along with a large surface area. The concentration of silicate loading required for the preparation of nanocomposites is a fraction of what is typically required with conventional composites. Moreover, the improvement in these nanocomposites is matchable or higher than those for the conventional composites regarding the filler content. Till now, various nanocomposites have been developed and are being developed using layered silicate clay minerals as a reinforcing phase due to its easy and abundant availability, low cost, and, more importantly, environmental friendliness [1]. The rapid growth of the use of plastics in our daily life has become an important issue due to the resulting global environmental problems. Generally, polyethylene, polypropylene, polystyrene, poly(vinyl chloride), and so forth are the most produced, used, and then discarded in the environment. Especially, their packaging materials end up as dangerous, undegradable wastes. Due to the unavailability of satisfactory landfill sites, production of large amounts of carbon dioxide and toxic gases during incineration, which again contribute to the environmental pollution, and the more expensive and reduced quality of recycled plastic products, there has been a drive to concentrate on biodegradable polymeric materials. Keeping these reasons in mind, the development of biodegradable materials with superior properties has become a challenging and interesting subject for researchers, scientists, and engineers throughout the world today. It is expected that, like other polymer nanocomposites, biodegradable polymer-nanocomposites will also demonstrate great enhancements in physical and material properties. Therefore, these biodegradable nanocomposites may become a possible solution for the global environmental problem caused by plastic wastes.
Various researchers [2,3] have studied biodegradable polymers which are derived from renewable and petroleum resources; these include poly(L-lactide) (PLA), poly(butylene succinate) (PBS), poly(butylene adipate terephthalate), poly(ethylene terephthalate succinate), and poly(εcaprolactone). The PLA, which is produced from a feedstock of corn rather than petroleum, has recently been used for packaging material. Packaging material for food should provide sufficient barriers against water vapour to prevent the food degradation, against atmospheric gases to prevent the oxidation of the food material, and against volatile organic compounds to preserve the aromas and the flavors of the food [4]. In food packaging, mass transfer both from the food to the polymer and vice-versa is responsible for the contamination of food and also causes degradation of the packaging material. Since the PLA is produced from a lactide monomer coming from a lactic acid which is a nontoxic substance already found in the human body, food contamination can be neglected when the PLA is used as a packaging material. Therefore, it can be assumed that the determination of permeability of oxygen, water vapour, and aromas in PLA used for food packaging material is an important parameter.
PLA is a linear thermoplastic polyester. It has good mechanical properties, thermal plasticity, and good biocompatibility and it can be easily fabricated. However, the cost associated with the ring-opening polymerization of lactide monomer and brittleness of the PLA limit the commercial use of PLA polymer. Therefore, blending of PLA with another biodegradable polyester polymer may be necessary such that the resultant blend would overcome the limitations of PLA achieving good qualities by the addition of another polymer and also reduced the final cost.
In this study, the polymer selected to blend with PLA was PBS which is one of the most promising polymers in the family of synthetic biodegradable polyesters. PBS, chemically synthesized by the polycondensation of 1, 4-butanediol with succinic acid [5], has high flexibility, excellent impact strength, and thermal and chemical resistance [6]. Excellent processability of the PBS might make processability PLA better. Among various physical characteristics, gas permeability properties of the nanocomposites are mainly focused and then correlated with their morphological properties from both XRD and TEM.

Background of Gas Barrier Modeling
Theoretical models have been developed for predicting the properties of composite materials based on the properties of the pure components and the morphology of the composite. The general objective of such theories has been to predict the performance of the composite for a given set of components [7]. These theories enable a simple route for evaluation of the individual contribution of the component's properties, such as matrix type, volume fraction, filler aspect ratio, filler orientation, and filler distribution. Various studies [8,9] have shown that aspect ratio and particle distribution in the polymer matrix are important factors for predicting the properties of polymer nanocomposites; suitable values help in achieving high enhancement in properties of nanocomposites.
Enhancements of various properties are well known in polymer-silicate nanocomposites; their main dependence is on the distribution of the silicate layers in the polymer matrices. However, the main factors behind the improvement in gas barrier properties are not yet fully understood. As clays are crystalline materials, they are believed to increase the barrier properties by creating a maze or "tortuous path" that restricts the progress of the gas molecules to pass through the polymer matrices [10]. Note that the idea of "tortuosity" has been already introduced by Michaels and Parker [11] accounting for the impedance to flow offered by the irregular intercrystallite passages of polyethylene.
Nielsen [10] developed a theory for diffusion through filled polymer systems. According to this theory, if the fillers are impenetrable to a diffusing gas or liquid molecule, then the diffusing molecule must go around the filler particles. He utilized the tortuosity factor (τ) which is defined as the ratio of the actual distance (d ) that a penetrant must travel to the shortest distance (d) that it would have travelled in the absence of the layered silicate. Nielson added that thin, plate-like fillers, having a large aspect ratio (L/2W) (where L is the length of the filler particle and W is the width of the filler plates), can dramatically reduce the permeability only if the particles are oriented in such a way that their flat surfaces are parallel to the surfaces of the film. Nielsen also stressed that incomplete filler dispersion, voids, and only partial orientation of the particles should result in a reverse affect on permeability. The tortuosity factor (τ) is expressed as where φ s is the volume fraction of the sheets. The effect of tortuosity on the permeability can be expressed as where P s and P p represent the permeabilities of the nanocomposite and neat polymer, respectively. Bharadwaj [12] invoked a simple model, developed earlier by Nielsen, correlating the sheet length, concentration, relative orientation, and state of aggregation of the filler in the polymer matrix. This model gave further guidance in the design of better barrier materials for nanocomposites. Bharadwaj [12] predicted, from (2), that the relative permeability (P s /P p ) is a function of the silicate sheet length. The relative permeability of H 2 O in a polyimidemontmorillonite nanocomposite reproduced the proposed concentration dependence according to (2) [13], showing a progressive decrease in relative permeabilities as a function of increasing concentration and sheet length. Extremely large length-to-thickness ratio of the silicate sheets (30 to ∼2000 nm in length, 1 nm thick) showed a significant increase in tortuosity. It is interesting to note that no significant reduction in the relative permeability was noticed, regardless of the clay sheet length, beyond Φ s = 0.05. This is also in agreement with experimental observations [13][14][15][16][17][18].
Bharadwaj modified the tortuosity factor to include the orientational order (S), writing the relative permeability as follows: The above expression reduces to (2) when S = 1 (planar arrangement, i.e., maximum tortuosity) and provides approximately the permeability of the neat polymer when S = −1/2 (orthogonal arrangement). Further, Bharadwaj concluded that incorporating sheets with L > 500 nm with random orientation was more beneficial for the barrier properties than the case where the sheets were aligned perpendicular to the diffusing path.
Exfoliation appears to be the critical factor in determining the maximum performance of polymer nanocomposites for barrier applications. A significant decrease in the permeability of O 2 gas through the near exfoliated PLAnanocomposite relative to the neat polymer (PLA) was clearly observed [8]. Nielsen [10] explained that at L = 1000 nm the relative permeability was ∼0.05 for W = 1 nm and 0.2 for W = 7 nm. While for L = 500 nm, relative permeability changed from 0.1 to 0.3, respectively, for W = 1 to 7 nm, thus underscoring the need for complete delamination of the silicate sheets.
Messersmith and Giannelis [19] observed a dramatic reduction in water permeability through composite films of poly(epsilon caprolactone)/organically modified mica-type silicate containing modest amounts of silicate. This reduction was due to the dispersion of impermeable, high aspect ratio silicate layers within the polymer matrix. On the other hand, Cava et al. [20] compared the barrier properties of biodegradable plastics and their nanocomposites with PET for food packaging applications. Biodegradable materials showed higher oxygen permeability compared to PET at 21 • C and 40% relative humidity; however, their nanocomposites demonstrated marginally lower permeability to the test gas than the neat polymer.
Both the Nielsen [10] and Bharadwaj [12] models are based on arguments related to the tortuosity factor. Further, equations related to the permeability and tortuosity were developed to understand the diffusion of small molecules in conventional composites; however, they do extremely well in predicting and relating the experimental results for the relative permeability in polymer/layered silicate nanocomposites. The main assumption of the above models is that the sheets are oriented, such that the sheet normal is coincident with the direction of diffusion (S = 1). Clearly, this arrangement results in the highest tortuosity. Any deviation from this arrangement would lead to deterioration in the barrier properties. Also, a sheet-like morphology (W = 1) is particularly efficient at maximizing the path length of the penetration due to the large length-to-width ratio compared to spheres or cubes.
The purpose of this research was to better understand the influence of clay concentration on the gas barrier properties and morphology of biodegradable nanocomposites of PLA/PBS/C30BX (an experimental grade related to Cloisite 30B) nanocomposites having a fixed composition of PLA and PBS, at 80 wt% and 2 wt%, respectively, with various clay loadings. Various aspects of selected morphological, oxygen, and water vapour permeability properties of the nanocomposites were compared to those of the polymer blend without clay. In particular, this study mainly aimed to obtain an extensive understanding of the gas barrier properties of biodegradable polymer nanocomposites using the Bharadwaj model based on detailed morphological analysis from TEM images. Comparison of oxygen relative permeabilities obtained through experimental investigations with relative permeabilities calculated from the Bharadwaj model was done. Bharadwaj model predictions were also discussed in terms of the morphological structure of the nanocomposites reanalyzed from TEM images reported [21].

Experimental
3.1. Materials. Both biodegradable poly(lactic acid) (4032Dgrade, Nature Works, Cargill Dow LLC) and poly(butylene succinate) (PBS) (grade G4460, Ire Chem. Co., Republic of Korea) were used in this study for blending. Densities of PLA and PBS were 1.240 g/cm 3 and 1.230 g/cm 3 , respectively. The PLA had crystalline melting point range between 150 and 170 • C, while PBS had a melting range between 95 and 114.7 • C. Furthermore, an experimental grade of Cloisite 30B (SCPX 3016, Southern Clay Products (SCP) Inc., USA) organoclay, indexed as C30BX, was used. The experimental grade clay is known to provide better dispersion in biodegradable polymers, such as PLA and PBS matrices, as compared to the commercially used C30B. This clay has basically the same chemical treatment as C30B but it was achieved via a different mechanism. The C30BX, was produced by a cation exchange reaction whereby natural MMT (Na + -MMT) was chemically modified with bis-(2hyroxyethy) methyl (hydrogenated tallow alkyl) ammonium cations (C 18 H 37 )-N + (C 2 H 4 OH) 2 CH 3 ).

Preparation of PLA/PBS/Clay Nanocomposites.
Before processing through an extruder, the pellets of both PLA and PBS were first dried in a vacuum oven at a temperature of 50 • C for 48 h to remove moisture. Drying of these polymers is very essential as moisture in these polymers induces in hydrolytic chain scission during the processing at high temperature. All the PLA/PBS/clay nancomposites were prepared by melt compounding using a Brabender twinscrew extruder at temperature 180 • C and screw speed at 40 rpm. All the nanocomposites were passed twice through the extruder to achieve better and homogeneous mixing of all the compositions. The composition of the binary blends of the PLA/PBS was kept fixed at 80/20 (wt%/wt%) for all the nanocomposites. Nanocomposites of various clay loadings, 1, 3, 5, 7, and 10 by wt%, were prepared for the investigations in this study.
All the components were premixed in a plastic bag by tumbling well and were fed simultaneously into the extruder to maintain the overall consistency. The extruded nanocomposites were pelletized and dried again at 50 • C in a vacuum oven over night. For the determination of oxygen and water vapour transmission rates, film samples of approximately 0.3 mm thickness, obtained by compression molding at 200 • C and pressure of 40 kN for 5 minutes, were used and the samples were then cooled under pressure (20 MPa) for 2-3 min. All these samples were used for morphological and gas barrier property evaluations.

Analysis of PLA/PBS/Clay Nanocomposites.
To examine morphology of the nanocomposites, wide angle X-ray diffraction (WAXD) (Philips PW1847 X-ray diffractometer) was carried out with a reflection geometry and CuK α radiation (wavelength λ = 0.154 nm) operated at 40 kV and 30 mA within the range of scattering angles (2θ) from 1 to 10 • . To investigate the internal structure and dispersion state of the nanocomposites, ultrathin sections (less than 100 nm) were microtomed using a Super NOVA instrument with diamond knife. Consequently, the ultrathin samples were investigated using a transmission electron microscopy (TEM) (Philips CM200) operated at 100-kV accelerating voltage [22].
For gas permeability study, at first oxygen permeability was examined with a MOCON Ox-Tran 2/21 instrument (Modern Controls Inc., USA) according to ASTM D3985 standard as previously described [23]. The values of oxygen permeability were obtained at 23 • C and 0% RH, with the exposed area and thickness of 50 cm 2 and 0.3 mm, respectively. A mixture of 97% N 2 with 3% H 2 was used as a carrier gas and 99.9% oxygen was used as the test gas. The permeability of Mylar (polyester) film (MOCON manufacturer) was used to calibrate the instrument, in which the permeability of this Mylar film was already known. The film samples were placed in two test (permeability) cells. After installing the samples, a continuous flow of gas was maintained on both sides of the barrier material. Firstly, nitrogen gas was passed through both surfaces to remove any oxygen in the sample. The nitrogen on one side was then replaced by oxygen, and the nitrogen flow on the other side then swept the surface to extract any oxygen that diffused through the material. The diffusing oxygen was measured by a coulometric sensor detector that is sensitive only to oxygen. For the completion of one test after films were loaded in the test cells, it takes 12 to 16 h. Before each measurement, the samples were kept at 50 • C in a vacuum oven for an overnight, at least four samples of each film for their analysis.
The water vapour transmission rate (WVTR) of each PLA/PBS/C30BX nanocomposite sample was measured using a Modern Controls Inc. (MOCON) Permatran-W1 (USA), according to ASTM standard F 1249 at 100% relative humidity (RH) and 37.8 • C temperature. The exposure area for the WVTR measurements was 50 cm 2 and thickness of the film sample was 0.3 mm, as above. Water vapour transmission rate through the plastic film was measured using a modulated infrared sensor. Water vapour transmission rates can be evaluated for the films by using other solutions, that is, sodium chloride, sodium acetate, and magnesium chloride giving different relative humidity (RH). In this study, very pure water (100% RH) was used as the samples did not absorb much water. Calibration of this instrument was also done using a Mylar (polyester) film of 5 mil thickness supplied by MOCON, which has a WVTR of 4.55 g/m 2 ·day at 37.8 • C and 100% RH. Test films mounted in the diffusion cell were initially exposed to a continuous flow of air across the upper side, while the bottom side was exposed to water vapour from the moistened pad (reverse osmosis water) in the humid cavity. This is called conditioning which took one night. Conditioning of the sample is required to allow the establishment of an equilibrium transmission rate. The flow rate of the air during the test was set to 60 cm 3 /min. After conditioning, actual water vapour transmission (WVTR) was noted, usually after 12-16 hours. An average of five values is reported in this work after normalizing the unit values according to the thickness of the sample.

Results and Discussion
The characteristic peak of pristine C30BX in WAXD patterns occurred at the diffraction angle 4.96 • . By using Bragg's law, the d-spacing, also called the basal spacing, was calculated as 1.78 nm for C30BX [23]. This is in close agreement with that of commercially available C30B organoclay (i.e., 1.75 nm). The WAXD patterns of the PLA/PBS/C30BX nanocomposites from our previous work ( Figure 1 [23]) demonstrated that the nanocomposite having 1 wt% of the C30BX did not show any diffraction peak in the range used, probably due to its low clay content. The nanocomposite of 3 wt% of C30BX exhibited a broad and extremely weak reflection at approximately 2 • diffraction angle. Moreover, it also showed a slightly visible peak at approximately 6 • diffraction angle. However, it is not sufficiently clear to be called a peak. For 5, 7, and 10 wt% of clay nanocomposites the peak shifted to lower angles at 2.3 • , 2.2 • , and 2.2 • , respectively. This indicated that the polymer chains were intercalated into the C30BX layers. The peak for the 5 wt% sample was relatively broad. Compared to the characteristic peak of neat C30BX the gave an interlayer spacing (d 001 ) of 1.78 nm, the interlayer d-spacing increased to 3.84 nm, 4.01 nm, and 4.01 nm in nanocomposites having 5, 7, and 10 wt% of C30BX concentration, respectively. Shifting of the peak occurred due to the intercalation of the polymer chains into the clay galleries, while broadening of the peak in 3 and 5 wt% of polymer/clay nanocomposite indicated a distribution of intercalated layer spacings. Krikorian and Pochan [24] revealed that the broadening of the peak can also occur due to the partial disruption of parallel stacking Journal of Nanomaterials or layer registry of the pristine organoclay, which means the existence of some delamination of clay platelets. Therefore, it can be concluded that a mixture of delaminated and intercalated structure was observed. On the other hand, nanocomposites with 7 and 10 wt% of clay clearly demonstrated predominantly intercalated structure, in which the intensity of these intercalated clay peaks gradually increased with the addition of the clay, due to more diffracted X-rays from the parallel assembly of the additional intercalated clay [24]. Interestingly, it was noticed that for the 3, 5, 7, and 10 wt% clay, the main diffraction peaks were accompanied by the appearance of another peak at 2θ of approximately 6 • , at the same location with the same basal spacing for each nanocomposite. Note that an additional small peak at a diffraction angle (2θ) below 10 • for various PLA/organoclay nanocomposites has been reported [25]. Based on this assumption, the secondary peak could be correlated to the XRD spectra of the PLA and PBS polymers itself in PLA/PBS/C30BX nanocomposites. On the other hand, the WAXD profiles of nanocomposites of PLA/7, PBS/7 and PLA/PBS/7 are shown in Figure 1. The nanocomposite of 7 wt% of C30BX clay in PLA had an increased interlayer distance of 3.08 nm (2θ = 2.86 • ), while the PBS/C30BX (93/7 wt%) nanocomposite had a clay interlayer distance of 3.06 nm (2θ = 2.88 • ), which indicated that PBS polymer also had a tendency to intercalate between C30BX layers. A better association or interaction between C30BX clay and PLA polymer could be concluded because of the larger interlayer distance of PLA/C30BX nanocomposite than that of PBS/C30BX nanocomposite. It was interesting to note further that when both PLA and PBS polymers were mixed with C30BX (PLA/PBS/C30BX: 80/20/7 wt%), intercalation occurred. On the other hand, Wang et al. [26] studied the effect of clay in polypropylene (PP)/polystyrene (PS) blends, finding that when organophilic clay montmorillonite (OMMT) was mixed with either PP or PS, the layer distance of OMMT was around 3.2-3.3 nm. But, when both PP and PS were mixed with OMMT, the layer distance was increased further to 3.79 nm. Thus, it was concluded that both immiscible polymers intercalated between the OMMT clay layers. Lim et al. [27] and Chen et al. [28] also confirmed the intercalation of immiscible polymer blend matrices within OMMT, which is in contrast to the results obtained in this study. It is noted that Chen et al. [28] also used a blend of PLA and PBS, but normal C30B, and they did not study the barrier properties.
It can also be found from Figure 1 that the addition of PBS to the PLA/C30BX (C30BX-7 wt%) nanocomposite system shifted the basal peak towards the lower diffraction angles (2θ) from 2.86 • to 2.2 • for PLA/PBS/C30BX (C30BX-7 wt%). Intercalated peaks for both nanocomposites (PLA/C30BX and PBS/C30BX nanocomposites) were obtained at almost similar diffraction angles but the intensities were quite different. The characteristic peak height in PLA/C30BX nanocomposite was smaller as compared to PBS/C30BX nanocomposite. The low intensity of the peak can be explained by the tendency for delamination of C30BX in the PLA/C30BX mixture [29]. Whether C30BX has better affinity with PLA or PBS polymers can be determined by using the X-ray data of the ternary mixture. Figure 1 also showed that both PLA and PBS nanocomposites have peaks at the same diffraction angle; however, peak of PBS/C30BX nanocomposite is very sharp and has high intensity as compared to PLA/C30BX nanocomposite. This suggested that PLA chains have better affinity towards C30BX clay. Lee et al. [30] also reported similar results with biodegradable aliphatic polyester blends.
Meanwhile, our previously reported SEM image of the PLA/PBS blend of 80/20 by wt% showed that the blend was immiscible [21]. Bhatia et al. [31] noticed a clear dispersed phase of PBS in all their compositions of PLA/PBS blends while investigating the compatibility between PLA and PBS polymers and concluded that for PBS contents higher than that in the 80/20 (PLA/PBS) blends, that is, 20%, the traditional morphology of immiscible blends could be seen. A similar type of immiscible morphology trend was reported for PLA and poly(butylene succinate adipate) blends by S. Lee and J. W. Lee [32]. On the other hand, TEM can further confirm the observations and the conclusions made from the WAXD data for the PLA/PBS/C30BX nanocomposites as previously reported [21]. Nonetheless for detailed analysis of gas barrier properties using Bharadwaj model, we need to reanalyze TEM images. Figure 2 shows TEM images of the PLA/PBS/C30BX nanocomposites with various clay loadings with the images chosen to show clay rich regions. Although the WAXD pattern of the nanocomposite having 1 and 3 wt% of C30BX showed featureless diffraction, which was interpreted as a delaminated morphology, in the TEM image (Figures 2(a) and 2(b)), delaminated/intercalated and stacked structure of the silicate layers could be readily observed for both 1 and 3% clay loadings. More precisely, 1 and 3 wt% of C30BX exhibited a disordered intercalated structure of stacked silicates. Therefore, nanocomposites with 1 and 3 wt% of C30BX are expected to achieve the greater enhancement in barrier properties.  Big, clear, and flocculated silicate layers tactoids can be seen in the clay content 5 wt% (Figure 2(c)) and higher contents of nanocomposites as we previously observed [21]. It can be seen that with the increase of clay concentration, the thickness of the tactoids increased. However, these tactoids themselves gain some degree of disorderliness, which may be due to the clay concentration as well as the shearing forces during the melt extrusion process in the preparation of the nanocomposite. TEM images (not shown) clearly demonstrated that the stacked and intercalated silicate layers were nicely dispersed in the PLA and PBS matrices which were supported by WAXD analysis also. A previous section proved that both polymers were intercalated between the clay layers of C30BX (refer to Figure 1).
Sheet-like platelets (clay particles) are believed to increase the gas barrier properties in nanocomposites. The barrier property improves due to the tortuous path being created by the clay particles which retards the diffusion of the gas molecules through the matrix region. The theory behind the improvement in the gas barrier properties of the polymer layered nanocomposites was explained above.
All the nanocomposite films showed better oxygen barrier properties than neat PLA films as well as PLA/PBS blend films, as shown in Figure 3, which is previously observed [23]. Amongst all the nanocomposites, the nanocomposite having 5 wt% of C30BX content exhibited the highest reduction in oxygen permeability, approximately 26%. XRD and TEM (Figures 2(a) and 2(b)) results indicated that a mixture of exfoliated and disordered intercalated morphology occurred for 1 and 3 wt% of clay nanocomposites. However, improvement in 5 wt% of nanocomposite was the maximum. As the clay content was increased, aggregates of the clay particles formed which gave less improvement in oxygen barrier property. Research has shown that the degree of tortuosity also reflects geometrical influences like shape and state of delamination/intercalation of platelets and their orientation in the polymer matrix [12,[33][34][35][36][37]. For instance, Ray et al. [34] found a significant decrease in oxygen permeability for compression molded PLA nanocomposites samples, while observing some difference between the experimental value and calculated value due to the coexistence of intercalated stacked and disordered/delaminated clay layers. On the other hand, compression-molded specimens are generally known to give lower transmission rates than extruded specimens, while blown films generally show a good improvement in barrier properties [35], probably due to the fact that the biaxial orientation of silicate platelets in the polymer matrix produced during the blown film process would increase the tortuosity enough to enhance the barrier properties [36]. Therefore, the overall improvement in the barrier property was not that high because the specimens used in this study for obtaining oxygen permeability were compression molded. Nonetheless, it is interesting to note that, even though the oxygen permeability did not decrease further at higher clay contents (7 and 10 wt%) in the PLA/PBS/C30BX nanocomposites, the decrease can be considered significant in comparison to the PLA/PBS (80/20) blend due to the presence of impermeable crystalline clay particles restricting the passage of oxygen. For the relatively modest improvement of the barrier properties in their compression molded nanocomposites, Krook et al. [35] explained two reasons of the void-content, which provided means for non-diffusion transport through the film and the nonuniformity of the clay particles. The improvement of the barrier properties of PLA/PBS/C30BX nanocomposites compared to PLA and PBS blend alone prepared by compression molding was probably because of the presence of the clay which served as barriers.
The PLA/PBS/C30BX nanocomposites also exhibited a modest improvement in water vapour barrier properties relative to those of the PLA/PBS blend films. Water vapour transmission rates (WVTR) of all the samples are shown in Figure 4. Nielsen [10] and Gerlowski [37] suggested a general principle for the transportation of water vapour. They stated that water penetrates the galleries between the montmorillonite layers, forming water clusters in the nanocomposites, and does not allow the water to pass through quickly, and thus the diffusivity and the overall transport of water across the films is decreased.
In the PLA/PBS/C30BX nanocomposites, the water vapour permeation rate was not reduced as much as in the case of the oxygen permeation rate due to the hydrophilic nature of the polymers. When the PBS was added to the PLA in the ratio of 80/20 wt% (PLA/PBS), the water vapour transmission rate increased. During the transport of water through the PLA/PBS films, water first absorbed through the one side of the film and once the saturation level was attained, the water started to diffuse through the film to the other side and transportation took place. The water vapour transmission rates decreased the most, up to 18%, when only 5 wt% of C30BX was added in the binary blend. This decrease in water vapour transmission rate should be considered as quite modest, which implies that the C30BX also acted as impermeable layers which decreased the amount of tortuosity.
Note that the polymer crystalline regions in PLA and PBS polymers are considered to be impermeable to small molecules such as gases, which can lead to modifications in the final properties. It has been suggested that the presence of the filler (clay) may nucleate [38] crystallization, which leads to an increase in degree of crystallinity. This can have a direct effect on the transport properties. The significant drop in permeability with an increased crystallinity is usually explained in two ways [39,40]. Firstly, inclusion of impermeable crystallites decreases the amount of amorphous material through which the permeant can diffuse. Secondly, impermeable clay crystallites increase the tortuosities of the transport path. This improvement in the gas barrier properties can be attributed to the achievement of the mixed morphology suggested by XRD and TEM results and a decrease in void density. Gas barrier properties were improved due to the distribution of clay in the polymer matrix. On the other hand, regarding the effect of clay on crystallization, it was observed from our previous study using DSC that while 1 and 3 wt% of clay present in PLA/PBS/clay nanocomposites may not be enough to notice effect on crystallinity, the crystallinity was the highest for 5 wt% of clay content [23].
As discussed earlier, according to the Nielsen model [10], the permeability of the nanocomposite (P s ) is related to the permeability of the neat polymer (P p ) and the volume fraction of the sheets (Φ s ), length (L), and width (W) of the sheets (clay particles), as given in (3), when S is equal to 1. The thickness of the basic structural unit of montmorillonite is well defined and is approximately given as W ∼ 1 nm; however, the length and width show large distribution, ranging from 100 nm to several microns depending upon the concentration of the filler in the polymer. The effective width of the sheets of the nanocomposites can change through aggregation of clay layers. As seen from (2), the volume fractions of the sheets and the sheet length and sheet width are the required parameters for the calculations of relative permeability for nanocomposites.
Importantly, two main reasons made the estimation of sheet length quite complicated. Firstly, on the micrograph images of the nanocomposites, only the projections of the particle transect were seen. Therefore, the real dimensions could not be estimated. Secondly, the features appearing on the micrograph images were not straight; therefore, a computer assisted analysis method was required [41]. Unfortunately, a fully automated image analysis was excluded, because of the poor visibility of the particles on the image. However, a semiautomated image analysis was used to estimate the sheet length of the clay particles as given in Figure 5. The volume fractions of the sheets in the nanocomposites were calculated from the weight fraction using the density (1980 kg/m 3 ) of C30BX.
TEM images of polymers normally consist of different shades of gray colour which primarily indicate electron density contrast through the sample. Regions of high atomic number appear dark, while those of low atomic number appear light. The discrimination between the greyish polymeric regions and the dark lines corresponding to the alumino silicate platelets was relatively easy by eye. In our semiautomated image analysis method, conversion of the original TEM photomicrograph into a black and white image was necessary.   The semi-automated method [7] consisted of, first, manually tracing the dispersed platelets that were parallel to the beam and the agglomerates on a transparent film overlaid on the micrograph image print using a black permanent pen. The isolated platelets were considered as single platelets. The resulting transparent film was then electronically scanned and converted into a greyscale tagged-image film format (TIFF) image file. A micrograph image and its scanned image obtained by manually tracing the particles are shown in Figure 5. This greyscale image can then be converted into a black and white or binary image by using image analysis software (Scion Image Beta 4.02, Scion Corporation, USA). This software easily converts a TIFF file into a binary image and creates a duplicate of the image. From this file, the sheet lengths of the particles were calculated to obtain histograms for the various nanocomposites using Minitab 15 software. Figure 6 shows histograms of the C30BX particle lengths and pertinent statistical data obtained on all the PLA/PBS/C30BX nanocomposites having various clay loadings. Here, the effects of clay concentration can be easily seen as sheet length increased with the addition of C30BX clay. Thus, average sheet lengths along with standard deviations were obtained. The calculated values of tortuosity and relative permeability of all PLA/PBS/C30BX nanocomposites using (2) are presented in Table 1. By comparing the relative permeabilities of oxygen obtained from the experiments and from the model, it can be seen that the experimental data very well validate the the Bharadwaj model for nanocomposites for the lower concentrations of C30BX. The nanocomposite of 1 wt% had the shortest sheet length and with the increase in clay loadings sheet length increased. At low concentrations of clay content, the clay particles were dispersed well as compared to the nanocomposites of higher content of C30BX (discussed earlier in XRD and TEM results). Therefore, lower concentration gave short clay lengths, while at higher concentrations, sheet length started to increase. From TEM images, it was seen that as concentration of clay increased more stacks of clay particles were formed. Therefore, the effective thickness of the clay sheets (tactoids) also increased. Instead of 1 nm, it becomes 10 or 14 nm or more, depending upon the number of clay layers in the intercalated tactoids. However, here S = 1 (planar arrangement of silicate layers to the path of gas molecule to diffuse) was taken as assumptions while doing the calculations using the Bharadwaj model based on the clay being a gas barrier. Due to the stacking of the clay layers, now the width of the clay particles becomes effective width with the value like 6, 10, 14, 21, and 33 nm approximately for the clay loading of 1, 3, 5, 7, and 10 wt%. If these values are used in the Bharadwaj model while calculating the relative permeability, then curves for both the relative permeability of experimental and Bharadwaj model would match as shown in Figure 7. Figure 7 shows that experimental data validated the Bharadwaj model up to a sheet length of approximately 80 nm (for 1 and 3 wt%). After that, at high sheet lengths larger and larger deviations from the Bharadwaj model were observed. The reason behind this may be that certain assumptions, like good dispersion of clay particles in the polymer matrix, S = 1 and W∼1 nm were used in this model. Due to agglomeration/intercalation in PLA/PBS nanocomposites, the thickness of the sheet did not remain as 1 nm. At high concentrations, thickness becomes effective thickness (effective width), which depends on the number of sheets in the stacks. The tendency of the silicate sheets to agglomerate was indeed expected to become more severe as the clay concentration increased. Due to the change in the thickness of the silicate stacks, its aspect ratio (L/2W) also changed and the high enhancement in gas barrier was not noted. Thus, for high content of C30BX particles, the Bharadwaj model was not validated by the experimental data of PLA/PBS/C30BX nanocomposites.
If the actual values of width instead of 1 nm were used in (2), the relative permeability of PLA/PBS/C30BX nanocomposites would validate the Bharadwaj model for all the concentrations of C30BX.

Conclusion
PLA/PBS/Cloisite 30BX clay nanocomposites containing 1, 3, 5, 7, and 10 wt% of clay were prepared by a melt extrusion method, and their morphology was examined using both XRD and TEM. Improvement in oxygen permeability was significantly noticed for PLA/PBS/C30BX nanocomposites with the addition of C30BX; however, the water vapour permeation rate showed only modest improvements in this system. In this study, compression molded samples were used for the determination of gas permeabilities. Due to the stacks of clay and non-uniform dispersion of clay particles, tortuosity path for the gas molecules did not increase sufficiently to make a significant improvement in gas barrier properties. PLA/PBS/C30BX nanocomposites validated Bharadwaj model for only up to 3 wt% of clay content. For higher concentrations of C30BX, deviation from

Introduction
Layer-by-layer (LbL) nanoassembly is considered as one of the most precise, readily scalable, and cost-effective techniques for the thin film deposition of polymers and nanoparticles [1,2]. Cu(InGa)Se 2 (CIGS) nanoparticles have been widely considered in thin film solar cells due to their high theoretical efficiency and tunable band gap. Typically, CIGS solar cells are fabricated using a coevaporation technique requiring a vacuum environment, and therefore resulting in higher cost [3,4]. Manufacturing costs for CIGS-based solar cells must be reduced to make economically viable. Many researchers have focused on lowering fabrication costs through alternate, nonvacuum deposition processes, such as blade coating and spray deposition technics [5,6]. However, these methods do not provide sufficient and accurate film thickness and produce undesirable rough surfaces [7]. LbL processes can be used to fabricate precisely tailored thin film on various types of substrates, both rigid and flexible. In LbL nanoassembly, positive and negatively charged species are sequentially deposited on a substrate via alternate dipping [8]. LbL nanoassembly does not require sophisticated lab facilities or precise process control, thus yields a cost-effective thin film manufacturing process. In this study, LbL nanoassembly is utilized to deposit controlled thin films of oppositely charged CIGS nanoparticles and polymers. Film deposition and electrical characterization results are presented and discussed. Novel aspects of this research include the functionalization of CIGS nanoparticles to create stable aqueous dispersions with desired surface charges, fabrication of thin CIGS film using LbL nanoassembly, and characterization and utilization of the films for solar cell application.

Experimental
2.1. CIGS Nanoparticle Synthesis. CIGS nanoparticles were synthesized using copper chloride (CuCl), indium chloride (InCl 3 ), gallium chloride (GaCl 3 ), and selenium (Se) in oleylamine (OLA) solvent. OLA (70%) was heated at 200 • C for 2 hours under nitrogen. Then, in nitrogen-filled glove box, 10 ml of heated OLA was added to Se and heated at 250 • C for 1 hour while being stirred. At the same time OLA was added into two different flasks containing InCl 3 and CuCl, each heated at 250 • C for 30 min while being stirred. All three flasks were then mixed and GaCl 3 was added. This reaction mixture was heated at 250 • C for 75 min while being stirred [9][10][11]. The resulting nanoparticles were washed through suspension in chloroform and precipitation by ethanol. The size of the synthesized nanoparticles was analyzed using Malvern Nanosizer, size and zeta-potential measurement instrument, and atomic force microcopy (AFM), and material composition was studied through electron dispersive spectroscopy (EDS).
CIGS nanoparticles were cleaned through sonicating in ethanol for several hours and precipitating by centrifuging at a high speed. The resulting nanoparticles were dispersed in water. Zeta-potential (ζ) of the nanoparticles dispersed in water was controlled through pH regulation by slowly adding 0.1 g/ml NaOH solution to the CIGS dispersion. In another process, CIGS nanoparticles were coated with polystyrene sulfonate (PSS) to allow the uniform dispersion in water (CIGS-PSS).

Film Fabrications and Characterization.
LbL nanoassembly is based on the sequential deposition of oppositely charged polyelectrolytes or nanoparticles. LbL films can be deposited on a wide variety of rigid and flexible substrates of different shapes and sizes [8]. LbL nanoassembly of oppositely charged CIGS nanoparticles (CIGS/CIGS), CIGS nanoparticles and PSS (CIGS/PSS), and PSS-coated CIGS nanoparticles and polyethylenimine (CIGS-PSS/PEI) was conducted.
To measure thickness, films were scratched using micromanipulator probe and were analyzed using atomic force microscopy (AFM). Confocal scanning microscopy was used to take the images of the first and last layer of a CIGS-PSS/PEI film composed of 25 bilayers. The bottom layer was lablled using Invitrogen's 20 nm fluorospheres in nile red (red-colored fluorescent dye) and the top layer was lablled using fluorescein isothiocyanate (FITC) (green-colored fluoerescent dye). To study the resisitivity and the photoelectric properties CIGS/CIGS, CIGS/PSS, and CIGS-PSS/PEI LbL films were deposited across a 200 μm × 58 cm channel on indium-tin-oxide-(ITO-) coated glass substrate. The current-voltage measurements of the films were conducted using Keithly Semiconductor Characterization instrument.  The optical absorbance of a CIGS-PSS/PEI film deposited on a glass substrate was measured with ultraviolet-visible-near infrared (UV-Vis-NIR) spectroscopy.

Solar Cell Fabrications and Testing.
A solar cell device consisting of CIGS-PSS/PEI absorbing layer, cadmium sulfide (CdS) n-type layer, intrinsic zinc oxide (ZnO 2 (i)) buffer layer, and aluminum-doped zinc oxide (ZnO 2 (n)) layer was fabricated and tested. A 219 nm thick CIGS-PSS/PEI LbL film was deposited on an ITO-coated glass substrate. A thin cadmium sulfide film was deposited on the CIGS-PSS/PEI through chemical bath deposition. Intrinsic zinc oxide ZnO 2 (i) and ZnO 2 (n) films were deposited by spincoating of the nanopowders dispersed in isopropyl alcohol. The J/V characteristics of the solar cell were measured with the Keithly Semiconductor Characterization instrument.

CIGS Nanoparticles Characterization. The EDS results
showing the material composition of the synthesized CIGS nanoparticles are presented in Figure 1. The gold and silicon peaks are attributed to the gold-coated silicon substrate used.  A scanning electron microscopy (SEM) image of the CIGS nanoparticles is shown in Figure 2. The sample for the image was prepared by drying a drop of CIGS dispersion on a gold-coated silicon substrate. It is observed that while the individual particle sizes are in 10 nm range, it forms an aggregation when dried. Zeta-potential (ζ) of the CIGS nanoparticles dispersed in water was measured to be +61 mV. The pH of the solution was raised by slowly adding 0.1 g/ml NaOH solution to the CIGS dispersion. The change in the ζ of the nanoparticles dispersed in water with increasing pH is shown in Figure 3(a). It is observed that the ζ changes from +61 mV at pH 7 to −51 mV at pH 10.5. As PSS was slowly added to the CIGS dispersed in water at pH 7, the ζ of the nanoparticles slowly decreased from

Scratch
Step height measurement 60 nm +61 mV to −56.9 mV. The size distribution of the CIGS-PSS nanoparticles in water after separating bigger particles by centrifuging the dispersion at 15000 rpm for 15 minutes is shown in Figure 3(b). The size distribution of the particle is observed to spread from 5-10 nm. with PSS shows a better adsorption to the surface than bare particles. It is attributed to a better attachment between the two polymers PSS and PEI, compared to the attachments between CIGS and PSS or CIGS and CIGS. Moreover, the comparable thicknesses of CIGS/CIGS and CIGS/PSS films further show that the attachment of CIGS to PSS is better than CIGS to CIGS. AFM image showing the top surface of a CIGS-PSS/PEI film composed of 6 bilayers is presented in Figure 5. A scratch made using micromanipulator probe is also seen in the figure. The depth of the scratch was measured to be 60 nm using step height measurement tool within AFM. The depth of the scratch measured is smaller than the film thickness given by the QCM result. It shows that the scratch does not completely penetrate through the film. SEM images of a 6 bilayer CIGS-PSS/PEI film are shown in Figure 6. Figure 6(a) shows the top surface of the film illustrating a scratch made by the micromanipulator probe. Figure 6  The optical absorbance of a CIGS-PSS/PEI film deposited on a glass substrate is shown in Figure 8. The results show that the absorbance spectrum edge of the film is near 920 nm wavelength (1.3 eV band edge). The inset of Figure 8 shows the current-voltage (J/V) characteristic of a 6-bilayer CIGS-PSS/PEI film deposited on a 200 μm × 58 cm channel, in the dark and when illuminating with a 100 W light source. It is observed that the current density increases by three folds when illuminated by the light source. However, the polymers employed in the LbL deposition process might be susceptible to photodegradation effects, which will be studied in the future, and may require further material design.

Solar Cell
Results. The measured J/V characteristic of the solar cell with 1.0 cm 2 active area, when illuminated by a 50 W light-source, is shown in Figure 9. The considered solar cell structure is shown in the inset of the figure. The open circuit voltage (V), short circuit current density (J), and power density under the illumination were measured to be 0.7 V, 0.3 mA/cm 2 , and 0.168 mW/cm 2 , respectively. The efficiency of the presented solar cell was measured to be 3.5%, which is comparable to the recently reported devices by emerging fabrication techniques such as spray-deposition, but lower than the devices fabricated through coevaporation of the CIGS materials. The LbL assembled CIGS films contain defects that increase carrier recombination. These films are also loosely packed and include polymers that contribute to lower the carrier mobility and thus affect the efficiency of the solar cell. However, it should also be noted that LbL assembly is simple, cost-effective, and highly scalable.

Conclusions
Fabrication of thin CIGS films using cost-effective, nonvacuum, and highly scalable LbL nanoassembly process has been presented. CIGS nanoparticles were synthesized, dispersed in water, and desired zeta-potential were obtained through pH regulation (from +61 mV at pH 7 to −51 mV at pH 10.5) and coating with PSS (−56.9 mV). The LbL film deposition process of the oppositely charged CIGS nanoparticles (CIGS/CIGS), CIGS nanoparticles and PSS (CIGS/PSS), and PSS-coated CIGS nanoparticles and polyethylenimine (CIGS-PSS/PEI) was studied. The J/V and photoelectric characterization results of the films have been presented and discussed. Prototype solar cell devices using the developed films have been fabricated and tested. LbL deposited CIGS films are applicable in cost-effective solar manufacturing.

Introduction
The use of nanoparticles (NPs) in biomedical field is one of the most important branches of nanobiotechnology. Many biomedical applications require nanoparticles with low toxicity, biocompatibility, and stability with high preferential accumulation in the target tissue or organ [1]. Polymernanoparticle composites may possess the unique property superior to that of either the polymer or nanoparticles. Metal nanoparticles with polymer coatings or dispersed in polymeric matrices display the increased stability, improved processability, recyclability, and solubility in a variety of solvents [2].
Thermal therapy, also termed as hyperthermia, has thousands years history as one of most important methods for the tumor treatment [9]. In this technique, moderate heating is generated in the target region, resulting in tumors selectively destroyed, due to their reduced heat tolerance compared to normal tissue [10]. In the past decades, because many new heating sources such as radiofrequency [11], microwave [12], magnetic field [13], and laser [9] were employed, there were remarkable progresses achieved in this area. Photothermal therapy (PT) is a promising field within thermal therapy, incorporating laser techniques and photoabsorbing agents. Light absorbing dyes, chemical photosensitizers, and solid optically responsive nanomaterials have been applied as the photoabsorbing agents in PT [9]. For small metal nanoparticles, their optical properties are dominated by collective oscillation of conduction electrons. An absorption band results when the incident photon frequency is resonant with the collective oscillation of the conduction band electrons, and this is known as the surface plasmon resonance (SPR) [14]. Au NPs possess strongly enhanced visible and near infrared light absorption, which is several orders of magnitude more intense compared to conventional chromophores. Various gold-based nanostructures with NIR 2 Journal of Nanomaterials absorption have been developed. Gold nanospheres [15], gold nanoshells [7], gold nanorods [16], and gold nanocages [17] are the popular nanostructures that have been demonstrated in photothermal therapeutics due to their strongly enhanced absorption in the visible and NIR regions on account of their SPR effects.
In our work, Au NPs with NIR absorption were prepared by the one step reaction of chloroauric acid and sodium thiosulfate. Their NIR absorption wavelengths were well controlled by adjusting molar ratio of HAuCl 4 /Na 2 S 2 O 3 [18]. These Au NPs were coated with chitosan-(CS)based coatings and used as a therapeutic agent for the photothermal ablation of cancer cells. Chitosan is a biobased polysaccharide derived from shrimp and crab shells [19]. It is nontoxic, biodegradable and biocompatible, and has good processability [20,21]. The cost associated with chitosan is much lower than that of other polymer coatings or chemical compounds, for example, polyethylene glycol (PEG) that are often used to protect and stabilize the nanoparticles by many researchers [5]. Hepatocellular carcinoma is the most common primary malignant tumor of the liver. It affects greater than half a million patients worldwide. US liverrelated cancer deaths account for 4% of all cancers or about 20,000 deaths annually. Two anchorage-dependent cell lines, hepatocellular carcinoma cell line (HepG2), and control cell line of human dermal fibroblast (HDF) were selected for the estimation of laser ablation efficiency of these novel gold nanocomposites. The efficacy was evaluated by live/dead stain images following photothermal ablation. These gold nanocomposites show great ability for laser ablation, and the surface coating components plays a very important role for selectivity to certain cancer cells. In addition, this composite material offers new and desirable advantages for cancer treatment. Beside the optically tunable gold nanoparticle for efficient conversion of light to heat, chitosan-based coatings provide tunable surface charges for applications in a variety of physiological pH environments and also allow further modification with targeting molecules or drug molecules by chemical reaction or electrostatic interaction.

Preparation of Au Nanoparticles.
Au NPs were prepared by the method we reported before [18,22]. In which, 3 mM sodium thiosulfate (Na 2 S 2 O 3 ·5H 2 O, Aldrich, 99.999%) solution was added into 1.71 mM chloroauric acid (HAuCl 4 ·3H 2 O, Alfa Aesar, Au 49.68%) solutions with desired volume ratio at room temperature and vortexed for 20 seconds for uniform mixing. All the water used in the experiments was purified by a Thermo Scientific Easypure II system, with a resistivity of 18.2 MΩ cm.

Preparation of Carboxymethylated Chitosan.
Low molecule weight chitosan (Sigma-Aldrich, MW 50-190 K) was used for the surface coating of the Au NPs. Based on this chitosan, O-carboxymethyl chitosan (CMCS) was synthesized as the reported method [23,24]. Briefly, 15 g sodium hydroxide was dissolved in a 20 : 80 mixture of deionized water and isopropanol (100 mL) in a 500 mL flask. 10 g chitosan was added to the flask to swell and alkalize at 50 • C for 1 h. 15 g monochloroacetic acid (ClCH 2 CO 2 H, 99%, Acros Organics) was dissolved in 20 mL isopropanol and added into the reaction mixture dropwisely in 30 min and reacted under vigorous agitation for 4 h at 55-60 • C. Adding 200 mL 80% ethyl alcohol stopped the reaction. The product was filtered and repeatedly rinsed by 80% ethyl alcohol, to desalt and dewater, until the pH value of the washing solution was less than 8.0. Finally, the product was vacuum dried for 1 day at 40 • C.

Preparation and Purification of Au NPs with Chitosan
Based Coatings. Stock solution of chitosan was prepared by dispersing chitosan (1 g) in 0.7% acetic acid solution (100 mL). CMCS solution was prepared by directly dissolving CMCS (1 g) in 100 mL DI water. CS and CMCS solutions were purified by centrifuge at 1000 rpm for 10 min, to remove the little amount of insoluble residuals, and then were dialyzed (Spectrum Dialysis Membranes, MWCO 3 KDa) for 2 days. The chitosan-coated Au NPs (Au/CS), CMCS coated Au NPs (Au/CMCS), and CS and CMCS blended coated Au NPs (Au/(CS+CMCS)) were prepared by, respectively, adding CS solution, CMCS solution, or CS and CMCS blending solution (CS and CMCS solutions were premixed) dropwisely into the as-synthesized gold particle suspensions, with vigorous agitation. The ratio between CS or CMCS to Au NPs was controlled as 0.2 mg CS or CMCS per OD Au NP suspension and between CS and CMCS blending to Au NPs is (0.15 mg CS + 0.05 mg CMCS) per OD Au NP suspension (OD is the optical density). Then these samples were set on the rocking bed for 1 day. In the end the gold nanocomposite suspensions were concentrated and separated from the solution by centrifugation at 1000 g for 20 min. This step will remove the byproducts in solution and the extra chitosan. Then the centrifuged pellets were dispersed in sterilized DI water to a concentration of 25 OD/mL and stored at 4 • C for the subsequent studies.

Characterization of Au NPs.
The optical absorption of Au NPs was measured by a UV-Visible spectrophotometer (Cary-50Bio, Varian). A Zetasizer (Nano-ZS90, Malvern) was used to assess the surface charge of the nanoparticle suspensions. A FEI Tecnai F30 transmission electron microscope (TEM) operated at 200 KV was used to determine the morphology of Au NPs. Several Au NPs suspensions that have different NIR peak wavelength were carefully prepared for the photo-induced temperature rising measurement. 1 mL of the Au NP suspensions were added into the wells of tissue culture plate (Multiwell, Primaria 24 Well), with the concentration of 1 OD/mL. A laser with 817 nm centered wavelength generated by FAT-System (Coherent, CA) was shone on the gold suspension. The laser spot has the same area with the well, and the power of the laser was fixed as 1 W/cm 2 . An infrared thermometer (CSmicro, Optris GmbH, Germany) detecting the temperature of the Au NP suspension was fixed 10 cm above the media. line (HDF, ATCC: PCS-201-012 male source) and hepatocellular carcinomas cell line (HepG2, ATCC: 59194) were seeded at monolayer confluence to allow for easier cell counting following staining (within 15 passages). All the cells were cultured at 37 • C in a 5% CO 2 environment in Dulbecco's modified eagle medium (DMEM) supplemented with 4 mM l-glutamine, 1% penicillin, and 10% fetal bovine serum. Prior to laser ablation experiments, cells were detached from culture flash with trypsin (0.25%) and resuspended in DMEM media and injected into the 24-well tissue culture plate (Multiwell, Primaria), cultured for 36-48 hours to achieve confluence. Laser ablation was performed based on methods described in [7]. Cells were incubated with 1 mL DMEM media which contains 1 OD Au NPs for 2 h to obtain the binding of the Au NPs on the cell. For this experiment, Au NPs and gold/chitosan nanocomposites were prepared with NIR absorption in the range of 820 nm to 850 nm, close to the wavelength of the laser source. Media were removed, and cells were rinsed with 0.5 mL PBS buffer, followed by addition of 0.5 mL fresh media to the culture wells. Next, the cells were exposed to 817 nm NIR laser at optimized conditions (5 W/cm 2 , 2 min). Well plates were incubated overnight after photothermal treatment, and a live/dead stain (Invitrogen, Carlsbad, CA) was performed, and fluorescent images were obtained and analyzed. Images were taken with a Nikon D90 camera on an Accu-Scope 3032 Microscope for green/red fluorescence, respectively, and cell numbers were counted manually.

Preparation of Gold Nanoparticles with Desired Near
Infrared Absorption. In our previous work [18,25], a convenient method was developed to synthesize Au NPs with controllable NIR absorption. Au NPs were synthesized by one-step reaction of chloroauric acid and sodium thiosulfate, without assistance of additional templates, capping reagents, or seeds assembly. The synthesized products consist of nanoparticles with different shape and size including small spherical colloid gold particles (<5 nm) and nonspherical gold crystals. The nonspherical gold crystals are mainly the truncated octahedron, pentagons, and cuboctahedron, as well as the triangular-shaped plate structures [18]. Two absorption peaks due to the surface plasmon resonance are observed from these samples. The first SPR peak centered at around 530 nm is the characteristic SPR of the spherical colloid gold particles, and the second SPR component is higher at NIR band, which is attributed to the multipolar SPR band from the nonspherical Au NPs [18,25]. Since the NIR absorption wavelengths increase with increasing molar ratio of HAuCl 4 /Na 2 S 2 O 3 , adjusting the molar ratio of HAuCl 4 /Na 2 S 2 O 3 during the reaction controlled the absorption wavelength of the Au NPs. These products were further purified by the centrifugation to remove the spherical colloid gold particles, improving the NIR absorption. Figures  1(a) and 1(b) are the Z-contrast STEM images of the assynthesized Au NPs and the same batch sample purified with centrifugation. The purified product possesses the NIR absorption wavelength at 820 nm. The temperature of the NIR absorbing Au NP suspension induced by the NIR laser radiation is affected by the concentration of nanoparticles, laser power, exposure time, and the absorbing wavelength of the Au NPs [26]. Figure 2 compares the heating effect of the Au NPs suspensions with NIR peak at 820 nm, 880 nm, 980 nm, and 1070 nm, respectively (showing in the insert figure) under the 817 nm wavelength diode laser (1 W/cm 2 ) exposures. After 3 minutes, the temperature reaches to 50.8, 48.2, 46.4, and 41.2 • C for the particles from low wavelength to high wavelength, respectively. The gold sample that has the NIR peak more close to the wavelength of laser source shows the best heating effect. The effective control of the NIR absorption of the gold nanoparticles for matching the wavelength of the laser source enhances the hyperthermia treatment. Meanwhile, the NIR wavelength is considered as the best light source for the photothermal therapy, because it is minimally absorbed by the normal tissues components of water and hemoglobin and can transmit deeply in the body.

Chitosan-Based Coatings on the Gold Nanoparticles.
Chitosan is a cationic polysaccharide of D-glucosamine and N-acetyl-D-glucosamine and has been shown to be a highly biocompatible material and have antibiotic properties. To improve its solubility or achieve new functionality, chemical modifications of chitosan were conducted and many derivatives have been synthesized. For example, modifications can be synthesized by carboxylation of the hydroxyl or amine groups in chitosan structure [20], resulting in the O-carboxymethyl chitosan or N-carboxymethyl chitosan respectively. O-carboxymethyl chitosan is achieved by substituting parts of the -OH at C-6 position of glucosamine units with -CH 2 COOH. Therefore, the reactive ligands such as -COOH and -NH 2 groups are amenable to chemical functions. Others have reported the stabilization of GNPs with chitosan. As chitosan in solution is protonated and Journal of Nanomaterials  Table 1 lists the hydrodynamic diameter and the surface charge of the gold nanoparticles with different chitosanbased coatings in water. The chitosan-based coating significantly changed the surface property of the gold nanoparticles. The average particle size of the bare gold nanoparticles is 36 nm; after being coated with polymer, the hydrodynamic diameter of Au/CS, Au/CMCS, and Au/(CS+CMCS) reaches 290, 236, and 262 nm, respectively. Meanwhile, their surface charge changes from −43.5 mV to +43.8, +20.2 and +24.3 mV, respectively. Figure 4 shows the optical stability of the Au NPs with different coatings in PBS and 1 wt% NaCl solution. Since the physiologic salinity is near 1%, optical properties of Au NPs were simulated in these solutions. The optical stabilities were measured by monitoring the optical absorbance of the Au NPs at their NIR peak wavelength as function of time. These nanoparticles have similar behavior in the PBS and NaCl solution. CMCS-coated particles show the highest stability, and there is no obvious decrease in its optical absorption in 3h. The CS and CMCS blended coated particles also show good stability, with a decrease in its absorbance less than 10%. The most unstable particles are the bare Au NPs. Its absorbance decreases more than 70% in saline environment. The Au/CS nanocomposites are more stable than the bare particles, but less stable than the Au NPs with CMCS component in the coating. The decrease of NIR absorption of the bare Au NPs results from interparticle coupling effect [30,31], in which its stabilized surface layer was disturbed due to the high ion strength, causing the particle aggregation (irreversible coalescence). This behavior may greatly decrease the efficiency of photothermal ablation. Chitosan-based coatings greatly improve the optical stability of the Au NPs. The instability of the Au/CS NPs mainly results from the sedimentation of particles caused by the agglomeration (reversible coalescence) of chitosan coating. At the pH value above 6, the solubility of chitosan becomes worse. CMCS has the best solubility in all pH ranges except its isoelectrical point [23]. The good solubility and the  increased chain flexibility [32] of CMCS may prevent the coupling effect and the sedimentation of the particles. By adjusting the surface components, the surface charge of the Au NPs can be modified. Figure 5 shows the zeta potential of Au NPs with different chitosan coatings as function of the pH value. The isoelectric points (IEPs) of Au/CMCS, Au/(CS+CMCS), and Au/CS nanocomposites are 6.1, 7.1, and 7.7, respectively. IEP of CMCS is close to value (5.5) reported by Zhao et al. [33]. CMCS presents amine and carboxylic acid groups in its structure. In solution, the groups turn to the ion state, −NH 3 + and -COO − , respectively, which improves the solubility of CMCS. Only at its IEP, the number of −NH 3 + is equal to that of -COO − , resulting in the deceasing of solubility. Au/CS particle shows much higher IEP value than that of the Au/CMCS particles, since the CS only presents the positively charged groups (−NH 3 + ). The IEP of the Au/(CS+CMCS) particle is 7.1, between the those of Au/CMCS and Au/CS. This shows that the surface charge of Au NPs can be tuned by the chitosan components on their surface, since CMCS possesses more negatively charged characteristic than CS. Figures 6 and  7 show the live/dead stain images of the HDF and HepG2 cells after being exposed to the 817 nm NIR laser at 5 W/cm 2 , 2 min. The live/dead stain provides a two-color fluorescence assay of cell viability based on plasma membrane permeability, for reliably and quantitatively distinguishing live and dead cells. The live cells show the green color and dead cells show the red color. The statistical cell death ratio of HDF cells and HepG2 cells counted from the live/dead stain images are shown in Figure 8.

Photothermal Ablation of the HepG2 Cells.
Without incubation with Au NPs, cells do not exhibit obvious ablation after the laser irradiation at 5 W/cm 2 , 2 min, as shown in Figures 6(b) and 7(b). The bare Au NPs do not show significant photo-destruction on either cell lines (Figures 6(c) and 7(c)), due to the interparticle coupling effect [31]. The relatively low photothermal ablation efficiency of Au/CMCS NPs is probably due to their less adsorption on the cell membrane (Figures 6(e) and 7(e)). The highest efficiency of the photothermal ablation is observed from the cells treated with Au/CS NPs. There are almost 100% HepG2 cells, and 90% HDF cells were killed after the laser treatment (Figures 6(d) and 7(d)). The CS and CMCS blended coated gold NPs present the promising result: around 96% HepG2 cancer cells were ablated and less than 30% HDF normal cells died (Figures 6(f) and 7(f)). The differences of the Au NPs in the photothermal ablation result from the density of the Au NPs adsorbed on the cell surface. The negatively charged cell surface allows Au/CS NPs to highly accumulate on the cell surface, and high density particles generate extraordinary heat, damaging the HepG2 cells as well as the HDF cells under the laser radiation. Au/(CS+CMCS) particles bind to the cell surface uniformly and generate suitable heat, which was high enough to kill the cancer cells but not enough to damage all the normal cells. This consists with laser ablation results by using gold nanorod [16]. It was reported that after exposure to continuous red laser, malignant cells require about half the laser energy to be photothermally destroyed than the nonmalignant cells [16].

Conclusions
Gold/chitosan nanocomposites were synthesized for the photothermal therapy applications. Au NPs were prepared by the reaction of chloroauric acid with sodium thiosulfate, and they have specific NIR absorption matching the wavelength of laser source. These Au NPs were coated with chitosan or O-carboxymethyl chitosan as well as a blending. These chitosan coatings protect the gold nanoparticles, avoiding irreversible coalescence of the bare Au NPs and improving   showing higher selectivity between the cancer cells and normal cells during the photothermal ablation. These novel gold/polymer nanocomposites have great potential for photothermal therapy, due to low-dosage particle injection, lowpower NIR laser radiation, and quick ablation. More broad clinical relevance can be envisioned, such as the breast cancer treatment, skin cancer treatment, esophageal cancer treatment, and laryngocarcinoma cancer treatment.

Introduction
Fast-activated anionic polymerization has been successfully used for the polymerization of lactams. The intended highpolymerization rate of the anionic polymerization is achieved by introducing an activator, which is a compound containing an N-acyllactam structure. In this sense, sodium hydride in caprolactam [1] and sodium caprolactamate (CLNa) [2] have been used as initiators for the initiation stage.
The technology of the in situ polymerization with sodium montmorillonite clay (NaMMT) was first established at the Toyota laboratories via the hydrolytic polymerization of ε-caprolactam intercalated in a complex of 12-aminododecanoic acid-NaMMT, in order to obtain a polyamide 6/clay nanocomposite (PA6/Clay) [3,4]. Combination of the in situ polymerization techniques with the anionic polimerization of lactams has been reported, where polyamide nanocomposites are readily obtained [5,6]. Jung et al. [7] recently reported the obtaining of PA6/Clay nanocomposites via the in situ anionic polymerization of caprolactam (CL) with clay, using CLNa as initiator and N-acetyl-caprolactam (ACL) as activator. While studying the synthesis of PA6/clay nanocomposites, Liu et al. [8] reported that when using NaMMT, they obtained exfoliated nanocomposites whereas when using organomodified montmorillonite they obtained intercalated nanocomposites. Ahmed and Lee [9] studied the effect of clay concentration on the mechanical and viscoelastic properties of PA6/clay nanocomposites synthesized via anionic polymerization.
On the other hand, studies on the synthesis of composites based on the in situ anionic copolymerization of CL and laurolactam (LL) with NaMMT have not received so much attention. In this sence, Ricco et al. [10] studied the copolymerization of CL with LL, using CLNa and sodium laurolactamate or a mix of both, as initiators, and found that the greatest conversion resulted when using CLNa. Budín et al. [11] studied the anionic copolymerization of CL with LL using either, CLNa or caprolactam magnesium bromide (CLMgBr) as initiators, and found that when using CLNa, there was a tendency to produce random copolymers, whereas when using CLMgBr, the tendency was to produce block copolymers. Although, the structure type of the copolymer was also dependant on the temperature of reaction and on the monomers molar ratio.
The purpose in this work, therefore, was to study the effect of the initiator type on the conversion degree and molecular weight of the final polymer, as well as the effect of the copolymer structure on the clay intercalation-exfoliation.

Copolymerization.
The in situ anionic polymerization was carried out in test tubes, where CL, LL, or a mix of both were introduced and heated to 180 • C. Thereafter, 0.5 mol% of CLNa or CLMgBr was added as initiator, then, 2 wt% of NaMMT was added, and finally, 0.5 mol% of N-acetyl caprolactam was added as activator, all percentages with respect to the total mols of CL + LL. The reaction was allowed to proceed for 30 min in a nitrogen atmosphere. The copolymerizations of CL with LL were carried out using 100/0, 80/20, 60/40, 40/60, 20/80, and 0/100 mol% of CL/LL, respectively. Nanocomposites obtained were then grounded, purified through Soxhlet extraction with methanol, and vacuum-dried at 110 • C for 8 hours.

Characterization.
Conversion was determined gravimetrically by Soxhlet extraction with methanol of the material from the test tubes, where the reaction took place, for 24 hours, using a 2 : 1 methanol-water mixture as the solvent, in order to extract the unreacted monomers and initiators. Thereafter, the remaining material in the Soxhlet tube was dried at 110 • C until constant weight. The conversion was taken as the final weight in the Soxhlet divided by the initial weight and multiplied by 100.
An Alliance 2695 Gel Permeation Chromatograph, with a UV detector model 2487 (248 nm) and styragel columns, was used to determine the evolution of molecular weight of all samples. Samples were first dissolved in trifluoroacetic acid and run in the GPC at a flow rate of 1 mL/min. X-ray diffraction (XRD) analyses were carried out in a Siemens D5000 using an anode of CuKα X-ray of 1.54Å at 35 Kv and 25 mA. Samples were scanned at diffraction angles from 2 to 12 • in 2(θ). Runs were performed with a step size of 0.02 • and step time of 1.0 s. Bragg's law (nλ = 2d sinθ) was used to convert the data of diffraction angle (2θ) to distance (d). Samples for X-ray analysis were obtained from the compression molded process.
Differential Scanning Calorimetry (DSC) analysis was performed on a TA Instruments 2920 Modulated DSC to establish the thermal transitions and structural characteristics of all samples. Samples of ca 10 mg were analyzed under a nitrogen atmosphere. Thermal analysis was carried out as follows: samples were first heated up to 250 • C at a heating rate of 10 • C/min and then cooled down to 0 • C at −10 • C/min. Finally, samples were heated again up to 250 • C at 10 • C/min.
The clay structural organization was studied through STEM observations using a Jeol-JSM7104F electron microscope equipped with STEM modulus and a field emission gun, at an accelerating voltage of 30 kV, using a LEI (low electron image) detector. Ultrathin sections for STEM analysis were previously cut from compression molded samples with a diamond knife at −30 • C using a Leica cryogenic ultramicrotome. Figure 1 shows the conversion and molecular weight of the in situ anionic homo-and copolymerizations of CL, LL, and CL with LL, respectively. The NaMMT content was kept at 2 wt%. All polymerizations were carried out for 30 min, using either CLNa or CLMgBr as initiators.

Physicochemical Properties.
It can be observed in Figure 1(a), that the highest conversion was achieved during the homopolymerization of CL for obtaining polyamide 6/NaMMT (PA6/NaMMT), independently of the initiator used. Also, it was observed that conversion decreased as the LL/CL ratio increased in the nanocomposites of poly (caprolactam-co-laurolactam)/NaMMT (PA6 : 12/NaMMT), but this seems to be dependent on the initiator used, since conversion was higher when using CLNa as initiator at all LL/CL ratios, except above ca. 90/10. This diminished catalytic activity of CLMgBr, as compared to CLNa, is attributed to the steric hindrance of the larger CLMgBr molecular size. Figure 1(b) shows that in all cases, the molecular weight of the resulting homo-or copolymer was greater when using CLNa as initiator. Additionally, with any of the two initiators, the molecular weight of the homopolymers was greater than that of the copolymers, showing a minimum when the LL content was ca. 60 mol%.
As expected, upon increasing the LL concentration, the molecular weight decreases. This is attributed to the much lower reactivity of the LL anions, as compared to those of CL [12]. These results coincide with those reported by Budín et al. [11], while studying the polymerization of CL or LL, without clay.
In all cases, CLNa produced a greater conversion and also a greater molecular weight, which coincides with Odian [13], who mentions that the ring opening anionic polymerization   proceeds in a way that the attained molecular weight varies linearly with conversion.
In addition, the greater molecular weights, when using CLNa, were indirectly assumed, as it was observed that during the copolymerization (CL/LL), the magnetic stirrer ceased turning after just 20 s of reaction when using CLNa, due to the viscosity increase, as a result of a greater molecular weight; whereas it took 35 s when using CLMgBr. Figure 2 presents the effect of composition of the nanostructured polymeric compounds on the fusion temperature T m . First, from Figure 2(a), it can be observed that the T m of the homopolymer compounds is not altered significantly, by the inclusion of the NaMMT, it remains at 221 • C (±1 • C) for PA6/NaMMT and 179 • C for Polyamide 12/NaMMT (PA12/NaMMT), in agreement with that reported elsewhere [14]. Second, from Figure 2  the T m of the copolymer PA6 : 12/NaMMT compounds, on the other hand, decreased as the concentration of LL increased, reaching a minimum (135 • C) at a LL content of 60 mol%. It is also observed that the fusion temperature of the copolymers is below that of the two homopolymers. This abatement is attributed to alpha and gamma crystalline phases of Polyamide 6 (PA6) and Polyamide 12 (PA12), respectively, trying to coexist in the copolymers, as reported by Budín et al. [11]. As the LL concentration increases, the alpha crystalline phase of PA6 tends to coexist with the gamma crystalline phase of PA12, generating many crystallite imperfections. On the other hand, upon increasing the LL content in the copolymers, the crystal size and the crystalline content decrease, decreasing the fusion temperature, as reported by Kubota and Nowell [15].

Thermal Properties.
Finally, it can also be observed in Figure 2(b) that the crystallinity (associated with the fusion enthalpy ΔH m ) of the copolymer compounds is less than that of the homopolymers and it decreases with the concentration of LL, as can be determined from the enthalpies of fusion (Table 1). Figure 3 shows the thermograms of the copolymers obtained using a CL/LL composition equal to 40/60. An important difference observed in the copolymers, when comparing that obtained with CLNa versus that obtained with CLMgBr initiator, is that, when using CLNa, the copolymer shows just one fusion peak, at 144.5 • C. This is indicative of a random copolymer. When using CLMgBr, on the other hand, the copolymer shows two fusion peaks, at 154.2 and 175.2 • C. This, on the other hand, is indicative of a block copolymer. These present a fair agreement with results reported by Budín et al. [11].
With respect to the fusion temperatures of the synthesized block copolymers, when using CLMgBr, it can be assumed that one block is a random block of CL (PA6) and LL (PA12) units. This randomness would hinder the formation of regular-perfect crystals, resulting in a diminished crystalline fusion temperature. The other block, on  the other hand, would be composed mostly of CL units, though not sufficiently long as to show the T m of the PA6 homopolymer, but sufficiently long as to permit the formation of more regular-perfect crystals, with higher crystalline fusion temperature.
In addition, the insertion of the long aliphatic chains of PA12 into the structure of the PA6 in the growing PA6/PA12 copolymer chains, would give rise to a change in the crystalline type, from alpha, typical of PA6, to gamma, typical of PA12. These changes in the crystalline structure of PA would also tend to abate the fusion temperature, as has been reported elsewhere [16]. Table 1 presents the fusion temperature (T m ) and the enthalpy of fusion (ΔH m ) for all the nanostructured polymer compounds studied, using  either CLNa or CLMgBr as initiators. It can be observed that the fusion temperature and the enthalpy of fusion of the copolymers are clearly affected by the increasing concentration of LL, as reported by Ricco et al. [10].
As an outcome, it can be said that the initiator type is the one that determines if the result is going to be a random or a block copolymer. In addition, the effect of LL content upon the fusion temperature of the copolymer is the same, no matter if the initiator is CLNa or CLMgBr. Finally, the addition of NaMMT has no apparent effect on the homopolymers fusion temperature (PA6 221 • C; PA12 179 • C) as shown in Figure 2(a).

Exfoliation and Dispersion of the Nanoclay.
For the case of the polyamide/clay nanocompounds obtained using CLNa as initiator, Figure 4 shows that all the nanocompounds present a progressive shift of the X-Ray diffraction peak to lower 2θ, with respect to that of the pure NaMMT. This indicates an increase in the gallery spacing, producing an intercalated structure, attributed primarily to the insertion of the CL and LL monomeric units into the clay galleries prior to the in situ polymerization. It is important to bear in mind that when using CLNa as initiator, there is a tendency to produce random type copolymers.
In Figure 4, it can also be observed that the diffraction peak of the PA6/NaMMT nanocompound occurs at the lowest 2θ angle, which would correspond to the largest intergallery spacing (17.3Å). On the contrary, the peak of the PA12/NaMMT nanocompound occurs at the highest  2θ angle, corresponding to the smallest intergallery spacing (13.1Å).
In all the PA6 : 12/NaMMT nanocompounds, the peak occurs at higher 2θ angles as the LL content increases. That is, the intergallery spacing diminishes with the concentration of LL. These differences can be attributed to the lower polarity of PA12, compared to that of PA6, due to the longer aliphatic, nonpolar segments (12 C atoms) along its chains, as compared to the shorter aliphatic segments (6 C atoms) in the PA6 chains. That is, for a given polymer chain length, the PA12 shows a smaller content of the polar amide groups than the PA6. This diminished polarity of the PA12 chains makes them less compatible to the highly polar NaMMT. Thus, the decreasing intergallery spacing with increasing LL content, up to the smallest spacing for the PA12/NaMMT nanocompound.
For the case of the polyamide/clay nanocompounds obtained using CLMgBr as initiator, Figure 5 shows that all the nanocompounds present the same shift of the X-Ray diffraction peak to lower 2θ, which would correspond to an intergallery spacing of ca. 16.5Å, with the exception of the PA12/NaMMT nanocompound, which presents an intergallery spacing of 13Å.
When using CLMgBr as initiators, there is no effect of the LL content on the 2θ angle of the X-ray diffraction peak, as observed in Figure 4. This no-effect is assumed to be due to the block copolymer structure formed when using CLMgBr as initiator, in which case, the CL short blocks act as the highly polar PA6 homopolymer does and shows the capability to intercalate into the clay galleries, as the highly polar PA6 homopolymer does. The micrographs in Figures 6(a), 6(b), and 6(c) show similar levels of exfoliation/intercalation, but not that in Figure 6(d), which corresponds to the PA12/NaMMT homopolymer compound, where very little or no exfoliation/intercalation is obtained. These results agree with those obtained via the XRD analysis, where the PA6/NaMMT as well as all the PA6 : 12/NaMMT copolymers prepared in the presence of CLMgBr initiator, present the same intergallery spacing of ca. 16.5Å, with the exception of the PA12/NaMMT homopolymer.
The incomplete clay exfoliation, even though the high affinity between the highly polar PA6 and NaMMT, can be attributed to the lack of application of any shear stresses during the preparation of the nanostructured PA compounds. It might be expected that the preparation of the same PA compounds, via reactive extrusion, could render a much higher level of clay exfoliation.

Conclusions
Homo-and copolyamide type nanocompounds were obtained, via in situ anionic polymerization of caprolactam and laurolactam with NaMMT, using two initiator types.
The use of CLNa as initiator tended to produce a random-type copolymer, whereas the use of CLMgBr tended to produce a block-type structure, which affected both the crystallinity of the copolymers and the intercalation/exfoliation of the nanoclay. In the random type copolymers, the intergallery spacing decreased with the LL content, whereas in the block type copolymer, the intergallery spacing remained the same (as that for PA6) for all the copolymers. The intergallery spacing of the PA12/NaMMT remained the lowest in all cases. The use of CLNa as initiator produced a higher conversion degree and higher molecular weight than the use of CLMgBr.
In all cases, increasing the LL content produced a decrease in the conversion degree and in the molecular weight of the resulting polymer compound.
Crystallinity of the copolymer compounds was found to be less than that of the homopolymers. Also, crystallinity was greatly affected by the LL concentration; as the LL concentration increased, crystallinity decreased, as observed from the enthalpies of fusion.

Introduction
Over the last two decades, blends of thermoplastic polymers with liquid crystalline polymer (LCP) received considerable interests from both academic and industrial researchers. The dispersed LCP domains tend to elongate during melt processing and subsequently form aligned fibrils capable of reinforcing the matrix polymer. The aligned fibril formation in these in situ composites greatly depended on the choice of polymer matrix and the selection of proper processing conditions. The droplets, short fibrils, networks of droplets and fibrils, and continuous fibrous LCP structures in thermoplastic polymers have been reported [1][2][3][4][5][6][7]. The research on in situ composites mainly dealt with unique morphology and consequent enhancement of mechanical properties, while those addressing the electrical properties are limited [8][9][10]. In this context, introduction of carbon nanotubes (CNTs) in thermoplastic polymer/LCP blends can offer additional improvements in mechanical properties in addition to offering excellent electrical and thermal conductivity. However, high cost and limited availability of CNTs often serve as deterrent to development of more widespread applications of composites containing CNTs. In this study, CNTs were introduced in blends of polycarbonate (PC) and LCP with the aim of obtaining synergistic effects, such as strong mechanical properties from fibrillar LCP domains and electrical conductivity from CNTs. The interplay of CNTs with the morphology of PC/LCP blends or the possibility of selective localization of CNTs in PC or LCP was not known a priori. A summary of existing literature on electrically conductive immiscible blends, especially those of carbon black, carbon nanofibers (CNFs), and CNTs is presented below.
Immiscible polymer blend systems have been effectively used in the production of electrically conductive composites via preferential localization of the conductive fillers, for example, carbon black, in one of the phases [11][12][13][14]. This gives rise to double percolation-whereby electrically conductive percolation networks develop in one of the cocontinuous phases. This phenomenon was mostly observed for blends with close to 50 : 50 weight ratios of the component polymers. Dharaiya and coworkers recently exploited chaotic mixing conditions to produce fibrillar morphology of polypropylene (PP) in blending of 90 parts by weight of polyamide 6 (PA6) with 10 parts by weight of PP and used the cocontinuous fibrils of PP to obtain electrically conductive PP/PA6 compounds at only 1 wt% conductive grade carbon black [15].
Other researchers also exploited the preferential localization concept in immiscible blends filled with carbon nanofibers [16,17], or CNTs [18][19][20][21], and reported success in terms of much reduced electrical percolation threshold compared to individual polymer compounds. A schematic representation of the double percolation phenomenon in involving CNTs is shown in Figure 1. Pötschke et al. [18,19] studied electrical conductivity of polycarbonate (PC)/polyethylene blends filled with CNTs and claimed that CNT connected the two polymer phases. The resulting composites showed electrical percolation threshold at 0.41 vol.% CNT. Pötschke et al. [20] used metallocene-based complex for in-situ polymerization of ethylene directly from the CNT surface which resulted in good filler dispersion upon blending with polycarbonate and reduced the electrical percolation from 0.75 to 0.25 wt%.
It should be noted that the key factors governing the preferential localization of typical fillers are thermodynamic, kinetic, and polymer melt viscosity [22]. Thermodynamic considerations state that the filler might reside in one of the polymer phases or at the interface as a result of the overall free energy of the existence of three types of interfaces: polymer 1/polymer 2, filler/polymer 1, and filler/polymer 2. Thermodynamically, the filler location might be predicted from the spreading coefficient parameters [23]. Kinetic effects, on the other hand, are related to the mixing process. An equilibrium state of filler dispersion may not be reached in typical mixing experiments due to high viscosity of the polymers. If the system is quenched before reaching an equilibrium state of filler dispersion, incomplete migration of the filler to the thermodynamically favored phase may result [24]. Another important factor is the melt viscosity of the blend component. It has been reported that the melt viscosity factor promotes the localization of filler in a less viscous phase [21,22,24].
The concept of selective localization was used in this study to develop electrically conductive in situ composites of LCP in PC. It was hypothesized in view of fibrillar domains of LCP that the selective localization of CNTs in the continuous LCP fibrils would facilitate the anisotropic electrical conductivity in PC/LCP blends and consequently reduce the electrical percolation threshold compared to PC/CNT composites. Continuous fibrils of LCP have been observed previously in blends of polycarbonate [2], polypropylene [25], poly(ether imide) [26], and polyamide 66 [3], with LCP content in the range of 15-35 wt%. Mukherjee et al. [27] recently reported improvements in mechanical properties of PC/LCP blends in the presence of up to 5 wt% carboxylated CNTs and attributed such improvements to compatibilization effects produced by hydrogen bonding interactions between the COOH groups on CNTs and the carbonyl groups in PC and LCP. These authors observed much smaller droplets of LCP phase with the increase of CNT concentration. Although not explicitly stated in their paper, such interactions would be most effective if carboxylated CNTs remained at the PC-LCP interfaces. Electrical conductivity also showed an increase from an insulator at 1 wt% loading of caboxylated CNT to conductive composite with volume electrical conductivity of approximately 5.5 × 10 −3 S/cm and 6.5 × 10 −3 S/cm, respectively, at 3 wt% and 5 wt% CNT. Bose et al. observed higher percolation threshold in the cocontinuous polyamide 6 (PA6)/acrylonitrile-butadiene-styrene (ABS) blends containing CNTs compared to the ABS/CNT composites [28]. They found that the nanotubes preferentially resided in PA6 as PA6 displayed a closer value in surface energy as well as a lower melt viscosity. The authors explained their results in terms of good wettability of CNTs by insulating PA6 that led to the suppression of electrical conductivities observed in the PA6/ABS/CNT system.
In this contribution, PC/LCP blends were produced with 20 wt% LCP and the rheological, morphological, electrical, and mechanical properties of the resultant PC/LCP/CNT composites were characterized. The morphological changes in the blends caused by the introduction of CNTs were related to the changes in viscosity and the consequent reduction of electrical conductivity. Several specific questions were answered in this work, which were not addressed in the work of Mukherjee et al. [27,29]. First, it was not clear from the outset if unmodified CNTs would remain at the interface, move to either the LCP or PC domains, or distribute both in PC and LCP domains in the blend. The images provided in Mukherjee et al. [27] and taken via a high-resolution transmission electron microscope were inconclusive about the location of CNTs. Specific knowledge of CNT localization

Composite Preparation.
Compounds were prepared in a twin-screw, minicompounder (HAAKE MiniLab) at 290 • C for neat PC and PC/V400P/CNT, and at 300 • C for PC/VA950/CNT, with a rotational speed of 60 rpm. All the components were dried in a vacuum oven at 120 • C for 8 hrs.
The feeding sequence was as follows: PC and CNT were melt-mixed in a minicompounder for 5 minutes, taken out, then melt-mixed again for 5 minutes with the addition of a certain amount of an LCP. Samples with the prefix "REV" are of the reversed mixing sequence, where CNT was first mixed with LCP followed by mixing the compound with PC. The composite compositions were kept the same in both sequences. Injection-molded samples were prepared using a mini-injection-molding machine from DSM research with the injection-molding temperature and mold temperature of, respectively, 310 • C and 150 • C for PC, V400P and PC/V400P/CNT, and 300 • C and 80 • C for VA950 and PC/VA950/CNT. The injection pressure was 6 × 10 5 kg m −1 s −2 for all samples. The injection-molded rectangular bars were cut into the dimensions of 36.0 mm × 12.0 mm × 2.0 mm and 30.0 mm × 6.5 mm × 2.0 mm for the determination of electrical conductivity and dynamic mechanical properties, respectively. The discs of PC, V400P, and VA950 with 28 mm diameter and 0.8 mm thickness were used in measurements of contact angle.

Characterization
The rheological properties of the samples were investigated using ARES (TA Instruments) rotational rheometer. The frequency sweep measurement was performed with 25 mm parallel plate geometry, in oscillatory shear mode, and a fixed strain amplitude of 2%. A Rame-Hart contact angle goniometer (Model 100-00) operating at 1 atm and a temperature of 24 ± 2 • C was used for measurement of contact angle of two reference liquids, di-iodomethane and deionized water. A Keithley sub-Femtoamp Remote SourceMeter four-point conductivity probe (model 6430) was used for measurement of electrical conductivity of the compounds with a current of 10 μA for low resistivity samples and a voltage of 100 V for high resistivity samples. The morphology of LCP in PC/LCP/CNT compounds was examined by scanning electron microscopy (SEM, JEOL SM-71510) and transmission electron microscopy (TEM, Tecnai 12 TEM). For this purpose, PC was dissolved by immersing the blends in chloroform at room temperature for 12 h and the LCP residue was recovered by filtration. Therefore, all SEM images presented in this paper show only the LCP phase. The dynamic mechanical analyzer (RSA III) was used to study the mechanical response, such as storage modulus and loss modulus of the samples at temperature ranging from 35 • C to 200 • C and a heating rate of 4 • C/min. These tests were conducted in single point bending mode with a fixed strain of 0.2% and a vibration frequency of 1 Hz.

Melt Rheology.
The melt rheological response in frequency sweep was recorded at two temperatures according to the injection molding conditions, that is, 300 • C for the systems comprising of VA950, and at 310 • C for those comprising of V400P. As evident in Figures 2(a) and 2(b), at both temperatures, the PC showed Newtonian behavior and both LCPs exhibited non-Newtonian behavior, that is, the complex viscosity (|η * |) decreased with increasing frequency. It is widely accepted that fibril formation of the LCP in a thermoplastic matrix is governed by two major factors-the ratio of viscosity of LCP and the matrix polymer and the value of interfacial tension between the polymer components. The LCP with lower viscosity than the matrix polymer tends to deform into fibrils. In other words, the viscosity ratio of less than unity is promising for fibrillation of the LCP phase. In addition, sufficiently low interfacial tension between LCP and the matrix polymer is also required for good fibril formation. Also, the LCP content must be high to yield substantially large LCP domains, which in turn would facilitate elongation of the LCP domains. It was found that the values of |η * | of VA950 at 300 • C were slightly greater than those of PC at very low frequencies (from 0.1 to 0.8 rad/s), and lesser at higher frequencies (from 0.8 to 100 rad/s), indicating that the viscosity ratio of VA950 and PC was less than unity during injection molding. The blend of PC with 20 wt% VA950 showed viscosity higher than the individual polymer components. This can be attributed possibly to trans-esterification at the interfaces of PC and VA950 as observed by Tovar et al. [30]. It is also not surprising that the viscosity increased in compounds with the addition of CNTs. The viscosity of PC/VA950 with 5 wt% CNT was slightly lower than PC with 5 wt% CNT and was greater than VA950 with 5 wt% CNT. At 310 • C, V400P exhibited the upward concavity at low frequency ( Figure 2(b)) which is often found in LCP. The V400P grade of LCP also showed higher viscosity than PC and, therefore, was expected to form spheres instead of fibrils in PC. As expected, the viscosity of this LCP also increased in the presence of CNTs. The viscosity of PC/V400P with 5 wt% CNT was found to be lower than the composites of the corresponding polymer components containing 5 wt% CNT. This can be attributed to the lubricating effect of the V400P domains brought about by immiscibility between a flexible coil-like polymer such as polycarbonate and a rigid-rod LCP as supported by arguments from statistical thermodynamics [31].

Calculation of Spreading Coefficient.
The two-liquid geometric model was used to calculate the surface energy (1) [32,33] of the polymeric solids with the aid of contact angle measurements: In (1), γ LV and γ S are the surface energies of the reference liquid and the polymer, respectively, the superscript d refers to the dispersion part and p refers to the polar part, and θ is the measured contact angle. It should be noted that the surface energy is the sum of the dispersion and the polar parts:   Table 3 using γ d LV and γ p LV of the reference liquids provided by Kaelble [36]. Although there was some disagreement in literature for the contact angle data for PC [37][38][39], and VA950 [40,41], those values still cover our experimental values. The surface energy of V400P has not yet been reported by others and the values for CNT were taken from [42]. The polarity of the materials was found in the following order: CNT > V400P > VA950 > PC. Since the carbonaceous materials tend to localize in more polar polymers [43], we anticipated that CNT would migrate towards the LCP phase. The interfacial energy of PC/VA950 was found to be much lower than that of PC/V400P (see Table 4) suggesting that the former should be a more compatible blend than the latter and, therefore, should be more conducive to fibrillation by LCP. The spreading coefficients of PC/V400P/CNT and PC/VA950/CNT were found to be 0.92 and 1.45 mJ/m 2 , respectively, after considering CNT as component 1, PC as component 2, and LCP as component 3, and after substituting the interfacial energy values in Harkins equation. The positive value of λ 31 indicated that carbon nanotubes would be encapsulated by the LCP domains. It should be noted that the discussion in this paragraph was based on the values of surface tension data obtained at room temperature. In this context, it is recognized that the changes in surface tension of most polymers with respect to temperature were comparable, for example, 0.06-0.08 dyne · cm deg K −1 [34,44,45].

4.3.
Morphology. SEM images of LCP residue extracted from PC/V400P and PC/VA950 blends are presented in Figure 3. The V400P phase in injection molded specimens remained in the form of droplets with the number-average diameter of 0.61 μm and a weight average diameter of 0.67 μm (Figure 3(a)) as a result of lower viscosity of PC. The average diameter was computed from the diameter of at least 100 droplets measured from the SEM images.
The as-compounded PC/VA950 blends taken from the minicompounder also show spherical VA950 domains with number-average diameter of 0.49 μm and a weight-average diameter of 0.60 μm (Figure 3(b)). The VA950 domains in both fibrillar and spherical forms were found in injection molded specimen as in Figure 3(c). This can be attributed to lower pressure generated in the mini-injection molder and low rate of shear in mold filling, typically around 1.0 s −1 . One may expect more elongated fibrils in a high speed injection molding process, where the shear rate greater than 10 s −1 is typically encountered. First, higher rate of deformation and second, lowering of the viscosity ratio at higher shear rate may cause more extensive fibrillation. Subdued fibrillation and underdeveloped fibrillar structures of LCP domains may also have resulted from the feeding sequence. For example, prior mixing of CNT with LCP increased the viscosity of the LCP phase and hence increased the viscosity ratio of LCP/PC blends. In this case, only spherical domains were obtained as seen in Figure 4. Similar observations were made by Mukherjee et al. [27] in the presence of carboxylated CNTs. The surface of the droplets became rough as a result of the presence of LCP-coated CNT of ∼50 nm diameter, although the diameter of as received nanotube was reported to be 13 nm. The mixing sequence was found to influence both the size and the shape of the LCP phase morphology. For example, larger and irregular dispersed VA950 domains were found when CNT was first mixed with VA950 (Figure 4(c)) compared to those of the reversed mixing sequence (Figure 4(b)). In the reversed mixing sequence, CNT increased the viscosity of the PCphase and accordingly produced much lower viscosity ratio of LCP to PC than in mixtures without CNT. Figure 5 shows the TEM micrographs of the PC/CNT and PC/LCP/CNT composites. All these samples contained 5 wt% of CNT. The PC/CNT composite shows good CNT dispersion ( Figure 5(a)) whereas PC/LCP/CNT composites (Figures 5(b) and 5(c)) clearly reveal nonuniform CNT dispersion. The darker areas present in Figures 5(b) and 5(c) in the form of spheroids are actually the LCP domains. Clearly, the CNTs mainly resided inside the LCP droplets especially in composites of V400P LCP, in agreement with the SEM images seen in Figure 4. Some differences are noticeable in CNT dispersion between the composites PC/VA950/5 wt% CNT and PC/V400P/5 wt% CNT. For example, the TEM image in Figure 5(c) shows that CNT particles were present in both PC-phase and inside LCP domains when VA950 was used, while as per Figure 5(b), CNTs were mostly contained in V400P LCP domains. The interfacial energy data presented in Table 4 shows that CNT had very close affinity for PC and VA950 LCP, with interfacial energy of respectively 10.9 and 9.4 mJ/m 2 . On the other hand, CNTs showed more affinity to V400P LCP with interfacial energy of 0.09 mJ/m 2 . From the melt viscosity point of view, the filler tends to reside in the less viscous phase. This is in accordance with the localization of CNTs in the VA950 domains. On the other hand, the presence of V400P, having higher viscosity than PC, would result in the suppressed migration of the nanotubes into V400 phase and thus competing with the thermodynamic driving force. According to Persson and Bertilsson, viscosity effects are weak and dominate only when the difference of interactions between the filler/polymer1 and     filler/polymer 2 is small [24,46]. Since such difference in the PC/V400P/CNT system is quite large, it is not surprising that the thermodynamic factor prevailed on the viscosity factor, that is, the nanotubes selectively localized in V400P domains.  Figure 6 provides the specific electrical conductivity of five compounds as function of CNT content. The electrical conductivity at 5 wt% CNT can be expressed in the following order with the numbers in parentheses representing electrical percolation threshold: PC/CNT (1 wt%) > PC/VA950P/CNT (1 wt%) > PC/V400P/CNT (3 wt%). In this case, the percolation threshold was taken as the starting point of sharp rise of electrical conductivity. In light of the SEM and TEM images presented in Figures  4 and 5, it can be inferred that the nanotube conductive pathways were destroyed due to subdued fibrillation in the presence of either type of the LCP. The composites of LCP and CNT were found to be insulators even when the CNT content was as high as 10 wt%. Compounds of V400P and VA950 with up to 15 wt% CNT did not produce conductive composites. Therefore, in addition to restricted fibrillation, the increase in the electrical percolation threshold as well as the lowered electrical conductivity of the blends may result from encapsulation of the nanotubes with the insulating LCP as predicted by thermodynamic factors. Similar observation in blends of ABS/PA6/CNT was reported by Bose et al. [28]. Nevertheless, the composite of PC/VA950P/CNT showed percolation at 1 wt% CNT loading. This can be attributed to the presence of CNTs on the rough surfaces of VA950/CNT domains as seen in Figure 4(b) and corroborated by TEM image presented in Figure 5(c). A distinct shift in the percolation threshold to the higher CNT loading was observed in composites of V400P LCP. Although the melt viscosity of V400P was significantly higher than that of the PC matrix, the good affinity between this LCP and the nanotubes rendered the confinement of a large number of CNTs within the V400P phase. Consequently, only a small amount of CNT was available to span the matrix or to bridge across the LCP domains to form an electrical percolation network. For blends of PC/VA950/CNT, the spreading coefficient also predicts the migration of CNTs to the better-wetting VA950 phase, but not to the same extent as the blend with V400P. Subsequently, the formation of the electrical network by CNTs was more favored in the blend containing VA950 as compared to the one with V400P. Overall, the strong insulating properties of an LCP and the tendency of CNT to localize in the LCP-phase lowered the values of electrical conductivity when incorporated into the PC/CNT compounds. The affinity of CNT with both LCPs was demonstrated very clearly by the electrical conductivity values of the REV-PC/LCP/CNT composites. Here, CNT was first mixed with the LCP to obtain composite specimens REV-PC/V400P/CNT, and REV-PC/V950/CNT. During the second mixing step with PC, most of CNT remained in the LCP phase indicating that filler migration to the PC matrix was not favored. Consequently, these composites did not show electrical conductivity even at a loading of 5 wt% CNT. These results comply with the SEM image shown in Figure 4(c). Table 5 shows the values of storage modulus at 40 • C of three groups of sample-the neat polymers (PC, V400P, VA950), the blends of PC with 20 wt% LCP (PC/V400P, PC/VA950) and the composites of either PC or LCP with 5 wt% CNT (PC/CNT, V400P/CNT, VA950/CNT) or both PC/LCP with 5 wt% CNT (PC/V400P/CNT, PC/VA950/CNT). The storage moduli at less than 110 • C of compounds containing V400P showed the following order: V400P/CNT > V400P > PC/V400P > PC/CNT > PC, PC/V400P/CNT. The most surprising result is the lower value of storage modulus of PC/V400P/CNT composite in comparison with PC/V400P blend and V400P/CNT composite. It was expected that the addition of CNT to PC/V400P would increase the modulus further in view of the fact that CNT remained primarily in V400P domains and that V400P/CNT composite showed higher modulus than V400P (Table 5). The only possible explanation of this reduction of storage modulus of PC/V400P/CNT composite may be attributed to potential degradation of molecular weight of PC due to longer exposure (10 minutes) to compounding temperature compared to 5 minutes in other materials. However, reduction of PC molecular weight was not verified using chromatographic method. For VA950containing compounds, the storage moduli at low temperature show the following trend: VA950/CNT > VA950 > PC/VA950/CNT ∼ PC/VA950 ∼ PC/CNT > PC. Once again, the storage modulus of PC/VA950/CNT composite remained in the neighborhood of PC/VA950 and PC/CNT, possibly due to molecular weight reduction of the PC phase, the latter attributed to longer exposure of PC to high compounding temperature.

Conclusions
Two PC/LCPs blends were designed by adjusting the melt viscosity ratio and the shear rate used in processing, such as injection molding to interpret the trends in mechanical properties and electrical conductivity in terms of CNT localization in LCP domains, PC phase, or at the interfaces. The PC/VA950 blend showed fibrillar structures and PC/V400P blend showed spherical LCP domains. Attempt was made to distribute the CNT in both PC/LCP blends and to determine the percolation threshold of CNT for electrical conductivity, although the positive values of the spreading coefficient parameter for CNT particles promoted preferential localization of CNT inside the LCP domains. PC showed surface energy and polarity comparable to VA950 but different from V400P, due to which double percolation was found to occur only for PC/VA950. Accordingly, PC/CNT and PC/VA950/CNT composites showed the same electrical percolation threshold (1 wt%) while PC/V400P/CNT needed 3 wt% CNT to show electrical percolation. Prior addition of CNT to LCP increased the viscosity of LCP phase further, which increased the viscosity ratio to greater than unity and hindered LCP fibril formation.

Introduction
CNT-doped polymers have emerged as popular materials in the production of microelectronic components and devices. These composites possess distinct attributes characteristic of both CNTs (high strength and electrical conductivity) and polymers (light weight, flexible, and customizable), making them ideal selections for the production of a variety of devices including chemical or mechanical sensors [1], flexible electromagnetic interference (EMI) shielding [2], and "smart" textiles [3]. One critical quality essential to the functionality of all microelectronic applications is the existence of sufficient conductivity throughout the composite material. Several groups have reported that conductivity of a CNT/polymer composite will increase with increasing CNT concentration in a nonlinear or nonisotropic manner [4][5][6][7]. The complex relationship between CNT concentration and conductivity is related to the principle of "percolation," which predicts a sharp rise in conductivity when CNT concentration surpasses a threshold value where a continuous network of CNTs exists within the surrounding polymer matrix. Furthermore, processing of composites into fiber geometries has been reported to align the CNTs along the fiber axis [5,6]. However, the effect of this alignment on conductivity seems unclear, as indicated by conflicting reports [6,7] in the literature.
Micro/nanofibrous architectures have proven effective in a variety of microelectronic applications due to intrinsic advantages associated with fibrous structures including improved flexibility, maximum surface area, and increased strength. Melt spinning has emerged as a popular technique to process these bulk composites into fibers, with reported generation of fibers composed of CNT-doped polyamide 12 [8], polyamide 6 [9], PMMA [10], and polypropylene [11]. Additionally, CNT/polyvinyl alcohol (PVAc) composite fibers have been produced via wet spinning [3,12] and CNT-doped poly(lactide) (PLA) [13] and polystyrene [14] via electrospinning. In recent work by Pabba et al. [5,15], a technique was reported in which viscous CNT/PMMA solutions were manually brushed across microfabricated pillar arrays to form networks of micro-and nanoscale solution filaments, which solidified to form suspended fibers. While each of these fiber fabrication strategies enables production of large quantities of fibers, none facilitates the construction of individual fibers that can be precisely positioned on any three-dimensional substrate. This limitation imposes constraints on the design of next-generation microelectronics, particularly those with 3D features. In this investigation, CNT-doped polymer fibers that were precisely positioned in 3D were produced using a previously described direct-write technique [16,17]. The directwrite system consists of a micromanipulator-controlled syringe that was programmed to pattern arrays of fibers on a substrate. The direct-write syringe was first loaded with a test solution and positioned 500 μm above a glass substrate (Figure 1(a)). The syringe was then pressurized to expel a pendant droplet of solution, which contacts and adheres to the substrate (Figure 1(b)). Next, the syringe was lifted 1 mm and translated laterally to extend this droplet into a filament of solution (Figure 1(c)). The fiber endpoint was established by lowering the syringe to its initial height after the translation, thereby allowing the pendant drop to make an additional contact point (Figure 1(d)). Meanwhile, during extension, the solution filament underwent surface tension-driven capillary thinning, while solvent evaporation promoted the solidification of the composite fiber. By repeating this process, arrays of composite fibers were created in series in a "connect-the-dots" manner.
In our previous work, we demonstrated that the diameters of the fibers with the direct-write system from polymer solutions alone (i.e., no CNTs) were dependent on both solution properties and direct-write system parameters. Specifically, fiber diameter increased with increasing viscosity, increasing solvent evaporation rate, decreasing surface tension, decreasing fiber length, and decreasing fiberdrawing rate [16,17]. Unfortunately, the applicability of these relationships is limited by the assumption that the polymer solution behaves in a "Newtonian-like" manner (i.e., that the capillary velocity of the solution, defined as the ratio of surface tension to viscosity, remains constant over the course of direct-write processing). Although some polymer solutions were found to meet this criterion, this critical assumption represents a significant shortcoming for the previously described model. Therefore, the purpose of the current investigation is to extend the direct-write technique to CNT/polymer composites and determine how the non-Newtonian nature of these materials affects directwrite outcomes.

Solution Characterization.
The effect of axial strain on extensional viscosity was determined for PMMA solutions with and without the addition of CNTs. A capillary breakup rheometer (CaBER) (Thermo Haake 1, Thermo Electron Corp.) was utilized to measure the midpoint diameters, D mid (t), of filaments of each of the test solutions as they experienced surface tension-driven thinning. These filaments were formed by stretching cylinders of each solution in a near-instantaneous manner (t stretch = 20 ms) from a length of 2 mm to a length of 8 mm. The midpoint diameters of the elongated filaments were continuously monitored for 10 seconds with a laser micrometer at a sampling rate of 3,000 Hz.
The midpoint diameter data collected by the CaBER was analyzed in order to determine the axial strain rate,ε, of the solution filaments. Assuming uniaxial extension, the axial strain rate is correlated to midpoint diameter by [18]: This strain rate was then employed to calculate the apparent extensional viscosity, η ext , of the test solutions using [19]: η ext was plotted as a function of axial strain rate over the course of the CaBER experiments to determine the level of dependency of extensional viscosity on strain rate. The dependency of η ext onε was quantified to determine if any extensional thinning (or thickening) of the test solutions was apparent, suggesting non-Newtonian behavior during extensional flow [19]. Surface tension was measured via the Wilhelmy technique [20], and solvent mass transfer coefficient was measured using thermogravimetric analysis (TGA), as previously described [16].

Direct-Write Fiber
Fabrication. Fibers were drawn to three different lengths (4, 8, and 16 mm) from each of the test solutions at a rate of 20 mm/s, which was found to work well in previous studies [16,21]. A total of 8 drawing attempts were made for each combination of solution and length. The syringe was pressurized to expel a constant flow of approximately 20 μL/min through a 22 gauge needle (I.D. = 394 μm). After drawing, each array was coated with a thin layer (t = 10 to 20 nm) of gold/palladium alloy and the diameter of each fiber was measured in three locations (at the midpoint and 200 μm from each end) using a scanning electron microscope (SEM) (Zeiss Supra 35VP, Thornwood, NY). Process yield was determined by calculating the ratio of unbroken fibers to fiber drawing attempts.

Electrical Transport Measurement.
The conductivity of the MWNT-doped fibers was measured to quantify the electrical transport along the fibers. The ends of each fiber were metalized with silver paint to minimize contact resistance, and the resistance of each fiber was analyzed with a highly sensitive parameter analyzer (Agilent 4156, Santa Clara, CA). Conductivity for each fiber was calculated by where G is conductivity, L is the length of the fiber, R is the measured fiber resistance, and D avg is the average fiber diameter [22].

Cross-Sectional
Imaging. Cross-sections of fibers from each test solution were inspected with an SEM to determine the homogeneity of the fibers. Fibers were frozen in liquid nitrogen for 1 minute, cleaved with a razor blade, and coated with a thin layer (t = 10 to 20 nm) of gold-palladium prior to imaging. The images were then used to determine the level and distribution of CNT aggregation within the fibers.

Solution Characterization.
Solutions containing PMMA and CNTs were observed to possess surface tensions that were slightly higher than pure chlorobenzene (σ = 33.28 mN/m at 20 • C). The presence of CNTs in the solutions was found to have no significant effect on the surface tensions of the PMMA solutions (Table 1). The inclusion of CNTs in the test solutions was found to impede the evaporation of the chlorobenzene from the solutions as evidenced by the lower values in the mass transfer coefficients. The CaBER was utilized to instantaneously extend volumes of each test solution into liquid bridges with initial aspect ratios (L/D) of approximately 2.7. The diameters of these liquid bridges, D mid (t), were monitored for 10 seconds as surface tension-driven thinning occurred (Figure 2(a)). This figure illustrates that solutions with no CNTs were observed to thin more rapidly than those with 1% CNTs and the 24% PMMA solution with no CNTs thinned faster than the corresponding 26% PMMA solution. Both solutions with 1% CNTs thinned at approximately the same rate, suggesting that the presence of CNTs is the dominating factor in the surface tension-driven thinning of these solutions. The measurement of D mid (t) was used with (1) and (2) and the data from Table 1 to calculateε and η ext , which are plotted against each other in Figure 2(b). Unfortunately, the CaBER instrument was only able to apply a limited range of axial strain rates to the test solutions and this range was different for each solution (i.e., each of the curves in Figure 2(b) is for a small range of strain rate, making comparison between solutions more difficult). However, based on this limited data, it appears that the CNT composites experience shear thinning while the extensional viscosity of the PMMA-only solutions is relatively independent of shear rate.

Direct-Write Production of Fibers.
The direct-write process was successfully employed to produce CNT-doped fibers with diameters ranging from 7 μm to over 100 μm (Figure 3). The fiber diameter data is given in Figure 4(a). ANOVA analysis of this data determined that fiber diameter increased with increasing PMMA concentration (P < 0.001) and decreasing length (P < 0.001). However, the presence of 1% CNTs did not significantly influence fiber diameter (P = 0.2) except for the 4 mm long fibers drawn from the 24% polymer solution (P < 0.005). Paired t-Test analysis confirmed that the fibers possessed significantly smaller diameters midway along their lengths relative to their ends (P < 0.001). Also, the ends of the fibers where drawing was begun were significantly smaller than the ends where drawing was completed (P < 0.001). The average coefficient of variation (diameter standard deviation/diameter average) for all fibers along the length of the fibers was 34%. The process yield for the direct-write process exceeded 85% under all test conditions (Figure 4(b)).
The characterization of test solution rheology suggests that the CNT-doped solutions experience strain thinning when exposed to axial strain, whereas nondoped solutions behave in a more Newtonian manner, even at higher strain rates (Figure 2(b)). A similar trend was observed   by Rahatekar et al. who concluded that epoxy doped with CNTs will thin when exposed to increasing shearing strain rates until it asymptotically approaches the viscosity of pure epoxy [23]. Although the CaBER system was not capable of inducing high axial strain rates (ε > 0.1) on the CNTdoped solutions, the orientations of the curves in Figure 2(b) suggests that such an asymptotic relationship may exist (i.e., the extensional viscosity of the CNT-doped solutions could approach that of the pure PMMA solutions if axial strain rate was increased). During direct-write drawing, the solution experiences axial strain due to surface tension-driven thinning in a manner similar to the CaBER experiments. However, the motion of the direct-write syringe, which imposed additional axial strain on the solution, augmented this strain. The magnitude of this strain generated by the motion of the direct-write syringe, ε DW , is given by [24]: where L(t) is the length of the direct-write fiber as it is drawn and L 0 is the initial length of the solution droplet on the direct-write syringe before drawing (L 0 ≈ 394 μm).
Equation (4) is differentiated to obtain the axial strain rate imparted by the direct-write process: where dL(t)/dt is the drawing velocity of the direct-write system. Figure 4(c) illustrates the magnitude of the axial strain rate generated by the direct-write process for fibers drawn up to lengths of 16 mm at a test velocity of 20 mm/s. Comparison of these axial strain rates with those generated by the CaBER (Figure 2(b)), which expresses the magnitude of axial strain rate during surface tension-driven thinning only, suggest that the motion of the drawing process influences axial strain rate by an approximate order of magnitude more than the surface tension-driven thinning motion alone. Furthermore, the axial strain rates encountered during the direct-write process may be sufficiently large to thin the CNT-doped solutions into the asymptotic regime where η ext approaches η ext of the homopolymer system, providing a potential explanation as to why CNT concentration was shown not to influence fiber diameter, except in the case of the shortest fibers (i.e., lowest applied strain) with the lower polymer concentration (i.e., more thinning of the undoped solution).