DOCTORAL DISSERTATION Design of high energy hybrid capacitors and ageing analysis by electrochemical on-line mass spectrometry Patryk Przygocki

Patryk Przygocki

Design of high energy hy brid capacitors and ageing analysis by electrochemical

on-line mass spectrometry

P r o j e k t o w a n i e w y s o k o e n e r g e t y c z n y c h k o n d e n s a t o r ó w h y b r y d o w y c h o r a z a n a l i z a p r o c e s ó w s t a r z e n i a z u ż y c i e m e l e k t r o c h e m i c z n e j

s p e k t r o m e t r i i m a s w c z a s i e r z e c z y w i s t y m

DOCTORAL DISSERTATION

Pro moter:

prof. François Béguin

A ss i st i ng Pr o mo t e r:

d r Qa ma r Abb a s

P o zna ń 2018

F

C

T

I

F

: C

T

Patr yk Prz ygocki 2 Badania do niniejszej pracy prowadzone były przy wsparciu projektu OPUS UMO-2014/15/B/ST4/04957 finansowanego przez Narodowe Centrum Nauki (NCN) w Polsce.

Kierownik projektu: Profesor François Béguin

This thesis research was supported by the OPUS project UMO-2014/15/B/ST4/04957 funded by the National Science Centre in Poland.

Project leader: Professor François Béguin

Patr yk Prz ygocki 3 Badania do niniejszej pracy były częściowo prowadzone przy wsparciu przez projekt ECOLCAP realizowanego w ramach Programu Welcome, finansowanego przez Fundację Nauki Polskiej (FNP) zgodnie z Działaniem 1.2. „Wzmocnienie potencjału kadrowego nauki”, Programu Operacyjnego Innowacyjna Gospodarka wspieranego przez Unię Europejską.

Kierownik projektu: Profesor François Béguin

A part of this research was supported by the ECOLCAP project funded in the frame of the Welcome Programme implemented by the Foundation for Polish Science (FNP) within the Measure 1.2. ‘Strengthening the human resources potential of science’, of the Innovative Economy Operational Programme supported by European Union.

Project leader: Professor François Béguin

Patr yk Prz ygocki 4 I am grateful to the promoter of this thesis - Prof. François Béguin for unmeasured time and invaluable guidance he has put into this work.

Merci!

My gratitude is dedicated to Prof. Elżbieta Frąckowiak for encouraging me to develop the knowledge in electrochemistry.

Dziękuję!

I am thankful to dr Qamar Abbas for his help and full commitment to my PhD work.

Thank you!

I would like to express my gratitude to dr Paula Ratajczak for her support and contribution into work related with the ageing analysis of electrochemical capacitors.

Dziękuję!

I would like to also acknowledge all the present and former members

of the Power Sources Group at the Poznan University of Technology.

Patr yk Prz ygocki 5

Table of Contents

General introduction ... 8

Chapter I... 16

Literature review 1. Introduction ... 17

2. Electrochemical capacitors ... 17

2.1 Models of the electrical double-layer ... 17

2.2 Electrical double-layer capacitors ... 20

3. Activated carbon electrodes for electrochemical capacitors ... 23

3.1 Porous texture of activated carbons ... 24

3.2 Effect of porous texture of activated carbons on capacitance ... 30

3.3 Surface functionality of carbons ... 32

3.4 Oxidation of activated carbons ... 35

4. Classification of electrochemical capacitors in aqueous electrolytes ... 37

4.1 Electrochemical capacitors with symmetric carbon electrodes ... 38

4.2 Electrochemical capacitors with asymmetric electrodes ... 39

4.3 Hybrid electrochemical capacitors in aqueous electrolytes ... 40

5. Aqueous electrolytes for ECs based on carbon electrodes ... 43

5.1 High voltage ECs using neutral aqueous electrolytes ... 43

5.2 Redox-active aqueous electrolytes ... 46

6. Analysis of electrochemical capacitors ageing in aqueous electrolytes ... 51

7. Conclusion and perspectives ... 55

Chapter II ... 56

Appropriate methods for evaluating the efficiency and capacitive behavior of different types of supercapacitors 1. Summary of the publication ... 57

2. Publication ... 59

Patr yk Prz ygocki 6

Chapter III………..…60

Confinement of iodides in carbon porosity to prevent from positive electrode oxidation in high voltage aqueous hybrid electrochemical capacitors 1. Summary of the publication ... 61

2. Publication ... 64

Chapter IV ... 65

Capacitance enhancement of hybrid electrochemical capacitor with asymmetric carbon electrodes configuration in neutral aqueous electrolyte 1. Summary of the publication ... 66

2. Publication ... 69

Chapter V ... 70

Environmentally friendly and cost-effective aqueous electrolyte for high-energy electrochemical capacitor with asymmetric carbon electrodes operating at sub-ambient temperature 1. Introduction ... 71

2. Conductivity and thermal behaviour of aqueous solutions based on choline salts ... 72

3. Performance of the symmetric and hybrid capacitors at room temperature ... 74

4. Electrochemical performance of symmetric and asymmetric hybrid pouch cells at low temperature ... 79

5. Conclusions ... 85

6. Experimental ... 85

6.1 Preparation and properties of the aqueous electrolytes ... 85

6.2 Manufacturing of electrodes and cells ... 86

6.3 Electrochemical investigations of electrodes and cells ... 87

Chapter VI ... 89

Self-consistent approach based on using electrochemical mass spectrometry (EMS) to analyze the degradation phenomena taking place during ageing of electrochemical capacitors in aqueous electrolyte 1. Introduction ... 90

2. Identification of the reactions contributing to the leakage current profile during ECs ageing ... 91

3. Qualitative and quantitative analysis of the reactions taking place during ECs ageing .... 94

Patr yk Prz ygocki 7 4. Analysis of the leakage current resulting of a charge distribution mechanism during ECs

ageing ... 100

5. Calculation of the residual charge spent for other parasitic reactions taking into account all the already analyzed contributions ... 102

6. Conclusions ... 104

7. Experimental ... 105

7.1 Electrodes, electrolyte and cell assembly ... 105

7.2 EMS measurements ... 106

7.3 Post-mortem analysis of electrodes by TPD ... 107

General conclusion... 108

Symbols and abbreviations ... 112

References ... 117

Scientific achievements ... 126

Abstract ... 131

Streszczenie... 136

Co-authorship statements ... 141

Patr yk Prz ygocki 8

General introduction

Patr yk Prz ygocki 9 The industrial development of western countries in the second half of the twentieth century, followed later by the development of China and other countries, is at the origin of a global rise of fossil fuels consumption and green-house gases emissions. Considering the fact that these fuels draw on finite resources which will be eventually emptied, and taking into account all the ecological issues related with global warming and air pollution, it is a duty of humans’ population to reduce dramatically their usage, either by improving the efficiency of energy production systems and/or by introducing renewables (sun, wind, water, …) in the energy mix.

One possible concept to improve energy efficiency is based on collecting unused or spoiled energy (heat, mechanical) and to use it when needed; for example, in a vehicle, the energy harvested during braking can be stored in a battery, and then used in an electric motor during the acceleration steps of the vehicle. Similarly, the strong dependence of renewables on the geographical location, climate and day/night cycles imposes the concept of Smart grid for an optimized utilization of these energy resources. Here again, energy storage systems are unavoidable to store the energy when it is produced and not needed, and to deliver it when it is demanded. Hence, energy storage appears as an unavoidable solution to improve energy management, and by this way reduce the dependence to fossil fuels.

Energy storage systems can be classified into two main categories: thermal (TES) and electrical

(EES). The main TES technologies include heat storage (energy is stored through the temperature

difference of storage means), latent storage (the phase of the thermal medium is changed during

the energy storage process) and thermochemical energy storage (the collected heat is used to

initiate a reversible endothermic reaction). Although TES is important for the integration of

renewable energy sources like sunlight and geothermal heat, it is still required to reduce heat losses

in order to enable its commercial application. Besides, EES includes all the technologies which

accept and return the stored energy as electric power. In general, it can be divided into mechanical

energy storage systems (pumped-hydro storage- PHS, compressed air energy storage – CAES,

flywheel energy storage – FES), superconducting magnetic energy storage (SMES), and

electrochemical energy storage (EES). For large scale energy storage, the most mature and widely

used technology is pumped hydro storage (PHS). It is utilized to store excess energy and supply

high peak demands by pumping water from a lower reservoir to an upper one and using the gravity

force from the upper to the lower reservoir to activate turbines which generate electricity. Although

Patr yk Prz ygocki 10 there are more than 300 PHS installations around the world, the scarcity and strong topographical dependence of available areas for two large reservoirs, together with the high construction cost makes PHS not a perfect energy storage system. Compressed air energy storage (CAES) is another large scale EES system in which energy is stored by compressing air in an underground cavity.

Unfortunately, there are only 2 plants constructed in the world, hence this technology is considered to be under research now due to the difficulty to find adapted locations. Flywheel energy storage (FES), which works by accelerating a rotor (flywheel) to a very high speed and maintaining the energy in the system as rotational energy, is another approach developed for energy storage.

However, the presence of friction forces significantly lowers their efficiency (even down to 45%

after 2 days) and increases their stand-by losses what makes them not suitable for long term storage. Besides, in superconducting magnetic energy storage (SMES), the energy is stored in the magnetic field created by a direct current using a superconducting coil cryogenically cooled below its superconducting critical temperature. The advantages of this system include fast response time (from 1 to 5 ms), high energy storage efficiency (more than 95%) and life time reaching around 30 years. However, due to high processing cost of SMES devices, this concept is still under research development, but mentioned advantages make this kind of energy storage system promising for the integrated application to smart grid.

Accumulators (currently called batteries) and electrochemical capacitors (ECs) belong to

electrochemical energy storage (EES) systems, which is a group of devices knowing recently a

significant development. Energy storage in batteries (Lead-acid, Ni-Cd, Ni-MH, lithium-ion

batteries (LIBs) …) is based on diffusion-controlled redox reactions occurring within electrode

materials. As a consequence of this charge storage mechanism, the energy density of LIBs can

reach high values (up to 240 Wh kg

), yet the power density is relatively low (not higher than 3

kW kg

) and the lifespan short (usually not more than a few thousand cycles). Small LIBs are

successfully applied in mobile phones, laptops, cameras, whereas bigger ones are used also in

traction applications for electric and hybrid vehicles as well as for standby power. Besides,

electrical double-layer capacitors are devices with high capacitance, low internal resistance and

consequently ultrahigh power density (up to 40 kW kg

), and extremely long lifespan. Such good

properties are owing to the electrostatic accumulation of ions, without any electronic transfer, at

the surface of electrodes. ECs are generally applied when high power is needed for short periods

Patr yk Prz ygocki 11 of time, e.g., forklifts, harbour cranes, security systems in aircrafts or portable devices (such as laptops and smartphones). They are also utilized for kinetic energy recovery system (KERS, also called regenerative braking) in automobiles, buses and tramways, providing then power impulses for acceleration, enabling by this way a reduction of fuel consumption. However, the relatively low energy density (ca. ten times lower than in batteries) and high cost per kWh of EDLCs (electrical double-layer capacitors e.g., 5 times higher than lead-acid batteries) are significant disadvantages of ECs which limit the spectrum of their applications.

The energy density of ECs, expressed by the equation E = ½ CU

, is determined by the operating voltage U (which depends on the stability window of the electrolyte) and the capacitance C (which is mainly controlled by the electrode material). Therefore, high specific surface area (up to 2000 m

g

) activated carbon (AC) electrodes and organic electrolytes (allowing cell voltage up to 2.7- 2.8 V to be reached) are the most popular components of commercially available ECs. However, due to the high toxicity and flammability of organic solvents (e.g., acetonitrile) and high cost of cells production, requiring assembling under inert atmosphere and extensive drying of electrodes, more eco-friendly technologies are strongly desirable. Therefore, nowadays important research efforts are focused on enhancing ECs energy density, while utilizing sustainable and, what is crucial, cheap components in order to realize environmentally friendly systems.

Considering the aforementioned formula of energy density, in the recent years, the interest for aqueous solutions in AC/AC electrochemical capacitors has been revived by using i) neutral salts (Li

SO

, Na

SO

, K

SO

, …) enabling to enhance the voltage U up to 1.5 – 1.6 V, owing to a high overpotential of water reduction in the porosity of the negative AC electrode; ii) redox electrolytes (e.g., KI) enabling to enhance C, owing to the hybridization between a battery-type positive electrode and a typical EDL negative one; iii) bi-functional electrolytes (i.e. Li

SO

+ KI) merging the advantages of high voltage owing to the supporting electrolyte (i.e. Li

SO

) and high capacitance owing to electrodes hybridization provided by e.g., iodide redox species.

However, several researches in the literature wrongly report the capacitance values and mostly

ignore the energy efficiency of electrochemical capacitors in aqueous electrolyte. In addition, there

is a lack of understanding the charge storage mechanisms at each electrode in hybrid cells

implementing aqueous potassium iodide. Furthermore, the relatively low solubility of electrolytes

Patr yk Prz ygocki 12 based on alkali sulfates and their freezing close to sub-zero temperatures (ca. -10°C for 1 mol L

Li

SO

) is a bottleneck in the development of such capacitors using water-based media. Therefore, in-depth study leading to better understanding the mechanisms which govern the ECs operation and proposing new strategies for the improvement of the electrochemical performance of ECs in aqueous electrolyte is required. In order to meet these challenges, the main objectives of the doctoral thesis are i) to establish an appropriate method for the evaluation of capacitance and efficiency of ECs based on aqueous electrolytes ii) to improve the ECs performance by designing environmentally friendly hybrid cells implementing bi-functional electrolytes which enable to obtain high cell capacitance (and energy) and cover the domain of the low temperatures and iii) to get an insight on the ageing processes within the ECs based on aqueous electrolyte using Electrochemical Mass Spectrometry (EMS).

Overall, the dissertation is divided into six chapters:

The first chapter introduces the literature background on electrochemical capacitors, including the models of the electrical-double layer, the principle of cells operation and the description of their most important properties (energy, power, capacitance, time constant). A special attention is paid to the activated carbon due to the fact that it is the most commonly utilized material for electrodes preparation. The state-of-the-art on ECs using aqueous electrolytes is introduced with the emphasis on neutral salts-, redox-active based ones and the combination of both (bi-functional electrolytes) to show the pros and cons of systems based on these electrolytes. Moreover, a systematic ECs classification is presented focusing on different charge storage mechanisms of various kinds of cells and their proper nomenclature. Besides, the review on ECs ends with the description of ageing analysis for the cells in aqueous electrolyte to bring the knowledge about the possible reasons and effects of this phenomenon.

The second chapter includes a summary of the publication entitled “Appropriate methods for

evaluating the efficiency and capacitive behavior of different types of supercapacitors” together

with the publication itself as attachment. Due to the fact that electrochemical capacitors using

neutral aqueous electrolytes often behave differently from an ideal capacitor (displaying non-

linear galvanostatic charge/discharge profiles or non-rectangular cyclic voltammograms), their

characteristics should be evaluated very carefully. Therefore, this chapter is dedicated to the

Patr yk Prz ygocki 13 description of problems encountered when using the traditional evaluation methods and also to the proper calculation methodology of capacitance and efficiency. It is then shown that some metrics are wrongly used in the literature, especially when the specific capacitance is reported for battery- type electrodes which do not display capacitance. Hence, exemplary calculations of efficiency and capacitance are documented for the cells in organic, ionic liquids, aqueous neutral and redox-active electrolytes. This methodology will be further applied in the next sections of the manuscript.

Due to the relatively high scarcity and cost of lithium sulphate, the chapter III deals with the implementation of manganese sulfate as supporting electrolyte for hybrid carbon/carbon cells. It presents a summary of the publication entitled “Confinement of iodides in carbon porosity to prevent from positive electrode oxidation in high voltage aqueous hybrid electrochemical capacitors” together with the publication which is attached. Manganese sulfate was selected for its low cost, high abundance, environmentally friendly character and slightly acidic pH of the aqueous solution (pH = 3), which allows to enhance the iodide activity (when used with KI). The state of iodine-based species in the positive carbon electrode extracted from hybrid cells is analysed by Raman spectroscopy. Moreover, post-mortem investigations are performed on aged positive electrodes in both MnSO

and MnSO

+KI by gas adsorption, temperature programmed desorption (TPD) and energy dispersive spectrometry (EDS). The obtained results help in understanding the reasons of much better performance of the hybrid cell utilizing MnSO

+KI than the symmetric cell based on aqueous solution of MnSO

. It will be also shown that, owing to the more pronounced redox activity of the iodide species at slightly acidic pH, higher capacitance is obtained in the hybrid cell using MnSO

+KI as compared to Li

SO

+KI.

The fourth chapter introduces the advantages of an asymmetric configuration of AC electrodes for

improving the electrochemical performance of hybrid capacitors in aqueous Li

SO

+KI bi-

functional electrolyte. It contains a summary of the publication entitled “Capacitance

enhancement of hybrid electrochemical capacitor with asymmetric carbon electrodes

configuration in neutral aqueous electrolyte” together with the publication itself. Based on the

requirements for optimizing the performance of the two electrodes in this hybrid cell, the idea is

to implement a microporous carbon for the negative electrode in order to enhance its EDL

capacitance and a micro/mesoporous carbon for the positive electrode to better trap bulky

polyiodides within the porosity. Owing to the adapted porosity of both electrodes, it is expected

Patr yk Prz ygocki 14 that the cell will display better electrochemical characteristics during ageing (both lower capacitance drop and resistance increase).

Although it was definitely proved in the previous chapters that hybrid cells using an aqueous bi- functional electrolyte may serve as a promising eco-friendly alternative to capacitors in organic electrolytes, their poor performance at sub-zero temperatures remains a cornerstone. Few strategies to extend the application of cells using aqueous electrolyte to low temperatures have been proposed in the literature, but none of them is satisfactory due to the implementation of highly diluted and low conductivity solutions. In order to solve these issues, an alternative, cheap and eco-friendly aqueous electrolyte containing choline nitrate and choline iodide is proposed in chapter V to realize a hybrid cell working safely down to -40°C. The low temperature properties of choline nitrate/choline iodide mixture are analysed by differential scanning calorimetry (DSC). Then, the electrochemical performance of the AC/AC cell using the selected mixture is extensively studied in the temperature range from -40°C to +24°C. Finally, it is attempted to improve the characteristics of the described system by i) implementing different carbons with a proper porous texture adapted to the species stored in each electrode and ii) by realizing pouch cell models of the two-electrode devices.

The objective of chapter VI is both to identify and quantify the reactions and other kinds of

processes occurring at the carbon/electrolyte interface of both electrodes during ageing of AC/AC

capacitors in a water-based medium. For this purpose, potentiostatic floating is applied to AC/AC

cells in aqueous Li

SO

, and the thereof produced gases are analysed by mass spectrometry, while

pressure inside the cells is simultaneously monitored. The carbon electrodes are analysed by

temperature programmed desorption after the floating period in order to get an insight on the

surface functionality modifications during ageing. The analysis of all the obtained data relies on

the quantitative estimation of the charge calculated from the leakage current (during floating at 1.5

V) and the amount of charge spent for the production of gases, oxidation of both electrodes surface

and other parasitic reactions like ionic charge diffusion and charge distribution. Such approach is

expected to facilitate the understanding of the various processes/reactions taking place during

ageing of ECs in aqueous electrolytes, and to further contribute to determining strategies for

improvement of their performance.

Patr yk Prz ygocki 15

The manuscript ends with a general conclusion which highlights and discusses the most important

results presented in the manuscript. Finally, perspectives for further research work in the field of

electrochemical capacitors based on aqueous electrolytes are suggested.

Patr yk Prz ygocki 16

Chapter I

Literature review

Patr yk Prz ygocki 17

1. Introduction

This bibliography part aims at providing the available state-of-the-art on electrochemical capacitors (ECs) in aqueous electrolytes. Therefore, it will first introduce briefly the general properties of electrical double-layer capacitors, and then quickly shift to describing the properties of activated carbons used for electrodes manufacturing, and particularly their porous texture and surface functionality, which strongly affect the ECs performance. Then the properties and operation mechanisms of symmetric and hybrid ECs in neutral aqueous electrolytes will be extensively presented with special emphasis on the advantages of using bi-functional electrolytes (including a support salt and a source of redox active species) and on the degradation mechanisms occurring during ageing of ECs in aqueous media.

2. Electrochemical capacitors

2.1 Models of the electrical double-layer

Helmholtz was the first scientist who suggested a charge separation model in 1853, to explain that

the interface between a polarized metallic plate and an electrolytic solution has the ability to store

electric charges. It was the beginning of the so-called electrical double-layer (EDL) model which

got its name from the two compact layers of charges of opposite signs at the electrode/electrolyte

interface (Fig. 1a). However, in this model, both diffusion of ions and thermal motions in the

electrolyte, and the interaction between the electrode and the dipole moment of the solvent

molecules were neglected [1, 2]. Between 1910 and 1913, Gouy and Chapman proposed the so-

called diffuse double-layer model where ions serve as point charges which are included within a

single diffuse layer (Fig. 1b). In this model, the thermal motion of ions encapsulated in the charged

surface was taken into account. However, it was not satisfactory enough to give an insight on the

actual capacity-potential curves which were experimentally established [3, 4]. Later, after

modifying the previously described models, Stern distinguished between the ions which adhere to

the electrode surface, as it was suggested by Helmholtz, and ions forming a diffuse layer (Fig. 1c)

[5]. In the Stern model, two capacitors in series are effectively operating in determining the total

capacitance of the interface, and the differential capacitance of the double layer C

is expressed by

equation (1):

Patr yk Prz ygocki 18

= + (1)

where C

is the capacitance due to the charges in the outer Helmholtz plane (OHP) and C

is the capacitance due to the charges in the diffuse layer.

In 1947, Grahame proposed an improved model in which ions (but not solvent molecules) or some uncharged species occupy the closest region to the electrode [6]. The ions in direct contact with the electrode are called to be specifically adsorbed. In this model, it is possible to distinguish three different contributions:

• the inner Helmholtz plane (IHP), which passes through the centres of the specifically adsorbed ions,

• the outer Helmholtz plane (OHP), which passes through the centres of solvated ions,

• the diffuse layer: located from the OHP to the bulk solution.

Figure 1. Models of the electrical double-layer a) Helmholtz model, b) Gouy-Chapman model,

c) Gouy-Chapman-Stern model [1].

Patr yk Przygo cki 19 Nowadays, another EDL model proposed by Bockris, Devanatham and Muller (BDM) takes into account the predominant role of the solvent at the electrode/electrolyte interface (Fig. 2). It claims that solvent molecules reorient depending on the charges excess on the electrode and the presence or absence of specifically adsorbed ions on the surface. The presence of primary and secondary water layer is distinguished in this model, being the main difference from the Grahame’s one. The water molecules form an oriented layer on most of the electrode surface, and at certain sites they are replaced by specifically adsorbed anions that have shed their hydration shell [7].

The mechanism of double-layer capacitance relies on storing charges by reversible electrostatic adsorption of ions of the electrolyte onto an electrode. Due to the charge separation occurring on polarization at the electrode-electrolyte interface, the double-layer capacitance C is given by equation (2) [8]:

C = A (2)

Figure 2. Graphical illustration of the BDM model [adapted from 7].

Patr yk Prz ygocki 20 where ε

is the relative dielectric constant of the electrolyte, ε

the dielectric constant in vacuum, d is the thickness of the double-layer and A is the available surface area of the electrode/electrolyte interface [8].

2.2 Electrical double-layer capacitors

An electrical double-layer capacitor (EDLC) consists of two high surface area carbon electrodes of and a porous membrane immersed in an ionically conductive electrolyte (Fig. 3) [1]; the separator electronically insulates the electrodes while allowing diffusion of ions between the compartments.

The capacitance C is the device characteristic (expressed in F) which correlates the electrical charge Q with the applied voltage U. For the voltage range of a test (from U

to U

), C is calculated accordingly to equation (3) [10]:

C =

∆

=

∆

= (3),

Figure 3. Scheme of an electrical double-layer capacitor in charged state and its corresponding

equivalent circuit. R

, R

, R

stands for resistors and C

, C

for capacitance of the positive and

negative electrode, respectively [9].

Patr yk Prz ygocki 21 where I is the discharge current and t

, t

stand for the time limits.

Taking into account that the two electrodes of capacitance C

or C

are connected in series, the capacitance of the two-electrode cell is given by equation (4):

= + (4),

which means that the capacitance of the device is essentially determined by the electrode with the lowest capacitance value.

The energy E stored in an electrochemical capacitor is expressed by equation (5):

E = CU (5).

Consequently, the energy of ECs is generally enhanced by applying high stability window electrolytes, namely organic electrolytes (e.g., 1 mol L

TEABF

in acetonitrile) enabling to reach voltage values as high as 2.7 V – 2.85 V [11, 12], together with high surface area activated carbons (equation (2) shows that the capacitance C is controlled by the surface area of the electrode/electrolyte interface). Further in this chapter (section 4), it will be shown that high energy values can be also obtained in aqueous neutral electrolytes, especially in presence of redox active iodide species.

The power of ECs is calculated either according to equation (6):

P =

(6)

by dividing the energy E by the discharge time ∆t, or according to equation (7):

P = (7),

where U stands for the voltage expressed in V and ESR for the equivalent series resistance

expressed in Ω. The ESR includes the electronic resistance of the electrode material, the contact

resistance between the electrodes and the current collectors, the electrolyte resistance, the ionic

diffusion resistance and the ionic resistance caused by the separator. In case of redox reactions, the

charge transfer resistance (R

) also contributes to the overall ESR [13]. However, the latter

Patr yk Prz ygocki 22 contribution should not exist in case of a pure electrostatic charge storage mechanism. Therefore, the series resistance of EDLCs is lower as compared to that of batteries, staying at the origin of higher values of power density of EDLCs than batteries. As previously stated for the energy, the stability window of the electrolyte has an important influence in determining the power of EDLCs.

The so-called Ragone plot is used to compare the specific power and specific energy of electrochemical energy storage devices (Fig. 4). Nowadays, Li-ion batteries (LIBs) can deliver high energy densities (from 120 to 240 Wh kg

) with relatively low power densities (up to only 3 kW kg

). Besides, EDLCs are able to deliver very high power (up to 40 kW kg

) with lower energy density than batteries (from 5 to 15 Wh kg

) [14]. The EDLCs charging time is usually between 1 and 10 seconds whereas for LIBs it generally lasts a few hours.

Figure 4. Ragone plot showing the correlation of specific power and specific energy for various

electrochemical energy storage systems [15].

Patr yk Prz ygocki 23 Key indicators of ECs performance are its capacitance C and its time constant τ (which is an indicator of power capability), which are primarily dependent on the technology used for the device construction. By definition, the time constant of a resistor-capacitor system reflects the time necessary to either charge the capacitor up to 63.2% or discharge it to 36.8% of its full charge [10].

This parameter is expressed in seconds and given by equation (8):

τ = !C (8)

where R stands for the resistance and C the capacitance of the capacitor [10].

3. Activated carbon electrodes for electrochemical capacitors

Together with the electrolyte, which mainly determines the operational voltage of the cell, electrodes are a key part of an electrochemical capacitor owing to their dominant contribution to the cell capacitance. There are many forms of carbon (carbon nanotubes, graphite, carbon blacks etc.), however activated carbons are the most commonly utilized electrode materials in the ECs technology. Their good physico-chemical properties and availability make them an attractive and cost effective component. In this section, a special attention will be paid to the properties of activated carbons, their porous texture and surface functionality which have a significant impact on ECs electrochemical performance.

As aforementioned, activated carbon is a proper choice for electrode material in ECs, owing to its combination of well adapted chemical and physical properties, namely [12]:

High specific surface area (up to 2500 m

g

estimated by Density Functional Theory - DFT),

Controllable porous texture Good electrical conductivity, High thermal stability,

Easily compatible in composite materials, Low cost.

In general, ACs are synthesized in two steps, i.e. carbonization and activation. A carbon-rich

organic precursor is heat treated in an inert atmosphere to eliminate heteroatoms, and further

Patr yk Prz ygocki 24 physically or chemically activated in order to develop the porosity. ACs can be produced from natural sources such as coconut shells, wood, pitch, coke, lignite as well as sucrose, corn grain, coffee ground, straw etc. [16-22]. However, advanced ACs displaying reproducible properties and uniform texture are better produced from synthetic polymers, such as polyacrylonitrile (PAN), polyvinylidene chloride (PVDC), polypyrrole (PPy) or polyaniline (PANI) [23-24].

In physical activation, the surface area is developed by gasification realized by partial etching of carbon during its annealing with oxidizing agents, such as CO

and H

O, according to equations (9) and (10) [25]:

C + CO

→ 2CO (9)

C + H

O → CO + H

(10).

The preparation of chemically activated carbons involves redox reactions between the carbon precursor and a chemical reagent, such as potassium hydroxide KOH [26], phosphoric acid H

PO

[27], zinc chloride ZnCl

[28] at moderate temperature (up to 700-800°C). Usually, this route provides carbons with smaller pores and more uniform pore size distribution (PSD). In order to remove inorganic impurities remaining on the surface of carbons after activation, it is necessary to extensively wash the materials. For EC applications, thermal post-treatment at 800°C-1000°C under neutral gas is often realized to remove all functional groups formed on the carbon surface.

3.1 Porous texture of activated carbons

According to the International Union of Pure and Applied Chemistry (IUPAC), there are three

types of pores dependently on their size (Fig. 5) [29]: micropores: < 2 nm (further divided into

supermicropores between 0.7 and 2 nm, and ultramicropores with pore size < 0.7 nm), mesopores

from 2 to 50 nm, and macropores having pore size larger than 50 nm. Essentially micropores

together with mesopores contribute to high specific surface area (SSA up to 2500 m

g

using

DFT) of activated carbons. Micropores are perfect for adsorption processes whilst mesopores play

an important role in facilitating the transport of adsorbates to the micropores [30].

Patr yk Przygo cki 25 The most popular methodology for porosity investigation is the adsorption of a gas or vapor onto an outgassed solid in a confined space at a given temperature and pressure. This process relies on the transfer and accumulation of adsorbate molecules in an interfacial layer as a consequence of attractive forces with the surface of the solid. Then, the thermodynamic adsorption equilibrium is realized at stable pressure. Depending on the interactions between the adsorbent and the adsorbate, there are two types of adsorption: physical and chemical; in the latter case, irreversible interactions of chemical nature take place. On the contrary, during physisorption, weaker and attractive both dispersive and electrostatic van der Waals forces occur between the permanent electrical moments of gas molecules and the electrical field of the solid. Physical adsorption has lower binding energy (from 10 to 100 meV) than chemical adsorption (from 1 to 10 eV). It is a reversible process which allows crucial information such as surface area and pore volume of the solid under study to be obtained [32].

An adsorptive gas has to possess a few specific features in order to be successfully employed for adsorption on the surface of carbon. The most important is that it has to be chemically inert to avoid any interaction with the adsorbent. Moreover, it must display a relatively high saturation pressure (P

) in order to cover the whole porosity range, and finally its shape should be as spherical as possible to avoid any error during calculation of transversal section [33]. There are a few gases

Figure 5. Scheme of the various pores in an activated carbon [adapted from 31].

Patr yk Prz ygocki 26 which fulfill the mentioned above conditions, such as N

, CO

, Ar, H

and He. CO

adsorption at 273 K is usually applied to analyze the porous texture of carbon molecular sieves and especially ultramicroporous carbons [34, 35]. At 273 K, both the diffusion and equilibration processes of CO

molecules in micropores are much faster than those of N

or Ar at cryogenic temperatures.

Fortunately, N

or Ar adsorption brings complementary information about the larger pores when the measurements are conducted at cryogenic temperatures and relative pressures higher than 0.001 (where the diffusion and equilibration are faster than at low pressures) [36]. Nitrogen is the most commonly utilized in porosity measurements, owing to its low price and easy availability;

the analysis temperature of 77 K is achieved by dipping the probe tube containing the sample to be analyzed in boiling liquid nitrogen under atmospheric pressure. At such low temperature, the adsorbed N

covers the whole porosity owing to the wide range of relative pressure possible to be obtained (from 10

to 1). However, it is necessary to use high vacuum pumps and pressure transducers for accurate pressure measurements. Moreover, for very small micropores (size < 0.7 nm), the kinetics of N

adsorption is slow, so the equilibrium time and consequently the analysis of the sample is quite long. It has been also recently proposed that the combination and fitting of adsorption isotherm data from N

and CO

is a more accurate method than the ones based on the individual gas adsorption isotherms [36]. A similar methodology was applied to the N

and H

adsorption data at 77 K for a series of activated carbons [37] and Ar and H

data at 87 K for molecular sieving carbons [38].

For a concrete solid sample and adsorptive gas at constant temperature, the amount of gas adsorbed (n – usually in cm

g

at standard temperature and pressure – STP) on the surface of the adsorbent is proportional to pressure, according to equation (11) [32]:

n = " ($)

(11)

In general, the gas is adsorbed in 4 steps as pressure increases (Fig. 6): i) at low pressure, the gas molecules are adsorbed on isolated sites on the sample surface; ii) together with an increase of pressure, the adsorbed molecules cover the surface forming a monolayer (one molecule thick);

then a multi-layer is formed (smaller pores in the sample are firstly filled); finally, the surface is

completely covered and all pores are filled [39].

Patr yk Prz ygocki 27

The amount of adsorbed gas onto the surface of a substrate versus relative pressure (P/P

) is

reflected by the adsorption/desorption isotherm (which in case of N

is realized at 77 K). The first

classification of the 5 types of adsorption/desorption isotherms (of any gas) was given by Brunauer

et al. [33, 34] in the 1940s. However, in 1985, the IUPAC proposed an extended version of this

classification by adding a sixth type – the stepped isotherm [35, 40]. The 6 types of isotherms are

shown in Figure 7 [40]. The Langmuir isotherm (type I) is a typical curve for microporous

substrates having pores below 2 nm (activated carbons, zeolites, metal organic frameworks). At

relative pressures close to 1 the amount of gas adsorbed may reach a limiting value or slightly

increase if larger pores are present. The type II represents macro- or nonporous materials (e.g.,

graphite, carbon black, carbon nanotubes - CNTs) with monolayer-multilayer adsorption. At the

point B, a monolayer is formed and the multilayer adsorption starts to occur. The type III is

characterized by very weak attractive forces between the adsorbate and the adsorbent (e.g., water

vapor on graphitized carbon blacks). On the type IV, a hysteresis loop can be noticed meaning that

the material is mesoporous and capillary condensation occurs (e.g., templated carbons, compacted

carbon blacks). However, for the carbons having both micro- and mesopores, one can expect the

Figure 6. Graphical representation of gas adsorption onto the surface of a porous sample [adapted

from 39].

Patr yk Prz ygocki 28 combination of type I and type IV isotherms. The type V suggests weak attractive forces, and the only difference between this one and type III is the presence of a hysteresis loop, where the desorption branch does not overlap with the adsorption one (e.g., steam adsorption on charcoal).

The type VI occurs for some materials having a uniform nonporous surface, and each step on the isotherm indicates the formation of a monolayer (e.g., noble gas adsorption on graphitized carbon blacks) [32, 40].

Usually, the BET (Brunauer-Emmett-Teller) specific surface area is calculated from the nitrogen

adsorption isotherm, however, for activated carbon materials, dependently on the size of pores, it

frequently leads to an overestimation (in case of pores larger than 0.9 nm) or underestimation (in

case of pores smaller than 0.9 nm) of S

. Neimark and Ravikovitch [41-43] were the ones who

started to develop customized Density Functional Theory (DFT) methods applicable for materials

of various porous texture characteristics. A drawback of the conventional Non-Local Density

Functional Theory (NLDFT) method is the fact that it does not take into account both chemical

and geometrical heterogeneities of the pore walls, but it considers a chemically and geometrically

Figure 7. The six types of adsorption isotherms according to the IUPAC classification [40].

Patr yk Prz ygocki 29 smooth surface model. As a consequence, a mismatch between the theoretical assumption of smooth and homogeneous surfaces and the inherent molecular scale heterogeneity of real adsorbents is that the theoretically calculated NLDFT isotherms display multiple steps related to the formation of a monolayer, a second adsorbed layer and a multilayer; these artificial steps cause a gap in pore size distributions (PSDs) of microporous samples between 1 and 2 nm [44].

Therefore, there were several attempts to improve this method in order to avoid this problem [44- 49]. In the Quenched Solid Density Functional Theory (QSDFT), the solid is considered as a heterogeneous structure owing to the surface roughness which is taken into account, improving the previous NLDFT models for carbon samples that assumed flat, structureless and graphitic pore walls. Both N

and Ar isotherms calculated by these models do not show layering transitions. Thus, this model which was initially established using non-graphitized carbon blacks can be applied to process the experimental data obtained with other carbon materials [50, 51]

Recently, the 2D-NLDFT method (two dimensional – Non-Local Density Functional Theory) established by Jagiełło and Olivier has been proposed for the estimation of cumulative surface area as well as pore size distribution (PSD) of activated carbons, carbon blacks, carbon nanotubes and zeolites [52]. It gives a better agreement between the NLDFT and experimental N

adsorption isotherms on porous carbon materials owing to the introduction of a two-dimensional finite and energetically heterogenic pore model instead of the standard one-dimensional method which evaluates infinite graphite-like pores; various shapes and sizes of pores are also taken into account in this model [42].

Examples of adsorption/desorption nitrogen isotherms at 77K for essentially microporous (DLC

Supra 30, Norit) and mesoporous carbon (BP 2000, Cabot) are presented in Figure 8a and 8b,

respectively. DLC Supra 30 exhibits a classical type I isotherm (Fig. 8a) in which the majority of

gas is adsorbed at low relative pressure (P/P

< 0.02). In this kind of sample, the desorption branch

overlaps with the adsorption one, which can be easily visible by the lack of hysteresis loop. On the

contrary, BP2000 is characterized by a combination of type I and type IV isotherms in which some

amount of nitrogen is adsorbed in micropores at very low relative pressure, yet the highest increase

of adsorbed gas can be noticed at high relative pressure (from 0.9 to 1), which indicates the

presence of large mesopores. The corresponding PSDs obtained using the 2D-NLDFT method for

these two carbon materials are given in Figure 8c and 8d and they confirm the mentioned porous

Patr yk Przygo cki 30 texture analysis. Indeed, DLC Supra is an activated carbon material containing pores of maximum size around 3 nm, whereas BP 2000 is a mesoporous carbon of external surface having pores even up to 45 nm.

3.2 Effect of porous texture of activated carbons on capacitance

In order to find correlations between the capacitance properties of carbon electrodes and their porous texture, many researchers assume that the accessible surface area of the electrode/electrolyte interface is equal to the specific surface area measured by gas adsorption [53].

Figure 8. Nitrogen adsorption/desorption isotherms at 77 K for a) DLC Supra30 (blue curve) and

b) BP-2000 (red curve). PSDs using the 2D-NLDFT method for c) microporous

DLC Supra30 (blue curve) and d) mesoporous BP-2000 (red curve).

Patr yk Prz ygocki 31 However, such statement is not correct, because of the different pore filling mechanisms, and also because of the different size of the gas probe and electrolytes ions [9]. In Figure 9a, which shows as example the relationship between the gravimetric capacitance and the BET specific surface area (S

) of a series of carbon materials (ACs and carbon blacks), it can be seen that the gravimetric capacitance is proportional to S

up to around 1500 m

g

, and that it does not increase anymore for higher values [54]. In the DFT model which considers a slit-shaped pore geometry [54], the proportionality between the DFT specific surface area (S

) is slightly better than in the previous case, but still the same “saturation” phenomenon is observed for S

higher than 1200 m

g

(Fig. 9b) [54]. As suggested by Barbieri et al. [54], the average pore wall thickness declines with the increase of pore volume (i.e. degree of activation to produce the material), until it approaches the screening length of the electric field (δ

). Hence, the “saturation” in capacitance observed for DFT specific surface area values higher than 1200 m

g

is due to too thin pore walls and overlapping of adjacent space charge regions within the carbon materials leading to incomplete charge screening. Notwithstanding, it is important to mention that, due to the molecular sieving effect, the EDL capacitance of nanoporous activated carbons may be negligible when the size of pores becomes smaller than the size of electrolyte ions [55-59].

While attempting to better understand the effect of pore size on the capacitance of carbon

electrodes, it has been reported that the rise of normalized capacitance (capacitance divided by the

Figure 9. Gravimetric capacitance vs a) BET and b) DFT specific surface area of a series of ACs

and carbon blacks [9]

Patr yk Prz ygocki 32 specific surface area) claimed for average micropore size (L

) lower than 1 nm in capacitors implementing various carbons and using organic electrolyte (tetraethylammonium tetrafluoroborate in acetonitrile) [60], depends strongly on the applied model for the estimation of surface area [61]. This rise occurs when the BET model is applied (Fig. 10a), and it is explained by the fact that this model overestimates or underestimates the specific surface area when pores are larger or smaller than 0.9 nm, respectively [61]. By contrast, when considering the average specific surface area (S

, calculated from several methods) corresponding to pores larger than 0.63 nm which is the equatorial diameter of the Et

N

cation, the normalized capacitance is constantly independent of the average pore size (Fig. 10b) [61]. This kind of results inconsistency shows the importance of careful data manipulation and their critical evaluation in order to get the right conclusions.

3.3 Surface functionality of carbons

Surface functionalities can be introduced on edges or in-built (depending on the precursor) in the graphene sheets of high surface area carbon materials, either on purpose (by modification and functionalization [62-64]) or by exposure to the atmosphere at ambient conditions. Recently, special attention has been paid to the incorporation of such heteroatoms as nitrogen, boron, sulphur or phosphorus, owing to the fact that such treatment may lead to capacitance increase [65-67].

Figure 10. Dependence of normalized capacitance vs average micropore size for a variety of

porous carbons (black squares) and carbon monoliths with pore widths around 0.7 nm

(green circles): a) capacitance normalized to S

; b) capacitance normalized to S > 0.63 nm [9].

Patr yk Prz ygocki 33 Besides, hydrogen can be also present on the surface as a part of some functional groups such as - COOH and -NH

. The nature and amount of functionalities depend strongly on the carbon precursor and on the activation or modification method [68, 69]. In fact, the accessible active sites which are on the surface of carbon, such as edge carbons and defected carbon rings, are the ones which stay at the origin of heteroatoms incorporation [70]. The most important heteroatom being covalently bonded to carbon is oxygen. Oxygenated surface functionalities have an influence on hydrophilicity/hydrophobicity, surface behaviour, electron density of graphene layers [71-72]. The main reasons for a relatively rich oxygenated surface functionality on carbons are i) the existence of free edge sites and ii) the presence of unpaired (delocalized) electrons on the basal planes. The groups are responsible for the change of pH at the surface and formation of a surface charge [73];

therefore, the point of zero charge (pH

) is an important parameter to characterise carbons. In fact, when a carbon of given surface pH is immersed in water, its low (or high) point of zero charge pH

is the consequence of the two following phenomena: i) at acidic pH (lower than pH

), the surface is positively charged, OH

ions are preferentially adsorbed and there is an excess of H

O

ions in the bulk solution; ii) at basic pH (higher than pH

), the surface is negatively charged, H

O

ions are adsorbed and OH

in excess stays in the solution. The adsorption of H

O

ions is due to the presence of dissociated surface OH groups (like phenolic or carboxylic) at the edges of graphene layers. On the contrary, the adsorption of OH

occurs owing to the formation of C

-H

O

donor-acceptor sites on the basal-plane surface of the graphene layers [73].

The variety of heteroatom-based surface functional groups is presented in Figure 11. These groups can be divided into basic and acidic. The latter include carboxyl, hydroxyl, lactone, anhydrides, lactol and phenol, whereas the former consist of pyrone, carbonyl and ether. All the nitrogen-based surface functionalities are basic (or slightly basic/neutral in case of quaternary ammonium group), whereas the sulfonic-type related with the presence of sulfur is obviously an acidic group [74].

In order to characterize the surface functionality of activated carbons, a number of experimental techniques may be utilized, and can be divided into wet and dry methods. Among the former, we can distinguish two titration methods, such as: chemical (developed by Boehm to estimate the amount of both basic and acidic oxygenated surface functional groups on carbon samples) [75]

and potentiometric (PT – to get an insight on the population of acidic sites which can be described

Patr yk Prz ygocki 34 by a continuous pKa distribution function) [76]. Dry methods include X-ray photoelectron spectroscopy (XPS), infrared spectroscopy methods (both FTIR and DRIFTS) and temperature- programmed desorption (TPD) [77].

TPD is the most commonly used (along with XPS), and it is based on the coupling of

thermogravimetric analysis (TGA), aiming at desorbing the groups from the solid surface by

increasing the temperature, with mass spectrometry (MS) needed for the detection of evolving

gases (the most often CO

, CO and H

O). After deconvolution (by Gaussian or Lorentzian

function) of the TPD profiles, and taking into account the information on the relationship between

the desorption temperature of a given gas and the decomposition of particular groups (based on

Figure 11. Functional groups on carbon surface related to the presence of: a) oxygen,

b) nitrogen and c) sulfur. The acidic group are marked in red and the basic groups are marked in

blue [74].

Patr yk Prz ygocki 35 the literature), it is possible to get an insight about the nature and amount of functionalities present on the carbon surface [78, 79].

3.4 Oxidation of activated carbons

In general, the oxidation of activated carbons can be divided into i) dry oxidation by a direct contact with steam, CO

, O

, O

etc. at temperature higher than 700

C; ii) wet oxidation by reaction with a solution of oxidizing agents, such as H

O

, HNO

, H

SO

, (NH

)

S

O

, KClO

, KIO

, KMnO

, etc. (under mild reaction conditions) and iii) oxygen plasma treatment [69, 79]. Besides, the electrochemical modification of activated carbon surface [80-84] presents several advantages, such as: i) the possibility of in situ analysis, ii) an easy supply of electrons by a direct current (DC) source, iii) the treatment can be done in moderate conditions (room temperature and atmospheric pressure), iv) the experiments are easy to reproduce and v) the electrode potential can be fully controlled during both the oxidation and reduction processes.

The effect of an electrochemical treatment (3 hours of constant current from 0.2 to 2 A) on the surface of granular activated carbon (GAC) in aqueous NaCl, NaOH and H

SO

electrolytes has been analyzed by Berenguer et al. [85]. By anodic treatment, oxygenated surface groups are created without any significant change in porous texture (only a slight blockage of pores was observed). Interestingly and unexpectedly, the cathodic treatment of activated carbon also leads to an increase in the surface oxygenated complexes as shown by TPD (for both CO

- and CO- evolving groups), whatever the used electrolyte; the higher the current value, the higher the degree of oxidation. It was proposed that hydrogen peroxide and oxygen-containing radicals (such as:

hydroxyl HO

, superoxide HO

or superoxide ions O

) are generated during the electroreduction of di-oxygen dissolved in the electrolyte, staying at the origin of the cathodic functionalization. At high current values (close to 1 A g

), either a competitive mechanism of indirect oxidation by hydrogen peroxide and oxygen-containing species or a direct reduction reaction may take place.

The electroreduction mechanism of O

occurs in two stages (Fig. 12), such as i) direct four-electron

reduction of O

, ii) reduction of di-oxygen to peroxide and reduction of peroxide to water or

hydroxide. Both processes may occur on different parts of the electrode surface and are strongly

dependent on the electrolyte pH, cathodic overpotential and nature of the electrode [85].

Patr yk Prz ygocki 36

The proposed mechanisms of direct and indirect electrooxidation were suggested in another study

with zeolite templated carbon (ZTC) [86]. The anodic treatment of ZTC sample in NaCl using

various currents and polarization durations causes an increase of CO

- and CO-evolving groups

and gradual decrease in both the specific surface area and micropore volume; the oxidation degree

progressively increases with the applied current and time. Besides, the effect of electrolyte on the

electrooxidation of ZTC was investigated using NaCl, NaOH and H

SO

. At higher currents, the

highest oxidation degree was obtained in NaOH electrolyte, whereas at lower values of current,

the treatment using H

SO

is the most oxidizing one. In both types of current conditions, an

intermediate behavior is observed in NaCl. In addition, longer treatments at lower currents are able

to give rise to a larger amount of oxygenated functionalities in the three kinds of electrolytes, but

essentially in the case of H

SO

electrolyte. Besides, mainly CO

-evolving groups were formed

after the treatment in NaOH, whereas in H

SO

the formation of CO-type evolving groups was

promoted. In case of modification in NaCl, an intermediate ratio of CO/CO

-evolving groups was

observed, which indicates that depending on the electrolyte, different amount and type of

oxygenated functionalities can be produced on the surface of ZTC [86]. It was proposed that the

anodic polarization may withdraw some electrons from the aromatic rings, promoting their

destabilization and the nucleophilic attack by electron-donor species, like water molecules or

hydroxyl species. It was also experimentally confirmed that the amount of formed surface

functionalities is better controlled by electrochemical treatment than chemical oxidation. The

reasons for that can be due to the higher control of the oxidation kinetics by adjusting the relative

Figure 12. Electroreduction pathways of di-oxygen. k

stands for the rate constant related with

the direct 4e

reduction of O

to H

O (in acidic pH) or hydroxide (in basic pH); k

is the rate

constant for the 2e

reduction of O

to peroxide; k

is the rate constant for the 2e

reduction of

peroxide to H

O (in acidic pH) or hydroxide (in basic pH) [85].

Patr yk Prz ygocki 37 participation of the direct and indirect oxidation pathways, which can be achieved by the right choice of the multiple electrochemical variables (current, potential, time etc.) [86].

Considering the example of anodic oxidation of a Ti/RuO

electrode on which carbon was deposited, it has been concluded that carbon can be oxidized either directly (by the presence of water) or indirectly through such oxidants as the OH

radical or Cl

. In both cases, it results in the formation of several functionalities on the carbon surface (e.g. carboxylic, carbonyl, phenolic etc.).

Thus, as shown in Figure 13, it has been assumed that the electrochemical oxidation involves two contributions: i) the direct and indirect oxidation of carbon owing to its polarization and ii) the in- situ electrochemical formation of oxidizing species [86].

4. Classification of electrochemical capacitors in aqueous electrolytes

As the charges might be stored with various mechanisms in the carbon electrodes, it is necessary

to distinguish between the types of electrochemical cells. Based on the criteria of electrolytes used,

Figure 13. Proposed mechanism for the electrochemical oxidation of a carbon material in

NaCl-based electrolyte [adapted from 86].

Patr yk Prz ygocki 38 electrochemical capacitors using organic [11, 12], aqueous [87-90] and ionic liquids [91-94]

electrolytic media have been thoroughly investigated. However, the configuration of electrodes inside the cell is an important criterion for classifying the cells into symmetric [95, 96], asymmetric [97-99] and hybrid [100-102] ones. Besides, there are also reports [103-109] showing that cells using redox active species in the electrolyte or grafted on the electrode materials do not fully exhibit the characteristics of a capacitor (e.g., rectangular cyclic voltammograms and linear charge/discharge curves). In these cases, the redox activity is either displayed on both electrodes or the redox potential is far from the equilibrium potential of the cell, causing non-linear charge/discharge curves; consequently, the cells could be classified as so-called “supercapattery”

[110] and will not be considered in this presentation.

Consequently, taking into account the scope of the thesis, this section will detail the principles and properties of different types of electrochemical cells using carbon electrodes and aqueous electrolytes, while giving capacitor-type charge/discharge characteristics.

4.1 Electrochemical capacitors with symmetric carbon electrodes

The most common electrochemical capacitors are designed in carbon/carbon symmetric configuration, e.g., identical carbon material and nearly similar mass of electrodes [87, 88]. The absence of faradaic charge storage infers the absence of chemical or structural changes of the electrode material. Both positive and negative electrodes display rectangular cyclic voltammograms and linear galvanostatic charge/discharge characteristics. The commonly used electrode material in these EDLCs is nanoporous activated carbon which is a high surface area material, ideal for storing a large amount of charges. The carbon material is generally coated on a current collector and the two electrodes are impregnated with aqueous KOH [87], H

SO

[90] or neutral aqueous solutions (Li, Na and K sulphates or nitrates) [111-114] and separated by a porous membrane. The charges are stored essentially in the electrical double-layer of the positive or negative electrodes until reaching the thermodynamic stability potential of the electrolyte, beyond which faradaic processes start to occur at each electrode. The stability window of the electrolyte plays an important role in determining the maximum voltage values reached by these capacitors.

For symmetric capacitors in aqueous KOH and H

SO

electrolyte, the voltage is generally limited