|UFDC Home||myUFDC Home | Help|
1 GRAPHITIC NANOMATERIALS AND POROUS 1 3 NM THICK MEMBRANES FOR NANOMOLAR DETECTION OF SMALL BIOMOLECULES By ABRAHAM BOATENG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMEN T OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 Abraham Boateng
3 To the memory of my Stepfather, O panin Robert Yaw Akuamoah Boateng
4 ACKNOWLEDGMENTS Praise be to the Lord God Almighty for He is my strength. He has been faithful to me and will forever be. He has caused the boundary lines to fall for me in pleasant places and crowned me with success. The Lord has done this, and it is marvelous in our eyes. To the glory of God! I thank my late step father, Agya Akuamoah so much for giving me the springboard in life; my mother Comfort Adwoa Kanin Afriyie for her love and prayers; my wife, Nana Nyarko for her prayers, great love, care and support; my sons, Nana Akuoko and Papa Sarpong for their company and the joy they always bring into our lives each day; my grandies, Lawyer Akuoko, Sisi, Daddy Kufuor my mother in law, Madam Gla dys and uncle in law, Rev. Sarpong for great support; my unc le, Nana Kwame Asante Otumfuo Funtunfunafudenkyemfunafuhene and aunt, Mama C harlotte for inspiring me to achieve greatness; my brothers and siste rs for their support; my best friend Vincent Mensah for urging me on throughout the years. I wish to express my heartfelt gratitude to my research advisor, Dr Anna Brajter To th, for her direction and encouragement. I owe much of my professional development in many respects to her guidance. I am also grateful to Dr. Young, Dr. Denslow Dr. Fanucci and Dr. Smith for offering to serve on my graduate research committee. I appreci ate the camaraderie of both the past (Dr. Mautjana, Dr. Kathiwala, Rachel, Andrews, Florian) and the present Toth G roup members (Gelin, Imran, Tim, Tho Da ni Elliot Larry and Annette ). I also appreciate the ass istance offered by Brian Todd (m achine shop) in designing the cell for carbon nanotube experiments as well as the training by Dr. Rajib Das Kumar (Rinzler Group), Bill and colleagues at the nanoscale research facility, on e lectron beam lithographic techniques. I also thank Dr. Williams for her time in proofread ing and constructive critique of my dissertation.
5 TABLE OF CONTENTS p age ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 8 LIST OF FIGURES ................................ ................................ ................................ ......................... 9 LIST OF SCHEMES ................................ ................................ ................................ ...................... 12 ABSTRACT ................................ ................................ ................................ ................................ ... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 15 Forms of Graphitic Nanomaterials ................................ ................................ ......................... 15 Carbon Fibers ................................ ................................ ................................ .................. 16 Carbon Nanotubes ................................ ................................ ................................ ........... 17 Graphitic Ultramicroelectrodes ................................ ................................ ....................... 19 Nanostructured Graphitic Ultramicroelectrodes ................................ .............................. 21 Nanoensemble Ultramicroelectrodes ................................ ................................ ...................... 23 Diffusion Fields at Nanoensembles ................................ ................................ ................. 24 Current and Kinetics ................................ ................................ ................................ ........ 26 Electrochemical Characterization of Nanostructured Graphitic Surfaces ....................... 27 Nanoporous Ultrat hin Membranes ................................ ................................ ......................... 28 Permeability of Ultrathin Membranes ................................ ................................ ............. 28 Diffusive Transport through Nanoporous Ultrathin Membranes ................................ .... 30 Fabrication of Adhered Porous Ultrathin Membranes at Nanoensemble Microelectrodes by Electropolymerization of Pyrrole ................................ ................. 31 Electrochemical C haracterization of Nanoporous Ultrathin Membrane Coated Nanoensemble Microelectrodes ................................ ................................ ................... 34 Analysis with Nanostructured Graphitic Ultramicroelectrodes ................................ .............. 36 Stability of Graphitic and Membrane Nanostructures ................................ ..................... 37 Fast Scan Cyclic Voltammetry with Background Subtraction ................................ ........ 38 Adsorpti on at Nanostructured Graphitic Ultramicroe lectrodes ................................ ....... 39 Study Overview ................................ ................................ ................................ ...................... 40 2 EXPERIMENTAL ................................ ................................ ................................ .................. 48 Reagents and Solutions ................................ ................................ ................................ ........... 48 Preparation and Purification of Chemicals ................................ ................................ ...... 49 Instrumentati on ................................ ................................ ................................ ....................... 50 Electrodes ................................ ................................ ................................ ............................... 51 Fabrication of Nanostructured Carbon Fiber Microdisks ................................ ................ 52
6 Fabrication of Porous Ultrathin OPPY Coated Nanostructured PAN Carbon Fiber Microdisks ................................ ................................ ................................ .................... 53 Fabrication of Single Walled Carbon Nanotube Film Microdisks ................................ .. 55 Analytical Measurements ................................ ................................ ................................ ....... 56 Fundamentals of Electrochemical Methods Used ................................ ................................ .. 56 Cyclic Voltamm etry ................................ ................................ ................................ ........ 56 Semiintegral Analysis ................................ ................................ ................................ ...... 69 Chronocoulometry ................................ ................................ ................................ ........... 70 Statistical Treatment ................................ ................................ ................................ ............... 72 Statistical Significance of Linear Correlation between Measured Variables .................. 72 Can the Median of Nanoensemble Dimens ions in OPPY Membrane Represent the Range? ................................ ................................ ................................ .......................... 73 Error Analysis ................................ ................................ ................................ .................. 73 3 EFFECT OF NANOSTRUCTURED CARBON FIBER MICRODISK ELECTRODES FRO M DIFFERENT PRECURSOR MATERIALS ON ANALYTICAL SENSITIVITY AND ELECTRODE KINETICS ................................ ................................ ............................ 81 Background ................................ ................................ ................................ ............................. 81 Results and Discussion ................................ ................................ ................................ ........... 84 Electrochemical Characterization of Nanostructured Carbon Fiber Microdisk Radius and Electron Transfer Kinetics ................................ ................................ ........ 84 Stability of Nano structured Carbon Fiber Microdisk Background Current .................... 87 Carbon Fiber Precursor Material and Adsorption of Dopamine and Uric Acid .............. 89 Correlation between Adsorption and Sensitivity of Dopamine and Uric Acid ............... 91 Determination of Electrode Kinetics after iR u Drop Correction of Peak to Peak Separation ................................ ................................ ................................ .................... 92 Conclusions ................................ ................................ ................................ ............................. 96 4 FABRICATION OF POROUS 1 3 NM THICK MEMBRANES OF OVEROXIDIZED POLYPYRROLE (OPPY) ON NANOSTRUCTURED PAN BASED CARBON FIBER MICRODISK ELECTRODES ................................ ................................ ............................. 105 Background ................................ ................................ ................................ ........................... 105 Results and Discussion ................................ ................................ ................................ ......... 107 Ferricy anide Kinetics at Nanostructured Carbon Fiber Microdisks .............................. 107 Characterization of Nanostructured Carbon Fiber Microdisks Area in Acetonitrile .... 112 Challenges to Overcome in Fabrication of Porous Ultrathin OPPY Membranes by Electropolymerization and Overoxidation ................................ ................................ 113 Effect of Electroactive Area of Nanostructured Surface on PY Electropolymerization ................................ ................................ ................................ 114 Electrochemical Characterization of OPPY Membrane Coated Nanostructured Carbon Fiber Microdisks ................................ ................................ ........................... 117 Effect of Electroactive Area of Nanostructured PAN Microdisk Surface on Porous Structure of PPY Membrane ................................ ................................ ...................... 121 Conclusions ................................ ................................ ................................ ........................... 122
7 5 LOW NANOMOLAR DETECTION OF SMALL BIOMOLECULES AT NANOENSEMBLE MICROELECTRODES COATED WITH POROUS 1 3 NM THICK OPPY MEMBRANES ................................ ................................ ............................. 130 Background ................................ ................................ ................................ ........................... 130 Results and Discussion ................................ ................................ ................................ ......... 132 Nanoensembles of Electroactive Elements at Bare and OPPY Coated Nanostructured PAN Carbon Fiber Microdisk Electrodes ................................ ........ 132 High Permeability of Porous Ultrathin OPPY Membranes at OPPY Coated Nanostructured Electrodes ................................ ................................ ......................... 137 Low Nanomolar Detection Limits of Dopamine and Uric Aci d at OPPY Coated Nanostructured PAN Microdisk Electrodes ................................ ............................... 138 Effect of OPPY Membrane Porosity and Thickness on Analytical Sensitivity ............ 1 42 Kinetics of Dopamine and Uric Acid Reaction at OPPY Coated Nanostructured PAN Microdisk Electrodes ................................ ................................ ........................ 142 Effect of OPPY Membrane Fabrication Procedure on Nanostructured Substrates of D ifferent PAN Materials and Implications for Analytical Sensitivity ....................... 143 Conclusions ................................ ................................ ................................ ........................... 145 6 NANOMOLAR DETECTION OF p NITROPHENOL VI A IN SITU GENERATION OF p AMINOPHENOL AT NANOSTRUCTURED CARBON FIBER MICRODISK ELECTRODES ................................ ................................ ................................ ..................... 158 Background ................................ ................................ ................................ ........................... 158 Results and Discuss ion ................................ ................................ ................................ ......... 160 Detection of p AP at Nanostructured PAN Carbon Fiber Microdisks .......................... 160 Nanomolar Detection of p NP at Nanostructured PAN Ca rbon Fiber Microdisks ....... 163 Evidence for Reduction of p NP to p AP and in situ Detection of p AP ...................... 163 Development of TRIS Buffer Me dium for Detection of p NP ................................ ...... 165 Optimization of p NP Sensitivity with Concentration of Supporting Electrolyte ......... 167 Conclusions ................................ ................................ ................................ ........................... 168 7 FABRICATION OF SINGLE WALLED CARBON NANOTUBE FILM MICROELECTRODES ................................ ................................ ................................ ........ 179 Background ................................ ................................ ................................ ........................... 179 Results and Discussion ................................ ................................ ................................ ......... 181 Nanostructured CNT Film Microdisk ................................ ................................ ........... 181 Determination of Electroactive Radius o f CNT Film Microdisk ................................ .. 181 Conclusions ................................ ................................ ................................ ........................... 183 8 SUMMARY AND FUTURE DIRECTIONS ................................ ................................ ....... 185 LIST OF REFERENCES ................................ ................................ ................................ ............. 188 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 201
8 LIST OF TABLES Table p age 1 1 Dependence of voltammetric characteristics on nanoensemble dimensions, diffusion layer thickness and scan rate ................................ ................................ ............................. 43 3 1 Carbon fiber material data ................................ ................................ ................................ 98 3 2 Surface properties of carbon fiber microdisk electrodes ................................ ................... 98 3 3 Sensitivity and kinetics of dopamine and uric acid ................................ ........................... 99 3 4 Resistance, background current and peak to peak separation ................................ ........... 99 4 1 Properties of electrodeposited PPY membranes at nanostructured PAN electrodes. ...... 123 5 1 Physical proper ties of bare and OPPY membrane coated nanostructured PAN microdisk electrodes ................................ ................................ ................................ ........ 147 5 2 Sensitivity and peak to peak separation of dopamine and uric acid at OPPY coated nanostructured PAN microdisks ................................ ................................ ...................... 147 6 1 Detection sensitivity for p aminoophenol, p nitrophenol, and dopamine in different buffer media at bare nanostructur ed PAN electrodes ................................ ...................... 169 6 2 Average peak potentials of fast scan voltammograms of p aminoophenol and p nitrophenol ................................ ................................ ................................ ....................... 169
9 LIST OF FIGURES Figure p age 1 1 Illustration of surface functionalities of oxygen in pitch and of both oxygen and nitrogen in PAN carbon fibers ................................ ................................ ........................... 43 1 2 A struct ure of graphite show ing basal and edge planes of a carbon material .................... 44 1 3 Illustration of nanoensemble and regular (hexagonal) array of active and nonactive elements and their dimensions on a microdisk ................................ ................................ .. 44 1 4 Representation of diffusion profile categories at nanoensembles of active elements and corresponding shape s of voltammogram s produced. ................................ .................. 45 1 5 Background current subtraction in fast scan cyclic voltammetry ................................ ...... 46 2 1 Set up for pyrrole purification ................................ ................................ ........................... 75 2 2 Schematic instrumental set up for mi croelectrodes at slow scan rate ............................... 76 2 3 Schematic instrumental set up for microelectrodes at fast scan rate ................................ 77 2 4 Representation of the carbon fiber microdisk electrode ................................ .................... 78 2 5 E lectron beam patterned CNT film microelectrode ................................ .......................... 79 2 6 Illustration of cell set up for electrochemical measurements with CNT film microdisk ................................ ................................ ................................ ........................... 79 2 7 Triangular waveforms and cyclic voltammograms at a microelctrode .............................. 80 3 1 Slow scan voltammetry of ferricyanide and background current at nanostructured PAN T650, PAN HCB and PCH P25 microdisk electrodes ................................ ............ 100 3 2 Semiintegrals of background subtracted fast scan voltammograms illustrating correlation between adsorption and sensitivity of nanostructured microdisks ................ 101 3 3 Background subtracted fast scan voltammograms at nanostructured microdisks showing the effect of carbon fiber material on current density of dopamine .................. 102 3 4 Background subtracted fast scan voltammograms at nanostructu red microdisks showing the effect of carbon fiber material on current density of uric acid. ................... 103 4 1 Slope of E vs log[( i L i )/ i ] plot of ferricyanide steady state current as a function of elec trode background current at nanostructured PAN carbon fiber microdisks .............. 124 4 2 Plots of electrode background charge vs background current densities, illustrating correlation between electroac tive surface areas in acetonitrile and in aqueous buffer .... 124
10 4 3 Anson plots of chronocoulometric curves for bac kground, and first and second pyrrole electropolymerization steps at nanostructu red PAN carbon fiber microdisks .... 125 4 4 Correlation between background charge densities in acetonitri le solutions with and without pyrrole at nanostructured PAN microdisk electrodes ................................ ........ 125 4 5 Amount of electrodeposited polypyrrole as a function of electroactive surface area of nanostructured PAN microdisk electrodes ................................ ................................ ....... 126 4 6 Ferricyanide steady state current at bare and OPPY membrane coated nanostructured surfaces of PAN HCB microdisk electrodes ................................ ................................ .... 126 4 7 Fraction of ferricyanide limiting current remaining at O PPY membrane coated nanostructured PAN microdisks as a function of the membrane thi ckness ..................... 127 5 1 Slow scan voltammograms of ferricyanide at naostructured PAN microdisk electrodes before and a fter fast scan voltammetric measurements of dopamine ............. 148 5 2 Fit of Butler Volmer kinetics to slow scan voltammograms of ferricyanide at bare and OPPY coated nanostructured PAN microdisk elec trodes ................................ ........ 148 5 3 Correlation between number and radius of active elements (pores in OPPY membranes) at the OPPY coated nanostructured PAN microdisk electrodes ................. 149 5 4 Background subtracted fast scan voltammograms illustrating nanomolar measurements of dopamine at OPPY coated nanostructured PAN microdisks .............. 150 5 5 Semiinte grals of background subtracted fast scan voltammograms of dopamine and uric acid at bare and OPPY coated nanostructured PAN HCB mircodisk electrode ...... 151 5 6 Dopamine and uric acid calibration p lots at bare and OPPY coated PAN HCB microdisk electrodes ................................ ................................ ................................ ........ 152 5 7 Background subtracted fast scan voltammograms of uric acid at OPPY membrane coated nanostructured PAN HCB microdisk electrode ................................ .................. 153 5 8 Sensitivity for dopamine as a function of membrane porosity and membrane thickness at OPPY coated nanostructured PAN microdisk electrodes ............................ 154 5 9 Background subtracted fast scan voltammograms of dopamine at OPPY membrane coated nanostructured PAN microdisk electrodes ................................ ........................... 155 5 10 Repetitive chronocoulometric curve s for background charge at nanostructured PAN microdisk electrodes ................................ ................................ ................................ ........ 156 5 11 Slow scan voltammogram of ferricyanide at OPPY membrane coated nanostructured PAN microdisk electrodesbefore and aft er fast scan measurements of dopamine .......... 156
11 6 1 Background subtracted fast scan voltammograms of p aminophenol illustrating submicromolar detection at nanostructured PAN T650 microdisk electrode s ................ 170 6 2 Background subtracted fast scan voltammograms of p nitrophenol showing nanomolar detection at nanostructured PAN T650 microdisk electrodes ...................... 171 6 3 Evidence for detection of p nitrophenol as p aminophenol confirmed by overlap of background corrected fast scan voltammograms of their separate solutions .................. 172 6 4 Shift of peak potentials of p aminophenol with pH of medium ................................ ...... 173 6 5 Slow scan voltammograms of ferricyanide after background current measurement in phosphate buffer at different pH at bare nan ostructured PAN microdisk electrodes ...... 174 6 6 Effect of buffer medium on slow scan voltammogram of ferricyanide and background current at bare nanostructured PAN microdisk electrodes ........................... 175 6 7 Effect of supporting electrolyte on background current and electrical conductivity of TRIS buffers ................................ ................................ ................................ ..................... 176 6 8 Effect of supporting el ectrolyte concentration on magnitude of anodic peak current of p nitrophenol in TRIS Ac/NaAc buffer at nanostructured PAN electrodes ................ 177 7 1 AFM images of CNT film at different stages of the m icrodisk electrode processing ... 184 7 2 Slow scan voltammograms of ferricyanide and dissolved oxygen at CNT film microdisk electrode ................................ ................................ ................................ .......... 184
12 LIST OF SCHEME S Schem e p age 1 1 Proposed reactions for electrochemical etch of graphitic materials ................................ .. 46 1 2 A mechanism for electropolymerization of pyrrole to form polypyrrole .......................... 47 2 1 Reaction for preparation of tetra n butylammonium perchlorate from the bromide analogue ................................ ................................ ................................ ............................. 80 2 2 Reduction of ferricyanide to ferrocyanide ................................ ................................ ......... 80 3 1 Redox reactions of dopamine (cation) and uric acid (anion) at pH 7.4 ........................... 104 4 1 Schematic representation of axial and cross sectional view of PAN carbon fibers ........ 128 4 2 Proposed structure of overoxidized polypyrrole ................................ .............................. 128 4 3 Models used in calculations of PPY and OPPY membrane thickness ............................. 129 5 1 Illustration of analyte transport and preconcentration at porous membrane coated electrode ................................ ................................ ................................ ........................... 157 5 2 Model of the partially blocked electrode ................................ ................................ ......... 157 6 1 Detection of p nitrophenol by electrochemical reduction to p aminophenol and subsequent measurement of analytical signal due to p aminophenol .............................. 178
13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Req uirements for the Degree of Doctor of Philosophy GRAPHITIC NANOMATERIALS AND POROUS 1 3 NM THICK MEMBRANES FOR NANOMOLAR DETECTION OF SMALL BIOMOLECULES By Abraham Boateng May 2012 Chair: Anna Brajter Toth Major: Chemistry The research p resented in thi s dissertation was undertaken to enhance analytical sensitivity of electrochemical biosensors for nanomolar detection. The effects of properties of graphitic nanomaterials, including carbon nanotubes and nanofeatures of polyacrylonitrile (PAN) and pitch b ased carbon fiber micro disks, were investigated as well as i mprovement in sensitivity with porous ultrathin (1 3 nm thick) membranes of overoxidized polypyrrole (OPPY) Both the n itrogen defects and the disorder of the carbon fibers determine the nanostruc tures fabricated on their microdisks by an electrochemical etch method. In analytical determinations by fast scan voltammetry (FSCV) at 500 V s 1 electrode sensitivity for dopamine and uric acid is adsorption controlled, as confirmed by semiintegral analy sis, and is higher for dopamine t han for uric acid. The a nalytical sensitivity for biomolecules is higher at nanostructured PAN microdisks, but the kinetics of the biomolecule reactions is slower than at pitch. Heterogeneity in surface nanostructures of P AN microdisks results in differences in the electrodeposited membrane structures of OPPY. Slow scan vo ltammetry of ferricyanide allowed interrogation of the coverage of the microdisks with the porous ultrathin OPPY membranes. The electrochemical characteri zation verified formation of nanoensembles and allowed estimation of
14 the dimensions of active sites (membrane pores) to be in the range of 0.04 to center separation of the sites (pores) of ca. 0.7 A model reflecting the porous stru cture of the membrane was proposed. Sensitivity enhancement was realized at the fabricated porous ultrathin OPPY membranes adhered to the nanostructured PAN microdisks. The highly permeable me mbranes at these surfaces provide the best limits of detection f or dopamine (15 nM) reported to date at microdisk electrodes of small 2 ). T he high sensitivity for dopamine at OPPY coated electrodes is due to its preconcentration on the membrane and high diffusive flux through the porous membrane which shows permselectivity against uric acid. In new investigations of nanostru ctured PAN microdisks for applications in electrochemical immunosensing, nanomolar detection of p nitrophenol ( p NP) was demonstrated by FSCV at 500 V s 1 P reliminary investigations of a well characterized single walled carbon nanotube film patterned into microd isk electrodes demonstrated possi bilities for electrochemical biosensing.
15 CHAPTER 1 INTRODUCTION Forms of Graphitic Nanomaterials When used as ultramicroelectr o des (UMEs) in electrochemical biosensing, graphitic nanomaterials in the form of carbon nanotubes and nanofeatures of carbon fibers (i.e. nanofibers and nanosheets) can provide high analytical sensitivity and enable nanomolar detection limits. These figures of merit are highly desirable in applications involving in vivo sensing, 1 measurements of single cell release, 2 microdetectors, 3, 4 SECM imaging of cellular release 5 and fundamental studies of prototype redox reactions inside biological cells 6 In addition, their applications in investigations of enzymatic kinetics 7, 8 and sensitive large biomolecule detection 9, 10 have been demonstrated. These qualities derive in part, from a combination of their nanometer and micron sized dimensions that result in improved mass transport and higher analytic al sensitivity than other electrode materials such as gold and platinum. 11, 12 The popularity of graphitic materials in electrochemical biosensing is due also to their biocompatibility, their rich surface chemistry of oxygen functionalities, a wide electrochemical potential window in aqueous solution, and ready availability. 13 Of the graphitic nanoma terials that can be fabricated into UMEs, carbon fiber based materials are popular for practical electrochemical biosensing applications. 1, 14 Carbon nanotubes are also emerging as UME materials, but issues related to material purity still need to be addressed. 15 In spite of their popularity as electrochemical biosensors, the rationale behind the choice of a given graphitic nanomaterial for an electrochemical biosensor is rarely stated in the liter ature. Graphitic nanomaterial produced from different precursors and by different processes can have different properties, which in turn, affect their electrochemical performance. Thus, an understanding of the relationship of material property to analytica l sensitivity should provide
16 fresh insights into strategies for optimizing graphitic nanomaterial properties for highly sensitive electrochemical biosensing, the primary goal for which the investigations herein were conducted. Carbon Fibers Carbon fibers t ypically have diameters of 5 16 and le ngths on the order of many millimeters. 17 They can be produced from various carbonaceous materials, but the common precursor materials include polyacrylonitrile (PAN), mesophase pitch and viscose rayo n, 17 all of which provide the required mechanical properties for the desired end use. Today, PAN is the most commonly used precursor, followed by mesophase pitch. 18 PAN and pitch based carbon fibers are manufactured by three main stages. The first involves spinning the polymeric precursors into fibers under a tensile stress to improve the orientat ion of molecular structures along fiber axis. In the second stage, oxidative stabilization in an oxidizing atmosphere at 180 400C is carried out to maximize the final properties of carbon fibers. This procedure increases the stiffness of the precursor fib er through dehydrogenation, cross linking mediated by oxygen functionalities, and cyclization in the case of PAN, and makes the fibers thermally stable for high temperature carbonization. Finally, the stabilized fiber is carbonized at ~1500C in an inert a tmosphere to remove heteroatoms and produce a graphitic material. 17 19 Because of the structural defects produced at this stage, the graphitic material is said to be highly disordered. The crystalline properties, specifically, the microcrystallite sizes can be increased and interplanar distance between graphene sheets, decreased by graphitization, which involves heat treatment at temperatures up to 3000C 17 to form low disorder ed graphite. Consequently, the desired mechanical, electrical and thermal properties can be improved by controlling the crystallinity of the material.
17 It has been suggested that different crystalline structures result in different electro chemical behavior. 16 Perhaps the most impor tant properties worth considering are the surface chemistry and nanostructure, because of the interfacial nature of heterogeneous electrochemistry. The surface of pitch typically has oxygen functionalities including carbonyl, carboxyl, hydroxyl and quinony l (Figure 1 1A). 13 In addition to these, PAN carbon fibers have quaternary like, pyridinic and pyrrolic nitrogen functionalities, 20 as well as nitro and nitroso groups, 21 as shown in Figure 1 1B. The different chemistries of the precursors can affect the orientation of graphen e sheets and, thus, the nanostructural features produced in the resulting carbon fiber material. 19 Sheet like morphologies for pitch 18 and intermingled nanofibers for PAN 22 24 carbon fibers have been verified by microscopic techniques. Pitch based fibers typically have r adial and concentric disk cross sections, while tho se of PAN can be random. Clearly, the possible implications of the precursor material and heat treatment history, which affect both the surface chemistry and the nanostructure, underscore the important considerations for the choice of a carbon fiber materi al as an electrochemical biosensor. Carbon Nanotubes Carbon nanotubes have dimensions ranging from ~1 to 30 nm diameter 15 depending on whether they are single double or multi walled, and can be up to many tens of microns long. 25 Recently, carbon nanotubes as long as 5 mm have also been reported. 26, 27 Structurally, carbon nanotubes can be considered as graphene sheets that are rolled into tubes. While car bon nanotubes are hollow nanostructures, carbon fibers can have filled nanostructural features similar 28 Carbon nanotubes can be produced by electric arc discharge 29 and laser ablation 30 of carbon precursors, as well as chemical vapor deposition (CVD) methods 25 from carbonaceous
18 vapors (e.g. methane, acetylene, xylene and pyridine) onto transition metal nanoparticle catalysts such as iron and cobalt. These catalysts commonly remain as impurities in the graphitic nanomaterials produced. Of these methods, CVD has been suggested to yield materials with lowest content of catalyst impurities. Nonetheless, significant variability in the quality (i.e. conten t of catalyst impurities) and structure (in the form of single double or multi walled nanotubes) of these materials produced from different precursors and under different processing conditions can be expected. In spite of the excitement about the pros pects of CNT as advanced electrochemical biosensors, concerns have been raised about the purity of CNTs employed in electrochemical studies, with evidence presented that catalytic metals (required for CNT growth) are responsible for some observed electroch emical responses. This suggests the requirement for high quality standards requiring purification and thorough characterization of the material prior to electrochemical applications. The most common purification method is nitric acid washing of the catalys ts, but this has been reported to be inefficient in removing residual catalyst and damages the nanotubes by shortening them and creating defects on their sidewalls. The use of HCl has been suggested to be efficient in removing catalyst residues, in additio n to being mild on the nanotube structure. 15 Alternative purification strategies have focused on electrochemical removal of metal catalyst without apparent detrimental effect on the electrochemical properties of the electrodes. 31 In addition, passivation of iron as Fe 2 O 3 has been suggested. 32 Although, more investigations are being conducted to lessen and if possible remove completely, the catalyst residues, it is expected that template approaches for catalyst free synthesis 33 35 will more directly address this issue.
19 Graphitic Ultramicroelectrodes Because of their small dimensions, ranging from nm to a few 2 Such electrodes allow for improved mass transport associated with the edge effect. 36 In addition, they are beneficial for rapid measurements on micro to millisecond time scales due to their short time constants and low Ohmic drops. 37 Their applic ations in measurement in small environments have also been demonstrated as stated above. As shown in Figure 1 2, the plane of hexagonal rings of sp 2 hybridized carbon atoms that form graphene sheet is referred to as the basal plane; the plane of edges of g raphenes stacked parallel to each other, is called the edge plane. 16 With the exception of single crystal graphite, highly ordered pyrolytic graphites and carbon nanotube tips, a typical surface of other forms of graphitic materials can be expected to have both basal and edge planes. At both carbon fibers and carbon nanotubes, as in the case of other graphitic materials, it has been suggested that the electron transfer kinetics of redox process at the edge is faster than that at the basal plane of a carbon material 13 Consequently, electrode surface having a higher density of edge plane such as carbon fiber microdisk, 38 edge plane 39 and rough 16 pyrolytic graphite can provide optimum biosensing capabilities. Formation of edge plane defects by acid treatments, on carbon nanotube sidewalls can also offer such possibilities, but this can be undermined by the increased resistance of the material. Graphitic materials in the form of intact graphite 40 and carbon paste 41 have been used as electrodes in ele ctroanalysis since the mid 1950 s, but the use of carbon fibers as working electrodes 42 was initial ly reported in the late 1970 s. Carbon fiber microelectrodes can be of disk elliptical and cylindrical shaped. Typically, part of the fiber is drawn/aspirated into a micropipette tip or glass capillary and seale d with epoxy. Electrical connection between the
20 carbon fiber and a metallic wire can commonly be made with either silver epoxy 37 colloidal graphite 43 or an electrolyte sol ution. 44 The protruding fiber tip can then be polished at an angle of ~90, <90 or cut to desired lengths with a scalpel to form disk elliptical or cylindrical shaped electrode, respectively. Unlike carbon fib er electrodes, the earliest report of carbon nanotube electrode material was in the form of a paste macroelectrode in the mid 1990 s. 45 Nonetheless, recent advances in nano and micro fabrication techniques have made it possible to pattern the nanotubes as UME s. 46, 47 Carbon nanotube UMEs in the form of either networks 46, 48 or bundles 49, 50 lithographically fabricated on the growth substrate are common, but th e possible presence of catalyst impurities can contribute to the electrochemical signal. Even when purified on the growth substrate by typical processes such as acid washing, some loss of intact nanotubes through dislodging can be expecte d leading to irrep roducibility in the electrode processing Alternatively, the nanotube can be transferred from the growth substrate purified and then fabricated as electrodes. In fact, individual carbon nanotubes fabricated into nanoelectrodes 47, 51 have been demonstrated, as well as 50 52 and bundled nanotubes 53 UMEs. Transferrable CNT networks/films can commonl y be formed by vacuum filtration of a nanotube suspension onto a membrane, 54 56 although electrostatic spraying of the material onto a substrates has also been reported. 57 Electri cal contact with the CNT network can be made with thermall y evaporated metal onto a section of the network prior to lithographic patterning of the UME. 58 It is clear from above that UMEs fabricated from tran sferrable, but purified or template synthesized, carbon nanotube networks can be promising as practical electrochemical biosensors as demonstrated in this work with single walled carbon nanotube films.
21 Since knowledge of the CNT network/film electrode geom etric area is desired, lithographic patterning becomes an indispensable tool, although the procedure can affect the integrity of the material. Some lessons can be learned from lithographic expo sure of carbon nanofibers (50 500 nm), another form of graphiti c nanomaterials. Fabrication of their nanoelectrode ensembles highlights some advances in lithographic and processing techniques relevant for maintaining the integrity of graphitic nanomaterials and achieving good electrochemical properties at such surface s. 59 61 It has been shown that some photo resist material can remain coated onto the CNT, resulting in deteriorated electrochemical performance. Evaporation of SiO x layer before photoresist coating and subsequent p atterning has been shown to reduce such contamination effects on electrochemical performance of nanoelectrodes, 47 as is the case for microelectrode a rrays. 50 Not only are lithographic lessons for patterning of quality carbon nanotubes microelectrodes important, but also the electrochemical properties of the carbon nanofiber can provide fresh insights into under standing the electrochemical behavior of nanostructured carbon fibers. For example, the diameters of the nanofibers are similar to those of the nanostructural building units of PAN carbon fibers. 23, 24, 62 Nanostr uctured Graphitic Ultramicroelectrodes There has been significant interests in nanostructured surfaces in recent years because their high surface to volume ratio in combination with further improved mass transport to their surfaces have led to impressive L fM range. 63 66 At graphitic microelectrodes, nanofeatures of both PAN 38 and pitch 67 carbon fibers have been exploited for electroanalytical applicatio ns. At a microdisk cross section of PAN, nodule nanofeatures are randomly arranged whereas nanosheets of radial and onion structure are typical of a pitch carbon fib er. 17
22 It has been demonstrated that these nanofeatures can be exposed/enhanced in a fashion top down approach involving mild electrochemical etch, 10, 38, 68 which is believed to occur through the reactions shown in Scheme 1 1. Formation of surface oxides (Scheme 1 1 V .), 37 which are undesirable, can be minimized by proper cho ice of etching conditions. 38 At present, electrochemical etch in aqueous medium, such as one developed by Bravo, is the most practical and reproducible method for exposing such nanofeatures. This procedure involves continuous potential cycling between 1.0 and +1.5 V vs. SCE at 10 V s 1 for 30 min in phosphate buffer. 38 It is also important to mention that electrochemical etch not only exposes surface nanostructures, but al so activates the electrode surface, possibly by creating more edge defects on the basal plane of graphite. In this work, In an electrochemical etch procedure, it was shown that PAN etches at a faster rate than pitch based carbon fiber. 69 Consequently, under the same conditions, t he extent of nanostructuring can be expected to be higher for PAN than pitch carbon fiber UMEs. Also, among carbon fibers fabricated from the same precursor (e.g., PAN or pitch), even though the more disordered material can etch faster, it is also more sus ceptible to overoxidation. 67, 70 These characteristics emphasize the need to tailor the electrochemical etch procedure to surface nanostructuring of a given fiber material, so as to maximize sensitivity in electroc hemical biosensing. By forming a carbon nanotube network/film in a bottom up approach, the surface nanostructure shows randomly arranged individual/bundled nanotubes and voids. 46, 56 In lithographic patterning of their microelectrodes, some fraction of the carbon will inevitably be passivated or partially covered by processing materials. Even if the network is activated a priori
23 by such methods as acid washing, the influence of such undesirable attributes of litho graphic processing can still be expected to manifest in poor electrochemical performance. As emphasized above, lithographic procedures for fabricating carbon nanotube network/film UMEs should aim for complete removal of the processing material. This requir ement is necessary to maintain the integrity of the material and allow total access to the exposed geometric area during electrochemical measurements, as has been demonstrated for carbon nanofiber ensembles. 59 61 N anoensemble Ultramicroelectrodes At nanostructured graphitic UMEs, the parts of the surface available for electron transfer constitute the active elements, with the rest of electrode surface being nonactive 71 as illustrated in Figure 1 3A according to model of Scharifke r. 72 In electrochemical measurements, analytes access the active elements. For nanostructured carbon fiber UMEs, nonactive elements typically comprise sur face oxides (insulating elements); for carbon nanotubes, sites covered with lithographic materials are also non active. 47 For both nanostructured carb on fiber microdisks 68 and carbon nanotube films/networks, 46 their inherent surface pores also contribute to the nonactive elements. The random array of nanosized active ele ments that are separated by nonactive elements constitutes nanoensembles on nanostructured graphitic surfaces. The formation of a nanoensemble of carbon characterizes the electrochemical signal as predicted by theory. 73 It is important to mention that microelectrodes of carbon nanofiber ensembles 59 61 and carbon nanotube arrays, 50 as well as nanoelectrode ensembles of gold in nanopor ous polycarbonate membrane, 74 have been reported. In those reports, the microelectrodes were fabricated by controlled manipulation of the dimensions of the active elements (i.e. gold or carbon). Therefore, the elect rochemical behavior of these surfaces can provide some
24 explanations of the observations that are associated with changes at nanostructured carbon fiber and carbon nanotube network surfaces during electroanalysis. Diffusion Fields at Nanoensembles At nanoen sembles, 75, 76 as in the case of fabricated regular arrays (Figure 1 3B) of microensembles, 77 the relative dimensions and density of the active elements have significance for the diffusion behavior (radial or linear or mixed) that occurs. The sizes of active elements, R a and their average center to center separation, 2 R 0 at an UME of geometric radius, r are illustrated in Figure 1 3 according to the models of Amatore 73 and Scharifker 72 For a cyclic voltammetric experiment, the diffusion layer thickness, and the characteristic time of the experiment, t are given by 73, 76 (1 1) and (1 2) where D 0 is diffusion coefficient (cm 2 s 1 ) R is the molar gas constant (J mo l 1 K 1 ), T is the absolute temperature (K), n is number of electrons, F is the Faraday constant (C mol 1 ) and is scan rate. According to Equation 1 1, the diffusion layer thickness increases with the experimental time scale. 36 Consequently, a combination of the time dependent diffusion layer thickness, and the relative dimensions of the nanoensemble elements determines whether a radial (i.e. > R a ) or linear (i.e. < R a ) diffusion field is created at the active elements The same set of conditions, > r and < r for radial and linear diffusion, re spectively, applies to an electrode for which all parts of the surface are active. 36 Since determines t and therefore a critical scan rate, c exists at which the diffusion field transitions from radial to linear or vice versa 7 6
25 As shown in Table 1 1, the theoretical characteristics of the voltammetric responses that can be expected at both micro and nanoensembles, depend on the relative dimensions of R a 2 R 0 and as well as on and c 76, 78 The corresponding categories (A, B and C) of diffusion behavior can be illustrated as shown in Figure 1 4 and these produce the different voltammetric characteristics. When linear d iffusion forms (Figure 1 4A), a peak shaped voltammogram (Figure 1 4D) is produced since the analyte is consumed at a faster rate than its diffusion to the el ectrode surface. On the other hand a steady state voltammogram (Figure 1 4E) forms in the case of radial diffusion (Figures 1 4B and C) since the analyte is consumed at the same rate as its diffusion to the electrode surface. As illustrated in Figure 1 4, the diffusion behavior of categories A and B are independent and their current responses are gi ven respectively by 76 (1 3) and (1 4) where i p and i L are peak and limiting currents (nA), r espectively, A Ra is the area (cm 2 ) of an active site of radius, R a in (m), C 0 is bulk concentration of analyte (M), N is active site density (cm 2 ), A is the geometric area of the electrode (cm 2 ) and n F D 0 and have their usual meanings. The contant value in Equation 1 3 has units of C V 1/2 mol 1 It can be verified mathematically, that the diffusion category B (Figure 1 4) produces the highest faradaic to background current ratio, as a result of which the design of microelectroelectrode array is tail ored to achieve establishment of this diffusion behavior. 78
26 In category C, the diffusion fields at the individual active elements overlap and the current produced, is proportional to the area of the electrode. For a steady state response, this current can be approximated to that of a microdisk as 74 (1 5) where r is the radius of the electrode (m). The diffusion behavior of category C was observed in this work and Equation 1 5 was used to characterize the system under investigation. Current and Kinetics At nanostructured graphitic UMEs where diffusion fields are created at the individual active elements, the total current at such surface is the sum of the contributions from the individual elements 74 as shown in Equations 1 3 and 1 4. The maximum that can be measured corresponds to the geometric area of the unblocked microelectrode, 76, 79 At nanoensemble UME, this maximum current is produced when significant/total overlap of the diffusion fields at the active elements occurs 46, 75 With partial overlap of the fields, competition for a nalyte in the zone between the active elements reduces the magnitude of the limiting current. 80 It is well known that a decrease in fractional coverage of the active elements due to blockage of the electrode surface results in a decrease in both the magnitude of the limiting current and the apparent heterogeneous rate constant relative to those at an unblocked surface. 73 It has also been shown theoretically that the change of shape of the active elements from oblate to prolate has significance for mass transport and, thus, directly determines the electrochemical behavior (magnitude of current and of heterogeneous electron transfer kinetics) at the electrode. 81 Therefore, for the same coverage with activ e elements having different geometries on different electrode surfaces, differences in their electrochemical behavior should be expected.
27 In the literature, theoretical investigations of the relation between nanoscopic active elements and the current and h eterogeneous electron transfer kinetics at nanoensemble UMEs is rare. In the early 1990s, C.G Phillips presented a theoretical treatment for a system that is most closely related to nanoensemble UMEs. In this theory, a model consisting of small active elem ents on an insulating UME was developed to understand the behavior of the active element assemblies and the extent of insulating effects. 79 Simulations on these nanoensemble systems have recently been investigated by Zoski and co workers. 74, 76 On the contrary, significant theory and simulations exist in the literature on microelectrode arrays on macroelectrodes. 73, 78, 80, 82 The on e that underpins the outcome of the other theories is the theory of charge transfer at partially blocked surfaces developed by Amatore et al 73 Although it was developed for and has a strong predictive power for ele ctrochemical behavior at microensembles, its predictions are also consistent with empirical data at nanoensemble UMEs. 74 In the same manner, an independent microensemble theory 78 could adequately simulate an experimental observation at a nanoensemble. 83 These theories were adapted to characterize the nanoensemble behavior of the fabricated surface nanostructures in this work. It is w orth mentioning that a review 84 and a theory 85 of electrochemistry at individual nanoscopic electrodes have been reported recently. Moreover, a theory for evaluating the kinetic activity of platinum (Pt) nanoparticle ensembles on a glassy carbon macroelectrode has been reported most recently. 86 These recent reports may indicate that related interests in theoretical investigations of the nanoensembles on UMEs could be in the pipeline. Electrochemical Characterization of Nanostructured Grap hitic Surfaces Electrochemical methods, voltammetry in particular, directly provide information on the fraction of the surface area that is electroactive. In effect, these studies allow for estimation of the fraction of the electrode surface that is inacce ssible to a probe. It is the most practical method
28 for characterizing nanoensembles at UMEs, since the small area of such surfaces benefits from the interfacial nature of the technique. 38, 46 In addition to its cap abilities for verifying modifications at nanostrucutured graphitic surfaces, voltammetry provides rapid and direct monitoring of changes (such as from fouling by adsorbed matter) that can occur at the bare nanoensembles during use in electrochemical biosen sing. 87 Nanoporous Ultr athin Membranes The rationale for designing a nanoporous ultrathin film /membrane coated electrode is to lower the limits of detection (LOD) and enhance the selectivity for biomolecules at such surfaces relative to the bare surface. 44, 88 92 Nanoporous ultrathin membranes can also be free standing, and they typically have pores and thickness of few nanometers, 93 bu t they must adhere to the electrode surface in order to realize their practical application as electrochemical biosensors, as is shown in this work. 71 Note also that a more compact ultrathin membrane coated electrochemical sensor for size selective detection of peroxide has also been reported. 90 Permeability of Ultrathin Membranes From the standpoint of analytical sensitivity, ultrathin membranes are desired because of their short response times and high permeability 88 90, 93, 94 that are beneficial for electrochemical sensing. In addition, it has been suggested that a porous material having surface functionalities can result in increased local concentration of analytes, 95, 96 and, thus, offer possibilities for lowering LOD. More recently, fresh theoretical insights have been gained into the dependence of permeability of nanoporous ultrath in membranes on their geometric parameters. 93 According to this theory, the permeability, k (cm s 1 ), of such membranes is given by (1 6)
29 where, R a is the average pore radius (nm), N is density of pores (cm 2 ) and D 0 is bulk solution diffusion coefficient (cm 2 s 1 ). Equation 1 6 has been shown to reasonably predict the correlation between k and D 0 of different probes. The theory also shows that k is independent of thickness when the membrane is ultrathin (i.e. 10 15 nm thick). More importantly, Equation 1 6 accounts for the fact that a membrane design strategy for producing large pore size and high density should offer possibilities fo r achieving high permeability. Therefore, a tailored porous ultrathin membrane design can offer possibilities short response time 97 and high analytical sensitivity at membrane coated electrodes. In this work, a strategy was developed for fabrication of membranes coated onto nanostr uctured carbon fiber microdiks to achieve such analytical goals. In con trast, the permeability of a thick membrane is controlled by translocation and is given by 98 (1 7) where is the membrane porosity (%) and m is the membrane thickness (nm). This form of the equation is similar to the classical membrane permeability, P m equation given by (1 8) where C ] m / [ C ] 0 ); [ C ] m and [ C ] 0 are analyte concentrations in the membrane and bulk solution, respectively, and D m is the diffusion coefficient in the membrane (cm 2 s 1 ). Intuitively, it should be expected that a highly porous membrane (i.e. high in Equation 1 7) s hould correspond to a high parti tion coefficient (i.e. high in Equations 1 8), and thus a high permeability in both cases. This agreement with the classical equation can be a corner stone of the recent theory. 98
30 Diffusive Transport through Nanoporous Ultrathin Membranes In general, diffusive transport of probes through nanopores is governed by pore radius, le ngth, surface charge density, 96 surface charge, 99 and the size of probe, 98, 100, 101 as well as the ionic strength and pH of the electrolyte. 93, 99, 102 It has been demonstrated experimentally that permselec tivity of nanopores (radius <15 nm ) in silica against hexacyanoferrate (III) 93 and in sulfonic acid covered silica against hexachloroiridate (III) 99 can be modulated by changing the pH and ionic strength of the electroly t es. The selectivity derives from electrostatic effects, for which the Debye screening length (~1 nm) of the electric field inside the pore 96 can be very important, especially for these small pore radii. In this mechanism, probe ions having the same charge as the nanopore wall are expelled, the reby limiting diffusion through the free solution inside the nanopores 93, 99 while diffusion of oppositely charged species is facili tated. A difference in the size of probes, allowing for size exclusive selectivity 90, 94, 100 is also an interesting mechanism. This mechanism can protect the underlying substrate surface of membrane coated electrodes from adhesion of cellular debris and direct molecular adsorption, factors beli eved to contribute to fouling when a bare electrode is used in complex biological media. A theoretical investigation by Vlassiouk and co workers, on the importance of these nanopore parameters has shown that the selectivity of charged long nanochannels (>1 high when the pore size is small ( radius < 32 nm). 96 Thus, for long pores of compact and thick (e.g. > 100 nm) membranes, permeab ility of a probe can be limited by, for example, tortuous paths taken in the membrane 103 W ith short and large nanopores of ultrathin membranes, 88 the effect of tortuosity can expected to be insig nificant, permitting diffusion limited (Equation 1 6) rather than membrane limited (Equations 1 7 and 1 8) permeability. It is clear from theory 96 that for pore radii >> 32 nm, the Debye screening length relative to the pore size is insignificant and
31 can result in low permselectivity, while showing the high permeability necessary for multianalyte detection in electrochemical biosensing applications. Fabrication of Adhered Porous Ultrathin Membranes at Nanoensemble Microelectrodes by Electropolymerization of Pyrrole Ele ctropolymerization of pyrrole (PY) offers versatile approaches for constructing functional surface architectures that allows improved sensitivity and enhanced selectivity in electrochemical sensing. 88 90, 97 Intere st in this monomer stems from its ease of oxidation to form polypyrrole that adheres and deposits uniformly on the sensor surface. 94, 104 More importantly, PY electropolymerization by a constant potential step tech nique is very reproducible, 90, 105 and the practical route for controlled modification at UMEs; an advantage not accessible with conventional methods like dip coating for Nafion. 106 Reproducibility of the technique is due to the ability to control thickness of electrodep osited polypyrrole (PPY) via choice of deposition parameters including current density, 94 anodic potential limit in a pulse experiment, and time. 88, 97 Other factors including electrode size, 10, 107, 108 electrode surface structure, 97 monomer concentration, 108 nucleophilicity of th e medium, 109 and the basic character of the dopant 104, 105 are important considerations for fabrication of porous ultrathin PPY membrane. The PY electropolymerization proce ss illustrated in Scheme 1 2, involves electrochemical oxidation of PY monomer to form a short lived (lifetime = 30 s 110 ) delocalized cation ra dical, A having the greatest position, and dimerization by radical positions followed by release of protons, 111 driven by establishment of aromaticity. The chain is th en propagated by further electrochemical oxidation of the neutral dimer, B 1 ms 112 ), which is more readily oxidized 112 than the PY monomer 113 to form dimeric cation radical, C This moiety subsequently reacts with other monomeric, dimeric and/or short chain oligomeric cation radi c als to form long chain
32 oligomers, D D is further easily oxidized, after whi ch it couples with other cation radical in termediates for example, A or C to form polypyrrole, E To form an electronically conducting PPY, F 112, 114 117 E is oxidized and then doped with an anion, perchlorate in this work. It has been suggested that a critical chain length, for which the solubility of D is exceeded, is necessary for precipitation from the solution/electrode interface onto the substrate to form nuclei. 108 Growth of the resulting doped PPY 114 occurs largely by further precipitation of other F chains, and this is more favored on top of the nuclei than at the bare substrate. 108 Although modification of macroelectrodes by PY electropolymerization and the associated mechanism has been investigated in great detail, li ttle is known concerning electrodes of nanometer and micron sized dimensions, especially as to how the improved mass transport rate at their surfaces can influence the coulombic efficiency of electropolymerization. 1 0, 107, 108 As stated above, one of the ultimate goals for fabricating UMEs is to improve mass transport. Unfortunately, when the time scale (typically seconds) for PY electropolymerization is such that a radial diffusion is created at such surfaces, the mass transport rate of PY to and of PY cation radical intermediates, A D from the electrode surface could occur simultaneously. 107 Consequently, the rate of chemical coupling between radical intermediates, which i s important for producing the critical chain length of D required for nucleation and subsequent polymer growth, could be low and could prevent successful electrodeposition of E and, thus, F A high monomer concentration (> 2 times) has been demonstrated to overcome this obstacle at platinum UMEs. 108 Even though PPY electrodeposition at UMEs is not very well understood, the possible effects of improved mass transport can also present an additional challenge, especially at nanostructured graphitic surfaces, where radial diffusion can be created at the i ndividual active
33 nanofeatures. At such surfa ces, escape of radical intermediates from the electrode surface prior to participation in further redox or chemical processes 107 can limit successful PPY electrodeposition. 108 Nonetheless, this challenge can be overcome on short (milliseconds) time scales, during which linear diffusion is created at th e nanostructured substrates, as has been demonstrated for other substrates. 88, 92, 97 Nucleation and subsequent polymer growth by rapid coupling on the millisecond time scales occur because of the short lifetimes ( < 1 ms) of the monomeric and dimeric cation radicals. 110, 112 Thus, electrodeposition can be realized because of rapid chemical coupling within the resulting thin diffusion layer, in a fashion similar to a relay ra ce, in which each athlete in a team relays the baton to the next quickly in order to win the race. Strategies for making ultrathin (~2 nm), but porous, polypyrrole membranes adhered to UMEs rely on passing a low charge density (C cm 2 ) on millisecond time scales. 88 Electrodeposition of ultrathin (5 10 nm) membranes onto Pt macroelectrodes on second time scales 94, 104 has also been demonstrated. These membranes had a compact structure and a low current density (C cm 2 ) could not result in significan tly increased membrane porosity. In one of such membranes, more pores could onl y be created by templating with polyethylene glycol nanoparticles. 94 On the other hand, factors that increase the rate of PPY growth such as high current density, result in rough s urface morphologies, 94, 118 similar to those observed for deposition of rough gold (Au) nanostructures at increased plating potential. 63 Clearly, high polymerization rate alone yields only rough morphologies. Perhaps, the competition between the rates of mass transport and chemical coupling could allow development of a hypothesis for formation of highly porous membranes at nanostructured surfaces. The effect of improved mas s transport, and thus high diffusion rate of cation radical intermediates from the substrate surface, 107, 108 can be hypothesized to result in n o nuclei formation at the parts of the
34 substrate with the highest mass transport rate and, thereby creating voids. This hypothesis may be unlike metallic deposition, where nuclei growth occurs at all zones with improved mass transport, 46 probably because of the characteristic deposit ion mechanism of metals. The above hypothesis could explain the elegant of work of Allan et al who demonstrated two decades ago that the rough surface structure of rough pyrolytic graphite (RPG) substrate accounts for a highly porous structure of electr odeposited PPY membrane, whereas a more compact membrane structure forms onto the smooth surface of glassy carbon (GC). 97 More recently, Kannan et al have also suggested that this hypothesis could account for the highly porous structure of electrodeposited polypyrrole on rough carbon fiber microdisk, than on glassy carbon. 10 Such insights into improved mass transport at nanofeatures, 107 coupled with the hi gh rate of chemical coupling of PY cation radical intermediates, 108 can allow for control ling the polypyrrole electrodeposition process. More importantly, this novel idea can be expected to open new possibilities for very easy and simple design of nanoporous ultrathin polypyrrole membranes for highly permeable electrochemical biosensors, as de monstrated with nanostructures on carbon fiber microdisks in this work. Electrochemical Characterization of Nanoporous Ultrathin Membrane Coated Nanoensemble Microelectrodes Characterization of the presence and physical parameters (e.g., membrane thicknes s and pore size) of thin to ultrathin membranes at nanostructured graphitic UMEs present technical challenges. These challenges originate from the high surface roughness of the substrate relative to the thickness of the membrane, as well as the electronic and chemical properties of the membrane/film. Thus, the choice of characterization method (electrochemical, microscopic, etc) should be capable of answering relevant questions pertaining to the system under investigation.
35 Few reports on microscopic charact erization of ultrathin films/membranes on other substrates and other forms of carbon are known in the literature. At smooth substrates such as Au (111) film, 119 amorphous carbon nitride (a CN x ) film 104 and Pt, 94 membrane pores were either visible or their presence had to be confirmed by electrochemical methods. Rarely are such observations made at rough UMEs most likely because of the above mentioned challenges. To the best of our knowledge, the only report of microscopic visualizarion of a porous thin PPY film on a rough carbon fiber microdisk is the work of Kannan et al 10 because of the electronical ly conducting character of the PPY and the high film thickness (10 20 nm) relative to the surface roughness of the substrate. Conversely, when the membrane material is insulating and its thickness is low relative to the surface roughness of the substrate, as is the case in this work, characterization of such membrane coated UME by microscopy becomes difficult, but possible, with electrochemical methods. 88 With electropolymerization being the practical coating technique, the membrane deposition. 120 Also, verification of the presence of the membrane by electrochemical techniques derives from its interfacial character and sensitivity to the unblocked electroactive area of the substrate. In this method, the response at the bare electroactive a rea is required a priori for comparison and subsequent extraction of relevant data. Knowledge of the surface chemistry of membrane/film is also required a priori to allow selection of appropriate probe for interrogation. Electrochemical strategies capable of indicating the presence of a nanoporous film/membrane are based on electrostatic repulsion between the probe and membrane or film coated surface, 88, 121, 122 as well as size selectivity in the case of compact membranes/films. 91, 94 For example, in overoxidized polypyrrole (OPPY) membranes, the negative charge of
36 ferricyanide is strongly repelled by the net negative charge density on the surface of the membrane. 88 This is also the situation between hexammineruthenium (III) and quaternary ammonium groups of thiol mediated self assemb led monolayers on Au. 122 Because of these interactions, the probe cannot access the part of the electrode that is blocked by the membrane/film. Consequently, the faradaic current and electrode transfer kinetics at the membrane or film coated sur face decrease relative to those at the bare electrode, in accordance with charge transfer at partially blocked surfaces. 73 As in the case of bare nanoensemble UMEs, there is no theory at present for predicting the behavior of nano dimension electrode ensembles formed at membrane coated UMEs. Nonetheless, as in the case of bare nanoensembles, 74 their behavior is consistent with that at macroelectrodes. In view of this, not onl y is the presence of the membrane verified by electrochemical methods, but also the current and kinetic data provide possibilities for estimating their pore size and center to center separations using electrochemical theories, 72, 73, 123 as demonstrated in this work. Analysis with Nanostructured Graphitic Ultramicroelectrodes As in the case of any useful analytical tool, practical application of both bare and membrane coated nanostructured graphitic UMEs as electroch emical biosensors requires that a calibration be performed. The analytical sensitivity of small biological molecules at these surfaces can be increased by scanning the potential limit to +1.5 V vs SCE in fast scan cyclic voltammetry (FSCV). 38, 124 The challenge to the use of such electrodes having small surface area (~4 2 vs SCE is that changes in the electrode surface nanostructure can easily manifest in changes in the background current, which can limit the reusability, sensitivity, and ultimately the LOD in electrochemical biosensing. It is also worth noting that moderate surface overoxidation can result in increased analytical sensitivity, 125 but the electrode
37 response can be deteriorated with e xcessive formation of such surface oxide 3 7 These surface changes can be minimized by proper choice of techniques. 38 Stability of Graphitic and Membrane Nanostructures Because of the high current density at UMEs, graphitic nanostructures at such surfaces can be readily overoxidized to form surface oxides (resistive elements) when etched at high potential limits (e.g. with lower potential limit, 1.0; upper potential limit +2.0 V vs. SCE at 70 Hz) on minute time scales. 37 In addition to the potential limit, a higher degree of material disorder can cause a higher extent of overoxidation 70 as observed in this work. Also, redox active membrane nanostructures such as OPPY can be oxidatively degraded in aqueous medium. 109 At the potential limits typically required for biosensing, such oxidative processes of surface nanostructures can be minimized by using the FSCV technique. The advantage at such fast scan rates, as it relates to stability of both graphitic and membrane nanomaterials, is due in part to the fact that the residence time at individual potentials within a potential window can be on the nano to millisecond time scale, resulting in limited electrolysis of such surfaces. The Toth research group demonstr ated in the latter part of 1990 s that potential limits of 1.0 and 1.5 V vs. SCE are important for regenerating nanostructured surfaces of carbon fiber microdisks during their use i n analytical determinations at 500 V s 1 38, 124 Since then, variants of this technique have been adopted by others 1, 126 to achieve similar goals. Such in situ renewal of electrode surfaces may not be produced with other electrochemical techniques, such constant potential amperometry. 87 As a result of a stable background current that is associated with limited changes in surface nanostructure with FSCV not only can calibrations be performed, 38, 68 but most importantly, improved LOD can be achieved, as demonstrated in this work.
38 Fast Scan Cyclic Voltammetry with Background Subtraction Fast scan cyclic voltammetry (FSCV) involves application of a triangular potential waveform to an electrode at potential scan rates above 10 2 V s 1 and measurement of the resulting current during the potential scan. The scan rate that can be applied to an electrode is determined by the cell time constant and Ohmic drop. A short cell time constant results in rapid charging of the double layer; the low Ohmic drop produces minimal distortion of the voltammetric data. Since both of these parameters are directly proportional to the electrode area, their lower values at the small geom etric areas of UMEs 37 permit fast scan rates up to 10 6 V s 1 127, 128 allowing rapid measurements on the millisecond to nanosecond time scales. In fact, it is difficult to perform such scan rates at macroelectrodes for the converse reasons. Unlike constant potential amperometry, FSCV provides excellent chemical resolution. 2, 68 Perhaps most of the advantages of the technique are conn ected with its high speed. FSCV can act as a kinetic filter to allow selective detection of analytes like dopamine, which react rapidly at the electrode surface over those with slow kinetics of reaction such as ascorbic acid. 129, 130 In addition, high temporal resolution and associated signal averaging (of a large number of acquired scans) capabilities of the technique result in improved signal to noise (S/N) ratio, 129 allowing LODs at submicromolar levels for small biomolecules to be achieved at small geometric areas (~40 m 2 ) of UMEs. 68 The FSCV technique is a very powerful biosensing tool because of it s speed of measurement and possibility for nM detection at small geometric areas (~40 m 2 ) of UMEs In this technique, a triangular waveform is applied and the background current of a buffer (electrolyte) is recorded. This is followed by recording current for a solution of the analyte in the same buffer. The analytical signal is then obtained by digital subtraction of current of the background from that of the analyte solution. 129 As depi cted in Figure 1 5, a few nA of peak
39 1, 68 Therefore, a combination of the sequential data acquisition 129 and the stable background current, 38 as well as other pragmatic strategies such as analyte preconcentration in a membrane coated at the electrode surface, can pr ovide the impetus for the high sensitivity and low nM detection desired at nanostructured graphitic UMEs. It is important to note that the biosensing capability of FSCV is typically limited to scan 1 1 since the LOD can be high when the scan rate is increased beyond this value. 124, 129, 131 Yet, these scan rates can be sufficient for lowering LOD when combined with analyte preconcentration strategies in memb rane coated electrochemical biosensor. At scan rates >> 500 V s 1 it can be very difficult to resolve the analytical signal from the digitized noise 129 because the background currents 36 can be >> 10 2 times higher than peak current produced by for example nM concentrations of analyte at UMEs. At such scan rates, the LOD can be compromised while allowing fast electron transfer kinetics to be addressed. Adsorpti on at Nanostructured Graphitic Ultramicroe lectrodes The high sensitivity for small biomolecules at graphitic microelectrodes at fast scan rates is due to weak adsorptio n. 129 The sensitivity is even higher at low submicromolar concentrations, 1, 44 exhibiting a characteristic feature of the Langmuir isotherm acco rding to 1 (1 9) and (1 10) 0 s are the surface excess and saturation coverage (mol cm 2 ), respectively, K is the equilibrium constant for adsorption (cm 3 mol 1 ), i p and A have their usual meanings. At bare
40 129 adenosine, 131 and xanthine 68 2 Improved LODs in the 5 10 nM range have been reported for dopamine, but at cylindrical carbon fiber microelectrodes having 2 and whose surfaces were either overoxidized 1 or flame etched. 132 For measurements at small environments such as release from single cells, microdisks are capable of high spatia l resolution 2, 5, 6 and simple device miniaturization. 4, 133 Highly sensitive miniaturized biosensors are also in high demand for applications involving limited fluid sampl es, such as those produced from small biological samples. 3 Moreover, such devices can significantly improve LOD in immunosensing. 9 These advantages and interesting possibil ities with micron sized electrochemical biosensors are currently the motivation for novel surface design strategies and one of such approaches was investigated in this work. A surface nanostructure for enhancing mass transport to a gold nanoparticle netwo rk on a microelectrode has demonstrated high sensitivity in exocytotic dopamine release from single cells. 87 However, transport to the electrode surface was hindered in this strategy. Alternative surface strategies that combine highly permeable 3D constructs/coatings as in the ca se of porous ultrathin membranes, 71, 93 which also allow significant surface preconcentration, 95, 96 have the capability to achieve nM detection driven by high diffusive fl ux across such memb ranes, as demonstrated in Chapter 5 of this dissertation. Study Overview Strategies for optimizing the analytical sensitivity of electrochemical biosensors for nanomolar dete ction of biomolecules are described in this dissertation. C hapt er 1 has presented the background to the challenges and state of the science at nanostructured graphitic ultrami croelectrodes, as well as recent r esearch to achieve nanomolar (nM) limits of detection
41 (LOD) at novel porous ultrathin surface constructs and t o advance electrochemical performance at nanostructur ed surfaces In Chapter 2, details of the experimental procedure and instr umentation are presented. This C hapter also highlights the fundamentals of the electrochemical methods as they relate to the rese arch investigation as well as interpretation of the acquired data. Illustrations are al so provided to help readers grasp the concepts of the methods and their applications. Chapter 3 describes the effects of properties of carbon fibers produced from polyac rylonitrile and pitch precursors on analytical sensitivity and kinetics of dopamine and uric acid reactions at their nanostructured microdisk surfaces. Results are explained in terms of surface chemistry and the material disorder of the carbon fibers. Addi tionally, the impact of the fabricated nanostructures on surface interacti ons with the biomolecules is described. Chapter 4 exploits efficient mass transport at nanostructure surfaces for fabrication of porous ultrathin (1 3 nm thick) membranes of overoxid ized polypyr role (OPPY) on nanos tructured PAN microdisks. This C hapter highlights a strategy for successful electr odeposition of these membranes. Additionally, characterization of the presence, partial coverage and high perme ability of the porous membranes using slow scan voltammetry of ferricyanide and electrochemical theories is described. Sensitivity enhancement and nanomolar detection at the porous ultrathin membrane coated nanotructured PAN microdisks ar e the focus of Chapter 5. This C hapter discusses the driving force for nM LOD for dopamine and the membrane permselectivity against uric acid. It also presents electrochemical theories for estimating the dimensions of the random array of pores and their center to center separations in the electrodeposite d membranes, allowing a model to be proposed for the membrane structure.
42 A new application of the nanostructured carbon fiber microdisks for electrochemical immu no sensing is the subject of Chapter 6, which highlights a strategy for nanomolar detection of p nitr ophenol via in situ electrogeneration and measurement of p aminophenol. This C hapter also discusses development of a 2 amino 2 (hydro xymethyl)propane 1,3 diol/sodium acetate buffer medium to provide suitable conditions for high catalytic activity of alkaline phosphatase, the intended reporter enzyme for the immunosensing scheme. In Chapter 7, preliminary investigations of single walled carbon nanotu be film microdisk electrodes are presented and challenges relating to electrochemical accessibility of t he geometric area of the exposed sections are addressed. Chapter 8 summarizes the conclusions of the results from the entire dissertation and proposes ideas for future advancements
43 Table 1 1. Dependence of voltammetric characteristics on nanoensemble di mensions, diffusion layer thickness and scan rate Property Category A B C vs. R a < R a > R a > R a vs. 2 R 0 < 2 R 0 < 2 R 0 R 0 vs. c > c < c < c Response Type Clear peak Steady state Steady state [Adapted from Zoski and Wi jesinghe, Isr. J. Chem. 2010 Davies and Compton, J. Electroanal. Chem. 2005 ] Figure 1 1. Illustration of surface functionalities of oxygen in (A) pitch and of both oxygen and nitrogen in (B) PAN carbon fibers [Adapted (A) from McCreery, Chem. Rev. 2008 ; (B) from Ma ldonado et al Analyst 2006 Pietrzak, F uel 2009 Laffont et al., Carbon 2004 ]
44 Figure 1 2. A structure of graphite showing basal and edge planes of a carbon material [Adapted from http://mrsec.wis c.edu/Edetc/nanoquest/carbon/index.html ] Figure 1 3. Illustration of (A) nanoensemble and (B) regular (hexagonal) array of active and nonactive elements and their dimensions on a microdisk. (Not drawn to scale). [Adapted (A) from Scharifker, J. Electroanal. Chem. Interfacial Electrochem. 1988 ; (B) from Amatore et al., J. Electroanal. Chem. Interfacial Electrochem. 1983 ]
45 Figure 1 4. Representation of diffus ion profile categories at nano ensembles of active elements in and corresponding shapes of voltammogram that is produced: (A) Linear diffusion generates (D) a peak shaped; radial diffusion in both (B) and (C) produce (E) steady state voltammograms. [Adapted ( A ) (C) from Davies and Compton, J. Elec troanal. Chem. 2005 ]
46 Figure 1 5. Background current subtraction in fast scan cyclic voltammetry: (A) background current in electrolyte, (B) current of low micromolar analyte concentration in the same electrolyte and (C) analyte signal, obtained by subtraction of (A) from (B). Scheme 1 1. Proposed reactions for electrochemical etch of graphitic materials [ Kathiwala et al., The Analyst 2008 ]
47 Scheme 1 2. A mechanism for elec trop olymerization of pyrrole to form electronically conducting polypyrrole : (A D) Pyrrole radical cation intermediates, (E) and (F) are insulating and conducting polypyrrole, respectively. [Adapted from John and Wallace J. Electroanal. Chem. Interfacial E lectrochem. 1991 Atana soska et al., Chem. Mater. 1992 ]
48 CHAPTER 2 EXPERIMENTAL Reagents and Solutions All chemicals were reagent grade and were used as received, unless otherwise stated. Potassium chloride, potassium ferricyanide, sodium chloride, sodium hydroxide, sodium phosphate monobasic monohydrate, anhydrous sodium phosphate dibasic disodium salt of ethylene diaminetetraacetic acid (EDTA), phosphoric acid (85%), perchloric acid (70%), hydrochloric acid (36.5%), glacial acetic acid, p aminophenol ( p AP), and neutral alumina (80 200 mesh) were obtained from Fisher Scientific, Fair Lawn, NJ. The 2 (3,4 dihydroxyphenyl)ethylamine hydrochloride (dopamine, DA), and 2,6,8 trihydroxypurine (uric acid anhydrous, UA), p nitrophenol ( p NP) (> 99%), magnesium a cetate tetrahydrate, 2 amino 2 (hydr oxymethyl)propane 1,3 diol (Tris ), pyrrole (PY) (98%) and HPLC grade acetonitrile (0.05% water) were from Sigma Aldrich, St. Louis, MO. The following chemicals were obtained from the indicated manufacturers: anhydrous so dium acetate (J.T. Baker Chemical Company, Phillipsburg, NJ), magnesium chloride hexahydrate (Mallinckrodt Chemical Works, St. Loius, MO), tetra n butyl ammonium bromide (Eastman Kodak Co., Rochester, NY) and absolute ethanol (Decon Labs, Inc, King of Prus sia, PA). Tetra n butyl ammonium perchlorate (TBAP) was synthesized in house from aqueous solutions of tetra n butyl ammonium bromide and perchloric acid ( vide infra ). Both pyrrole and p aminophenol were purified as described below. Aqueous solutions were prepared in doubly distilled water. Buffer pH was adjusted with 1 M NaOH or a stock solution of indicated acid. The composition of the electrolytes/buffers were as follows: 500 mM KCl, 100 mM NaCl, 2100 mM MgCl 2 6H 2 O standardized with EDTA, 500 mM Na 2 HPO 4 500 mM NaCH 3 CO 2 40 mM Mg(CH 3 CO 2 ) 2 4H 2 O, 31 mM phosphate buffer pH 7.4 (12.1 mM NaH 2 PO 4 H 2 O and 19.34 mM Na 2 HPO 4 ), 225 mM phosphate buffer pH 7.0
49 (138.14 mM NaH 2 PO 4 H 2 O and 86.45 mM Na 2 HPO 4 ), 100 mM Tris/ Acid buffer pH 9.0 (Acid component: CH 3 CO 2 H, HCl or H 3 PO 4 ). Nonaqueous solutions of TBAP and PY were prepared in acetonitrile. 40 mM PY sol ution was made in 0.1 M TBAP. Preparation and Purification of Chemicals Tetra n butyl ammonium perchlorate: TBAP was synthesized using aqueous solutions of 0.233 M tetr a n butyl ammonium bromide (TBAB) and 1.17 M perchloric acid (Scheme 2 1) 1 Briefly, 90 mL solution of the acid was added gradually to 400 mL of TBAB solution with continuous stirring using a glass rod. The added volume ratio was such that the acid was ~10 mole % excess of the mole equivalent for the reaction in Scheme 2 1. The resulting suspension was filtered to obtained a wet, crude, tainted yellow precipitate of TBAP (~62 g; theoretical yield is 32 g). The solid was recrystallized in three stages from a bsolute ethanol after dissolving in the solvent at 76C, using the following order of solid (g): solvent (mL) ratios: 25:100, 30:100, 35:100. During each stage, the mixture was allowed to cool to room temperature for at least 12 h after which it is was kep t in refrigerator at 5C for 5 6 h. In the first stage, white sheet like crystals (~22 g) were formed. Weight loss (40 g) should have significant contribution from the excess perchloric acid and residual hydrobromic acid (pale yellow), which were almost co mpletely removed at this stage. The TBAP crystals formed at the third stage were dried under vacuum and 13 g of solid (40 wt% yield) was finally produced. Pyrrole: PY was purified in three stages by passing 1 mL sample through three glass wool plugged past eur pipettes, in the order of increasing mass of the columns packed with 0.4, 0.6 and 0.8 g of neutral alumina (Figure 2 1). A nitrogen pressure from a tube was applied over 1 Liu, D. ; University of Florida; Personal communication.
50 the first pipette, such that the yellowish brown PY passed through the column to t he next within 10 the pipette tip), the tube was transferred to the next pipette and the procedure repeated. At the third stage, the color of PY changed from yellow ish brown to pale yellow to colorless. Purified PY (0.1 mL; 10 vol% yield) was stored in an amber vial under nitrogen and at 5C until its use. p aminophenol: p AP was purified five times according to a general procedure in the literature. 134 Briefly, a 20 g sample of brown colored p AP was added to 10 mL of acidic ethanol, pH 4 (50:0.025 v/v of absolute ethanol: stock HCl) an d stirred to dissolve brown matter covering the grains. The brown liquid was removed by decantation. At the end of the third stage, a beige colored solid remained. Each of last two purification stages was carried out using 5 mL of the solvent. After the fi fth stage, the precipitate was dried under vacuum for over 24 h and the resulting faintly beige to off white colored fine crystals of p AP (0.2 g; 10 wt% yield) were stored in a nitrogen filled amber vial. Instrumentation A BAS 100 electrochemical analyzer (Bioanalytical Systems Inc., West Lafayette, IN) connected to a home built amplifier 135 (Figu re 2 2) was used for slow scan cyclic voltammetry (SSCV) at 0.050 V s 1 potential step chronocoulometry (CC) and bulk electrolysis. The minimum current that can be measured by the commercial BAS 100 instrument alone is 100 nA. Measurement of currents <100 pA could be permitted by the gains of the amplifier. With the exception of bulk electrolysis, all electrochemical measurements were done in a two electrode configuration in an Al foil Faraday cage. The data obtained were transferred to a personal computer The instrumental set up for fast scan cyclic voltammetric (FSCV) measurements has been described. 135 A function generator, EG&G PARC model 175 Universal Programmer, (EG&G
51 Instruments, Princeton, NJ) with a home built current transducer (Figure 2 3) was used for the FSCV at 500 V s 1 135 The same set up was also used in electrochemical etching at 10 V s 1 The function generator was used to generate a continuous triangular waveform, which was applied to the SCE reference electrode, in a two electrode system in a copper mesh Faraday cage used to reduce noise in FSCV. Current produced at the working electrode was converted to voltage and amplified by the current transducer. Both the voltage output of the current transducer and potential waveform applied to the SCE were stored b y a digital oscilloscope, LeCroy 9350A, 500 MHz (LeCroy Corporation, Chestnut Ridge, NY). Data were transferred to a personal computer using a home written program in LabView (National Instruments, Austin, TX) In both cases, data were analyzed using Origi n 8.5 software (OriginLab, Northampton, MA). Electrodes A saturated calomel reference electrode (SCE) (Fisher, Pittsburg, PA) and home made silver (Ag) wire quasi reference were used in aqueous and non aqueous media, respectively. The Ag wire was used in s uch media to prevent formation of liquid liquid junction potentials. 136 The Ag wire was also used as an auxiliary electrode in the bulk electrolysis procedure. The Ag wire electrode was made by soldering a 5 cm long Ag wire (0.5 mm dia.) to a conductor. The electrode assembly was then supported in 200 L micropipette tip, using EPOXI PATCH 1C WHITE Kit (The Dexter Corporation, Seabrook, NH). The epoxy was formed by squeezin g equal lengths of resin (Part A) and hardener (Part B) and then mixing thoroughly to form a uniform color. After filling the micropipette tip, the epoxy was cured at 60C for 2 h. Nanostructured graphitic microdisk electrodes fabricated from single walled carbon nanotube films (The Rinzler Group, UF, Gainesville, FL) and three different carbon fiber materials, Polyacrylonitrile (PAN ) based HCB (Textron Specialty Materials, Lowell, M A), PAN
52 T650 and pitch based P25 (Cytec Engineered Materials, Greenville, S C ) having 7.0, 6.8 and 11 respectively, were used as working electrodes Fabrication of Nanostructured Carbon Fiber Microdisks A carbon fiber (~3 cm long) was connected to copper wire (~6 cm long, 0.3 mm dia, Fisher Scientific, Pittsburgh, PA) wit h electrically conductive silver epoxy resin (E4110 part A, Epoxy Technology, Billerica, MA) mixed with 43 wt% curing agent (hardener) (E4110 Part B). About 1 cm of the connected fiber and wire were covered by the silver epoxy mixture, after which the asse mbly was cured at room temperature for 48 h. The copper wire end of the electrode assembly was drawn into a 200 L micropipette tip (5 cm long) so that approx. 1 cm of the fiber was drawn in while the rest protruded at the tapered end of the pipette tip. When in a vertical position with the tapered end pointing downwards, t he electrode was sealed by filling the pi pette tip with a hot (80 90C) mixture of epoxy (EPON TM Resin 828, Miller Stephenson Chemical Co., Danbury, CT) with 13 wt% m phenylenediamine flaked hardener (DuPont Specialty Chemicals, Wilmington, DE). To obtain a good quality seal, the resin was cured at room temperature for 72 h. The previously reported resin composition having 12 wt% of the hardener 38 did not work well with a fresh resin, which remained sticky after heat curing and appeared to cover the microdi sk surface during polishing. The sealed electrodes were further cured at 90C in an oven, Precision Thelco (Precision Scientific, Chicago, IL) for 1 h to further improve the sealing and hardness of the epoxy ( Figure 2 4 ). The disk cross sectional surface of the fiber was exposed by touch polishing with 600 grit SiC polishing paper (Mark V Laboratory, East Granby, CT), mounted on a polishing wheel (Ecomet I, Buehler Ltd, Evanston, IL) with doubly distilled water as a lubricant. Electrodes were held at ~ 90 to the surface of the polishing paper with the aim of obtaining a disk shaped surface.
53 Polished electrodes were tested by SSCV at 0.050 V s 1 from 0.5 to 0.2 V vs SCE with 5 mM Fe(CN) 6 3 in 0.5 M KCl to verify the electroactive radius ( vide infra ). 38 Electrodes that did not show a response were re polished. Touch polished electrodes were considered to be of good quality when the separation between forward and reverse scans of pseudo steady state voltammograms o E quality electrodes were the n nanostructured by the electrochemical etching Microdisk electrodes were nanostructured as described previously by the electrochemical etch method to expose the nanofeatures of the carbon fiber microdisk. Briefly, this involved continuous potential cycling from 1.0 to 1.5 V, at 10 V s 1 for 30 min in 31 mM phosphate buffer, pH 7.4 124 The electrochemical etch of the f iber surface exposes the nodules and nanosheet features as well as the nanopores that form the carbon fiber nanostructure. Potential window, scan rate, and electrolyte identity were selected to limit the overoxidation of the electrode surface and to allow reconditioning of the electrode surface for repeated use. 38 The background current of the electrodes was recorded from 1.0 to 1.5 V vs SCE at 10 V s 1 in 31 mM phosphate buffer, pH 7.4, before and after the elec trochemical etch to verify the formation of the nanostructure, and to confirm the stability of the nanostructured electrodes during use. Nanostructured electrodes were tested in ferricyanide to verify the electroactive radius and improved electron transfe r kinetics. Electrodes that showed well developed steady state voltammograms of ferricyanide were selected for fabrication of overoxidized polypyrrole (OPPY) membranes on their surfaces and/or for F C SV measurements at 500 V s 1 Fabrication of Porous Ultr athin OPPY Coated Nanostructured PAN Carbon Fiber Microdisks The membranes were fabricated by electropolymerization of pyrrole (PY) to polypyrrole 108 onto the nanostructured electrode, followed by electrolytic
54 overoxidation of PPY to overoxidized polypyrrole (OPPY). In the PY electropolymerization, the double p otential step method was used to minimize changes at the electrode surface nanostructure. 71 Prior to PY electropolymerization, the electrode radius was determined in 5 mM Fe(CN) 6 3 in 0.5 M KCl as described above. However, in spite of careful washing and drying of the electrodes, remnants of the solution that was used in the determination of electrode radius interfered with electrodeposition o f PPY. We speculate that the interference is due to a film of chloride on the electrode nanostructures. As a result, after the determination of the electrode radius, the nanostructured electrodes were cleaned in 31 mM phosphate buffer, pH 7.4, by sweeping the electrode potential from 1.0 to +1.5 V vs SCE at 500 V s 1 using 250 scans. The electrodes were then washed with doubly distilled water and air dried at room temperature for 90 min before electrodeposition of PPY from an acetonitrile solution of pur ified PY. Three of the electrodes were dried overnight, because our results showed that 90 min drying was insufficient for successful formation of PPY membranes in one PY electropolymerization step (Chapter 4). In th e procedure for PPY electrodepo sition, t he PAN electrodes were equilibrated in 0.1 M TBAP/acetonitrile solution for 10 min and the background charge was acquired by chronocoulometry (CC) using a 10 ms potential step from 0.650 to 0.950 and then back to 0.650 V vs Ag wire. With the electrode in acetonitrile solution of 40 mM purified PY in 0.1 M TBAP, the potential step was repeated to electropolymerize PY to form the PPY membrane. For electrodes dried for 90 min, the PY electropolymerization step was done twice. Overoxidation of PPY (electronic ally conductive) to OPPY (insulator) occurs readily in nucleophilic environments. 104, 109 PPY was overoxidized at a constant potential of 0.950 V vs SCE in 0.23 M phosphate buffer, pH 7.0. The ferricyanide limitin g current was then measured at the OPPY
55 membrane coated PAN carbon fiber nanostructured microdisk electrodes as described above to verify coverage of the nanostructured PAN electrodes with the OPPY membranes. Fabrication of Single Walled Carbon Nanotube Fi lm Microdisks The carbon nanotube (CNT) films (50 nm thick) on glass slides with P d electrode contacts (Figure 2 5 B) were prepared according to a reported procedure. 56 Briefly, a mixture of single walled CNT in 1% v/v Triton X 100 was filtered onto a cellulose ester membrane substrate, after wh ich the surfactant was wa ed walled CNT/cellulose ester membrane was transferred onto a glass slide and the membrane was dissolved with acetone, rinsed with methanol and then with water. After drying, a section of the CNT film close to the middle of the glass slide was masked. Pd was then sputtered onto both sections of the unmasked CNT and the glass slide extending to the edge along its length as illustrated in Figure 2 5A. A 200 nm thick layer of poly(methyl methacrylate ) (PMMA) photoresist was spin coated onto the SWCNT film. This layer thickness was obtained with PMMA A4 in anisole at 2000 rpm using the spin coater, WS 400 6NPP (Laurell Technologies Corporation, North Wales, PA). The PMMA layer was then baked at 170C f or 30 min. A microdisk elect patterned by e beam lithography using a nano and microfabrication device, Raith 150 (Raith USA Inc., Ronkonkoma, NY) and developed by immersing in methyl isobutyl ketone: isopropyl alcohol (1:3 v/v) for about 30 s. The immersion was r epeated once. Part of the photoresist on sections of the film having sputtered Pd (i.e. about 3 mm from the edge of the glass slide and along its length) was wiped with acet one to expose the Pd (Figure 2 5 B) for electrode contact. The electrode was stored under nitrogen until use. The resulting CNT film ultramicroelectrode was housed in a cell made from teflon and polycarbonate blocks as illustrated in Figure 2 6 The response of these electrodes were also
56 tested by SSCV at 0.050 V s 1 from 0.5 to 0.2 V vs SCE with 5 mM Fe(CN) 6 3 in 0.5 M KCl to verify the electrode radius and electron transfer kinetics. Analytical Measurements Analytical determinations were performed using FSCV at 500 V s 1 A syringe was used to inject a buffer solution into the electroc he the solution and electrodes as shown in Figure 2 3 The background current at the nanostructured microdisk electrodes in the buffer was recorded at 500 V s 1 with 250 averaged scans, by continuous potenti al cycling from 1.0 to 1.5 V and back to 1.0 V vs SCE. These data were stored in the digital oscilloscope. A solution of the analyte in the same buffer was then injected and the procedure was repeated to measure and store the analyte current in a separa te file in the oscilloscope. The background current was then digitally subtracted from the analyte current to obtain the background corrected analyte signals. This procedure was repeated for all calibrations in the 1 00 nM using the Adjacent Averaging routine in Origin 8.5 and setting points of window to 5. Data in the nanomolar range were smoothed 2 times. In analytical determinations of p AP, buffer solutions were purged with nitrogen for 30 min. Solutions of this analyte were also kept in the dark until measurements were made. Nanostructured microdisk electrodes with excessively noisy background currents at 500 V s 1 were discarded, since such behavior led to irreproducible background subtracted signals. Fundamentals of Electrochemical Methods Used Cyclic Voltammetry In cyclic voltametry (CV), 137 a triangular waveform is applied to an electrode and current is measured as the potential is scanned from an initial to a final value as shown in Figure 2 7 The initial potential is selected a priori such that no current due to faradaic process(s) of the
57 analyte is produced. As the potential is scanned past a value at which the analyte oxidizes or reduces, the current increases gradually until it reaches a maximum and then either remains steady (sigmoidal shaped in Figure 2 7B ) or decreases gradually and stabi lizes (peak shaped in Figure 2 7 D) during the scan to the final potential. When the direction of the potential scan is switched, the current signal corresponding to the oxidized or reduced form of the analyte can be observed if the corresponding redox reaction is chemically reversible. The range of potentials scanned between the initial and final values is referred to as the potential window, and is determined by electrochemical behavior of the analyte, electrode material and electrolyte identity; the current potential plot is called a voltammogram (Figures 2 7B and D). The current produced in a CV is controlled by adsorption and/or dif fusion. For a diffusion controlled process at a microdisk, radial and linear diffusion fields are produced at slow and fast scan rates, respectively, 36 since the diffusion layer thickness is related to the time scale of the experiment according to Equations 1 1 and 1 2, 73, 76 as illustrated in Figure s 1 4A C. Consequently, a steady state voltammogram is produced when a radial diffusion field is created, because analyte undergoes a redox process at the same rate as it diffuses to the electrode surface. On the other hand, a peak shaped voltammogram forms when a linear diffusion field is created, since th e rate of analyte redox process is faster than its rate of diffusio n. It is important to note that for an analyte that adsorbs irreversibly at an electrode surface such as benzoquinone, 138 a peak shaped voltammogram can be observed irrespective of the time scale because of its high surface excess at the electrode surface relative to the bulk solution. Although both radial and linear diffusion are explicit, mixed behaviors can also be observed experimentally. 75, 77 In this work steady state voltammograms recorded at slow scan rates (Figures 2 7 A and B) we re used to characterize the bare and OPPY membrane coated nanostructured graphitic
58 microelectrodes At fast scan rates, pea k shape d voltammograms (Figures 2 7 C and D) were used to test their electrochemical biosensing capabilities. Electroactive Radius of Microdisks : As shown in Figure 2 7B a steady state voltammogram has an slope region that is kinetically limited (current increases with applied potential) and a plateau region (independent of applied potential) having mass transport limitation. The limiting current, i L (nA) is determined from the plateau region and the baseline. For steady state disk current, 36 i L is given by (2 1) where n is number of electrons, F is the Faraday constant (C mol 1 ), C o is bulk concentration of analyt e (mM), and D 0 is the diffusion coefficient (cm 2 s 1 ) of the electroactive species. 36 In this work, the e lectroactive radius, r a (m) of th e microdisks was determined from the limiting current for electrochemical reduction of ferricyanide (Scheme 2 2) using Equation 2 1. The electroactive area of electrode, A a was then calculated by (2 2) Because of the possible vari ations in the polishing procedure, non homogeneity in the fibe r diameter along its length and structural het erogeneity in the exposed cross sections, variations in r a and therefore, in A a can be expected. At this point, it is very important to make a clear distinction between the electroactive, A a and geometric, A areas of the microdisks. The geometric area, A, is derived from the physical dimension of the electrode material supplied by the manufacturer, whereas the electroactive area, A a is the area of the microdisk (Equation 2 2) that contributes to electrochemical reactions. At nanostructured graphitic ultramicroelectrodes, not all parts of the microdisk area are electroactive, as stated in Chapter 1. Part s of the electrode that are non active do not co ntribute to
59 faradaic processes. Consequently, the current produced is due to electrochemical reactions only at the active sites. For a perfect ly smooth micro disk electrode, A a can be A In the case of a rough electrode surface, A a is typically >, but can also be A Improved Electron Transfer Kinetics at Nanostructur ed Carbon Fiber Microdisks: In electrochemical reactions, a fast redox process is said to be electrochemically rever sible or Nernstian; otherwise, the process is quasi reversible or totally irreversible. For a one step, one electrode redox reaction of ferricyande (Scheme 2 2) that was used to test the electrode response, the electrode potential, E of such a reaction as sumed to be in equilibrium is given by the Nernst equation 36 (2 3) where E 0 is the formal potential (mV) of the re dox system, R is the molar gas constant (J mol 1 K 1 ), T is the absolute temperature (K), i is the current (nA) at a specific potential, m R and m O are mass transfer coefficients of the reduced and oxidized forms of the redox system, respectively, and n F and i L have their usual meanings. Since at the slope region of the steady state voltammogram (Figure 2 7B) current ator for electron transfer kinetics at the electrode. A high slope indicates fast kinetics and vice versa Data in this region were analyzed by plotting E vs. log[( i L i )/ i ], in accordance with Equation 2 3, 36 to determine kinetics of ferricyanide redox reaction. Theoretically, the slope of the resulting straight line is 59.1 mV for a reversible, one step, one electron reaction at 25C. Values of slope > 60 mV indicate a degree of irreversibility. 36 A slope value of ~60 mV, extracted from the steady state voltammetry of ferricyanide at the nanostructured carbon fiber microdisks, was sufficient to indicate improved kinetics at such fabricated surface nanostructures.
60 It is important to note that a fast track to determine the reversibility of a redox system is to calculate the absolute value of the difference between quartile potentials, E Q accor criterion given by 36 (2 4) where E 3/4 and E 1/4 are quartile potentials (mV) corresponding to third and first quarter, respectively, of the limiting current. In this case, E Q has a value 56.4 mV at 25C for a reversible one step, one electron reaction, and > 56.4 mV when the system is irreversible. In this work, the Nernstian route was used, because another important parameter, the half wave potential, E 1/2 which is an estimate of the formal potential, can be directly extracted from the intercept of the E vs. log[( i L i )/ i ] plot when i = i L /2. Under this situation, the a bscissa is zero and Equation 2 3 rearr anges to (2 5) The formal potential, E 0 ', can then be estimated by assuming that the mass transfer coefficient ratio, m R / m O in Equation 2 5 is unity. Microdisk Electrode Capacitance : In addition to the electroactive radius and elect rode kinetics, the apparent specific electrode capacitance, C ( F cm 2 ) at 10 V s 1 was determined before and after fabrication of the surface nanostructures, as well as after FSCV measurements at such surfaces. The capacitance was calculated from the bac kground current, at 0.75 V vs SCE, i b by assuming that i b results predominantly from double layer charging, with limited contributions of surface residual faradaic reactions 38 using the relationship (2 6)
61 where b is the difference between anodic and cathodic sections of the background current (nA), is the scan rate (V s 1 ), and A is the geometric area of the microdisk (cm 2 ) calculated based on the geometric radius of the fiber. The ratio of C of the nanostruc tured microdisk electrode to the specific capacitance of the edge plane of highly ordered pyrolytic graphite (HOPG) (60 F cm 2 ) 139 was calculated to estimate the roughness factor of the fabricated nanostructures. F ractional Coverage of Nanostructured Carbon Fiber Microdisks with OPPY Membranes : Steady state electrochemical signals are advantageous for characterizing modified electrodes, since time can be eliminated as a variable allowing, mathematical treatments to be simplified. 36 When part of an electroactive area is blocked by a membrane/film, it manifests in decreased magnitude of the limiting curr ent of a steady state voltammogram, in accordance with charge transfer at partially blocked surfaces. 73 After fabricating the OPPY membranes in this work, the steady state voltammogram of ferricyanide was recorded a t OPPY coated microdisk electrodes as described above. A lower limiting current of ferricyanide is measured after the bare electrode is coated with OPPY membrane, primarily because of its permselectivity against ferricyanide. 88 In other words, sections of the electrode blocked by OPPY membrane are inaccessible to ferricyanide. In addition, the decreased magnitude of t he current can be due to the reduced size of the active elements ( vide infra ). Thus, only a fraction of the bare electrode response (to ferricyanide) is observed at the OPPY coated electrodes. Under the experimental conditions at 0.050 V s 1 radial diffu sion fields surrounding the individual active elements can be considered to overlap totally, as observed by Zoski et al 74 and this behavior has also been suggested for CNT nanoensemble microelectrodes 46 under similar conditions. In such a situation, the current is due to the entire geometric area of the electrode The limiting current expression can be approximated to that of a microdisk (Equation 1 5), 74
62 allowing the use of Equation 2 coated nanostructured microdisk electrodes. Thus, the decrease in ferricyanide response, f can be calculated from (2 7 ) where i L and i OP (nA) are the limiting currents of ferricyanide at bare and OPPY coated electrodes, respectively. The term i OP / i L membranes at OPPY coated nanostructured electrodes. The decrease in the limit ing current for ferricyanide reduction at an OPPY coated electrode relative to the corresponding limiting current at the bare nanostructured microdisk electrode can be used to determine the fraction of the electroactive area that was covered with OPPY, as shown below. Since it can be deduced from Equation 2 1, that i 2 is directly proportional to r 2 (a measure of area), the fraction of the electrode area accessible to ferricyanide, p as it diffuses through the membrane pores to such sites, is given by (2 8) The fraction of the electrode covered by the OPPY membrane, can then be expressed as (2 9) In Equation 2 8, p also represents the porosity of the porous ultrathin OPPY membrane, through which ferricyanide accesses the uncovered electroactive surface. Thus, the OPPY coated surface can be d epicted as a nanoensemble of electroacti ve sites (membrane pores) and nonactive sites formed due to blocking by the membrane. OPPY Membrane Pore Size and Center to Center Separ ation at OPPY Coated Microdisks: The use electrochemical theory to characterize the dimensions of pores and their
63 separation is complicated by their different dimensions and non uniform distributions in the membrane/film. Nonetheless, idealized models are often sufficient for estimation of these parameters. A theoretical treatment bas ed on chronoamperometry has been proposed for estimating such dimensions in thin films, but its application is limited by more complex models required to describe the system. 36 As pointed out above, steady state behavior can permit such estimations, for which the model for charge transfer at partially blocked surfaces 73 can be very useful toward this e nd. We apply this model, to our knowledge for the first time, to determine the dimensions of pores and their separation in the ultrathin O PPY membranes fabricated here. At OPPY coated surfaces, the apparent electron transfer rate constant of ferricyanide, k app 0 is low relative to that at the bare electrode, k 0 These experimental observations are consistent with the model of charge transfer at partially blocked electrode surfaces (developed for microscopic active sites) in the limit when the fraction of th e covered electrode surface area, approaches unity. 73 Under these conditions: (2 10) and (2 11) and the limiting current of steady state voltammogram for disk type active elements decreases accordin g to (2 12) where R a is the radius of active element (assumed to be disk type) R 0 is half the distance between their centers (Figure 1 3B), i L,BLK is the limiting current of ferricyanide at a blocked surface (nA), A is the geometri c surface area of the bare electrode; F C 0 and D 0 have their usual
64 meaning. The values at OPPY coated electrodes were calculated using Equation 2 10. Note that when approaches zero, R 0 approaches r a and Equation 2 12 can be reasonably approximated to Equation 2 1, as expected. Zoski et al used Equation 2 10 to predict the kinet ensembles (NEE) consisting of 30 nm Au discs supported in a track etched polycarbonate membrane. 74 microdis k electrodes in this work. Therefore, Equations 2 10 through 2 12 were adapted to estimate the size and the distance between the electroactive elements at OPPY coated nanostructured microdisk electrodes, allowing an estimate of the pore sizes in the OPPY m embrane. Unlike the test for improved electrode kinetics with nanostructuring, for which characterization with slope values using Equation 2 3 would suffice, it was imperative to determine heterogeneous rate constants to allow theoretical calculation of (Equation 2 10), needed for subsequent calculations of dimensions of the random array of membrane pores. The values of k 0 and k app 0 were determined by fitting the equation for cathodic current (Equation 2 13) derived from Butler Volmer model of electrode kinetics 123 to the steady state voltammograms at the bare and OPPY coated electrodes respectively, obtai ned for ferricyanide reduction. (2 13) The calculated rate constants are fundamentally the lower limits, since the departure of E from E 0 is not large. 36
65 In the derivation of Equation 2 13, the heterogeneity function, which reflects the extent to which the solutes diffusing from the bulk solution have different accessibility to the different parts of the electrode surface, was assumed to be zero. In such case, the accessibility to all points on the electrode surface is uniform, as is the case for geometries such as hemispheroids. In addition, it was assumed that D ox = D red 123 The anodic current analogue to Equation 2 13 has also been derived and utilized to determine heterogeneous rate constants elsewhere. 47, 140 Using Equations 2 11 and 2 12, the theory of parti ally blocked surfaces 73 allows estimation of the sizes of the electroactive elements, which are accessed by ferricyanide after diffusing through the pores in the OPPY membrane. The theory predicts that sections of t he electrode surface that are blocked by the OPPY membrane are not accessible to ferricyanide and do not make contributions to the faradaic process. In this regard the estimated sizes of the electroactive elements are approximately the same as the pore si zes in the membrane. The density of the active elements in the random array (Figure 1 3A) N formed by pores in the OPPY membranes at OPPY coated electrodes, was calculated from the equation for the nearest neighbor distribution in a random array as 72 (2 1 4) As pointed out in Chapter 1, the permeability of a porous ultrathin membrane is related to its geometric parameters according to Equation 1 6. A rearrangement of Equations 1 6, 2 11 and 2 14 allows the permeability of the porous ultrathin OPPY membranes to be determined in this work as (2 15)
66 where 1 and R a are independently determined parameters from Equations 2 9 and 2 11, respectively. Analytical Sensitivity : For a peak shaped voltammogram, peak currents are measured between the maximum peak and the baseline (Figure 2 7D) The diffusion controlled peak currents, i p for reversible and irreversible systems are given respectively by the Randles Sevcik equations 36 (2 16) and (2 17) n is the number of electrons in the rate determining V s 1 ), and n A D 0 C 0 have their usual meanings. When the reaction is controlled by adsorption, peak currents, i p for reversible and irreversible systems are also given respectively by 36 (2 18) and (2 19) In Equations 2 18 and 2 19, 0 is the surface excess (mol cm 2 ), the surface concentration of the analyte, which can be over ten times higher tha n the bulk concentration. In Equations 2 16 and 2 17, contant values have units of C V 1/2 mol 1 ; they are in C V mol 1 in Equations 2 8 and 2 19. For a diffusion controlled process, i p is proportional to the square root of the scan rate (Equations 2 16 and 2 17), whereas i p is directly proportional t o the scan rate (Equations 2 18 and 2 19) for a redox process controlled by adsorption. Thus, at the same scan rate, high surface excess should result in a higher sensitivity for processes controlled by adsorpti on compared to
67 diffusion controlled processes. In this work, the sensitivity of the model biomolecule probes was found to be adsorption controlled. In addition Equations 2 17 and 2 19 apply because the FSC voltammograms of redox reactions at the electrodes did not meet the reversibility criteria described below. In the analytical determinations, sensitivity was obtained from the slope of calibration plot of anodic peak currents vs analyte concentrations. For limits of detection, the standard deviation of t he noise, was obtained from peak to peak noise, N p p using = 0.2 N p p at 99 % confidence level. 141 N p p was calcula ted from 50 mV window centered at the anodic peak potential. Electrode Kinetics from Peak to Peak Separation : For a peak shaped voltammogram, the potentials corresponding to the anodic and cathodic peak currents, E p,a and E p,c are called peak potentials ( Figure 2 7D) Their difference, E p referred to as peak to peak separation is given by (2 20) which equals 58/ n mV at 25C for an electrochemically reversible redox system. In addition, the ratio of the anodic to cathodic currents should be unity. The peak potentials also allow calculation of E 1/2 using mV (2 21) Therefore, for a Nernstian redox system, E p is independent of scan rate. 36 In the case of an irreversible redox reaction, E p increases anodically or cathodically with scan rate, resulting in peak broadening (i.e. E p > 58/ n mV), which occurred in this work. To extract kinetic data from such voltammograms, typical for fast scan rates, an Ohmic drop correction becomes necessary after background subtraction.
68 When the current passage does not affect the reference electrode potential, the relationship between the applied cell potential, E app (V), the true potential of the working electrode, E w and the O hmic potential drop in the cell, iR u is given by 36 (2 22) where i is working electrode current (nA) (anodic current is negative). In effect, the uncompensated resistance displaces peak potentials by E = iR u resulting in peak broadening. 36 In addition, uncompensated resistance R u decreases peak current by decreasing the effective scan rate, w (V s 1 ), according to 37, 142 (2 23) where app is the applied scan rate (V s 1 ), and di/dt is the time rate of change of the working electrode current (A s 1 ). The conventional way for de termining electron transfer kinetics from E p is the use of Nicholson Method. 143 However, this method is not valid for irreversible redox reactions controlled by adsorption. In such a situation, the standard reaction rate constants, k o (s 1 ) can be obtained using the equation developed by Laviron as 144 (2 24) where p,cor is the iR u corrected p Equation 2 24 is valid for p > 200/ n mV. Correction of uncompensated resistance can be achieved with an on line positive current feedback compensation circuit 37 Alternatively, w hen the resistances of the components of the cell are known, a correction can be performed by simple calculations, as demonstrated in Chapter 3 67 of this work. Although, this approach is limited by its inability to account for decreased faradaic current, the faradaic current itself is only a small fraction of the electrode current, and hence the
69 effect of the decreased current is insignificant. Therefore, these calculations can provide reasonable estimates of iR u for correcting p values of distorted peak shaped voltammograms Semiintegral Analysis Semiintegration of voltammetric current as a function of time, i (t), can be performed by applying the semiintegral operator on the i (t) function as follows: (2 25) where I (t) (A s 1/2 ) is the semiintegrated current. The operation in Equation 2 25 can be performed by numerical integration of the voltammetric data with the help of a computer. Semiintegration can be used to obtain R u 145 and for correction of distorted voltammograms, 142 as well as for verifying adsorption controlled redox reactions. 146 The semiintegra ted voltammogram has a sigmoidal shape for a redox reaction controlled by diffusion only. When the reaction has a contribution from adsorption, a peak shaped semiintegrated voltammogram is observed. In Chapter 3 of this work, semiintegration is used to ver ify the contribution of adsorption to the sensitivity of the nanostructured carbon fiber microdisks at 500 V s 1 Contribution of adsorptive and diffusive components to voltammetric current was also characterized by semiintegration as demonstrated in Chapt er 5. According to an elegant theory developed by Freund and Toth, a linear form of the I (t) function in Equation 2 25 is obtained in the limit as time, t approaches infinity as 138 (2 26) A plot of I (t) vs t 1/2 yields nFAC R D R 1/2 as the intercept and the slope is nFA R 1/2 The intercept is an indicator of diffusion and the slope provides information about adsorption. These
70 parameters are diagnostic tools for the dominant process contributi ng to the voltammetric current. Chronocoulometry In chronocoulometry (CC), po tential step waveform is applied by starting at an initial value, at which no electrolysis occurs and then changing instantaneously to a final value, where an analyte undergoes oxidation or reduction under diffusion limited conditions. The potential is the n held at the final value for a predefined time, during which the charge passed is measured. 36 In the presence of adsorption, 147 as in the case of PY electropolymerization in this work, the total charge density passed, Q (C cm 2 ), can be written as (2 27) where D 0 is the diffusion coeff icient (cm 2 s 1 ) of dopant (perchlorate in this work) in the resulting electrodeposited PPY, Q PY is the sum of the charge density, ( 0 ), (C cm 2 ) due to the instantaneous electrolysis of the surface excess, 0 (mol cm 2 ) of PY, and the capacitive charge density Q B (C cm 2 ). Note that at each electrode, the charge passed was normalized to its electroactive area (Equation 2 2) to obtain the charge density. Unlike the specific capacitance (Equation 2 6), normalization of charge to electroactive area here was necessary, because each nanostructured electrode surface has different electroactive area and shows different electron transfer kinetics. Thus, the differences in these surface characteristics could impact the rate of electropolymerization proportiona tely. Ideally, Q equals Q B in the absence of PY, and the relationship in Equation 2 27 should be time independent. 36 However, a time depen dent behavior is observed at the nanostructured carbon fiber microdisks (Chapter 4). In the presence of PY, the first term in Equation 2 27 is time dependent whereas Q PY is not. Thus, a plot of Q vs. t 1/2 called the Anson plot, should yield
71 a straight lin e with an intercept of Q PY and a slope, m PY of 2 nFC 0 D 0 1/2 1/2 from which the diffusion coefficient of perchlorate in the electrodeposited PPY can be determined. From the charge passed, other important information, including thickness of the electrodeposited PPY membrane, can be obtained as described below. In this work, a short (10 ms) double potential step CC was used to fabricate the PPY membrane onto the nanostructured carbon fiber microdisks. The electropolymerization charge density, Q PPY 2 ), was calculated from the difference between the charge dens ity after 10 ms of electropolymerization of PY and the intercept, Q PY of the corresponding Anson plot. PPY membrane thickness, PPY (nm), was calculated from Q PPY assuming a uniform compact packing density of PPY according to 120 (2 28) where MW is the molecular weight of a PPY repeat unit ( MW = 98.24 g mol 1 ), 148 n is number of electrons per PPY repeat unit ( n = 2.25) 114 F is Fa raday constant (C mol 1 ) ; is density of porous nm thick PPY membrane, assumed to be equal to the density of acetonitrile ( = 0.7857 g cm 3 ). 149 The thickness of PPY membrane that was formed by two successive electropolymerization steps as described above, was determined from Q PPY obtained from the charge acquired during the second PY electropolymerization step. The morphology of PPY, has been shown not to change significantly after the overoxidation of PPY because the integrity of the PPY backbone is maintained during the overoxidation. 115 Thus, OPPY membrane thickness, OP and that of PPY, PPY are the same. The apparent thickness, M (nm), of the porous ultrathin OPPY membranes was obtaiended from Equations 2 9 and 2 28 as
72 (2 29) Equation 2 29 accounts for higher thickness of the porous membrane than would be calculated from the compact model using the same value of Q PPY The amount of PPY electrodeposited at 10 ms, n PP Y (amol), wa s calculated from (2 30) with n = 2.25 114 The term Q PPY /nF represents surface coverage of PPY (pmol cm 2 ). The capacitive charge density, Q B in the TBAP electrolyte is determ ined a priori and assumed to be the same both in the absence and presence of PY. The charge due to adsorbed PY is obtained by correcting Q PY for Q B Statistical Treatment Statistical Significance of Linear Correlation between Measured Variables In Chapter 4 of this work, data scatter is observed in the measurements of picocoulomb (pC) charge at the nanostructured electrodes with surface areas in the 10 20 x 10 8 cm 2 range. Significant data scatter has also been reported in measurements of charge in the nano coulomb (nC) range at nanostructu res with surface areas in the 6 9 x 10 5 cm 2 range. 150 In Chapter 4 of this work, data are fitted as straight lines using Origin 8.5 software (OriginLab, Northampton, MA) and suspect outliers are removed. The software also performs an F test on the linear model and compares it by default to a horizontal line to determine whether the linear relationship best fits the population or the data is best represented as random noise and a p value is calculated as an output. The p value provides a measure of the rel iability of the analysis, where p = 0.05 is customarily treated as a "border line
73 acceptable" error level, with a 5% probability that the relation between the variables found in the sample is not valid. 151 In physical sciences and engineering, a border line error level of 0.1 is acceptable. We therefore judge the statistical significance of the linear regressions in Chapter 4 using the follo wing criteria: p p p > 0.1 (no relationship). We use the 0.1 error level because of the significant data scatter expected in dia.) electrodes. The lines in Figures 4 1, 4 2 and 4 5 show adjusted R 2 values in the range 0.599 0.984. The same electrodes were also used to obtain the results shown in Figures 4 4 and 4 7. Can the Median of Nanoensemble Dimensions in OPPY Membrane Repr esent the Range? Differences in the structure of the nanostructured electrodes are reflected in the range of pore sizes, separation of pores and pore number densities in the fabricated OPPY membranes on the OPPY coated electrodes, reported in Table 5 1. Th e acquired data represents a skewed population distribution, and thus, the median values reported in Table 5 1 reflect observed data clustering and are within 95% confidence interval of the p opulation. A written code in R ( The R Foundation for Statistical Computing, Vienna, Austria) was used to calculate the Bootstrap 95% CI for the median 2 Error Analysis The error bars in the calibration plots (Figure 5 6) represent the standard deviation of three measurements at each concentration. The error bars in Fig ure 5 8 represent the standard deviation of the slopes of the calibration plots determined with Origin 8.5 software (OriginLab, Northampton, MA). The errors in Tables 3 2 represent the standard deviation for n number of 2 Giurcanu, M.; R 2.14.0 Statistical Software Code ; IFAS Statistical Consulting Unit, University of Florida; Personal communication
74 data. In the construction of Anson plots 36 shown in Figure 4 3, CC data at short times (1 3 ms) are discarded. 152
75 Figure 2 1. Set up for pyrrole purification. (Not drawn to scale).
76 Figure 2 2. Schematic instrumental set up for microel ec trodes at slow scan rate. BAS 100 Electrochemical analyzer as potentio stat, CT/A: current transducer/ amplifier, WE: working electrod e (Carbon microelectrode), RE: r eference electrode (SCE), RS232: connection port between potentiostat and PC: personal com puter.
77 Figure 2 3. Schematic instrumental set up for mi croelectrodes at fast scan rate. DO: digital oscilloscope, C1: channel 1 to monitor applied waveform, C2: channel 2 to monitor response of electrode, T: trigger of FG, the function generator, FS: frame synchronizer of FG, SO: signal output of FG; CT/A: current transducer/amplifier, WE: working ele ctrode (carbon fiber microdisk), RE: reference electrode (SCE), which al so works as a counter electrode, RS232: connection port b etween oscilloscope and PC, personal computer.
78 Figure 2 4 Representation of the carbon fiber microdisk electrode. (Not drawn to scale).
79 Figure 2 5 Electron beam patterned CNT film microelectrode s howing (A) Top and (B) Side views. (Not drawn to scale). Figure 2 6 I llustration of cell set up for electrochemical measurements with CNT film microdisk (3 m dia.). (Not drawn to scale).
80 Figure 2 7. Triangular waveforms (A,C) and cyclic voltammograms (B,D) at a microelctrode (~ 7 m). (A) Slow and (C) Fast scans; (B) Sigmoidal shaped (steady state) and (D) p eak shaped voltammograms. Note that at scan rates used in this work, the time scales of scans in (A) and (C) are second and milliseconds, respectively. i L i p,a i p,c are limiting, anodic peak and cathodi c peak currents, respectively. E 1/2 E p,a E p,c are half wave, anodic peak and cathodic peak potentials, respectively. [Adapted (C) from Van Ben schoten, J. J. ; Lewis, J. Y.; Heineman et al., J. Chem. Educ. 1983 ] TBAB TBAP Scheme 2 1. Reaction for preparation of tetra n butylammonium perchlorate (TBAP) from the bromide analogue (TBAB) A B Scheme 2 2. Reduction of (A) ferricyanide to (B) ferrocyanide.
81 CHAPTER 3 EFFECT OF NANOSTRUCT URED CARBON FIBER MI CRODISK ELECTRODES F ROM DIFFERENT PRECURSOR MATERIALS ON ANALYTI CAL SENSITIVITY AND ELECTRODE KINETICS Background Stability and ease of fabrication make electrochemical sensors highly desirable in demanding bioana lytical applications. 38, 63, 68, 130, 153, 154 However, the question as to why a given electrode material is preferred for sensors has been rarely answered. To develop high performance sensors for emerging new appli cations, better insights into material properties are needed. Among materials that have been used for in vivo sensing applications, which are among the most demanding of electrode materials, carbon fibers are the most common, because they are more sensiti ve than gold and platinum wires, and are easy to fabricate as microelectrodes. 11, 12 Since different precursor materials are used to fabricate carbon fibers, there is a need to characterize the sensitivity and elect ron transfer kinetics of carbon fiber electrodes fabricated from different precursors in bioanalytical applications. Attempts to explain the differences between heterogeneous electron transfer kinetics at carbon fiber electrodes have mainly considered th e propensity for adsorption at such surfaces. Since unwanted adsorption has a significant effect on electrode activity, different procedures for surface cleaning/pretreatment have been developed. 13 In addition, it has been proposed that exposed active graphitic surface edge sites can improve the kinetics and sensitivity of reactions at carbon electrodes. 155, 156 Furthermore, covalently bound surface oxygen functionalities, which control surface hydrophilicity 70 and adsorption properties, 157 continue to be investigated. N ew investigations of activity of graphitic electrodes have addressed the role of surface nitrogen functionalities, which are coordinated as pyrrolic pyridinic and quaternary like structures in nitrogen doped carbon nanotubes (N CNT). 20 These groups have been identified
82 during structural evolution of polyacrylonitrile (PAN) into carbon fibers. 70, 158 The contribution of these groups to the improved sensitivity of N CNT elec trodes to 2 (3,4 dihydroxyphenyl)ethylamine (dopamine, DA) and 3,4 dihydroxyphenylacetic acid (DOPAC) has been highlighted by Maldonado et al 20 Nanostructured carbon fiber electrodes have been developed to improv e sensitivity of biological measurements 129 and are among the most sensitive and stable electrode materials. 38, 68, 154 The nanostructured carbon fiber electrodes are renewable, because irreversible changes of the electrode surface nanostructure and chemistry are minimized during surface fabrication by electrochemical etching and during electrode use 38, 68, 154 through the choice of applied potential, potential scan rate and waveform. 124 Thus, these electrodes are now preferred for in vivo use, 14, 159, 160 and there is continue d interest in their characterization. 68, 154, 161 However, carbon fiber material properties such as precursor material, crystallite size, interplanar spacing, and material disorder, all of which impact electrode sur face structure and chemistry, have attracted less attention. 13 Carbon fiber properties have been related to precursor materials such as PA N or pitch, and heat treatment temperature and atmosphere (either nitrogen or air) during material processing. 18, 30 The analytical response of carbon fiber electrodes has been related to tensile modulus of the fibe rs, 162 in agr eement with the direct relationship between carbon fiber tensile modulus and electrical conductivity. 18 It has been reported that the sensitivity of c ylindrical carbon fiber electrodes to cationic catecholamines correlates with fiber conductivity, but not with the precursor material of the fiber. 163 There appears to be ambiguity in the effect of the chemistry of the electrode surface on sensitivit y in the investigations by Huffman 163 and Maldonado 20 and their co workers. However, the method of carbon fiber electrode surface fabrication may have
83 contributed to the observed behavior. 163 Carbon fib ers from different precursor materials have different surface chemistry and structure. Since electrochemistry is based fundamentally on interfacial phenomena, the structure and chemistry of the electrode surface is of obvious importance. 13 To our knowledge, the effect of common carbon fiber precursor materials (PAN and pitch) on electroanalytical performance (i.e. sensitivity and electron t ransfer kinetics) of carbon fiber electrodes has not been reported. In the investigations conducted in this Chapter, we sought to determine the effect of precursor material structure on the electrochemical performance of carbon fiber microdisk electrodes The electrodes were fabricated by an established method, which produces stable and renewable electrode surfaces, 38, 68, 154 due to the limited overoxidation of the electrode surface. The microstructure and the surfa ce nanostructure of the electrodes have been characterized. 38, 68, 154 As stated in Chapter 1, short time constant 135 and low Ohmic drop allow fast scan cyclic voltammetric (FSCV) measurements to be performed at such electrodes. In addition, fast scan methods facilitate the acquisition of a large number of s ignals that can be averaged in a short period of time and allow for kinetic filtering, thereby decreasing interference effects. 124 These advantages have been exploited in monitoring dynamic changes in the concentrat ion of neurotransmitters in vivo and at cells. Both DA and uric acid (UA) adsorb weakly at carbon fiber electrodes and undergo fast electron transfer kinetics. 124, 129 Additionally, the observed analytical signals for DA and UA are highly reproducible at carbon fiber electrodes, unlike some other reaction probes, such as oxidation of p aminophenol. Therefore, DA and UA are suitable as electrochemical probes for characterizing the nanostructured surfaces by FSCV at 5 00 V s 1 It was hypothesized that differences in material structure would result in differences in electroanalytical performance of carbon fiber electrodes produced from different precursor
84 materials. DA and UA, which are structurally different, biologic ally important molecules, positively and negatively charged, respectively, at pH of 7.4, were expected to show different interactions with nanostructured carbon fiber microdisk electrode surfaces because of the different structure of the carbon fiber mater ials. The results confirm contributions of material properties to electroanalytical performance of carbon fiber electrodes. The observed correlations between material properties and electrochemical parameters are discussed. Results and Discussion Electroch emical Characterization of Nanostructured Carbon Fiber Microdisk Radius and Electron Transfer Kinetics Surface micro and nanofeatures (such as nanocracks and edge nanostructures) can be fabricated by etching the carbon fiber electrode surface using electr ochemical methods. 38, 68, 154, 161 The electrolytic top down fabrication of the electrode surface can lead to irreversible side reactions: electrolysis of water, production of carbon dioxide and the formation of n on conducting surface oxides (Scheme 1 1). These reactions can contribute to irreversible behavior of carbon fiber electrodes. 38, 68, 154, 164 However, the electrolyte composition, potential window, potential scan r ate and waveform used during surface fabrication can be optimized to control these reactions. 38, 124 The heterogeneous electron transfer kinetics of ferricyanide are determined by electrode surface structure and are thus, useful as markers of electrode activity. 13, 38 Figure 3 1A shows cyclic voltammograms of 5 mM Fe(CN) 6 3 in 0.5 M KCl at PAN T650, PAN HCB and PCH P25 carbon fiber microdisk electrodes The results show that slow electron transfer kinetics of ferricyanide is typically observed at polished carbon fiber electrodes, irrespective of ca rbon fiber precursor material. This is apparent from the slow rise in ferricyanide current and the poorly defined limiting current plateau of ferricyan ide (Figure 3 1A, solid ). In contrast, ferricyanide
85 kinetics are fast at electrodes fabricated by electrochemical etch of the carbon fiber surface, again irrespect ive of the electrode material. This is apparent from higher slope values and well defined plateaus of the sigmodal i E cur ves (Figure 3 1A, large dash). As reported previously (and discussed below), nanostructured carbon fiber microdisk electrodes fabricated by electrochemical etching of the surface have a stable background cur rent and the electrode surface can be easily renewed by FSCV. 38, 68, 154 However, after the electrochemical etch of the electrode surface, differences in response to ferricyanide are observed at the resulting nanost ructured microdisk electrodes fabricated from different car bon fiber precursor materials. At PAN T650 carbon fiber electrodes, the limiting current of ferricyanide decreases, while it increases at PAN HCB and PCH P25 carbon fiber microdisk electrodes (Figu re 3 1A, large dash). Material properties must account for the differences, since the conditions of electrode surface fabrication were the same. Electrodes from the two types of PAN fibers have similar geometric radii. However, carbon assay and electrical conductivity of PAN T650 fibers a re relatively low (Table 3 1). Properties such as carbon assay have been related to the carbonization temperature of PAN fibers. 158 The correlation between electrical conductivity, material disorder/crystalline structure of PAN based carbon nanofibers and heat treatment temperature has been confirmed by XRD and Raman spectroscopic analysis. 165 Thus, lower carbon assay likely reflects a more am orphous or less crystalline structure 16 of PAN T650 fibers, presumably because of lower carbonization temperature. Less ordered fiber structure facilitates oxidation 70 of PAN T650 carbon fiber electrode surfaces during fabrication of its surface nanostructur e by electrochemical etch. The greater susceptibility of less crystalline carbon fibers to oxidation has been verified by XPS and thus, greater surface oxide coverage should be expected. 70 Ferricyanide access to oxi de rich surfaces can be suppressed, 166
86 as observed here. Consequently, P AN T650, PAN HCB and PCH P25 carbon fiber microdisk electrodes, although fabricated by the same method, will have different surface structures and chemistry, resulting from the different properties of the respective carbon fiber materials. Furthermore, ele ctron transfer kinetics of ferricyanide are faster at polished PAN T650 carbon fiber electrodes than at electrodes fabricated from the other two materials, as indicated by the less cathodic E values and lower slopes of E vs. log[( i L i )/ i ] plots (Equations 2 3 and 2 5) for ferricyanide at these electrodes (Table 3 2). Faster kinetics at polished PAN T650 electrodes are consistent with the greater material disorder 39 of PAN T650 carbon fiber material and can be attributed to its hig h surface edge defect density. Kinetics of ferricyanide reduction are even faster at these electrodes after the surface nanostructure is produced, as verified by the positive shift in E and the lower slope values of E vs. log[( i L i )/ i ] plots close to 59 mV, expected for a one electron reversible reaction. 36 In addition, E values are close to E o of 217 mV vs. SCE reported for Fe(CN) 6 3/ 4 in 0.5 M KCl. 167 It is intere sting to note that, after the surface nanostructure is produced, the slope values of E vs log[( i L i )/ i ] plots are similar (~59 mV) at all electrodes, (two types of PAN and pitch). This is due to the low resolution of slow scan voltammetry method that was used in the kinetic analysis. Thus, electrode kinetics of ferricyanide reduction at the nanostructured electrodes appears independent of the electrode material (Table 3 2). However, small differences in slope rved. These likely reflect differences in electrode resistance, 142 due to the different structures of carbon fibers from different precursor materials. Electrodes fabricated from (more d isordered) PAN T650 carbon fibers are the most resistive (Tables 3 1 and 3 2) and these electrodes are most susceptible to oxidation, including electrochemical etch of the electrodes, which can further increase electrode resistance. 37 The
87 slow electron transfer kinetics at nanostructured PAN T650 microdisks is not surprising in the light of significant decrease (22%) in the limiting current at such surfaces (Table 3 2). Nonetheless, after their use in measurements of dopamine and uric acid by FSCV at 500 Vs 1 the electrode radius of all the carbon fiber materials decrease d by ~5% (Table 3 2). Thus, ~95 % of the ferricyanide signal at the newly nanostructured surface was re produced, an indication of the stability of t he fabricated electrode nanostructures during their use in analytical determinations. Stability of Nanostructured Carbon Fiber Microdisk Background Current High electrode background current confirms the high electrode surface area of nanostructured carbon fiber electrodes. 38, 68, 154 The higher background current at nansotructured (dash) than at the touch polished (solid) surface shown in Figure 3 1B, illustrates this at electrodes fabricated from dif ferent carbon f iber materials. The low background current at PAN T650 electrodes is due to their small geometric radius (Table 3 1). However, specific capacitance calculated from the normalized background currents (Equation 2 6) is similar at electrodes fabricated from t he two types of PAN materials. Specific capacitance is lower at pitch electrodes (Table 3 2), which have a larger geometric radius. Background current of stable electrodes is expected to remain constant during electrode use. The results in Figure 3 1B and C showing the overlap of large dash and dotted curves confirm that the background current of all these electrodes is stable during FSCV measurements of DA and UA. Consequently, the ratio of the background current (or electrode capacitance) after and befor e many analytical measurements is expected to remain close to 1, as observed at the nanostructured electrodes from different carbon fiber materials (Table 3 2). In addition to stability, which confirms reusability of the electrodes in FSCV measurements, th e smaller ratio of the background currents (i.e. less positive deviation from 1) resulting in better precision of
88 analytical measurements at PAN HCB (shown by the results) translates into higher correlation coefficients of linear working curves of DA and U A determined at these electrodes. Even though the carbon content, electrical conductivity and tensile modulus of PAN HCB fibers are higher than those of PAN T650 fibers (Table 3 1), electrodes from the two types of PAN fibers have similar surface roughnes s factors due to their similar (apparent) specific capacitance (Table 3 2). A lower surface roughness of pitch fiber electrodes is detected (Table 3 2), as expected based on the properties of pitch, including high carbon assay, large geometric area, high d ensity and large surface to volume ratio (Table 3 1). Physically, the roughness factor indicates the density of rough features on the surface of a graphitic material relative to that on the surface of highly ordered pyrolytic graphite (HOPG), which is amon g least disordered graphitic materials, as its name implies. Quantitatively, it was obtained by normalizing the capacitance of the microdisks to that for edge plane of HOPG ( C HOPG = 60 F cm 2 ) 139 The high relative standard deviation of the calculated roughness factors is due to heterogeneity in the nanostructured electrode surfaces fabricated from the same carbon fiber material, say, PAN T650. Additionally, non uniformity of microstruct ures at each exposed disk cro ss section along the fiber length contributes to the lower precision of data obtained at several electrodes that were pooled for analysis (Tables 3 2 and 3 3). The capacitive current at 0.75 V vs SCE (Figure 3 1B) was used in the calculation of specific c apacitance (Equation 2 6). As shown in Figure 3 1B, the small current at 1.5 V vs SCE is due is limited overoxidation of the carbon fiber electrode surface. 38 The peak s within 0.1 and 0.2 V vs SCE are associ ated with redox process of quino ne hydroquinone surface functional groups. 16 The slow rise in current form 0.5 to 1.0 V vs SCE in Figure 3 1B may be due to hydrogen evolution through reduction of water 168 and has been suggested to contribute to
89 formation of surface roughness. 149 At the electrodes from the different carbon fiber materials, the background currents at 500 V s 1 (Figure 3 1C) are higher than at 10 V s 1 (Figure 3 1B) consistent with the direct relationship between background current and scan rate 36 (Equation 2 6). Noticeably, the ratio of the background currents at 1.5 V to that 0.75 V vs SCE is considerably lower at 500 V s 1 than at 10 V s 1 pointin g to significantly de creased surface overoxidation, and thus high nanostructure stability at fast scan rates. Carbon Fiber Precursor Material and Adsorption of Dopamine and Uric Acid Both dopamine ( DA ) and uric acid ( UA ) undergo 2e 2H + oxidation reactio ns at similar potentials under the experimental conditions of this work to form dopamine o quinone and a diimine, respectively (Scheme 3 1). Oxidation of DA results in a broad oxidation peak observed between 0.55 to 0.65 V vs SCE (Figure 3 2B). Oxidation of UA produces multiple smaller peaks (Figure 3 2D) rather than one bro ad large peak observed for DA. The processes responsible for the individual peaks in UA voltammograms in Figure 3 2D have not been identified. Currents for UA oxidation peaks shown w ith arrows in Figure 3 2D, observed between 0.47 to 0.60 V vs SCE were measured to obtain UA calibration curves. Current traces between 0 and 0.3 V vs SCE and ~1.0 and 1.5 V vs SCE (Figures 3 2B and D) are likely due to artifacts resulting from the inc omplete background subtraction. The large surface area of nanostructured carbon fiber electrodes favors adsorption and can enhance sensitivity of analytical measurements. 38, 68, 154 Large values of the ratios of exp erimental to theoretical sensitivity of DA and UA (Table 3 3) reflect this behavior. If sensitivity were controlled by diffusion, this ratio would be ~1. Based on the equation for peak current of an irreversible diffusion controlled system (Equation 2 17) 36 the calculated contribution of diffusion to FSCV peak current is < 3 %. This suggests that the high anodic peak currents of DA and UA at 500 V s 1 are due to adsorption controlled processes.
90 The contribution of adsorption to peak currents of DA and UA was confirmed by semiintegral analys is of fast scan voltammograms. Semiintegration of voltammetric curves (Equation 2 25) involves point by point numerical transformation of the current time data 142 obtained from the electroch emical experiment (i.e. FSCV). Semiintegrals have a sigmoidal (steady state) shape when the faradai c process is diffusion controlled 142 and show peaks on the steady state response in the presence of adsorption. 20, 146 Semiintegral analysis was used to deconvolute the contributions of diffusion and adsorption to the oxidation processes of DA and UA (Figures 3 2A and C). In these semiintegrals, the presence of peaks rather than steady state current plateaus, 37, 142, 145 confirms adsorption of DA 20 and UA 37 at all electrodes that were fabricated from different carbon fiber materials. The relative differences in the semiintegrate d peak heights indicate different contributions of adsorption 20 to the response of electrodes from different carbon fiber materials. Additionally, they reflect the differences in response of DA an d UA at these elect rodes. Differences in response can be identified (in the semiintegrated current curves) between 0.68 to 0.85 V vs SCE for DA and between 0.60 to 1.10 V vs SCE for UA (Figures 3 2A and C). Larger peaks on the semiintegrated current curves of DA confirm greater adsorption of DA. In the proton coupled electron transfer of DA hydrogen bonding with surface oxides at carbon based electrodes has been proposed. 157 The positive charge of DA at pH 7.4 (Scheme 3 1A) can a id interactions with surface oxides; UA is negatively charged (Scheme 3 1B). The higher analytical sensitivity of DA is in agreement with greater relative semiintegral peak heights due t o DA adsorption (Figure 3 2A). Thus, the results verify the correlatio n between higher sensitivity and greater adsorption of DA.
91 Adsorption behavior at carbon fiber electrodes reflects structure of the analytical probes (UA or DA) and of the electrode surfaces. 157, 169 For DA, the re lative magnitudes of the semiintegrated peak currents at different electrodes are PCH P25 < PAN HCB < PAN T650 indicating more adsorption at PAN T650 electrodes. UA adsorbs less than DA, and the semiintegrated peak heights of UA are similar at PAN electrod es (Figure 3 2C). Adsorption behavior of DA and UA at PAN electrodes agrees with the higher sensitivity reported for N doped rather than non doped electrodes. 20 The results show that DA adsorbs more at more disorder ed PAN T650 than at PAN HCB electrodes. Correlation between Adsorption and Sensitivity of Dopamine and Uric Acid The adsorption and sensitivity of DA and UA is low at pitch electrodes (Figure 3 2 and Table 3 3), as expected because of the lower surface ro ughness of these electrodes (Table 3 2). However, in spite of the similar surface roughness of the two types of PAN electrodes, PAN T650 electrodes are significantly more sensitive in DA measurements (Figure 3 3). The greater material disorder of PAN T650 must contribute to this behavior, because it facilitates overoxidation of the electrode surface, producing greater surface oxide coverage and leading to greater adsorption of DA. DA sensitivity at PAN electrodes correlates inversely with electrical conduc tivity and tensile modulus of the two types of PAN materials (Table s 3 1 and 3 3), in agreement with a previous report. 163 Howe ver, although Pitch P25 has low er electrical conductivity, sensitivity of DA is higher at PAN HCB. Thus, conductivity effec ts can be correlated only at electrodes from the same precursor materials. It has been reported that, even when values of electrical conductivity of carbon fibers were similar, PAN was observed to have higher capacitance than pitch electrodes. 163 Thi s is in agreement with the higher sensitivity of PAN (less conductive) to DA than pitch electrodes determined in this work (Table 3 3).
92 Interactions of UA with the nanostructured electrodes are clearly different from those of DA. Adsorption and sensitivit y of UA is higher at PAN HCB than at PAN T650 electrodes (Figures 3 2C and 3 4). This agrees with lower material disorder and higher electrical conductivity and tensile modulus of PAN HCB fibers. However, although PCH P25 has lower material disorder and hi gher conductivity than PAN T650, the sensitivity of UA is low at pitch carbon fiber electrodes. Thus, the structure of the precursor material rather than electrical conductivity of the carbon fiber has the major impact on electroanalytical performance of c arbon fiber electrodes. Determination of Electrode Kinetics after iR u Drop Correction of Peak to Peak Separation Uncompensated resistance, R u and large double layer capacitance, C dl can contribute to distortions of cyclic voltammograms, requiring data c orrection to obtain accurate kinetic information. 37 As stated in Chapter 2, the effect of the R u is to displace peak potentials by an iR u drop, resulting in p (Equation 2 20) broadening and, thus, the need for its correction. An iR u corrected cyclic voltammogram can be obtained straightforwardly by an on line positive current feedback compensation circuit. 37 This appr oach compensates cell resistance to negligible values so that the R u ( di/dt ) term in Equation 2 23 can be nullified. 136 Uncompensated resistance can arise from electrolyte/ solution resistance, R s 142, 145 the inherent resistivity of electrode materials, R m ; 170 may also be due to additional resistance of insulating/resistive surface films, R i 36 and the resistance of the electrical contact leads, R c 170 Experimental and calculated R s values have been shown to agree when the electrode materials have a negligible resistance. 142, 145 At PAN HCB microdisk electrodes, Hsueh et al reported ~8 times higher resistance after electrochemical treatment of the electrode surface (by a square p otential waveform from 1.0 to 2.0 V vs. SCE at 70 Hz for 5 min), which was ascribed to
93 surface oxide formation. 37 used. 36 Thus, resistance contributions to R u can be expressed as (3 1) The solution ohmic resistance, R s was calculated from the expression 171, 172 (3 2) where is the solution conductivity (S cm 1 ) and r the carbon fiber geometric radius (cm). The conductivity of 31 mM pH 7.4 phosphate buffer was measured as 3.16 mS cm 1 Equation 3 2 was derived from the expression for R s 172 at a disk shaped working electrode of radius, r at a distance, Z R from the reference electrode given by (3 3) For the cell configuration used in the fast scan measurements, Z R is ~ 4 mm. Fo r the micron radius nanostructured electrodes, the ratio Z R / r 3 for which arctan[ Z R / r rearrangement of Equation 3 3 to Equation 3 2. Resistance of carbon fibers, R m (Table 3 4) was calculated from (3 4) whe re is the electrical conductivity (S cm 1 ) of the carbon fiber material (Table 3 1), l is the fiber length (1 cm) and A is the geometric area (cm 2 ) of microdisk. The calculated contributions of carbon fiber resistance to uncompensated resistance are low since the calculated R m values are small (1 R s (160 4). It has been suggested that at potential scan rates below 1000 V s 1 R u compensation is not necessary because the working electrode current magnitude is small (1 100 nA). 37 Therefore,
94 iR u can be neglected from Equation 2 22 so that E w = E app allowing kinetic information to be extracted from the uncorrected values of the difference between vo ltammetric peak potentials, p However, unless R u values are independently verified, this assumption may not be accurate. To estimate iR u the value of solution resistance in Equation 3 2 was used after making the following assumptions: (i) R m + R i + R c is negligible, 36, 37 (ii) working electrode current, i is due to the background current at 0.75 V vs SCE, i b measured in buffer; (iii) the effect of iR u on anodic and cathodic peak potential displacement is the same. Based on the above a ssumptions and Equation 3 1, R u = R s Correction factor, E ', was calculated from Equation 3 5, where the assumption (iii) accounts for the factor of 2 in this equation. (3 5) The calculated E values are greater than 2 mV (Table 3 4) and are, therefore, significant. 36 Thus, experimental peak separation, p (Table 3 4) was corrected using Equation 3 6: (3 6) Since in FCSV, adsorption dominates the responses of DA and UA, 129 the kinetic equation for an adsorbed ir reversible system, (Equation 2 24), which is valid for p > 200 / n mV, 144 was used to obtain standard reaction rate constants, k o (s 1 ) from p,cor for n = 2, = 500 V s 1 T = 25 C and by approximating as 0.5. The order of p values for DA and UA at nanostructured carbon fiber microdisk electrodes fabricated by electrochemical etch is PCH P25 < PAN HCB < PAN T650. Therefore, the kinetics of DA and UA reactions are slower at PAN than at pitch electrodes (Table 3 3). There are differences between individual values of the rate constants (from p ), but the order of p values is the same for UA and DA. Thus, material structure impacts electr ode kinetics, as expected. Slower kinetics at nanostructured PAN T650 than at PAN HCB electrodes correlates
95 with the higher material disorder of the former and likely reflects the higher nitrogen content and higher degree of surface oxidation of PAN T650. For PAN, but not for pitch fibers, slower kinetics correlates with lower conductivity of the materials (Table s 3 1 and 3 3). The electrical conductivity of pitch is lower than that of PAN HCB, but the kinetics of DA and UA reactions are faster at pitch e lectrodes. The k o values of DA (Table 3 3) are inversely related to the magnitude of the adsorption peaks in the semiintegrals in Figure 3 2A and to the sensitivity of DA. Thus, carbon fiber surface structure which favors DA adsorption likely contrib utes to slower kinetics of DA. Kinetics of UA oxidation are slow at PAN T650 fiber electrodes but adsorptio n of UA is also low (Table 3 3 ). Previously, differences in k o values at carbon fiber ultramicroelectrodes have been related to differences resulting from different conditions of surface treatment 37 and recently lower p values for DA at flame etched than at electrochemically treated (by 6V sine wave in 0.5 mM K 2 Cr 2 O 7 in 5 M H 2 SO 4 ) PAN T650 electrodes have been att ributed to higher edge plane and surface defect densities, 132 as observed for HOPG. 39 It is clear from results in Table 3 3 that, in addition to adsorption of the probe material properties have a significant impact on electrode kinetics. Formation of graphite like nitrogen in graphene layers of PAN heated to 1000 C has been observed. 158 The presence of nitrogen in the graphene layer represents a defect and results in lower order of the material. 13 The high conductivity of PAN HCB, which appears to contradict the presence of defects, may result from the donor character of the pyridinic structures 173 and a highly conjugated carb on network. Presumably, because of these defects and smaller geometric radii, PAN materials can be more susceptible to oxidation than PCH P25.
96 Although the results show that greater material disorder and increased oxidation of carbon fiber electrode surfaces can increase electrode sensitivity to DA, likely because of greater DA adsorption, DA kinetics becomes slower, 1 presumably because of higher electrode resistance. 36, 37 For UA, faster kinetics at more ordered (and presumably less oxidized) PAN HCB carbon fiber electrodes correlates with more adsorption and greater sensitivity, pointing to different surface interactions of UA and DA. Furthermore, the kinetics of UA reaction are ~50 times faster than those of DA (Table 3 3), irrespective of the electrode material. The kinetics of UA reaction are also fast at pitch electrodes, which are not oxidized, where adsorption and sensitivity of UA is low. It is interest ing that the subtle differences in ferricyanide kinetics determined by slow scan voltammetry (Table 3 2) correlate with th ose of DA and UA in Table 3 3. Thus, structure face nanostructure and chemistry, do impact the surface interactions and el ectrode kinetics of the probes. Conclusions Adsorption of DA, but not of UA, is favored at more disordered and more oxidized PAN T650 electrodes. The greater disorder of PAN T650 ma terials is indicated by the lower carbon assay and lower conductivity of this material, and contributes to greater susceptibility of this material to oxidation, including during electrochemical etch of carbon fiber electrodes fabricated from this material. The resulting surface oxides favor adsorption of DA, but not of UA, and contribute to the sl ower electrode kinetics of DA. Adsorption of UA is favored at surfaces that are less oxidized, such as PAN HCB electrodes that contain nitrogen defects. The lower background current at PAN based fibers, which have smaller geometric radii than pitch fibers, contributes to better signal subtraction and increased sensitivity for DA, but has less imp act on the sensitivity for UA. Nitrogen defects, which are absent in pi tch but are present in PAN
97 materials, can account for different kinetic s at PAN and pitch electrodes. In addition, the greater geometric radius of PCH P25 electrodes in combination with inherent slow etching rate of pitch carbon fiber 69 can limit formation of nanofeatures and surface oxides at the electrode surface during electrochemical fabrication of the su rface nanostructure. The somewhat surprising outcome of this work is the absence of a simple correlation between electrode sensitivity a nd electron transfer kinetics. It may be that the assumption of a negligible contribution of resistance, R i of the in sulating/resistive surface film to the total uncompensated resistance, R u leads to an error in the estimated k o values. When corrected, k o values may be similar, especially at PAN T650 and PCH P25 electrodes. Nevertheless, PAN electrodes prepared under th e same conditions of surface etching are more useful for sensitive analytical determinations, whereas pitch based electrodes facilitate fast electron transfer kinetics. Thus, the performance of nanostructured carbon fiber microdisk electrodes fabricated b y the same method depends on structure of carbon fiber material, although the correlation with material properties is not simple. In addition to the precursors, processing conditions define material properties. The most common application of carbon fibers is in structural materials, for which processing is tailored to maximize strength. The results that were obtained provide new guidelines for tailoring material properties of new carbon electrode materials to meet electroanalytical considerations such as se nsitivity and electron transfer kinetics.
98 Table 3 1. Carbon fiber material data a a Adapted from material safety data sheets of manufacturers. b Calculated from the product of specific surface area and density. Table 3 2. Surface properties of carbon fiber microdisk electrodes Property Touch polished Nanostruct ured a PAN T650 PAN HCB PCH P25 PAN T650 PAN HCB PCH P25 r a b ( m) 2.9 0.3 3.0 0.8 4.2 0.4 2.1 0.4 3.2 1.2 4.3 0.8 c (%) 22 15 5 21 1 15 d (%) 5 12 4 13 9 36 E e (mV) 160 47 141 38 119 44 207 4 208 3 209 4 Slope f ( mV) 111 48 126 54 129 30 66 6 62 2 61 3 C g ( F cm 2 ) 911 219 932 394 740 235 C / C HOPG h 15 4 16 7 12 4 C aft / C i 1.2 0.4 1.1 0.2 1.3 0.9 a Fabricated by the electrochemical etch method. 38 b Electroactive radius of polished and nanostructured electrodes from slow scan voltammetry of 5 mM K 3 Fe(CN) 6 in 0.5 M KCl; 50 mV s 1 Equation 2 1. c,d Change in electrode radius after electrochemical etch c and after FSCV measurements of dopamine and uric acid (~10,000 scans). d Co nditions same as in b. e,f Intercept e and slope f of E vs log[( i L i )/ i ] plot (Equation 2 3) of voltammograms in b. Slope = 2.3 RT / nF n = 1, T = 298 K. g Apparent specific capacitance of nanostructured electrodes; 0.75 V vs SCE Equation 2 6; 31 mM phosphat e buffer, pH 7.4; 10 V s 1 50 scans averaged. h C HOPG = 60 F cm 2 139 i C aft is the apparent specific capacitance after FSCV measurements of dopamine and uric acid (~10,000 scans). Conditions as in g. Property PAN T650 PAN HCB PCH P25 Geometric radius ( m) 3.4 3.5 5.5 Carbon assay (%) 94 99.5 99+ Electrical conductivity (S cm 1 ) 667 909 769 Density (g cm 3 ) 1.77 1.75 1.77 1.90 Tensile modulus (GPa) 255 266 159 Specific surface area (x 10 4 cm 2 g 1 ) 0.50 0.62 0.70 Surface to volume ratio b (x 10 4 c m 1 ) 0.885 1.091 1.330
99 Tabl e 3 3. Sensitivity and kinetics of dopamine (DA) and uric acid (UA) a Fiber Material Sensi. b (mA cm 2 M 1 ) (mA cm 2 M 1 ) Exp./Cal. c Sensitivity p,cor d (mV) k 0e (s 1 ) DA UA DA UA DA UA DA UA PAN T650 5.2 0.5 1.3 0.6 314 73 851 651 0.001 2 0.060 PAN HCB 2.8 0.2 1.6 0.6 170 87 735 525 0.012 0.70 PCH P25 2.0 0.3 0.7 0.3 124 37 714 509 0.018 0.96 a At nanostructured electrodes fabricated by the electrochemical etch method. b DA sensitivity from a calibration curve; averaged UA sensitivity us ing two calibration curves and three one data point calibrations; 1 7 M; 31 mM phosphate buffer, pH 7.4; 500 V s 1 250 scans averaged. c Theoretical sensitivity calculated using Equation 2 17; n = 2, = 0.5 n = 1, = 500 V s 1 D DA = 6.0 x 10 6 cm 2 s 1 D UA = 5.0 x 10 6 cm 2 s 1 A is geometric area (cm 2 ), C o = 1 M. d Using Equation 3 6. e Using Equation 2 24. Table 3 4. Resistance, background current and peak to peak separation a Fiber Material R m b R s c i b d (nA) E e (mV) p (mV) DA UA PAN T650 4.1 257.1 76 39 890 8 690 5 PAN HCB 2.9 249.8 130 65 800 63 590 10 PCH P25 1.4 158.9 207 66 780 29 575 7 a At nanostructured electrodes fabricated by the electrochemical etch method. b Using Equation 3 4. c Using Equation 3 2. d At 0.75 V vs SCE; 31 mM phosphate buffer, pH 7.4; 500 V s 1 ; 250 scan s averaged. e Using Equation 3 5.
100 Figure 3 1. Slow scan voltammetry of ferricyanide and background current at PAN T650 (3.4 m geometric radius), PAN HCB (3.5 m) and PCH P25 (5.5 m) microdisk electrodes: (A) 5 mM K 3 Fe(CN) 6 in 0.5 M KCl, 0.050 V s 1 ; (B) 31 mM phosphate buffer, pH 7.4, 10 V s 1 50 sc ans averaged; before (solid), after (dash ) electrochemical fabrication of nano structure and after (dot ) FSCV measurements of dopamine and uric acid (~10,000 scans); (C) Same medium as in (B), but at 500 V s 1 250 scans averaged.
101 Figure 3 2. Semiintegrals (A, C) of background subtracted fast scan voltammograms (B, D) illustrating the correlation between adsorption and sensitivity of nanostructured PAN T650 PAN HCB and PCH P25 electrodes for 7 M (A, B) dopamine and (C, D) uric acid; 31 mM phosphate buffer, pH 7.4, 500 V s 1 250 scans averaged. Arrows point to the peaks used in calibrations.
102 Figure 3 3. Background subtracted fa st scan voltammograms at n anostructured PAN T650, PAN HCB and PCH P25 microdisk electrodes showing the effect of carbon fiber material on cur rent density of dopamine: (A) (D) 1, 4, 7 and 10 M dopamine, respectively; 31 mM phosphate buffer, pH 7.4, 500 V s 1 250 scans averaged.
103 Figure 3 4. Background subtracted fast scan voltammograms at n anostructured PAN T650, PAN HCB and PCH P25 microdisk electrodes showing the effect of carbon fiber material on current density of uric acid: (A) (D) 1, 4, 7 and 10 M uric acid, respectively; 31 mM phosphate buffer, pH 7.4, 500 V s 1 250 scans averaged.
104 Scheme 3 1. Redox reactions of (A) dopamine (DA) cation and (B) uric acid (UA) anion at pH 7.4.
105 CHAPTER 4 FABRI CATION OF POROUS 1 3 NM THICK MEMBRANES OF OVEROXIDIZED POLYPYRROLE (OPPY) O N NANOSTRUCTURED PAN BASED CARBON FIBER MICRODISK ELECTRODES Background Our group has developed a top down method for fabricating random arrays of carbon nanostructures on 7 m d iameter carbon fiber microdisk surfaces. The method involves exposing the nanofeatures and nanopores of the polyacrylonitrile (PAN) based carbon fibers via an electrochemical etch of the microdisk surface of the fibers. These nanostructures are heterogeneo us and consist of graphitic nodules separated by nanopores. 38, 68 Related carbon nanotubes. 27, 54, 59 In analytical applications of the nanostructured microelectrodes, significantly improved limits of detection have been demonstrated in measurements at biological cells compared to those with smooth surfaces; 154 such electrodes have been adapted to measurements in vivo 1 As reported in Chapter 1, the nanostructured PAN T650 and PAN HCB show higher analytical sensitivity than those of PCH P25 microdisk elect rodes. Recent reports have shown unique permeability properties of porous 10 15 nm thick silicon based membranes. 93 Because of fresh insights into the depe ndence of permeability on the physical parameters of such membranes, it was thought that fabrication of related membrane structures on the nanostructured PAN T650 and PAN HCB could allow the design of a highly permeable electrochemical biosensor. In this Chapter, the properties of porous ultrathin (1 3 nm thick) membranes electrodeposited on the nanostructured microdisk electrodes were investigated. The ultrathin porous membranes were deposited directly on the nanostructured microelectrodes by
106 electropolym erization of pyrrole (PY) to polypyrrole (PPY), followed by constant potential electrolytic overoxidation of the electronically conducting PPY to ionically conducting overoxidized polypyrrole (OPPY). Electropolymerization allows control over the amount of charge deposited during the electrooxidation of PY, allowing direct control of the PPY film thickness. 88 Improved mass transport to and from the nanofeatures on the nanostructured substrates can drive the formation of a highly porous membrane. During PY electropolymerization the polymer deposits on the nanostructured electrodes by el ectroprecipitation 108 producing porous ultrathin membranes. 88 To our knowledge, there have been no reports of electrochemical characterization of highly porous nm thic k surface architectures, such as the one described here, composed of a porous nm thick membrane OPPY deposited directly onto nanostructured carbon based microdisk electrodes as substrate. Electrochemical characterization can aid the development of practica l ultrathin membranes 93, 100 For example, in analysis of cell supernatants, 1 3 nm thick membranes that are highly permeable, and thus have a fast response time, can act as effective protective layers for the nanos tructured transducers In addition, the selectivity and sensitivity of the transducers can be controlled through the selective interactions of the sample with the membrane. More recently, Kannan et al reported deposition of a porous 10 20 nm PPY on a roug h carbon fiber microdisk. 10 Due to the electronic conducting property of PPY, its microscopic chara cterization was possible. Compact insulating materials of OPPY, 10 nm thick on smooth surface of CN x films 104 and as composites with CNT on cylindrical carbon fibers 174 have also been observed by SEM. But the porous structure of 1 3 nm thick insulating materials, suc h as the OPPY membranes fabricated in this work, are difficult to image by microscopy, especially when
107 they are adhered at nanoscopically rough substrates. A recently developed XPS depth profiling and 3D imaging technique holds promise for the visualizatio n of 1 3 nm thick architectures. 175 Electrochemical methods can be used to characterize the permeability of 1 3 nm thick films 88, 97 using electroactive probes, 99 and this approach was used in this work. PAN carbon fiber materials are heterogeneous in surface edge defect density, surface area and surface chemistry. The material properties of the carbon fiber determine the electroactive surface area and the kinetics of electron transfer at the nanostructured carbon fiber microdisk electrodes as demonstrated in Chapter 3. 67 We hypothesized that these material properties would determine the structure, and, thus, the permeability of the porous ultra thin OPPY membranes electrodeposited on the nanostructured electrodes. The results illustrate how the thickness and permeability of the 1 3 nm thick OPPY membranes are affected by the surface heterogeneity of the nanostructured substrates of PAN carbon fib er microdisk electrodes. Results and Discussion Ferricyanide Kinetics at Nanostructured Carbon Fiber Microdisks Mesoporous carbon fibers made from PAN are heterogeneous in defect density, surface area and chemistry. 67 Scheme 4 1A illustrates the axial and the cros s sectional microstructure of PAN fibers. According to the model of Diefendorf and the cross sectional texture proposed by Paris, the fibers consist of elongated and intermingled bundles of graphitic nanofiber ribbons, enclosing small elongated 10 50 nm ne edle like pores. 22 24, 176 179 sections of the nanofibers can be etched into nodules by a mild electrochemical method developed in our lab. Kathiwala et al has confirmed by SEM imaging th at the electrochemical etch of PAN carbon fiber surfaces produces irregular nanostructures, with 300 600 nm dia. nodule features, separated by pores of about 10 50 nm diameter. 68 Because of the microstructure of the fiber, a random array of carbon nanofeatures is exposed on the etched microdisk electrode
108 surface, as shown in Scheme 4 1B. Similar roughness of surface nanofeatures at a carbon fiber microdisk has also been observed recently by Kannan et al. 10 In addition, PAN nanofibers are chemically heterogeneous because of the varying C:N content. 70 microdisk electrodes from PAN HCB and PAN T650 carbon fibers, as described previously 67 Ferricyanide was used as the electroactive probe in the ch aracterization of the nanostructured electrodes, because of the limited adsorption and high sensitivity of electron transfer kinetics of ferricyanide to the electrochemical activity of graphite, 39 and to the passivation of the graphite surface by surface oxides. 38 Figure 4 1 shows that at the nanostructured microdisk carbon fiber electrodes, electron transfer kinetics of ferricyanide, represented by the slopes of E vs. log[( i L i )/ i ] plots (Equation 2 3) of steady state voltammograms of ferricyanide, depend strongly on the electroactive surface area of the nanostructured electrodes. The closer the slope value is to 60 mV, the faster is the electron transfer kinetics of the f erricyanide reaction (Scheme 2 2 ). The results in Figure 4 1 show that the electron transfer kinetics of ferricyanide are fastest (lowest slope) at the microdisk electrodes with the largest surface area. The surface area of the nanostructured elec trodes is directly proportional to the background current, i b shown in Figure 4 1, and the background current is a function of scan rate (Equation 2 6) as shown in Figure 3 1 B and C Voltammetric background current reflects charging of the double layer of the electroactive surface during the voltage scan 36 Because of the smaller range of areas of the nanostructured PAN T650 microdisk electr odes, ferricyanide kinetics shows a weaker dependence on the surface area of these electrodes (Figure 4 1).
109 The results in Figure 4 1 show differences in the electrode kinetics of ferricyanide at nanostructured electrodes fabricated by the same method from the same type of fiber, for example, PAN HCB This result is somewhat surprising in view of the identical diameter (ca. 7 surface of the microdisk electrodes, which are formed by the mild electrochemical etch of the electrode surface, form nanoensembles with different areas (Chapter 3) This behavior is ascribed to the differences in defect structure and density in the carbon fiber material and partly to small differences in the angle of contact during me chanical touch polishing of the electrodes. Thus, the nanostructured electrodes have a distribution of surface areas; the nanostructures are heterogeneous and form a random array of electroactive elements and non active pores depicted in Scheme 4 1B As sho wn in Figure 4 1, slower kinetics of ferricyanide reduction is observed at nanostructured PAN T650 than at PAN HCB electrodes. This is because the nanostructures on PAN T650 microdisk electrodes are more easily oxidized than those on PAN HCB, as a result of the more disordered microstructure of PAN T650 fibers compared to PAN HCB fibers. 67 The results in Figure 4 1 can be explained by a model that was developed to analyze the response of partially blocked surface s with microscopic active and non active sites. 73 At the nanostructured carbon fiber microdisk electrodes, the blocked part of the surface is due to the nanopores, as well as surface oxides, which partly cover the nodules that form the electroactive elements of the na noarray. Formation of surface oxides due to electrode surface oxidation increases the center to center distance, 2 R 0 between the active elements, as their sizes, R a shrink because of having been partially covered with the oxides. Slower kinetics of ferr icyanide reaction is predicted by the model, as the active area of the random array decreases. 67, 73
110 According to the model, under the conditions where the fractional coverage of the electrode surface by the non ac tive elements, is close to unity, the limiting current can be expressed as shown in Equation 2 12, by assuming that disk shaped active elements are formed in the nanoensemble. 73 It is important to note that with a priori knowledge of Equation 2 12 can be used to estimate 2 R o Because of the dimensions of the nanoensemble elements formed on 600 nm dia. nodules and the 10 50 nm nanopores, diffusion fields at the e lectroactive elements of the random array overlap at scan rates of 0.050 V s 1 as illustrated in Scheme 4 1D. As a result, the ferricyanide steady state current ( Equation 2 1). The ferricyanide current at the nanostructured microdisk electrodes reflects the distribution and density of the electroactive surface sites. The large range of electroactive areas of the nanoensembles that are produced on the surfaces of c arbon fiber microdisk electrodes from the same batch of PAN HCB fibers is a consequence of the heterogeneity of PAN materials. In addition, the etched nanofeatures can have different geometries at distal ends. After the electrochemical etch of the microdis k surfaces, the background current at the microdisk electrodes increases by a factor of 2. 67 The magnitude of the background current can be expected to change with a change in the geometry of the surface nanofeatures from, for example, a disk ( A = 2 ) to a hemisphere ( A = 2 2 ). 81 Zhu et al have recently reported an increase in the theoretical limiting current of electrocatalytically active surfaces with the change in geometry of surface nanofeatures with the same surface area, from oblate to spherical to prolate. Very low coverage by prolate spheroid geometries produced high catalytic activity. 81 Although the fine features of the nodules at the electrochemically etc hed PAN carbon fibers were not resolved by SEM, 68 the barely resolved distal ends of nanofeatures
111 on a rough carbon fiber microdisk 10 suggest that they may have different shapes (geometries), as observed at other electrochemically etched surfaces. 43, 69, 180 As has been s uggested recently, the behavior of infinite (i.e. macro ) arrays and ensembles of micro and nanoelectrode sites is well understood, but this is not the case for finite (i.e. ultramicro ) arrays and ensembles, where the perimeter sites play a large role in the overall steady state behavior. 76 The carbon nanoensembles that were fabricated by the mild etch of the 7 f carbon fiber electrodes fall in the latter category. Nevertheless these random arrays can be described semi quantitatively by the model for a partially blocked surface active sites, 73 as demonstrated by Zoski et al. 74 Future simulations are needed for a more detailed analysis of the i E curves obtained at the nanostructured carbon fiber microdisks in this work. Such simulations can be conside red foreseeable with continuous interest in the field since microelectrode array theory 78 has been shown to adequately simulate i E curve for a random array of nanoelectrode s 83 Because their behavior is consistent with theoretical predictions, 73 the nanostructures etched on PAN fiber microdisk electrode surfaces can be thought of as a random array of active and non active sites or as a partially blocked surface having a distribution of surface nodule and nanopore features with different geometries. High relative coverage by active nodule geometries results in higher limiting current and faster electron transfer kinetics of ferricyanide. Slower kinetics of ferricyanide reduction are expected in the presence of resistive surface oxides that partially block the nodules. 67 The fabrication of heterogeneous surfaces was an advantage in this work because it allowed characterization of electrodes with a range of electroactive areas of the nanostructures and to study their effect on PY electropolymerization. The random nanoarrays on
112 the microdisk electrodes served as substrates for the electrodeposition of porous ultrathin OPPY membranes, discussed below. Characterization of Nanostructured Carbon Fiber Microdisks Area in Acetonitrile Aqueous media were used in the fabrication of the nanostructures on the surface of carbon fiber microdisk electrodes and were used in the determination of the electroactive sur face area of the nanostructured electrodes by cyclic voltammetry. 67 However, to limit the overoxidation of PY, which occurs in aqueous solutions 105 during the electropolymerization of PY to PPY, PY was electropolymerized fro m acetonitrile using a potential step method, which produced 108 Potential step chronocoulometry (CC) was used to evaluate the surface area available for the deposition of PPY from acetonitrile. The results in Figure 4 2 verify a direct correlation between the background charge density, Q B of the nanostructured electrodes in acetonitrile solution with tetra n butylammonim perchlorate (TBAP) and the background current density, I B of the same nanostructured electrode s in aqueous phosphate buffer. The results in Figure 4 2 indicate that charging of the active areas of the nanostructured electrodes from PAN HCB is easier in acetonitrile than in water. Conversely, charging of the active areas of the nanostructured electrodes from PAN T650 is easier in water that in acetonitrile. PAN T650 is more hy drophilic because it is easier to oxidize; oxidation results in hydrophilic oxygen containing functionalities at the surface, 70 which can influence wetting properties and, thus, double layer charging. It is also pos sible that residual faradaic processes at PAN T650 surfaces can also contribute to the high background current densities.
113 Challenges to Overcome in Fabrication of Porous Ultrathin OPPY Membranes by Electropolymerization and Overoxidation It was difficult t o electrodeposit PPY membranes onto the nanostructured substrates due in part to the challenges outlined in Chapter 1 and partly because of the small picocoulomb (pC) amount of charge required. Since Anson plots allow for verification of different processe s in the absence and presence of PY, information extracted from such plots allowed us to directly confirm successful deposition of the membranes. Figure 4 3 shows Anson plots obtained from CC data that were acquired during the electropolymerization of PY o nto nanostructured surfaces of PAN T650 and PAN HCB microdisks. For the first electropoymerization step, the slopes and intercepts are about the same as for the background (Open triangle and circle in Figure 4 3) Slope, m PY and intercept, Q PY of Anson p lots 36 shown in Figure 4 3, were obtained from CC data for PY electropolymerization using another form of Equation 2 27. Likewise, data fro m background charge CC in 0.1 M TBAP solution in acetonitrile yielded m B and Q B The ratio of intercepts, Q PY / Q B and slopes, m PY / m B were used to verif y electropolymerization of PY. It was obse rved that one/both of the ratio( s ) was/were < 2 at electrodes that were dried for 90 min (Table 4 1). Furthermore, subsequent electrochemical characterization of the coated electrodes ( vide infra ) did not reveal PPY deposition. As a result, electrodes dried for 90 min required repetition of the PY electropolymerizati on step to successfully form the PPY membrane and this yielded rati 1). At these electrodes, electropolymerization of PY was of aqueous phospha te buffer. 104 We speculate that during second electropolymerization step this barrier was removed. For the overnight dried electrodes, which were only a few, the ratios were step and subsequent electrochemical characterization
114 verified successful deposition of PPY membrane. For both types of dried electrodes, membrane coated electrodes at which the electrodeposition of PPY film was successful showed similar characteristics. T o form the overoxidized polypyrrole (OPPY) membrane, the PPY was overoxidiz ed by bulk electrolysis at 0.950 V vs SCE for 2 min in 22 5 mM phosphate buffer, pH 7.0. During the overox idation the initial current (~8 12 nA) decreased and then stabilized around 0 nA, indicating complete overoxidation. 88 The OPPY structure is proposed to have oxygen fucnctionalities on the hydroxy pyrrole units, 109, 115 with disruption of conjugation as shown in Scheme 4 2. It has also been proposed to have carboxylic functionalities, all of which confer a ne t negative charge density on the OPPY material. 181 In OPPY, the mechanical integrity of polymer maintained 115 as verified by almost constant C:N ratio in OPPY as in PPY while O:N ratio increases. 181 OPPY film thickness, OP is the same as the thickness of the electrodeposited PPY film since the integrity of the PPY backbone is maintained during the overoxidation 115 Successful formation of OPPY membranes on the nanostructured elec trodes was confirmed from the limiting curre nt for ferricyanide reduction. At the membrane coated electrodes the ferri cyanide current decreased by 11 75% relative to ferricyanide current at the bare electrode. 71 Effect of Electroactive Area of Nanostructured Surface on PY Electropolymerization Anson plots of the CC curves confirm that in acetonitrile solutions with TBAP, PY adsorbs on the surfa ce of the nanoarrays fabricated at both types of PAN fibers as indicated by higher intercepts in PY than in TBAP (Figure 4 3). The charge density related to the adsorbed PY, Q PY was plotted as a function of the background charge in acetonitrile solution w ith TBAP, Q B and the re sults are shown in Figure 4 4. Note that Q PY and Q B are indicators for coverage with adsorbed PY and the electrode area, respectively. Statistical analysis (see Chapter 2) does not
115 confirm the apparent increase of coverage, Q PY wit h electrode area (Figure 4 4). This observation is surprising in view of direct relationship between surface concentration and area. 36 Howe ver, the amount of electrodeposited PPY, n PPY (amol) during the 10 ms electropolymerization of PPY (Equation 2 30), increases with increase in the electroactive area of the nanostructured carbon fiber microdisk electrodes (Figure 4 5) as supported by the statistical analysis A related observation has been made by Fletcher et al who used arrays of microband electrodes for the electropolymerization of PY. 107 They observed an increase in the amount of deposited PPY with a increase in the dimensions of the microband explained this behavior as due to faster mass transport of PY cation radical intermediates into the bulk solution than th eir rate of chemical coupling as the electrode size decreases The dimensions of the random array element s of carbon nanostructures on the surface of PAN carbon fiber microdisk electrodes are significantly smaller than those of the electrodes in the work of Fletcher et al. 107 Thus, the increase in the area of nanoarray s on the microdisk electrodes may not significantly impact adsorption of PY, in light of the significantly improved mass transport to be expected with these carbon nanofeatutes. Data scatter was observed in this work in the measurements of pC charges at na nostructured electrodes with surface areas in the 10 20 x 10 8 cm 2 range as has also been reported for nanocoulomb (nC) range at nanostructur ed surfaces with areas in the 6 9 x 10 5 cm 2 range. 150 Therefore, data sca tter could be associated with such small areas, especially ca. 10 8 cm 2 (Figure 4 5). Nonetheless, more data will be necessary for detailed quantitative analysis. Lines in Figures 4 4, 4 5 and 4 7 are shown to aid visualization of the data.
116 The surface cov erage of adsorbed PY determined from CC was in the range of 143 880 pmol cm 2 Based on the reported monolayer coverage at Pt (111) of pyridine (450 pmol cm 2 ) 182 with dia. d 2 monolayers of PY are adsorbed on the nanostructured electrodes. Because of the 1 3 nm thickness of PPY films, PPY film thickness is also reported as number of PY layers in the deposited PPY membranes as shown below. The overall process for electropolymer ization of PY to PPY and subsequent doping with anions to form electronically conducting polymer (Scheme 1 2) can be expressed as 104 (4 1) where w is number of PY repeat units, is PPY doping level and X represents the dopant anion, which is perchlorate in this work. For perchlorate doped PPY, the ratio of PY to perchlorate is 3:1, 148 for which the theoretical number of electrons is 2.3 3 (Equation 4 1) per molecular weight equivalent, MW of 98.24 g mol 1 for the PPY repeat unit 148 Theoretically, monolayer surface PPY was calculated as 320 pmol cm 2 using Equation 4 2 (4 2) In this calculation, it was assumed that (i) density of nm thick PPY membrane equals to that of acetonitrile ( = 0.7857 g cm 3 ) 149 and (ii) vertical arrangement on the short dimension of the repeat unit having a thickness, d which is equal to diameter of PY (0.4 nm). 88 The surface coverage by PPY was 560 1660 pmol cm 2 corresponding to ~2 6 PY units in the PPY film (Table 4 1) based on the theoretical value of 320 pmol cm 2 In these PPY films diffusion coefficients of perchlorate dopant, D app determined from CC using Anson plots, 36 are ~1.7 17. x 10 8 cm 2 s 1 It is noteworthy that D app is a quantitative measure of the rate of charge transport in the me mbrane and of switching of neutral to doped states of PPY 183 T hus, it should be expected to be high for porous ultrathin membranes. The
117 values obtained in this work are in agreement with the diffusion coefficient v alues reported for porous 100 nm thick ionically conducting OPPY films. 97 Additionally, the D app is 10 to 100 times higher than the perchlorate diffusion coefficient reported in 100 nm thick PPY films polymerized and doped ex situ 113 probably due to a compact structure. The range of values of perchlorate diffusion coefficients in the porous ultrathin PPY membranes determined in this work reflects the distribution of PPY membrane thickness on the membrane coated nanostructur ed microdisk electrodes. Electrochemical Characterization of OPPY Membrane Coated Nanostructured Carbon Fiber Microdisks There have been no previous reports of the characterization of porous 1 3 nm thick OPPY films on random arrays of carbon nanostructures on the surface of carb on fiber microdisk electrodes. Such architectures were characterized in this work, using electrochemical methods. SEM of 10 20 nm thick PPY (electronically conducting) on rough carbon fiber microdisk 10 and A FM imaging of 1 3 nm thick layers of PPY on smooth substrates of Au (111) film electrodes have been reported. 119 Submicron thick PPY films on carbon nanofiber arrays 184 and other substrates are well k nown. 185 188 Microscopic characterization has been possible due in part to electronically conducting properties of the material and partly because of smooth texture of the substrates relative to the thickness of the deposited films. To our knowledge, porous 1 3 nm thick films of insulating materials such as OPPY, adhered at nanoscopically rough substrates of micrometer diameter, have not been characterized previously. Electrochemical characterization was expected to provide insight into film permeability. OPPY electrodes, suppress the response of ferricyanide when the pores and defects in the film are filled. 88 Permeability of ferricyanide in porous 1.6 nm thick OPPY films decreases, and is
118 eventually completely suppressed, as the PY electrodeposition/PPY overoxidation s teps are repeated to increase the packing density of the film by filling the pores and defects. 88 Permeability, P m depends on membrane thickness, m and other parameters as given by Equation 1 8 Ultrathin, 1.6 nm thick compact OPPY films completely suppress ferricyanide transport because of the negative charge of ferricyanide and th e net negative charge density of hydroxyl, carbonyl and carboxylate functional groups of the OPPY films. 88, 89, 109, 181 However, 1.6 nm thick pinhole free OPPY films do not completely suppress transport of the posi tively charged Ru(NH 3 ) 6 3+ Permeability of the positively charged Ru(NH 3 ) 6 3+ which partitions into the negatively charged OPPY films, increases as the film thickness decreases. 88 Differences in molecular permeability in compact 10 nm thick OPPY membrane films have been exploited in selective detection of hydrogen peroxide. 90 In this work, the fabricated po rous 1 3 nm thick OPPY membranes do not completely suppress, but only decrease the rate of ferricyanide transport. 88 Consequently, the steady state limiting current of ferricyanide at electrodes coated with porous 1 3 nm thick OPPY films decreases compared to the current at the bare elec trodes as shown in Figure 4 6. In addition, the electron transfer kinetics of ferricyanide reduction are slower at OPPY coated than at the bare nanostructured microdisk electrodes, as shown by slow rise of current with applied potential in Figure 4 6. These observations are consistent with the model of charge transfer at partially blocked surface s with microscopic active and non active sites. 73 The thickness of PPY membranes, PPY was calculated first from Equation 2 28 using Q PPY the charge deposited during the electropolymerization of PPY and the number of electrons per PPY repeating unit ( n = 2.25), 114 assuming uniform surface coverage by a compact film and
119 by ignoring the presence of voids and/or pinholes in the film. 92, 97, 120 It was also assumed that the deposited PPY density is the same as that of acetonitrile by virtue of its high porosity, and that the electropolymerization efficiency is 100%. Polymerization efficiency of 37 44% has been reported for mg amounts of electrodeposited PPY. 189 Table 4 1 shows the properties of the fabricated membranes at the nanostructured surfaces of the PAN microdisks. The calculated thickness of PPY films at P AN HCB was in the range of 0.7 2.3 nm, and higher than those at PAN T650 nanostructured electrodes (0.8 1.8 nm). However, the ultrathin OPPY membranes must be porous since the limiting current for ferricyanide reduction, is not completely suppressed at the membrane coated nanostructured surfaces (Figure 4 6). Thus, the films cover only a fraction of the nanostructured surface of the microdisk electrodes as illustrated in Scheme 4 1C. As a result, thickness values determined from Equation 2 28 are smaller t han the actual thickness of the porous OPPY membranes. The same amount of charge will result in electrodeposition of thinner compact membranes (Scheme 4 3A) and will produce thicker films if the membranes are porous as illustrated in Scheme 4 3B. The fract ion of the microdisk electrode surface area covered by the porous OPPY membranes, was determined from the ratio of the limiting current of ferricyanide at the OPPY coated to the current at the same the bare electrode using Equations 2 8 and 2 9. The use of these equations is valid because of the overlap of diffusion fields at the active elements on the OPPY coated electrodes (Scheme 4 1E) t hat result in the observed steady state responses. The apparent thickness of the porous OPPY films, M calculated f rom Equation 2 29 based on the fraction of the covered surface is 0.9 2.6 nm, corresponding to ~2 7 PY units in the PPY membrane. The ferricyanide limiting current at the OPPY coated membrane coated electrodes is 25 to 65% of the limiting current at the b are electrodes. This suggests that only a fraction of the bare
120 electrode response to ferricyanide ( i OP / i L ), is observed after the electrode is coated with 1 3 nm thick OPPY films. The fraction of the response that is decreased/suppressed, f was calculated from Equation 2 7. Figure 4 7A is a plot of the ratio, i OP / i L as a function of the inverse of the film thickness, PPY calculated from Equation 2 28. The observed behavior in Figure 4 7A appears to be consistent with that expected based on the classical permeability equation (Equation 1 8) and suggests that the term, i OP / i L can be PPY increases, ferricyanide current decreases. However, statistical analysis of the data indicates that the dependence is weak, which is not surprising in view of the low nanometer thickness of the membranes in addition to the significant data scatter at the random arrays of carbon nanofeatures on the microdisk electrodes. Next, a plot o f i OP / i L vs 1/ M in Figure 4 7B, shows only a weak dependence of the M of the porous membrane. In the absence of theory and in the presence of data scatter, the exact dependence cannot be extracted from the available data and such confirmation will require further experimental verification. Recently, permeability has been related to porosity of nanoporous ultrathin membranes with large void volume. High permeability of such membranes is due to their high porosity and the high driving force for transport across the nanopores. More importantly, the permeability of such membranes is independent of membrane thickness, 93 consis tent with the results obtained in this work (Figure 4 7B). Our results indicate that porosity contributes to the high permeability of the 1 3 nm thick OPPY membranes.
121 Effect of Electroactive Area of Nanostructured PAN Microdisk Surface on Porous Structure of PPY Membrane Ideally, improved mass transport at an electrode with smaller dimension should lead to high amount of deposited PPY. In the light of PPY electrodeposition mechanism 107, 108 at finite sites, where ma ss transport of PY cation radical intermediates to the bulk solution competes with their rate of chemical coupling at the electrode surface, the small amount of electrodeposited PPY at electrodes having very small sizes in the work of Fletcher et al 107 should be expected. Another consequence of the competition between mass transport rate and chemical coupling of reactive intermediates at nanofeatures that is not explored in detail is the possibility of forming an el ectrodeposited PPY with a highly porous structure. The results in Figure 4 6 show that the decrease in ferricyanide current is low when OPPY is coated on a nanostructured electrode, having a large electroactive surface. In other words, a highly porous mem brane structure is electrodeposited on a surface with large electroactive area (Figure 4 6A) and vice versa (Figure 4 6B). A higher density of active sites that contribute to large electroactive area as stated above, could contribute to high rate of mass t ransport of the PY cation radical intermediates, and, thus, produce a highly porous OPPY membrane structure in this work. The effect of roughness features of macroelectrode substrate on porous structure of PPY membranes was also demonstrated by Witkowski e t al. 97 This hypothesis has also been proposed by Kannan et al 10 to account for porous structure of PPY deposited onto rough carbon fiber microdisk surface. The observations in this work could be consistent with the hypothesis that a highly porous membrane should form on a surface with density of active nanofeatures. However, observations such as those in Figure 4 6 A and B were random in this work. This suggests that not only is the active site d ensity of the nanofeatures important, but also, probably their geometries
122 Conclusions Because of the heterogeneity of PAN carbon fiber materials, nanostructured microdisk electrodes fabricated from the same batch of PAN fibers have a range of electroactive surface areas. Differences in the electroactive areas of the nanostructured electrodes are additionally due to the formation of electrodes with surface nanofeatures having different geometries and different oxide coverages. At larger electrodes, faster el ectron transfer kinetics of ferricyanide are observed, in agreement with the model of charge transfer at a partially blocked surfaces with microsco pic active and non active areas. A similar electrochemical behavior of nanostructured carbon fiber electrodes is indicated in aqueous and in acetonitrile solutions. The electroactive surface area of the nanostructured electrodes determines the amount of electrodeposited PPY and the resulting thickness of OPPY films. Surface oxidation to resistive surface oxides c an slow the electropolymerization of PY at nanostructured electrodes with smaller electroactive surface areas while at nanostructured electrodes having larger electroactive areas, rapid transport facilitates the electropolymerization. At PAN T650, which is more susceptible to oxidation than PAN HCB, oxidation of the surface can result in thinner films during the electropolymerization of PY. The results lead to a new model of porous OPPY membrane films. The key finding is that the high permeability of the me mbranes is due to the apparent high driving force for transport through the porous 1 3 nm thick membranes. These membranes can present selective and sensitive platforms that can act as functional surfaces for biosensing and for the immobilization of biolog ical molecules.
123 Table 4 1. Properties of electrodeposited PPY membranes at nanostructured PAN electrodes Parameter PAN T650 PAN HCB m PY / m a 1.0 2.2, (1.5 6.7) 1.1 2.1, (1.2 8.9) Q PY / Q b 1.7 2.6, (3.0 6.5) 0.1 1.7, (2.1 8.4) Adsorbed PY layer c ( pmol cm 2 ) 307 877 143 879 Surface Coverage with PPY d ( pmol cm 2 ) 664 1429 560 1660 PPY e (nm) 0.8 1.8, [0.9 2.5] 0.7 2.3, [0.9 2.6] N PPY f 2 4, [2 6] 2 6 ,[2 7] D app g (x 10 8 cm 2 s 1 ) 2.3 10. 1.6 17. f h (%) 56 75 35 73 i 0.88 0.94 0.58 0.93 a,b Ratio of slope a and intercept b of Anson plot ( a form of Equation 2 27) for PY electr opolymerization t o that for background CC curves. Ratios for the repeated PY electropolymerization in parenthesis. c Calculated from the difference between Q PY and Q B using Equation 2 30. d Calculated from Q PPY using Equation 2 30. e Calculated thickness of P P Y membranes using Equation 2 29. Values in square brackets based on porous membrane model (Scheme 4 3B) calculated using Equation 2 29. f Number of PY layers deposited as PPY. Values in square brackets based on porous membrane model (Scheme 4 3B). g D app is diffusion coefficient of perchlorate in PPY membrane calculated from slope of Anson plot (Equation 2 27). h Fraction of decreased ferricyanide limiting current at OPPY membrane coated electrodes (Equation 2 7). i Fraction of electrode area covered with OPPY using Equation 2 9.
124 Figure 4 1. Slope of E vs log[( i L i )/ i ] plot of ferricyanide steady state current as a function of electrode background current at nanostructured PAN carbon fiber microdisk electrodes. Background current measured at 0.75 V vs SCE from cyclic voltammetry at 10 V s 1 in 31 mM phosphate buffer, pH 7.4 ( p < 0.05). Slope determined from Equation 2 3 Inset in PAN HCB is a sample E vs log[( i L i )/ i ] plot, from whi ch the slope values were obtain ed Figure 4 2. Plots of electrode background charge vs background current densities, illustrating correlation between electroactive surface areas in acetonitrile and in aqueous buffer of nanostructured PAN ca rbon fiber microdisk electr odes. Q B is area normalized background charge determined from intercept of Anson plot of CC in acetonitrile solut ion with 0.1 M TBAP; I B is area normalized background current in Figure 4 1 ( p < 0.05).
125 Figure 4 3. Anson plots of CC curves for background (triangle), and first (open circle) and second (filled circle) PY electropolymerization steps at nanostructured PAN ca rbon fiber microdisk electrodes: 40 mM PY in 0.1 M TBAP; 10 ms pulse width; 0.650 to 0.950 and back to 0.650 V v s Ag/wire quasireference electrode. Figure 4 4. Correlation between background charge densities in acetonitrile solutions with and without PY, Q PY and Q B respectively, at nanostructured PAN microdisk electrodes: Charge densit ies determined as in Figure 4 1 ( p > 0.1).
126 Figure 4 5. Amount of electrodeposited PPY, n PPY as a function of electroactive surface area of p p < 0.05 at PAN HCB ) Figure 4 6. Ferricyanide steady state current at bare and OPPY membrane coated nanostruc tured surfaces of PAN HCB microdisk electrode with (A) 3.8 and (B) 3.0 m radii: 5 mM Fe(CN) 6 3 in 0.5 M KCl, 0.050 V s 1
127 Figure 4 7. Fraction of limiting current of ferricyanide measured at OPPY membrane coated relative to th at at bare nanostructured PAN microdisk electrodes as a function of the reciprocal of (A) membrane thickness of PPY, PPY calculated accordi ng to compact membrane model using Equation 2 28 ( p thickness, M according to porous membrane model using Equation 2 29 ( p > 0.1). OPPY membrane thickness is the same as PPY membrane thick ness.
128 Scheme 4 1. Schematic representation of (A) axial and cross sectional view of PAN carbon fibers microstructure ; (B) formation of random nanoarray electroactive elements of carbon nanostructures after electrochemical etch of the microdisk surface ; (C) porous structure of electrodeposited overoxidized polypyrrole (OPPY) on nanostructured electrodes ; (D) and (E) proposed diffusion profiles at (B) and (C) respectively, at 0.050 V s 1 Sizes of electroactive elements as sh own by contact with inner diffusion layer in (D) are larger than in (E). The clockwise tilt in (B) and (C) is used to enhance the 3D view. (N ot drawn to scale). [ Adapted (A) from Diefendorf and Tokarsky, E. Polym. Eng. Sci. 1975 Paris et al., Carbon 2002 ] Scheme 4 2. Proposed structure of overoxidized polypyrrole (OPPY ). [Adapted from Jaramillo et al. Analyst 1999 ]
129 Scheme 4 3 Models used in calculations of PPY and OPPY membrane thickness: (A) comp act membrane model for calculation of PPY ; (B) porous membrane model illustrating fraction of electrode area covered by the porous membrane for calculation of M with = 50 % (N ot drawn to scale )
130 CHAPTER 5 LOW NANOMOLAR DETECT ION OF SMALL BIOMOLE CULES AT NANOENSEMBLE MICROELECTRODES COAT ED WITH POROUS 1 3 NM THICK OPPY MEMB RANES Background Electrochemical sensors can achieve submicromolar detection limits (LOD s ) in direct measurements in biological samples, such as biological fluids and single cells, 87, 190 with the best LOD s obtained at nanostructured electrodes. 64, 65 In spite of this impressive performance, sensors with even lower LOD are needed for emerging new applic ations in early detection of infections and disease, where electrochemical sensors that rely on simple instrumentation are highly desirable. 3, 191 195 Atto igns made by modifying the surface of the electrochemical sensor with nanoparticles 64 and biocatalysts 4, 65 for signal amplification. The high sensitivity of nanostructured e lectrochemical sensors is due in large part to efficient mass transport, as well as the large surface to volume ratio, 64, 87 which can also contribute to high adsorption of small biomolecules. 67 Unfortunately, in biological analysis mass transport to the electrode can be compromised by adsorption. The adsorbed layer creates a diffusion barrier at the sensor surface, there by lowering sensitivity. 2 When the sensor surface is protected with a membrane 40 200 nm thick that can also preconcentrate analytes, concentrations in the 150 500 nM range can be measured. 92, 106 A 30 nM LOD has been reported in dopamine measurements in vivo 2 ; r = 5.0 p sulfobenzene, where contribution of diffusion to the low LOD h as been suggested. 44 15 nm, wi th nanopores of ca. 6 nm radi us. Depletion of analyte on one side of a confined volume (e.g. of the nanopores
131 present in the ultrathin membranes) creates a high local concentration gradient that results in a high rate of diffusive flux and ion current. 93 Mass transport through the ensemble of nanopores in an ultrathin membrane is much more efficient than diffusive transport through compact membranes of sim ilar thickness. When combined w ith analyte preconcentration in the membrane as illustrated in Scheme 5 1, low LOD can be expected. Similarly, highly efficient mass transport has been reported in thin constrained 3 196 Recent theory has shown that the permeability, k of nanoporous ultrathin (10 15 nm) membranes with arrays of nanopores of nm radius is independent of the membrane thickness, but depends only on pore size, pore density, and the diffusion coefficient of th e probe in solution, according to Equation 1 6. 93 Higher flux through nanoporous ultrathin silica, with pore radii in the 2 7 nm range, than through commer cial dialysis membranes has been demonstrated. 100 It has also been shown that transport through a nanoporous membrane under the influence of a combination of electric field and diffusion, is more efficient than mass transport by diffusion alone. 98 Based on these results, nanoporous membranes tailored with functional grou ps for sample preconcentration, 95, 96 can be expected to allow fabrication of sensitive (membrane co ated) electrochemical sensors. This principle is illustrated in Scheme 5 1. It was shown in Chapter 3 that polyacry lonitrile (PAN) show higher electroanalytical sensitivity than pitch based carbon fiber materials. A surface design strategy (Chap t er 4) that can enhance their sensitivity in electrochemical biosensing was also demonstrate d. In this strategy, porous 1 3 nm thick OPPY membranes were fabricated onto nanostructured surfaces of PAN T650 and PAN HCB microdisks. As a result, a nanoensemble of small radii active elements was formed by coverage of the electrode surface with the po rous OPPY membrane. The membrane fa electro 108 ont o the
13 2 electrode, followed by electrolytic overoxidation of the PPY to OPPY. Electrochemical characterization of the OPPY membrane coated electrodes showed that the porous ultrathin OPPY membrane covers 21 94% of the nanostructured microdisk electrode surfa ce. 71 In this Chapter, a significant improvement in sensitivity for dopamine (a model biomolecule) at the porous ultrathin membrane coat ed ele ctrodes is demonstrated. The low nanomolar detection limits that were achieved can be attributed to high diffusive flux through the porous membrane, which also preconcentrates dopamine. Ultrathin porous OPPY membranes are permselective, as shown here, as a re the pinhole free OPPY membranes. 88, 90 A lower LOD of 15 nM for dopamine (cation) at the porous OPPY membrane coated electrodes (~4 2 ) at physiological pH 7.4, is demonstrated in this Chapter compared to previously reported value of 30 nM, at p 2 44 The observed lower sensitivity for uric acid (a nion) is also addressed in this Chapter. The effect of the membrane fabrication procedure on the stability of underlying nanostructured surfaces of PAN materials, and the potential impact on electroanalytical sensitivity is also discussed. In this work, th e use of electrochemical theories 72, 73, 123 to estimate the dimensions of pore size and their separation was also demonstrated for the first time. Results and Discussion Nanoensembles of Electroactive Elements at Bare and OPPY Coated Nanostructured PAN Carbon Fiber Microdisk Electrodes Bare nanostructured carbon fiber electrodes can be v iewed as arrays of active and non active elements, fo rmed by graphitic nodules and non active pores and by surface oxides. 190 T he relative dimensions of the elements control mass transport and reaction kinetics at the nanostructured electrodes. 73 Nodule features of ca. 300 600 nm dia, separated by 10 50 nm pores, have been identified by SEM on the surface of electrochemically etched PAN based
133 carbon fiber, PAN HCB. 190 SEM images of the surface of the electrochemically cleaned carbon fiber showed nanofeatures with similar dimensions. 10 Although the surface oxides are not resolved by SEM on the surface of the fiber electrodes, the electrochemical results discussed in Chapter 3 and shown in this work confirm their presence on the etched surfaces. Thus, the nanostructured surfaces can be depicted as a nanoensemble of electroactive car bon nanofeatures separated by non active elements, which comprised surface oxides and pores (Chapter 3). The effective radius of the electroactive surface elements, R a in the nanoensembles formed after the electrochemical etch of the carbon fiber surface, and the distance between their centers, 2 R 0 are illustrated in Scheme 5 2A. The dimensions can be estimate d from the analysis of cyclic voltammograms of ferricyanide and the theory for charge transfer at partially blocked electrodes using Equations 2 10 to 2 12. 73 Note that this theory is valid when the fractional cover age of the electrode surface by blocking elements, approaches unity. At nanostructured PAN microdisks, the effects of nanoensemble formation on their electrochemical behavior agree with the predictions of the theory even though the fraction of the block ed electrode surface, << 1. Nanoensemble formation was verified by cha nges in steady state voltammograms of ferricyanide reduction that occurred at the bare nanostructured surface after its use in analytical measurements at 500 V s 1 as described in Cha pter 3. As shown in Figure 5 1, blocking of the electrode results both in decreased magnitude of ferricyanide limiting current and decreased rise of the current with applied potential in the slope region. These characteristics are significantly pronounced at the PAN T650 microdisk because of high instability of its nanostructure, 67 attributable to its highly disordered material structure. At bare nanostructured PAN T650 electrodes, the blocked fraction of the electrode surface de termined from cyclic voltammograms of ferricyanide (Figure 5 1) is ~0.2, after 10,000 scans
134 of the analytical measurements at the electrode, increasing from ~0.1 at the newly nanostructur ed electrodes. For ~0.2, the values of R a and 2 R 0 estimated using the theory 73, 123 are in the range of 0.61 2.2 and 2.6 At PAN HCB electrodes, the corresponding values are estimated as ~1.7 The PAN HCB electrodes are more stable, 67 and the formation of surface oxides that block the surface is limited, preventing ac quisition of many data points. Even though very approximate, these values account for the electrochemical behavior of the nano structured electrodes. Since the electrodes behave as partially blocked surfaces, the overlapping diffusion fields at the active elements on the surface can compete for the diffusing analyte. 80 As the ferricyanide limiting current decreases (Figure 5 1), a decrease in the rate const ant is also observed. To determine at the bare surfaces, the cathodic current for Butler Volmer kinetics (Equation 2 13) was fitted to the cathodic sections of the ferricyanide voltammograms in Figure 5 1. Electron transfer kinetics data extracted from the fits were then used to calculate (Equation 2 10), from which R a and 2 R 0 were obtained using Equations 2 11 and 2 12. The same curve fitting was also performed for ferricyanide responses at OPPY coated electrodes shown in Figure 5 2, from which the nanoensemble dimensions were subsequen tly calculated using the same set of equations. The fitted curves in Figure 5 2 had high correlation coefficients of 0.9999 and 0.9987 at the bare and corresponding OPPY coated electrodes, respectively. It is also worth noting that can also be calculated experimentally from Equation 2 9. A plot of the theore tically calculated (Equation 2 10 ) and the experimentally determined values (Equation 2 9 ) has a slope of 1.2, reasonably close to the expected value of 1. At OPPY coated electrodes, the nanostructur ed microdisk electrodes are blocked further by the OPPY membrane. As a result, the ferricyanide reduction current decreases (Figure 5 2)
135 because of blocking by the OPPY membrane; this excludes anions. The observed 11 75% decrease in ferricyanide current a t the OPPY coated electrodes, f (Equation 2 7), reflects the coverage of the electrode by the OPPY membrane and the fraction of the blocked electrode surface, (Equation 2 9) is now 0.21 0.94 as shown in Table 5 1. The lower limit of the membrane coverage is smalle r than the previously reported value in Chapter 4 (Table 4 1) 71 because more experimental da ta were included in this work. The values summarized in Table 5 1, indicate that PAN T650 electrodes, are on the average, more blocked by the OPPY membran e than the PAN HCB electrodes. The differences in the material properties and the surface structure of the carbon fibers contribute to the structural differences in the electrodeposited membranes. 67, 97 The blocked OPPY coated electrode can be visualized by covering a brown table top with a white table mat that has disk shaped holes, and viewing fr om above. Sections of the brown substrate can be seen through the holes in the mat, but at parts of the mat that have no holes, the underlying substrate cannot be seen. Note also that the radii of visible brown sections will be the same as those of the hol es. By analogy, the radii of the active elements at the membrane coated electrode can be considered to be the same as those of the OPPY membrane pores, since ferricyanide accesses the active elements through those pores. At OPPY coated electrodes, the rand om array of membrane pores (Scheme 4 1 C ), serve as channels for analyte transport to the underlying active elements of carbon nanofeat ures on the electrode surface. Diffusion through the compact parts of the membrane is significantly slower (x10 2 ) because of permselectivity of the membrane against ferricyanide. 88 After coating the electrodes with the OPPY membrane, t he radius of the active elements at the corresponding bare surface decreases to 0.04
136 the bare electrode (Scheme 5 2 and Table 5 1). In addition, their center to center separations decrease to 0.89 3.2 m. These sizes are approximately the same as the membrane pore radii (with most radii in the 0.10 3) as well as their separations. The significance of smaller dimension of the nanoensemble elements at the OPPY coated electrode is that the diffusion fields can be expected to overlap. Under the experimental conditions of thi s work, the calculated diffusion layer thickness, of 20 m (Equations 1 1 and 1 2) in combination with the median 2 R 0 value of 1.6 m, suggests that radial diffusion fields surrounding the individual active elements can be considered to overlap totally. 74 It is important to note that in such a situation, the current is due to the entire geometric area of the electrode Thus, the observed decrease in ferricyanide limiting current (Figure 5 2) is due to blocking by the membrane. 88 At OPPY coated electrodes the separation between the active elements (pores) is ~2 to 4 times les s than at the bare electrodes ( comparison of Schemes 5 2A and B ), leading to higher active site density than at t he bare electrode (Table 5 1). At the high limit of the separation (3.2 m), the sensitivity of the OPPY coated electrodes can be expected to b e low because of the dearth of active sites, if their sizes are very small as was shown with gold (Au) nanoensembles in polycarbonate membranes. 75 From the equation for the nearest neighbor distribution in a random array (Equation 2 14), the density of the active elements (membrane pores) in the random array 72 at the OPPY membrane coated el ectrodes was determined to be 3 50 x 10 6 cm 2 The range of pore density of the fabricated membranes reflects heterogeneity of the nanostructured surfaces used as substrates. This density indicates that ~1 20 pores (active sites) are present at the membrane 2 radius ). The pore density in this work is 1 2 orders of magnitude lower than that (60 x 10 7 cm 2 ) in polyca rbonate membranes, in which
137 nanoensembles of 30 nm Au nanoelectrodes have been fabricated with inter nanoelectrode distance of 200 nm. 74 For these Au nanoensembles, a calculated 231 active Au sites would fit on a 7 m dia. microdisk, in agreement with the expected higher density. This, in turn, results in a higher nu mber of pores of smaller size. The results in Figure 5 3 show that at OPPY coated electrodes, a higher number of pores (active sites) are formed when the pores (active sites) have smaller radii. An unsurprising outcome of the electrochemical model for estimating the nano en semble dimensions in this work is that, the pore (active site) density is about the same order of magnitude as that (10 60 x 10 6 cm 2 ) of a random nanoarray of Au having about the same range of STM determined 2 R 0 values (1.3 3.2 m) 83 as that estimated here, using electrochemical theories. This agreement between our calculations and experimental da ta of Wendy et al 83 as well as their consistency with the report of Zoski et al 74 can confirm the validity of our model. In the work of Wendy et al 83 the Au nanoensembles were prepared by cyanide etching of hexadecanethiol monolayers confined to copper underpotential deposition modified Au (111). High Permeability of Porous Ultrathin OPPY Membranes at OPPY Coated Nanostructured Ele ctrodes Because of the differences in membrane structure, which are expected from the differences in the nanostructured substrates, membranes with a range of pore sizes, porosity (Equation 2 8) and therefore, permeability, k (~2 50 x 10 3 cm s 1 Equation 2 15) are produced (Table 5 1). The permeability of porous ultrathin membranes at OPPY coated electrodes is of the same order of magnitude as the permeability reported for the free standing nanoporous ultrathin Si membranes, when transport was controlled b y electric field and diffusi on 93 This confirms efficient mass transport through the porous ultrathin OPPY memb ranes fabricated in this work. The permeabil ity of the OPPY membranes (Equation 2 15) is ~2 50 times higher than the rate of
138 mass transport across the free standing Si membrane ( D 0 / 0 10 3 cm s 1 ) reported from bulk transport experiments when the thickness of the stagnant layer adjacent to the nan oporous membrane is 0 100 More importantly, at the lower limit, the permeability of the OPPY membranes is approximately the same as the rate of radial mass transport in solution by diffusion ( m 0 = D 0 / 0 = 4 x 10 3 cm s 1 ) during slow scan voltammetry of ferricyanide (0.050 V s 1 ) when the thickness of the diffusion layer, 0 Since the membrane thickness is much smaller than that of the diffusion layer (i.e. OP << 0 ) at the solution membrane int erface at 0.050 V s 1 radial diffusion fields that develop at the interface overlap and extend into the bulk solution. 197 Mass transport is efficient because of high permeability of the membrane an d efficient radial diffusion. T herefore, the lower limiting current of ferricyanide in slow scan voltammetry at the porous ultrathin OPPY coated electrodes compared to the bare electrodes (Figure 5 2), is a result of blocking by the OPPY membranes. 88 When fast scan cyclic voltammetry (500 V s 1 ) was used for analysis, the diffusion layer thickness, 0 3 cm s 1 for dopamine) is ~100 times higher than the permeability of the membrane. However, because the membrane is permselective, it favors interactions with cationic analytes (e.g., d opamine at pH 7.4), 89 and preconcentration of dopamine in the membrane can be expected. Together, efficient mass transport to the membrane, dopamine preconcentration and high membrane permeability shou l d enhance the sensitivity of dopamine at the OPPY coated electrodes. The effect of membrane thickness 93 and porosity p 94 on analytical sensitivity is also discussed below. Low Nanomolar Detection Limits of Dopamine and Uric Acid at OPPY Coated Nanostructured PAN Microdisk Electrodes 2 ( r = 3. current of 500 nM dopamine in FSCV at 500 V s 1 i s 0.35 nA. This calculation is based on the
139 equations for Langmuir isotherm (Equation 1 9) and for the adsorption controlled peak current for an irreversibl e process (Equation 2 1 9), when K = 2.8 x 10 7 cm 3 mol 1 s = 10 10 mol cm 2 1 A current of this magnitude can be easily measured, since the S/N is > 3 at the bare electrode. Preconcentration of dopamine, resulting from dopamine intera ctions with the electrode surface, results in a LOD of 300 600 nM. 67 However, lower concentrations lead to lower currents that cannot be reliably distinguished from the peak to peak noise in the background subtracted voltammograms. 38, 129, 190 In the concentration range of 1 1 ) at the OPPY coated PAN HCB electrodes is ~ 1.5 times higher than at the bare PAN HCB electrode (0.9 nA M 1 ). The sensitivity enhancement is concentration dependen t, but is lower at OPPY coated PAN T650 electrodes (~0.7 nA M 1 ). As a result, 100 nM and 200 nM dopamine can be detected at OPPY coated PAN HCB (Figure 5 4C and D). Surprisingly, 200 nM each of dopamine (Figure 5 4A) and uric acid (Figure 5 4B) were also detected at the OPPY coated PAN T650 electrode, in spite of its lower sensitivity. Those results could not be reproduced readily, and more experimental data will be needed to clear this ambiguity between the lower sensitivity and nanomolar measurements at the OPPY coated PAN T650 electrodes. In the concentration range of 0.4 to 1 M, where the sensitivity is ~1.8 nA M 1 at the OPPY peak to peak noise of 100 nM dopamine signal shown in Figure 5 4 C. To the best of our knowledge, this is the first demonstration of a well resolved signal of 100 nM dopamine at a 2 ) This corresponds to the detection of ~300 dopamine molecules within the 0.125 m diffusion layer at the electrode/solution interface A 30 nM LOD of dopamine has been reported at elliptical pitch based carbon fiber electrodes (~100
140 2 ) grafted with a multilayer of p sulfobenzene film, 44 whe reas such signals could not be 2 ; r 130 Su ch low concentrations (40 nM) can be detected at bare cylindrical 130 and carbon nanotube (CNT)/OPPY coated 174 2 ; r = 3.5). As discussed in Chapter 3, semiintegrated voltammograms allow verification of adsorption contribution to the measured current. 146 Figure 5 5 shows the semi i ntegrated voltammograms that were obtained at the po rous ultrathin OPPY coated and the bare nanostructured PAN electrodes from fast scan voltammograms of dopamine and uric acid. The semiintegrated voltammograms at the bare electrode (dash in Figures 5 5A and C) show pronounced peaks, confirming that adsorpt ion makes a contribution to the analytical sensitivity, as was demonstrated in Chapter 3. In contrast, the semiintegrated voltammograms of at the OPPY co ated electrode (solid in Figure 5 5A ) do es not show a pronounced peak confirming diffusion control of dopamine currents at the OPPY coated electrod e. The higher current that results in high sensitivity of dopamine at the OPPY coated electrode is a consequence of preconcentration and efficient flux of dopamine at the porous ultrathin OPPY coated electrode ( Scheme 5 1). This results in sensitivity enhancement for dopamine at the OPPY coated PAN HCB electrode, shown in Figure 5 6. Uric acid, which is an anion at pH 7.4, was not expected to be preconcentrated in the anion permselective OPPY membranes. 89 However, at the OPPY coated PAN HCB electrodes, in the concentration range of 1 10 M, uric acid signal increased at the coated electrodes by a fixed value, unlike dopamine, for which the sensitivity was concentration dep endent (Figure 5 6). Also, uric acid sensitivity was approximately the same at both OPPY coated and bare PAN HCB
141 electrodes (Figure 5 6) A comparison of the results in Figures 5 7 and 3 4 show s significant improvement in S/N ratio at the OPPY coated PAN H CB electrodes. Figure 5 5C compares the semiintegrated voltammogram for uric acid at the OPPY coated and the bare electrodes. The results show that semiintegrated voltammogram of uric acid at the OPPY coated electrode (Figure 5 5C; solid) shows a significa nt peak. By contrast, the semiintegrated voltammogram of dopamine at the OPPY coated electrode shows a different behavior (Figure 5 5A; solid). Uric acid may be adsorbed in the membrane after it enters the porous OPPY membrane, resulting in a systematicall y higher signal, which does not change the sensitivity of the coated electrode compared to the bare electrode (Figure 5 6). The LOD of uric acid at the OPPY coated PAN electrodes is 3 0 nM (Table 5 2), which is ~15 times lower than that at the bare electrod e, because of the higher S/N at the coated electrode. Not only is the LOD lowered at the OPPY coated microdisk electrodes, but also the dynamic range of DA can be extended from 7 to 10 M as shown by the calibrations in Figure 5 6. Because UA adsorbs less than DA at bare electrode, it would saturate at a higher concentration (Equation 1 9) than 10 M as shown in Figure 5 6. Therefore, the dynamic ranges at both OPPY coated and bare electrodes should be indistinguishable up to 10 M UA. Current enhancement by enhanced diffusive flux can be quantitated from the ratio of semiintegrated diffusion currents, D C/B at OPPY coated to bare electrodes, using the intercepts of Equation 2 26. 138 Likewise, the relative contributi on of adsorption, A C/B can be determined from the slope of Equation 2 26. The results show that for dopamine D C/B = 11, but the diffusion behavior of dopamine at the coated electrode does not allow the determination of the A C/B ratio. For uric acid D C/B = 4 and A C/B = 0.5. The significant enhancement of diffusive flux, together
142 with dopamine preconcentration in the OPPY membrane, accounts for the higher sensitivity and lower LOD of dopamine than uric acid at the ultrathin porous OPPY coated electrodes. E ffect of OPPY Membrane Porosity and Thickness on Analytical Sensitivity Figure 5 8A shows the effect of membrane porosity, p (Equation 2 8) on dopamine sensitivity. In the range of porosities of ~20 55%, the sensitivity of the OPPY coated electrodes is i ndepend ent of OPPY membrane porosity. The sensitivity decreases when p < ~15% (when there are fewer pores) and is low when p > ~80%. At the high limit of membrane porosity, the coverage of the electrode by the OPPY membrane is low, and the sensitivity of t he coated electrode approaches that of the bare electrode, as expected. Similar behavior has been reported at polyethylene glycol (PEG) nanoparticle templated 5 nm thick OPPY membranes, 94 where the maximum membrane permeability was achieved using small PEG nanoparticles (i.e. 0.7 1.4 kDa), in agreement with the results obtained in this work (Figure 5 8A). The permeability of nanoporous ultrathin membranes is predicted by theory to be independent of membrane thickness. 93 This has been confirmed at 1 3 nm thick porous OPPY membrane coated electrodes, 71 in agreement with recent investigations 10 and confirmed by the results in Figure 5 8B. Kinetics of Dopamine and Uric Acid Reaction at OPPY Coated Nanostructured PAN Microdisk Electrodes The results in Table 5 2 show that, p for UA is higher indicating slower kinetics at OPPY coated than at bare PAN HCB microdisk electrodes (Table 3 4) as observed by Bravo. 149 This can be ascribed to OPPY permselectivity against UA consistent wit h decreased electron transfer kinetics during interrogation of the OPPY coated electrode surface with negatively charged ferricyniade. For DA, the p values are about the same at both OPPY coated and bare PAN HCB microdisks and seems to contradict predictions of the behavior at partially blocked
143 surfaces. 73 But, because of its high permeability, the efficient transport of DA in the OPPY membrane pores can possibly preclude decrease in DA kinetics even at partially blocked surfaces. Not e that efficient mass transport allows access to fast kinetic reactions. 36 Effect of OPPY Membrane Fabrication Procedure on Nanostructured Substrates of Different PAN Materials and Implications for Analytical Sensitivity As emphasized above, different material prop erties of the nanostructured PAN substrates, contribute to the apparent differences in structure of OPPY membranes formed on these substrates (Table 5 1) in agreement with previous reports. 97 Differences in material properties of PAN carbon fiber s are also known to contribute to 2 times higher dopamine sensitivity at bare nanostructured PAN T650 than at PAN HCB microdisk electrodes, as shown in Chapter 3. 67 Thus, it was expected that higher dopamine sensitivity at OPPY coated PAN T650 than at PAN HCB should be observed in this work, in accordance with the observed differences at their bare surfaces. However, th e opposite behavior was observed at the OPPY coated surfaces (Figure 5 8B). This observation is clear whe n the cyclic voltammograms at the OPPY coated (Figure 5 9) and bare (Figure 3 3 ) PAN electrodes are compared. The behavior could be due to a (i) high b lockage of the nanostructured PAN T650 by OPPY membrane, stated earlier and shown in Table 5 1 and/or (ii) deterioration of the nanostructured substrate of PAN T650 during the OPPY membrane fabrication process. In the results of the analytical measurements at OPPY coated electrodes shown in Figure 5 8A, the membrane porosities at PAN T650 are within the range of those at PAN HCB nanostructured surfaces. Thus, the effect of (i) is less likely. To determine a possible effect due to (ii), the signals from the membrane fabrication procedure (Chapter 4) were evaluated. In the membrane fabrication procedure as described in Chapter 4, the chronocoulometric (CC) signals for background in TBAP and PY electropolymerization were acquired. Anson plots
144 of background CC d ata (Figure 4 3) showed a linear increase of Q with t 1/2 indicating diffusive behavior during background charging instead of the expected capacitive behavior. 36 The non zero Anson slope could be due to (I) water oxidation and/or (II) electrode surface oxidation during charging of the nanostructured electrode. 164 The Anson slope for the reverse st e p was found to be ~50 % of that for the forward step, suggesting that the background charging process is ~50 % reversible In a separate experiment, reasonable overlap of the background charging curves was observed when the procedure was repeated (n = 6) at the same electrode (Figure 5 10), indicating that the electrode surface nanostructure does not change significantly during the acquisition of the background charge. This indicates that the ~50% reversibility could be inherent with associated processes for (I) and that process (II) is less likely. These results show that changes at the nanostructured electrode substrate should not be significant during both background recordings and PY electropolymerization. Changes in the electrode radius of both bare and OPPY coated nanostructured electrodes after analytical determinations provided a clue as to why the sensitivity of PAN T650 deteriorates. Investigations on nanostructured surfaces of PAN carbon fiber microdisks show that they are very stable after analytic al measurements (~10, 000 scans) at 500 V s 1 as discussed in Chapter 3. Indeed, after analytical determinations at bare electrodes, the electrode radii of both nanostructured PAN T650 and PAN HCB decreased by only ~5% (Table 3 2) indicating high stabilit y. After analytical determinations at OPPY coated electrodes, the OPPY membrane degrades, by oxidative mechanism, 109, 115 for which chain scission/rapture along the C N bond or between two C=O groups 109 occurs. Also, oxidation of the pyrrolidine segments to release CO 2 115 possibly through formation of COOH groups, 181 has been proposed
145 After analytical determinations at OPPY coated PAN HCB, the ferricyanide limiting current was approximately the same as that at the newly nanostructured electrode, as shown in Figure 5 11 indicating high nanostructure stability. However, the res ults at OPPY coated PAN T650 show that after analytical measurements, the ferricyanide limiting current (i.e., electrode radius) decreased tremendously (~50%) relative to that at its newly nanostructured electrode. Since the decrease in electrode radius of bare PAN T650 is only 5% after analytical determinations alone, it is possible that the overoxidation stage of the OPPY membrane fabrication procedure may have caused the further 45% reduction its radius. This significant reduction in electrode radius is an attribute of a deteriorated electrode surface nanostructure. In the light of its high susceptibility to overoxidation, 67, 70 we ascribe the deterioration at OPPY coated PAN T650 electrode to formation of excessiv e surface oxides (nonactive elements), as a consequence of which a tremendous decrease in analytical sensitivity of dopamine at that surface was observed. Conclusions High permeability of the porous ultrathin (1 3 nm thick) OPPY membranes results in effici ent diffusive flux at the OPPY coated electrodes. A low nM LOD was achieved for 2 surface area because, in addition to high permeability, OPPY membran es can preconcentrate cations. Differences in structure of the membranes that are electrodeposited on the nanostructured PAN T650 and PAN HCB electrodes are a result of the differences in the structure of the fiber substrates. The LOD for dopamine at the OPPY coated nanostructured PAN electrodes is 15 nM. Lower LOD s can be achieved by significantly increasing the surface area of the electrode. The LOD of ur ic acid is 30 nM, which is ~15 times lower than that at the bare electrode, because of the highe r S/N at the coated electrode. A significant contribution to the improvement in LOD for
146 dopamine is due to the enhanced diffusive flux thro ugh the porous membrane, which wa s veri fied by semiintegral analysis. The results provide insights into the principles of construction of functional nanoporous ultrathin membranes ad hered to nanostructured sensor surfaces, and offer the possibility of detection of nanomolar concentrations of multiple redox active metabolites. Because of the more defective microstructure of PAN T650 than PAN HCB carbon fiber, nanostructures fabricated on the former are partly overoxidized during the overoxidation step of the membrane fabrication procedure, causing a decreased electroactive site density whereas nanostructures on PAN HCB remain reasonably unchanged. A modified PPY overoxidation procedure, for example, 1 min electrolysis, can minimize excessive surface oxides formation at the OPPY coated PAN T650, while at the same time being sufficient for complete PPY overoxidation.
147 Table 5 1. Physical properties of bare and OPPY membrane coated nanostr uctured PAN microdisk electrodes PAN T650 PAN HCB Parameter Bare Membrane coated Bare Membrane coated r a a 1.7 2.9 1.7 2.9 1.6 5.5 1.6 5.5 f b (%) 21 75 11 73 p c (%) 6 62 7 79 d 0.38 0.94 0.21 0.93 R a e 1.2 [0.61 2.2] 0. 16 [0.065 0.65] 1.7 0.18 [0.039 0.66] 2 R o f 3.8 [2.6 6.6] 1.5 [0.89 3.0] 5.1 1.7 [0.71 3.2] N g (x 10 6 cm 2 ) 2.0 [0.59 3.5] 11 [2.8 32] 0.97 9.0 [2.5 50] k h (x 10 3 cm s 1 ) 4.8 [2.4 50] 8.8 [4.1 31] a Electroactive radius of the electrode usin g Equation 2 1 b Fraction of decreased ferricyanide limiting current at OPPY membrane coated electrodes (Equation 2 7). c Membrane porosity (Equation 2 8). d Fraction of electrode area covered with OPPY using Equation 2 9. e,f Radius of active element e and sep aration between their centers f using Equations 2 10 to 2 12. g Density of active elements (membrane pores) in the nanoensemble (Equation 2 14). g Membrane permeability (Equation 2 15). e,f,g,h Median values; range shown in square brackets. n = 1. Table 5 2 Sensitivity and peak to peak separation of dopamine (DA) and uric acid (UA) at OPPY coated nanostructured PAN microdisks Fiber Material Sensitivity a (nA M 1 ) (mA cm 2 M 1 ) LOD b (nM) p c (mV) DA UA DA UA DA UA PAN T650 0.7 [0.61 0.94] 4.1 30 816 [776 833] PAN HCB 1.4 [0.89 2.1] 0.31, 0.46 ** 15 789 [756 859] 670, 708** a,c Median values; range shown in square brackets. b N p p at 99 % confidence level. 141 One data point calibration for 200 nM in Figure 5 4B. ** From two calibrations.
148 Figure 5 1. Slow scan voltammograms of ferricyanide at naost ructured PAN microdisk electrodes before (dash) and after (dot) FSCV measurements of dopamine (~10,000 scans); 5 mM K 3 Fe(CN) 6 in 0.5 M KCl, 0.050 V s 1 Figure 5 2. Fit of Butler Volmer kinetics (Equation 2 13) to slow scan vol tammograms of ferricyanide at bare (dash) and OPPY coated (solid) nanostructured PAN microdisk electrodes. Conditions same as in Figure 5 1.
149 Figure 5 3. Correlation between number and radius of active elements (pores in OPPY membranes) at the OPPY coated nanostructured PAN microdisk electrodes. Data obtained from Equations 2 10 to 2 14
150 Figure 5 4. Background subtracted fast scan voltammograms illustrating nanomolar measurements at OPPY membrane co ated nanostructured PAN microdisk electrodes. 200 nM (A) dopamine and ( B) uric acid at PAN T650; (C) (F ) 100, 400, 700 and 1000 nM dopamine, respectively, at PAN HCB. 100 nM dopamine peak at 0.66 V vs SCE confirmed by calibration. Peak at 0.21 V vs SCE i s an artifact of background subtraction. 31 mM phosphate buffer, pH 7.4, 500 V s 1 250 scans averaged.
151 Figure 5 5. Semiintegrals (A, C) of background subtracted fast scan voltammograms (B, D) of 7 ) uric acid at bare (dash) and OPPY coated (solid) nanostructured PAN HCB electrodes, illustrating signal enhancement at the coated electrodes. Conditions same as in Figure 5 4.
152 Figure 5 6. Dopamine and uric acid calibration p lots at bare and OPPY coated PAN HCB microdisk electrodes ( r a bottom). Data for each calibration were obtained at the same electrode surface; n = 3 for dopamine and n = 2 for uric acid measurements at the same concentration. Error bars show uncertainty in nth measurements. C onditions same as in Figure 5 4.
153 Figure 5 7. Background subtracted fast scan voltammograms of uric acid at OPPY membrane coated nanostructured PAN HCB microdisk electrode. C onditions same as in Figure 5 4.
154 Figure 5 8. Sensitivity for dopamine as a function of (A) membrane porosity and (B) membrane thickness at OPPY coated nanostructured PAN HCB and PAN T650 (Insets) microdisk electrodes Sensitivity from calibration plo t s obtained in 1 7 M analyte; c onditions same as in Figure 5 4. Error bars show uncertainty in the slope of the calibration plot at the same electrode surface, obtained from linear regression analysis. p and OP calcu lated using Equations 2 8 and 2 29.
155 Figure 5 9. Background subtracted fast scan voltammograms at OPPY membrane coated nanostructured PAN microdisk electrodes, illustrating higher sensitivity for dopam ine at PAN HCB than at PAN T650. (A) (D) 1, 4, 7 and 10 M dopamin e, respectively. Conditions same as in Figure 5 4.
156 Figure 5 10. Repetitive CC curves for background charge at nanostru ctured PAN microdisk electrodes. 0.1 M TBAP; 0.650 to 0.950 and back to 0.650 V vs Ag/wire quasireference el ectrode. Figure 5 11. Slow scan voltammograms of ferricyanide at nanostru ctured PAN microdisk electrodes. Nanostructured surface (dash), after OPPY coating (solid) and after FSCV measurements (dot) of dopamine (~10,000 scans) a t the OPPY coated surface; Conditions same as in Figure 5 1.
157 Scheme 5 1. Illustration of analyte transport and preconcentration at porous membrane coated electrode. Reduced analyte (black dot) is oxidized (red circle) b y the applied anodic potential creating a concentration gradient that drives diffusive flux in bulk solution and in the membrane. Scheme 5 2. Model of th e partially blocked electrode. Active elements, R a and their center to cen ter separation, 2 R 0 at (A) bare and (B) OPPY coated electrode. The active elements at the bare electrode (A) are divided into smaller active elements at the coated electrode (B) by the membrane coating.
158 CHAPTER 6 NANOMOLAR DETECTION OF P NITROPHENOL VIA IN SITU GENERATION OF P AMINOPHENOL AT NANOS TRUCTURED CARBON FIB ER MICRODISK ELECTRO DES Background In Chapter 5, a general strategy for improving limits of detection (LODs) by coating nanostructured PAN carbon fiber microdisks with porous ultrathin membran e was demonstrated. In this Chapter, the application of these nanostructured microdisks to solve an interesting problem in alkaline phosphatase (ALP) based electrochemical immunosening the detection of p nitrophenol ( p NP) (Scheme 6 1) is demonstrated. This compound is released by a hydrolytic reaction between ALP and its p nitrophenyl phosphate ( p NPP) substrate when ALP is used as an enzyme label in sandwich immunoassays. Because of electrochemical challenges, the p NP is commonly detected spectrophot ometrically to indicate the number of immunoreactions, and therefore, the protein level in a given biological sample. When the amino analogue of the substrate, p aminophenyl phosphate ( p APP) is used, the released p aminophenol ( p AP) is detected by electr ochemical methods. It has been shown that electrochemical immunosensing using p AP/ p APP as a reporter offers better LODs for ALP than spectrophotometric detection of the nitro analogue, p NP/ p NPP. 198 Since the first use of p AP for immunosensing was described by Tang et al 199 many articles describing its applications have appeared in the literatu re. 8, 200, 201 Impressive advancement in this field has resulted from signal amplification by chemical and/or enzymatic redox cycling strategies. 202 205 Limoges et al 206 presented in depth theory of these strategies based on the 2,6 dicholoro derivative of p AP. Detection with p AP appeared to be a promising approach to adapt miniaturized electrochemical devices to immunosensing. Yet, as noted by the first advocates of this analyte, this detection is limited by the inherent chemical instability of the p AP/ p APP system. 198 The p
159 APP substrate undergoes slow autohydrolysis and can contribute to a large background in p AP detection. In alkaline media (e.g. pH 8.0 10), which are optimum for catalytic activity of ALP, the p AP solution turns brown quickly if not protected from light, making p AP calibrations difficult. Limoges et al 207 demonstrated that the use a moderately stable dichloro substituted p AP/ p APP system can offer possibilities for overcoming these problems These challenges with maintaining chemical integrity have hampered commercialization of electrochemical devices that could utilize the p AP/ p APP system, while for high stability reasons, spectrophotometric detection using the nitro analogue p NP/ p NPP s ystem is well established and incorporated in commercial assay kits. 208 Unlike spectroscopic detection in ALP based immunoassays, direct electrochemical detection schemes using p NP is limited in the literature beca use of electrochemical challenges, including its irreversible behavior 199 and high anodic potentials (~0.9 V vs SCE at pH 9.0) for p NP oxidation. 209 212 Reduction of p NP also requires high cathodic potentials ( 0.84 vs SCE at pH 7), at which oxygen reduction interferes with the resulting analytical signal. 213 Such interference has been suggested to be absent at a bi smuth film electrode, where the ALP assay was performed by reduction of p NP at pH 7 214 Moreover, the oxygen interference was overcome in acid phosphatase assays involving reduction of p NP in moderately acidic med ium, thereby allowing a thermodynamically lower reduction potential. 215, 216 In spite of p AP instability, its higher analytical sensitivity than p NP 199 has been the motivation for alt ernative enzyme free strategies involving generation of p AP catalytically from p NP and detection in situ 217, 218 Although, these catalytic strategies are very complex, they point to the significance of catalytic nanostructures in facilitating reduction of p NP to p AP, which can then be detected electrochemically to allow a highly sensitive assay. A simple alternative,
160 mediated by direct reductive catalysis of p NP at an electrode surface, can offer possibilities for direct detection at simply designed electrode surfaces, as has been demonstrated at a silver nanoparticle coated GC surface. 219 Improved mass transport at nanofeatures of the electrode increased the local concen tration of p NP at the electrode/solution interface resulting in lower overpotential for reduction of p NP. Herein, the direct reductive catalysis of nanomolar concentrattions of p NP to p AP and in situ detection of the generated p AP at nanostructured ca rbon fiber microdisks by fast scan cyclic voltammetry (FSCV) at 500 V s 1 is demonstrated. Together, this combination of electrode surface nanostructure and a fast instrumental technique allow rapid reduction of p NP to p AP at lower potentials, thereby av oiding the concomitant kinetically slow reduction of interfering oxygen. A literature search located several reports of kinetic and mechanistic studies of p NP in aprotic solvents by fast scan voltammetry, 127, 220 b ut none for quantitative analysis. Nevertheless, the application of the FSCV technique to the study of deaminase kinetics by monitoring the decrease in adenosine anodic current 7 at 400 V s 1 suggests easy implementa tion of the detection strategy in this work with ALP based immunoassay platforms. The development of a buffer system to adapt the p NP detection strategy to these platforms is also described. Results and Discussion Detection of p AP at Nanostructured PAN C arbon Fiber Microdisks Because of their high analytical sensitivity in electrochemical biosensing, nanostructured PAN carbon fiber microdisks were used as the working electrodes in these investigations. To realize their practical application, it was desira ble to detect p AP at micro to submicromolar concentrations. The large active surface area and rapid rate of mass transport to their surface nanofeatures contribute to the high analytical sensitivity of nanostructured PAN carbon fiber microdisks for biolo gically relevant molecules.
161 Figure 6 1 shows background subtracted voltammograms of micromolar concentrations of p AP at 500 V s 1 at both OPPY coated and bare nanostructured PAN T650 microdisks. The results in Figure 6 1A show that 500 nM p AP could be d etected with significant S/N ratio at the OPPY coated surface. Analytical sensitivity is 4 times higher at the OPPY coated than at bare electrodes (Table 6 1) using the anodic peaks at 0.59 V and 0.56 V vs SCE, respectively. The enhanced analytical sensit ivity of p AP at the OPPY coated PAN T650 microdisk was surprising in light of the decreased dopamine sensitivity at such surfaces, suggested to result from surface overoxidation (Chapter 5). Nevertheless, this result demonstrates that the sensitivity can be enhanced with this strategy. The reaction of p AP at electrode surfaces involves a two electron, two proton redox process to form p quinoimine ( p QI) ( Detection in Scheme 6 1). The acquired background subtracted voltammograms of p AP show multiple peaks labeled I VI (Figure 6 1), the peak potentials of which are summarized in Table 6 2. The results in Figure 6 1 show major anodic (II) and major cathodic (IV) peaks at ca. 0.58 and 0.25 V vs SCE, respectively, assigned to the redox processes of the p AP/ p QI couple. In addition to these peaks, the results in Figure 6 1 also show other anodic peaks (I and III) at ca. 1.2 and 0.21 V vs SCE, as well as cathodic peaks (V and VI) at ca. 0.54 and 0.86 V vs SCE, specifically reported in Table 6 2. The origin of peaks I and VI at ca. 1.2 and 0.86 V vs SCE, respectively, is unclear, since they are not concentration dependent. However, peaks III and V at ca. 0.21 and 0.54 V vs SCE, respectively, are concentration dependent, and thus can be produced by an imp urity in p AP or a secondary process (i.e. the hydroquinone/quinone redox system). It is well documented in the literature that p QI hydrolyzes to quinone, 221, 222 which can undergo further reversible redox processe s to form hydroquinone. 223 The hydroquinone produces
162 about the same anodic potential as p AP, whereas the quinone is reduced at a higher cathodic potential than p QI. 224, 225 The absence of these characteristic features in Figure 6 1 is consistent with the slow rate of hydrolysis of p QI, 226 making the hydroquinone/quinone system electrochemically undetectable. I n fact, an estimated half life of 2.8 s at pH 7.4 for p QI 221 precludes its hydrolysis at 500 V s 1 since within 8 ms of p QI formation ( Detection in Scheme 6 1), it is converted back to p AP. As will be shown belo w, data analysis indicates that peaks III and V can be ascribed to the redox process due to the presence of p (hydroxyamino)phenol/ p nitrosophenol couple ( the zoom into Reduction in Scheme 6 1 ) in accordance with a previous report, 225 to account for the impurity in p AP (Figure 6 1). Note that the peak at potentials positive of 1.2 V vs SCE is speculated to be an artifact of the background subtraction. In Chapter 4, it was demonstrated that dopamine, a phenol derivative, could be measu red reproducibly at the carbon fiber nanostructured microdisk electrodes, since the electrode surface is regenerated by the chosen potential waveform and potential limits. In direct detection of p AP, although submicromolar can be realized, these results a re highly irreproducible most probably because of instability of the analyte at pH 7.4, even under stringent deoxygenated and light protected conditions. As pointed out in the Background, interest in the use of p AP as a reporter molecule in the electroche mical detection of immunoassays stems from its high sensitivity and reversibility of its electrode reactions. 199 In addition, p AP is easily oxidized, since it has a low one electron redox potential of 0.063 V vs SCE, 227 compared to the value of 0.940 V vs SCE for p NP 228 However, the poor stability of the p AP/ p APP system limits its practical applications involving the use of ALP as the enzyme label in sandwich immunoassays. The nitro analogue, p NP/ p
163 NPP, is demonstrated herein to be the most practical reporte r system by electrochemical detection of nanomolar concentrations of p NP as described below. Nanomolar Detection of p NP at Nanostructured PAN Carbon Fiber Microdisks Since p NP could be easily reduced at a Ag nanoparticle coated substrate due to improved mass transport, 219 it occurred to us that the nanostructured substrate fabricated herein should have the same capability. Although reduction of p NP is difficult at conventional electrodes in alkaline media, 214, 229 the improved mass transport to the nanofeatures fabricated in this work can result in decreased overpotential for the reduction process to produce p AP. This strategy was exploited at 500 V s 1 The background s ubtracted voltamm ograms of 0.4 1 and 1 10 M solutions of p NP are shown in Figures 6 2A C and D F respectively. The results demonstrate the detection of 400 nM p NP at a microdisk electrode of small geometric area (~40 m 2 ). Using a 35 s incubation time for the reaction between ALP and p 1 for p NPP, 230 th is measurement (400 nM p NP) allows detection of ~1 ng mL 1 of ALP. Although the affinity of ALP for p NPP is about one half that for p APP, 198 this impressive result attests to th e possible practical utility of p NP nanomolar detection in ALP based immunoassays. Evidence for Reduction of p NP to p AP and in situ Detection of p AP To verify the generation of p AP from p NP, volt ammograms obtained from separate solutions of p NP and p AP were compared. The results in Figure 6 3 demonstrate reasonable overlap of representative voltammograms from separate solutions of the two analytes, thus providing the evidence that p NP was reduc ed to p AP, which was then detected in situ As summarized in Table 6 2, the voltammograms in Figure 6 2 have features (anodic peaks at 0.62 and 0.22; cathodic peaks at 0.29 and 0.65 V vs SCE clearly seen in Figure 6 2F ) similar to those in Figure 6 1. The high reproducibility of the overlap of the representative data (Figures 6
164 3 A, B and C) attest to this observation. It is also very interesting to note that the sensitivity of the two analytes is about the same (~0.35 nA M 1 ) when determined under the same conditions (Table 6 1). Reduction of p NP is known to be a six electron, six proton redox process that ultimately generates p AP ( the zoom into Reduction in Scheme 6 1 ). The proposed mechanism for this process assumes that formation of p AP occurs via p nitrosophenol ( 1 ), p (hydroxyamino)phenol ( 2 ) and p QI intermediates in both aqueous 229, 231, 232 and aprotic 233 media. The formation of ( 1 ) has been reported to be chemi cally irreversible in aprotic media. 127, 233 The key feature about this mechanism is that each subsequent intermediate has a lower reduction potential than its precursor. For example, ( 1 ) is easier to reduce than p NP 233 and electrochemical reduction of the p QI intermediate also occurs at a lower cathodic potential than that of ( 1 ), 225 thereby precluding the need for a high cathodic potential to reduce t he intermediates. Consequently, the reduction of p NP easily leads to the formation of p AP, since rapid dehydration of ( 2 ) to p QI also occurs. 225 As a matter of fact, the half life for the dehydration step is estimated to be 1.5 ms a t pH 9.0, 234 short e nough to generate p AP that will be detected 2 ms after the positive potential change at the 1.0 V switching potential (Figure 6 2). The effect of pH on shifts in peak potentials of p AP has been reported previously, 199 and su ch signatures were expected to confirm the in situ detection of p AP. The results in Figure 6 4 show that with an increase in pH from 4.8 to 9.0, the overpotential for oxidizing p AP decreases (i.e., decrease in E a,II from black to red curves in Figure 6 4 ) whereas that for reduction of p QI increases (i.e., increase in E c,IV from black to red), as expected. The same behavior also applies to E a,III and E c,V of the suggested p (hydroxyamino)phenol/ p nitrosophenol couple since its
165 reversible redox reaction i s pH dependent. 225 one of them is probably due to oxidation of p (hydroxyamino)phenol at pH 4.8. In Figure 6 4, the presence of multiple anodic peaks on the voltammogram obtained at pH 4 .8 is consistent with complexity of follow up redox reactions between protonated p QI and unreacted p AP (half life of rate determining step (RDS) < 3.5 s ). 235 As a result, it is possible that during the 2.5 s of 250 scans at 500 V s 1 such complex follow up reactions can occur to produce the multiple anodic peaks. On the other hand, this follow up reaction at pH 9.0, now between neutral p QI and unreacted p AP is precluded, since the half life (~73 s ) 235 of its RDS is ~30 times longer than the signal acquisition time (2.5 s). Consequently, the acquired voltammogram at the pH of interest is free from such complex features, allowing easy identification of peaks for calibration and, thus, easy implementation of this detection strategy with an ALP based immunoassay. Development of TRIS Buffer Medium for Detection of p NP Until now, the detection of p NP has been tested in phosphate buffer at 7.4. However, inorganic phosphate inhibits the release of p NP from p NPP in an ALP based immunoassay. 230 As a res ult, transphosphorylating agents such as aminoalcohols are commonly used as buffers in assaying ALP. 207, 236 During ALP catalysis of reaction with substrates such as p NPP, transphosphorylation results in a net tran sfer of the released phosphate to a hydroxyl group of the buffer ing agent This transfer has been suggested to be more rapid than transfer to water, 237 thereby enhancing ALP activity by limiting formation of inorganic phosphate. It is also known that magnesium and zinc ions 238 as well as alkaline media activate ALP, with optimum activity achieved in the pH 8 10 range. Therefore, these key factors were considered in the design of the buf fer medium for the reaction of ALP with p NPP and subsequently for detection of the released p NP using the strategy described above. In this work,
166 a model TRIS buffer based on the transphorylating agent, 2 amino 2 (hydroxymethyl)propane 1,3 diol (Tris), w as developed using a composition modified from those reported previously, 207, 236 because Tris enhances the activity 237 and specificity 239 of ALP towards a substrate. From ele ctroche mical point of view, Tris is relatively 1.05 V vs SCE ). 239 To check the effect of pH, background phosphate, but at pH 9.0. Figure 6 5 shows a comparison of the ferricyanide voltammograms after recording the background currents in the phosphate buffer at pH 7.4 and 9.0 at the same nanostructured microdisk. The results show that the magnitudes of the currents and other voltammetric features are about the same. Therefore, at pH 9.0, any change(s) in the voltammetric characteristics in a different buffer can be ascribed to redox process(es) of the buffer compon ents. Figure 6 6 shows the same comparison as in Figure 6 5, but with 31 mM phosphate at pH 7.4 and 100 mM TRIS Cl at pH 9.0. The results in Figure 6 6B show an onset of an anodic peak at ~1.0 V vs SCE at the nanostructured electrodes in 100 mM TRIS Cl at pH 9.0 and at 10 V s 1 This feature is also observed at 500 V s 1 (onset at 1.3 V), but has a lower faradaic/background current ratio (i.e., one third and one sixth at PAN T650 and PAN HCB, respectively) than at 10 V s 1 This suggests that the resid ual faradaic process(es) that produce(s) this peak is/are slow at 500 V s 1 and this characteristics should be beneficial for good background subtraction in FSCV. It is also noteworthy th at the background current in TRIS Cl is lower than that in the phosph ate buffer. Although appears to be also beneficial for good background subtraction, the low electrical conductivity of TRIS Cl buffer consequently decreases analytical sensitivity as discussed below.
167 At pH 9.0 of TRIS Cl buffer and potentials 1.1 V vs SCE, pH dependent oxidation of the amine group of T ris 239, 240 and chloride oxidation ( E 0 = 1.12 vs SCE ) 36 can occur, suggesting that one/both of these components can contribute to the residual faradaic current in Figures 6 6B and C. When the chloride is replaced with acetate, the results at 500 V s 1 show that the residual faradaic curre nt starting at 1.3 V vs SCE almost disappears (Figure 6 7B), indicating that this process could be due to chloride oxidation as has been suggested for the behavior of other chloride containing physiological media. 24 1 To prevent chloride interference, acetate was chosen as a counter ion because of its use as a buffer in immunoreactions of ALP labeled IgG, 239, 242 and as a source of magnesium ions in a commercial assay kit, Qua ntichrom TM Alkaline Phosphatase Assay Kit 208 Additionally, no residual faradaic current was produced in the acetate media at 500 V s 1 even though acetate oxidizes easily in nonaqueous medium via the Kolbe reactio n. 243 Optimization of p NP Sensitivity with Concentration of Supporting Electrolyte Using dopamine detection as a standard, it was found that sensitivity for dopamine in the TRIS Cl buffer, pH 7.4 (0.45 nA M 1 ) was one half of that in the phosphate b uffer, pH 7.4 at nanostructured PAN HCB electrode (Table 6 1). Also, the electrical conductivity of TRIS Cl was found to be lower, in agreement with the lower background current compared to the phosphate buffer (Figure 6 6C). These observatio ns are also in accord with ~ 25 % lower analytical sensitivity of p NP in TRIS Cl than in phosphate buffer (Table 6 1). Figure 6 7A shows that by increasing the concentration of 31 mM phosphate electrolyte in TRIS Cl buffer, the background current approached that of the phosphate buffer as was expected. The same concept was used to modify the TRIS Ac buffer with 31 mM sodium acetate (Figure 6 7B). Although the concentrations of the phosphate and acetate electrolytes were the same, the higher background current of the form er (Figure 6 7A) is in agreement with its higher electrical conductivity (3 .2 m S cm 1 ) than that (1 .7 m S cm 1 ) of the latter in Figure 6 7B. The
168 results in Figure 6 7C also demonstrate that the electrical conductivity of the resulting medium (with increase d background current in Figure 6 7B) correlates with the electrolyte concentration. Therefore, by increasing the electr ical conductivity of the TRIS Ac buffer (with addition of sodium acetate) to produce the same magnitude of background current as that of the phosphate buffer, the optimum analytical sensitivity of p NP could be realized. As an inaugural test of this hypothesis, a plot of signal intensity of 2 M p NP versus sodium acetate concentration in TRIS Ac buffer was prepared (Figure 6 8). With addit ion of 250 mM acetate, the results show that the sensitivity of p NP can be doubled to 0.84 nA M 1 which is about the same as that of dopamine at the PAN HCB nanostructured electrode (Table 6 1). In fact, the background signal of this buffer composition overlapped with that of the 31 mM phosphate, suggesting that the optimum sensitivity for p NP achievable at the bare nanostructured PAN HCB microdisk is ca. 1 nA M 1 Conclusions The detection of p NP by reducing it to p AP and subsequent in situ measurem ent of p AP has been demonstrated at nanostructured carbon fiber microdisk electrodes. The conversion of p NP to p AP at the nanostructured surface was confirmed by overlapping signals of separate solutions of the two compounds. An optimum buffer, based on TRIS Ac to adapt this detection strategy to ALP based immunoassays was developed and optimized to enhance the sensitivity for p NP detection. This sensitivity (0.84 nA M 1 ) is reasonably close that of dopamine (0.9 nA M 1 ) pointing to the high capabilit y of this strategy for achieving impressive LODs in ALP based immunoassay applications. The nanomolar capability of this strategy can also be applied to environmental monitoring of released p NP from industrial processes.
169 Table 6 1. Detection sensitivity for p aminoophenol ( p AP ), p nitrophenol ( p NP), and dopamine (DA) in different buffer media at bare nanostructured PAN electrodes Analyte Sensitivity (nA 1 ) a Phosphate b TRIS Cl TRIS Ac p AP 0.38 [1 10] 0.39  d p AP 1.59 c [0.5 2] p NP 1.22 [0 .4 1] p NP 0.34 [ 4 10] 0.26 [ 1 10] e 0.62 [ 2] g 0.84  h,* DA 0.90 [1 10] 0.45 [1 10] f,* a Data collected at bare PAN T650 unless stated otherwise; 500 V s 1 250 scans averaged. Value(s) in square brackets is/are concentration(s) (M). b 31 mM phosphate buffer, pH 7.4. c At OPPY coated PAN T650. d 50 mM Tris/HCl/1 mM MgCl 2 /2 mM Na 2 HPO 4 pH 9.0. e 2 2 HPO 4 pH 9.0/ 9 mM phosphate buffer, pH 7.4. f Same medium as in d, but pH was adjusted to 7.4 with a few drops stock solution of H 3 PO 4 g 2 /2 mM Na 2 HPO 4 pH 9.0/62.5 mM NaAc. h 12.5 mM Tris/AcH 2 /2 mM Na 2 HPO 4 pH 9.0/250 mM NaAc. At bare PAN HCB Table 6 2. Average peak potentials of fast scan voltammograms of p amin ophenol ( p AP) and p nitrophenol ( p NP) a Anodic Potential, E a (V vs SCE) Cathodic Potential, E c (V vs SCE) Analyt e I II III IV V VI p AP b 1.2 [1.1] 0.56 [0.59] 0.19 [0.23] 0.26 [ 0.23] 0.49 [ 0.59] 0.87 [ 0.86] p NP 0.97 0.62 0.22 0.29 0.65 0.89 a n = 3; Data at nanostructured PAN T650 microdisks; 500 V s 1 250 scans averaged. b Data in square brackets colle cted at OPPY coated surface.
170 Figure 6 1. Background subtracted fast scan voltammograms of p aminophenol illustrating detection of (A) (D) 0.5, 1, 2 and 4 M concentrations, respectively, at OPPY membrane coated, and (E) (H) 1, 4, 7 and 10 M at bare nanostructured PAN T650 microdisk electrodes. Peaks, II, III, IV and V assigned to electrode reaction of p aminophenol, p (hydroxyamino)phenol, p quinoneimine and p nitrosophenol; origin of I and VI is unclear. 31 mM phosphate buffer pH 7.4, 500 V s 1 250 scans averaged.
171 Figure 6 2. Background subtracted fast scan voltammograms of p nitrophenol at bare nanostructured PAN T650 microdisk electr odes showing detection of (A) (C) 400, 700 and 1000 nM concent rations, respectively, and (D ) (F) 4, 7 and 10 M. Peak assignments and conditions same as in Figure 6 1.
172 Figure 6 3. Evidence for detection of p nitrophenol ( p NP) as p aminophenol ( p AP) confirmed by overlap of background co rrected fast scan voltammograms of separate solutions of p NP and p AP at bare nanostruct ured PAN T650 electrodes. (A) (C) Highly reproducible overlap of representative data. Conditions same as in Figure 6 1.
173 Figure 6 4. Shift of peak potentials of p aminophenol ( p AP) with pH of medium. p AP generated by electroreduction of p nitrophenol at different pH and detected in situ Peaks at potentials E a,II and E c,IV assigned to p AP/ p QI; E a,III and E c,V to the suggested p (hydroxyamino)phenol/ p nitrosophenol redox couple; origin of peaks drops of 1 M NaOH or a stock solution of H 3 PO 4 ; 500 V s 1 250 scans averaged.
174 Figure 6 5. Slow scan voltammograms of ferricyanide after background current measurements in phosphate at different pH at bare nanostructured PAN microdisk electrodes. (A) Ferr icyanide signal after recording background current in 31 mM phosphate, at pH 7.4 (solid) and 9.0 (dash); 5 mM K 3 Fe(CN) 6 in 0.5 M KCl 0.050 V s 1 ; (B) Background currents in 31 mM phosphate medium, pH 7.4 (solid) and 9.0 (dash) at 10 V s 1 50 scans averaged. (C) Same medium as in (B), but at 500 V s 1 250 scans averaged.
175 Figure 6 6. Effect of buffer medium on slow scan voltammograms of ferricyanide and background current at bare nanostructured PAN microdisk electrodes. (A) Ferr icyanide signal after recording background current in 31 mM phosphate buffer, pH 7.4 (solid) and in TRIS Cl, pH 9.0 (dash); 5 mM K 3 Fe(CN) 6 in 0.5 M KCl, 0.050 V s 1 ; (B) Background currents in 31 mM phosphate medium, pH 7.4 (solid) and in TRIS Cl pH 9.0 (dash). TRIS Cl is 100 mM Tris/HCl/1 mM MgCl 2 /2 mM Na 2 HPO 4 Tris/HCl is 2 amino 2 (hydroxy methyl)propane 1,3 diol hyd roch loride; 10 V s 1 50 scans averaged. (C) Same medium as in (B), but at 500 V s 1 250 scans averaged
176 Figure 6 7. Effect of supporting electrolyte on background current and electrical conductivity of TRIS buffers (A, B) Background currents at bare nanostructured PAN T650 electrode in different volume ratios of (A) TRIS Cl:Phosphate and (B) TRIS Ac:NaAc. (C) Effect of sodium acetate supporting electrolyte concentration on conductivity of TR IS Ac/NaAc buffer; TRIS Ac/NaAc is MgAc 2 2 HPO 4 pH 9.0/31 mM NaAc; Tris/AcH is 2 amino 2 (hydroxymethyl)propane 1,3 diol acetatic acid; TRIS Cl medium has chloride instead of acetate and has the same composition as TRIS Ac, but with 31 mM phosphate b uffer, pH 7.4; 500 V s 1 250 scans averaged.
177 Figure 6 8 p NP in TRIS Ac/NaAc buffer medium at bare nanostructured PAN electr odes. TRIS Ac/NaAc is 12 2 /2 mM Na 2 HPO 4 pH 9.0 /NaAc ; 500 V s 1 250 scans averaged. Error bars for n = 3 at the same electrode surface.
178 Scheme 6 1. Detection of p nitrophenol ( p NP) by electrochemical reduction to p aminophe nol ( p AP) and subsequent measurement of analytical signal due to p AP. The zoom into Reduction is p roposed mechanism for electrochemical reduction of p NP to p AP) via p nitrosophenol ( 1 ), p (hydroxyamino)phenol ( 2 ) and p quinoimine ( p QI) intermediates.
179 CHAPTE R 7 FABRICATION OF SINGL E WALLED CARBON NANOTU BE FILM MICROELECTRODES Background The research work described in Chapters 3 through 6 was conducted at nanostructured r microdisk approach is used to fabricate microdisk electrodes using a new carbon material in the form of single walled carbon nanotubes (CNT). This Chapter explo res the capabilities of the nanos tructured surface of CNT film ultramicroelectrodes (3 m dia.) for electrochemical biosensing. 15 and many tens of microns length, 25 the surface nanostructure formed from connected nanotubes can provide improved mass transport for high analytical sensitivity and rapid measurements. Additionally, the small dimensions of CNTs offer the possibility for fabrication of ultramicroelectro des having smaller diameters compared to carbon fibers (~10 m). Although, carbon fibers can be electrochemically etched to form ~2 m dia. microdisks, 69 such surfaces can be expected to have a low density of nanofeatures (300 600 nm dia. 10, 68 ), thus limiting the capability for electroanalytical sensitivity. Therefore the use of CNTs offers possibilities for constructing a surface having a high density of nanofeatures, while allowing the freedom to fabricate ultramicroelectrodes of small (< 5 m) dimensions, as demonstrated herein. CNTs have been suggested to be prom ising new materials for electrochemical biosensing. 20, 52, 244 As a result, several reports in the literature have explored such possibilities, and these have been reviewed recently. 245 The typical approach for making CNT electrodes is to coat the nanotubes on conducting substrates including glassy carbon 246 and carbon fiber 174, 247
180 However, such substrates contribute to the electrochemical current, thereby making it difficult to determine the inherent electroanalytical capabilities of the CNT. Alternative procedures have focused on fabric ating CNT electrodes by processing the nanotubes into networks (< 1 % surface coverage) on insulating supports, 248 mats, 20 and fibers, 5 2, 53 permitting the sensing capabilities solely due to reactions at the CNT material to be determined. Other studies have involved electrodes obtained from nanotubes processed into towers 27 and papers. 54 Unlike the CNT fiber and tower, the network and mat ultramicroelectrodes can be fabr icated reproducibly because the processing and density of such networks and mats can be easily controlled. In spite of these developments, reports in the literature show that metallic nanoparticle catalysts, which are occluded between graphitic sheets of CNTs, contribute to the electrochemical activity. 15, 249 This underscores the need for highly purif ied CNT materials in order to explicitly determine their biosensing capabilities. To this end, the first report of a purified CNT network micro electrode was probably made by the MacPherson Group. 248 However, the Mac pherson Group noted that low analytical reproducibility and high Ohmic drop are produced at these networks. Nonetheless, these challenges have provided the motivation for making better materials mats (highly dense) that allow ed fabrication of robust e lectrodes. 250 In this work, highly pure (> 99 %) single walled CNTs processed into a film (50 nm thick) was used to fabricate 3 m dia. ultramicroelectrodes by electron beam lithography. The sheet resistivity of thes e materials (0.075 1 ) 56 is significantly lower than that previously reported for mats (15 1 ). 250 Together with their small geometric radii, rapid measurements can be expected at these ultramicroelectrodes. The preliminary findings on
181 investigations of t hese nanostructured single walled CNTs for electrochemical biosensing is discussed. Results and Discussion Nanostructured CNT Film Microdisk The nano sized dimensions of CNTs make it easy to construct the carbon material into a nanostructure. Figure 7 1 sh ows the AFM images of the nanostructured CNT film electrode at different stages of processing. As shown in Figure 7 1A, bundled single walled CNTs (few tens of nm dia.) form the network, and sections of dense aggregation are apparent, indicating high conne ctivity between the intertwined CNTs. The random array of these bundles/aggregates of nanometer dimensions constitute active sites that are separated by pores to produce a nanoensemble surface. Because of small dimensions of these active sites, mass transp ort to the nanofeatures is efficient. These features and the high surface to volume ratio allow high analytical sensitivity, and thus low detection limits of biomolecule detection. 248 After spin coating with poly(m ethyl methacrylate) PMMA and subsequent exposure of the underlying microdisk electrode (Figure 7 1C) by electron beam lithography, the image in Figure 7 1B shows that the resolution of the bundles in the starting CNT network decreases. It is possible that PMMA was not efficiently removed from the CNT network. Also, the density of fine features between the aggregates in Figure 7 1A is decreased in Figure 7 1B after the electrode processing. This can be an indication of decreased connectivity between the aggr egates, and thus lower conductivity, which can have implications for electrochemical performance. Determination of Electroactive Radius of CNT Film Microdisk For purposes of electroanalytical applications, it is important to determine the electroactive ra dius/area. At the scan rate of 0.050 V s 1 diffusion layers around the nanofeatures overlap,
182 permitting the use of Equation 2 1 to determine the electroactive radius. 46, 74 The theoretical limiting current 36 of reduction of 5 mM ferricyanide at an electrode with 3 m diameter, according to Equation 2 1, is 2.2 nA. Ideally, this should be the minimum me asurable limiting current needed to confirm that the exposed diameter of the CNT film microdisk is 3 m. However, as shown in Figure 7 2A, the electroactive radius, r a (i.e. 0.3 nm from experimentally determined limiting current) is ~3 orders of magnitude lower than expected. The extremely small electroactive radius suggests that a significant part of the electroactive area must have been blocked, preventing access of the analytes to those sites, consistent with the theory of charge transfer at partially bl ocked surfaces 73 Figure 7 2B shows the background current recorded in sodium chloride. The anodic peak onset at 0.6 V vs Pd wire can be due to oxidation of the CNT material. The results also show a current trace f or reduction of dissolved oxygen (~260 M at 1 atm) in water. Using the first reduction wave enveloped within 0.25 and 0.65 V vs Pd, and similar calculations as for the ferricyanide, the outcome of electroactive radius (1.2 nm) is about the same. More i mportantly, the results in Figure 7 2B show that dissolved oxygen acts as an internal standard for characterizing the electroactive radius. Analysis of E vs log[( i L i )/ i ] plots of Figure 7 2A using Equation s 2 3 and 2 5 gives a slope value of 0. 103 V, in dicating a slow kinetics of ferricyanide reduction ; reduction of oxygen (slope of 0.0 77 V from Figure 7 2B) also has slow kinetics T he se slope values are significantly higher than the corresponding values of 0.059 / n V 36 for a reversible electrode reaction, consistent with the very small fraction active sites 73 suggested to be present. Overall, these d ata show that the reactions of ferricyanide and oxygen at the CNT microdisk electrode are very slow.
183 A slower kinetics indicates a higher electrical resistivity of the CNT network. It has been shown that acid treatment of the nanotube material can result i n increased resistivity of single walled compared to double walled 244 and multi walled CNT 54 and manifest in decreased electron transfer rates at the electrode surface. However, the high density of (0.71 g cm 3 ) and c onductivity of the starting CNT network (Figure 7 1A) should have prevented the poor electrochemical performance observed here. Incomplete surface exposure can contribute to significant blockage of the active CNTs and cause the observations in this work. I n fact, PMMA has been reported to contaminate the CNT surface and result in poor electrochemical response, but a pre coating with SiO x layer was found to decrease this effect 47 It is unknown at present, as to whether the electron beam could damage the CNT material and contribute to the observed response. Conclusions This Chapter has described initial attempts to fabricate a 3m dia. CNT film ultramicr oelectrode by electron beam lithography. Interestingly, it is possible to determine the electroactive area in a blank electrolyte using oxygen reduction as an internal standard. The exposed geometric area is significantly less than that determined electroc hemically suggesting that a large fraction of the CNT network was blocked
184 Figure 7 1. AFM imag es of CNT network (A) before, (B) after PMMA spin coating and exposure and (C) 3 m dia. microdisk electrode exposed by electron bea m patterning of the spin c oated 200 nm PMMA layer Fi gure 7 2. Slow scan voltammograms of (A) ferricyanide and ( B) dissolved oxygen at CNT film microdisk electrode. (A) 5 mM K 3 Fe(CN) 6 in 0.5 M KCl and (B) 0.1 M NaCl; 0.050 V s 1
185 CHAPTER 8 SUMMARY AND FUTURE D IRECTIONS The material properties of graphitic nanomaterials have great impact on their electroanalytcal performance when used as electrochemical biosensors. It was demonstrated in this work that nanostructured microdisk surfaces of polyacrylonitrile (PAN) show higher electroanalytical sensitivity for dopamine and uric acid than pitch based carbon fibers. The electrode sensitivity for these biomolecules is controlled by adsorption and is higher for dopamine than for uric a cid. Further, the overall high sensitivity of PAN materials is due to the nitrogen defects and high material disorder that dictate the nanostructures fabricated on their microdisk surfaces. On the other hand, the electrode kinetics of dopamine and uric re actions are faster at pitch than at PAN based carbon fiber microdisks. The inherently faster kinetics of the uric acid reaction was demonstrated in this work. It was also demonstrated that peak to peak separations could be corrected by simple calculation of the resistances of the cell components. Although the electrochemical etch method used in this work to fabricate the nanostructured carbon fiber microdisk allowed a good comparison of the material properties as demonstrated in Chapter 3, this method had originally been optimized to enhance the analytical sensitivity at PAN HCB. 124 Future considerations should focus on optimizing the electrochemical etch method for PCH P25 and PAN T650 in order to harness their ful l capabilities for electrochemical biosensing, since it has been demonstrated that etching rate is directly related to the precursor material 69 as well as disorder of the carbon fiber. When their nanostructures are fabricated by the same electrochemical etch procedure, PAN materials should be suitable for measurements of analytical sensitivity, whereas pitch should allow access to reactions having very rapid rates
186 A novel concept for enhancing analytical sensitivity at porous ultrathin membrane coated electrodes was proven in this work (Chapters 4 and 5) It was demonstrated for the first time that such highl y permeable membranes provide capabilities for nanomolar detection of biomolecules at electrochemical biosensors having small geometric areas. The improved S/N ratio at these biosensors allowed nM LODs for dopamine and uric acid. Therefore, they should fin d applications in vivo and in measurements at single cells. The demonstrated strategy for fabricating the porous ultrathin membrane coated microdisks should open possibilities for tailoring surface architectures for selective detection in complex biologica l samples. Using related electrochemical theories, 73, 123 it was also possible for the first time, to estimate the dimensions of the random array of pores and their separation in the membranes at the membrane coate d nanostructured microdisks. This allowed a model to be developed for the membrane structure. In light of recently reviewed 84 and use of microensemble theory to simulate 78 th e behavior of random array of nanoscopic features on electrodes, development of theories for their behavior can also be foreseen and should permit more accurate modeling of such surface architectures. In an effort to improve the sensitivity of one of the b est materials, PAN T650, by membrane coating, the biosensing capabilities were surprisingly deteriorated. The high disorder of the PAN T650 material could have resulted in excessive surface oxidation, 70 thereby cont ributing to this behavior. The effect of the membrane fabrication should be investigated to address this issue. In light of the high susceptibility of PAN T650 to overoxidation, the investigations should focus on optimization of: first, the electrochemical etch procedure to minimize surface oxide formation, and second, the membrane fabrication procedure to prevent
187 deterioration of the underlying substrate of carbon nanofeatures. The knowledge base accrued from such investigations should be applicable to rel ated graphitic nanomaterial systems. An interesting application of the nanostructured PAN carbon fiber microdisks for nanomolar detection of p nitrophenol ( p NP) was demonstrated in Chapter 6 The rationale for this investigation was to develop alternative electrochemical detection schemes for ALP based immunoassays using the p NP/ p nitrophenyl phosphate ( p NPP ) reporter system. A challenge is expected to be presented by this system because of interference from concurrent redox processes of the starting p N PP substrate as well as its electrolytic and hydrolytic products. 199, 215 Using the potential waveform in this work, oxidation of p NP was not detected, and thus that of p NPP will not be observed (oxidation peak o ccurs at 0.550 V higher than that for p NP at pH 9. 199 On the other hand, reduction of p NPP interferes with that of p NP at pH 5 since their reduction peak potentials are 0.070 V apart 2 15 Nonetheless, larger E p E p/2 for p NPP than p NP suggests slower kinetics of p NPP reduction 215 Even if electroreduction of p NPP to p APP occurs at 500 V s 1 the anodic peaks of p APP and p AP may be distingui shable since their peak potentials are 0.420 V apart. 199 Future investigations should test this hypothesis, since it is imperative to use p NPP in order to couple the proposed detection to ALP based immunoassays. Additionally, potential limits could also be optimized, since anodic and cathodic potentials of the p NPP and p A PP are higher than those of the ir corresponding hydrolytic products. 199, 215 The fabrication of nanotructured CNT fi was also demonstrated in Chapter 7 The cause of the smaller electroactive area than expected should be investigated with focus on optimizing processing parameters during exposure to minimize blocking of th e CNT surface. 47 Alternative exposure techniques like photolithography could also be explored as demonstrated elsewhere. 46
188 LIST OF REFERENCES (1) Heien, M. L. A. V.; Phillips, P. E. M.; Stuber, G. D.; Seipel, A. T.; Wightman, R. M. Analyst 2003, 128 1413 1419. (2) Zhang, B.; Heien, M. L. A. V.; Santillo, M. F.; Mellander, L.; Ewing, A. G. Anal. Che m. 2011, 83 571 577. (3) Fang, H.; Vickrey, T. L.; Venton, B. J. Anal. Chem. 2011, 83 2258 2264. (4) Rossier, J. S.; Girault, H. H. Lab Chip 2001, 1 153 157. (5) Mauzeroll, J.; Bard, A. J.; Owhadian, O.; Monks, T. J. Proc. Natl. Acad. Sci. 2004, 101 17582 17587. (6) Z han, W.; Bard, A. J. Anal. Chem. 2006, 78 726 733. (7) Xu, Y.; Venton, B. J. Phys. Chem. Chem. Phys. 2010, 12 10027 10032. (8) Pariente, F.; Hernandez, L.; Abruna, H. D.; Lorenzo, E. Anal. Chim. Acta 1993, 273 187 193. (9) Prabhulkar, S.; Alwarappan, S.; Liu, G.; L i, C. Z. Biosens. Bioelectron. 2009, 24 3524 3530. (10) Kannan, B.; Williams, D. E.; Booth, M. A.; Travas Sejdic, J. Anal. Chem. 2011, 83 3415 3421. (11) Zachek, M. K.; Hermans, A.; Wightman, R. M.; McCarty, G. S. J. Electroanal. Chem. 2008, 614 113 120. (12) Berberia n, K.; Kisler, K.; Fang, Q.; Lindau, M. Anal. Chem. 2009, 81 8734 8740. (13) McCreery, R. L. Chem. Rev. 2008, 108 2646 2687. (14) Huffman, M. L.; Venton, B. J. Analyst 2009, 134 18 24. (15) Dumitrescu, I.; Unwin, P. R.; MacPherson, J. V. Chem. Commun. 2009, 45 6886 6901. (16) McCreery, R. L. In Electroanalytical Chemistry ; Bard, A.J., Ed.; Marcel Dekker: New York, 1991 17 ; pp 222 364. (17) Donnet, J. B.; Rebouillat, S.; Wang, T. K.; Peng, J. C. M., Ed., Carbon Fiber s 3rd ed.; Marcel Dekker: New York, 1998 (18) Minus, M. L.; Kum ar, S. JOM 2005, 57 52 58. (19) Edie, D. D. Carbon 1998, 36 345 362.
189 (20) Maldonado, S.; Morin, S.; Stevenson, K. J. Analyst 2006, 131 262 267. (21) Pietrzak, R. Fuel 2009, 88 1871 1877. (22) Diefendorf, R. J.; Tokarsky, E. Polym. Eng. Sci. 1975, 15 150 9. (23) Paredes, J. I .; Martinez Alonso, A.; Tascon, J. M. D. Fuel Process. Technol. 2002, 77 78 293 300. (24) Shi, D.; Liu, N.; Yang, H.; Gao, J.; Pang, S.; Wu, X.; Ji, Z. J. Mater. Res. 1997, 12 2543 2547. (25) Fan, S.; Chapline, M. G.; Franklin, N. R.; Tombler, T. W.; Cassell, A. M .; Dai, H. Science 1999, 283 512 514. (26) Gong, K.; Chakrabarti, S.; Dai, L. Angew. Chem., Int. Ed. 2008, 47 5446 5450. (27) Yun, Y.; Shanov, V.; Schulz, M. J.; Dong, Z.; Jazieh, A.; Heineman, W. R.; Halsall, H. B.; Wong, D. K. Y.; Bange, A.; Tu, Y.; Subramaniam S. Sens. Actuators, B 2006, B120 298 304. (28) Rodriguez, N. M. J. Mater. Res. 1993, 8 3233 3250. (29) Ebbesen, T. W.; Ajayan, P. M. Nature 1992, 358 220 222. (30) Thess, A.; Lee, R.; Nikolaev, P.; Dai, H.; Petit, P.; Robert, J.; Xu, C.; Lee, Y. H.; Kim, S. G.; et a l. Science 1996, 273 483 487. (31) Heras, A.; Colina, A.; Lopez Palacios, J.; Ayala, P.; Sainio, J.; Ruiz, V.; Kauppinen, E. I. Electrochem. Commun. 2009, 11 1535 1538. (32) Lyon, J. L.; Stevenson, K. J. Langmuir 2007, 23 11311 11318. (33) Parthasarathy, R. V.; Phani, K. L. N.; Martin, C. R. Adv. Mater. 1995, 7 896 897. (34) Kyotani, T.; Tsai, L. f.; Tomita, A. Chem. Mater. 1995, 7 1427 1428. (35) Han, C. Y.; Xiao, Z. L.; Wang, H. H.; Lin, X. M.; Trasobares, S.; Cook, R. E. J. Nanomater. 2009 ID562376, http://www.hindawi.com/journals/jnm/2009/562376/ (36) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications 2nd ed., Wiley: New York, 2004 (37) Hsueh, C. C.; Brajter Toth, A. Anal. Chem. 1993, 65 1570 1575. (38) Brajter Toth, A.; Abou El Nour, K.; Cavalheiro, E. T.; Bravo, R. Anal. Chem. 2000, 72 1576 1584. (39) Banks, C. E.; Compton, R. G. Analyst 2006, 131 15 21.
190 (40) Lord, S. S., Jr.; Rogers, L. B. Anal. Chem. 1954, 26 284 295. (41) Adams, R. N. Anal. Chem. 1958, 30 1576 1576. (42) Gonon, F.; Cespuglio, R.; Ponchon, J. L.; Buda, M.; Jouvet, M.; Adams, R. N.; Pujol, J. F., C. R. Hebd. Seances Acad. Sci., Ser. D 1978, 286 1203 1206. (43) Chen, S.; Kucernak, A. Electrochem. Commun. 2002, 4 80 85. (44) Hermans, A.; Seipel, A. T.; Miller, C. E.; Wightman, R. M. Langmuir 2006, 22 1964 1969. (45) Britto, P. J.; Santhanam, K. S. V.; Ajayan, P. M. Bioelectrochem. Bioenerg. 1996, 41 121 125. (46) Dumitrescu, I.; Unwin, P. R.; Wilson, N. R.; Macpherson, J. V. Anal. Chem. 2008, 80 3598 360 5. (47) Heller, I.; Kong, J.; Heering, H. A.; Williams, K. A.; Lemay, S. G.; Dekker, C. Nano Lett. 2005, 5 137 142. (48) Fung, A. O.; Tsiokos, C.; Paydar, O.; Chen, L. H.; Jin, S.; Wang, Y.; Judy, J. W. Nano Lett. 2010, 10 4321 4327. (49) Wang, K.; Fishman, H. A.; Dai, H.; Harris, J. S. Nano Lett. 2006, 6 2043 2048. (50) Liu, X.; Baronian, K. H. R.; Downard, A. J. Anal. Chem. 2008, 80 8835 8839. (51) Campbell, J. K.; Sun, L.; Crooks, R. M. J. Am. Chem. Soc. 1999, 121 3779 3780. (52) Punbusayakul, N.; Talapatra, S.; Ci, L.; Surareun gchai, W.; Ajayan, P. M. Electrochem. Solid State Lett. 2007, 10 F13 F17. (53) Guo, X.; Yun, Y.; Shanov, V. N.; Halsall, H. B.; Heineman, W. R. Electroanalysis 2011, 23 1252 1259. (54) Li, J.; Cassell, A.; Delzeit, L.; Han, J.; Meyyappan, M. J. Phys. Chem. B 2002, 106 9299 9305. (55) Barisci, J. N.; Wallace, G. G.; Baughman, R. H. J. Electrochem. Soc. 2000, 147 4580 4583. (56) Das, R. K.; Liu, B.; Reynolds, J. R.; Rinzler, A. G. Nano Lett. 2009, 9 677 683. (57) Kim, J. H.; Nam, K. W.; Ma, S. B.; Kim, K. B. Carbon 2006, 44 196 3 1968. (58) Wilson, N. R.; Guille, M.; Dumitrescu, I.; Fernandez, V. R.; Rudd, N. C.; Williams, C. G.; Unwin, P. R.; Macpherson, J. V. Anal. Chem. 2006, 78 7006 7015.
191 (59) McKnight, T. E.; Melechko, A. V.; Guillorn, M. A.; Merkulov, V. I.; Doktycz, M. J.; Culberts on, C. T.; Jacobson, S. C.; Lowndes, D. H.; Simpson, M. L. J. Phys. Chem. B 2003, 107 10722 10728. (60) Baker, S. E.; Tse, K. Y.; Lee, C. S.; Hamers, R. J. Diamond Relat. Mater. 2006, 15 433 439. (61) Baker, S. E.; Colavita, P. E.; Tse, K. Y.; Hamers, R. J. Chem. Mater. 2006, 18 4415 4422. (62) Correia, A.; Gil, A.; Garcia, N.; Madronero, A. J. Thermoplast. Compos. Mater. 1999, 12 373 387. (63) Soleymani, L.; Fang, Z.; Sargent, E. H.; Kelley, S. O. Nat. Nanotechnol. 2009, 4 844 848. (64) Soleymani, L.; Fang, Z.; Lam, B.; Bin, X.; Vasilyeva, E.; Ross, A. J.; Sargent, E. H.; Kelley, S. O. ACS Nano 2011, 5 3360 3366. (65) Munge, B. S.; Coffey, A. L.; Doucette, J. M.; Somba, B. K.; Malhotra, R.; Patel, V.; Gutkind, J. S.; Rusling, J. F. Angew. Chem., Int. Ed. 2011, 50 7915 7918. (66) Zhou, W. J.; Chen, Y.; Corn, R. M. Anal. Chem. 2011, 83 3897 3902. (67) Boateng, A.; Cohen Shohet, R.; Brajter Toth, A. Electroanalysis 2010, 22 2915 2923. (68) Kathiwala, M.; Affum, A. O.; Perry, J.; Brajter Toth, A. Analyst 2008, 133 810 816. (69) Schulte, A.; Chow, R. H Anal. Chem. 1998, 70 985 990. (70) Pamula, E.; Rouxhet, P. G. Carbon 2003, 41 1905 1915. (71) Boateng, A.; Cohen Shohet, R.; Brajter Toth, A. Electrochim. Acta 2011, 56 7651 7658. (72) Scharifker, B. R. J. Electroanal. Chem. Interfacial Electrochem. 1988, 240 61 76 (73) Amatore, C.; Saveant, J. M.; Tessier, D. J. Electroanal. Chem. Interfacial Electrochem. 1983, 147 39 51. (74) Zoski, C. G.; Yang, N.; He, P.; Berdondini, L.; Koudelka Hep, M. Anal. Chem. 2007, 79 1474 1484. (75) Hulteen, J. C.; Menon, V. P.; Martin, C. R. J. Che m. Soc., Faraday Trans. 1996, 92 4029 4032. (76) Zoski, C. G.; Wijesinghe, M. Isr. J. Chem. 2010, 50 347 359. (77) Davies, T. J.; Ward Jones, S.; Banks, C. E.; Del Campo, J.; Mas, R.; Munoz, F. X.; Compton, R. G. J. Electroanal. Chem. 2005, 585 51 62.
192 (78) Davies, T. J.; Compton, R. G. J. Electroanal. Chem. 2005, 585 63 82. (79) Phillips, C. G. J. Electrochem. Soc. 1992, 139 2222 2230. (80) Alfred, L. C. R.; Oldham, K. B. J. Electroanal. Chem. 1995, 396 257 63. (81) Zhu, H.; Tolmachev, Y. V.; Scherson, D. A. J. Phys. Chem. C 2010, 114 13650 13656. (82) Davies, T. J.; Banks, C. E.; Compton, R. G. J. Solid State Electrochem. 2005, 9 797 808. (83) Baker, W. S.; Crooks, R. M. J. Phys. Chem. B 1998, 102 10041 10046. (84) Murray, R. W. Chem. Rev. 2008, 108 2688 2720. (85) Liu, Y.; He, R.; Zhang, Q.; Che n, S. J. Phys. Chem. C 2010, 114 10812 10822. (86) Zoski, C. G.; Fernandez, J. L.; Imaduwage, K.; Gunasekara, D.; Vadari, R. J. Electroanal. Chem. 2011, 651 80 93. (87) Adams, K. L.; Jena, B. K.; Percival, S. J.; Zhang, B. Anal. Chem. 2011, 83 920 927. (88) Hsueh, C.; Brajter Toth, A. Anal. Chem. 1994, 66 2458 64. (89) Witkowski, A.; Brajter Toth, A. Anal. Chem. 1992, 64 635 641. (90) Debiemme Chouvy, C. Biosens. Bioelectron. 2010, 25 2454 2457. (91) Debiemme Chouvy, C.; Gallois, M. Surf. Interface Anal. 2010, 42 1144 1147. (92) Pihel K.; Walker, Q. D.; Wightman, R. M. Anal. Chem. 1996, 68 2084 2089. (93) Kim, E.; Xiong, H.; Striemer, C. C.; Fang, D. Z.; Fauchet, P. M.; McGrath, J. L.; Amemiya, S. J. Am. Chem. Soc. 2008, 130 4230 4231. (94) Gros, P.; Gibson, T.; Bergel, A.; Comtat, M. J. Elec troanal. Chem. 1997, 437 125 134. (95) Farhat, T. R.; Schlenoff, J. B. Langmuir 2001, 17 1184 1192. (96) Vlassiouk, I.; Smirnov, S.; Siwy, Z. Nano Lett. 2008, 8 1978 1985. (97) Witkowski, A.; Freund, M. S.; Brajter Toth, A. Anal. Chem. 1991, 63 622 6. (98) Ishimatsu, R.; Kim, J.; Jing, P.; Striemer, C. C.; Fang, D. Z.; Fauchet, P. M.; McGrath, J. L.; Amemiya, S. Anal. Chem. 2010, 82 7127 7134. (99) Smith, J. J.; Zharov, I. Langmuir 2008, 24 2650 2654. (100) Striemer, C. C.; Gaborski, T. R.; McGrath, J. L.; Fauchet, P. M. Nature 200 7, 445 749 753.
193 (101) Ignacio de Leon, P. A. A.; Zharov, I. Chem. Commun. 2011, 47 553 555. (102) Cheng, Q.; Brajter Toth, A. Anal. Chem. 1992, 64 1998 2000. (103) Newton, M. R.; Morey, K. A.; Zhang, Y.; Snow, R. J.; Diwekar, M.; Shi, J.; White, H. S. Nano Lett. 2004, 4 875 880. (104) Debiemme Chouvy, C. Electrochem. Solid State Lett. 2007, 10 E24 E26. (105) West, R.; Josowicz, M.; Janata, J.; Minet, I.; Hevesi, L. J. Electrochem. Soc. 2009, 156 F55 F59. (106) Baur, J. E.; Kristensen, E. W.; May, L. J.; Wiedemann, D. J.; Wightman, R. M. Anal. Chem. 1988, 60 1268 1272. (107) Fletcher, B. L.; Fern, J. T.; Rhodes, K.; McKnight, T. E.; Fowlkes, J. D.; Retterer, S. T.; Keffer, D. J.; Simpson, M. L.; Doktycz, M. J. J. Appl. Phys. 2009, 105 124312/1 124312/6. (108) John, R.; Wallace, G. G. J. Electroanal Chem. Interfacial Electrochem. 1991, 306 157 167. (109) Beck, F.; Braun, P.; Oberst, M. Ber. Bunsen Ges. Phys. Chem. 1987, 91 967 9 74. (110) Andrieux, C. P.; Audebert, P.; Hapiot, P.; Saveant, J. M. J. Am. Chem. Soc. 1990, 112 2439 2440. (111) Bose, C. S. C.; Basak, S. ; Rajeshwar, K. J. Phys. Chem. 1992, 96 9899 9906. (112) Zotti, G.; Martina, S.; Wegner, G.; Schlueter, A. D. Adv. Mater. 1992, 4 798 801. (113) Genies, E. M.; Bidan, G.; Diaz, A. F. J. Electroanal. Chem. Interfacial Electrochem. 1983, 149 101 103. (114) Diaz, A. F.; Cas tillo, J. I. J. Chem. Soc., Chem. Commun. 1980, 9 397 398. (115) Christensen, P. A.; Hamnett, A. Electrochim. Acta 1991, 36 1263 1286. (116) Atanasoska, L.; Naoi, K.; Smyrl, W. H. Chem. Mater. 1992, 4 988 994. (117) Genoud, F.; Guglielmi, M.; Nechtschein, M.; Genies, E.; Salmon, M., Phys. Rev. Lett. 1985, 55 118 121. (118) Montemayor, M. C.; Vazquez, L.; Fatas, E. J. Appl. Phys. 1994, 75 1849 1851. (119) Marandi, M.; Kallip, S.; Sammelselg, V.; Tamm, J. Electrochem. Commun. 2010, 12 854 858.
194 (120) Holdcroft, S.; Funt, B. L. J. Electroan al. Chem. Interfacial Electrochem. 1988, 240 89 103. (121) Cheng, Q.; Brajter Toth, A. Anal. Chem. 1995, 67 2767 2775. (122) Doblhofer, K.; Figura, J.; Fuhrhop, J. H. Langmuir 1992, 8 1811 1816. (123) Zoski, C. G. In Modern Techniques in Electroanalysis ; Vanysek, P., Ed. ; John Wiley & Sons, Inc.: New York, 1996 ; pp 258 261. (124) Bravo, R.; Hsueh, C.; Brajter Toth, A.; Jaramillo, A. Analyst 1998, 123 1625 1630. (125) Gonon, F. G.; Fombarlet, C. M.; Buda, M. J.; Pujol, J. F. Anal. Chem. 1981, 53 1386 1389. (126) Cooper, S. E.; Venton, B. J. Anal. Bioanal. Chem. 2009, 394 329 336. (127) Forryan, C. L.; Lawrence, N. S.; Rees, N. V.; Compton, R. G. J. Electroanal. Chem. 2004, 561 53 65. (128) Andrieux, C. P.; Garreau, D.; Hapiot, P.; Saveant, J. M. J. Electroanal. Chem. Interfacial Electrochem. 1988, 2 48 447 450. (129) Hsueh, C.; Bravo, R.; Jaramillo, A. J.; Brajter Toth, A. Anal. Chim. Acta 1997, 349 67 76. (130) Cahill, P. S.; Walker, Q. D.; Finnegan, J. M.; Mickelson, G. E.; Travis, E. R.; Wightman, R. M. Anal. Chem. 1996, 68 3180 3186. (131) El Nour, K. A.; Brajte r Toth, A. Electroanalysis 2000, 12 805 810. (132) Strand, A. M.; Venton, B. J. Anal. Chem. 2008, 80 3708 3715. (133) Liu, X.; Barizuddin, S.; Shin, W.; Mathai, C. J.; Gangopadhyay, S.; Gillis, K. D. Anal. Chem. 2011, 83 2445 2451. (134) Keum, S. R.; Gregory, D. H.; Brui ce, T. C. J. Am. Chem. Soc. 1990, 112 2711 2715. (135) Hsueh, C. C.; Brajter Toth, A. Anal. Chim. Acta 1996, 321 209 2 14. (136) Hsueh, C. C. PhD. Dissertation, University of Florida, 1996 (137) Van Benschoten, J. J.; Lewis, J. Y.; Heineman, W. R.; Roston, D. A.; Kissinge r, P. T. J. Chem. Educ. 1983, 60 772 776. (138) Freund, M. S.; Brajter Toth, A. J. Phys. Chem. 1992, 96 9400 9406. (139) Rice, R. J.; McCreery, R. L. Anal. Chem. 1989, 61 1637 1641.
195 (140) Li, W.; Tan, C.; Lowe, M. A.; Abruna, H. D.; Ralph, D. C. ACS Nano 2011, 5 2264 22 70. (141) Skoog, D .A.; Holler, F. J.; Crouch, S. R. Principles of Instrumental Analysis 6th ed.; Brooks/Cole: California, 2007 (142) Whitson, P. E.; VandenBorn, H. W.; Evans, D. H. Anal. Chem. 1973, 45 1298 1 306. (143) Nicholson, R. S. Anal. Chem. 1965, 37 1351 135 5. (144) Laviron, E. J. Electroanal. Chem. Interfacial Electrochem. 1979, 101 19 28. (145) Bond, A. M.; Oldham, K. B.; Snook, G. A. Anal. Chem. 2000, 72 3492 3496. (146) Bowling, R.; McCreery, R. L. Anal. Chem. 1988, 60 605 608. (147) Anson, F. C.; Osteryoung, R. A. J. Chem. Educ 1983, 60 293 296. (148) Harraz, F. A. J. Electrochem. Soc. 2006, 153 C349 C356. (149) Bravo, R. PhD. Dissertation, University of Florida, 1999 (150) Bin, X.; Sargent, E. H.; Kelley, S. O. Anal. Chem. 2010, 82 5928 5931. (151) Statsoft, Statistical significance ( p value), http://www.statsoft.com/textbook/statistics glossary/s/?button=0#Statistical_Significance_(p level) (152) Bott, A. W.; Heineman, W. R. Curr. Se p. 2004, 20 121 126. (153) Kozminski, K. D.; Gutman, D. A.; Davila, V.; Sulzer, D.; Ewing, A. G. Anal. Chem. 1998, 70 3123 3130. (154) Kathiwala, M.; Abou El Nour, K. M.; Cohen Shohet, R.; Brajter Toth, A. Analyst 2010, 135 296 301. (155) Ranganathan, S.; Kuo, T. C.; McC reery, R. L. Anal. Chem. 1999, 71 3574 3580. (156) Fagan, D. T.; Hu, I. F.; Kuwana, T. Anal. Chem. 1985, 57 2759 2763. (157) Runnels, P. L.; Joseph, J. D.; Logman, M. J.; Wightman, R. M. Anal. Chem. 1999, 71 2782 2789. (158) Laffont, L.; Monthioux, M.; Serin, V.; Mathur, R. B.; Guimon, C.; Guimon, M. F. Carbon 2004, 42 2485 2494. (159) Robinson, D. L.; Venton, B. J.; Heien, M. L. A. V.; Wightman, R. M. Clin. Chem. 2003, 49 1763 1773. (160) Venton, B. J.; Wightman, R. M. Synapse 2006, 61 37 39.
196 (161) Takmakov, P.; Zachek, M. K.; Keithley R. B.; Walsh, P. L.; Donley, C.; McCarty, G. S.; Wightman, R. M. Anal. Chem. 2010, 82 2020 2028. (162) De Carvalho, R. M.; De Oliveira Neto, G.; Kubota, L. T. Electroanalysis 2001, 13 131 136. (163) Huffman, M. L.; Venton, B. J. Electroanalysis 2008, 20 2422 2428 (164) Goss, C. A.; Brumfield, J. C.; Irene, E. A.; Murray, R. W. Anal. Chem. 1993, 65 1378 1389. (165) Kim, C.; Park, S. H.; Cho, J. I.; Lee, D. Y.; Park, T. J.; Lee, W. J.; Yang, K. S. J. Raman Spectrosc. 2004, 35 928 933. (166) Cheng, Q.; Brajter Toth, A. Anal. Chem. 1996, 68 4180 4185. (167) Bard, A. J. Encyclopedia of Electrochemistry of the Elements Marcel Dekker: New York, 1928 9 Part A, pp 236. (168) O'Brien, J. C.; Shumaker Parry, J.; Engstrom, R. C. Anal. Chem. 1998, 7 1307 1311. (169) O'Mullane, A. P.; Zhang, J.; Brajter To th, A.; Bond, A. M. Anal. Chem. 2008, 80 4614 4626. (170) Myland, J. C.; Oldham, K. B. Anal. Chem. 2000, 72 3972 3980. (171) Newman, J. J. Electrochem. Soc. 1966, 113 501 502. (172) Oldham, K. B.; Stevens, N. P. C. Anal. Chem. 2000, 72 3981 3988. (173) Wiggins Camacho, J. D.; Stevenson, K. J. J. Phys. Chem. C 2009, 113 19082 19090. (174) Peairs, M. J.; Ross, A. E.; Venton, B. J. Anal. Methods 2011, 3 2379 2386. (175) Hajati, S.; Tougaard, S. Anal. Bioanal. Chem. 2010, 396 2741 2755. (176) Paris, O.; Loidl, D.; Peterlik, H. Carbon 2002, 40 5 51 555. (177) Oya, N.; Johnson, D. J. Carbon 2001, 39 635 645. (178) Peng, M.; Li, D.; Shen, L.; Chen, Y.; Zheng, Q.; Wang, H. Langmuir 2006, 22 9368 9374. (179) Liu, J.; Yue, Z.; Fong, H. Small 2009, 5 536 542. (180) Ito, T.; Sun, L.; Crooks, R. M. Electrochem. Solid State Le tt. 2003, 6 C4 C7. (181) Palmisano, F.; Malitesta, C.; Centonze, D.; Zambonin, P. G. Anal. Chem. 1995, 67 2207 22 11.
197 (182) Stern, D. A.; Laguren Davidson, L.; Frank, D. G.; Gui, J. Y.; Lin, C. H.; Lu, F.; Salaita, G. N.; Walton, N.; Zapien, D. C.; Hubbard, A. T. J. Am. Chem. Soc. 1989, 111 877 891. (183) Penner, R. M.; Van Dyke, L. S.; Martin, C. R. J. Phys. Chem. 1988, 92 5274 5282. (184) Fletcher, B. L.; McKnight, T. E.; Fowlkes, J. D.; Allison, D. P.; Simpson, M. L.; Doktycz, M. J. Synth. Met. 2007, 157 282 289. (185) Iroh, J. O .; Chen, Y. Polym. Compos. 1999, 20 482 492. (186) Wood, G. A.; Iroh, J. O. J. Appl. Polym. Sci. 1996, 61 519 528. (187) Suarez, M. F.; Compton, R. G. J. Electroanal. Chem. 1999, 462 211 221. (188) Debiemme Chouvy, C. Electrochem. Commun. 2009, 11 298 301. (189) Wood, G. A.; Iroh, J. O. Synth. Met. 1996, 80 73 82. (190) Kathiwala, M.; Affum, A. O.; Perry, J.; Brajter Toth, A. The Analyst 2008, 133 810 816 (191) Muller, M. Free Radical Biol. Med. 2006, 41 1670 1677. (192) Muller, M.; Li, Z.; Maitz Peter, K. M. Burns 2009, 35 500 508. (193) Rennem eier, C.; Frambach, T.; Hennicke, F.; Dietl, J.; Staib, P. Infect. Immun. 2009, 77 4990 4997. (194) Sharp, D.; Gladstone, P.; Smith, R. B.; Forsythe, S.; Davis, J. Bioelectrochemistry 2010, 77 114 119. (195) Bukelman, O.; Amara, N.; Mashia ch, R.; Krief, P.; Meijl er, M. M.; Alfonta, L. Chem. Commun. 2009, 20 2836 2838. (196) Monson, C. F.; Majda, M. Anal. Chem. 2007, 79 9315 9320. (197) Guo, J.; Amemiya, S. Anal. Chem. 2005, 77 2147 2156. (198) Thompson, R. Q.; Barone, G. C., III; Halsall, H. B.; Heineman, W. R. Anal. Biochem. 1 991, 192 90 9 5. (199) Tang, H. T.; Lunte, C. E.; Halsall, H. B.; Heineman, W. R. Anal. Chim. Acta 1988, 214 187 195. (200) Rao, A. K.; Creager, S. E. Anal. Chim. Acta 2008, 628 190 197. (201) Baldrich, E.; Munoz Francesc, X.; Garcia Aljaro, C. Anal. Chem 2011, 83 2097 2 103. (202) Das, J.; Jo, K.; Lee, J. W.; Yang, H. Anal. Chem. 2007, 79 2790 2796.
198 (203) Kwon, S. J.; Yang, H.; Jo, K.; Kwak, J. Analyst 2008, 133 1599 1604. (204) Mie, Y.; Kowata, K.; Hirano, Y.; Niwa, O.; Mizutani, F. Anal. Sci. 2008, 24 577 582. (205) Astier, Y.; Bartlett, P. N. Bioelectrochemistry 2004, 64 53 59. (206) Limoges, B.; Marchal, D.; Mavre, F.; Saveant, J. M. J. Am. Chem. Soc. 2008, 130 7276 7285. (207) Limoges, B.; Marchal, D.; Mavre, F.; Saveant, J. M.; Schoellhorn, B. J. Am. Chem. Soc. 2008, 130 7259 7275. (208) BioAssay Syste ms, Data sheet: Quantichrom TM alkaline phosphatase assay kit, http://www.bioassaysys.com/file_dir/DALP.pdf (209) Wang, J.; Ibanez, A.; Prakash Chatrathi, M. J. Am. Chem. Soc. 2003, 125 8444 8445. (210) Fanj ul Bolado, P.; Gonzalez Garcia, M. B.; Costa Garcia, A. Anal. Bioanal. Chem. 2006, 385 1202 1208. (211) Kim, H. J.; Kwak, J. J. Electroanal. Chem. 2005, 577 243 248. (212) Sun, K.; Ramgir, N.; Bhansali, S. Sens. Actuators, B 2008, B133 533 537. (213) Matsue, T.; Fujihira M.; Osa, T. Anal. Chem. 1981, 53 722 723. (214) Anik, U.; Cubukcu, M.; Cevik, S.; Timur, S. Electroanalysis 2010, 22 1519 1523. (215) Calvo Marzal, P.; Rosatto, S. S.; Granjeiro, P. A.; Aoyama, H.; Kubota, L. T. Anal. Chim. Acta 2001, 441 207 214. (216) Yamauchi, Y.; I do, M.; Ohta, M.; Maeda, H. Chem. Pharm. Bull. 2004, 52 552 555. (217) Tang, J.; Tang, D.; Su, B.; Huang, J.; Qiu, B.; Chen, G. Biosens. Bioelectron. 2011, 26 3219 3226. (218) Das, J.; Aziz, M. A.; Yang, H. J. Am. Chem. Soc. 2006, 128 16022 16023. (219) Casella, I. G.; C ontursi, M. J. Electrochem. Soc. 2007, 154 D697 D702. (220) Amatore, C.; Capobianco, G.; Farnia, G.; Sandona, G.; Saveant, J. M.; Severin, M. G.; Vianello, E. J. Am. Chem. Soc. 1985, 107 1815 1824. (221) Wang, Z.; Li, X.; Wu, Y.; Tang, Y.; Ma, S. J. Electroanal. Che m. 1999, 464 181 186. (222) Snead, W. K.; Remick, A. E. J. Am. Chem. Soc. 1957, 79 6121 6127. (223) Pelekourtsa, A.; Missaelidis, N.; Jannakoudakis, D. Chim. Chron. 1997, 26 49 56.
199 (224) Salavagione, H. J.; Arias, J.; Garces, P.; Morallon, E.; Barbero, C.; Vazquez, J. L. J. Electroanal. Chem. 2004, 565 375 383. (225) Nicholson, R. S.; Shain, I. Anal. Chem. 1965, 37 190 195. (226) Heras, A. M.; Avila, J. L.; Ruiz, J. J.; Garcia Blanco, F, Electrochim. Acta 1984, 29 541 545. (227) Steenken, S.; Neta, P. J. Phys. Chem. 1982, 86 3661 3667. (228) Lind, J.; Shen, X.; Eriksen, T. E.; Merenyi, G. J. Am. Chem. Soc. 1990, 112 479 482. (229) Stutts, K. J.; Scortichini, C. L.; Repucci, C. M. J. Org. Chem. 1989, 54 3740 3744. (230) Della Ciana, L.; Bernacca, G.; Bordin, F.; Fenu, S.; Garetto, F. J. Electroanal. Che m. 1995, 382 129 135. (231) Herman, H. B.; Bard, A. J. J. Phys. Chem. 1966, 70 396 404. (232) Heras, A. M.; Munoz, E.; Avila, J. L.; Camacho, L.; Cruz, J. L. J. Electroanal. Chem. Interfacial Electrochem. 1988, 243 293 307. (233) Farnia, G.; Roque Da Silva, A.; Vianello, E. J. Electroanal. Chem. Interfacial Electrochem. 1974, 57 191 202. (234) Nicholson, R. S.; Wilson, J. M.; Olmstead, M. L. Anal. Chem. 1966, 38 542 545. (235) Brown, K. C.; Corbett, J. F. J. Chem. Soc., Perkin Trans. 2 1979, 308 311. (236) Farrell, S.; Ronkainen Matsuno, N. J.; Halsall, H. B.; Heineman, W. R. Anal. Bioanal. Chem. 2004, 379 358 367. (237) McComb, R. B.; Bowers, G. N., Jr. Clin Chem 1972, 18 97 104. (238) Besman, M.; Coleman, J. E. J. Biol. Chem. 1985, 260 11190 1193. (239) Thompson, R. Q.; Porter, M.; Stuver, C.; Halsall H. B.; Heineman, W. R.; Buckley, E.; Smyth, M. R. Anal. Chim. Acta 1993, 271 223 229. (240) Neta, P.; Maruthamuthu, P.; Carton, P. M.; Fessenden, R. W. J. Phys. Chem. 1978, 82 1875 1878. (241) El Nour, K. A.; Brajter Toth, A., Analyst 2003, 128 1056 1061. (242) Xu, Y.; Halsall, H. B.; Heineman, W. R. J. Pharm. Biomed. Anal. 1989, 7 1301 1311. (243) Belanger, G.; Lamarre, C.; Vijh, A. K. J. Electrochem. Soc. 1975, 122 46 50. (244) Pumera, M. Nanoscale Res. Lett. 2007, 2 87 93.
200 (245) Jacobs, C. B.; Peairs, M. J.; Venton, B. J. Anal. Chi m. Acta 2010, 662 105 127. (246) Alwarappan, S.; Prabhulkar, S.; Durygin, A.; Li, C. Z. J. Nanosci. Nanotechnol. 2009, 9 2991 2996. (247) Swamy, B. E. K.; Venton, B. J. Analyst 2007, 132 876 884. (248) Bertoncello, P.; Edgeworth, J. P.; Macpherson, J. V.; Unwin, P. R. J. Am. Chem. Soc. 2007, 129 10982 10983. (249) Dai, X.; Wildgoose, G. G.; Compton, R. G. Analyst 2006, 131 901 906. (250) Dumitrescu, I.; Edgeworth, J. P.; Unwin, P. R.; MacPherson, J. V. Adv. Mater. 2009, 21 3105 3109.
201 BIOGRAPHICAL SKETCH Abraham Boateng was born in Asante Mampong, in the Ashanti region of Ghana, in 1975. He passed his ordinary and advanced levels of the General Certificate of Examinations in 1994 and 1997, respectively. In 1998, he attended the Kwame Nkrumah University of Science and Technology, w here he received a B.S. degree in Chemistry (First Class Honors) in 2002. After graduation, he worked as a regulatory officer at Food and Drugs Board for one year and as a chemistry teacher for six months, in Ghana. In 2004, he received a Norwegian governm ent scholarship QUOTA Program for a two year M.S. degree in materials technology at the Norwegian University of Science and Technology (NTNU) under the direction of Prof. Kjell Wiik. At NTNU, he developed barium sulfate alumina silica composites, as corros ion resistant refractory castables. Because of his passion for solving problems, he began further graduate studies in analytical chemistry at the University of Florida (UF) under the guidance of Prof. Anna Brajter Toth in 2006. As a research assistant at U F, he identified and characterized graphitic nanomateria ls for electroanalysis; demonstr ated cont rolled electrodeposition of porous, ultrathin membran es for nanomolar detection, detection of p nitrophenol via in situ generation of p aminophenol and fabrication of nanostructured carbon nanotube film ultramicroelectrode (3 m dia.) Upon completion of his graduate studies, he will join Eastman Chemical Company