Layer-by-Layer Carbon Nanotube Electrodes for Flexible Chemical Sensor and Energy Storage Applications by ARCHNEW MASSACHUSETTS INN YlE OF TECHNOLOGY Kittipong Saetia M.S. Chemical Engineering Practice Massachusetts Institute of Technology, 2010 B.S. Chemical Engineering Northwestern University, 2007 MAR 2 5 2014 LIBRARIES SUBMITTED TO THE DEPARTMENT OF CHEMICAL ENGINEERING IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF: DOCTOR OF PHILOSOPHY IN CHEMICAL ENGINEERING AT THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY SEPTEMBER 2013 0 2013 Massachusetts Institute of Technology. All rights reserved. Signature of Author.................. ... ................... Kittipong Saetia Department of Chemical Engineering September 4, 2013 C ertified B y ........................... .......................... A ccepted B y ......................................... ....................................... Paula T. Hammond Professor of Chemical Engineering ,Thesis Supervisor ......................... Patrick S. Doyle Professor of Chemical Engineering Chairman, Committee for Graduate Students Layer-by-Layer Carbon Nanotube Electrodes for Flexible Chemical Sensor and Energy Storage Applications by Kittipong Saetia Submitted to the Department of Chemical Engineering On September 4, 2013, in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Chemical Engineering At the Massachusetts Institute of Technology Abstract The spray-assisted layer-by-layer (LbL) technique has been investigated for creating multi-walled carbon nanotube (MWNT) films on porous electrospun fiber mats via LbL assembly of surface-functionalized MWNTs. Negative and positive charges were introduced on outer walls of MWNTs by surface functionalization to enable the assembly of MWNTs into multilayer films without binder via the LbL technique. Conformal coating of MWNT films on individual electrospun fibers was achieved by applying a pressure gradient across the electrospun fiber mat to generate flow throughout the porous membrane during the spray-LbL deposition. The resulting MWNT films were shown to have interpenetrating network of unbundled MWNTs with nanoporous texture, which is well suited as electrode material for numerous applications. These LbLMWNT/electrospun fiber electrodes were utilized for applications in chemical sensing and energy storage as described below. First, the LbL-MWNT/electrospun fiber electrodes were employed as flexible chemiresistive sensors for real-time detection of a nerve agent simulant (dimethyl methylphosphonate (DMMP)). Available functional groups (-NH2 and -COOH) on the MWNT films offer a direct route for covalently attaching a receptor of interest to the MWNTs, in order to tailor the chemical specificity in a modular fashion. Here, a thiourea-based receptor was covalently attached to the -NH 2 functional groups on the MWNT/ES fiber electrodes to enhance sensitivity toward DMMP via hydrogen bonding interaction, resulting in an up to 3-fold increase in sensing response. Chemiresistive sensors based on the engineered textiles displayed reversible responses and detection limits for DMMP as low as 10 ppb in the aqueous phase and 5 ppm in the vapor phase. Employing the LbL-MWNT/electrospun fiber electrodes described above, we created hierarchical porous (HP) MWNT electrodes by removing the electrospun fiber substrate. The resulting HP-MWNT electrodes, which were self-standing and tens of micron in thickness, contained 3D interconnected macropores and mesoporous network of LbL-MWNT films. HP-MWNT electrodes can deliver a high gravimetric energy of ~100 Wh kg-1 at a gravimetric power of -25 kW kg' in lithium nonaqueous cells. Compared to similar MWNT electrodes without macropores, HP-MWNT electrodes 3 exhibited superior rate capability with an up to 10-fold increase in energy retention when the power was increased from ~100 W kg-1 to ~100 kW kg~1 . These HP-MWNT electrodes possess great potential as electrode materials for high-rate electrochemical energy storage applications such as electric vehicles. In summary, this thesis presents a versatile and easily scalable strategy for depositing functionalized MWNT films on 3D porous substrates to create tailored nanostructured electrodes with for various applications, including flexible chemical sensors and energy storage devices. Thesis Supervisor: Paula T. Hammond, Professor of Chemical Engineering 4 Acknowledgements I would like to first thank my advisor Prof. Paula Hammond for giving me the opportunity to work in her lab and helping me grow as a scientist. My thesis would not have possible without her continuous support, encouragement and advice. I am grateful for the freedom she gave me to explore my own projects during my thesis work. I thank my thesis committee members Prof. Bill Deen, Prof. Tim Swager, and Prof. Karen Gleason for all of their helpful suggestions and guidance. I feel very fortunate to have had the opportunity to interact with and learn from such a distinguished group. I also thank Prof. Greg Rutledge for his constructive and insightful comments. I thank Kevin Krogman for introducing spray-LbL to me when I first joined the lab and mentoring me during my first semester as a Hammond Group member. I had a pleasure of working with two very hard-working, smart collaborators, Jan Schnorr (Swager Group) and Matthew Mannarino (Rutledge Group). I thank Jan Schnorr for his assistance with the chemical sensor work (Chapter 2) and helpful discussions about carbon nanotube sensors. I thank Matthew Mannarino for his expertise on electrospinning and supply of electrospun fiber mats. I thank my fellow grad students, Simon Choong, Armon Sharei, Kevin Lin, Harry An, Yuko Kida, Bridget Navarro, Jessie Wong, Yin Fan, Jin Sunthivich, Tanya Shatova, and Agustin Lopez, for their friendship and making my experience at MIT a very enjoyable one. I thank former and present Hammond group members for their friendship and guidance. I have really enjoyed interacting with all the Paula-mers: Ben Almquist, Julio D'Arcy, Avni Argun, Dan Bonner, Eunice Costa, Nicole Davis, Peter DeMuth, Erik Dreaden, Amanda Engler, Kevin Huang, Bryan Hsu, Nasim Hyder, Sung Yeol Kim, Jason Kovacs, Kevin Krogman, Becky Ladewsky, Jung-Ah Lee, David Liu, Younjin Min, Isa Montiero, Josh Moskowitz, Grinia Nogueira, Jim Gifford, Megan O'Grady, Mike Petr, Zhiyong Poon, Ray Samuel, Anita Shukla, Jessie Wong, Wei Li, Myoung-Hwan Park, Mohiuddin Quadir, Steven Castleberry, Po-Yen Chen, Sun Hwa Lee, Samantha Collins, Noemie Dorval-Courchesne, Jonathon Harding, Mackenzie Martin, Jouha Min, Shanon Morey, Stephen Morton, Nisarg Shah, Marianna Sofman, and Brian Aitken. I thank Steve Kooi and Bill DiNatale for their help with lab instruments at the ISN. Many of my experiments would have been much more difficult to accomplish without their help. I thank the undergraduate students who have worked with me over the years, including Paul Heyse, Cadet Samuel Lowell and Cadet Joshua Dillard. 5 I thank Elizabeth Galoyan and Christine Preston for helping me with administrative matters and being super nice. This work was supported primarily by the U.S. Army through the Institute for Soldier Nanotechnologies, under contract DAAD-19-02-D-0002 with the U.S. Army Research Office. The content does not necessarily reflect the position of the government, and no official endorsement should be inferred. This work was also supported financially by the National Science Foundation Center for Chemical Innovation (CCI) under NSF grant number CHE-0802907 as well as ENI SpA as part of the Solar Frontiers Research Program. We acknowledge the use of facilities supported by the Institute for Soldier Nanotechnologies (ISN), Center for Materials Science and Engineering (CMSE) and Koch Institute at MIT. Finally, I thank my parents, Tongchai Saetia and Somjai Saebae, brothers, Srith, Kittikhun, Chettha, Kittiphun and Ben Saetia, sisters, Parinda and Nathika Saetia, grandmother, and sister-in-law, Maureen Saetia, for their constant support, encouragement and love. 6 Table of Contents ACKNOW LEDGEMENTS .......................................................................................................................... 5 TABL E OF CONTENTS .............................................................................................................................. 7 LIST OF FIGURES....................................................................................................................................... 8 1. INTRODUCTION ............................................................................................................................... 1.1 LAYER-BY-LAYER ASSEMBLY............................................................................................................. 1.2 APPLICATION TO CHEMICAL DETECTION........................................................................................... 1.3 APPLICATION TO ELECTROCHEMICAL ENERGY STORAGE SYSTEMS................................................. 1.4 R EFE REN CE S ........................................................................................................................................ 2. CARBON NANOTUBE/ELECTROSPUN FIBER ELECTRODES FOR FLEXIBLE CHEMIRESISTIVE SENSOR APPLICATIONS................................................................................ A B ST RA C T ................................................................................................................................................. 2 .1 IN TRO D U CTION .................................................................................................................................... 2.2 EXPERIMENTAL SECTION..................................................................................................................... 2.3 RESULTS AND DISCUSSION .................................................................................................................. 12 12 14 16 20 21 21 22 26 32 2.1. Fabricationand Characterizationof MWNT-Film-CoatedElectrospun FiberMats............... 32 2.2. Functionalizationof Thiourea-basedReceptor to the MWNT/ES FiberElectrode.................. 45 2.3. Response of Chemiresistive Sensor Based on MWNT/ES FiberElectrodeto DMMP ............. 47 2 .4 CO N CLU SION S ...................................................................................................................................... 2 .5 R E FE REN CE S ........................................................................................................................................ 57 58 3. SELF-STANDING CARBON NANOTUBE ELECTRODES WITH 3D HIERARCHICAL POROUS STRUCTURE FOR HIGH-RATE ENERGY STORAGE APPLICATIONS..................61 AB STRA C T ................................................................................................................................................. 3 .1 IN TRO D U CTIO N .................................................................................................................................... 3.2 EXPERIMENTAL SECTION..................................................................................................................... 3.3 RESULTS AND DISCUSSION .................................................................................................................. 3.3.1 Electrode Preparationand Characterization........................................................................... 3.3.3 Electrochemicalcharacteristicsof MWNT electrodes........................................................... 61 62 64 69 69 80 3 .4 CO N C LU SION S ...................................................................................................................................... 3 .5 RE FE REN CE S ........................................................................................................................................ 90 91 4. CONCLUSION AND FUTURE RECOMMENDATIONS ............................................................. 93 7 List of Figures Figure 1-1. Schematic of Dip-LbL assembly process to create thin films on a substrate. Adapted from reference 3 (Aiche Journal 57, 2928-2940 (2011)). ........................ 13 Figure 1-2. Cross-sectional scanning electron micrographs of electrospun fibers before (left) and after (right) conformal coatings. Adapted from reference 14 (Nat Mater 8, 5 12-5 18 (2009)) ................................................................................................... .. 14 Figure 1-3. Nerve agents 1) Tabun, 2) Sarin, 3) Soman, 4) VX ................................... 15 Figure 1-4. Competitive Inhibition Mechanism........................................................... 16 Figure 1-5. Schematic of charging and discharging processes in lithium-ion batteries and ECs. Adapted from reference 6 (Energy & Environmental Science 4 (6), 1972-1985 (2 0 1 1))....................................................................................................................... 17 Figure 1-6. Ragone plot showing gravimetric energy and power densities of commercial lithium-ion batteries and ECs, and performance targets for hybrid-electric (HEVs) and plug-in hybrid-electric vehicles (PHEVs). Adapted from reference 16 (Energy & Environmental Science 4 (6), 1972-1985 (2011)) ................................................. 19 Figure 2-1. Schematic representation of the fabrication steps of a chemiresistive sensor based on MWNT-film coated electrospun fiber mat. ........................................... 26 Figure 2-2. Setup for liquid flow sensing measurements. Liquid flow between syringes was directed by controlling the independent syringe pumps................................ 31 Figure 2-3. Illustration of layer-by-layer assembly on electrospun (ES) fiber mat. (a) Parallel-plate electrospinning technique (diagram, left) is employed to create polysulfone ES fiber mats of 5-6 inch diameter (digital image, right). (b) Schematic representation (left) and top-down (center) and cross-sectional (right) scanning electron microscopy (SEM) images of the annealed polysulfone ES fiber mat. (c) A spray-LbL deposition technique with a pressure imposed across the mat is demonstrated to create conformal coatings of (SPS/PDAC)10 thin films on individual fibers. (d) (MWNT-COO~/MWNT-NH 3+)15 films are uniformly coated on ES fibers independent of spray direction, as shown in the cross-sectional SEM image. ......... 33 Figure 2-4. Synthesis scheme of (a) carboxylic acid (MWNT-COO) and (b) amine (MW NT-NH 3+)functionalized MW NTs. .............................................................. 35 Figure 2-5. X-ray photoelectron spectroscopy (XPS) survey spectra of functionalized MWNTs. (a) Wide scan survey, (b) atomic concentration and (c) C1 , spectra of unmodified MWNTs, MWNT-COO and MWNT-NH 3 ..................... ........ ........ . . 37 Figure 2-6. Zeta potential as a function of pH of aqueous dispersions of MWNT-COO and M WN T -NH 3..................................................... .................... ................ ....... . . 38 8 Figure 2-7. Scanning electron microscopy (SEM) images of entire cross-section of an electrospun fiber mat coated with 15 bilayers of (MWNT-COO~/MWNT-NH 3+) films (inset shows higher magnification view).................................................... 39 Figure 2-8. Photographs of an electrospun fiber mat (a) before and (b) after coating with (M WNT-COO/M W NT-NH 3+)20 ........................................ ................................. . . 39 Figure 2-9. (a) Scanning electron microscopy images of ES fibers coated with MWNT films, where n indicates the number of bilayers (2, 5 and 10, respectively) in (MWNT-COO/MWNT-NH 3+)n films (scale bar, 400 nm). (b) Correlation of cumulative weight [mg cm-3] to layer pair number of MWNT films. The straight line is a linear fit. (c) Sheet resistance [ohm square-'] of the MWNT/ES fiber electrodes. 40 ................................................................................................................................... Figure 2-10. Scanning electron microscopy images of an electrospun fiber with 15 bilayers of (a) (MWNT-COO~/MWNT-NH 3+) and (b) (MWNT-COO/MWNTC o o ~)film s.............................................................................................................. 41 Figure 2-11. Electrical conductivity, a, of MWNT/ES fiber electrode as a function of volume fraction of MWNTs, D. Insert: a log-log plot of a versus reduced volume fraction of MWNTs, (eD-Do). The solid line is a fit to a power law (Equation (3)).. 43 Figure 2-12. Covalent attachment of thiourea group to the -NH 2 functional groups on the surface of MWNT-NH 3+ in the MWNT multilayer films. For clarity, a single-walled carbon nanotube segment is shown with only a single functional group at the end of 45 th e tub e ...................................................................................................................... Figure 2-13. X-ray photoelectron survey spectra of ES fiber mats coated with (MWNTCOO-/MWNT-NH 3+)2 films before and after functionalization with thiourea, showing the presence of Ols, N ls, C is, S2s, S2p, and Cl2p. The appearance of fluorine (F 1s) as highlighted in the box indicates the successful attachment of the 47 thiourea functional group to the surface of MWNTs........................................... Figure 2-14. (a) Schematic of the sensor device showing the MWNT/ES fiber electrodes enclosed in a custom-made flow chamber and electrical connectivity for current measurements. (b) I-V characteristics of MWNT/ES fiber electrodes (scan rate, 1 mV s-1). (c) Variation of responses (Al/Jo [%]) of thiourea-functionalized MWNT/ES (TF-MWNT/ES) fiber electrode upon cyclic exposure to aqueous DMMP (10 ppb to 50 ppm) and water. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to water at 'arrow with asterisk'. (d) Comparison of average responses of thiourea-functionalized and unfunctionalized MWNT/ES fiber electrodes as a function of aqueous DMMP concentration. (e) Variation of responses of TF-MWNT/ES fiber electrodes upon cyclic exposure to diluted DMMP vapor (5, 20, 100 ppm) and nitrogen stream. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to nitrogen at 'arrow with asterisk'. (f) Comparison of average responses of thiourea-functionalized and unfunctionalized MWNT/ES fiber electrodes as a function of DMMP vapor concentration.............................. 48 9 Figure 2-15. Signal response (JI/Io [%]) of thiourea-functionalized MWNT film upon cyclic exposure to aqueous DMMP (0.5 ppm) and water. Aqueous DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to water at 'arrow w ith asterisk'. ........................................................................................... 51 Figure 2-16. Variation of responses (u/Io [%]) of unfunctionalized MWNT/ES fiber electrode upon cyclic exposure to aqueous DMMP (10 ppb to 50 ppm) and water. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to w ater at 'arrow w ith asterisk'. ............................................................................. 52 Figure 2-17. Variation of responses of TF-MWNT/ES fiber electrodes upon cyclic exposure to diluted DMMP vapor (5, 20, 100 ppm) and nitrogen stream. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to nitrogen at 'arrow w ith asterisk'. ............................................................................................. 53 Figure 2-18. Average sensing response of unfunctionalized and thiourea-functionalized MWNT/ES fiber electrodes to different analytes. The concentration of each analyte represents 1% of its equilibrium vapor concentration at room temperature except DMMP, which was used at -6% of its equilibrium vapor concentration at room temp erature. .............................................................................................................. 55 Figure 2-19. Signal response (-Il/Io [%]) upon exposure to a pulse of DMMP vapor (20 ppm). DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to N2 at 'arrow with asterisk'. The dots represent the experimental data while the solid lines are the best fit for the DMMP desorption step..................... 57 Figure 3-1. X-ray photoelectron spectroscopy (XPS) survey spectra of functionalized MWNTs. (a) Wide scan survey, (b) atomic concentration and (c) Cis spectra of unmodified MWNTs, MWNT-COO- and MWNT-NH 3+.(d) Schematic representation of functional groups on MWNT-COO~ and MWNT-NH 3.. ...... . . . 71 Figure 3-2. Scanning electron microscopy (SEM) images of VF-MWNT electrodes. (a) Low and (b) high magnification cross-section micrographs. (c) Low and (d) high magnification top-down view micrographs. ......................................................... 73 Figure 3-3. (a) Nitrogen adsorption-desorption isotherm of VF-MWNT electrodes. The insert is the corresponding BJH pore size distribution. (b) Pore size distribution calculated using top-down view SEM image of VF-MWNT electrodes.............. 74 Figure 3-4. Schematic illustration of the fabrication of HP-MWNT electrode from aqueous dispersion of MWNT-COO and MWNT-NH 3+ at room temperature. The resulting MWNT electrode is freestanding and binder-free (bottom-left photo)...... 76 Figure 3-5. (a) Top-down view SEM image of polysulfone electrospun fiber mat. (b) Plot of differential mercury intrusion vs. pore diameter of electrospun fiber mat, obtained using mercury porosim etry. .................................................................................. 76 10 Figure 3-6. Scanning electron microscopy (SEM) images of the HP-MWNT electrodes. (a) Low and (b) high magnification cross-section micrographs. (c) Low and (d) high magnification top-down view micrographs of the conformal side of HP-MWNT electro d e.................................................................................................................... 79 Figure 3-7. Pore size distribution calculated using top-down view SEM image of HP80 M WN T electrode. ................................................................................................ Figure 3-8. Cyclic voltamograms (CVs) at different scan rates of (a) HP-MWNT and (b) VF-MWNT electrodes in 1 M K2 S0 4 at room temperature. (c) Capacitance retention ratio at different scan rates. (d) Nyquist plot of electrochemical impedance spectra m easured in 1 M K 2 S0 4................................................... ......................................... 82 Figure 3-9. Ragone plot of gravimetric energy and power performance of HP-MWNT and VF-MWNT electrodes. The thicknesses of MWNT electrodes were 74 ptm for 84 HP-MW NT and 66 ptm for VF-MW NT............................................................... Figure 3-10. Cyclic voltammograms of HP-MWNT electrodes with different cycling numbers in 1 M K2 S0 4 at room temperature. A scan rate of 20 mV s-1 was used. (b) Capacity retention of HP-MWNT electrodes as a function of cycling number........ 84 Figure 3-11. Potential-dependent cyclic voltammograms (1 mV s1) of HP-MWNT (a) and VF-MWNT (b) electrodes measured in lithium cells. The current was normalized to the mass of MWNT electrodes. The thickness of MWNT electrodes were 66 pm for HP-MWNT and 72 ptm for VF-MWNT. Electrolyte was 1 M LiPF 6 in a mixture of ethylene carbonate and dimethyl carbonate (1:1 v/v). ................. 86 Figure 3-12. Galvanostatic charge-discharge curves of (a) HP-MWNT (66 [tm) and (b) VF-MWNT (72 pm) electrodes in lithium cell at different current rates. The specific capacities were normalized to the weight of MWNT. (c) Gravimetric discharge capacity of HP-MWNT and VF-MWNT electrodes as a function of gravimetric discharge current. (d) Cycling stability of HP-MWNT electrodes tested by chargedischarge cycling tests. MWNT electrodes were charged and discharged at 10 A g 1 between 1.5 V and 4.5 V for 99 cycles, and at every 1 0 0 th cycle, it was charged and discharged at 0.1 A g1. The potential of the MWNT electrodes was maintained at 1.5 V or 4.5 V for 30 minutes to completely discharge or charge the electrode before charging (or discharging) the electrode from 1.5 V (or 4.5 V). The insert shows 87 charge-discharge curves for the 1 0 0 th and 10 0 0th cycles. ..................................... Figure 3-13. A Ragone plot showing electrode-level gravimetric energy and power performance of HP-MWNT and VF-MWNT, calculated from galvanostatic data in Figure 12a-b. Only MWNT electrode weight was considered in the gravimetric energy and power calculations. The gravimetric energy and power of sLbL-MWNT (ref. 10) were calculated from galvanostatic discharge curves normalized by the weight of MWNT and current collector. The performance of electrochemical capacitor and Li-ion batteries is estimated from gravimetric energy and power in the 89 Ragone plot of a previous review paper (ref. 3). .................................................. 11 1. Introduction 1.1 Layer-by-Layer Assembly The layer-by-layer (LbL) assembly technique is a versatile method for creating thin films based on sequential adsorption of complementary multivalent species on a substrate via electrostatic interactions, hydrogen bonding, or other secondary interactions.' The technique is illustrated in Figure 1 below for LbL films that use electrostatic interactions between polymers with positive and negative charges (polyelectrolytes), commonly referred to as polycations and polyanions respectively. In this process, a substrate with an innate charged surface is sequentially dipped into solutions of oppositely charged polymers, which then adsorb to the developing film. The application of polyelectrolytes, or species with more than one charges, is essential to the LbL deposition process because it enables the surface to reverse its charge once the molecules adsorb on the surface, provides self-limiting regulation during the assembly. Once the surface charge is reversed, the surface is ready for the next adsorption step. This approach enables a fine control over the composition and structure of the LbL films by adjusting the assembly conditions (e.g. salt concentration, pH), as well as a variety of components that can be incorporated into the thin films. 2 3 12 substrate polycation solution polyanion solution LbLFilm Figure 1-1. Schematic of Dip-LbL assembly process to create thin films on a substrate. Adapted from reference 3 (Aiche Journal 57, 2928-2940 (2011)). The versatility of the LbL technique has been demonstrated for creating thin films from solutions of various materials, including polyelectrolytes4 , dendrimers , nanoparticles6 , carbon nanotubes 7, and proteins8 , for numerous applications such as drug delivery systems9, energy storage devices'0 , fuel cell membranes", etc. Although dipLbL process is depicted in the figure above, variations of the LbL techniques, including spin-LbL12 and spray-LbL 13 , have been employed to affect the processing time and film characteristics. Recently, the application of LbL technique has been extended by spraying solutions of charged species, such as polyelectrolytes and nanoparticles, onto porous electrospun materials. By imposing a pressure gradient to force flow through the porous substrate during the spray-LbL process, conformal coatings on individual fibers throughout the bulk of the electrospun fiber mat can be achieved as shown in Figure 2.14 Spray-LbL coating of porous non-wovens provides a rapid, scalable approach to introduce new functionalities to materials while retaining their mechanical robustness. 13 This spray-LbL strategy is investigated in this thesis as a versatile tool for creating thin films with multiple functionalities on 3D substrates. m Figure 1-2. Cross-sectional scanning electron micrographs of electrospun fibers before (left) and after (right) conformal coatings. Adapted from reference (Nat Mater 8, 512- 518 (2009)) 1.2 Application to Chemical Detection Chemical weapons can be broadly grouped into three categories; (1) nerve agents, (2) blister agents and (3) choking agents. All nerve agents are organo-phosphorous (OP) compounds. They include phosphorus-containing chemicals such as ethyl N, Ndimethylphosphoroamidocyanidate, commonly known as Tabun (GA), isopropyl methylphosphonofluoridate, commonly Sarin (GB), pinacolyl methyl phosphonofluoridate, commonly Soman (GD), and O-ethyl-S-(2-diisopropylaminoethyl) methyl phosphonothiolate, commonly known as VX. Nerve agents are considerably more toxic than other chemical warfare agents. They may cause pronounced effects within seconds and death within minutes. In this thesis work, emphasis is placed on detection of nerve agent because it is the most lethal chemical weapon, and OP compounds are also 14 commonly used in agriculture as pesticides. The nerve agents are all liquids at room temperature. They can penetrate through ordinary clothing quickly and can be absorbed into the blood stream through the skin, or by inhalation. Figure 3 shows examples of some of the known nerve agents 5 . 0 CH 3CH 2-1) CH0 11/C 3 FP-N I H CN CH 3 CH 3 11 1 CH 3 -P-0-CH I I F 2) CH 3 0 CH 3CH-0-P -S-CH2CH 4) 0 CH 3 11 CH 3-P-O-CH-C-CH CH 3 3) /CH(CH I I F CH I 3 3 CH 3 3)2 2-N CH(CH 3)2 Figure 1-3. Nerve agents 1) Tabun, 2) Sarin, 3) Soman, 4) VX In the nervous system a neurotransmitter known as acetylcholine (ACh) is used to transmit the signal from one neuron to another across the synaptic cleft. Normally, the enzyme acetylcholinesterase (AChE) binds to and hydrolyzes the ACh, which terminates the activity of the ACh molecule at the receptor site, thereby decomposing it into acetate and choline which can then be removed from the system. Upon exposure, the nerve agent acts as a competitive inhibitor and binds to AChE, making it unable to bind with ACh. As a result, ACh is not hydrolyzed. The accumulation of ACh causes hyperactivity of the body organs stimulated by cholineraic neurons, and eventually leads to nervous system failure.15 The competitive inhibition mechanism of AChE is shown in Figure 4. The lethality of nerve agents is caused by the magnitude of the reaction rate constant in the irreversible binding step to AChE. This accumulative property of nerve agents allows them to be toxic at much lower concentrations than other chemical warfare agents. 15 AChE + ACh < E * S -> AChE + acetate + choline AChE + I -> E I Figure 1-4. Competitive Inhibition Mechanism. As the threat of chemical terrorism both at the domestic and international level continues to persist, soldiers and emergency responders are likely to be exposed to such harmful environments. Laboratory-based analytical techniques for detecting toxic chemicals, such as chromatography and spectroscopy, are not suited for in-field use because they require bulky equipment and considerable skills to operate. Wearable sensors that continuously detect toxic chemicals in the surrounding environment in real time can offer advantages over laboratory-based analytical methods and hand-held devices because the wearers can receive timely information without having to operate a complex device or compromising the functionality and discomfort of their garments. The spray-LbL technique of creating conformal thin films on porous substrates allows us to incorporate functional thin films into fabrics to develop engineered textiles that combine chemical sensing capabilities along with lightweight and flexible characteristics. In Chapter 2 of this thesis, the spray-LbL technique is employed to deposit coatings of functionalized carbon nanotube films on electrospun materials to create flexible chemical sensors capable of detecting nerve agents. 1.3 Application to Electrochemical Energy Storage Systems Development of novel energy storage systems has been widely considered as a critical link between sustainable energy supply and demand for applications in 16 microelectronics, electric vehicle, and load leveling. Two promising candidates for these applications are lithium-ion batteries and electrochemical capacitors (ECs). Although batteries and ECs have different fundamental energy storage and conversion mechanisms, they exhibit similar structure and operation. Figure 5 shows schematic of (a) lithium-ion batteries and (b) ECs. Both systems have negative and positive electrodes, which are separated by electrolyte medium. Electrons travel from negative to positive electrode through an external circuit, and corresponding ions migrates through electrolyte medium. Electrochemical reactions (generation and consumption of ions and electrons) only take place at the interface between electrodes and electrolyte. Thus, electrode materials need to have fast transport of ions and electrons within the electrodes to achieve high energy output. + (b) (a) (b Ilpd 4&)0 Positive electrode (LiCoO2) cnarge e -+ ++ Li* Negative electrode (Graphite) Pos jtive electrode Doublelayer capacitence Negative electrode Pseudocapacitance 0~ 0 + Ex. Activated Carbon 0+ 0 00 Figure 1-5. Schematic of charging and discharging processes in lithium-ion batteries and ECs. Adapted from reference 16 (Energy & Environmental Science 4 (6), 1972-1985 (2011)) 17 Electrochemical performance of lithium-ion batteries and ECs can be evaluated by comparing the energy and power of these energy storage systems using a Ragone plot shown in Figure 6. Gravimetric power (W/kg) is plotted versus gravimetric energy (Wh/kg) in logarithmic scales to compare energy capacity and rate capability of different energy storage systems. Li-ion batteries can store high energy density but at relatively low power, while ECs can deliver high power but with limited energy storage. These differences can be understood by examining their energy storage mechanisms. Li-ion batteries store energy chemically in electrode materials through lithium intercalation/deintercalation between two host structures, 16 as depicted in Figure 5a. During the charging process, lithium ions migrate from the positive to negative electrode, where lithium ions intercalate into the graphite sheets. When energy is utilized during discharging, lithium ions are de-intercalated from the graphite sheets, and migrate to the positive electrode, where they are inserted into the cathode. Lithium intercalation process allows Li-ion batteries to have high energy capacity due to high utilization of electrode material, but the rate capability (power) is limited due to slow solid-state diffusion of lithium. In contrast, ECs store charges on the surface of electrode materials through the formation of an electrical double layer capacitance or through fast Faradaic reactions (pseudocapacitance) near the surface, as shown in Figure 5b. These processes are significantly faster than solid-state diffusion of lithium. Thus, ECs have much high power capability, but because only the electrode surface is utilized for charge storage, energy density of ECs is limited. 18 10 goal HE 103 u PHEV goals 40Passenger Car 10 Electrochemical Capacitors 101 E Li-ion Batteries 100.... 10- 100 101 102 103 (Wh/kgvic) Gravimetric Energy Figure 1-6. Ragone plot showing gravimetric energy and power densities of commercial lithium-ion batteries and ECs, and performance targets for hybrid-electric (HEVs) and plug-in hybrid-electric vehicles (PHEVs). Adapted from reference16 (Energy & Environmental Science 4 (6), 1972-1985 (2011)) A major challenge in the field of energy storage devices is to bridge the performance gap between batteries and ECs by developing electrode materials that combine high power and energy. A promising strategy is to employ the Faradaic reactions of surface functional groups on carbon electrodes, which can store more energy than the electrical double-layer capacitance while providing high power capability. Considerable efforts have also been made to investigate the effects of pore structure of electrodes on electrochemical performance with the aim of improving ion transport." Carbon materials with hierarchical porous structure that combines macroporous cores, mesoporous walls, and micropores have been shown to exhibit great potential as electrode materials due to the synergistic benefit of the hierarchical pore texture. The macropores (>50 nm) function as ion-buffering reservoirs to provide decreased ion transport distance to the interior surfaces. The mesoporous walls (2-50 nm) provide fast 19 ion-transport pathways with low resistance. The micropores enhance the electricaldouble-layer capacitance" Thus, numerous efforts have been devoted to develop carbonbased electrodes with tailored porous structure. In Chapter 3 of this thesis, the spray-LbL technique is utilized to create hierarchical porous electrodes based on functionalized multi-walled carbon nanotubes (MWNT) that include pseudocapacitive functional groups. A combination of large micro-size channels that serve as ion-buffering reservoirs and mesoporous network of MWNTs that facilitate fast ion transport enables HP-MWNT electrodes to have excellent energy retention at high power, showing promise as electrode materials for high-rate energy storage applications. 1.4 References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] Decher, G. Science 1997, 277, 1232. Hammond, P. T. Adv. Mater. 2004, 16, 1271. Hammond, P. T. Aiche J2011, 57, 2928. Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213. Tsukruk, V. V.; Rinderspacher, F.; Bliznyuk, V. N. Langmuir 1997, 13, 2171. Kotov, N. A.; Dekany, I.; Fendler, J. H. J.Phys. Chem. 1995, 99, 13065. Lee, S. W.; Kim, B. S.; Chen, S.; Shao-Hom, Y.; Hammond, P. T. J.Am. Chem. Soc. 2009, 131, 671. Lvov, Y.; Ariga, K.; Ichinose, I.; Kunitake, T. J.Am. Chem. Soc. 1995, 117, 6117. Hammond, P. T. A CS Nano 2011, 5, 681. Lee, S. W.; Yabuuchi, N.; Gallant, B. M.; Chen, S.; Kim, B. S.; Hammond, P. T.; Shao-Horn, Y. Nat Nanotechnol 2010, 5, 531. Argun, A. A.; Ashcraft, J. N.; Hammond, P. T. Adv. Mater. 2008, 20, 1539. Jiang, C.; Markutsya, S.; Tsukruk, V. V. Adv. Mater. 2004, 16, 157. Krogman, K. C.; Zacharia, N. S.; Schroeder, S.; Hammond, P. T. Langmuir 2007, 23, 3137. Krogman, K. C.; Lowery, J. L.; Zacharia, N. S.; Rutledge, G. C.; Hammond, P. T. Nat Mater 2009, 8, 512. Lillie, S. H.; School, U. A. C., Ed. 2005. Lee, S. W.; Gallant, B. M.; Byon, H. R.; Hammond, P. T.; Shao-Horn, Y. Energ Environ Sci 2011, 4, 1972. Jiang, H.; Lee, P. S.; Li, C. Z. Energ Environ Sci 2013, 6, 41. Wang, D.-W.; Li, F.; Liu, M.; Lu, G. Q.; Cheng, H.-M. Angewandte Chemie 2008, 120, 379. 20 2. Carbon Nanotube/Electrospun Fiber Electrodes for Flexible Chemiresistive Sensor Applications Abstract Development of a versatile method for incorporating conductive materials into textiles could enable advances in wearable electronics and smart textiles. One area of critical importance is the detection of chemicals in the environment for security and industrial process monitoring. Here, we report the fabrication of a flexible, sensor material based on functionalized multi-walled carbon nanotube (MWNT) films on a porous electrospun fiber mat for real-time detection of a nerve agent simulant. The material is constructed by layer-by-layer (LbL) assembly of MWNTs with opposite charges, creating multilayer films of MWNTs without binder. The vacuum-assisted spray-LbL process enables conformal coatings of nanostructured MWNT films on individual electrospun fibers throughout the bulk of the mat with controlled loading and electrical conductivity. A thiourea-based receptor is covalently attached to the primary amine groups on the MWNT films to enhance the sensing response to dimethyl methylphosphonate (DMMP), a simulant for sarin nerve agent. Chemiresistive sensors based on the engineered textiles display reversible responses and detection limits for DMMP as low as 10 ppb in the aqueous phase and 5 ppm in the vapor phase. This fabrication technique provides a versatile and easily scalable strategy for incorporating conformal MWNT films into 3dimensional substrates for numerous applications. 21 2.1 Introduction Wearable electronics can be defined as portable electronic devices that can be seamlessly incorporated into the user's outfit as part of the clothing or accessoryJ1 Their unique functionalities, including flexibility, lightweight, and stretchability, enable new device designs for many applications previously unfeasible with conventional electronics technology. Examples of these novel applications include wearable displays,[2 a] health monitoring,[2b] energy storage devices, 2c security and environmental monitoring.[2d] One area of critical importance is chemical sensing due to the increased threat of chemical terrorism at the domestic and international level. Chemical warfare agents (CWAs), especially nerve agents such as tabun (GA), sarin (GB), soman (GD), and VX, are of great concern due to their acute toxicity at low concentration and colorless characteristics. Soldiers and emergency responders are particularly susceptible such harmful environments. Laboratory-based analytical techniques, such as chromatography and spectroscopy, are not suited for in-field use because they require bulky equipment and considerable skills to operate. Wearable sensors that continuously detect toxic chemicals in the surrounding environment in real time can offer advantages over laboratory-based analytical methods and hand-held devices because the wearers can receive timely information without having to operate a complex device or compromising the functionality and discomfort of their garments. To achieve high-performance wearable sensor with good sensitivity and selectivity, the sensing device need to have the following characteristics: simple architecture to remain lightweight and portable, high surface area for enhanced interaction, and chemical functionality to achieve selective response. 22 Carbon nanotube (CNT) resistance-based chemical sensors have been extensively investigated in recent years, and they are recognized as highly sensitive chemical sensors.E31 A typical CNT-based chemiresistor consists of one or several pairs of electrodes that make electrical contact with a thin film or network of CNTs. A change in the electrical resistance of the CNT film upon exposure to an analyte is measured as the output signal. The advantages of chemiresistors include minimal electronic components, low power consumption, and simple operation. E3 These features enable miniaturization of the sensing unit and unobtrusive integration into a lightweight, wearable system. There are three general methods for fabricating CNT-based resistive sensors: (1) directly growing CNTs using vapor deposition, (2) transferring a CNT array or CNT thin film, and (3) solution-depositing CNT suspension, such as drop-cast, spin-coat, and ink-jet, on a rigid (glass or silicon wafer) or flexible plastic substrate. 4 3 Although these fabrication techniques can generate CNT network with various geometries and structures, they do not allow for uniform, scalable deposition of CNT film on a porous, textile material, which is an ideal substrate for wearable electronics due to its flexible and breathable characteristics. Fabrics made of electrospun (ES) polymeric fibers provide flexible, porous scaffolds with high surface area, which enhances the specific surface area of the sensing material. The utilization of ES fiber mats as sensor substrates has been shown by Wang et al.E 5a] and Bhattacharyya et al.[5bl for fabrication of polymer-based biosensors to improve sensitivity and response time due to enhanced surface area. These studies suggest that an electrospun fiber mat is an ideal candidate as a non-conducting substrate for fabrication of a chemiresistive sensor and also serves as a platform for a wearable electronics system. 23 The layer-by-layer (LbL) assembly technique is a versatile method for creating uniform thin films through sequential adsorption of complementary multivalent species on a substrate via electrostatic interactions, hydrogen bonding, or other secondary interactions.E63 A substrate with an innate charged surface is successively dipped into solutions of oppositely charged species, which adsorb to the developing film. This approach enables a fine control over the composition and structure of the LbL films by adjusting the assembly conditions (e.g. salt concentration, pH), as well as a diverse choices of components that can be incorporated into the thin films.93 In our previous work, we have extended this LbL technique by spraying solutions of charged species, such as polyelectrolytes and nanoparticles, onto a porous ES fiber mat in conjunction with a pressure gradient across the substrate to deposit uniform coatings composed of polyelectrolytes and nanoparticles on individual fibers throughout the bulk of the ES fiber substrate.E8 1 Spray-LbL coating of porous non-wovens provides a rapid, scalable approach to introduce new functionalities to materials while retaining their mechanical robustness. In addition, we have demonstrated the ability to construct multi-walled carbon nanotube (MWNT) multilayer films without binder on planar substrates and carbon paper using conventional dip-LbLE91 and spray-LbLE'03, respectively, to create electrodes for energy storage applications. These studies suggest that the LbL technique is well suited for depositing conformal MWNT films on porous ES fiber substrates. Recently, dip-LbL assembly in which conductive carbon-based materials, such as CNTs and graphene, have been alternated with polymers has been utilized to deposit CNTs and graphene on electrospun fiber mats. 12 1 However, the LbL films in these studies contain polymer as part of the multilayers, which functions as a binder to 24 facilitate the assembly process. Spray-LbL assembly of binder-free CNT films provides a better and more rapid approach to obtain nanostructured thin films composed of only CNTs. The resulting systems provide access to the surfaces of the functionalized carbon nanotubes, which is key to achieving high levels of active surface area as binding sites for sensing. An added advantage to Spray-LbL is the ability to control the coating of fine pore within the electrospun mat without introducing bridging across pores, and thus limiting accessible surface area with increased film thickness. In this work, we present the design and fabrication of flexible, real-time chemiresistive sensors based on conformal LbL all-MWNT films on ES fiber mats. The sensory material is composed of two high-surface-area materials: functionalized MWNTs as a sensing component and an ES fiber mat as a flexible, porous substrate. Figure 2-1 depicts a schematic representation of the steps involved in the fabrication of this chemiresistive sensor. Here we demonstrate the facility of the vacuum-assisted, spray-LbL technique to generate conformal, binder-free MWNT films on ES fibers to yield a MWNT/ES fiber electrode with electrical conductivity that can be varied with MWNT loading. The available functional groups on the MWNT films are used to covalently attach a thiourea-based receptor to the MWNT/ES fiber electrode to enhance sensitivity, and the thioureafunctionalized composite electrode is employed as a chemiresistive sensor for detecting dimethyl methylphosphonate (DMMP), a sarin nerve agent simulant, in both aqueous and gas phases. We believe this fabrication approach is a versatile, scalable method to generate flexible, high-surface-area CNT devices for applications ranging all the way from wearable sensors to energy storage devices. 25 Real-time amperometric measurement Electrospun MWNT coated Thiourea attached (ES) fiber mat ES fiber mat ES fiber mat * spray-LbL of MWNT films Electrodes Thiourea Functionalization with thiourea-based receptor Figure 2-1. Schematic representation of the fabrication steps of a chemiresistive sensor based on MWNT-film coated electrospun fiber mat. 2.2 Experimental Section Materials: All chemicals, unless stated otherwise, were purchased from Sigma Aldrich (St. Louis, MO) and used without further purification. pH adjustment of all solutions was done by addition of dilute hydrochloric acid or sodium hydroxide. Poly(sodium 4styrenesulfonate) (SPS, Mw = 70,000) and poly(diallyldimethylammonium chloride) (PDAC, Mw = 150,000, 20% aqueous solution) were prepared as 0.01 M solutions, on the basis of the repeat-unit molecular weight, in deionized water (Milli-Q, 18.2 M9e cm) and titrated to pH 4. Branched polyethylenimine (BPEI, Mw = 75,000) was purchased from PolySciences (Warrington, PA). Pristine MWNTs (95% purity, length 1-5 pm, outer diameter 15 ± 5 nm) synthesized by chemical vapor deposition were purchased from NanoLab (Newton, MA). Preparationof electrospunfiber materials: The electrospinning apparatus used is similar to previous report [13]; briefly, it consists of two aluminum disks 10 cm in diameter oriented parallel to each other and separated by a distance of 30 cm. A 25 wt.% solution 26 of bisphenol A polysulfone (UDEL P3500, Solvay Advanced Polymers) in 1:1 N,Ndimethyl formamide (DMF):N-methylpyrrolidone (NMP) is pumped through a Teflon tube with a syringe pump (Harvard Apparatus PHD 2000) at a rate of 0.010 mL/min through a 0.040" ID needle in the top aluminum disk. A power supply provides 18.0 kV potential to the upper aluminum disk and the polymer solution is drawn to the bottom grounded disk where fibers of 1.69 ± 0.17 Rm in diameter are collected. The thickness of the mat can be controlled by the time allowed for deposition (2 hours for ca. 100 tm thick mat), after which the sample is annealed in an oven at 150 'C for 2 hours in order to remove any residual solvent as well as to improve the strength of the electrospun (ES) fiber mat. Mean and standard deviation of fiber diameter were determined from measurements, using ImageJ [13], of 100 fibers selected manually and at random from an SEM micrograph with a magnification of 5,OOOX. The thickness of the ES mat was measured by a digital micrometer (Mitutoyo CLM1) with a constant measuring force of 0.5 N. Prior to LbL deposition, 2.5-inch diameter disk of polysulfone ES mat was cut and treated with one minute of air plasma (Harrick PDC-32 G plasma cleaner) to improve the wettability by introducing charged carboxylated groups on the surface. The mat was then immediately immersed into a 0.2 wt.% aqueous solution of BPEI for 10 min to impart positive charge on the surface of ES fibers. Functionalization of MWNTs: MWNTs modified with negatively charged carboxylate ions (MWNT-COO) were prepared by an oxidation treatment according to a published protocol [9]. MWNTs were refluxed in strong acids (3/1 v/v, 95% H2 SO 4 , 70% HNO 3) at 70'C for 2 hrs, and then washed several times with deionized water by filtration using a polypropylene carbonate membrane (Whatman, pore size = 50nm). The MWNT-COO~ 27 powder was dispersed in deionized water to attain 0.5 mg mlV suspension. The resulting suspension was dialyzed (MWCO 6000-8000) for a few days to remove any residuals and byproducts from functionalization. MWNTs modified with positively charged ammonium ions (MWNT-NH 3+) were prepared as previously reported by reacting MWNT-COO~ suspension (400 mL, 0.5 mg ml 1 ) with excess ethylenediamine (40 mL) in the presence of 4 g of 1-[3-(Dimethylamino)propyl]-3-ethylcarbodiimide methiodide [10]. The resulting suspension was also dialyzed for a few days to remove any residuals and byproducts. The pHs of MWNT suspensions were adjusted after synthesis and sonicated (Branson Bransonic 2510 ultrasonic cleaner) briefly prior to LbL assembly. LbL film assembly: Conformal (MWNT-COO/MWNT-NH 3+)n films were coated on ES fibers using a vacuum-assisted spray-LbL technique with a modification of the reported protocol [8a]. Typically, a 2.5-inch diameter disk of polysulfone ES mat was placed on a stainless steel mesh (56 mesh, 0.004", TWP Inc.) and fixed inside a modified filter holder (Pall Life Science) by a cap. The back of the filter holder was connected via rubber tubing to a vacuum source to create airflow through the ES mat during spray deposition. All solutions were atomized into spray and delivered by a home-built automated spray system using high purity N2 (20 psi). A base film of (SPS/PDAC) 10 was deposited on the ES mat with the following steps. SPS solution (0.01 M, pH 4) was sprayed for 4 seconds at a rate of 0.3 mL s and drained for 10 seconds under vacuum. Subsequently, pH 4 rinse water was sprayed for 4 seconds and drained for 10 seconds under vacuum. This half cycle was repeated for PDAC solution (0.01 M, pH 4) to complete one bilayer of (SPS/PDAC) film. The total cycle (56 s) was repeated 10 times to complete the base layers. (MWNT-COO7/MWNT-NH 3+)n films were deposited on top of the (SPS/PDAC)io 28 base layers by repeating the above procedure with an increased drain time of 90 s, resulting in about a 4 min process cycle to complete one layer of (MWNTCOO~/MWNT-NH 3+) film. The total number of cycles, n, was adjusted to achieve a target number of film layers. Characterization:The surface chemistry of functionalized MWNTs and MWNT filmcoated ES fibers was analyzed using a Kratos AXIS Ultra Imaging X-ray photoelectron spectrometer (XPS). Zeta potential of MWNT aqueous suspension was measured using a Zeta PALS instrument (Brookhaven Instrument Corp.). MWNT mass loading was measured by weighing the ES mat before and after coating with MWNT films using a micro-balance (Sartorius M2P). Sheet resistance of MWNT coated ES mats was measured using a four-point probe method (Keithley SCS-4200). Four measurements were taken on each film, and the measurements were averaged to give the final reported value with the standard deviation. Morphologies of the surfaces and the cryo-fractured cross-sections of LbL films were examined using a scanning electron microscope (JEOL 6320FV Field-Emission High-Resolution SEM) operating at 2 and 5 kV. Functionalization of MWNT/ES fiber electrode with thiourea-based receptor: The MWNT-film coated ES mat was cut into rectangles (2 x 10 mm). Subsequently, the rectangles were immersed separately in 1-isothiocyanato-3,5-bis(trifluoromethyl)benzene solution (1 mL of 0.09 M isothiocyanate solution in ethanol). After 4 days, 1 mL of isocyanate solution was added (0.09 M in ethanol). After 6 additional hours, the liquid was removed and the film was washed six times by immersion in ethanol (2.5 mL each), followed by immersion in ethanol (three times, 15 mL ethanol each) and sonication for 3 times 5 sec in 1.5 mL ethanol. The last washing solution was tested for isocyanate via 29 thin layer chromatography (TLC), showing no residual reagent or other UV-active impurities. The functionalized mats were air dried overnight and then heat-treated at 150'C for 12 h in vacuum to improve film stability. Procedureforsensing measurements: The MWNT/ES fiber electrodes (2 x 10 mm) were placed inside a homemade flow chamber. The electrodes made electrical contacts with platinum wires (5 mm apart) to connections on the outside of the chamber. Two ports on opposite ends of the chamber allowed continuous flow of liquid or gas through the chamber. Up to 9 samples could be tested simultaneously. Two samples were used for each measurement. For aqueous sensing measurements, the inlet was attached to a tubing system with a Y connector attached to two syringes (Figure 2-2). The two syringes were controlled using two separate syringe pumps (Harvard Apparatus, Holliston, MA) with a constant flow rate of 0.5 ml min-1 . In a typical aqueous sensing measurement, the MWNT/ES fiber electrodes were first exposed to MilliQ water. The flow was then switched from water to diluted DMMP solution for 100 s and changed back to MilliQ water for 700 s. In each test, the MWNT/ES fiber electrodes were exposed to at least three pulses of DMMP solution. 30 Syringe 1: MiIHQ water Dvice enclosure Syringe 2: DMMP in MiIQ water Figure 2-2. Setup for liquid flow sensing measurements. Liquid flow between syringes was directed by controlling the independent syringe pumps. For gas sensing measurements, the chamber inlet was connected a KIN-TEK gas generator system where a DMMP/nitrogen gas mixture was generated. A trace amount of DMMP vapor produced in a heated chamber at 70*C was mixed with a nitrogen stream to create an oven flow, which was further diluted with another nitrogen stream (dilution flow) to achieve a target concentration. In a typical gas sensing measurement, the MWNT/ES fiber electrodes were first exposed to dilution flow. Subsequently, a mixture of oven and dilution flows was run through the device for 30 s, followed by dilution flow for 400 s. Electrochemical measurements were conducted using a PalmSens potentiostat connected to a laptop computer and controlled through PSTrace software. Electrical current was monitored in real time at an applied voltage of 0.05 V. 31 2.3 Results and Discussion 2.1. Fabrication and Characterization of MWNT-Film-Coated Electrospun Fiber Mats Figure 2-3 illustrates the steps involved in the fabrication of MWNT film-coated electrospun (ES) fiber mats. Parallel-plate electrospinning apparatus (Figure 1 a) was used to create flexible nonwoven mats of polysulfone fibers (average fiber diameter of 1.69 ± 0.17 pim and mat thickness = 103 ± 2 jim, Figure lb) from organic polymer solutions using a protocol modified from a previous report.E1 3 Polysulfone was chosen as a polymer substrate because it has high chemical resistance and good thermal and mechanical stability. 143 The nonwoven mats have an inherently high porosity of 85.1 ± 1.1%, which was determined by a gravimetric method described elsewhere. 1 3 Once the ES fiber mat was generated, it was first treated with one minute of air plasma to improve the wettability by introducing hydroxyl and charged carboxylated groups on the surface. 15 3 The ES fiber mat was then immediately immersed into a 0.2 wt.% aqueous solution of branched polyethylenimine (BPEI) for 10 min to impart positive charge on the surface of ES fibers. Ten bilayers of poly(sodium 4-styrene-sulfonate) (SPS) and poly(diallyldimethylammonium chloride) (PDAC) film were deposited on the ES fiber mat as base layers to facilitate a more effective electrostatic assembly of MWNT films by creating uniformly charged surfaces on the ES fibers. 32 (a) electrospun fiber (b) electrospun polysulfone fiber mat (C) anionic SPS pressure + cationic PDAC (d) MWNT-COO~ pressure >77w ~ + MWNT-NH 3 1 1 pm. pm Figure 2-3. Illustration of layer-by-layer assembly on electrospun (ES) fiber mat. (a) Parallel-plate electrospinning technique (diagram, left) is employed to create polysulfone ES fiber mats of 5-6 inch diameter (digital image, right). (b) Schematic representation (left) and top-down (center) and cross-sectional (right) scanning electron microscopy 33 (SEM) images of the annealed polysulfone ES fiber mat. (c) A spray-LbL deposition technique with a pressure imposed across the mat is demonstrated to create conformal coatings of (SPS/PDAC)10 thin films on individual fibers. (d) (MWNT-COO~/MWNTNH 3+)15 films are uniformly coated on ES fibers independent of spray direction, as shown in the cross-sectional SEM image. To achieve conformal coating on the ES fiber mat, we employed the vacuumassisted Spray-LbL technique described in previous work,[8 a] which forces flow throughout the porous membrane. By selecting the flow rate that yields Red (Equation (1)) below the critical value of 6, where Red is the Reynolds number based on a cylinder diameter for flow separation from the downstream side of the cylinder,E163uniform films can be created on the ES fibers. At Red> 6, LbL film growth has been observed only near the stagnation point on the front of the ES fibers. 8 a3 The flow rate was adjusted in this work to give an Red = 1.4, which allowed conformal films to be developed on the ES fibers. Ten base layers of (SPS/PDAC) were deposited by alternately spraying aqueous solutions of 10 mM SPS and PDAC (pH 4) onto the ES substrate to coat the ES fibers uniformly with (SPS/PDAC) 10 films with a total thickness of ca. 25 nm (Figure Ic). Red = (D V)/[(1- &)v] (1) where D is the diameter of ES fibers, V is the superficial fluid velocity, F = 0.85 is the void fraction of the ES fiber mat and v = 15.7 x 10~6 m2 s-I is the kinematic viscosity of air at 300 K.[8 a] 34 To create electrostatic LbL films of MWNTs, negatively and positively charged MWNT suspensions were prepared by chemical functionalization of the MWNT surfaces based on previous reports (Figure 2-4).9'14] In brief, carboxylic acid groups were created at defect sites and ends of unmodified MWNTs through acid oxidation to impart aqueous solubility and anionic surface character (MWNT-COO~). Positively charged MWNTs with free amine groups were prepared by introducing amine groups through a carbodiimide-mediated coupling reaction of MWNT-COO~ and ethylenediamine to form MWNT-NH 3+. (a) MWNT-COO~ Q OH o OH C 1) H 2S0 4/HNO 3 2) Filtration/dialysis H2 C OH C NH NH2 (b) MWNT-NH 3 + 7 0H O /CH /H -- 2 7 CH 2 H2 C /NH NC 2) NH 2(CH 2)2NH 2 3) Dialysis Figure 2-4. Synthesis scheme of (a) carboxylic acid (MWNT-COO~) and (b) amine (MWNT-NH 3+) functionalized MWNTs. 35 X-ray photoelectron spectroscopy (XPS) elemental analysis was performed to verify the surface functional groups on the modified MWNTs (Figure 2-5a-b). The XPS analysis of the C1 , spectra for the unmodified MWNTs, MWNT-COO, and MWNTNH3 + shows detailed surface functional groups on the MWNTs (Figure 2-5c). The unmodified MWNTs consisted of primarily sp 2 -hybridized carbons centered at 284.5 eV (in red) and some sp 3 -hybridized carbons centered at 285.2 eV (in magenta). Additional peaks at higher energies indicated small traces of C-O (286.2 ± 0.1 eV) and carbonyls C=O (286.8 ± 0.2 eV).E91 These oxygen-containing groups on the unmodified MWNTs were attributed to atmospheric oxidation and residual oxides resulting from the MWNT synthesis process.E 1] The C i peak of MWNT-COO shows two additional higher binding energies that were attributed to hydroxyls C-OH (287.5 ± 0.1 eV) and carboxyls O=C-H (288.9 ± 0.2 eV).E91 The intensities of these peaks increased relative to the sp 2 -hybridized carbons due to the oxidation process. The XPS analysis of the Cl s spectrum of MVNTNH 3 + indicated the presence of amide bonds N-C=O (287.9 eV) and amine C-N (286.2 eV),E9'1]1 which suggested the successful formation of amide bonds and introduction of amine groups. 36 (a) (C) Cls Unmodified MWNTs C is 2 - Sp3 -C Sp -C - 01S Cc=O -0 c-0 C=O MWNT-NH3+ C C MWNT-COO- MWNT-COO- Unmodified MWNT 600 400 500 300 Binding energy [eV] 2C0 M C-O H C COOH C -0 OH; ~ (b) MWNT-NH3* Atomic Ratio via XPS C 0 N Unmodified MWNTs 0.99 0.01 0.00 MWNT-COOMWNT-NH 3 H= C-N -CN C=O: 0.89 0.11 0.00 0.86 0.07 0.06 CONH 292 290 284 288 286 Binding energy [eV] 282 Figure 2-5. X-ray photoelectron spectroscopy (XPS) survey spectra of functionalized MWNTs. (a) Wide scan survey, (b) atomic concentration and (c) Ci, spectra of unmodified MWNTs, MWNT-COO~ and MWNT-NH 3+. The colloidal stability of aqueous dispersions of MWNT-COO and MWNT-NH 3+ was determined using zeta potential analysis. Both functionalized MWNT suspensions are sufficiently stable across the studied pH range from 3 to 8, with zeta potential values of -43.3 mV (MWNT-COO) at pH 4 and +50.1 mV (MWNT-NH 3+) at pH 4 (Figure 26). With these stable MWNT suspensions, (MWNT-COO7/MWNT-NH 3+)n films, where n denotes the number of film bilayers, were deposited on top of the (SPS/PDAC)10 base layer by sequentially spraying MWNT-COO and MWNT-NH 3+ suspensions onto the ES 37 fiber mat. Figure Id shows top-down and cross-sectional SEM images of ES fiber mats coated with (MWNT-COO~/MWNT-NH 3+)1 5 . The SEM images show that the LbLMWNT films were conformally coated on individual fibers with uniform cross-section, showing no preference toward the direction of fluid flow. .60 -60 E -50- 50 -40- 40 0 aa) -30- + MWNT-COO- 30 + MWNT-NH.' -20 3 4 5 pH 6 7 8 '20 Figure 2-6. Zeta potential as a function of pH of aqueous dispersions of MWNT-COO~ and MWNT-NH 3+. Furthermore, SEM images of a cross-section of a coated ES fiber mat (Figure 27) demonstrate that the convective flow of MWNT suspension penetrated throughout the depth of the 103 + 2 ptm thick ES fiber mat, enabling uniform growth of MWNT films from the front to the back side of the substrate. The ES fiber mat also retained its flexibility after coating with the MWNT films (Figure 2-8). 38 Figure 2-7. Scanning electron microscopy (SEM) images of entire cross-section of an electrospun fiber mat coated with 15 bilayers of (MWNT-COO7/MWNT-NH 3 ) films (inset shows higher magnification view). I Figure 2-8. Photographs of an electrospun fiber mat (a) before and (b) after coating with (MWNT-COO/MWNT-NH 3+)20. 39 Figure 2-9a shows a series of SEM images consisting of an ES fiber coated with 2, 5 and 10 bilayers of (MWNT-COO-/MWNT-NH 3'), respectively. The SEM images clearly show that the MWNT loading increases with the number of deposited layers, gradually covering the ES fiber until the underlying fiber is not visible, as demonstrated for the films with 10 bilayers. (a) [ (b) (c) 40- 106 .rU) C> E E 30- 20- 105- 104- i 10 ion-. z 10 I* ------------------........ r 5 0 xz 10 15 20 0 Number of Bilayers 10 20 30 40 50 Number of Bilayers Figure 2-9. (a) Scanning electron microscopy images of ES fibers coated with MWNT films, where n indicates the number of bilayers (2, 5 and 10, respectively) in (MWNTCOO7/MWNT-NH 3+)n films (scale bar, 400 nm). (b) Correlation of cumulative weight [mg cm~3] to layer pair number of MWNT films. The straight line is a linear fit. (c) Sheet resistance [ohm square'] of the MWNT/ES fiber electrodes. 40 The buildup of MWNT films on ES fibers was not observed when only a MWNTCOO~ suspension was used (Figure 2-10), suggesting that an electrostatic cross-linking between positively and negatively charged MWNTs is the major driving force for the multilayer assembly process rather than simple film accumulation. Figure 2-10. Scanning electron microscopy images of an electrospun fiber with 15 bilayers of (a) (MWNT-COO7/MWNT-NH 3+) and (b) (MWNT-COO/MWNT-COO) films. The porous network structure of the MWNT films also shows that the electrostatic repulsion between well-dispersed nanotubes dominated the attractive Van der Waals forces, resulting in randomly oriented arrangements of unbundled MWNTs. Wellexposed tubes in the MWNT films increase the effective surface area available for interactions with surrounding molecules compared to more traditional methods of assembling nanotubes, making the MWNT coated ES fiber mat highly attractive as a CNT based electrode for sensor applications. Figure 2-9b displays the cumulative weight of (MWNT-COO /MWNT-NH 3+) films on the ES fiber mat as a function of the number 41 of bilayers. As expected for film thicknesses that are much less than the fiber diameter, a linear growth profile is observed, consistent with our previous observation for LbLassembled films of MWNTs on a flat Si wafer.E93 The sheet resistance of the MWNT/ES fiber electrodes was measured by four-point probe and presented as a function of the number of bilayers (Figure 2-9c). The sheet resistance decreases with increasing number of bilayers, indicating that added layers of MWNT films enhance the conductivity of the composite material. The MWNT/ES fiber electrode underwent a transition from an electrically nonconductive to a conductive thin film after just 1 bilayer of (MWNTCOO7/MWNT-NH 3 ) film was deposited, suggesting that an interconnected network of MWNTs well above the percolation threshold was formed. The control over MWNT loading and conductivity of the MWNT/ES fiber electrodes demonstrate their potential use in energy and sensor devices. The electrical conductivity ((T) of the MWNT/ES fiber electrodes was calculated from the sheet resistance according to Equation (2) and presented as a function of the volume fraction of MWNTs (D=VMWNT/Vmat) in the MWNT/ES fiber electrode (Figure 211). The volume of MWNTs (VMWNT) was derived using the total weight of the composite electrode that is due to deposited MWNT films and the density of freestanding LbL-assembled MWNT films of 0.83 g cm 3 reported by Lee et al.J9 The apparent volume of the polysulfone ES fiber mat (Vmat) was determined by measuring dimensions of the mat specimen. a= i(RsT) (2) where R, is sheet resistance, and T is thickness of MVWT/ES fiber electrode. 42 1 I I * I * I 10-1 E . 10--2 -10-1 E 10-2 UI 0 10-3 10-2 10-1 10-4 - 0 0.04 0.03 Volume fraction of MWNTs (0) 0.01 0.02 0.05 Figure 2-11. Electrical conductivity, cy, of MWNT/ES fiber electrode as a function of volume fraction of MWNTs, D. Insert: a log-log plot of Y versus reduced volume fraction of MWNTs, (F-(l,). The solid line is a fit to a power law (Equation (3)). The conductivity value of 0.7 ± 0.03 S cm-1 for the MWNT/ES fiber electrode with 50 bilayers of LbL-MWNT films is of the same order of magnitude as the value of approximately 1 S cm-1 reported by Lee et al. for free-standing LbL-assembled MWNT films.193 The conductivity of the MWNT/ES fiber electrode increases dramatically after the percolation threshold (ID = 0.0022 at 1 bilayer) is reached and approaches a plateau with increasing MWNT loading. This behavior closely resembles the electrical conductivity of polymers reinforced with randomly distributed, electrically conductive 43 particles. According to the percolation theory, a power law can describe the relationship between electrical conductivity of a composite material and conductive filler concentration as shown in Equation (3). T a (-(D)' (3) where (D is the volume fraction of conductive filler, FD is the volume fraction at the percolation threshold, and / is the critical exponent, which reflects the dimensionality of the system. Theoretical values for P are around 1.33 and 2.0 for two and three17 dimensional systems, respectively.E 1 The log-log plot (Figure 2-11 insert) of electrical conductivity versus reduced volume fraction of MWNTs, (0-0c), shows that the electrical conductivity of MWNT composite agrees well with the percolation behavior described by Equation (3). The best fit to experimental values resulted in / = 1.5. The low / suggests that the conductivity of the MWNT/ES fiber electrode behaves similarly 8 to that of a two-dimensional lattice of a CNT film (network) on a flat substrate,f[1 consistent with the relatively thin nature of the films, ranging from approximately 15 to 100 nm in thickness. The percolation threshold (Dc) of approximately 0.22 vol% falls within the range of the values reported in literature for MWNT-epoxy composites (0.1 to 3.8 vol%). 19 ] The aspect ratio and dispersion state of CNTs have a significant impact on 20 the percolation threshold, resulting in the wide range of reported values.E 1Balberg et al. used the excluded volume concept to describe a statistical distribution of conductive filler sticks shaped as capped cylinders and correlated the relationship between Dc and aspect ratio (ii) as (Dc a j-2 for r > 15.[2 This dependence explains why the observed percolation threshold ((Dc = 0.22 vol%, r ~ 60 for MWNTs used in this work) is higher than that reported for other CNT-epoxy composites with higher aspect ratio CNTs (e.g., Dc = 44 0.0025 wt% for q ~ 340),22a] but is still relatively low when compared to spherical particles ((D,= 1 6 %),[22b] such as carbon black, as anticipated. 2.2. Functionalization of Thiourea-based Receptor to the MWNT/ES Fiber Electrode The LbL-MWNT films deposited on the MWNT/ES fiber electrode are composed of carboxylic acid (MWNT-COO) and amine (MWNT-NH 3+) functionalized MWNTs. These functional groups on the MWNT surfaces can be utilized for coupling other molecules directly to the surface of the MWNT/ES fiber electrode to tailor its functionality and properties. Here, the available -NH 2 functional groups on the surface of MWNT-NH 3+ were reacted with an electron poor isothiocyanate derivate to covalently attach a thiourea-based receptor on the MWNT/ES fiber electrode (Figure 2-12). 0 NCS F 3C CF3 \ 0 N--,,H2 2 TEthanol 2 days, RT H H N-N YN HNNH S CF 3 F3C Figure 2-12. Covalent attachment of thiourea group to the -NH 2 functional groups on the surface of MWNT-NH 3+in the MWNT multilayer films. For clarity, a single-walled carbon nanotube segment is shown with only a single functional group at the end of the tube. 45 The presence of the thiourea group is intended to promote strong interactions between the MWNT/ES fiber electrode and DMMP through the hydrogen-bonding affinity of thiourea (pKA= 21.1 in DMSO)[23' toward the phosphate ester in DMMP, which is common to G-type nerve agents. Specifically, the phosphate ester acts as a strong hydrogen-bond acceptor that binds to the acidic thiourea protons. 24 1The electronwithdrawing group (3,5-bis(trifluoromethyl)phenyl) is expected to increase the acidity of the thiourea protons and thus further strengthens the interactions of DMMP with the thiourea moiety. Figure 2-13 shows the X-ray photoelectron spectroscopy (XPS) survey spectra of an ES fiber mat coated with 2 bilayers of (MWNT-COO/MWNT-NH 3+) films before and after functionalization with the thiourea unit. The XPS survey scans show the presence of C, 0, and N, which are characteristic of the carboxyl and amine groups of MWNT-COO and MWNT-NH 3+, respectively. Sulfur peaks (S2s and S2p) can be assigned largely to the polysulfone electrospun fibers. Chlorine was present as expected from the chloride counterions in the LbL-MWNT films. Appearance of a Fis peak (atomic concentration ~1%) after the functionalization suggests that the thiourea-based receptor, which contains a unique fluorine element, was successfully incorporated onto the MWNT/ES fiber electrode. Functional group density based on FIs and CIs signals by XPS was estimated to be 1 thiourea unit per 540 CNT carbons which approximately corresponded to 1 thiourea unit per 68 CNT carbons on the outer wall. 46 CIs O1s Fis Functionaized with thiourea to 0 NIs Unfunctionalized US 900 800 700 600 500 400 Binding energy [eV] 300 U 200 Figure 2-13. X-ray photoelectron survey spectra of ES fiber mats coated with (MWNTCOO7/MWNT-NH ) 2 films before and after functionalization with thiourea, showing the presence of Ols, NIs, C1s, S2s, S2p, and Cl2p. The appearance of fluorine (F 1s) as highlighted in the box indicates the successful attachment of the thiourea functional group to the surface of MWNTs. 2.3. Response of Chemiresistive Sensor Based on MWNT/ES Fiber Electrode to DMMP The MWNT/ES fiber electrodes (2 x 10 mm) were integrated into a custom-made sensor enclosure (Figure 2-14a), for which the electrodes were electrically contacted with two platinum wires that were connected to a potentiostat. 47 (a) (b) Analyte inlet/outlet 10- - 5 0Platinum wires 0 -5 0 H3CO" o 10 bilayers a 5 bilayers o 2 bilayers -10 WNTIES fiber electrode PCH 3 CH3 DMMP . 30 -20 Potentiostat (d) (c) 1.5 50 ppb 1.2- -lppb 0.9- aqueous DMMP Functionalized 1.51.2 -10 0 10 Voltage [mV) Unfunctionalized - 3C 20 aqueous DMMP ------ - 0.9 0.6 0.31 0.0 vt 0 500 1500 1000 Time [s] (e) 1.0 0.6- .' 0.3. 0 .0 '3 - 0.0 0.01 2000 0R lOppni 10 20 ppm -5ppm DMMP gas 0.8- 0.6- (1) I1.. 0.8 0.05 0.5 5 DMMP Concentration [ppm] - Functionalized -A- - -- --50 DMMP gas Unfunctionalized 0.6 0.4 0.4 0.2 0.2 0.0 0.01 500 1000 Time [s] 1500 5 10 20 50 DMMP Concentration [ppm] 100 Figure 2-14. (a) Schematic of the sensor device showing the MWNT/ES fiber electrodes enclosed in a custom-made flow chamber and electrical connectivity for current measurements. (b) I-V characteristics of MVNT/ES fiber electrodes (scan rate, 1 mV s-1). (c) Variation of responses (AI/o [%]) of thiourea-functionalized MWNT/ES (TFMWNT/ES) fiber electrode upon cyclic exposure to aqueous DMMP (10 ppb to 50 ppm) and water. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to water at 'arrow with asterisk'. (d) Comparison of average responses of 48 thiourea-functionalized and unfunctionalized MWNT/ES fiber electrodes as a function of aqueous DMMP concentration. (e) Variation of responses of TF-MWNT/ES fiber electrodes upon cyclic exposure to diluted DMMP vapor (5, 20, 100 ppm) and nitrogen stream. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to nitrogen at 'arrow with asterisk'. (f) Comparison of average responses of thioureafunctionalized and unfunctionalized MWNT/ES fiber electrodes as a function of DMMP vapor concentration. Figure 2-14b shows the current-voltage (I-V) curves of the MWNT/ES fiber electrodes that were coated with 2, 5 and 10 bilayers of (MWNT-COO7/MWNT-NH 3+) films. All MWNT/ES fiber electrodes exhibited ohmic behavior from their I-V characteristics. Although not measured, the contact resistance with the sensor enclosure was assumed to be small compared to the resistance of MWNT/ES fiber electrodes (0.20.5 Mn). MWNT/ES fiber electrodes with more bilayers of the MWNT film also displayed higher electrical current because denser packing of MWNTs led to a lower resistance for electron transport across the conductive electrode. Using the sensor enclosure described above, the sensing capability of the thiourea-functionalized MWNT/ES (TF-MWNT/ES) fiber electrode (with 2 bilayers of (MWNT-COO/MWNTNH 3+) films) was investigated for detecting DMMP in both aqueous and vapor phases. The sensor signal (S=(I-Io)/Io where Io is the initial current) is a normalized change in electrical current monitored in real time between the two platinum wire electrodes at a constant voltage of 0.05 V. Figure 2-14c shows the sensor signal upon alternating exposure to aqueous DMMP solutions (100 s) and pure water (700 s) at varying concentrations (10 ppb to 5 ppm). Exposing aqueous DMMP to the TF-MWNT/ES fiber 49 electrode caused the current to increase rapidly, then reach a plateau. The increase in electrical current of our MWNT sensor upon exposure to DMMP solution is consistent with observation by Roberts et al. for a resistive sensor based on a single-walled carbon nanotube network. 25 3 In an aqueous environment, a dipole interaction model is proposed as the mode of interaction between CNTs and DMMP because the polar DMMP has a strong dipole moment which can cause local induced-dipole electrostatic interactions between DMMP and carbon nanotubes.[3c1 After a pure water stream replaced DMMP, the current initially decreased below its original value and gradually recovered back to its initial level. We speculate this dip below the initial current can be attributed to swelling and/or relaxation of the polymer substrate when the flow was switched from DMMP to water. Exposing aqueous DMMP to a drop-cast film of thiourea-functionalized MWNTs that was deposited on a glass substrate did not elicit the drop in the current below the baseline value (Figure 2-15). A normalized current change of 0.3% was observed in response to 10 ppb DMMP in water. This result is excellent sensitivity for MWNT based sensors operating in an aqueous environment. Our sensor was able to detect DMMP at concentration below the recommended maximum permissible concentration (MPC) for Sarin nerve agent in field drinking water of 14 ppb. 26 The modulation in current was directly correlated to the concentration of DMMP, as shown Figure 2-14d. An approximately linear response was observed for the TF-MWNT/ES fiber electrodes to DMMP solution in the concentration range of 10 ppb to 0.5 ppm. A saturation of response was observed at concentrations above 0.5 ppm. The sensing response of the TFMWNT/ES fiber electrodes to aqueous DMMP was compared to the unfunctionalized MWNT/ES fiber electrodes (Figure 2-14d). The TF-MWNT/ES fiber electrodes exhibited 50 a 1.2-3.0-fold increase in the response when compared to the MWNT/ES fiber electrodes without the thiourea-based receptor. See Figure 2-16 for the response curves of unfunctionalized sensors to pulses of aqueous DMMP. The improved response of the unfunctionalized MWNT/ES fiber electrodes to DMMP is expected, as a result of the hydrogen-bond interactions between DMMP molecules and the available carboxyl and amine groups on the unfunctionalized MWINT/ES fiber electrodes. DMMP, however, formed stronger hydrogen bonding with the acidic thiourea -NH protons on the TFMWNT/ES fiber electrodes, resulting in a larger response. In addition, the observed response of the unfunctionalized MWNT/ES fiber electrodes showed little correlation with DMMP concentration. The enhancement in the sensory response through the functionalization of the thiourea-based receptor demonstrates the utility of the MWNT/ES fiber electrode as a platform for creating chemical/biological sensors. 8 6- ~'20 0 500 1500 1000 Time [s] 2000 Figure 2-15. Signal response (AZ/Jo [%]) of thiourea-functionalized MWNT film upon cyclic exposure to aqueous DMMP (0.5 ppm) and water. Aqueous DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to water at 'arrow with asterisk'. 51 1.5 aqueous DMMP 1.2- .0 - - 50 ppm 50 ppb - 10 ppb 0.90.60.3- -0.30 -- 500 1000 1500 Time [s] 2000 Figure 2-16. Variation of responses (AI/Io [%]) of unfunctionalized MWNT/ES fiber electrode upon cyclic exposure to aqueous DMMP (10 ppb to 50 ppm) and water. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to water at 'arrow with asterisk'. The response of the TF-MWNT/ES fiber electrode was further examined upon cyclic exposures to different concentrations (5 to 100 ppm) of DMMP vapor (30 s) and N2 stream (400 s). The sensing response (Figure 2-14e) gave a reduction in current (increase in resistance), which was reversible. The response was rapid with the introduction of DMMP vapor, and gradually recovered back to its original level when the flow was switched from DMMP to N2 . This decrease in conductance of the MWNT sensor upon exposure to DMMP vapor can be attributed to the adsorption of DMMP 52 molecules onto the surface of MWNT network and the subsequent change of the intertube (junction) resistance. 27 3 The sensor response to DMMP vapor was directly dependent on the concentration of DMMP, as shown in Figure 2-14f. See Figure 2-17 for the response curves of unfunctionalized sensors to pulses of diluted DMMP vapor. The incorporation of the thiourea-based receptor on MWNT/ES fiber electrodes led to a 1.22.0-fold increase in response compared to the unfunctionalized MWNT/ES fiber electrodes. 1.0 DMMP gas 100 ppm 0.8- i - 20 ppm 5 ppm 0.60.4- '0.2-0.20 500 1000 Time [s] 1500 Figure 2-17. Variation of responses of TF-MWINT/ES fiber electrodes upon cyclic exposure to diluted DMMP vapor (5, 20, 100 ppm) and nitrogen stream. DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to nitrogen at 'arrow with asterisk'. 53 The effect of the incorporation of the thiourea-based receptor was further evaluated by exposing the thiourea-functionalized and unfunctionalized MWNT/ES fiber electrodes to a variety of analytes (Figure 2-18). The thiourea-funcitonalized sensors showed a significant enhancement in sensing response to analytes that can form hydrogen-bonding interactions, e.g., DMMP, acetone and water. Non-polar analytes such as toluene and hexane induced a much lower response. These results are in line with the expected interaction of the tested analytes with the thiourea-functionalized and unfunctionalized MWNT/ES fiber electrodes. In this study, the thiourea-based moiety was the only receptor incorporated into the MWNT/ES fiber electrode; thus the functionalized sensor was not designed for high selectivity. Adding different classes of receptors to the MWNT/ES fiber electrodes can improve sensor selectivity and will be investigated in the future. 54 0.9 0.8 0.7 EUnfunctionalized 0 H Functionalized 0.6 0., 0.5 0.4 0.3 0.2 T U.1I 0.0 I S ,,.ZRf IQ n3 ez NZ ze 10 Figure 2-18. Average sensing response of unfunctionalized and thiourea-functionalized MWNT/ES fiber electrodes to different analytes. The concentration of each analyte represents 1% of its equilibrium vapor concentration at room temperature except DMMP, which was used at -6% of its equilibrium vapor concentration at room temperature. The kinetics of adsorption and desorption of DMMP vapor on the sensors was modeled to extract kinetic parameters, such as the adsorption rate constant (k), desorption rate constant (k.), and equilibrium constant (K=k/k. 1 ). The desorption step was first fitted to a reversible binding model described elsewhere [28a] using a first-order reaction rate to obtain k 1, which is directly related to the DMMP desorption rate.[ 2 8b The desorption part of the sensor signal (S = AI/Io, where So is the signal when the analyte is removed) was 55 fitted to Equation (4) to obtain k4 = 1.7 ± 0.2 x 10-2 s' for the thiourea-frnctionalized sensor and 2.6 ± 0.4 x 10-2 s- for the unfunctionalized sensor (see Figure 2-19 in for curve-fitting). These values are within the range of previously reported values for DMMP desorption rate of 3.67 x 10-4 s1 and 0.525 s4 for resistive sensors based on single-walled carbon nanotubes.[ 7' 48 The desorption rate of the unfunctionalized sensor is 55% higher than the value obtained for the thiourea-functionalized sensor. This result suggests that the presence of the thiourea-based receptor slowed down the desorption process of DMMP molecules because DMMP interacted more strongly with the sensor with the thiourea-based receptor. Fitting the adsorption portion of the data to a reversible binding kinetic model, as described elsewhere 47 (Equation (5)), at an analyte concentration CA, gives k = 9.7 ± 0.5 x 10-7 ppb' s- and K = 7.3 ± 0.7 x 10- ppb 1 for the thiourea- functionalized sensor, and k = 4.9 ± 0.4 x 10~7 ppb' s- and K = 3.2 ± 0.5 x 10~5 ppb 1 for the unfunctionalized sensor. The K value of the thiourea-functionalized sensor is very close to the previously reported equilibrium constant of 6.98 x 10' ppb' as reported by Lee et al. for a DMMP gas sensor based on functionalized single-walled carbon nanotubes. S(t) = S(t)= So exp(-k 1 t) ( kC (kCA +k- 1 ) (4) (Ii-exp(-(kCA+ kI)t)) (5) 56 0.4- - 0 Thiourea functionalized Unfunctionalized 0.30.20.1- ---............................ .. 0.0 0 100 200 300 Time [s] 400 500 Figure 2-19. Signal response (-AI/Io [%]) upon exposure to a pulse of DMMP vapor (20 ppm). DMMP is introduced at 'arrow'. The flow to the sensor is switched from DMMP to N2 at 'arrow with asterisk'. The dots represent the experimental data while the solid lines are the best fit for the DMMP desorption step. 2.4 Conclusions We have presented a versatile, scalable method for depositing conformal, binderfree MWNT multilayer films on flexible, porous electrospun (ES) fiber substrates. The vacuum-assisted spray-LbL assembly technique enables rapid, uniform electrostatic assembly of positively and negatively charged MWNTs on individual ES fibers, creating highly porous networks of conformal MWNT multilayers with controlled loading and electrical conductivity. The available -NH 2 and -COOH functional groups on the 57 MWNT films offer a direct route for covalently attaching a receptor of interest to the MWNTs, in order to tailor the chemical specificity of the MWNT/ES fiber electrode for sensing applications and a means of greatly increasing properties such as selectivity and sensitivity in a modular fashion. Here, a thiourea-based receptor was covalently attached to the -NH 2 functional groups on the MWNT/ES fiber electrodes to enhance sensitivity toward DMMP, resulting in an up to 3-fold increase in sensing response when compared to unfunctionalized MWNT/ES fiber electrodes and very high levels of sensitivity for aqueous multi-walled carbon nanotube sensors. This work shows that the engineered textile can function as an ultrasensitive sensing platform for real-time detection of an analyte of interest in an aqueous or gas phase. More importantly, this work demonstrates another example of the simple, versatile, water-based method of spray-LbL coating for introducing new functionalities to complex three-dimensional structures. We believe this method will provide an economical, yet powerful tool for developing lightweight, flexible carbon-nanotube-based sensors, intelligent textiles and wearable electronics. 2.5 References [1] [2] [3] Lukowicz, P.; Kirstein, T.; Troster, G. Method Inform. Med 2004, 43, 232. a) Nomura, K.; Ohta, H.; Takagi, A.; Kamiya, T.; Hirano, M.; Hosono, H. Nature 2004, 432, 488; b) Windmiller, J. R.; Wang, J. Electroanal2013, 25, 29; c) Hu, L. B.; Pasta, M.; La Mantia, F.; Cui, L. F.; Jeong, S.; Deshazer, H. D.; Choi, J. W.; Han, S. M.; Cui, Y. Nano Letters 2010, 10, 708; d) McAlpine, M. C.; Ahmad, H.; Wang, D. W.; Heath, J. R. Nat. Mater. 2007, 6, 379. a) Kong, J.; Franklin, N. R.; Zhou, C. W.; Chapline, M. G.; Peng, S.; Cho, K. J.; Dai, H. J. Science 2000, 287, 622; b) Novak, J. P.; Snow, E. S.; Houser, E. J.; Park, D.; Stepnowski, J. L.; McGill, R. A. Appl. Phys. Lett. 2003, 83, 4026; c) Snow, E. S.; Perkins, F. K.; Houser, E. J.; Badescu, S. C.; Reinecke, T. L. Science 2005, 307, 1942; d) Wang, F.; Gu, H. W.; Swager, T. M. J.Am. Chem. Soc. 2008, 58 [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] 130, 5392; e) Snow, E. S.; Perkins, F. K.; Robinson, J. A. Chem. Soc. Rev. 2006, 35, 790. Park, S.; Vosguerichian, M.; Bao, Z. A. Nanoscale 2013, 5, 1727. a) Wang, X. Y.; Kim, Y. G.; Drew, C.; Ku, B. C.; Kumar, J.; Samuelson, L. A. Nano Letters 2004, 4, 331; b) Bhattacharyya, D.; Senecal, K.; Marek, P.; Senecal, A.; Gleason, K. K. Advanced FunctionalMaterials2011, 21, 4328. a) Decher, G. Science 1997, 277, 1232; b) Hammond, P. T. Materials Today 2012, 15, 196. a) Hammond, P. T. Aiche J2011, 57, 2928; b) Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213; c) Choi, J.; Rubner, M. F. Macromolecules 2005, 38, 116. a) Krogman, K. C.; Lowery, J. L.; Zacharia, N. S.; Rutledge, G. C.; Hammond, P. T. Nat Mater 2009, 8, 512; b) Liu, D. S.; Ashcraft, J. N.; Mannarino, M. M.; Silberstein, M. N.; Argun, A. A.; Rutledge, G. C.; Boyce, M. C.; Hammond, P. T. Advanced FunctionalMaterials2013, 23, 3087. Lee, S. W.; Kim, B. S.; Chen, S.; Shao-Horn, Y.; Hammond, P. T. J.Am. Chem. Soc. 2009, 131, 671. Kim, S. Y.; Hong, J.; Kavian, R.; Lee, S. W.; Hyder, M. N.; Shao-Horn, Y.; Hammond, P. T. Energ Environ Sci 2013, 6, 888. Ramanathan, T.; Fisher, F. T.; Ruoff, R. S.; Brinson, L. C. Chem. Mat. 2005, 17, 1290. a) Zhou, K.; Thouas, G. A.; Bernard, C. C.; Nisbet, D. R.; Finkelstein, D. I.; Li, D.; Forsythe, J. S. A CS Appl. Mater. Interfaces 2012, 4, 4524; b) He, L. M.; Zhao, P. P.; Han, Q.; Wang, X. Y.; Cai, X.; Shi, Y. F.; Zhou, L. B.; Zhang, Y. M.; Xue, W. Carbon 2013, 56, 224; c) Luo, Y.; Wang, S. G.; Shen, M. W.; Qi, R. L.; Fang, Y.; Guo, R.; Cai, H. D.; Cao, X. Y.; Tomas, H.; Zhu, M. F.; Shi, X. Y. Carbohyd Polym 2013, 91, 419. Mannarino, M. M.; Rutledge, G. C. Polymer 2012, 53, 3017. a) Kapantaidakis, G. C.; Koops, G. H. Journalof Membrane Science 2002, 204, 153; b) Deimede, V.; Voyiatzis, G. A.; Kallitsis, J. K.; Qingfeng, L.; Bjerrum, N. J. Macromolecules 2000, 33, 7609. Asfardjani, K.; Segui, Y.; Aurelle, Y.; Abidine, N. JournalofApplied Polymer Science 1991, 43, 271. Deen, W. M. Analysis of TransportPhenomena; 6th ed.; Oxford University Press: New York, NY, 1998. a) Weber, M.; Kamal, M. R. Polym Composite 1997, 18, 711; b) Stauffer, D.; Aharony, A. Introduction to PercolationTheory; Taylor & Francis: London, 1992. Hu, L. B.; Hecht, D. S.; Gruner, G. Chem. Rev. 2010, 110, 5790. a) Du, J. H.; Zhao, L.; Zeng, Y.; Zhang, L. L.; Li, F.; Liu, P. F.; Liu, C. Carbon 2011, 49, 1094; b) Li, Q.; Xue, Q. Z.; Hao, L. Z.; Gao, X. L.; Zheng, Q. B. Compos Sci Technol 2008, 68, 2290; c) Lin, B.; Sundararaj, U.; Potschke, P. MacromolecularMaterials andEngineering2006, 291, 227; d) Jiang, X. W.; Bin, Y. Z.; Matsuo, M. Polymer 2005, 46, 7418; e) Logakis, E.; Pandis, C.; Peoglos, V.; Pissis, P.; Pionteck, J.; Potschke, P.; Micusik, M.; Omastova, M. Polymer 2009, 50, 5103. 59 [20] [21] [22] [23] [24] [25] [26] [27] [28] Li, J.; Ma, P. C.; Chow, W. S.; To, C. K.; Tang, B. Z.; Kim, J. K. Advanced FunctionalMaterials 2007, 17, 3207. Balberg, I.; Anderson, C. H.; Alexander, S.; Wagner, N. PhysicalReview B 1984, 30, 3933. a) Sandler, J. K. W.; Kirk, J. E.; Kinloch, I. A.; Shaffer, M. S. P.; Windle, A. H. Polymer 2003, 44, 5893; b) Balberg, I.; Binenbaum, N. PhysicalReview B 1987, 35, 8749. Bordwell, F. G. Accounts Chem. Res. 1988, 21, 456. Jiang, Y. B.; Li, A. F.; Wang, J. H.; Wang, F. Chem. Soc. Rev. 2010, 39, 3729. Roberts, M. E.; LeMieux, M. C.; Bao, Z. N. ACSNano 2009, 3, 3287. Guidelinesfor Chemical WarfareAgents in MilitaryFieldDrinking Water; The National Academies Press, 1995. Wang, F.; Swager, T. M. J.Am. Chem. Soc. 2011, 133, 11181. a) Lee, C. Y.; Strano, M. S. Langmuir 2005, 21, 5192; b) Lee, C. Y.; Sharma, R.; Radadia, A. D.; Masel, R. I.; Strano, M. S. Angew. Chem.-Int. Edit. 2008, 47, 5018. 60 3. Self-standing Carbon Nanotube Electrodes with 3D Hierarchical Porous Structure for High-Rate Energy Storage Applications Abstract We report the fabrication of hierarchical porous multi-walled carbon nanotube (HPMWNT) electrodes using using polysulfone electrospun fiber mat as a macroporous template and spray layer-by-layer (LbL) assembly of functionalized MWNTs. These HPMWNT electrodes are self-standing and free of binder and current collectors. Scanning electron microscopy (SEM) show that HP-MWNT electrodes contain three-dimensional interconnected macropores and mesoporous network of LbL-MWNT films. HP-MWNT electrodes can deliver a high gravimetric energy of 100 Wh kg_1 at a power of 25 kW kg_1 in nonaqueous cells. Compared to similar MWNT electrodes without micron-size pores, HP-MWNT electrodes exhibit superior rate capability with a 10-fold increase in energy retention when the power was increased from 100 W kg-1 to 100 kW kg-'. We speculate that a combination of large macropores that serve as ion-buffering reservoirs and mesoporous network of MWNTs that facilitate fast ion transport enabled HP-MWNT electrodes to have excellent energy retention at high power, showing promise as electrode materials for high-rate electrochemical energy storage applications such as electric vehicles. 61 3.1 Introduction The growing demand for energy storage systems for applications in electric and hybrid electric vehicles, load-leveling of energy supply and demand, and portable electronics has attracted significant research efforts for developing new electrode materials with high power, high energy, and long cycle life for use in lithium-ion batteries and electrochemical capacitors (ECs).1 3 Carbon nanotubes (CNTs) are considered to promising electrode materials for electrochemical energy storage due their excellent physicochemical properties including high electronic conductivity, high surface area, chemical and mechanical stability. 4 Numerous efforts have been devoted to developing processing methods for creating nanostructured CNT electrodes with high electronic and ionic transport properties. Various strategies such as chemical vapor deposition,5 spray, 6 filtration,'' 8 and layer-by-layer (LbL) deposition,9 '10 have been suggested for the fabrication of CNT electrodes without the use of insulting binders to maximize the electrochemically active surface area while minimizing electronic resistance. Considerable efforts have also been made to investigate the effects of pore structure of electrodes on electrochemical performance with the aim of improving ion transport." Carbon materials with hierarchical porous structure that combines macroporous cores, mesoporous walls, and micropores have been shown to exhibit great potential as electrode materials due to the synergistic benefit of the hierarchical pore texture.1 2 13 The macropores (>50 nm) function as ion-buffering reservoirs to provide decreased ion transport distance to the interior surfaces. The mesoporous walls (2-50 nm) provide fast ion-transport pathways with low resistance. The micropores enhance the 62 electrical-double-layer capacitance. 12 Thus, numerous efforts have been devoted to develop carbon-based electrodes with tailored porous structure. Recently, Lee and coworkers 7 have reported the fabrication of self-standing oxidized CNT electrodes of tens of microns in thickness using a vacuum-filtration process of oxidized few-walled CNTs. These mesoporous CNT electrodes exhibit a high gravimetric energy -200 W h kg~ 1 at a high power of ~10 kW kg~1 . However, significant decrease in energy density was observed at high power (>10kW kg'). Kinetic problems related to inner-pore ion transport are speculated to be the contributing factors causing the reduction in electrode performance particularly for electrodes with thickness on the order of tens of microns. Incorporation of large macropores into these mesoporous CNT electrodes could improve their electrochemical performance at high rates by reducing ion-transport resistance. In this work, we report the synthesis of free-standing hierarchical porous multi-walled carbon nanotube (HP-MWNT) electrodes with three-dimensionally interconnected macropores and mesoporous MWNT network. Here, we utilize a template method to incorporate micron-size channels into electrode structure using a sacrificial electropsun fiber substrate (fiber diameter ~I Rm). A vacuum-assisted spray-LbL technique is employed to deposit functionalized MWNT films on the sacrificial electrospun fiber mat. HP-MWNT electrodes are shown by scanning electron microscopy (SEM) to contain highly interconnected macropores and mesoporous network of unbundled MWNTs. We compare electrochemical performance of HP-MWNT electrodes (thickness = 66-74 pm) to vacuum-filtered MWNT electrodes (thickness = 58-71 pm) that do not contain macropores. HP-MWNT electrodes exhibit much higher rate 63 capability compared to VF-MWNT electrodes with a 10-fold increase in energy retention when the power is increased from 100 W kg-1 to 100 kW kg 1 . 3.2 Experimental Section Materials:All chemicals, unless stated otherwise, were purchased from Sigma Aldrich (St. Louis, MO) and used without further purification. pH adjustment of all solutions was done by addition of dilute hydrochloric acid or sodium hydroxide. Branched polyethylenimine (BPEI, Mw = 75,000) was purchased from PolySciences (Warrington, PA) and prepared at 0.2 wt.% in deionized water (Milli-Q, 18.2 MQe cm) and titrated to pH 4. Pristine MWNTs (95% purity, length 5-20 gm, outer diameter 15 ± 5 nm) synthesized by chemical vapor deposition were purchased from NanoLab (Newton, MA). Preparationof electrospunfiber materials: The electrospinning apparatus, similar to that as previously reported, consisted of two aluminum disks 10 cm in diameter oriented parallel to each other and separated by a distance of 30 cm. A 22 wt.% solution of bisphenol A polysulfone (UDEL P3500, Solvay Advanced Polymers) in 1:1 N,Ndimethyl formamide (DMF):N-methylpyrrolidone (NMP) was delivered with a syringe pump (Harvard Apparatus PHD 2000) at a rate of 0.010 mL min-1 through a 0.04" ID needle in the top aluminum disk. A power supply provided 19.5 kV potential to the upper aluminum disk and the polymer solution was drawn to the bottom grounded disk where fibers of 1.12 ± 0.16 gm in diameter were collected. The thickness of the mat could be controlled by the time allowed for deposition (- 100 pm thick mat in 2 hours). The electrospun sample was annealed in an oven at 170 'C for 2 hours to remove any residual 64 solvent and to improve the mechanical strength of the electrospun fiber mat. Fiber diameter was determined from measurements, using ImageJ, of 100 fibers selected manually and at random from an SEM image with a magnification of 3,OOOX. Prior to LbL deposition, a 2.5" diameter disk of polysulfone electrospun mat was cut and plasmaetched (Harrick PDC-32 G plasma cleaner) in air for one minute to improve the wettability. The mat was then immediately soaked in a 0.2 wt.% aqueous solution of BPEI (pH 4) for 10 min to impart positive charge on the surface of electrospun fibers. FunctionalizationofMWNTs: MWNTs with negatively charged carboxylate ions (MWNT-COO) and positively charged ammonium ions (MWNT-NH 3+) were prepared according to published protocol with some modifications.14 As-received MWNTs were refluxed in strong acids (3/1 v/v, 95% H2 SO 4 , 70% HNO 3) at 70'C for 4 hours, and then washed several times with deionized water by filtration using a polypropylene carbonate membrane (Whatman, pore size = 50 nm). The MWNT-COO powder was dispersed in deionized water to attain 0.5 mg mlV dispersion. The resulting dispersion was dialyzed (MWCO 10,000) for a few days to remove any residuals and byproducts from functionalization and sonicated (Branson Bransonic 2510 ultrasonic cleaner) for 1 hour to improve the quality of dispersion. MWNT-NH 3+was prepared by reacting MWNT-COO dispersion (500 mL, 0.5 mg mlV, pH 5) with excess ethylenediamine (50 mL) in the presence of 5 g of 1-(3-(Dimethylamino)propyl)-3-ethylcarbodiimide methiodide. The resulting dispersion was dialyzed for a few days to remove any residuals and byproducts, pH adjusted to 4 and sonicated for 1 hour to improve the quality of dispersion. Prior to electrode assembly, MWNT dispersions were sonicated for 30 min. 65 Electrode Preparation: Vacuum-FiltrationProcess: Vacuum-filtered multi-walled carbon nanotube (VFMWNT) electrodes were synthesized using a vacuum filtration method on Whatman polycarbonate track-etched membranes (0.2 mm pore diameter), which generated mechanically robust MWNT films attached to the membrane. The first step was an addition of MWNT-COO dispersion (0.5 mg ml- 1, 10 mL, pH 4) to the filter funnel. Immediately upon completion of filtration, MWNT-NH 3+ dispersion (0.5 mg mr', 10 mL, pH 4) was added to the filter funnel. Once the filtration of MWNT-NH 3+ dispersion was finished, this step concluded one bilayer of MWNT film. This process was repeated four more times to achieve 5 bilayers using 50 mL of MWNT-COO and 50 mL MWNTNH 3+. Free-standing electrodes were obtained by removing air-dried MWNT films from the membrane using tweezers, and electrodes were further dried overnight (1 50'C) under vacuum. The dried electrodes were cut or punched to the desired size. The thickness of VF-MWNT electrodes was 58-71 pm. Spray-LbL Assembly: A 2.5" diameter disk of electrospun fiber mat (thickness = 92 ± 5 tm) was placed on a stainless steel mesh (56 mesh, 0.004", TWP Inc.) and mounted inside a filter holder (Pall Life Science) using a screw cap. A vacuum source was connected to the back of the filter holder generate an airflow through the electrospun fiber mat during spray deposition. All solutions were atomized into spray and delivered by a home-built automated spray system, as previously described elsewhere, using high purity N 2 (20 psi).2 2 An automated program, containing spray recipe, executed by a logic relay controlled the apparatus. For conformal coatings of MWNT films, the deposition 66 process started with spraying MWNT-COO~ (0.25 mg mlf1, pH 4) and MWNT-NH3 + (0.25 mg ml 1 , pH 4) for 3 s each, with 60 s of drain time after MWNT spray, 5 s of rinse water and 60 s of drain time after water spray in between the MWNT sprays. The process was repeated to achieve a target number of bilayers. Bridging MWNT films were deposited using the same process described above but the drain time was reduced to 2 s. The MWNT-coated mat was dried under vacuum at room temperature overnight before it was cut or punched to the desired size. The electrospun fibers were removed by immersing the cut samples in DMF (100 mL) at 80'C for 12 hours with periodic stirring. The samples were transferred to 20-mL scintillation vial with fresh DMF and placed on a shaker for 6 hours at room temperature. After that, the samples were washed with Milli-Q water and dried in a vacuum oven at 150'C for 12 hours. The thickness of HP-MWNT electrodes was 66-74 pm. Characterization:A Kratos AXIS Ultra Imaging X-ray photoelectron spectrometer (Kratos Analytical) was used to analyze the surface chemistry of functionalized MWNTs and MWNT electrodes. Microstructure and the cryo-fractured cross-sections of MWNT films were examined using a scanning electron microscope (JEOL 6320FV FieldEmission High-Resolution SEM) operating at 2 and 5 kV. Thickness of electropsun fiber mats and MWNT electrodes was measured using a digital micrometer (Mitutoyo CLM 1) with a constant measuring force of 0.5 N in air at room temperature and cross-sectional SEM images. Nitrogen adsorption and desorption isotherms of the MWNT films were measured at 77 K on a Micromeritics ASAP 2020 system. The samples were degassed at 70 K under vacuum (<104 mbar) for several hours prior to measurements. Surface areas 67 of the MVNT films were calculated using the Brunauer-Ennett-Teller (BET) method. Mercury porosimetry measurements were performed using an Autopore IV porosimeter (Micromeritics, Norcross, GA) according to published protocol. [23] Zeta potential of MWNT aqueous suspension was determined using a Zeta PALS instrument (Brookhaven Instrument Corp.). Sheet resistance was measured using a 4-point probe (Signaton) with a device analyzer (Keithley 4200). ElectrochemicalMeasurements: Electrochemical tests in aqueous electrolyte were conducted using two-electrode cells (Swagelok@ type cell) with two identical selfstanding MWNT electrodes as the working and counter electrodes, two sheets of microporous polymer separators (Celgard 2500) wetted in 1 M K2 SO 4 solution. For electrochemical tests in nonaqueous electrolyte, two-electrode coin cells were assembled in an argon-filled glove box (02 and H20 level below 1 ppm) using lithium foil as a negative electrode, self-standing MWNT electrode, two sheets of microporous polymer separators (Celgard) and non-aqueous electrolyte of 1 M LiPF6 dissolved in the mixture of ethylene carbonate (EC) and dimethyl carbonate (DMC) 1:1 volume ratio (Novolyte). The assembled cells were tested electrochemically using a Solartron Analytical 1470E potentiostat at room temperature for cyclic voltammetry and charge-discharge tests. Electrochemical impedance spectroscopy (Solartron 1260 impedance analyzer) was used to measure the impedance of self-standing MWNT electrode as a function of the frequency. The electrodes were equilibrated in the electrolytes for at least 5 min at a direct current (dc) bias potential of +100 mV vs Ag/AgCl. A 10 mV root mean square 68 alternating-current amplitude with a frequency ranging from 0.1 Hz to 300 kHz was superimposed on the dc bias potential. 3.3 Results and Discussion 3.3.1 Electrode Preparation and Characterization 3.3.1.1 Vacuum-filtered MWNT electrodes The primary objective of this work is to investigate the effect of hierarchical porous structure on electrochemical performance of MWNT electrodes. In particular, the incorporation of interconnected micron-size channels into mesoporous MWNT electrodes is anticipated to facilitate fast ion transportation by reducing ion transport distance and resistance in the electrode especially at high rates. We first prepared mesoporous MWNT electrodes using a vacuum filtration method from negatively and positively charged functionalized MWNT suspensions. MWNTs were obtained from NanoLab (95% purity, length ~5-20 pm, outer diameter 15 ± 5 nm). Negatively charged MWNTs (MWNT- COO~) and positively charged MWNTs (MWNT-NH 3+) were prepared using surface functionalization process described in previous work.' 4 X-ray photoelectron spectroscopy (XPS) elemental analysis was performed to quantify surface functional groups on the modified MWNTs (Figure 3-1). XPS wide scan survey spectra show that oxygen functional groups were introduced onto surfaces of MWNT-COO~, and MWNT-NH 3+ contained oxygen and nitrogen functional groups. Figure 3-1c displays high-resolution C Is XPS spectra, which reveal detailed surface functional groups on the MWNTs. The unmodified MWNT C is peaks consisted of mostly sp 2 -hybridized graphitic carbons 69 centered at 284.5 eV (in red) and some sp 3 -hybridized diamond-like carbons centered at 285.2 eV (in blue). The C1, spectra of MWNT-COO show carbon atom peaks connected to oxygen atoms (C-O centered at 286.0 + 0.1 eV, C=O centered at 286.9 ± 0.2 eV, COOH(COOR) centered at 288.9 ± 0.2 eV).[15] The sp 3 -hybridized carbon peak also increased, indicating that the oxidation process breaks the outer graphitic wall of MWNTs and introduce oxygen functional groups. 7 The C Is peaks of MWNT-NH 3+ show the presence of amide bonds N-C=O (287.9 eV) and amine C-N (286.2 eV),15 which suggested the incorporation of amide bonds and amine groups. The introduced carboxylic acid and amine groups on the surface of MWNT-COO and MWNT-NH 3+ exist as carboxylate anions (COO~) and amine cations (NH 3 +)in aqueous solution (pH 4) with zeta potential values of -49.3±2.5mV and +50.5±1.1 mV respectively, which can provide electrostatic repulsion to prevent precipitation of MWNTs.7 70 (a)Cs (C) Unmodified MWNTs 2 Sp sP3-c - C1S -C MWNT-NH3+ MWNT-COOUnmodified MWNT 600 500 400 300 Binding energy [eV] 200 MWNT-COO_ c- Sp2-C CIS - SP 3-C (b) C-0 M Atomic Ratio Cis Ols Nis Unmodified MWNTs 1.00 0.00 0.00 MWNT-COO~ 0.82 0.18 0.00 MWNT-NH*3 0.79 0.11 COH 2c 0) - C: CO C-o C; 4C COOH 0.10 MWNT-NH 3 C1s - SP 2-C (d) 3 - SP -C N\\ NH2 H OH 10 MWNT-COO- ON -=O -CONH + C=OC CONH MWNT-NH 3 292 290 288 286 284 Binding energy [eV] 282 Figure 3-1. X-ray photoelectron spectroscopy (XPS) survey spectra of functionalized MWNTs. (a) Wide scan survey, (b) atomic concentration and (c) Ci spectra of unmodified MWNTs, MWNT-COO- and MWNT-NH 3+. (d) Schematic representation of functional groups on MWNT-COO and MWNT-NH 3+. Mesoporous MWNT electrodes were assembled via the vacuum filtration (VF) method using aqueous dispersions of MWNT-COO~ and MWNT-NH 3+ (0.5 g 1- 1, pH 4). The electrodes were built in a layer-by-layer fashion by first adding MWNT-COO~ dispersion then MWNT-NH 3+ dispersion immediately upon completion of the MWNT- 71 COO filtration. This process was repeated four more times to achieve 5 bilayers using 50 mL of MWNT-COO~ and 50 mL MWNT-NH 3*, yielding films of interwoven MWNTs that are self-standing with thickness of 58-71 ptm. SEM images (Figure 3-2) of VF-MWNT electrodes show the highly interpenetrating network structure of randomly oriented MWNTs. Density of the VF-MWNT electrodes was found to be -0.6 g CM~3, which is similar to reported density of VF electrodes of oxidized few-walled carbon nanotube electrodes (-0.4 to 0.8 g cm- 3). 7 The electrical conductivity of VF-MWNT electrodes was measured using a 4-point probe and found to be 9.1 ± 1.0 S cm 1 , which is comparable to 2-10 S cm- of previously reported LbL-MWNT electrodes fabricated using similarly functionalized MWNTs.15 72 Figure 3-2. Scanning electron microscopy (SEM) images of VF-MWNT electrodes. (a) Low and (b) high magnification cross-section micrographs. (c) Low and (d) high magnification top-down view micrographs. Pore size distribution of VF-MWNT electrodes was examined using nitrogen adsorption-desorption measurement as well as image analysis of top-down view SEM images. Figure 3-3a displays the nitrogen adsorption-desorption isotherm of the VFMWNT electrodes with a Type IV isotherm according to the IUPAC classification, indicating mesoporous characteristics. 16 The BET surface area of the VF-MWNT electrodes was 88.5 m 2 g-1. The pore size distribution (Figure 3-3a, insert) calculated 73 using Barrett-Joyner-Halenda (BJH) method shows the presence of pores ranging from 5 to 30 nm with a peak pore diameter around 13 nm. The pore size distribution calculated from top-down view SEM images (Figure 3-3b) exhibits similar pore sizes with equivalent pore diameter in the range of 5 to 40 nm. These results indicate that the VFMWNT electrodes contain only pores in the mesopore range (2-50 nm). (a) (b) 0.35 1.2- 200- 1.0 1.04=_0.30 .D 0 -0.8 .8- 150- 0.6 U. 0.4W0.20 W 12 0 0 a2 0.- . , 10 Pore Diameter [nm]L 100 LL. 50+ Desorption -u- Adsorption 0.0 a 0.25 0.6- 0.2 0.4 0.6 0.8 1.0 P/Po 0.15- 0.4Q 0.10 E 0.2 0.050 0 5 10 15 20 25 30 35 Equivalent Pore Diameter [nm] 40 0 Figure 3-3. (a) Nitrogen adsorption-desorption isotherm of VF-MWNT electrodes. The insert is the corresponding BJH pore size distribution. (b) Pore size distribution calculated using top-down view SEM image of VF-MWNT electrodes. 3.3.1.2 Hierarchicalporous MWNT electrodes Hierarchical porous MWNT (HP-MWNT) electrodes were fabricated by using a sacrificial template method to incorporate macropores into electrode structure and the vacuum-assisted spray-LbL process to deposit MWNT films. Figure 3-4 illustrates the steps involved in the fabrication of HP-MWNT electrodes. Polysulfone electrospun fiber mat (porosity = 91% by mercury porosimetry, mat thickness = 92+ 5 gm) was employed as a sacrificial substrate to introduce micron-size pores into electrode structure and to 74 facilitate the assembly of LbL-MWNT films via the spray-LbL approach. Electrospun fiber mat with fiber diameter = 1.12 ± 0.16 pim (Figure 3-5a) was chosen in this study because we aimed to investigate the effect of large macropores (>1 pm) on the electrochemical performance of MWNT electrodes. The pore size distribution of the electrospun fiber mat was calculated using mercury porosimetry. Figure 3-5b shows a plot of differential intrusion volume vs. pore diameter for the electrospun sample. It should be noted that the volume distribution has been adjusted to correct for pore buckling deformation according to published work.17 The profile shows the presence of pores with diameter ranging from 1 to 10 pm and a peak pore diameter (diameter at which the largest fraction of mercury is intruded during mercury porosimetry) centered at 3.7 pm. The pores of the electrospun template are at least two times larger than the pores of MWNT films (5 to 40 nm) described above. Tunable characteristics (fiber diameter and mat thickness) of electrospun fiber substrates also allow for tailoring of pore size distribution and thickness of electrodes. 75 macroporous electrospun fiber mat Spray-LbL Spray-LbL deposition of deposition of bridging conformal MWNT films MWNT films LbL-MWNT coated mat Remove electrospun fibers HP-MWNT mesopor es MWNT [macropores LULIVIVVI I s hierarchical porous (HP) MWNT electrode Figure 3-4. Schematic illustration of the fabrication of HP-MWNT electrode from aqueous dispersion of MWNT-COO and MWNT-NH 3 + at room temperature. The resulting MWNT electrode is freestanding and binder-free (bottom-left photo). (a (b) - 04 0A 0.1 1 10 Pore Diameter [pm] 100 Figure 3-5. (a) Top-down view SEM image of polysulfone electrospun fiber mat. (b) Plot of differential mercury intrusion vs. pore diameter of electrospun fiber mat, obtained using mercury porosimetry. 76 The vacuum-assisted spray-LbL technique was used to deposit MWNT films on electrospun fiber mat. In previous work, we have demonstrated conformal coatings of LbL-MWNT films on individual electrospun fibers by alternately spraying MWNTCOO~ and MWNT-NH 3+ suspensions in conjunction with a pressure gradient across the porous electrospun substrate.1 4 The application of the vacuum along with adequate draining time (1 minute) between each spray enabled uniform coatings of (MWNTCOO~/MWNT-NH 3+)n, where n denotes the number of bilayers, throughout the bulk of electrospun fiber mat. In this work, conformal coatings of (MWNT-COO/MWNTNH3+)n=10o were first assembled on the electrospun fibers. Next, bridging layers of (MWNT-COO/MWNT-NH 3)n= 1OO were deposited on one side of the conformally coated mat to completely fill the interstitial voids between fibers and build LbL-MWNT films on top of the mat. The electrospun fibers were then removed by immersed the coated mat in dimethylformamide, creating self-standing HP-MWNT electrodes that are comprised of only functionalized MWNTs with thicknesses of 66-74 pim (Figure 3-4, bottom-left photo). Density of HP-MWNT electrodes was -0.2 g cm, which is three-fold lower than that of VF-MWNT electrodes (~0.6 g cm' ) due to higher porosity. The electrical conductivity of HP-MWNT electrodes was 3.9 ± 0.2 S cm', which is roughly half of the value observed for VF-MWNT electrodes. This difference can be attributed to lower packing density of HP-MWNT electrodes. Figure 3-6a shows the entire cross-section SEM image of HP-MWNT electrode. The bridging section (-7 pm) account for 10-15% of the total electrode thickness. The bridging layers provide mechanical robustness after the electrospun fibers are removed as well as good electrical contact with electrochemical cells. Pore bridging during the 77 vacuum-assisted spray-LbL process was achieved by reducing the draining time between each spray from 1 minute (conformal films) to 2 seconds (bridging films). The bridging of pores (~1-20 pm in equivalent diameter) between fibers may be explained by the insufficient amount of time allowed for MWNTs (hydrodynamic radius of 1 pm) to travel through the porous mat from one side to the other. Inhomogeneous pressure gradient on the electrospun mat caused by broad pore size distribution (Figure 3-5b) and high pore tortuosity could also contribute to the pore-bridging phenomenon by reducing the hydrodynamic drag force experienced by MWNTs sprayed on areas with smaller pores. Instead of penetrating through the electrospun matrix, some MWNTs and water remain on the mat when the next spray droplets arrive at the surface during the spray-LbL process. As the serial deposition continues, the remaining MWNTs on the mat from the previous spray form electrostatic complexes with oppositely charged MWNTs from the next spray. This coating grows laterally to fill interstitial voids until all the pores are completely blocked. Subsequent layers of MWNT films are then built on top of the bridged pores as shown in Figure 3-6a. This bridging phenomenon of pores on porous media has been observed for other spray-LbL systems, including MWNT films assembled on carbon paper.' 0 The underlying mechanisms will be the subject of future investigation. 78 conformal coatings 1 00nm bridging layers Figure 3-6. Scanning electron microscopy (SEM) images of the HP-MWNT electrodes. (a) Low and (b) high magnification cross-section micrographs. (c) Low and (d) high magnification top-down view micrographs of the conformal side of HP-MWNT electrode. Figure 3-6b shows the cross-section SEM image of the conformal (MWNTCOO7/MWNT-NH 3+),=1 oo coated on the sacrificial electrospun fibers. The LbL-MWNT films (thickness ~100 nm) remain intact after the electrospun fibers were removed. The thickness of the conformal MWNT films is approximately uniform throughout the entire conformal section of HP-MWNT electrodes with slightly thicker films near the bridging section. A top-down view SEM image (Figure 3-6c) of the conformal section of HPMWNT electrodes clearly shows that MWNT films were uniformly coated on individual 79 electrospun fibers. A zoom-in SEM image (Figure 3-6d) reveals a highly porous network structure of randomly oriented, unbundled MWNTs. The pore size distribution calculated from top-down view SEM image (Figure 3-7) shows pore sizes with equivalent pore diameters, ranging from 5 to 40 nm, similar to the pore size distribution of VF-MWNT electrodes (Figure 3b). These microstructure analyses demonstrate that HP-MWNT electrodes contain 3D interconnected macropores (equivalent pore diameter -1 -20 pm) from the void space of the sacrificial electrospun fiber mat, in addition to the mesoporous network (5-40 nm) of the LbL-MWNT films. Self-standing HP-MWNT electrodes with hierarchical porous structure show promise as good electrode materials for energy storage devices. 0.35 1.0 0.30. > 0.25 Cr 0.20 0 06u D 0.15- 0.4 U- E 0.10 -0.2 0.05 0 5 10 15 20 25 30 35 Equivalent Pore Diameter [nm] 40 0 0 Figure 3-7. Pore size distribution calculated using top-down view SEM image of HPMWNT electrode. 3.3.3 Electrochemical characteristics of MWNT electrodes 3.3.3.1 Electrochemical performance in aqueous electrolyte system Electrochemical performance of self-standing MWNT electrodes in aqueous electrolyte (1 M K2 S0 4 solution) was obtained using cyclic voltammetry (CV) at 80 different scan rates (Figure 3-8a-b). In theory, a CV curve of an ideal capacitor with a pure electrical double layer (EDL) capacitance behavior will have a perfect rectangular profile because the current (1) should keep constant at a linear scan rate (dV/dt = constant) according to I=CxdV/dt where C is the double layer capacitance. However, slow ion transport in the electrode will lead to a distorted rectangular shape of a CV curve at a high scan rate because ions cannot respond fast enough at the switching potentials.[13] The CV curves of HP-MWNT electrodes (Figure 3-8a) kept a quasi-rectangular shape at all scan rates between 5 and 200 mV s 1 whereas the CV profiles of VF-MWNT electrodes (Figure 3-8b) showed high distortions especially at higher scan rates. At a scan rate of 5 mV s-, both samples displayed a rectangular shape, indicating that the porous structures of both electrodes can satisfy ion transport/diffusion at this low scan rate. However, as potential scan rate increased from 5 to 200 mV s4 , the CVs of VF-MWNT electrodes became more distorted, while the CVs of HP-MWNT electrodes retained a rectangular shape. These results indicate that the hierarchical porous structure of HPMWNT electrodes with interconnected macropores enabled faster ion transport at high charge-discharge rates compared to VF-MWNT electrodes. The gravimetric capacitances calculated from CVs (scan rate = 1 mV s') of HP-MWNT and VF-MWNT electrodes were 17 F g- and 21 F g-, respectively, which can be attributed primarily to EDL capacitance in neutral electrolyte18 (1 M K2 S0 4). Figure 3-8c shows a plot of capacitance retention as a function of potential scan rates. The high capacitance retention ratio supports the fast ion transport characteristics of HP-MWNT electrodes. At a high scan rate of 200 mV s-1, HP-MWNT electrodes still retained 58% of its capacitance, whereas the capacitance value was reduced to 12% for VF-MWNT electrodes. 81 (a ) , , I, , 3 - HP-MWVNT 2- , , , 200 mVs 100 M s4 50 mVsa ~ 1 b) - 50V- /0 0- (b ) - -2- ) I I II . 3 -VF-MWNT 2- 50 I 1 1 200 mVs 100 MVs-1 mVs-I 5 mVs- 1 - -2- -3 -3 0 0.2 0.4 0.6 0.8 0 0.2 0.4 E [V] 0.6 0.8 EMV ) loo- 150 * HP-MWNT A AA cc S60 40- k0 AU AA A4 - N] A \50- .% 20 -A 0 HP-MWNT a VF-MWNT e VF-MWNT 80 - A 00 0 20- 0 0 200 400 Scan Rate [mV/s] 600 0 50 100 Z' [Ohm] 150 Figure 3-8. Cyclic voltamograms (CVs) at different scan rates of (a) HP-MWNT and (b) VF-MWNT electrodes in 1 M K 2S0 4 at room temperature. (c) Capacitance retention ratio at different scan rates. (d) Nyquist plot of electrochemical impedance spectra measured in 1 M K 2 SO 4. Electrochemical impedance spectroscopy was employed to further analyze ion transport resistance in the porous electrodes. The Nyquist plots (Figure 3-8d) show a semi-circle at high frequency and a sharp increase in the imaginary part of the impedance at lower frequency. The diameter of the semicircle is commonly referred to as the polarisation resistance (Rp), which can describe the porous structure of the sample. In general, higher Rp value indicates higher transport resistance of ions into the pores of 82 electrodes. 13 ,19 The Rp value estimated from the Nyquist plot of HP-MWNT (-29 f) was approximately 5 times lower than that of VF-MWNT (-131 Q). This result supports the higher capacity retention observed for HP-MWNT electrodes (Figure 3-8c) and further confirms that hierarchical porous structure of HP-MWNT electrodes can promote a more effective transport of ions into pores with a minimized resistance as compared to VFMWNT electrodes. The gravimetric energy and power characteristics of MWNT electrodes are shown in Figure 3-9. Gravimetric energy density (E) was calculated using specific capacitance (C) obtained from CVs (Figure 3-8a-b) and formula: E=CV2/2. Gravimetric power density (P)was obtained from P=IV/2 using average current I from CVs. At low power of -10 W kg', both HP-MWNT and VF-MWNT electrodes delivered similar energy density of -2 Wh kg'. When the power was increased over 200 W kg', the gravimetric energy of VF-MWNT electrodes decreased significantly by up to an order of magnitude, as expected from the capacitance retention data shown in Figure 3-8c, whereas HP-MWNT electrodes retained 74% of the initial energy. HP-MWNT electrodes delivered much higher power (1.5 kW kg') compared to that of VF-MWNT electrodes (0.2 kW kg') at similar energy (-0.5 Wh kg'). The energy and power limitations observed at high rates are likely related to complexly distributed resistance and tortuous pathways for ion transportation within the porous electrode. Only some parts of the electrode surfaces can be accessed by the ions at high rates while most of the electrode surfaces are utilized for charge storage at low rates.[12] HP-MWNT electrodes with 3D hierarchical porous structure showed superior performance at high rates with a 7-fold increase in power density as compared to that of VF-MWNT electrodes. HP-MWNT electrodes also 83 exhibited good cycling stability up to 1000 cycles, where CVs shows small reduction in current (Figure 3-10a) and capacity (~6.5% decline) during cycling tests (Figure 3-10b). HP-MWNT E VF-MWNT * 103 A A U) 0 ~1) 0 a- 102 A0 10 1 10-1 Energy Density [Wh kg 1',Ntr,) Figure 3-9. Ragone plot of gravimetric energy and power performance of HP-MWNT and VF-MWNT electrodes. The thicknesses of MWNT electrodes were 74 pm for HPMWNT and 66 pm for VF-MWNT. (a) 0.4 Ist Cycling 100 0.21 0 0 th 1 0- Cycling U) t C Cycling 0 ---o---o - 0-0--_ 80 60 40 -0.2Scan Rate: 20 mV s l M M) C 20 Q. 0.8 0 1 -0.4 .- 0o-o- 0 0.2 0.4 E [V] 0.6 Scan Rate: 20 mV s-1 0 200 400 600 800 Cycle Number 1000 Figure 3-10. Cyclic voltammograms of HP-MWNT electrodes with different cycling numbers in 1 M K2SO 4 at room temperature. A scan rate of 20 mV s~ 1 was used. (b) Capacity retention of HP-MWNT electrodes as a function of cycling number. 84 3.3.3.2 Electrochemical performance in non-aqueous electrolyte system Electrochemical properties of these MWNT electrodes were also investigated in nonaqueous liquid electrolyte system using lithium foil negative electrode and 1 M LiPF6 (ED:DMC 1:1 v/v) electrolyte. Figure 3-11 shows voltage-dependent CV measurements of HP- and VF-MWNT electrodes at 1 mV s-1. HP-MWNT electrodes exhibited an average capacitance of 77 and 65 F g'1 for the forward and backward scan (1.5-4.5 V vs. Li), respectively, while an average capacitance of 75 and 70 F g1 was obtained for VFMWNT electrodes. The capacitances of these MWNT electrodes were highly dependent on the voltage window. The gravimetric current of both MWNT electrodes decreased significantly when the potential window was restricted to 1.5-3.0 V vs. Li or 3.0-4.5 V vs. Li, resulting in 23-37% reduction in average capacitance. Higher capacitance values observed with the wide voltage window (1.5-4.5 V vs. Li) can be attributed to additional charge storage via the possible Faradaic reactions between the surface oxygen groups (e.g., carboxylic acid and carbonyl groups) of functionalized MWNTs and Li ions at -3 V vs. Li (e.g., C=O + Li + e * C-OLi) as previously reported for similar LbL-MWNT films9 ' 10 and functionalized few-walled carbon nanotubes 7. 85 (a) 0.4 ( ) HP-MWNT 0.2- I 0.2 - VF-MWNT I I 1 mV s 1 1 mv s 0.1 b) bo) 0,- - <I, 5 to V --- 1.5 to 4.5V 1.5 to 3.OV - 3.0 to 4.5V 1.5 to 4.5V - -0.2 -- -0.4 . 1 2 2 3 1.5 to 3.OV 3.0 to 4.5V 4 4 -0.2 5 5 E [V vs. Li] 1 2 3 4 5 5 E [V vs. Li] Figure 3-11. Potential-dependent cyclic voltammograms (1 mV s'1) of HP-MV/NT (a) and VF-MWNT (b) electrodes measured in lithium cells. The current was normalized to the mass of MWNT electrodes. The thickness of MWNT electrodes were 66 ptm for HPMWNT and 72 pm for VF-MWNT. Electrolyte was 1 M LiPF6 in a mixture of ethylene carbonate and dimethyl carbonate (1:1 v/v). Galvanostatic testing (Figure 3-12a-b) showed that HP-MWNT and VF-MWNT electrodes exhibited similar gravimetric capacities at low rates (51 A g 1 ). The maximum discharge capacities were found to be 76 mAh g-1 and 71 mAh g~ 1 for HP-MWNT and VF-MWNT electrodes, respectively, at a discharge rate of 0.1 A g1. The capacity decreased with increasing current rate as shown in Figure 3-12c. Both MV/NT electrodes exhibited similar dropping rate in capacity with increasing discharge current rate at low rates (<5 A g'1), but HP-MWNT electrodes were able to retain capacity more effectively at high rates (>5 A g1), indicating lower ion transport resistance compared to that of VFMWNT electrodes. These results suggest that the hierarchical porous structure with large macropores enable more efficient ion transportation in the electrode and electrochemical accessible surface area at high current rates. The repetitive charge-discharge cycling test 86 (Figure 3-12d) shows that the gravimetric capacities of HP-MWNT electrodes were sustained over 1000 cycles with negligible capacity loss (retained capacity of 96% after 1,000 cycles). (a) (b)r 0 VF-MWNT HP-MWNT 4 4 LU 3 -J 3- 2 2 50 M 1 0 20 (C) 10 5 0.5 1 40 60 Q [mAh/g] 0.1Ag 80 1 100 10010 , 0 40 -I 10 0-00-0-0-0-0.000 80- 40- " 2-< 2 100t' h I ,M 20 40 60 80 100 Q [mAh g-] 1QOOth 20 - VF-MWNT 0 0.1 5 60- A + HP-MWNT . 80 1 A 60 20 1 0.5 0.1Ag 40 60 Q [mAh g-1] 20 (d) 80 -7 0 5 10 50 I 1 10 Discharge Current [A g-] 100 01 0 10 200 400 600 800 Cycle Number 1000 Figure 3-12. Galvanostatic charge-discharge curves of (a) HP-MWNT (66 pm) and (b) VF-MWNT (72 pm) electrodes in lithium cell at different current rates. The specific capacities were normalized to the weight of MWNT. (c) Gravimetric discharge capacity of HP-MWNT and VF-MWNT electrodes as a function of gravimetric discharge current. (d) Cycling stability of HP-MWNT electrodes tested by charge-discharge cycling tests. MWNT electrodes were charged and discharged at 10 A g~1 between 1.5 V and 4.5 V for 99 cycles, and at every 10 0 th cycle, it was charged and discharged at 0.1 A g'. The potential of the MWNT electrodes was maintained at 1.5 V or 4.5 V for 30 minutes to 87 completely discharge or charge the electrode before charging (or discharging) the electrode from 1.5 V (or 4.5 V). The insert shows charge-discharge curves for the and 10 0 0 th 1 0 0 th cycles. The gravimetric energy and power characteristics of HP-MWNT and VF-MWNT electrodes (based on the MWNT electrode weight) are shown in a Ragone plot (Figure 13). Gravimetric energy density was calculated by integrating the area under the discharge curves of the MWNT electrodes (Figure 12a-b), and gravimetric power density was obtained by dividing the energy density by the discharge time (4.5 V to 1.5 V vs. Li). At a relatively low power (~120 W kg-'), HP-MWNT and VF-MWNT electrodes delivered similar gravimetric energies of~200 Wh kg-1 and ~190 Wh kg1 , respectively. Upon further increase in power to -120 kW kg 1 , a significant decrease in energy density was observed for VF-MWNT electrodes (-2 orders of magnitude reduction) as compared to that of HP-MWNT electrodes (<1 order of magnitude). At high power of -120 kW kg 1 , HP-MWNT electrodes retained 27% of its energy density, whereas VF-MWNT electrodes preserved only 2.4%. This ~10-fold increase in gravimetric energy retention at high power of HP-MWNT electrodes can be attributed to enhanced ion transportation facilitated by 3D interconnected macropores that serve as ion buffering reservoirs at high rates. HP-MWNT electrodes exhibit higher energy and power densities than previously reported spray-LbL-MWNT electrodes10 (normalized by weight of MWNT and current collector) that were fabricated using similarly prepared MWNTs on carbon papers because HP-MWNT electrodes contain only electrochemically active MWNTs without current collector. HP-MWNT electrodes can deliver high gravimetric energy of -100 Wh 88 kg-' at a gravimetric power of~25 kW kg-1 , which exceeds the gravimetric energy of conventional electrochemical capacitors due to the utilization of oxygen-containing groups on MWNTs that enhances energy storage capacity. In addition, the gravimetric energy and power densities of HP-MWNT electrodes are comparable to reported lithiumion batteries. Given that the HP-MWNT electrodes with more practical thickness of~6070 prm are free of current collectors, higher gravimetric energy and power are expected in practical devices. The superior performance of self-standing HP-MWNT electrodes in both organic and aqueous electrolytes makes its promising as electrode materials for high-rate energy storage applications. 105 -A- HP-MWNT E VF-MWNT -9L sLbL-MWNT 3 10 4 :tC . ... .. - 106 1033 00 103 Q_ 10, - 1 10 102 103 Energy Density [Wh kg- electrodel Figure 3-13. A Ragone plot showing electrode-level gravimetric energy and power performance of HP-MWNT and VF-MWNT, calculated from galvanostatic data in Figure 12a-b. Only MWNT electrode weight was considered in the gravimetric energy and power calculations. The gravimetric energy and power of sLbL-MWNT (ref. 10) were calculated from galvanostatic discharge curves normalized by the weight of MWNT and current collector. The performance of electrochemical capacitor and Li-ion batteries is 89 estimated from gravimetric energy and power in the Ragone plot of a previous review paper (ref. 3). 3.4 Conclusions We report the fabrication of functionalized MWNT electrodes with hierarchical (HP) porous structure containing 3D interconnected meso/macropores. HP-MWNT electrodes are synthesized using macroporous electropsun fiber mats as sacrificial template to incorporate macropores into electrode structure. Functionalized MWNT films are assembled on electrospun fiber substrate using the vacuum-assisted spray-LbL process. After the electrospun fibers are removed, the interwoven network of LbLMWNT films supports the self-standing structure of the electrode that is free of binder and current collector. SEM images reveal that HP-MWNT electrodes contain mesoporous network of LbL-MWNT films combined with 3D interconnected micron-size channels left by the sacrificial electrospun fibers. HP-MWNT electrodes can deliver high gravimetric energy of ~100 Wh kg1 at a gravimetric power of -25 kW kg 1 . The incorporation of well-interconnected macropores in the MWNT electrodes enhances ion transport in the electrode, resulting in superior capacity retention compared to similar MWNT electrodes without macropores. Future research will focus on incorporating highenergy materials (e.g. metal oxides20 ) into HP-MWNT electrodes as well as optimizing the spray-LbL deposition conditions (e.g. number of MWNT layers, 13 pH) to further increase the gravimetric energy and power of electrodes. We believe this fabrication method for creating hierarchical porous electrodes based on functionalized MWNTs 90 provide an interesting strategy for designing self-standing, binder-free electrodes for high-rate energy storage applications. 3.5 References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] Liu, C.; Li, F.; Ma, L. P.; Cheng, H. M. Adv. Mater. 2010, 22, E28. Simon, P.; Gogotsi, Y. Nat. Mater. 2008, 7, 845. Lee, S. W.; Gallant, B. M.; Byon, H. R.; Hammond, P. T.; Shao-Horn, Y. Energ Environ Sci 2011, 4, 1972. Ajayan, P. M.; Zhou, 0. Z. CarbonNanotubes 2001, 80, 391. Futaba, D. N.; Hata, K.; Yamada, T.; Hiraoka, T.; Hayamizu, Y.; Kakudate, Y.; Tanaike, 0.; Hatori, H.; Yumura, M.; Iijima, S. Nat. Mater. 2006, 5, 987. Kaempgen, M.; Chan, C. K.; Ma, J.; Cui, Y.; Gruner, G. Nano Letters 2009, 9, 1872. Lee, S. W.; Gallant, B. M.; Lee, Y.; Yoshida, N.; Kim, D. Y.; Yamada, Y.; Noda, S.; Yamada, A.; Shao-Horn, Y. Energ Environ Sci 2012, 5, 5437. Byon, H. R.; Gallant, B. M.; Lee, S. W.; Shao-Horn, Y. Advanced Functional Materials2013, 23, 1037. Lee, S. W.; Yabuuchi, N.; Gallant, B. M.; Chen, S.; Kim, B. S.; Hammond, P. T.; Shao-Hom, Y. Nat Nanotechnol 2010, 5, 531. Kim, S. Y.; Hong, J.; Kavian, R.; Lee, S. W.; Hyder, M. N.; Shao-Horn, Y.; Hammond, P. T. Energ Environ Sci 2013. Jiang, H.; Lee, P. S.; Li, C. Z. Energ Environ Sci 2013, 6, 41. Wang, D.-W.; Li, F.; Liu, M.; Lu, G. Q.; Cheng, H.-M. Angewandte Chemie 2008, 120, 379. Xu, F.; Cai, R. J.; Zeng, Q. C.; Zou, C.; Wu, D. C.; Li, F.; Lu, X. E.; Liang, Y. R.; Fu, R. W. J.Mater. Chem. 2011, 21, 1970. Saetia, K.; Schnorr, J. M.; Mannarino, M. M.; Kim, S. Y.; Rutledge, G. C.; Swager, T. M.; Hammond, P. T. Advanced FunctionalMaterials 2013. Lee, S. W.; Kim, B. S.; Chen, S.; Shao-Horn, Y.; Hammond, P. T. J.Am. Chem. Soc. 2009, 131, 671. Sing, K. S. W.; Everett, D. H.; Haul, R. A. W.; Moscou, L.; Pierotti, R. A.; Rouquerol, J.; Siemieniewska, T. PureAppl Chem 1985, 57, 603. Rutledge, G. C.; Lowery, J. L.; Pai, C. L. JEng FiberFabr2009, 4, 1. Byon, H. R.; Lee, S. W.; Chen, S.; Hammond, P. T.; Shao-Horn, Y. Carbon2011, 49, 457. Toupin, M.; Belanger, D.; Hill, I. R.; Quinn, D. JPowerSources 2005, 140, 203. Poizot, P.; Laruelle, S.; Grugeon, S.; Dupont, L.; Tarascon, J. M. Nature 2000, 407, 496. Mannarino, M. M.; Rutledge, G. C. Polymer 2012, 53, 3017. 91 [22] [23] Krogman, K. C.; Zacharia, N. S.; Schroeder, S.; Hammond, P. T. Langmuir 2007, 23, 3137. Lowery, J. L.; Datta, N.; Rutledge, G. C. Biomaterials2010, 31, 491. 92 4. Conclusion and Future Recommendations In this thesis, the versatility of the spray-assisted layer-by-layer (LbL) technique has been demonstrated by depositing coatings of functionalized multi-walled carbon nanotube (MWNT) films on 3D porous electrospun fiber substrates, expanding its range of applications. MWINT/electrospun fiber composite electrodes were designed for applications in flexible chemical sensing (Chapter 2) and energy storage devices (Chapter 3). Conformal coating of MWNT films on individual electrospun fibers was achieved by applying a pressure gradient across the electrospun material to force flow through the porous membrane during repeated, sequential spray deposition of negatively and positively charged MWNT aqueous solutions. The resulting LbL-MWNT films have an interpenetrating network structure of unbundled MWNTs with nanoporous texture. In addition, the LbL-MWNT films contain no insulting binder, allowing fast electron transport throughout the entire electrode. MWNT loading and electrical conductivity of these electrodes can be tuned by simply changing the number of deposited layers. MWNT/electrospun fiber composite electrodes also retained their porous and flexible characteristics after MWNT deposition. This electrode platform was applied to generate two novel electrodes for applications in chemical detection and energy storage devices. The LbL-MWNT/electrospun fiber electrodes were utilized as a flexible chemiresistive sensor for real-time detection of a nerve agent simulant (dimethyl methylphosphonate (DMMP)) as described in Chapter 2. Chemiresistive sensors based on the engineered textiles display reversible responses and detection limits for DMMP as low as 10 ppb in the aqueous phase and 5 ppm in the vapor phase. However, the sensor selectivity still needs to be improved before it can be implemented in the field. One 93 promising strategy for improving selectivity is to use array sensing. A diverse set of molecular recognition receptors with different binding abilities can be functionalized on the MWNT/electrospun fiber electrodes to create a distinct response pattern that can be analyzed statistically when exposed to different classes of chemicals. Array sensors using multiple receptors and pattern recognition process can detect a wider range of analytes compared to individual sensors with only one type of receptor.1 2 Given a substantial improvement in sensor selectivity, MWNT/electrospun fiber electrodes can provide an attractive strategy for seamless integration of chemical sensor capability onto multifunctional textiles. Hierarchical porous (HP) structure of LbL-MWNT/electrospun fiber electrodes was investigated and employed to create HP-MWNT electrodes for high-rate energy storage devices. By removing the electrospun polymeric substrate, self-standing HPMWNT electrodes with 3D interconnected micron-size channels and mesoporous network of LbL-MWNT films can be obtained. Compared to similar MWNT electrodes without micron-size pores, HP-MWNT electrodes exhibit superior rate capability with an up to 10-fold increase in energy retention when the gravimetric power is increased from 100 W kg~' to 100 kW kg- 1 . Superior performance of HP-MWNT electrodes at high rates can be attributed to the synergy of large micro-size channels that serve as ion-buffering reservoirs and mesoporous network of MWNTs that facilitate fast ion transport. A number of strategies can be implemented to further increase electrochemical performance of HP-MWNT electrodes. First, the use of CNTs with fewer walls and longer tubes instead of MWNTs (-10 walls by TEM in this work) should have a positive impact on the performance by improving two parameters: (1) electronic conductivity of 94 the electrodes by reducing junction resistance due to longer tubes; (2) gravimetric accessible surface area due to fewer CNT walls. Recent report by Lee and coworkers 3 shows that CNT electrodes synthesized via a vacuum filtration method of oxidized fewwalled CNTs can have up to an order of magnitude higher electrical conductivity compared to LbL-MWNT electrodes. Second, incorporation of metal oxides, such as MnO2 and RuO 2 , that undergo fast surface redox (pseudocapacitive) reactions can increase the energy density of HP-MWNT electrodes. MnO 2 with the theoretical specific capacitance of ~1370 F/gMo2 based on a one-electron redox reaction per Mn atom4 is recommended due to lower cost compared to RuO 2 . The hierarchical porous structure of HP-MWNT electrodes should allow for efficient deposition of nanoscale MnO 2 due to the high porosity of electrodes. Lastly, spray-LbL conditions such as pH, number of bilayers that control electrode structure should be investigated and optimized to achieve electrode materials with high electronic conductivity and low ion transport resistance. References [1] [2] [3] [4] Wang, F.; Swager, T. M. J.Am. Chem. Soc. 2011, 133, 11181. Schnorr, J. M.; van der Zwaag, D.; Walish, J. J.; Weizmann, Y.; Swager, T. M. Advanced FunctionalMaterials 2013. Lee, S. W.; Gallant, B. M.; Lee, Y.; Yoshida, N.; Kim, D. Y.; Yamada, Y.; Noda, S.; Yamada, A.; Shao-Horn, Y. Energ Environ Sci 2012, 5, 5437. Toupin, M.; Brousse, T.; Belanger, D. Chem. Mat. 2004, 16, 3184. 95