7004
Langmuir 2008, 24, 7004-7010
Conductive Macroporous Composite Chitosan-Carbon Nanotube Scaffolds Carolin Lau and Michael J. Cooney* Hawaii Natural Energy Institute, UniVersity of Hawaii at Manoa, Manoa, Hawaii 96822
Plamen Atanassov Department of Chemical and Nuclear Engineering, UniVersity of New Mexico, Albuquerque, New Mexico 87131 ReceiVed February 27, 2008. ReVised Manuscript ReceiVed March 25, 2008 Multiwalled carbon nanotubes (MWCNTs) were used as doping material for three-dimensional chitosan scaffolds to develop a highly conductive, porous, and biocompatible composite material. The porous and interconnected structures were formed by the process of thermally induced phase separation followed by freeze-drying applied to an aqueous solution of 1 wt % chitosan acetic acid. The porosity was characterized to be 97% by both mercury intrusion porosimetry measurements and SEM image analysis. When MWCNTs were used as a filler to introduce conductive pathways throughout the chitosan skeleton, the solubilizing hydrophobic and hydrophilic properties of chitosan established stable polymer/MWCNT solutions that yielded a homogeneous distribution of nanotubes throughout the final composite matrix. A percolation theory threshold of ∼2.5 wt % MWCNTs was determined by measurement of the conductivity as a function of chitosan/MWCNT ratios. The powder resistivity of completely compressed scaffolds also was measured and was found to be similar for all MWCNT concentrations (0.7-0.15 Ω cm powder resistivity for MWCNTs of 0.8-5 wt %) and almost five times lower than the 20 kΩ cm value found for pure chitosan scaffolds.
1. Introduction Chitosan is a polyionic biopolymer that has been widely used as a material to which tissue cells are attached, biomolecules are entrapped, or enzymes are immobilized.1–4 A common approach has been to entrap or adsorb biomolecules to (or within) chitosan films via a dip-coating technique. Other methods have applied a freeze gelation technique to produce macroporous scaffolds for the slow release of drugs or as a support matrix for cell tissue growth.2,5,6 More recently, chitosan-carbon nanotube (CHITCNT) composite films were proposed as a conductive matrix supporting direct electron transfer in biosensors.7,8 In this work we present, to our knowledge for the first time, the concept of macroporous and highly conductive CHIT-CNT composite scaffolds as an advanced material for the fabrication of biocatalyzed electrodes that can support both mediated and direct electron transfer processes. Such biocatalytic electrodes can be tailored for both enzyme immobilization capacity and porosity to accommodate mass transfer in flow-through electrodes in biosensors and biofuel cells. The attraction to chitosan lies in its carboxyl and amine side groups that can support functional manipulations such as the direct binding of enzymes for immobilization,1,9,10 covalent * Corresponding author. E-mail:
[email protected]; tel.: (808) 9567337; fax: (808) 956-2336. (1) Rinaudo, M. Prog. Polym. Sci. 2006, 31, 603. (2) Madihally, S. V.; Matthew, H. W. T. Biomaterials 1999, 20, 1133. (3) Krajewska, B. Enzyme Microb. Technol. 2004, 35, 126. (4) Huang, Y.; Onyeri, S.; Siewe, M.; Moshfeghian, A.; Madihally, S. v. Biomaterials 2005, 26, 7616. (5) Ho, M. H.; Kuo, P. Y.; Hsieh, H. J.; Hsien, T. Y.; Hou, L. T.; Lai, J. Y.; Wang, D. M. Biomaterials 2004, 25, 129. (6) Leffler, C.; Muller, B. W. Int. J. Pharm. 2000, 194, 229. (7) Liu, Y.; Wang, M.; Zhao, F.; Xu, Z.; Dong, S. Biosens. Bioelectron. 2005, 21, 984. (8) Liu, Y.; Liu, L.; Dong, S. Electroanalysis 2007, 19, 55. (9) Klotzbach, T.; Watt, M.; Ansari, Y.; Minteer, S. D. J. Membr. Sci. 2006, 282, 276.
attachment of functional groups for the modification of polymer properties (such as relative degree of hydrophobicity), and because of the polymer’s ability to form a semirigid precipitate that can be manufactured into a variety of useful forms including gels, scaffolds, beads, fibers, or films.3,11,12 Chitosan scaffolds are produced by the process of thermally induced phase separation (TIPS), whereby the water within a mostly aqueous chitosan solution is frozen into ice crystals that are surrounded by thin layers of the precipitated polymer. A highly macroporous scaffoldlike structure is obtained when the water retained in the ice crystals is subsequently sublimed under vacuum. The final structure of the macropores can be controlled by a number of factors including freezing rate, orientation of the applied thermal gradients, pH, and concentration of chitosan.2,5,11,13–15 Since chitosan is nonconductive, its immediate application to biosensor and/or biofuel cell development is limited to mediator-based electron transfer systems such as the NADH dependent dehydrogenases.16 Although valuable, mediator-based systems are potentially limited by the requirement of mediator diffusion to the electrode surface. Its value, therefore, would be greatly enhanced if chitosan could be made conductive such that the diffusion distance is shortened (i.e., in mediator-based systems) or the need for mediators is eliminated (i.e., in direct electron transfer systems). One method would be to bind aromatic or metal-based mediators along the chitosan backbone, thus creating a matrix that supports electron (10) Wang, Q. Z.; Chen, X. G.; Liu, N.; Wang, S. X.; Liu, C. S.; Meng, X. H.; Liu, C. G. Carbohydr. Polym. 2006, 65, 194. (11) Roh, I. J.; Kwon, I. C. J. Biomater. Sci. 2002, 13, 769. (12) Siso, M. I. G.; Lang, E.; Carreno Gomez, B.; Becerra, M.; Espinar, F. O.; Mendez, J. B. Proc. Biochem. 1997, 32, 211. (13) Falk, B.; Garramone, S.; Shivkumar, S. Mater. Lett. 2004, 58, 3261. (14) Wan, Y.; Fang, Y.; Wu, H.; Cao, X. J. Biomed. Mater. Res., Part A 2007, 80, 776. (15) Cooney, M. J.; Lau, C.; Windmeisser, M.; Liaw, B. Y.; Klotzbach, T.; Minteer, S. D. J. Mater. Chem. 2008, in press. (16) Akers, N. L.; Moore, C. M.; Minteer, S. D. Electrochim. Acta 2005, 50, 2521.
10.1021/la8005597 CCC: $40.75 2008 American Chemical Society Published on Web 06/03/2008
Composite Chitosan-Carbon Nanotube Scaffolds
“hopping” along the polymer backbone.17 A second method would be to add a conductive material to the chitosan solution at concentrations that surpass a percolation threshold.18 Although composite materials have been made conductive in this manner,19 this approach has the requirement, in the context of biosensor and biofuel cell development, of being able to facilitate direct electron transfer between the enzyme and the conductive doping material.7,20 CNTs have been proposed as an additive to facilitate direct electron transport (between entrapped enzymes and electrode surface) in thin chitosan films.7 While this method relies solely on the mixing and drying process to create films within which the entrapped enzymes are in sufficient proximity to the CNTs to permit direct electron transfer, the enzymes also can be directly bound to the CNTs.21 The dispersion of CNTs in aqueous solutions generally requires their initial functionalization with hydrophilic side groups (such as carboxyl groups) through various chemical pretreatment techniques.22 This technique, however, can lower their conductivity. To counter this, researchers have proposed the use of surfactants (e.g., biomaterials, biomolecules, biopolymers, and other bionanostructures)21 that ease the surface tension between the hydrophilic aqueous phase and the hydrophobic surface of the CNTs.23 Chitosan has been proposed in this regard because it is a hydrophobic material that can be made more hydrophilic by the protonation of its amine side groups at pH values below approximately 5.0 (e.g., by the addition of acetic acid). In this sense, chitosan can be made to possess amphiphilic properties that give it a unique capacity to solubilize hydrophobic CNTs in aqueous solution.24–26 Because of this ability to combine the mechanical and electrical properties of CNTs with the macroporous scaffold forming properties of chitosan, CHIT-CNT composite scaffolds promise to provide a breakthrough platform technology for applications in biosensing or bioenergy.19,27–29 A key characteristic to a CHIT-CNT composite scaffold is its conductivity as defined by the charge transfer from one conductive particle to another. Since the conduction of electrical charge is established when the network of conductive CNTs reaches a critical percolation threshold density that provides direct electrical contact between particles, the effective conductivity of the CHIT-CNT composite scaffold will depend upon many factors such as size, shape, density, and distribution of CNTs within the chitosan precipitate, as well as chemical interactions between the two materials.18,19 In the case of a thin insulating chitosan layer surrounding the CNTs, the conductivity is expected to be dominated by tunneling effects.30 In the case of chitosan scaffolds, the conductivity is expected to be dominated by direct contact between CNTs. The contact resistance between particles (17) Katakis, I.; Heller, A. In Frontiers in Biosensorics I: Fundamental Aspects; Scheller, F. W., Schubert, F., Fedrowitz, J., Eds.; Birkha¨user Verlag: Basel, Switzerland, 1997. (18) Foygel, M.; Morris, R. D.; Anez, D.; French, S.; Sobolev, V. L. Phys. ReV. B: Condens. Matter Mater. Phys. 2005, 71, 104201. (19) Wescott, J. T.; Kung, P.; Maiti, A. Appl. Phys. Lett. 2007, 90, 33116/1. (20) Ivnitski, D.; Branch, B.; Atanassov, P.; Apblett, C. Electrochem. Commun. 2006, 8, 1204. (21) Katz, E.; Willner, I. Chem. Phys. Chem. 2004, 5, 1084. (22) Tasis, D.; Tagmatarchis, N.; Bianco, A.; Prato, M. Chem. ReV. 2006, 106, 1105. (23) Wang, J.; Musameh, M.; Lin, Y. J. Am. Chem. Soc. 2003, 125, 2408. (24) Liu, Y.; Tang, J.; Chen, X.; Xin, J. H. Carbon 2005, 43, 3178. (25) Tkac, J.; Whittaker, J. W.; Ruzgas, T. Biosens. Bioelectron. 2007, 22, 1820. (26) Watts, P. C. P.; Hsu, W. K.; Chen, G. Z.; Fray, D. J.; Kroto, H. W.; Walton, D. R. M. J. Mater. Chem. 2001, 11, 2482. (27) Liu, Y.; Qu, X.; Guo, H.; Chen, H.; Liu, B.; Dong, S. Biosens. Bioelectron. 2006, 21, 2195. (28) Qian, L.; Yang, X. Talanta 2006, 68, 721. (29) Zhang, M.; Smith, A.; Gorski, W. Anal. Chem. 2004, 76, 5045. (30) Zallen, R. The Physics of Amorphous Solids; John Wiley and Sons: New York, 1983.
Langmuir, Vol. 24, No. 13, 2008 7005
is therefore particularly important and will depend upon their concentration and physical geometry (i.e., if cut or bent or twisted). With respect to the latter, cut and nonpristine multiwalled (MW) CNTs are highly preferred. Assuming that these factors are wellunderstood and -controlled during the fabrication process, the reproducible production of conductive CHIT-CNT scaffolds is possible.
2. Experimental Procedures 2.1. Chemicals. Chitosan (medium molecular weight, SigmaAldrich) stock solutions (1 wt %) were prepared by dissolving the polysaccharide in 0.2 M acetic acid. To eliminate all visible air bubbles, the chitosan solution was allowed to sit for at least 24 h before use. A stock solution of MWCNTs was prepared by dispersing the MWCNTs (20-30 nm outer diameter, 10-30 µm length, 95 wt % purity from www.cheaptubesinc.com) in a 1 wt % chitosan (0.2 M HOAc) solution to a final MWCNT concentration of 5 wt %. The mixture was vortexed for 15 min before storage at room temperature under permanent stirring with a micromagnetic stirring bar. This was sufficient to yield stable “black ink” CHIT-MWCNT suspensions without additional pretreatment of the MWCNTs. More dilute solutions (0.83-3.5 wt %) of MWCNTs were made by mixing the stock solutions of pure CHIT and CHIT-MWCNT in the required ratios. The MWCNTs selected for this project were of average size and diameter for several reasons. First, inexpensive manufactured tubes generally possess more flaws in their construction (i.e., surface defectssbends and breaks) that enhance the formation of electrically conductive networks. Second, MWCNTs show metallic conductivity as compared to semiconductive single walled CNTs. Third, because of the underlying forces during the phase separation process, we wanted to avoid the use of very short tubes that would decrease the statistical likelihood of entanglements, and therefore require greater concentrations to achieve a percolation threshold, and tubes of a length that was significantly larger than the average diameter of the chitosan macropores. For these reasons, MWCNTs of a reasonable diameter (20-30 nm) and length (10-30 um) were selected as appropriate for these studies. 2.2. CHIT-MWCNT Scaffold Preparation. The chitosan scaffolds were prepared by freezing 500 µL of a CHIT-MWCNT dispersion (multiple MWCNT concentrations) into a specially fabricated aluminum mold (10 mm × 10 mm × 10 mm, 1 mL total volume). The mold was directly attached to a solid state cold plate (AHP-300CP, ThermoElectric Cooling America Corp.), which permitted cooling of the solution to -20 °C in just under 8 min. For resistivity measurements using the distributed pin heads, a 4 × 4 gold pin matrix (SULLINS part no. PRPNxx4RCCN) was centrically placed into the solution before freezing in situ. After freezing at -20 °C for 1 h, the mold was transferred to a freeze-drier (Sentry, VirTis Co.), where it was freeze-dried for at least 5 h until being stored at room temperature and under desiccation. 2.3. Resistivity Measurement. As described previously, all CHIT-MWCNT composite scaffolds were prepared in situ around commercially available 4 × 4 gold headers (pins), and the volume resistances R across the eight symmetrically arranged electrodes (Scheme 1) were measured using a digital multimeter (Keithley 238 High Current Source Measure Unit) operated in four terminal current-voltage (I-V) configurations to guarantee a homogeneous electrical field around the measured circuit. The scaffold volume resistivities F were averaged over eight individual measurements taken across distributed single resistances within the CHIT-MWCNT composite scaffold. Each header pin possessed a quantity of gold surface coating, surface area A, and interpin distance d. The measured volume resistances R were converted to volume resistivities F using
F ) RA ⁄ d
(1)
where A is the effective area of the measuring electrode (A ) 4.4 × 0.5 mm), and d is the distance between two measuring electrodes (d × 2.0 mm). 2.4. Powder Resistivity Measurements. From scaffolds that had already been used for the resistivity measurements, a cylinder-like
7006 Langmuir, Vol. 24, No. 13, 2008 Scheme 1. Schematic Drawing of Location of 4 × 4 Gold Electrode Matrix Inside a Scaffolda
Lau et al. voltage of 15 kV at various levels of magnification. The mean pore diameter was estimated by digital analysis of SEM images of crosssectioned scaffold areas with imaging software. The length and width of an individual pore was outlined and measured. The pore mean diameter was then calculated as the arithmetic mean of at least 50 estimated pore diameter values. Mean pore diameters were estimated from a single SEM image of a complete scaffold cross-section generally focused on the center of the scaffold. High magnification images of the MWCNT-CHIT scaffold surface morphology were obtained using a Hitachi (S-5200) scanning electron microscope operated at 10 kV and on samples that were not coated.
3. Results and Discussion
a For four terminal resistivity measurements, the outer four electrodes were connected to the outer electrical field, whereas the resistivity between each pair (R1-R4) of inner electrodes was measured. By rotating the scaffold around 90°, R5-R8 are measurable in the same way.
piece was cut with a hole-punch. The cut scaffold was then placed within a special conductivity device, consisting of two brass stamp electrodes whose distance between each other can be tightly controlled by compression screwing. The resistance across the two brass stamp electrodes was then measured with a digital multimeter with 1 GΩ internal resistance (Keithley 238 High Current Source Measure Unit), starting at the first measured contact and thereafter as defined compression distances (as measured with calipers). 2.5. SEM Image Analysis and Pore Size Estimation. SEM samples were prepared by slicing pieces of the freeze-dried scaffolds with sharp razor blades, and each slice was fixed onto a conductive stub. The samples were then sputter-coated with gold-palladium for 50 s (using a Hummer 6.2) prior to their placement into the SEM chamber (Hitachi S-800). Images were taken with an accelerating
3.1. Pore Structure Characterization. In this work, threedimensional and macroporous CHIT-CNT composite scaffolds were fabricated from aqueous mixtures of chitosan and MWCNTs using a two-step process in which the CHIT-MWCNT solution was first frozen and then freeze-dried. The resulting scaffold macroporous structure was characterized with respect to the average pore size, porosity, and pore size distribution using software assisted analysis of SEM images and mercury intrusion porosimetry. Typical macroporous and interconnected pore structures of pure chitosan and CHIT-MWCNT composites are shown in Figure 1A-D. Visual comparison of a pure chitosan scaffold (Figure 1A) against CHIT-MWCNT composite scaffolds (Figure 1B-D) suggests that under identical conditions of chitosan concentration (1 wt %) and freezing time (1 h, -20 °C), the pore structure of the CHIT-MWCNT composite scaffolds is relatively similar to that made from pure chitosan, although the surface texture of the CHIT-CNT composite scaffold is rougher and possessive of distinct agglomerates. The images also reveal that the MWCNTs are mostly sequestered within the concentrated chitosan precipitate, suggesting that the chitosan polymer works as a binder of the
Figure 1. SEM images of CHIT-MWCNT scaffolds of different CHIT/CNT ratios. All scaffolds were obtained with 1 wt % chitosan solution (0.2 M HOAc), frozen at -20 °C for 1 h. Panels A-D show macroporous structures at lower magnification, and panels E-G depict high magnification of scaffold walls to show the CNT distribution.
Composite Chitosan-Carbon Nanotube Scaffolds
Figure 2. High magnification SEM images of CHIT-MWCNT scaffold surfaces of (A) 2.5 wt % MWCNT and (B) 5 wt % MWCNT. All scaffolds were obtained with 1 wt % chitosan solution (0.2 M HOAc), frozen at -20 °C for 1 h.
MWCNTs that preferably concentrate within the more hydrophobic polymer phase. As the concentration of MWCNTs in the precipitate increases, however, the degree to which the tips of the MWCNTs puncture the precipitate’s surface increases. At the lowest MWCNT concentrations (Figure 1E), for example, the chitosan surface is mostly smooth with few distinguishable MWCNTs visible. At higher MWCNT concentrations (Figure 1E-H), however, the tips of the MWCNTs begin to puncture the surface in a “spaghetti”-like arrangement that can only be observed under much higher magnification (Figure 2). The diameter of the spaghetti-like appendages was estimated to be between 25 and 40 nm, correlating well with the manufacturer stated dimensions of the MWCNTs (20-30 nm outer diameter, 10-30 µm length). The slightly increased thickness is attributed to the presence of chitosan. It would appear that as the relative concentration of MWCNTs increases, a point is reached wherein the amount of chitosan present as a precipitate cannot cover the entire MWCNTs, thereby forcing portions of the MWCNTs to rest within the ice phase during the thermally induced phase separation process. The result, after the ice is sublimed away, is a lawn of MWCNTs lining the surfaces of the scaffold macropores that provides a point of direct electrical contact for biomolecules and enzymes capable of direct electron transport (e.g., as shown for glucose oxidase on CVD (chemical vapor deposition) grown MWNCTs on Toray carbon paper).20 The authors previously reported the fabrication of macroporous scaffolds made from chitosan that had been hydrophobically pretreated to possess mesopores that entrap enzymes.15 Although scaffold electrodes made from this material were applied to mediator-based electron transfer systems (i.e., NADH dependent glucose dehydrogenase), the authors clearly demonstrated the ability of macroporous scaffold electrodes to provide a greater power density than thin film electrodes (all other elements being equal). In the context of fabricating macroporous CHIT-MWCNT composite scaffold electrodes that support direct electron transport, the demonstration of functional pore structure, distribution, and interconnectivity is equally important. Figure 3 presents the mean pore diameter and distribution of CHIT-MWCNT composite scaffolds as a function of the MWCNT concentration in the original solution or final precipitate. As can be seen, the initial effect of adding MWCNTs is to decrease the mean pore diameter (i.e., the initial addition of ∼16 vol % MWCNT decreases the mean pore size by ∼40%), presumably due to the additional volume provided by the filling material
Langmuir, Vol. 24, No. 13, 2008 7007
Figure 3. Mean pore diameter of chitosan scaffolds as a function of MWCNT concentration after freezing for 1 h at -20 °C. Insets: examples of statistical histograms used to estimate pore size (A: pure chitosan (1 wt %) and B: CHIT-MWCNT (1 wt %/5 wt %).
(i.e., MWCNTs). Further increases in the concentration of MWCNTs, however, do not result in further pore size reduction, with a final mean pore size of ∼10-15 µm realized by 2.5 wt % MWCNTs in the original solution (or 50 wt % in the final precipitate). The statistical distribution of mean pore diameters (as shown in the histograms for both pure CHIT and CHIT-MWCNT scaffold of highest MWCNT concentration in Figure 3A,B) is almost symmetrical in shape, suggesting roughly spherical pores. This result agrees with our previous findings15 that chitosan scaffolds fabricated under conditions that favor mass transfer domination (i.e., -20 °C, 1 h) are more or less spherical in shape (as opposed to a layered sheet-like pore structure). By contrast, the CHIT-MWCNT composite scaffolds become more brittle and fragile with increasing MWCNT ratio and are devoid of the characteristic compressibility associated with pure CHIT scaffolds. We suggest that the higher MWCNT concentrations presented in this work represent an upper limit in terms of acceptable mechanical stability of such a CHITMWCNT composite. The scaffolds of higher MWCNT concentrations also possess more fiber-like structures embedded within the macroporous structure (Figure 1E,G). These structures, which serve as a skeleton for the establishment of chitosan bridges, presumably form because of the natural agglomerating behavior of CNTs. Finally, the pore walls are often fractured, making pore analysis more difficult. We also characterized the pore structure of our CHIT-MWCNT composite scaffolds using mercury porosimetry. Specifically, a single pure CHIT (1 wt %) scaffold and one CHIT-MWCNT (1 wt %/2.5 wt %) scaffold were characterized for their porosity and mean pore diameter. All mercury intrusion and SEM results are given in Table 1 along with the scaffold weight and volume that were used to estimate a density term that was then used to estimate (assuming no scaffold shrinking during the drying process and a MWCNT density of 1.75 g/cm3)19 the volume ratio of CNTs in precipitated chitosan (i.e., 0.83, 1.43, 2.5, 3.5, and 5 wt % are equivivalent to 16, 27, 48, 66, and 95 vol %, respectively). The scaffold porosity was measured to be 97% for scaffolds with and without MWCNTs, and the density of the MWCNT scaffold was roughly 1.5 times as dense as the pure chitosan scaffold. The differential intrusion data (not shown) suggest a bimodal pore size distribution for both scaffolds: pores larger than 100 µm and pores smaller than 50 µm (Table 1). Since mercury, which is a nonwetting liquid, penetrates all hydrophobic and hydrophilic
7008 Langmuir, Vol. 24, No. 13, 2008
Lau et al.
Table 1. Mercury Intrusion Porosimetry Data for Two Scaffold Samples pure CHIT scaffold content scaffold vol, scaffold wt bimodal pore diameters (µm) mean pore diameter (µm) bulk density, skeletal density (g/mL) porosity (%) mean pore diameter (µm)
1 wt % CHIT 800 µL, 0.0084 g Mercury intrusion porosimetry data 25 ( 8, 155 ( 65 47.1 0.0353, 1.4063 97.49 SEM image data analysis 36 ( 11
pores regardless as to whether they are blind or through pores, the measured pore diameter represents the largest diameter of any given pore. The mean reported by the software is the mean of all pore lengths. This suggests that the measurement software might confuse what SEM visual analysis would interpret as a series of smaller interconnected pores with a single long pore (otherwise termed a pore hole). That being said, the software porosimetry reported distribution of pores above 100 µm is most likely a reflection of unrealistic surface pores caused by sample preparation (i.e., cutting of the sample with razor blades) and occasional lengths of interconnected pores that are misinterpreted to be pore holes (i.e., throats). By contrast, the more narrow software reported distribution of pores below 50 µm in diameter likely reflects mercury’s intrusion into the internal and interconnected pores (as seen just below the surface in the SEM images). In general, the software reported smaller pore diameters (
