Layer-by-Layer Carbon Nanotube Electrodes for Flexible

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