THE PENNSYLVANIA STATE UNIVERSITY SCHREYER HONORS COLLEGE JOHN AND WILLIE LEONE FAMILY DEPARTMENT OF ENERGY AND MINERAL ENGINEERING

Transcription

1 THE PENNSYLVANIA STATE UNIVERSITY SCHREYER HONORS COLLEGE JOHN AND WILLIE LEONE FAMILY DEPARTMENT OF ENERGY AND MINERAL ENGINEERING DEVELOPMENT OF CARBON-BASED SUPERCAPACITORS AND A DISCHARGE MODEL DANHAO MA SPRING 2015 A thesis submitted in partial fulfillment of the requirements for a baccalaureate degree in Energy Engineering with honors in Energy Engineering Reviewed and approved* by the following: Ramakrishnan Rajagopalan Assistant Professor of Engineering, Penn State Dubois Thesis Supervisor Sarma Pisupati Professor of Energy and Mineral Engineering Honors Adviser * Signatures are on file in the Schreyer Honors College.

2 i ABSTRACT A unique carbon nanotube-based electrode processing technique, named post-synthesis self-assembly technique, was developed in this thesis. The prepared electrodes obtained superior properties including flexibility, binder-free, light-weight, and excellent pore structures for aqueous supercapacitors. The supercapacitors were demonstrated to be charged/discharged at high current load of 50A/g for over 10,000 cycles without any fading in the capacitance. In this work, a power density of 1040 kw/kg was obtained in the supercapacitors using carbon nanotubes-based electrodes. That was due to the unique porous electrode structures and low internal electrical resistance. The specific capacitance was enhanced by decorating the pseudocapacitive materials on the carbon nanotubes-based electrodes. By doing so the supercapacitor maintained the high power performance and doubled the energy performance. The supercapacitors showed promising applications in microelectronics, and flexible energy storage devices. A physicochemical mathematical model was developed to understand the discharge characterization of supercapacitors including electric double layer formation, chargeand mass-transfer limitations, and resistive contact at the electrode/current collector interfaces. The model helped design desired properties and minimize unfavorable reactions in the supercapacitors during the charge/discharge cycles. The model also served as a tool to diagnose the issues on the discharge behavior, and to improve the supercapacitor system.

3 ii Table of Contents List of Figures v List of Tables vii Acknowledgements......viii Chapter I: Introduction Overview of Supercapacitors Energy Storage Mechanisms in Supercapacitors EDLC Energy Storage Mechanism Pseudo-capacitor Energy Storage Mechanism...4 Chapter II: Carbon Nanotube Electrode Fabrication Techniques Overview of Carbon Nanotubes Electrodes in Supercapacitors A Wet Chemistry Post-synthesis Self-assembly Technique Characterizations of Self-assembled CNT Electrodes Purity Surface Area Porosity 9 Chapter III: Polyvinyl Alcohol Hydrogel Separator Fabrication of the PVA separator Characterizations of the PVA separator 12 Chapter IV: Electrochemical Characterization Techniques Cyclic Voltammetry Electrochemical Impedance Spectroscopy.14

4 iii 4.3 Constant Current Charge and Discharge.. 15 Chapter V: Electrochemical Performance of Aqueous EDLC Assembly of the Aqueous EDLC Cyclic Voltammetry Electrochemical Impedance Spectroscopy Constant Current Charge/discharge of the Aqueous EDLC..19 Chapter VI: Aqueous Pseudocapacitors Surface Oxidation of CNT Electrodes Assembly of the Aqueous Pseudocapacitors Constant Current Charge/discharge of the aqueous Pseudocapacitors Electrochemical Impedance Spectroscopy.24 Chapter VII: Physicochemical Model on Discharge Behaviors of Supercapacitors Introduction on Mathematical Models of Supercapacitors Mathematical Development Simulation Results Contributions from Double Layer Formation and Separator Contributions from Double Layer Formation, Ionic Resistive Separator, Decomposition Reactions, and Resistive Contact Parameters Variation as a Function of Current Density Internal Verification..40 Chapter VIII: Discussion..41 Chapter IX: Conclusions.. 43 References...44

5 Academic Vita.49 iv

6 List of Figures Figure 1. Ragone plot..2 Figure 2. Electrical double layer model.4 Figure 3. Schematic of the CNT electrode preparation 6 Figure 4. (a) Optical image of flexible CNT electrodes, (b) Raman spectra and (c) TGA curves of the CNT electrode materials through the self-assembly technique..7 Figure 5. (a) Isotherm pore volume evaluation (b) Cumulative pore volume vs. pore sizes 8 Figure 6. Image analysis on porosity of shiny side CNT electrodes with (a) original image, (b) filtered image, and (c) void area distribution..9 Figure 7. Flow chart of PVA fabrication procedure 11 Figure 8. Scanning electron micrograph images of the polyvinyl alcohol hydrogel membrane (a) cross section, and (b) top view.13 Figure 9. (a) Nyquist plot of the PVA separator s ionic conductivity evaluation and (b) the corresponding fitted curve from 220K to 298K..13 Figure 10. (a) Schematic of a supercapacitor with PVA separator (pink), CNT electrodes (blue) and gold current collectors (yellow); (b) supercapacitor testing cell model; (c) optical image of real testing cell..16 Figure 11. (a) Cyclic voltammogram of the CNT supercapacitor cycled at 1 V/s and (b) the specific capacitance vs. the CV scan rate 17 Figure 12. (a) Electrochemical Nyquist and (b) Bode plots of the aqueous supercapacitor. 18 Figure 13. (a) Specific capacitance of aqueous CNT EDLC and (b) Imaginary capacitance of the cell as funcitons of frequency v

7 vi Figure 14. (a) Charge/discharge profile of the CNT supercapacitor at the current density of 10 A/g and (b) capacitance of the cell vs. current density..20 Figure 15. (a) Specific cell capacitance and (b) corresponding energy efficiency vs. cycles of charge and dsicharage at 10 A/g.21 Figure 16. EDX analysis of CNT electrode after PSSA process and surface oxidation..23 Figure 17. (a) Specific cell capacitance vs. the cycles at 10 A/g using the CNT supercapacitors made with oxidized and pristine CNT and (b) the corresponding energy efficiency.. 24 Figure 18. (a) Comparisons of the specific cell capacitance and (b) the time constants as a function of frequency 25 Figure 19. Schematic of an electrochemical capacitor using porous electrodes 28 Figure 20. (a) Discharge data and fitting curves of the CNT supercapacitor; (b) specific cell capacitance of the CNT supercapacitor vs. cycles.33 Figure 21. Discharge data and fitting curves of supercapacitors using (a) NORIT, (b) PFA Sphere, and (c) PFA PEG carbon electrodes..34 Figure 22. Evaluation of (a) capacitance per interfacial area, (b) interfacial area per unit volume, and (c) effective conductivity in electrode pores as a function of cycles 36 Figure 23. Discharge data and fitting curves of PFA PEG supercapacitors under various discharge current load.37 Figure 24. Percentage variation of interfacial area per unit volume, capacitance per interfacial area, and exchange current density vs. current loads.39

8 List of Tables Table 1. Components concentration in different stages of carbon nanotubes electrodes.23 Table 2. Physical properties of carbon electrode materials..32 Table 3. List of discharge fitting parameters of CNT supercapacitors at various cycles.33 Table 4. List of discharge fitting parameters for NORIT, PFA Sphere, and PFA PEG supercapacitors under different discharge cycles.35 Table 5. List of discharge fitting parameters of the PFA PEG supercapacitor under different current loads.38 Table 6. Voltage distribution due to various contributions on supercapacitors 39 Table 7. Comparison of theoretical and simulated properties values..40 vii

9 viii Acknowledgements If passion is the seed that produces diligence and confidence, then expanded knowledge and independent ideas are the fertile soil in which the seed must grow. I am grateful that the Pennsylvania State University has given me the fertile soil I need to expand my knowledge and ideas. I finished my first two years of undergraduate program at the Pennsylvania State University, Altoona Campus. After basic engineering courses, I became involved in undergraduate research and joined Dr. Kofi Adu s research group. Projects focused on magneticinduced nanoparticle assembly and MgB 2 -Carbon-Nanotube composite superconductors taught me how to conduct literature searches, and how to analyze and plot data in comprehensive and understandable ways. Those physics projects catalyzed my interest in energy, and I decided to discover applications for materials and technology in energy fields. In my junior year, I declared my major as Energy Engineering under the guidance from Dr. Sarma Pisupati, who is my Schreyer Honor s advisor. I joined Dr. Clive Randall s research group with the supervision from Dr. Ramakrishnan Rajagopalan. I had the opportunity to work on a National Science Foundationfunded project. The project s long-term vision is to develop a medical monitoring wristband coupled with energy harvesting and storage devices. That derived into my Schreyer Honors thesis. I took a leadership role in the development of flexible supercapacitors using carbon nanotubes electrodes. My research experience working with Dr. Adu, Dr. Rajagopalan, and Dr. Randall at Penn State has allowed me to experience all aspects of academic research. Throughout my undergraduate study at Penn State, many professors, staffs, and students have given me tremendous advice, support and encouragements.

10 ix I acknowledge my advisors and mentors: Dr. Sarma Pisupati, Dr. Kofi Adu, Dr. Ramakrishnan Rajagopalan, and Dr. Clive Randall. I acknowledge the financial supports from Penn State Undergraduate Discovery Summer Grant, NSF Center for Dielectric Studies and NSF Nanosystems Engineering Research Center for Advanced Self-Powered Systems of Integrated Sensors and Technologies. I would also like to acknowledge the supports from Penn State Altoona Undergraduate Research Program, and Materials Characterization Laboratory at Penn State University Park..

11 1 Chapter I. Introduction 1.1 Overview of Supercapacitors Supercapacitors are also called ultracapacitors or electrochemical capacitors. Many research studies have been conducted on supercapacitors due to their excellent cycle life (>10,000 cycles), high power output, and sufficient energy density. 1,2 Ragone plot, as shown in Figure 1, has been used to compare various energy conversion and storage systems. Supercapacitors occupy a large region with both relatively high specific power and energy performance. 3 Compared to conventional dielectric capacitors; supercapacitors maintain large power capability with increase in stored energy. Compared to traditional batteries, supercapacitors could collect and deliver sufficient amount of energy within a shorter time period. In other words, supercapacitors have a bridging function between conventional dielectric capacitors and rechargeable batteries, which provides a pathway to meeting the increasing demand of energy storage devices in the near future. 3 There are many existing applications of supercapacitors, including their usage in electronics, sensors, and large-scale renewable energy storage and dispatching systems. 4-7 Supercapacitors are believed to serve an important function in electric cars. In those applications, supercapacitors are usually combined with batteries to function as a high power storage component, which can store energy when braking and release energy when accelerating. 8

12 2 Figure 1. Ragone plot Energy Storage Mechanisms in Supercapacitors Supercapacitors have relatively simple energy storage principles compared to other electrochemical energy storage devices (fuel cells, rechargeable batteries, etc.) Supercapacitors,, can be divided, according to storage mechanisms, into electrical double layer capacitors (EDLCs) and pseudo-capacitors. 9 Electrostatic charge accumulation occurs at the interfaces between electrode and electrolyte in EDLCs. Reversible faradic reactions take place in pseudo-capacitors. These two energy storage mechanisms can be designed to function simultaneously in a supercapacitor eventually to achieve both improvements on energy and power performances.

13 EDLC Energy Storage Mechanism The concept of the electrical double layer was introduced by von Helmholtz in the 19 th century. 10 Helmholtz described a model that two layers of opposite charges accumulate at the interface between electrode and electrolyte. The separation between the layers is at an atomic level. 10 This model is analogical to the model of conventional capacitors. Dielectric capacitors normally store small amount of charge because of their limitations on charge accumulation areas (specifically on metal plate areas in two-plate conventional capacitors) and separation between two charged plates. Compared to the model of conventional capacitors, the electrical double layer model provides an approach to achieving larger charge accumulation areas, as known as interfacial areas, and reduced the distance of separation in the atomic scale. Gouy and Chapman further modified Helmholtz s electrical double layer model. They introduced a concept of diffuse layer taking into account of a continuous distribution of ions in the electrolyte. 11,12 Stern further combined the original Helmholtz s model and Gouy-Chapman s diffuse layer model to include two regions of ion distribution. 13 As shown in the Figure 2, the inner region is named as a compact layer. The electrolyte ions are adsorbed on the electrode/electrolyte interfacial areas in the compact layer. For example, on a positively charged electrode a layer of negative electrolyte ions is first formed in the compact layer. That layer is defined as the inner Helmholtz plane. Consequently, in the same configuration positive electrolyte ions are then formed into a second plane, which is named as the outer Helmholtz plane, after a compactly accumulated plane of negative electrolyte ions. Besides the compact layer, the diffuse layer exists in the electrical double layer model, which starts from the outer Helmholtz plane.

14 4 Figure 2. Electrical double layer model Pseudo-capacitor Energy Storage Mechanism Fast and reversible faradic reactions take place in pseudo-capacitors. 9 Capacitance generated in this type of mechanism is also known as pseudo-capacitance. Electrochemical active species are important to the pseudo-capacitors. These species could be metal oxides, or oxygen- or nitrogen-rich surface functional groups, such as manganese oxide, vanadium nitrides or ruthenium oxide Pseudo-capacitors could provide higher capacitance than EDLCs. However, reversible faradic reactions normally occur much slower than the electrostatic adsorption/desorption process. 9 Hence, the limited power performance and cycling stability become the drawbacks of pseudo-capacitors. EDLCs and pseudo-capacitors can be designed to function simultaneously in a supercapacitor to achieve both improvements on energy and power performances.

15 5 Chapter II Carbon Nanotube Electrode Fabrication Techniques 2.1 Overview of Carbon Nanotubes Electrodes in Supercapacitors Carbon nanotubes (CNTs) have attracted researchers attention for energy storage applications, such as fuel cells, batteries and supercapacitors, due to CNTs unique chemical and physical properties. 18 Some of these features are unique atomic structures, excellent electrical properties, good chemical and mechanical stability There are two kinds of CNTs, which are single-walled and multi-walled carbon nanotubes. Scientific investigations have been performed on both types of carbon nanotubes as electrode materials and carbon substrates. In this thesis project, single-walled carbon nanotubes were studied as electrode materials in supercapacitor applications. Various fabrication methods of carbon nanotubes-based electrodes have been invented in recent years. CNTs-based electrodes could be prepared by depositing electrophoretically on a current collector/substrate, or directly growth on metal foils, or inclusion of binders. 19, These approaches have their own advantages and disadvantages. For instance, the majority of binders that are used in electrode fabrications are electrically non-conductive, which would increase the overall resistance and suppress power performance and energy efficiency. 25 Some of recent innovative methods have been proposed to eliminate some of these drawbacks. CNT buckypaper electrodes were prepared by the filtration of surfactant-dispersed CNTs, in which the method eliminates the usage of binders in electrodes. 26 Liquid-induced collapse approach was invented to prepare electrodes with vertically-aligned carbon nanotubes. 19 In such approach, swelling became a challenge in those vertically-aligned CNT electrodes. 19

16 6 2.2 A Wet Chemistry Post-synthesis Self-assembly Technique A new technique was introduced in this thesis project, which was named as a wet chemistry post-synthesis self-assembly (PSSA) technique. 27,28 In this new technique, commercial CNTs (Thomas Swan Elicarb) with ~ 3.5 wt% of residual noncarbonaceous content were purchased and used for the electrode preparation. The commercial CNTs were well dispersed in a highly dense organic liquid solution (Cargille Heavy Liquid) under continuous mixing for 24 hours. The resultant mixture was poured into a glass petri dish. The liquid was thermally evaporated, which formed the carbon nanotubes-based electrodes. The electrodes were thermally treated at 900 C under helium environment to ensure the removal of liquid remnant. The chlorination treatment was applied to eliminate the impurities in the commercial CNTs, which reduced the impurity content to part per million levels. The hydrogenation treatment was applied after the chlorination to remove the residual chlorine in the CNTs-based electrodes. Figure 3. Schematic of the CNT electrode preparation

17 7 2.3 Characterizations of Self-assembled CNT Electrodes The self-assembled CNT electrodes were characterized using the thermogravimetric analyzer (TA Q5000IR), the Micrometric ASAP 2020, the micro-raman (Renishaw InVia), and the scanning electron microscopy (FEI NanoSEM). The characterizations cover the purity, specific surface area, and porosity of the self-assembled CNT electrodes Purity A flexible, binder-free, and thin CNT electrode, as shown in the figure 4, was prepared through this new self-assembly technique. The Raman spectra were performed on the CNTs before and after the self-assembly. The D-band intensity of the CNT decreased after the electrode preparation, which indicated the improvement of the carbon nanotubes quality. The TGA curves were also applied to confirm the quality improvement. The residual weight percentage of the non-carbonaceous impurities decreased to below the 0.1wt%. Meanwhile, the combustion temperature increased to 633 o C. Figure 4. (a) Optical image of flexible CNT electrodes, (b) Raman spectra and (c) TGA curves of the CNT electrode materials through the self-assembly technique

18 Surface Area The Brunauer, Emmett and Teller theory was used to evaluate the specific surface area of the carbon nanotubes-based electrode. Degassing of the sample was conducted under a vacuum of 10-7 torr at 300 C for overnight. The Barrett-Joyner-Holenda model was applied to obtain the pore volume, pore widths, and pore size distributions. A hysteresis was shown on the nitrogen adsorption and desorption profile. That profile indicated the existence of meso- and macro-pores. Broad distributions of both meso-pores and the majority of macropores were observed in the fabricated CNT-based electrode. The electrode has the specific surface area of 792 m 2 /g. Figure 5. (a) Isotherm pore volume evaluation (b) Cumulative pore volume vs. pore sizes

19 Porosity Two-dimensional void percentage was analyzed using the image analysis software, ImageJ. 29 SEM images were taken on the as-prepared CNT electrode. Original images were calibrated through defining the corresponding length scales. 29 The images were filtered using threshold adjustment into a black-white mode for clear distinction between void and solid areas on the images. The black regions corresponded to the void areas on the electrode. The individual black regions were evaluated and summarized. The percentage of the void areas for each image was obtained as the ratio between the total black areas and the overall image area. The porosity values were evaluated to be 16.0% and 19.4% for the carbon nanotubes-based electrode on the shiny side and dull side, respectively. Figure 6. Image analysis on porosity of shiny side CNT electrodes with (a) original image, (b) filtered image, and (c) void area distribution

20 10 Chapter III: Polyvinyl Alcohol Hydrogel Separator In a supercapacitor, a porous membrane is required to separate an anode and a cathode in order to avoid internal electrical short circuits. At the same time, interconnected pores in a membrane offer accessible channels for ionic transport in liquid electrolyte. 30 This type of porous membrane is known as a separator. Generally, a separator is sandwiched between the cathode and the anode in a supercapacitor. Various separators have been studied and fabricated to meet specific system requirements. Those separators could be nonwoven mat, polymeric membranes, and composite separators. 30,31 Following features are normally taken into account when selecting a separator 32 : Electronic insulation High ionic conductivity Easiness of handling Mechanical stability Electrolyte wettability Uniform thickness Chemical resistance to decomposition reactions caused by electrolyte and impure electrode materials In this thesis project, a polyvinyl alcohol (PVA) hydrogel separator was fabricated and applied in the supercapacitors.

21 Fabrication of the PVA separator The PVA solution was prepared by mixing 1g of polyvinyl alcohol powders (Sigma- Aldrich) with 10ml of deionized water. The solution was heated and continuously stirred under the hot plate temperature of 130 C (solution temperature of 90 C). The resultant solution was slowly cooled to 25 C. To crosslink the polymer, the solution was mixed with 4μL of glutaraldehyde and 50% water solution (Sigma-Aldrich) for 5 minutes. The solution was casted on a glass plate using a doctor blade to achieve a uniform thickness of 75μm. After 12 hours, the polymer membrane was lifted off glass plate to be vacuum-dried overnight for 12 hours. Figure 7. Flow chart of PVA fabrication procedure

22 Characterizations of the PVA separator The PVA separator was characterized through scanning electron microscope (Hitachi S- 3500N) to confirm the uniform morphology. The separator was evaluated for the ionic conductivity as a function of temperature using electrochemical impedance spectroscopy. The PVA separators were soaked in 1M H 2 SO 4 for 24 hours before being the measurements. Gold foils were used as current collectors. The active area of the PVA separator was around 44 mm 2 with a thickness of 100µm as determined from the cross-sectional SEM images of the separator. The testing cell was placed in a furnace (Delta 9023). Electrochemical impedance data were collected at various temperatures from 25 C to -65 C with the temperature interval of 15 C. Ionic conductivity of the PVA separator at each temperature was calculated using the equation 1: d (1) R A Where is ionic conductivity in S/cm, d is the separator thickness in cm, A is the active area in cm 2, and R is the separator resistance in ohms. The SEM images of the fabricated PVA separator were shown in the figure 8. The ionic conductivity data were collected using electrochemical impedance measurements. The room temperature ionic conductivity of the PVA separator was 45mS/cm. The conductivity was largely reduced when the temperature reached below -25 C. The temperature dependence of ionic conductivity obeyed the Vogel Tamman-Fulcher expression. 26 B 0 exp T T 0 (2) As generated from the fitting results, the fitting parameters values were 0, B, and T 0 were 0.06 S/cm, K and K, respectively.

23 13 Figure 8. Scanning electron micrograph images of the polyvinyl alcohol hydrogel membrane (a) cross section, and (b) top view Figure 9. (a) Nyquist plot of the PVA separator s ionic conductivity evaluation and (b) the corresponding fitted curve from 220K to 298K

24 14 Chapter IV Electrochemical Characterization Techniques In this chapter, different electrochemical characterization techniques were introduced. Two-electrode supercapacitor characterizations were conducted using an electrochemical potential station (Gamry Reference 3000). The measurements included cyclic voltammetry (CV), constant current charge and discharge, and electrochemical impedance spectroscopy (EIS). Supercapacitors performance parameters, including specific capacitance, energy and columbic efficiencies, cycling stability, and power density, were introduced and expressed using equations. The results of supercapacitors characterizations are discussed in chapter Cyclic Voltammetry Cyclic voltammetry measures the current in corresponding to the change of potential between two electrodes in a supercapacitor. In the thesis project, the cyclic voltammetry was performed at the voltage window between 0V and 1V at scan rates from 10mV/s to 1000mV/s. Specific cell capacitance was obtained from the CV profiles through the equation 3: I C (3) m Where I is the current in A, is the scan rate in V/s, and m is mass of both the electrodes in g. 4.2 Electrochemical Impedance Spectroscopy Electrochemical Impedance Spectroscopy (EIS) is known as the AC impedance method. EIS is a common technique in chemistry and electrochemical studies. The EIS is usually applied in on EDLC, electrode characterization, or complex interfaces In this thesis project, the EIS measured the response of a supercapacitor when applied a periodic AC signal. A broad frequency range of AC signal was applied and the impedance data were collected at each

25 15 frequency. That is the reason why this type of measurement is known as impedance spectroscopy. To perform the EIS characterization of the supercapacitors, an AC 10mV perturbation was applied at the open circuit condition (around 0V). The AC perturbation was scanned from 10 5 Hz to 10-3 Hz. The equivalent series resistance (ESR) of a supercapacitor was obtained from the EIS data at 1 khz. The cell power density could be calculated using the following equation: 2 V P (4) 4Rm Where V is the cell voltage in V, R is the ESR in ohm, and m is the total mass of both the electrodes in g. 4.3 Constant Current Charge and Discharge Cycling stability of the supercapacitor was assessed by applying a constant current load to change and discharge the supercapacitor for 10,000 times. The applied current densities varied from 1A/g to 50A/g. The specific capacitance of the cell was calculated based on the charge and discharge cycles: I t C (5) V m Where I is the applied current in A, t is discharge time in s, V is the potential decrease during discharge in V, and m is the mass of both the electrodes in g. Energy efficiency was obtained as the ratio between the amount of the energy discharged to that of the energy charged as expressed in equation 6: E eff IVd t IVd t d ischarg e ch arg e (6)

26 16 Chapter V Electrochemical Performance of Aqueous EDLC 5.1 Assembly of the aqueous EDLC The carbon nanotubes-based electrodes were punched into disc electrodes with an area of 0.4cm 2, and a mass of 0.3mg for the individual electrode. Gold foils were applied as the current collectors. The electrodes were immersed in the electrolyte 30 minutes before the cell assembly in order to ensure a complete wetting of the electrodes. Similarly, the PVA separator was immersed in the electrolyte 2 hours before the cell assembly to ensure the saturation of the electrolyte in the separator. The PVA separator was sandwiched between the electrodes with the gold current collectors on both sides, as illustrated in the figure 10. Figure 10. (a) Schematic of a supercapacitor with PVA separator (pink), CNT electrodes (blue) and gold current collectors (yellow); (b) supercapacitor testing cell model; (c) optical image of real testing supercapacitor cell

27 Cyclic Voltammetry Results The cyclic voltammetry of the aqueous EDLC was measured and plotted in the figure 11. The CV curve exhibited a rectangular shape, which was indicative to a capacitive response at a 1V/s. As described in the equation 3, the specific cell capacitance was calculated at each CV scan rate. The specific cell capacitance was above 10F/g and decreased slightly with the scan rate. That indicated an excellent power stability of the supercapacitor. Figure 11. (a) Cyclic voltammogram of the CNT supercapacitor cycled at 1 V/s and (b) the specific capacitance vs. the CV scan rate 5.3 Electrochemical Impedance Spectroscopy Results In the Nyquist plot, a semicircle was observed at high frequencies followed by a Warburg behavior. That was attributed to the ionic diffusion. The capacitive behavior was shown as a steep line at the low frequencies. The knee frequency was defined as a transitional point beyond which the capacitive contribution ensued. The knee frequency in the aqueous EDLC was 1 khz.

28 18 That value was higher than the reported values of the most carbon nanotube based supercapacitors in the literature Carbon nanotubes-based supercapacitors reported in the literatures showed a knee frequency in the range from 100Hz to 300 Hz while the knee frequency of activated carbon electrodes-based conventional capacitors was lower than 10 Hz The knee frequency at around 1kHz was reported using the electrophoretically deposited carbon nanotubes electrodes. 22 In this thesis project, high knee frequencies in the supercapacitor were achieved using binder-free carbon nanotubes-based prepared from the PSSA method. The maximum power density of the supercapacitor was obtained to be 1040kW/kg. Figure 12. (a) Electrochemical Nyquist and (b) Bode plots of the aqueous supercapacitor

29 19 Figure 13. (a) Specific capacitance of aqueous CNT EDLC and (b) Imaginary capacitance of the cell as funcitons of frequency 5.4 Constant Current Charge and Discharge Results of the Aqueous EDLC The supercapacitor was charged and discharged at 10A/g. A typical saw-tooth profile was observed with non-observable IR drops in the figure 14. The stability of the supercapacitor was studied upto 50A/g with slight fadings in the specific cell capacitance.

30 20 Figure 14. (a) Charge/discharge profile of the CNT supercapacitor at the current density of 10 A/g and (b) capacitance of the cell vs. current density The cycling stability of the aqueous EDLC was evaluated in 1M and 3M aqueous sulfuric acid electrolytes as shown in the figure 15. The EDLC performed better in 1M sulfuric acid electrolyte. The specific cell capacitance was calculated to be 10 F/g and stable for 10,000 cycles. The columbic efficiencies were 100% and the energy efficiencies was higher than 90% for both conditions. The supercapacitor in 1M H2SO4 obtained a higher energy efficiency of 94%, which was attributed to the low ohmic drops as observed in the charge and discharge curves.

31 21 Figure 15. (a) Specific cell capacitance and (b) corresponding energy efficiency vs. cycles of charge and dsicharage at 10 A/g

32 22 Chapter VI Aqueous Pseudocapacitors Aqueous EDLCs were assembled and evaluated in the chapter 5, which demonstrated excellent power performance and cycling stability over 10,000 cycles. To further improve the specific cell capacitance, by which would increase the energy density of the cell, pseudocapacitance was introduced through decorating oxygen-rich functional groups on carbon nanotubes-based electrodes. Since both the EDLC and the pseudo-capacitor mechanisms were applied, this was defined as an aqueous hybrid supercapacitor. 6.1 Surface oxidation of CNT electrodes The self-assembled CNT electrodes were surface oxidized using a 0.002M potassium permanganate (KMnO 4 ) solution (Alfa Aesar). The pristine carbon nanotubes-based electrodes were immersed in the mild KMnO 4 solution at the solution temperature of 60 C for 30 minutes. The oxidized electrodes were washed thoroughly wish deionized water and then soaked in 1M H 2 SO 4 for 4 hours to remove manganese residues from the surface oxidized carbon nanotubesbased electrodes. The treated electrodes were washed with copious deionized water and dried at 100 C for overnight. The energy dispersive x-ray (EDX) spectroscopy was applied and coupled with SEM measurements in order to quantify the elemental components in carbon nanotubes-based electrodes. As indicated in the figure 16, the as-received carbon nanotubes were purified using the PSSA technique by which the oxygen, iron and other impurities were eliminated with only small residue of chlorine and bromine in the pristine carbon nanotubes-based electrodes. The purity could be improved with better controlled reaction conditions including temperature, treating duration, and gases concentrations. After the process of the surface oxidization, the

33 23 oxygen content increased to 8.9 weight percent in the surface oxidized carbon nanotubes-based electrodes. The surface oxidized electrodes were believed to induce pseudo-capacitance coupling with original EDLC mechanisms. The elemental weight percentages for each type of samples were listed in the table 1. Table 1. Components concentration in different stages of carbon nanotubes electrodes CNT C (wt%) O (wt%) Fe (wt%) Cl (wt%) Br (wt%) as-received Purified Oxidized oxidation Figure 16. EDX analysis of carbon nanotube electrode after PSSA process and surface

34 Assembly of the Aqueous Pseudocapacitors The aqueous hybrid supercapacitors were assembled using the same configuration as the aqueous EDLC. Only the electrodes are replaced with surface functionalized electrodes. 6.3 Constant Current Charge and Discharge of the Aqueous Hybrid Supercapacitors To enhance the specific cell capacitance of the carbon nanotubes-based supercapacitor, the electrodes were surface oxidized using acidified potassium permanganate solution Surface oxygen-rich functional groups were reported to cause reversible faradaic reactions, which could induce pseudocapacitance in aqueous acids. 42,43 The surface oxidation of the carbon nanotubes increased the specific capacitance of the supercapacitor from 10F/g to 18F/g at 10A/g. The combination of EDLC and pseudocapacitive contributions was stable up to 10,000 cycles as shown in the cycling study. Figure 17. (a) Specific cell capacitance vs. the cycles at 10 A/g using the CNT supercapacitors made with oxidized and pristine CNT and (b) the corresponding energy efficiency

35 Electrochemical Impedance Spectroscopy Results in the Pseudocapacitors Surface oxygen-rich functional groups have been studied and believed to improve the wettability of the electrolyte on the electrodes. That was reflected on the lower ESR of the supercapacitor using the surface oxidized electrodes. Reversible faradaic reactions on the electrode/electrolyte interface increased the charge-transfer resistance. Hence, the relaxation frequency of the supercapacitor using surface oxidized electrodes decreased from 40 Hz to 25 Hz. The specific cell capacitance at low frequencies was enhanced to 23F/g due to the inclusion of pseudocapative materials on carbon nanotubes-based electrodes. Figure 18. (a) Comparisons of the specific cell capacitance and (b) the time constants as a function of frequency

36 26 Chapter VII: Physicochemical Mathematical Model on Discharge Behaviors of Supercapacitors 7.1 Introduction on Mathematical Models of Supercapacitors In order to precisely understand contributions of components in an electrical double layer capacitor on its electrochemical behaviors, several mathematical models have already been developed into two categories: equivalent circuit models and physics models. In an equivalent circuit model, a resistor-capacitor circuit is normally used to interpret the electrical behavior of an EDLC. 44,45 However, the equivalent circuit approach has limitations when to understand complex physical elements in an EDLC system. It is because that parameter values in equivalent circuits do not directly indicate any physical meanings. The usage of equivalent circuit models therefore provides limited information on the improvement of a supercapacitor. Physics based models, on the other hand, are based on conservation and diffusion equations. A system of partial differential equations are normally applied to describe fundamental principles and to solve within constraint conditions. Srinivasan and Weidner have developed a porous-electrode method using a numerical approach to simulate electrochemical behaviors of EDLCs. 46 Such models are more popular than equivalent circuit models due to their abilities on catching the dynamics in a supercapacitor. Although the physics based models have been successful on capturing the energy storage mechanism in EDLC using physical parameters, parasitic reactions involving charge- and mass-transfer characteristics have not been taken into account. Those parasitic reactions often exist in real devices due to electrodes decomposition reactions, membrane degradation and other causes In order to comprehensively model the performance of an EDLC, mass- and charge-transfer controlled terms for the parasitic reactions

37 27 could be integrated into the porous electrode model to better map the voltage distributions in a supercapacitor under the constant current charge and discharge. In this chapter, a physicochemical mathematical model was developed to analyze the discharge behavior of supercapacitors taking into account of electric double layer formation, charge- and mass-transfer limitations, ionic resistance of the separator, and resistive contacts between the electrodes and current collectors. The model was applied and verified in various electrochemical capacitor systems. 7.2 Model Development A supercapacitor was consisted of current collectors, porous electrodes, and separateor. In the porous electrodes, pores in different sizes, including macro, meso, and micropores, were considered in the model. Four major contributions were concerned: electric double-layer formation, ionic resistance in the separator, decomposition reactions on the electrodes, and contact resistance on the electrode-current-collector interfaces. The mathematical expression describing the double layer contributions in the porous supercapacitor was adapted from the model developed by Srinivasan and Weidner in The following assumptions were made in the model: 1. uniform electrolyte concentration throughout the system. 2. physical properties of the electrochemical capacitor, such as effective conductivities and capacitance, do not change during a discharge.

38 28 Figure 19. Schematic of an electrochemical capacitor using porous electrodes The double layer formation on porous carbon electrode was analyzed first using the mathematical model developed by Venkat Srinivasan and John W. Weidner. 53 The current densities in the solid phase and solution phase can be described using Ohm s law in equations 7 and 8. Conservation of charge was applied to get the differential expression of total charge transfer in equation 9. The interfacial current density was also described in the term of interfacial potential difference. The equation 10 was nonlinear and converted into a dimensionless form (equation 11). It was solved using Laplace equations and boundary conditions to reach the description of the voltage drop across double layers and the separator under the constant current discharge as expressed in the equation 12. Details on the mathematical derivations on the contribution of double layer formations could be found in paper authored by Venkat Srinivasan and John W. Weidner. 53

39 29 (7) (8) (9) (10) (11) (12) Where a is volumetric interfacial area in cm 2 /cm 3 ; C is capacitance per interfacial area in F/cm 2 ; I is applied cell current density in A/cm 2 ; I* is dimensionless cell current; is current density in the solid phase of the porous electrode in A/cm 2 ; is current density in the solution phase of the porous electrode in A/cm 2 ; is current per interfacial area in A/cm 2 ; i o is exchange current on the cathode in A/cm 2 ; t is time in s; V* is dimensionless cell voltage; V cell is cell voltage in V; V o is initial cell voltage in V; x is distance in cm; is ratio of separator to electrode resistance; is solid phase potential in V; is solution phase potential in V; is ratio of solution to solid phase conductivity; is overpotential; κ is effective conductivity of the

40 electrolyte in the electrode in S/cm; κ s is effective conductivity of the electrolyte in the separator 30 in S/cm; σ is solid phase conductivity in S/cm; τ is dimensionless time; is dimensionless distance. The resistive contact was considered at the electrode/current-collector interfaces. Potential drop was expressed in terms of interfacial conductivity, current density, and electrode thickness. (13) Where is the voltage drop due to contact resistance; i is applied current density in A/cm 2 ; L is electrode thickness in cm; V o is initial cell voltage in V; к i is contact interfacial conductivity in S/cm. The decomposition reaction under the constant current charge and discharge was also taken into account. Voltage drop due to decomposition reactions was described in terms of charge- and mass-transfer resistances. For instance, when the charge-transfer resistance is much lower than the mass-transfer resistance, the mass-transfer will be the limiting factor. An equation of electrode reaction also obeys the form of Tafel equation. (14) Where is the voltage drop due to the decomposition reactions; i is current density in A/cm 2 ; R ct is charge-transfer resistance in Ohm; R mt,c is mass-transfer resistance on the cathode in Ohm; R mt,a is mass-transfer resistance on the anode in Ohm;

41 31 The equations were modified to account for the net parasitic reactions on the electrodes. As indicated in the equation, if the exchange current is much smaller than the limiting current, charge transfer will become the limiting factor. The terms describing decomposition reactions on electrodes were included into the voltage drop equation for the overall system. (15) Where is the voltage drop due to the decomposition reactions; F is the Faradays constant in C/equiv; i is cell current density in A/cm 2 ; i o is exchange current in A/cm 2 ; i l is limiting current in A/cm 2 ; L is electrode thickness in cm; Ls is separator thickness in cm; R is gas constant in J/k/mol; T is the cell temperature in K; t is time in s; is transfer coefficient in 0.5; The mathematical model was developed combining the voltage drops across the doublelayer formation, the separator, decomposition reactions and resistive contacts. (16) 7.3 Simulation Results Various carbon electrodes were assembled and tested to provide discharge data for the mathematical model. The four types of the carbon electrode materials were CNTs, NORIT activated carbons, PFA-derived spherical carbons, and PFA-PEG derived carbons. The physical properties of these carbons were measured to provide the practical information for the model.

42 32 Table 2. Physical properties of carbon electrode materials Sample Micropore volume (cc/g) Mesopore volume (cc/g) Macropore volume (cc/g) Total pore volume (cc/g) Specific Surface Area (m^2/g) CNT / NORIT 0.75 / / PFA Sphere 0.45 / / PFA PEG8K / Contributions from Double Layer Formation and Separator The aqueous supercapacitor using carbon nanotubes electrodes were demonstrated as a pure electrical double layer capacitor. The model with only double layer contribution and ionic resistance of the separator was applied to fit the real discharge data of the aqueous carbon nanotubes-based EDLC. The EDLC s performances were reported previously, which showed a cycling stability for 10,000 cycles of charge and discharge. As shown on the values of the fitting parameters, the effective conductivity in the separator was stable over the charge/discharge cycles. After the 10,000 cycles of charge/discharge, the effective conductivity in the pores of the carbon nanotubes electrodes increased slightly. That might indicate a slight improvement on the wettability and effective pore volumes. That was confirmed by the increased values of the interfacial area (a) and capacitance per interfacial area (C). In this case, the decomposition reactions did not occur in the EDLCs.

43 33 Figure 20. (a) Discharge data and fitting curves of the CNT supercapacitor; (b) specific cell capacitance of the CNT supercapacitor vs. cycles Table 3. List of discharge fitting parameters of CNT supercapacitors at various cycles CNT EDLC Ks (S/cm) K (S/cm) σ (S/cm) a (cm^2/cm^3) c (F/cm^2) Cycle E E E E E-06 Cycle E E E E E Contributions from Double Layer Formation, Ionic Resistive Separator, Decomposition Reactions, and Resistive Contact Electrical double layer capacitors using activated carbon electrodes in organic electrolyte could have decompositions reactions of electrodes or electrolytes due to the overcharge potential, kinetics and properties of the electrode materials. As reported in the literatures on three carbon electrode materials, NORIT carbon, PFA derived spherical carbon and PFA-PEG derived carbons had different pore size, pore volume distribution, conductivity and other physical properties. 49,50 The mathematical model including double layer formation, decompositions

44 34 reactions, ionic resistive separator, and resistive contact was applied for the discharge simulation of the supercapacitors using different types of carbon electrodes. Each supercapacitor was evaluated using the model for 1500 cycles of constant current charge and discharge at 200mA/g. The values of the physical properties were obtained as a function of cycles. The discharge curves were well fitted using the model as shown in the figure 22. Figure 21. Discharge data and fitting curves of supercapacitors using (a) NORIT, (b) PFA Sphere, and (c) PFA PEG carbon electrodes

45 35 Table 4. List of discharge fitting parameters for NORIT, PFA Sphere, and PFA PEG supercapacitors under different discharge cycles NORIT Cycle # K s K Zegma a C i o i l K i E E E E E E E E E E E E E E E E PFA Sphere K s K Zegma a C i o i l K i Cycle # E E E E E E E E E E E E E E E E PFA PEG K Cycle # s K Zegma a C i o i l K i E E E E E E E E E E E E E E E E The parameters values were generated from the curve fittings using the mathematical model. All three supercapacitors showed the same trends of the terms including capacitance per interfacial area (C), interfacial area (a), and the effective conductivity in the electrode pores (K). The capacitance and interfacial area terms slightly decreased as a function of cycles in the supercapacitors using PFA derived spherical carbon and PFA-PEG derived carbon electrodes. The supercapacitor using NORIT activated carbon electrode was shown larger decreases in the terms of capacitance per interfacial area, interfacial area, and the effective conductivity in the electrode pores as a function of cycles. This was attributed to the decompositions reactions of the electrolyte when the NORIT supercapacitor was charged and discharged between 0V and 3V.

46 36 The possible decomposition reactions have been investigated and reported in the literatures on the NORIT carbon electrodes. 49 Compared to NORIT derived carbon electrode materials, PFAderived spherical carbon and PFA-PEG derived carbon materials were demonstrated with better cycling stability for 1500 cycles of charge and discharge. Figure 22. Evaluation of (a) capacitance per interfacial area, (b) interfacial area per unit volume, and (c) effective conductivity in electrode pores as a function of cycles

47 Parameters Variation as a Function of Current Density The supercapacitor using PFA-PEG derived carbon electrodes was fitted using the model for its discharge behaviors as a function of constant current density from 200mA/g to 1000mA/g. The discharge data were well fitted as shown in the figure 24. Fitting parameters values were produced from the model to understand the change of properties in the system. Figure 23. Discharge data and fitting curves of PFA PEG supercapacitors under various discharge current load As listed in the table of the parameters values of the supercapacitor using PFA-PEG derived carbon electrodes, certain terms varied as a function of current density. The constant current charge and discharge experiment showed 10% capacitance decrease as the current density increased from 200mA/g to 1000mA/g. The effective conductivity in the separator, electrode pores, matrix phase conductivity and contact conductivity remained constant over the

48 38 range of applied current densities. The exchange current density was one order of magnitude lower than the limiting current density in the decomposition reactions. This indicated the case that the decomposition reaction was not mass-transfer limited. The exchange current density of the decomposition reactions increased and approached to the limiting current density as the discharge current increased. In that case, the system became more mass transfer limited; in other words, the decomposition reaction became less favored. The increase in the exchange current led to the decrease in the voltage drop due to the decomposition reaction. The decomposition reaction was suppressed with the increase in the applied current density. Table 5. List of discharge fitting parameters of the PFA PEG supercapacitor under different current loads current load K (ma/g) s K Zegma a C i o i l K i E E E E E E E E E E E E E E E E E E E E