Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Fabrication and modelling of an all-printed PEDOT:PSS supercapacitor on a commercial paper Yoo, Dan Sik 2010-06-29

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


24-ubc_2010_fall_yoo_dansik.pdf [ 9.37MB ]
JSON: 24-1.0065066.json
JSON-LD: 24-1.0065066-ld.json
RDF/XML (Pretty): 24-1.0065066-rdf.xml
RDF/JSON: 24-1.0065066-rdf.json
Turtle: 24-1.0065066-turtle.txt
N-Triples: 24-1.0065066-rdf-ntriples.txt
Original Record: 24-1.0065066-source.json
Full Text

Full Text

Fabrication and modelling of an all-printedPEDOT:PSS supercapacitor on a commercial paperbyDan Sik YooB.A.Sc., Sungkyunkwan University, 2000A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Electrical and Computer Engineering)The University Of British Columbia(Vancouver)June 2010c￿Dan Sik Yoo, 2010AbstractSupercapacitorsdemonstratesubstantialimprovementinchargestoragecapabilitycompared to the conventional capacitors. With the emergence of printed electron-ics such as RFID tags, smart cards, electronic paper, and wearable electronics,printed energy storage devices are desirable. Therefore, a flexible and printablesupercapacitor with a PEDOT:PSS electrode is fabricated with inkjet micropat-terning technology. Electroanalytical measurement techniques are employed tocharacterize the performance of the printed supercapacitor. It has been foundthat addition of the surfactant (Triton X-100) increases the porosity of the PE-DOT:PSS electrode. A volume capacitance of 9.36 F/cm3(adding surfactant) and9.09 F/cm3(without adding surfactant) are measured with cyclic voltammetry.The two devices have different capacitor charging times, e.g., 50.46 s for elec-trode added surfactant and 112.9 s for the electrode without adding surfactant.In order to investigate the rate limiting factors of capacitor charging, electro-chemical impedance measurements and equivalent circuit modelling is utilized.Instead of using a constant phase element (CPE), a multiple time constant modeliiis proposed in order to explain the physical origin of the distributed time constantbehaviour. Thickness variation of the PEDOT:PSS electrode is assumed as a pri-mary reason for the distributed time constants and thus actual thickness variationis incorporated in the modelling. Data fitting with the measured impedance areconsistent with this assumption. However. it also has been found that there aremore factors distributing capacitances than just variations in thickness. A log-normal distribution function (LNDF) is utilized in order to further investigate therelationship between the distributed capacitance and the capacitor charging. It isfound that the capacitance distribution likely influence the charging.Previous experimental results demonstrate that the distributed capacitance isthe physical cause of the distributed time constant behaviour in electrochemicalimpedance measurement. However, this is the first analytical report proving therelationship between the distributed time constant behaviour and a thickness de-pendent capacitance distribution.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Supercapacitor fundamentals . . . . . . . . . . . . . . . . . . . . 31.2 Commercial applications . . . . . . . . . . . . . . . . . . . . . . 71.3 Thesis overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Supercapacitorelectrodefabricationusinginkjetprintingtechnology 122.1 Motivation of using inkjet printing technology . . . . . . . . . . . 132.2 Characterization of the printing parameters . . . . . . . . . . . . 162.3 Equipment setup and electrode printing . . . . . . . . . . . . . . . 18iv3 Experiments and results . . . . . . . . . . . . . . . . . . . . . . . . 233.1 Porosity measurements with AFM, FESEM, and BET technique . 243.2 Ionic conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . 273.3 Cyclic voltammetry . . . . . . . . . . . . . . . . . . . . . . . . . 303.4 Electrochemical impedance spectroscopy . . . . . . . . . . . . . 344 Multiple time constant model . . . . . . . . . . . . . . . . . . . . . . 424.1 The origin of the CPE response . . . . . . . . . . . . . . . . . . 434.2 The origin of capacitance distribution of the PEDOT:PSS electrode 464.3 Multiple time constant modelling . . . . . . . . . . . . . . . . . . 494.4 Data fitting of the multiple time constant model . . . . . . . . . . 565 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75A Design of the printing nozzle housing . . . . . . . . . . . . . . . . . 76B Nozzle maintenance tips . . . . . . . . . . . . . . . . . . . . . . . . 79vList of Tables3.1 Volume capacitance decreases as scan rate increases, WT and NTrefer to electrodes with Triton X-100 and without Triton X-100respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.2 Volume capacitances calculated from the EIS data for both WTand NT at different scan rates. . . . . . . . . . . . . . . . . . . . 404.1 TheaveragethicknessofthetwoPEDOT:PSSelectrodes(WTandNT) and bare paper roughness along five different lines. . . . . . . 474.2 Experiment and modelling data comparison. C, σi, Rs, and Rfrefer to the capacitance, ionic conductivity, solution resistance,and faradaic resistance respectively . . . . . . . . . . . . . . . . . 62viList of Figures1.1 The energy storage mechanism of a regular capacitor. . . . . . . . 31.2 Aequivalentcircuitofthesupercapacitorelectrode. Re, C,Ri, andRs represent electrical resistance, capacitance, ionic resistance,and solution resistance respectively. . . . . . . . . . . . . . . . . 51.3 TheRagoneplotforvariousenergystoragedevices. Adaptedfrom[10]-[11]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.1 Dispensing frequency and continuous line patterning a) 333 Hz b)833 Hz c) 1666 Hz d) 2333 Hz e) 3000 Hz. . . . . . . . . . . . . 182.2 a) The inkjet printer stage setup b) the paper attachment on thesubstrate heater c) the substrate heater and the silicon gasket. . . . 192.3 PEDOT:PSS deposition on commercially available letter size pa-pers which are showing different particle agglomeration and pen-etration. #1-1, #2-1, and #3-1 are from Husky xerocopy (Canada),Hansol paper (Korea), and Korea paper (Korea), respectively. . . . 212.4 Electrode pattern on a commercial paper (whitestone, Neenah pa-per company, USA). . . . . . . . . . . . . . . . . . . . . . . . . . 22vii3.1 AFM images for both WT and NT samples. . . . . . . . . . . . . 263.2 Cross section images of the electrode from FESEM. . . . . . . . . 273.3 Ionic conductivity test apparatus and the equivalent circuit of theionic conductivity measurement. WE, RE, and CE refer to work-ing electrode, reference electrode, and counter electrode respec-tively. The sample is placed at the center, clamped between twocircular cups. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283.4 The cyclic voltammetry of the PEDOT:PSS electrodes at 10 mV/s(top) and 1 mV/s (bottom) scan rates. . . . . . . . . . . . . . . . . 323.5 Thecyclic voltammetryof thePEDOT:PSSelectrodes at0.1mV/sscan rate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.6 a) Electrode with aluminum foil substrate b) coil shape platinumwire and stainless steel sheet piece. . . . . . . . . . . . . . . . . . 363.7 Abnormal impedance response without stainless steel wire con-nection to RE at high frequency. . . . . . . . . . . . . . . . . . . 373.8 The EIS three electrode cell setup. . . . . . . . . . . . . . . . . . 383.9 Volumetric capacitance vs. frequency plot from the EIS and itscorresponding scan rate in the CV. . . . . . . . . . . . . . . . . . 394.1 Microscope image of the PEDOT:PSS electrode surface. . . . . . 454.2 a) The deposition direction (from top lines to bottom lines) of thePEDOT:PSS electrode b) the thickness scan direction of the PE-DOT:PSS electrode. . . . . . . . . . . . . . . . . . . . . . . . . . 47viii4.3 Thickness profile for both WT and NT electrodes at #3 (center)position scan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484.4 The total thickness deviation of the PEDOT:PSS electrode whichis measured from scan #1 to #5. . . . . . . . . . . . . . . . . . . . 504.5 The simple drawing of the multiple thickness electrode (top) andequivalent circuit for one section of the electrode (bottom). . . . . 514.6 The three electrode EIS measurement setup for measuring solu-tion resistance. A stainless steel sheet (SSS) is used as a workingelectrode. WE, RE, and CE refer to the working electrode, refer-ence electrode, and counter electrode respectively. . . . . . . . . . 524.7 The Bode plot of the NT electrode. . . . . . . . . . . . . . . . . . 574.8 The Nyquist plot of the NT electrode. . . . . . . . . . . . . . . . 584.9 The Bode plot of the WT electrode. . . . . . . . . . . . . . . . . . 594.10 The Nyquist plot of the WT electrode. . . . . . . . . . . . . . . . 604.11 Phase angle response with different σtvalues. . . . . . . . . . . . 61A.1 Design and assembly of the inkjet printing nozzle housing. . . . . 77ixAcknowledgmentsI would like to give my special thanks to my supervisor, Professor John D. Mad-den, forgivingmeanopportunitytojointhelabandsupportingmethroughoutmystudies. He always inspired me with his great knowledge and experience and alsogave me clear guidance and advice. I also would like to thank Professor KonradWalus, who helped me to finalize the experiments and the project. The progress ofthisresearchwould havenotbeenpossiblewithouthis adviceandencouragement.I thank Professor Andre Ivanov for giving me a chance to join UBC and hav-ing a chance to explore academia. Many thanks to Dr. Arash Takshi for helpingme out when I first started the supercapacitor project. I would have had manydifficulties in a new environment without your help. Many thanks to all the labmembers for helpful comments, discussions, and cooperation. I greatly benefittedfrom their theoretical insights and experiences.I deeply appreciate my wife (Ji Hey Jung) for her sacrifice to support methroughout my Master’s. I also thank my daughter (Jean Ruhiyyih Yoo) for grow-xing without any problems and giving me joy and energy in life.xiChapter 1IntroductionWithin the last decade, we have seen incredible innovations in portable electronicdevices such as mobile phones, smartphones, MP3 players, digital cameras, etc.However, recent trends in portable devices also brought an issue to the recharge-able lithium-ion battery. By integrating many features in a portable device, thebattery also has to cope with high power consumption. This new challenge mayshorten life time of the battery or it may also cause performance failure. One ofthe main obstacles in battery design is maintaining voltage levels of the battery.However, due to the high current that is often required, it has become difficult tosustain a stable voltage level. When the voltage drops below an operational range,then the internal passivation layer of the lithium battery starts to break down andinduces considerable leakage current inside the battery. These kinds of constantleakage currents shorten the battery life time.1Supercapacitors can be a promising solution for increased power demands inconsumer electronics devices. By placing supercapacitors in parallel with the bat-teries, the dramatic voltage drops can be prevented and it can extend life time ofthe battery. Whenever a device needs high power, the supercapacitors will fill theincreased demand instead of a battery. The supercapacitor will then be rechargedin a short time[1].Supercapacitors have several names such as electrochemical double layer ca-pacitor (EDLC), ultracapacitor, and pseudocapacitor. They have remarkably highspecific capacitance (capacitance per unit mass) and energy density compared toconventional capacitors. If the supercapacitors can provide similar energy densityto batteries, this would provide significant value to many devices. However, dueto the lower energy densities but high power supply capabilities, supercapacitorsare getting attention primarily as a complementary energy source to batteries[2].The unique features (low cost, thin, lightweight, and also flexible) of theprinted energy storage devices are getting great attention with the emergence ofprinted electronics such as RFID tags, smart cards, electronic paper, and wearableelectronics. In this thesis, a rectangular (12 mm x 12 mm) PEDOT:PSS super-capacitor is fabricated using inkjet micropatterning and characterized by usingelectroanalytical methods21.1 Supercapacitor fundamentals!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"!""#" !""#" !""#"$%&'&()*%("+,)&*%,'"#"!"-&),'".',)&"-&),'".',)&"/"Figure 1.1: The energy storage mechanism of a regular capacitor.The basics of the capacitor are briefly reviewed before we discuss the super-capacitor fundamentals. The conventional capacitor consists of two metal plateswhich are sandwiching a dielectric material as shown in Figure 1.1. When a volt-age is applied between two metal plates, charges separate from one side of themetal plate and accumulate at the other side of the metal plate. The capacitanceof a simple capacitor is proportional to the area of the contact plates and the per-3mittivity of the dielectric. The capacitance of a parallel plate capacitor isC =εr·ε0·Ad, (1.1)where C, εr, ε0, A, and d are the capacitance, relative permittivity, electric con-stant, the area the metal plate, and the distance between two plates.Supercapacitors are a unique kind of capacitor. Whereas a common capac-itor stores charges at only the surface of the metal plate, supercapacitors storecharges through the whole volume of the electrode due to the porous nature of theelectrode materials. The Figure 1.2 describes a equivalent circuit of one superca-pacitor electrode. When voltage is applied to the electrode, ions penetrate insidethe electrode and charge the electrode. When the electrode is positively charged,the electrode repels positive ions and attracts negative ions, whereas a negativelycharged electrode behaves in opposite way and charges the entire volume of theelectrode.Supercapacitors are normally divided into two groups according to their en-ergystoragemechanism. Thefirstoneonlyuseselectrostaticchargeaccumulationbetween the electrode and the electrolyte rather than using electrochemical reac-tions. Carbon materials belong to this group and this is the reason why thesesupercapacitors are commonly classed as electrochemical double layer capacitors(EDLC). The second mechanism utilizes charge transfer reactions which are ob-4!"#$%Figure 1.2: A equivalent circuit of the supercapacitor electrode. Re, C, Ri,and Rs represent electrical resistance, capacitance, ionic resistance,and solution resistance respectively.served in electro-active materials (metal oxides or conducting polymers). Thecapacitance values for these materials are reported to be much higher than EDLCmaterials.Supercapacitors demonstrate substantial improvement in charge storage capa-5bility compared to the conventional capacitors because the distance between theelectrode and electrolyte ions are on the order of nanometers (∼ 1 nm) in theelectrical double layer [3]-[4]. The amount of charge stored at the surface of theelectrode is mainly determined by the accessible surface area. For porous mate-rials which are used in supercapacitor electrodes, the charge/discharge processestake place throughout the entire volume and not only at the surface of the material.Most of the energy stored in carbon materials is in the electrostatic chargeseparation rather than charge transfer reaction[9]. Carbon is one of the commonlyused materials for supercapacitors because of the high surface area, low materialcost, processability, and chemical/temperature stability. Additionally, the activa-tion procedures that control carbon porosity are well understood and developed.Activated carbon devices have demonstrated a wide range specific capacitances(100 - 400 F/g) depending on the surface area (m2/g), pore size distribution, andsurface double layer capacitance (µF/cm2) [5].Thereisanotherclassofelectrochemicalenergystoragesupercapacitorswhichutilize pseudo-capacitance. Unlike the EDLC, pseudo-capacitance utilizes thecharge transfer reactions between the electrode and the electrolyte associated withelectrosorption and surface redox processes with high surface area of the elec-trode. Metaloxidematerials(e.g. RuO2,MnO2)orconductingpolymers(polypyr-role, polyaniline, polythiophene derivatives) store charges with such faradaic pro-cesses but the electrical behaviour is like that of a capacitor. Whereas the capaci-6tance of ideal EDLC is constant, the pseudo-capacitance is usually dependent onpotential. However, almost constant capacitance response over the full voltagewindow (∼1.4 V) has been reported for RuO2. In addition to the constant capac-itance response, it also demonstrates high specific capacitance (650 F/g)[5], greatcyclability and high conductivity [6]. However, due to the high cost of the RuO2,it has mostly been used in military applications.Conducting polymers are less stable than RuO2. However due to their sig-nificantly lower cost, conducting polymers are getting attention as a substitutefor supercapacitor material. Conducting polymers can be charged positively ornegatively with inserted ions and pseudo capacitance arises with the faradaiccharge transfer reaction[7]. K. R. Prasad and N. Munichandraiah have demon-strated specific capacitance of 1300 F/g with polyaniline (PANI) on a stainlesssteel substrate[8].1.2 Commercial applicationsSince their first commercialization as backup power supply for computer memoryin 1978 (NEC company, Japan), supercapacitors have been used in applicationswhere power density is a higher priority than energy density. As shown in Fig-ure 1.3, supercapacitors are filling the gap between batteries and conventionalcapacitors. The energy density of the supercapacitor (∼10Wh/kg) is fairly highcompared to a conventional capacitor (∼0.05Wh/kg). However it is only about7!"!#$!"#$ #$ #!$ #!!$#!!!$!"!#$#!$#!!$#!%$#!&$#!'$#!($#!)$*+,-./$0,+123/$45678.9$:;<,-$0,+123/$4578.9$=,>0?>@20$$$$A>B,-/$$$C2D0$A>B,-/$$$=2362EF$$$A>B,-/$GE,H$D,HH1$D>I>@23;-1$JEI,-@>I>@23;-1$Figure 1.3: The Ragone plot for various energy storage devices. Adaptedfrom [10]-[11].one-tenth that of a battery. The specific power of supercapacitors is much higherthan batteries. As a result, supercapacitors can deliver energy in a short time butthey cannot provide a lot of energy, whereas batteries can provide a considerablequantities of energy but powering of a high power device is not an optimal ap-plication for battery. Supercapacitors can supply large amounts of charge veryquickly and can be recharged in short times. Supercapacitors demonstrate greatcyclic performance. Battery can only be charged/discharged approximately one8thousand times. However, supercapacitors generally show only little degradationafter thousands of cycles. Supercapacitors also demonstrate very low internal re-sistance, low heating levels, and good safety standards. Batteries supply steadyvoltage in the usable voltage range, whereas the voltage of a supercapacitor dropslinearly from the beginning of discharge and this hinders supercapacitors fromproviding full charge. About 75 % of the total energy of supercapacitors can beutilized by using a DC-DC converter [10].The hybrid electrical vehicle (HEV) power source application has substantialpotential as a new market for supercapacitors. By utilizing the complementaryproperties of batteries and supercapacitors, it can greatly enhance the HEV per-formance. Supercapacitors have great advantages of supplying bursts of powerand charging very quickly. For instance, supercapacitors can deliver the powerwhen vehicles need high power (acceleration or hill climbing). They also can beutilized to capture the regenerative braking energy[12].As they first found their application in backup power for computer memory,backup power sources for the consumer electronic products are one of the biggestmarkets for supercapacitors. For instance, we can find supercapacitor applicationin satellite TV (TV-channel setting, clock time), car audio systems (radio stationmemory), coffee machines (protect programmed functions), and programmablepocket calculator among others[10].91.3 Thesis overviewIn chapter 2, inkjet micropatterning technology is introduced and advantages andlimitations are discussed in various applications. All the parameters (voltage,pulse length and frequency) determining droplet formation and deposition speedare defined. The detailed fabrication procedures of the PEDOT:PSS electrode arealso described in the chapter.In chapter 3, electroanalytical measurement techniques, e.g., ionic conductiv-ity measurement, cyclic voltammetry (CV), and electrochemical impedance spec-troscopy (EIS) are employed to characterize the PEDOT:PSS electrode. The ionicconductivity measurements suggest that the surfactant Triton - X100, increasesthe porosity of the PEDOT:PSS electrode. In the CV measurements, varying scanrates (10 mV/s, 1 mV/s, and 0.1 mV/s) are applied to investigate the relationshipbetween increased porosity and the capacitance associated with the addition ofTriton X-100. EIS is used to measure the impedance of the system over a range offrequencies in order to characterize energy storage behaviour of the PEDOT:PSS.In chapter 4, a multiple time-constant model is proposed in order to explainthe physical origin of the constant phase element (CPE) behaviour in the EISmeasurements. Although it does not help to identify the origin of the response,the CPE is widely used in electrochemical impedance modelling in order to repre-sent time constant distributions in an equivalent circuit. The thickness variation of10the printed PEDOT:PSS was found to be the primary physical origin of the capac-itance distribution. However, there could be other possibilities that may lead toa distributed capacitance. All the modelling parameters (ionic conductivity, solu-tion conductivity, and capacitance) are also compared with the experimental data.11Chapter 2Supercapacitor electrode fabricationusing inkjet printing technologyThis chapter details the inkjet micropatterning technology that is utilized to fab-ricate the device studied in this thesis. In Section 2.1, we briefly introduce thecharacteristics of inkjet printing as a fabrication technology and discuss the ad-vantages and limitations in various applications. In Section 2.2, recent researchreports that investigate the relationship between droplet formation and control pa-rameters are introduced. We also describe features of key parameters (voltage,pulse length, frequency) that affect droplet shape and speed of deposition. In Sec-tion 2.3, we describe the inkjet system setup and electrode printing procedures.122.1 Motivation of using inkjet printing technologyInkjet printing technology is most commonly used for document printing ontopaper. In addition, it also has widespread commercial applications in printingproduct information onto cans or bottles. However, in recent years, the micro pat-terning capability of inkjet printing technology coupled with conducting polymersor novel nanoparticles is also getting great attention[13]-[14].Drop-on-demand (DOD) printers eject ink droplets when and where they areneeded to create patterns on a substrate. This approach eliminates the complexityof drop charging and deflection hardware, as well as the inherent unreliability ofink recirculation systems required for continuous inkjet technology. Thermal andpiezoelectric printing technologies are competing with each other in the marketwith this mode. Thermal inkjet printers have occupied most of the low cost officeprinter market. However, this system only can work with highly tailored inks.If other inks are used in thermal printers their performance/life can be degradedradically due to the vaporization process, whereas the piezoelectric inkjet systemsdo not modify the ink properties and have the capability to dispense a wide rangeof solutions. Thus, piezoelectric inkjet printing technology has become the mainchoice for experimental DOD systems.Piezoelectric printers are categorized by the deformation methods (squeeze,bend, push and shear[15]-[16]) of the piezoelectric ceramic used in the device.13The squeeze mode is used for our experimental apparatus. The core of the deviceconsists of a glass capillary which is surrounded by a tubular piezoelectric actua-tor. The actuator is controlled by the electrostatic field and its squeezing actuationcreates a pressure pulse inside the glass capillary. A small amount of liquid isejected from the orifice due to the pressure pulse and it forms a uniform droplet.Uniquefabricationfeaturesofhighresolutioninkjetmicropatterningisattract-ing great attention[17]-[18]. Direct material deposition is a more environmentallyfriendly process because it is not based on subtractive lithography steps that wastesignificant portion of materials. Furthermore, less material and simple fabricationsteps yield a lower cost process. Organic and biological materials that are notcompatible with semiconductor processes can also be deposited. Manufacturingon flexible substrates is also possible since the process is low temperature. Inrecent years, there has been extensive research and promising results that utilizeinkjet printing technology in various applications.Inkjet microdispensing technology is adaptable to high density DNA arrayfabrication [19]. The capability of dispensing nano to picoliter volumes onto thespecific DNA test site is a significant advantage of utilizing inkjet technology inbiomedical arrays. Liquid delivery capability of a small volume with high speedfeatures of inkjet microdispensing technology has promising potential of combi-natorial chemistry [20]. Whereas, chemical compatibility with a wide range ofsolvents has to be solved to be used as an automated chemical synthesis equip-14ment.Cantilever arrays coated with a range of printed sensing material have beenreported[21]. A chemical gas sensor was fabricated by printing thin layers of dif-ferent polymers from dilute solutions onto cantilevers. It has been demonstratedthat the inkjet method is faster and more versatile than other coating methods suchas microcapillaries or pipettes. Furthermore, it is capable of coating large arraysand arbitrary structures.Printing energy storage devices is also one of the emerging research areasaiming to satisfy the growing demands for portable electronic devices that are lowcost, thin, lightweight, and potentially also flexible. Especially, with the emer-gence of printed electronics such as RFID tags, smart cards, electronic paper, andwearable electronics, it has become increasingly important to develop printed en-ergy storage devices[22].A micro-supercapacitor which is integrated with a MEMS based energy har-vesting device is fabricated with inkjet printing technology and described in thereference[23]. Activated carbon particles are mixed with PTFE polymer binderin ethylene glycol. Triton X-100 surfactant was added to the solution in order toincrease the wettability and the stability of the emulsion. The inks were depositedonto the interdigited gold fingers. The substrate is then heated at 140◦C in orderto get an homogeneous activated carbon deposition. The device was characterized15with electrochemical techniques in 1 M Et4NBF4and demonstrated a wide poten-tial range (2.5V) with a cell capacitance of 2.1mF/cm2.There are research reports describing electroactive polymer ink preparationsthat can be used with inkjet printing methods. X. Li et al. have demonstrated adispersed conducting polyaniline/silica ink preparation by using hydrolysis andcondensation of tetraethyl orthosilicate (TEOS)[13]. Polyaniline (PANI) is a con-ducting polymer which has received significant attention due to its good stabilityand interesting redox response [14]. P. Gajendran and R. Saraswathi [24] havereported PANI-CNT (Carbon Nanotube) composite inks. PANI-CNT compositesare dispersed in aqueous solution in spite of the insoluble nature of PANI in water.A 1:3 nitric acid - sulfuric acid treatment converted the hydrophobic property ofCNTs to hydrophilic due to the incorporation of acid functionalities and allowedthe PANI-CNT composites to be dispersed in aqueous solutions.With motivations of increasing demands for the printable energy storage de-vices, an all printed flexible PEDOT:PSS supercapacitor is fabricated and charac-terized as a prototype device in this thesis.2.2 Characterization of the printing parametersTherehasbeenconsiderableresearchinvestigatingtherelationshipbetweendropletformationandcontrolparameters. Dropletformationofsilvernanoparticles(around164 nm) is investigated by M. H. Tsai et al. [25]. It is generally accepted thatdroplet size is determined by the radius of the nozzle orifice. However, an orderof magnitude reduction in drop volume with same nozzle has been demonstratedby adjusting the driving voltage wave form [26]. In this thesis, however, due tothe relatively large size of the pattern (12 mm × 12 mm) and the ease and relia-bility of drop formation using PEDOT:PSS solution, we do not perform detailedinvestigation into these characteristics. Here, we focus on the adjustable rangeof printing parameters (voltage, pulse length, frequency) that result in the mostuniform electrode pattern.Higher voltage causes larger displacement of the piezoelectric actuator andthus results in higher pressure inside the nozzle. This results in increased dropletacceleration and it enables the nozzle to eject higher viscosity inks. The voltagerange of the Microdrop system is 30 - 255 V. However, we observe that a voltagerange of 50 - 80 V is sufficient for printing the PEDOT:PSS solution.The voltage pulse length can be adjusted between 8 and 200 µsec. The pulselength can change droplet shape and velocity. However, it is not as critical whencompared to voltage or frequency in order to get a good pattern. It is noticed thatoptimal droplets are achieved between 20 and 40 µsec from the experiments.The adjustable droplet frequency range is from 1 to 3000 Hz. The fastest dis-pensing frequency is dependent on the properties of the liquid. Characterizing17Figure 2.1: Dispensing frequency and continuous line patterning a) 333 Hzb) 833 Hz c) 1666 Hz d) 2333 Hz e) 3000 Hz.dispensing speed is important in order to minimize electrode fabrication time. Asshown in Figure 2.1, slower dispensing frequency gives better line patterning andfrom 2333 Hz, the printed lines are no longer uniform. For the PEDOT:PSS solu-tion, 1666 Hz dispensing frequency which corresponds to 30 µm droplet distanceand 50 mm/s stage speed is used to pattern the electrode.2.3 Equipment setup and electrode printingFigure 2.2 shows the experimental setup used to print the electrode. In addition tothe nozzle housing, we have designed a substrate heater (right bottom of Figure2.2). Water in the PEDOT:PSS solution causes the paper to warp during the depo-sitionandcuringprocess(100◦C10minutes). Thepaperisplacedonthesubstrate18!"#$"#%"#&'('%)*#+!,-./#&0$,/1!/.#2.!/.1#Figure 2.2: a) The inkjet printer stage setup b) the paper attachment on thesubstrate heater c) the substrate heater and the silicon gasket.heater (100◦C is maintained) and kapton tape is used to hold the position of bothpaper and substrate heater. Rapid water evaporation using this setup allows us tofabricate thick and uniform electrodes without warping commercial paper.The substrate heater is fabricated using three glass slides, nickel-chromiumalloy wire (80 % Nickel, 20 % Chromium), and thermally conductive epoxy ad-hesive resin. The glass slide is wrapped uniformly with the nickel-chromium wire19and an epoxy resin is applied to both sides. The glass slide is sandwiched withtwo additional glass slides and kapton tape is used to hold the assembly. Theepoxy was cured using a hot plate (80◦C, for an hour). A silicon gasket (operatingtemperature 204 - 260◦C) is placed under the heater to reduce heat transfer to theprinter stage. The temperature of the glass surface is maintained around 100◦C byapplying a 1.6 A current to the nickel-chromium wire.The macro function of the inkjet printing system is used to print the elec-trode pattern. The macro dispensing feature enables us to control the pattern startposition, size of the pattern, droplets distance, etc. Micro patterning using themicrodrop printing system can be done with a repetition accuracy of ± 1 µm(x and y axis) and ± 5 µm z-axis. The PEDOT:PSS conducting polymer is de-posited on both sides of the paper in order to complete the supercapacitor. Sixlayers are deposited to minimize electrical resistance. Considering the sensitivenature of the droplet, the droplet is checked using the strobe station after eachlayer is deposited. Four different commercial papers are tested as a supercapaci-tor substrate as shown in Figure 2.3-2.4. Due to the different chemical additives ofeach commercial letter size paper (from Husky xerocopy (Canada), Hansol paper(Korea), and Korea paper (Korea) for #1-1, #2-1, and #3-1), each sample showsdifferent patterning properties such as PEDOT:PSS particle agglomeration, pene-tration. We choose the paper (from Neenah paper, whitestone) that is electricallyinsulating between the two sides and has a uniform PEDOT:PSS pattern on eachside. Figure 2.4 shows the electrode patterned on the paper.20Figure 2.3: PEDOT:PSSdepositiononcommerciallyavailablelettersizepa-pers which are showing different particle agglomeration and penetra-tion. #1-1, #2-1, and #3-1 are from Husky xerocopy (Canada), Hansolpaper (Korea), and Korea paper (Korea), respectively.In this chapter, the advantages and limitations of the inkjet micropatterningtechnology are described in various applications. With the recent research reportsof droplet formation and control parameters, all the parameters (voltage, pulselength and frequency) determining droplet formation of our system are discussed.Electroanalytical experiments including cyclic voltammetry, ionic conductivity,21Figure 2.4: Electrode pattern on a commercial paper (whitestone, Neenahpaper company, USA).electrochemical impedance measurement will be performed with the printed PE-DOT:PSS electrode in the next chapter.22Chapter 3Experiments and resultsIn this chapter, the capacitance of the printed supercapacitor is measured and theorigin of the capacitance is explored. It is also suggested that porosity can beincreased by using the surfactant Triton X-100. In Section 3.1, we discuss ex-perimental results and limitations of using atomic force microscopy (AFM), fieldemission scanning electron microscope (FESEM), and the BET (Brunauer, Em-mett, and Teller) techniques to characterize the porosity of the PEDOT:PSS elec-trode. Ionic conductivity measurement of the PEDOT:PSS film using a custommade apparatus are introduced in Section 3.2. A decrease in the ionic resistanceof the sample with Triton X-100 suggests that the addition of Triton X-100 leadsto increased porosity. In Section 3.3, electrode porosity is further investigated us-ingthecapacitorcharge/dischargeresponse. Varyingscanrates(10mV/s, 1mV/s,and 0.1 mV/s) are applied to study the relationship between increased capacitanceand the addition of Triton X-100. The electrochemical impedance spectroscopy23(EIS) measurements are detailed in Section 3.4 as a means to understand the fac-tors that affect the charging rates. EIS response is compared to the result of cyclicvoltammetry (CV) experiments in order to relate the frequency response to thetime domain response. A trend of increased capacitance at low frequencies andlong times is found in the measurements.3.1 Porosity measurements with AFM, FESEM,and BET techniqueInitially, Triton X-100 (C14H22O(C2H4O)n), which has a hydrophilic polyethy-lene oxide group and a hydrophobic hydrocarbon lipophilic group, is added toincrease hydrophilicity of aqueous PEDOT:PSS solution in order to enhance theadhesion to the paper substrate. However, in addition to the increased adhesion,it was also found to increase the capacitance of the electrode at high scan rates asshown in Table 3.1.Table 3.1: Volume capacitance decreases as scan rate increases, WT and NTrefer to electrodes with Triton X-100 and without Triton X-100 respec-tively.Sample Scan rate(mV/s) Volume capacitance(F/cm3) Increase(%)WT/NT 10 3.01/1.56 93WT/NT 1 5.98/4.09 46WT/NT 0.1 9.36/9.09 3There are research reports that suggest the addition of Triton X-100 increases24theporosityofpolymerelectrodesandyieldsahigherspecificcapacitances[27][28].To date, there are no reports demonstrating that addition of Triton X-100 increasesthe porosity of the PEDOT:PSS electrode. In order to investigate the effect of Tri-ton X-100 on the PEDOT:PSS electrode, we add 0.3 wt% Triton X-100 with 94.7wt% PEDOT:PSS solution and 5 wt% Ethylene Glycol and refer to this sample asWT (with Triton X-100). For the NT (without Triton X-100) sample, we only use95 wt% PEDOT:PSS solution and 5 wt% Ethylene Glycol.The surface morphology of two WT and NT samples were been measuredwith atomic force microscopy (AFM) as shown in Figure 3.1. WT (160.752 nm)shows higher standard deviation than NT (44.699 nm), which suggests that theaddition of Triton X-100 increases surface roughness. It has been found that thisapproach has limitations. The surface roughness of the electrode could not pro-vide the whole porosity property of the PEDOT:PSS electrode even though wecan see the effect of the Triton X-100 on the surface morphology.Cross sectional images were taken in order to characterize the porosity of theelectrode. Figure 3.2 shows the cross section image of the PEDOT:PSS electrodemeasured by field emission scanning electron microscope (FESEM). We foundthat the images do not reveal the porosity directly because we need to see bothsides of same fractured spot in order to determine whether or not the pore shapestructure came from the fracture process or it is pristine pores.25Figure 3.1: AFM images for both WT and NT samples.The BET (Brunauer, Emmett, and Teller) measurement was also taken to char-acterize the surface area. However, it has been found that there is no measurablesurface area with the N2gas adsorption BET test[29]. It seems likely that thestructure of the PEDOT:PSS is sufficiently dense so that there is no significantinsertion of nitrogen, and that any increase in porosity is at the molecular scale.In the next subsection, we will perform ionic conductivity measurements to26Figure 3.2: Cross section images of the electrode from FESEM.demonstrate that addition of Triton X-100 increases ion transport rates in PE-DOT:PSS electrodes.3.2 Ionic conductivityIn the equivalent circuit of the electrode, an ion conductive pathway is representedby an ionic resistance (Ri) which is related to the porosity of the electrode. Forinstance, higher porosity electrodes will demonstrate lower resistances and moredense electrodes will show higher resistances. Thus the Riof PEDOT:PSS elec-trode is measured to investigate the porosity effect of Triton X-100.27A custom made ionic conductivity test apparatus is used as shown in Figure3.3. The distance between two reference electrodes is designed to be 5.2 mm byusing curved tips inside the glass apparatus and the diameter of the joint part is 10mm. The sample is sandwiched between two electrolyte reservoirs and the posi-tion is fixed by using a spring clamp.Figure 3.3: Ionic conductivity test apparatus and the equivalent circuit of theionic conductivity measurement. WE, RE, and CE refer to workingelectrode, reference electrode, and counter electrode respectively. Thesample is placed at the center, clamped between two circular cups.The equivalent circuit of ionic conductivity measurement is shown in Figure283.3. Rs, Ri, Re, and Cdare solution resistance of the electrolyte reservoirs, ionicresistance of the sample, electrical resistance of the sample, and internal doublelayer capacitance respectively. Ri, Re, and Cdof PEDOT:PSS electrode are con-nected in series with the paper resistance (Rp). Both ends of the electrode andpaper are facing the electrolyte (0.1 M TBAP propylene carbonate), which is rep-resented as Rs.At high frequency, Cdbecomes a short circuit and Riis much higher than Reso the impedance approaches the sum of Rs, Re, and Rp. At low frequency, theimpedance approaches the sum of Rs, Ri, and Rpbecause Cdacts as an open cir-cuit. A separate test is performed to measure Rsand Rpof the bare paper. Theionic resistanceof the electrodeis calculated bysubtracting Rsand Rpvalues fromthe total resistance at low frequency.It has been noticed that the resistance of the paper (Rp) decreases over timeand saturates to approximately 840Ωafter soaking in the electrolyte for six hours.As a result, all samples were soaked for more than six hours in electrolyte prior tothe measurement.Ionic resistance (Ri) has the relationship with the ionic conductivity (σi) as,Ri=Tσi×π ·r2, (3.1)29where T is the thickness of the electrode and r is the inner radius of the apparatus.By using above equation, ionic conductivities of WT (2.57×10−4(S/m)) and NTelectrode (1.04×10−4(S/m)) are calculated.From the ionic conductivity test result, it appears that the addition of TritonX-100 increases the porosity of the PEDOT:PSS electrode. In the next subsection,itwillbeshownhowthischangeofRiaffectscapacitorcharge/dischargeresponse.3.3 Cyclic voltammetryCyclic voltammetry (CV) is one of the most widely used electroanalytical tech-niques used to study electrochemical systems in the time domain. During a CVexperiment, a potential sweep of constant rate (voltage/second) is applied to theworking electrode (WE) with respect to the reference electrode (RE). The currentthat flows from the working electrode to the counter electrode (CE) is measuredto determine the overall electrochemical response. When the CV reaches a setpotential, the potential sweep is reversed. This cycle is typically repeated and thedata is plotted as current (i) vs. potential (E).The basic relationship between stored charge (Q) to capacitance (C) at a givenvoltage (V), is given asQ =C·V, (3.2)30and the current,I =dQdt=C·dVdt. (3.3)Recognizing that dV/dt is potential scan rate v, we get,C =Iv. (3.4)With this simple derivation, we get an expression for the capacitance as a functionof current at a given scan rate.Cyclic voltammetry measurements are performed using a Solatron 1287A Po-tentiostat/Galvanostat with a three electrode system, which is composed of work-ing electrode (WE), reference electrode (RE), and counter electrode (CE). Propy-lene carbonate electrolyte with 0.1 M tetrabutylammonium perchlorate (TBAP)and Ag/Ag+ ion reference electrode (0.1 M TBAP, 0.01 M AgNO3), and ac-tive carbon (carbon-filled polytetrafluoroethylene, from W. L. Gore & Associates)counter electrode are used for these tests.The volume capacitance of both the WT and NT electrodes is measured at sev-eral scan rates to investigate the effects of porosity for capacitor charge/discharge.It has already been reported[30]-[32] that lower scan rate can yield higher capac-itance. When the electrochemical double layer capacitors are charged, the rate islimited by the RC charging times. At a high scan rates, only part of the electrode31is charged, whereas ions can have more time to penetrate deep inside the electrodeat a lower scan rate. Thus we measure higher volume capacitance at lower scanrates.�0.8�0.6�0.4�0.2 0�6�4�20246x 10�4VoltageCurrent  10mV/sscanrateWTNT�0.8�0.6�0.4�0.2 0�1.5�1�0.500.511.5x 10�4VoltageCurrent1mV/sscanrate  WTNTFigure 3.4: The cyclic voltammetry of the PEDOT:PSS electrodes at 10mV/s (top) and 1 mV/s (bottom) scan rates.32�0.8�0.6�0.4�0.2 0�3�2�10123x 10�5VoltageCurrent0.1mV/sscanrate  WTNTFigure 3.5: The cyclic voltammetry of the PEDOT:PSS electrodes at 0.1mV/s scan rate.As summarized in Table 3.1, both WT and NT electrodes show increased ca-pacitance at lower scan rate. The WT electrode demonstrates higher volumetriccapacitance than NT electrode, particularly at intermediate and fast scan rates.This can be explained with the reduced time constant of WT electrode due to theincreased porosity. However this volume capacitance difference is not apparentat 0.1 mV/s (3 %). This suggests that Triton X-100 can improve the charge/dis-charge speed but it does not increase the capacitance itself.Cyclic voltammetry measurements for both WT and NT at different scan rates(10 mV/s, 1 mV/s, 0.1 mV/s) are presented in Figures 3.4-3.5. The CV plot showsa deviation from the ideal rectangular shape. In particular the CV plots of 10mV/sscanrateshowsaslopethroughoutthecharge/dischargecycle, andtherefore33a slow increase in the capacitance. If the time constant for electrode chargingis lower than the complete scan time, capacitance can increase throughout thecharge/discharge cycle. From equation (3.4), we can see that current is functionof the capacitance. The CVs show sharp increase in the current at both ends of thevoltage window at lower scan rates. This can be explained by the charge transferreaction that is a function of the applied potential to the electrode. Equation (3.3),showsthatreducedelectrodecurrentcanmakethereactioncurrentsmoreapparentat slow scan rate3.4 Electrochemical impedance spectroscopyElectrochemicalimpedancespectroscopy(EIS,sometimesalsocalledACimpedance)is a powerful diagnostic tool in the analysis of electrochemical systems. By mea-suring the impedance of a system over range of frequencies, we can characterizethe energy storage behavior and utilize this information to improve the perfor-mance of the system.A low AC perturbation potential is applied to the cell in order to charge/dis-charge the electrode. Ion movement associated with the energy storage elementswill induce an AC which flows through the cell. Electrochemical impedance iscalculated as the rate of the applied potential and measured current into the de-vice. A frequency response analyzer (FRA) is used to measure the phase shiftwhich can determine whether the cell response is capacitive, inductive, or resis-tive at a particular frequency.34In the equivalent circuit analysis, the ionic resistance represents an ionic con-ductive path through an electrode and a capacitor forms the energy storage ele-ment. With the AC potential input, the resulting output current signal is at thesame frequency but is shifted in phase and has different amplitude. The phaseprovides information on the nature of the device impedance. For instance, withzero phase, the system is behaving in a purely resistive manner. For 45◦phase,ions diffuse inside but have not reached the end of the electrode[33]. For 90◦phase, it responds purely in a capacitive manner. The magnitude ratio betweeninput voltage and output current, along with phase, can be used to evaluate boththe real and imaginary impedance at any frequency.The measured EIS impedance and phase information is expressed graphicallywith Bode plots, which depict the magnitude and phase of the impedance as afunction of frequency, or with the Nyquist plots on which the real and imaginaryparts of impedance are plotted.The three electrode cell setup, with a Solatron 1287A Potentiostat/Galvanos-tat, 1260A Impedance/Gain-phase Analyzer, is used for the EIS experiments. Itis noticed that direct electrical connection to the electrode using an alligator clipinduces corrosion to the alligator clip due to reactions with the paper substratesoaked with electrolyte. In order to avoid the corrosion, aluminum foil is attachedon top of the paper using kapton tape. The paper is cut in half through the thick-35!"#$%&#$'()%"'*+' ,+'-.$$'Figure 3.6: a) Electrode with aluminum foil substrate b) coil shape platinumwire and stainless steel sheet piece.ness in order to minimize the gap between the paper and the aluminum foil. Asshown in Figure 3.6 the PEDOT:PSS electrode is deposited on both paper and alu-minum foil without electrical disconnection. After curing the electrode, electricalresistance is measured to ensure the solid electrical contact between aluminumfoil part electrode and the main rectangular electrode. Coiled platinum wire (top)and rectangular stainless steel sheet (bottom) are used to connect the alligator clipto the electrodes in order to prevent corrosion between the alligator clip and theelectrode.A stainless steel wire of 0.7 mm diameter is connected in parallel to the refer-36Figure 3.7: Abnormal impedance response without stainless steel wire con-nection to RE at high frequency.ence electrode (RE) to remove the high frequency abnormal response. Figure 3.7shows there is an unexpected impedance response from 105Hz to 103Hz. We be-lieve the RC delay caused by RC components inside reference electrode inducesthis phenomenon at high frequency. Thus a stainless steel wire is connected inparallel to the RE to make another repeatable ground path at high frequency. Inorder to see the input capacitance effect, 0.1 µF, 2.2 µF, 4.7 µF capacitance areconnected in series with the stainless steel wire. However, it was found that allcapacitance values corrected the high frequency response and as a result, only thestainless steel wire is connected as shown in Figure 3.8.37CE WE RE !"#$%&'(()("''&)*$+')CE WE RE Figure 3.8: The EIS three electrode cell setup.From 105Hz to 10−4Hz, 0 V DC, 50 mV AC input setup is used for the test.A 50 mV AC input is found to give good signal to noise while not including anon-linear response. A 50 mV AC input measurement data is compared with 10mV AC input response to ensure it is tested in the linear response regime.From the phase angles (45◦, 67.5◦, and 90◦) of the Bode plot, the capacitorcharge condition can be inferred. A phase angle of 45◦can be associated with thefact that the ions do not reach the end of the electrode[33] and therefore it is calledas an infinite response. The 90◦phase angle suggests that the electrode behavespurely capacitive manner. The phase angle 67.5◦is transition angle from infinite38diffusion response to capacitive response, which corresponds to the RC time con-stant. The RC time constant of the WT electrode is 50.46 s (3.15 x 10−3Hz) andof the NT electrode is 112.9s (1.41 x 10−3Hz) that is demonstrating the effect ofaddition of the Triton X-100.10�410�310�210�11001011021030123456x 106frequencyvolumecapacitance(F/m3)  WT from EISNT from EIS1 mV/s10 mV/sFigure 3.9: Volumetric capacitance vs. frequency plot from the EIS and itscorresponding scan rate in the CV.It was previously mentioned that the scan rate does affect the observed capac-itance in the CV measurement. We calculated a volume capacitance (F/m3) fromthe EIS measurement in order to see whether we can observe same response inEIS. Imaginary impedance of the EIS is used to calculate the capacitance by usingthe capacitance and impedance relationship,39Z =1jω ·C. (3.5)The volume capacitance is plotted as a function of frequency and two scan rates(10 mV/s and 1 mV/s) are noted in Figure 3.9. In order to calculate the frequencywhich corresponds to each scan rate, the relationship given asf =12·π ·τ, (3.6)is used, where f is the frquency and τ is the RC time constant which is calculatedfrom cyclic voltammetry experiment at each scan rate.The volume capacitance calculated from the equation (3.6) is summarized inTable3.2. Thescanrate0.1mV/s(6.12x10−5Hz)isoutoftheEIStestfrequencyrange thus it is not compared.Table 3.2: Volume capacitances calculated from the EIS data for both WTand NT at different scan rates.Sample Scan rate(mV/s) Volume capacitance(F/cm3) Increase(%)WT/NT 10 3.18/2.72 19WT/NT 1 4.56/3.84 17Unlike results of the CV test (Table 3.1), the capacitance ratio decreases be-tween the WT and NT at lower scan rates are not obvious in the EIS measurement.The voltage range and the signal difference might cause this differences. The volt-age range of the CV is ±0.6V which is in non-linear impedance response range,40whereas it is linear regime (±50mV) in the EIS. The voltage input for the CV islinearly increasing, whereas the EIS input is the sinusoidal.In this chapter, we have shown that the charging rate of the PEDOT:PSS elec-trodeisincreasedbyusingthesurfactantTritonx-100. WiththeCVmeasurement,it has been found that the WT electrode demonstrates higher volumetric capaci-tance than NT electrode at intermediate and fast scan rates. An attempt has beenmade to relate the time domain response (CV) to the frequency domain response(EIS). However, the capacitance ratio decrease between the WT and NT at lowerscan rate is not obvious in the EIS measurement.41Chapter 4Multiple time constant modelIn Chapter 3, electroanalytical measurement techniques are utilized in order tocharacterize the performance of the printed supercapacitor. Volume capacitancesof 9.36 F/cm3and 9.09 F/cm3are measured for both the WT electrode and NTelectrode. Although the magnitude of the capacitance is similar, different charg-ing times are observed, e.g., 50.46 s for the WT electrode and 112.9 s for the NTelectrode. In order to investigate the rate limiting factors of capacitor charging, anequivalent circuit model is utilized. A multiple time constant model is proposedin order to explain physical origin of the distributed time constant behaviour. Thethickness variation of the PEDOT:PSS electrode is assumed to be the primary rea-son of the measured behavior and thus actual thickness variation is incorporatedin the modelling. The CPE response and its physical origin are discussed in Sec-tion 4.1. The origin of capacitance distribution of the PEDOT:PSS electrode isdescribed in Section 4.2. In Section 4.3, a multiple time constant model is con-42structed based on thickness variation of the electrode. Data fitting with the EISexperiment of the WT electrode suggests that thickness variation is not the onlyfactor leading to increase in capacitance distribution. Finally, in Section 4.4, allthe parameters used in modelling are compared to the experimental data.4.1 The origin of the CPE responseA frequency dispersive response is often observed in EIS and cannot be explainedby simple arrangements of passive electrical elements such as resistors, capaci-tors, or inductors. A common form of the response observed in electrochemicalsystems, but for which concrete physical explanations are usually lacking, is theconstant phase element (CPE) impedance given byZ(ω)=￿1CCPE￿(jω)−α, (4.1)whereCCPEand α are the CPE coefficient and the CPE exponent respectively.The time constant dispersion mainly caused by the capacitance dispersion isthe critical aspect necessary to explain the capacitor charge/discharge response.Thus, extensive attention has been paid to investigate the capacitance dispersiveresponse. The CPE is widely used in electrochemical impedance modelling inorder to represent time constant distribution in an equivalent circuit [34]-[35]. It43also demonstrates excellent fit to the EIS results in our experiments. However, itdoes not help to identify the physical origin of this behaviour.There are research results demonstrating pore size distribution [36] or porelength distribution [37] as a physical origin of frequency dispersion. In thesetwo reports, analytical modelling is performed to explain the abnormal EIS re-sponse of activated carbon fiber cloth electrode (ACFCE). The measured phaseangle of the impedance spectrum was larger than 45◦at high frequencies andsmaller than 90◦at low frequencies. Song et al. [36], explained it using a dis-persed pore size distribution (PSD), whereas Lee et al. [37], demonstrated betterfitting with a pore length distribution (PLD). They also mentioned that the PSDof the ACFCE is quite narrow and thus the PSD has a negligible influence on thenon-ideal impedance behaviour of the AFCC. The PLD theory has demonstratedimproved fitting and it also has a clear physical basis. However, their approachwasrestrictedtocylindricallyporouselectrodes. Thisisnotapplicabletotheanal-ysis of nanoparticle composite electrodes used in this study.Transmission line models of micro-to-nanoporous activated carbon materialscomposed of agglomerates and grains have been reported [38]. Linear transmis-sion lines and self-affinity fractal geometry were used to create an equivalentcircuit model. In general, they demonstrated a successful impedance responsemodel with different electrode conditions such as electrode thickness, agglomer-ate radius, and pore size distribution. It has been demonstrated that size of the44agglomerated particle affects the capacitance and charging property; i.e., largeinterfacial area with small ion transport paths of small particles show higher ca-pacitance but slow charge/discharge time compared to large particles. However,faradaic impedance was not taken into account in the modelling.Figure 4.1: Microscope image of the PEDOT:PSS electrode surface.It has been noted that the specific capacitance per unit area of the carbon elec-trode ranges from about 1 µF/cm2for the basal plane to 70 µF/cm2for the edgeplane [39]. It has been demonstrated experimentally that the CPE exponent αdecreases with increasing amount of the edge orientation of the carbon electrode.This result suggests that increased distribution of capacitance induced by inho-mogeneity of the surface causes frequency dispersive response. Z. Kerner andT. Pajkossy have reported that atomic scale inhomogeneities of the electrode are45the reason of capacitance dispersion and experimentally demonstrated its effect inCPE response [40].4.2 The origin of capacitance distribution of thePEDOT:PSS electrodeThe capacitance distribution of the PEDOT:PSS electrode is also believed to bethephysicaloriginoftheCPEresponse. Thelikelyfactorswhichcancausetheca-pacitance distribution are investigated in this chapter. As mentioned in fabricationsection, aqueous PEDOT:PSS solution which consists of 30 - 100 nm particles[41] is deposited on commercial paper by an inkjet printing process as shown inFigure 4.1. During the deposition, the sample is cured by a substrate heater (100◦C) to evaporate water solvent quickly. This curing condition may enhance thehydrodynamic flow inside the deposited solution [42] and it can induce inhomo-geneous thickness variation.The thickness variation measurements are performed by using the Dektak 150surface profiler (from Veeco company, USA). PEDOT:PSS is deposited from topto bottom and thickness is measured along the five lines shown in Figure 4.2. Fig-ure 4.3 shows that thickness profile of the WT electrode is flatter than that of theNT electrode, which is believed to be the effect of the Triton X-100.It is observed through the experiment that a curing temperature of around46!"#$$#%&'()*+,#!"#$$#%-./*+,#012345066#/'789.:8(#/.;'<:8(#!"#$$#%&'()*+,#!"#$$#%-./*+,#4+.<=('99#9<>(#/.;'<:8(#?!#?"#?@#?A#?B#>,#C,#Figure 4.2: a) The deposition direction (from top lines to bottom lines) ofthe PEDOT:PSS electrode b) the thickness scan direction of the PE-DOT:PSS electrode.Table 4.1: The average thickness of the two PEDOT:PSS electrodes (WTand NT) and bare paper roughness along five different lines.Sample #1 (µm) #2 (µm) #3 (µm) #4 (µm) #5 (µm)WT 16.8 10.0 43.0 37.1 33.7NT 6.7 23.7 69.3 70.8 49.6Bare paper 3.5 NA 3.2 NA 3.1130◦C causes the PEDOT:PSS electrode to lift off from the paper substrate orbreak the pattern. Therefore, the temperature of the substrate heater was main-tained around 100◦C. However, the temperature is not sufficiently high to get in-stant water evaporation due to the reduced heat transfer through the paper.470 2000 4000 6000 8000 100001200014000012345678x 105Lengthoftheelectrode(µm)Thicknessoftheelectrode(˚A)  WT thickness profile of #3 scan0 2000 4000 6000 8000100001200014000024681012x 105Lenghtoftheelectroce(µm)Thicknessoftheelectrode(˚A)  NT Thickness profile of #3 scanFigure 4.3: Thickness profile for both WT and NT electrodes at #3 (center)position scan.It is observed that the parts of the PEDOT:PSS electrode along the top linesare evaporated faster than bottom lines and it took time to completely evaporate48water of the bottom of the electrode. We believe that slower evaporation timethan accumulation of droplets caused asymmetric evaporation time and it causedhydrodynamic flow towards the center of the electrode from the top lines.The thickness variation caused by nanofibers of the paper substrate is mea-sured. However it is found to be minor (3.1 - 3.5 µm) compared to the thick-ness of the electrode as shown in Table 4.1. The total thickness variations of thePEDOT:PSS electrode which is measured from scan #1 to #5 are shown in Fig-ure 4.4. The histogram suggests that the thickness variation is not following aGaussian distribution or lognormal distribution which is often used in analyticalmodelling [36] - [37]. In the next chapter, the measured thickness is incorporatedinto a proposed model and the results are compared to the fitting that has usedlognormal distribution.4.3 Multiple time constant modellingThe transmission line model using multiple time constants has been constructedto represent the dispersed capacitance behaviour of the electrode. Based on thethickness measurement data in Figure 4.4, a transmission line model with differ-ent thickness sections is proposed. A difference in height means that at differentlocations it takes more or less time to charge the sample, as the time for ion trans-port through the thickness will vary with height. This variation is simulated bymodelling each sample as 200 parallel transmission lines, with the distribution of49�1 0 1 2 3 4 5 6 7 8x 10500.511.522.5x 104Thicknessoftheelectrode(˚A)Frequency  WT thickness variation0 2 4 6 8 10 12x 10500.511.522.5x 104Thicknessoftheelectrode(˚A)Frequency  NT thickness variationFigure 4.4: ThetotalthicknessdeviationofthePEDOT:PSSelectrodewhichis measured from scan #1 to #5.transmission line lengths matching the measured height distribution. Using thisapproach, eachelectrodeisdividedinto20x10sectionswithdistributedthicknessvariation. As shown in Figure 4.5, each section is treated as a separate transmis-50!"#$%&'(()*#+,")-'&.,")Figure 4.5: The simple drawing of the multiple thickness electrode (top) andequivalent circuit for one section of the electrode (bottom).sion line within the multiple time constant model.The PEDOT:PSS electrode is attached to a paper substrate and the front side51Figure 4.6: The three electrode EIS measurement setup for measuring so-lution resistance. A stainless steel sheet (SSS) is used as a workingelectrode. WE, RE, and CE refer to the working electrode, referenceelectrode, and counter electrode respectively.of the electrode is facing the electrolyte phase. In the equivalent circuit in Fig-ure 4.5, the electrical resistance, Re, is treated as a short circuit due to its muchlower resistance (about 103) when compared to the ionic resistance. During theelectrochemical charging or discharging, ions from the electrolyte pass through aresistance of Riand charge/discharge the capacitance. In this model, we allow theelectrode to be charged from two sides, since ions enter the polymer both throughthe paper and from liquid electrolyte in contact with the polymer.52Thefrontsidetransmissionlineisconnectedtothesolutionresistance,whereasback side transmission line is connected to Rpand Rs. These two transmissionlines are connected in parallel to make one transmission line which representsone section of the electrode. A faradaic impedance, Rf, takes into account chargetransferreactionsthataretypicallyrelevantatextremelylowfrequency. Itisaddedparallel to the capacitance as shown in Figure 4.5.In order to calculate the ion penetration depth, the solution resistance (resis-tance between WE and RE) is measured with three electrode measurement setupas shown in Figure 4.6. As shown in Figure 4.5, there are two solution resistancesdue to the PEDOT:PSS electrode charging with ions penetrated from the papersubstrate. After covering one side of stainless steel sheet (SSS, 12 mm x 12 mm) completely with the kapton tape, the solution resistance, Rs, from the front side(SSS is facing RE) and back side (SSS is facing opposite direction of RE) is mea-sured. The solution resistance between WE and RE is measured to be the samewhether the stainless steel sheet is facing towards the RE or opposite direction tothe RE. Therefore, the solution resistance is cancelled out when calculating RCtime constant from front side charging (Ri· Cfront) and back side (after penetratingpaper) charging (Ri· Cback) in the equation below.Ri·Cfront= Ri·Cback. (4.2)53By substituting Ri,Cfront, and Cback, we haveRi￿TRT￿·C￿TRT￿=￿Rp+Ri￿T −TRT￿￿·C￿T −TRT￿. (4.3)The front side ion penetration depth is calculated asTR=Rp·T +Ri·TRP+2·Ri, (4.4)where Rp, TR, and T refer to paper resistance, front side ion penetration thickness,and total thickness of the electrode respectively.Now, we derive the impedance of the right side of the RC transmission line(RCTL)inFigure4.5. ThetotalvaluesoftherightsideoftheRCTLaredefinedasRiR(ionic resistance),CR(capacitance), RfR(faradaic resistance). We can expressthe voltage drop through the unit resistance RiasdV(x)=i(x)·Ri·dx, (4.5)where Riis given by RiR/ TR(Ω/cm) and TRis the right side ion penetrationthickness. We have the charging current of a unit capacitance, di(x), asdi(x)=V(x)Z·dx, (4.6)where Z is given by Z = RfRTR/(1+ jωCR·RfR)(Ω· cm). By using equation 4.5and equation 4.6, we get54d2V(x)dx2=RiZ·V(x). (4.7)The general solution of equation 4.7 has the formV(x)=A·exp￿￿RiZ·x￿+B·exp￿−￿RiZ·x￿. (4.8)The total current is divided into unit capacitance charging currents. Near the cen-ter of the electrode (see Figure 4.5), we can define boundary condition as x = 0,and ionic current i(0)=0, and thus we get A=B. The total current into the elec-trode must pass through the Z’s, so:i(TR)=￿TR0V(x)Zdx=A√Ri·Z￿exp￿￿RiZ·TR￿−exp￿−￿RiZ·TR￿￿. (4.9)The impedance of right side of the RCTL is obtained asZR(TR)=V(TR)i(TR)=√Ri·Z·coth￿￿RiZ·TR￿. (4.10)The impedance of both sides of the RCTL which are connected in parallel in the55three electrode system is obtained asZRL=11Zfront+1Zback, (4.11)where,Zfront=√Ri·Z·coth￿￿RiZ·TR￿+Rs, (4.12)andZback=√Ri·Z·coth￿￿RiZ·TL￿+Rp+Rs. (4.13)The total impedance of 20 x 10 RCTLs is obtained asZt=120∑j=110∑i=1￿1ZRL·i·j￿. (4.14)4.4 Data fitting of the multiple time constant modelThe multiple time constant modelling proposed in Section 4.3 is compared to theEISexperimentalresultinthissection. Inordertofittheexperimentaldata, capac-5610�410�310�210�1100101102103104105102103104105frequencyImpedance  NT simulationNT experiment10�410�310�210�1100101102103104105�80�70�60�50�40�30�20�100FrequencyPhaseangle  NT simulationNT experimentFigure 4.7: The Bode plot of the NT electrode.itance, ionic resistance, faradaic resistance, solution resistance, paper resistance,and thickness variation of the electrode are varied, starting with the experimen-tally determined values shown in Table 4.2. It has been found that capacitance570 0.5 1 1.5 2 2.5 3 3.5x 10400.511.522.533.5x 104RealimpedanceImaginaryImpedance  NT simulationNT experimentFigure 4.8: The Nyquist plot of the NT electrode.and thickness variation are the main factors determining the capacitive transitionresponseandtheionicresistanceisadominantfactoratfrequencyrange10−1Hz-102Hz. First, the NT electrode fitting is shown in Figure 4.7 - 4.8. It demonstratesgoodfittingovertheentirefrequencyrangeonboththeBodeandtheNyquistplot.This is believed to demonstrate that the distributed thickness variation may be thephysical origin of the time constant distributed response in the electrochemicalimpedance measurement.The multiple time constant model of the WT electrode is compared to the EISexperimental data in Figures 4.9 - 4.10. The phase angle of the WT simulationshows slightly different capacitive transition than the experiment. This suggests5810�410�310�210�1100101102103104105102103104105frequencyImpedance  WT simulationWT experiment10�410�310�210�1100101102103104105�90�80�70�60�50�40�30�20�100FrequencyPhaseangle  WT simulationWT experimentFigure 4.9: The Bode plot of the WT electrode.that the WT electrode has a capacitance distribution that is governed by more fac-tors than just the distribution of thickness.590 0.5 1 1.5 2 2.5 3x 10400.511.522.53x 104RealimpedanceImaginaryImpedance  WT simulation WT experiment Figure 4.10: The Nyquist plot of the WT electrode.The thickness is the primary parameter determining the total resistance andcapacitance of an RCTL. However, the measured thickness variation did not fullyexplain the frequency dispersion for WT electrode. In order to find the effectof the highly distributed resistances and capacitances, a lognormal distributionfunction (LNDF) is employed, i.e., different standard deviations of the thicknesscan control the capacitance and ionic resistance distribution in the electrode. Thethickness of the 20 x 10 sections is distributed with LNDF instead of using theactual measured thickness. Data fitting with different sigma values demonstrateresults consistent with experimental reports [39]-[40], i.e., the distribution of thecapacitance determines the slope of the capacitive transition. The multiple timeconstant modelling with LNDF demonstrates almost identical fitting at with the6010�410�310�210�1100101102103104105�90�80�70�60�50�40�30�20�100FrequencyPhaseangle  mt=1.1mt = 0.6mt = 0.1WT experimentmmFigure 4.11: Phase angle response with different σtvalues.EIS experimental result at a high standard deviation (σt= 1.1). This suggest thatthe thickness variation is not the only factor leading to increase in capacitancedistribution of the PEDOT:PSS electrode. We need more detailed investigation tofind out the cause of the distributed Riand C. However, the effect of the TritonX-100 can be the cause for the distributed RiandC. There are no research reportsof the Triton X-100 effect on the morphology of the PEDOT:PSS electrode. How-ever, if increased porosity is a local effect, Riand C values in that region will bemuch lower than other area.Therefore, the addition of Triton X-100 can increasethe distribution of resistance and capacitance values.By comparing simulation results of different sigma values, it also has been61found that the capacitance distribution is a limiting factor of capacitor charging.AsshowninFigure4.11,thesimulationwithhighsigmatakesmoretimetochargethe capacitor, i.e., capacitor charging time and σthas an inverse proportional re-lationship.Table 4.2: Experiment and modelling data comparison. C, σi, Rs, and Rfrefer to the capacitance, ionic conductivity, solution resistance, andfaradaic resistance respectively.Sample C (F) σi(S/m) Rs(Ω) Rf(Ω)WT simulation 0.058 2.32 ·10−4305 1.8·105WT experiment 0.075 2.57·10−4N/A N/ANT simulation 0.039 1.13·10−4330 1.5·105NT experiment 0.054 1.04·10−4N/A N/AAll the parameter values used in fitting are compared to the experimental re-sults in Table 4.2. The capacitance used in the simulation of both WT and NTelectrodes show lower value than the experimental result. The capacitance of theconnector,whichisnotincludedinsimulation,caninducethisdifferenceincapac-itance between experiment and simulation. The ionic conductivities (σi) for bothWT and NT electrode which are used in simulation show good agreement withexperiment because of the realistic ion penetration mechanism employed in themodelling. The solution resistances (Rs) of 305Ω(WT), 330Ω(NT) are used forsimulation. Unlike the solution resistance measurement apparatus (Figure 3.3), inthe electrochemical impedance measurement (Figure 3.8), the ion transport pathsare not restricted and thus simple comparison between simulation data and ex-periment data is not relevant. Therefore, the only simulation data is described in62Table 4.2. We believe that the experimental variation of positioning WE and REcontributed to the differences in resistance between WT and NT samples.The multiple time constant model is constructed based on the theories [39]-[40] that explain physical origin of the CPE response as a capacitance distribu-tion. Thickness variation of the PEDOT:PSS electrode is assumed as a primaryreason for the CPE response and thus actual thickness variation is incorporatedin the modelling. The NT electrode demonstrated good fitting through out thewhole frequency range. It suggests that the thickness variation can be the maincause of the capacitance distribution of the PEDOT:PSS electrode. Unlike the NTsimulation, the WT fitting suggests that there are more causes of the capacitancedistribution than just the thickness. We need further investigation to determine thecause. However Triton X-100 is the most probable origin for the capacitance dis-tribution. If the increased porosity of the WT electrode is not uniform, capacitorcharge time will be different locally and it can also cause the capacitance distri-bution.It has been shown that the distributed capacitances combined with ionic con-ductivities are the rate limiting mechanism for supercapacitor charging in Figure4.11. The addition of the Triton X-100 decreases the thickness variation. How-ever, data fitting suggests that the increased porosity caused by the addition ofTriton X-100 might also induce capacitance distribution. In chapter 3, we de-scribed that the addition of Triton X-100 can improve charging speed but it is not63improving capacitance itself. However, considering that the ideal charging timefor supercapacitors are between 10−2−102s [43], addition of Triton X-100 isbetter in practical applications.64Chapter 5ConclusionsIn this thesis, a flexible supercapacitor is fabricated with micropatterning tech-nology and charge/discharge response is measured and simulated in order to un-derstand the rate limiting mechanisms and assess its value. In contrast to the ini-tial purpose of the Triton X-100; i.e., increase the PEDOT:PSS solution adhesionto the substrate, the addition of the surfactant also demonstrated higher volumecapacitance at fast and intermediate scan rates (10 mV/s - 1 mV/s). Increasedporosity of the PEDOT:PSS electrode is the cause of the faster capacitor charingeffect. The ionic conductivity measurements have been performed to validate thehypothesis of increased porosity in the WT electrode.In order to investigate the rate limiting factors of capacitor charging, electro-chemical impedance measurements and equivalent circuit modelling are utilized.A frequency dispersive response that cannot be explained by simple arrangements65of electrical elements such as resistors, capacitors, or inductors is observed inEIS measurements. Among various theories explaining the physical origin of thefrequency dispersive responses, multiple time constant modelling is constructedbased on the theories that explain physical origin as a capacitance distribution.Thickness variations are measured and incorporated in a multiple time con-stant modelling which consists of 200 parallel RC transmission lines. The basicconcept of this modelling is ions will take more or less time to charge the ca-pacitors if the thickness is varying throughout the electrode. Therefore, unlike thesimpleequivalentcircuitmodelling, thismultipletimeconstantmodellingdemon-strates the time constant dispersive behaviour.Thickness variation did not fully explain the frequency dispersion in sam-ple with Triton X-100. However, sample without Triton X-100 demonstratedgood agreement between experiment and simulation which can be explained withmore highly distributed thickness distribution. A lognormal distribution function(LNDF) is utilized in order to investigate the relationship between the capaci-tance distribution and the capacitor charging, i.e., different standard deviations oftime constant can control the capacitance distribution in the electrode. A gooddata is demonstrated at high standard deviation (σt= 1.1), which suggests thatthere are more factors distributing capacitances than just thickness distribution.By comparing simulation results of different sigma values, it has been found thatthe capacitance distribution is also a capacitor charge limiting factor along with66total capacitance.The thickness measurements and data fitting suggested two effects of the Tri-ton X-100. The Triton X-100 decreases the thickness variation of the PEDOT:PSSelectrode. However, the data fitting suggests that there might be additional capac-itance distribution caused by the addition of Triton X-100. The comparison withthe sample without Triton, the WT electrode demonstrates substantial chargingspeed improvement at high scan rates (10 mV/s) and still faster charging speed atintermediate scan rates (1 mV/s). Therefore, considering ideal charging time forsupercapacitors (10−2−102s), addition of Triton X-100 is beneficial in practicalapplications.There are research reports which already demonstrated experimentally thatthe distributed capacitance is the physical cause of the time constant distributedbehaviour in electrochemical impedance measurement. However, there was noanalytical modelling supporting these experimental results. This is the first an-alytical model showing that the capacitance distribution of the electrode is themain cause of the time constant distributed response. In addition to the measuredthicknessvariation,comparisonresultsofmainparameters(capacitance,andionicconductivity)usedinmultipletimeconstantmodellingandexperimentresultsalsosuggest that this multiple constant model is explaining the real cause of the timeconstant distributed response.67Bibliography[1] J. P. Zheng and T. R. Jow, “High energy and high power density electro-chemical capacitors,” Power Sources, vol. 62, pp. 155-159, 1996. → pages2[2] A.ChuandP.Braatz, “Comparisonofcommercialsupercapacitorsandhigh-power lithium-ion batteries for power-assist applications in hybrid electricvehicles: I. Initial characterization,” Power Sources, vol. 112, pp. 236-246,2002. →pages 2[3] J. Chmiola et al., “Anomalous Increase in Carbon Capacitance at Pore SizesLess Than 1 Nanometer,” Science, vol. 313, pp. 1760-1763, 2006. →pages6[4] Celine Largeot et al., “Relation between the Ion Size and Pore Size for anElectric Double-Layer Capacitor,” J. Am. Chem. Soc., vol. 130, pp. 2730-2731, 2008. →pages 6[5] Andrew Burke, “R&D considerations for the performance and application of68electrochemical capacitors,” Electrochimica. Acta., vol. 53, pp. 10831091,2007. →pages 6, 7[6] A. V. Rosario et al., “Investigation of pseudocapacitive properties of RuO2film electrodes prepared by polymeric precursor method,” Power Sources,vol. 158, pp. 795-800, 2006. →pages 7[7] H.Zhouetal.,“Theeffectofthepolyanilinemorphologyontheperformanceofpolyanilinesupercapacitors,” Solid State Electrochem., vol.9, pp.574580,2005. →pages 7[8] K. R. Prasad and N. Munichandraiah, “Fabrication and evaluation of450 F electrochemical redox supercapacitors using inexpensive and high-performance, polyaniline coated, stainless-steel electrodes,” Power Sources,vol. 112, pp. 443-451, 2002. →pages 7[9] E.FrackowiakandF.Beguin,“Carbonmaterialsfortheelectrochemicalstor-age of energy in capacitors,” Carbon, vol. 39, pp. 937-950, 2001. →pages6[10] R. Kotz and M. Carlen, “Principles and applications of electrochemical ca-pacitors,” Electrochimica. Acta., vol. 45, pp. 2483-2498, 2000. →pages vii,8, 9[11] D. A. Scherson and A. Palencsar, “Batteries and Electrochemical Capaci-tors,” The Electrochemical Society Interface, 2006. →pages vii, 869[12] S. Pay et al., “Effectiveness of Battery-Supercapacitor Combination in Elec-tricVehicles,” IEEE Bologna Power Tech Conference, June23th-26th, Italy,2003. →pages 9[13] X. Li et al., “Surface modification of polyaniline using tetraethyl orthosili-cate,” Colloid and Interface Science, vol. 322, pp. 429-433, 2008. →pages13, 16[14] K. S. Ryu et al., “Symmetric redox supercapacitor with conducting polyani-line electrodes,” Power Sources, vol. 103, pp. 305-309, 2002. → pages 13,16[15] P. Glynne-Jones et al., “A feasibility study on using inkjet technology, mi-cropumps, andMEMsasfuelinjectorsforbipropellantrocketengines.”Uni-versity of Southampton, UK, SO17 1BJ →pages 13[16] J. Brunahl and A. M. Grishin, “Piezoelectric shear mode drop-on-demandinkjet actuator,” Sensors and Actuators, vol. 101, pp. 371-382, 2002. →pages 13[17] H. Sirringhaus et al., “High-Resolution Inkjet Printing of All-Polymer Tran-sistor Circuits,” Science vol. 290, pp. 2123-2126, 2000. →pages 14[18] S.E.Burnsetal.,“InkjetPrintingofPolymerThin-FilmTransistorCircuits,”Material Research Society, 2003. →pages 14[19] M. J. Heller, “DNA MICROARRAY TECHNOLOGY: Devices, Systems,70and Applications,” Annu. Rev. Biomed. Eng., vol. 4, pp. 129-153, 2002. →pages 14[20] A. V. Lemmo et al., “Inkjet dispensing technology: applications in drugdiscovery,” Current opinion in biotechnology, vol. 9, pp. 615-617, 1998. →pages 14[21] A. Bietsch et al., “Rapid functionalization of cantilever array sensors byinkjet printing,” Nanotechnology, vol. 15, pp. 873-880, 2004. →pages 15[22] M. Kaempgen et al., “Printable Thin Film Supercapacitors Using Single-Walled Carbon Nanotubes,” Nano Letters, vol.9, pp. 18721876, 2009. →pages 15[23] D. Pech et al., “Elaboration of a microstructured inkjet-printed carbon elec-trochemical capacitor,” Power Sources, vol. 195, pp. 1266-1269, 2010. →pages 15[24] P. Gajendran and R. Saraswathi, “Polyanilinecarbon nanotube composites,”Pure Appl. Chem., vol. 80, pp. 2377-2395, 2008. →pages 16[25] M. H. Tsai et al., “Effects of pulse voltage on inkjet printing of a silvernanopowder suspension,” Nanotechnology, vol. 19, 2008. →pages 17[26] A. U. Chen and O. A. Basaran, “A new method for significantly reducingdrop radius without reducing nozzle radius in drop-on-demand drop produc-tion,” Physics of Fluids, vol. 14, pp. L1-L4, 2002. →pages 1771[27] T.C. Girija, M.V. Sangaranarayanan, “Polyaniline-based nickel electrodesfor electrochemical supercapacitorsInfluence of Triton X-100,” PowerSource, vol. 159, pp. 1519-1526, 2006. →pages 25[28] S. Devaraj and N. Munichandraiah, “The Effect of Nonionic Surfactant Tri-ton X-100 during Electrochemical Deposition of MnO2 on Its CapacitanceProperties,” Electrochemical Society, vol. 154, pp. A901-A909, 2007. →pages 25[29] H. Yanazawa et al., “Precision Evaluation in Kr Adsorption for Small BETSurface Area Measurements of Less Than 1 m2,” Adsorption, vol. 6, pp.73-77, 2000. →pages 26[30] V. Subramanian et al., “Mesoporous anhydrous RuO2 as a supercapacitorelectrode material,” Solid State Ionics, vol. 175, pp. 511515, 2004. →pages31[31] JeffreyW.Long et al., “VoltammetricCharacterizationofRutheniumOxide-Based Aerogels and Other RuO2 Solids: The Nature of Capacitance inNanostructured Materials,” Langmuir, vol. 15, pp. 780-785, 1999. →pages[32] J.H. Chen et al., “Electrochemical characterization of carbon nanotubes aselectrode in electrochemical double-layer capacitors,” Carbon, vol. 40, pp.1193-1197, 2002. →pages 31[33] J. Bisquert et al.,“ Impedance of constant phase element (CPE)-blocked dif-72fusion in film electrodes,” Electroanalytical Chemistry, vol. 452, pp. 229-234, 1998. →pages 35, 38[34] H. E. Brouji et al., “Analysis of the dynamic behaviour changes of super-capacitors during calendar life test under several voltages and temperaturesconditions,” Microelectronics Reliability, vol. 49, pp. 1391-1397, 2009. →pages 43[35] P. J. Mahon et al., “Measurement and modelling of the high-power perfor-mance of carbon-based supercapacitors,” Power Sources, vol.91, pp. 68-76,2000. →pages 43[36] H. K. Song et al., “Electrochemical impedance spectroscopy of porous elec-trodes: the effect of pore size distribution,” Electrochimica Acta., vol. 44,pp. 3513-3519, 1999. →pages 44, 49[37] G. J. Lee et al., “Kinetics of double-layer charging/discharging of the ac-tivated carbon fiber cloth electrode: effects of pore length distribution andsolution resistance,” Solid State Electrochem., vol. 8, pp. 110-117, 2004. →pages 44, 49[38] M. Eikerling et al., “Optimized Structure of Nanoporous Carbon-BasedDouble-Layer Capacitors,” Electrochemical Society, vol. 152, pp. E24-E33,2005. →pages 44[39] C. H. Kim et al., “An investigation of the capacitance dispersion on the frac-73tal carbon electrode with edge and basal orientations,” Electrochem Acta.,vol. 48, pp. 3455-3463, 2003. →pages 45, 60, 63[40] Z.Kerneretal.,“Ontheoriginofcapacitancedispersionofroughelectrodes,” Electrochem Acta., vol. 46, pp. 207-210, 2000. →pages 46, 60, 63[41] G. F. Wang et al., “Modification of Conductive Polymer for Polymeric An-odes of Flexible Organic Light-Emitting Diodes,” Nanoscale Res. Lett., vol.4, pp. 613-617, 2009. →pages 46[42] D. Soltman and V. Subramanian,“Inkjet-Printed Line Morphologies andTemperature Control of the Coffee Ring Effect,” Nano Letters, vol. 24, pp.2224-223, 2008. →pages 46[43] R. Kotz and M. Carlen, “Principles and applications of electrochemical ca-pacitors,” Electrochimica. Acta., vol. 45, pp. 2483-2498, 2000. → pages6474Appendices75Appendix ADesign of the printing nozzlehousingWe purchased inkjet printing equipment from Microdrop technologies (so calledMicrodrop system, German). However the design of the inkjet printing nozzlehas a disadvantage, i.e., the glass orifice is not visible from the outside thus itwas difficult to clean the clogged nozzle. The nozzles are also very expensive.Thus we designed custom inkjet printing nozzle housing by purchasing and in-tegrating MJ-ABP-01 single jet dispensing nozzle (from Microfab Technologiescompany, USA). The nozzle assembly and complete nozzle housing is shown inFigure A.1. Microfab nozzles are compatible with the Microdrop system. How-ever, the Microdrop system is designed for a nozzle of resistance 2.6MΩ, whereasthe resistance of the Microfab nozzle is lower (1.53MΩ). Therefore, attentionneeds to be paid when it is used over 100V to prevent component failure of the76!"#$%&'()*+','-./0'()*+'12344'536$23#7'!"#$%&'()*+'8'9:#$%;'!"##$%&'"()*"$&+,-(.$&/"0+,(-&)%12%*.)0*%&'"()*"$&30%*&3"'4&56(-&7.2&Figure A.1: Design and assembly of the inkjet printing nozzle housing.piezo drive system.Chemically non-reactive PTFE is chosen for the tubing material. However, itisoftenobservedthattheconnectionbetweennozzleandPTFEtubeareweakenedand disconnected due to the high air pressure which is used to clean the cloggednozzle. In order to prevent the PTFE tube from disconnecting, inner adhesiveshrink tube is used as shown in Figure A.1 - shrink tube 1. The adhesion layerof the shrink tube prevents disconnection during the cleaning process. The shrinktube is also used to prevent wire disconnection and scratching from the assemblyprocess (Figure A.1 - shrink tube 2). The O-ring is added to the tip of nozzle toprevent water from getting inside the housing as well as fix the position of nozzle(FigureA.1 - O-ring). It is important not to soak the housing body even with thisO-ring shielding. Luer lock fitting is used to connect the air pressure supply to the77nozzle housing (Figure A.1 - Luer Lock fitting). In order to get a pressure leakfree connection, the thread of the luer lock fitting is wrapped with PTFE film tape.The tube path hole is made smaller to ensure pressure leak free assembly. 1.8 mmhole is designed to the cap with 2.38 mm diameter PTFE tube. The larger tube issqueezed inside the cap without any problems.78Appendix BNozzle maintenance tipsCleaning a clogged nozzle is quite a time consuming task when we are usinginkjet printing system. Here, two practical tips are shared to clean a cloggednozzle. First, the vacuum backflush can be an effective approach to remove anyloose clog. The clogged nozzle tip should be immersed in a solvent. The strongbackward stream with solvent can rapidly remove particles. The pressure controlunit (AD-E-130) of the Microdrop system can be used to apply vacuum to thenozzle. The other method involves the use of a low power ultrasonicator withnitrogen gas and solvent. It is often observed that dust moves inside the nozzleafter sonication. However, the dust clogs the nozzle tip again when it is used later.Byapplyinglowgaspressuretothecloggednozzle, wecancauseconstantsolventflow through the tip and this helps to maintain the dust position during sonication.In order to avoid damaging the electrical connection part, we should be carefulnot to immerse the whole nozzle into the sonication bath.79


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items