Study of electrolytic materials for the development of innovative energy storage devices Candidate: Lucia Lombardo Sapienza University of Rome, Chemistry Department PhD Material science, XXV cycle Supervisor: Professor Stefania Panero Coordinator: Professor Ruggero Caminiti

Index

1. Introduction .................................................................................................... 1 1.1. Aim of the work......................................................................................................1 1.2. Safety issues............................................................................................................2 1.3. Lithium batteries.....................................................................................................3

1.3.1. Electrolytes .............................................................................................................9 1.3.2. Electrodes .............................................................................................................12

2. Innovative electrolytes – State of the art...................................................... 20

2.1. Ionic Liquids (ILs)................................................................................................20 2.2. “Fluorine-free” salt-based electrolytes .................................................................21 2.3. Gel polymer electrolytes (GPEs) ..........................................................................22

3. ILs electrolytes ............................................................................................. 26 3.1. Pyrrolidinium-based ILs .......................................................................................26 3.2. Py24TFSI: synthesis ..............................................................................................27 3.3. Py24TFSI and Py14TFSI: characterization.............................................................28

3.3.1. Thermal characterization: DSC, TGA ..................................................................28 3.3.2. Electrochemical characterization: conductivity test, cathodic cyclic

voltammetry and anodic linear scan .....................................................................30 3.3.3. Conclusions...........................................................................................................31

4. IL-alkylcarbonates mixtures......................................................................... 33

4.1. IL-alkylcarbonates mixtures .................................................................................33 4.1.1. Thermal characterization: flammability test .........................................................34 4.1.2. Electrochemical characterization: conductivity test, cathodic cyclic

voltammetry and anodic linear scan .....................................................................35 4.1.3. Cell performances: galvanostatic cyclation ..........................................................37

4.2. Conclusions...........................................................................................................40

5. Fluorine-free salt-based electrolytes ............................................................ 42

5.1. Fluorine-free salt-based electrolytes: characterization .........................................42 5.1.1. Electrochemical characterization: conductivity test, cathodic cyclic

voltammetry and anodic linear scan .....................................................................42 5.1.2. Thermal characterization: DSC, TGA ..................................................................44 5.1.3. Cell performances: galvanostatic cyclation ..........................................................45

5.2. Conclusions...........................................................................................................47

6. Gel Polymer Electrolytes (GPEs)................................................................. 48

6.1. IL-based and carbonate-based GPEs: preparation procedures .............................48 6.2. IL-based GPE: characterization............................................................................49

6.2.1. Electrochemical characterization: conductivity test, cathodic cyclic voltammetry and anodic linear scan .....................................................................49

6.2.2. Thermal characterization: TGA and flammability test .........................................53 6.2.3. Cell performances: galvanostatic cyclation ..........................................................54

6.3. Carbonate-based GPE: characterization ...............................................................55 6.3.1. Thermal characterization: DSC, TGA ..................................................................55 6.3.2. Electrochemical characterization: cathodic cyclic voltammetry, anodic linear

scan and conductivity test.....................................................................................57 6.4. New “in situ” GPE: preparation procedure ..........................................................58 6.5. New “in situ” GPE: characterization ....................................................................58 6.6. Conclusions...........................................................................................................60

7. Technological approach to safety................................................................. 62

7.1. Getters...................................................................................................................62 7.2. Titanium-based metallic getters............................................................................62

7.2.1. Electrochemical characterization..........................................................................62 7.2.2. Results and discussion ..........................................................................................64

7.3. Polymeric composite getters.................................................................................72 7.3.1. Electrochemical characterization..........................................................................72 7.3.2. Results and discussion ..........................................................................................73

7.4. Conclusions...........................................................................................................74

8. General conclusions ..................................................................................... 76

Appendix – Experimental methods..................................................................... 78

Methods of thermal analysis.................................................................................................78 Differential Scanning Calorimetry (DSC)........................................................................78 Thermal Gravimetric Analysis (TGA) .............................................................................79

Techniques for electrochemical analysis..............................................................................80 Electrochemical impedance spectroscopy........................................................................80 Voltammetric techniques..................................................................................................82 Galvanostatic cycling .......................................................................................................84

Acknowledgements ............................................................................................. 87

1

1. Introduction 1.1. Aim of the work

Lithium ion batteries are used today in all the popular portable electronic devices, e.g. mobile phones, laptops, and others, thanks to the high energy content compared to any other electrochemical energy storage device. Commonly a lithium ion battery is constituted by a graphite anode, a lithium metal oxide cathode and a separator soaked with a liquid solution of a lithium salt (e.g., lithium-hexafluorophosphate, LiPF6) in an organic solvent mixture (e.g., EC–DMC mixture). This kind of batteries is light, compact and has an operational voltage averaging on 3.6 V, with an energy density that, according to the structure, ranges from 150 Whkg-1 up to 250 Whkg-1 [1-2]. A serious issue in lithium ion battery technology is safety. Electrodes and electrolytes are both hazard factors in Li-ion cells. In particular the use of graphite-based electrodes easily leads to the release upon cycling of gaseous products or even more dangerously to lithium plating on the electrode surface in high current regimes. Substitutes to graphite have been proposed (e.g. titanium oxide, TiO2) [3] and incorporated in advanced Li-ion cells [4]. However electrolytes are the most critical component for the control of safety of lithium batteries because of the high vapor pressure and the flammability of the LiPF6-organic carbonate solution electrolytes. Currently in the commercial lithium batteries it is mandatory to introduce some security systems like thermal switch (in order to prevent the overheating that can occur in case of overcharge) and a safety tab with vent valve (aimed to control the increase of internal pressure due to the formation of gaseous products) [5]. Despite of these safety systems, lithium-ion batteries can be subject of factory recall due to safety reasons. Moreover these safety systems occupy useful space inside the battery, thus worsening the energy density of the device both in term of volume and weight and increase the possibility of failure. The aim of my PhD thesis is to study innovative, non conventional electrolytes, in order to improve the safety and reliability of Li-ion batteries. This work is structured in three main parts:

1. Development and characterization of Ionic Liquid-based electrolyte solutions 2. Development and characterization of Fluorine-free salt-based electrolytes 3. Development and characterization of Gel polymer electrolytes, using solution

developed at point 1) and 2). During this work, thanks to a cooperation with Saes Getters S.p.A of Milan, the safety issue were also investigated from a technological point of view. In fact some electrochemical investigation were performed on a particular safety system, getter, that can adsorb gases released in batteries or supercapacitors.

2

1.2. Safety issues

Perhaps the word lithium itself has questions of safety tagged to it. In fact, safety is a recurring theme even with lithium-ion cells where metallic lithium is replaced with lithium-insertion active materials. Ridden with a poor understanding of the fledgling lithium-ion battery technologies, what manufacturers and consumers fear are accidents during use or inadvertent abuse [5]. For example, in an incident that occurred to Apple in 1995, lithium-ion batteries got overcharged during an in-house testing of a newly manufactured PowerBook 5300 portable computer [6]. Apple then removed all lithium-ion power packs from their product lines [7]. Hereabouts, Ericsson announced that its mobile phones and other portable electronic applications would wean away from lithium-ion batteries [8]. In fact, several other OEM manufacturers have also been proactive in recalling their products. In 2000, in cooperation with the U.S. Consumer Product Safety Commission, Dell voluntarily recalled 27,000 lithium-ion batteries, manufactured by Sanyo Electric Co. Ltd., and sold in notebook computers. Compaq also recalled 55,000 notebook lithium-ion batteries manufactured by Sony Corporation because of a defect in the circuit board that controls the recharge and discharge processes. One of the recent lithium-ion battery recalls with the USPSC was in 2002 when, upon receiving five reports of batteries overheating (in three of the instances they caught fire), EV Global Motors Company announced the recall of 2000 batteries in their electric bicycles. Withdrawal of products, loss of market and even a ban on lithium-ion batteries were part of a backlash prompted by these incidents. Thus arose the need for safety in commercial lithium-ion battery applications. Today, lithium-ion batteries are the state-of-the-art power sources for a variety of portable electronic devices. They combine high energy density and excellent cycle life, and have no memory effect. That no lithium battery-related accident has been reported in the recent past is testimony to improved safety characteristics of present-day lithium battery products. The excellent safety record has been brought about by regulations governing the safety of the cells [9]. Continual improvements in safety are being made especially with large battery packs as for electric traction and load leveling [6]. The gravity of the situation becomes evident considering the market share for lithium batteries. Of the US$ 37 billion battery market in 2000, about US$ 2.9 billion was shared by lithium batteries, the share for primary and secondary lithium batteries being US$ 1.1 and 1.8 billion, respectively [11][12]. Lithium-ion batteries combine highly energetic materials in contact with a flammable electrolyte based on organic solvents. They can suffer premature failure if subjected to conditions for which they are never designed. Any abuse, including disposing in fire, overcharging, external short circuiting or crushing, can trigger spontaneous heat-evolving reactions, which can lead to fire and explosion. The most flammable component of a lithium-ion cell is the hydrocarbon-based electrolyte. The hydrocarbon-based electrolyte in lithium-ion cells means that under fire conditions, these cells will behave in a fundamentally different way than lead acid, NiMH or NiCAD cells, which contain a water-based electrolyte. Although all charged cells contain stored electrical energy, even fully discharged lithium-ion cells contain appreciable chemical energy that can be released through combustion of the electrolyte. Water-based chemistries, under some charging conditions can produce hydrogen gas through electrolysis of the water; however, this hazard is seldom a concern during

3

storage where no charging occurs. If cells with water-based electrolyte are punctured or damaged, leakage of the electrolyte can pose a corrosive hazard; however, it does not pose a flammability hazard. In comparison, leakage or venting of lithium-ion cells will release flammable vapours. If fire impinges on cells with water-based chemistries, the water in the cells absorbs heat and reduces the total heat release of the fire. In comparison, fire impingement on lithium-ion cells will cause release of flammable electrolyte, increasing the total heat release of the fire (assuming well-ventilated conditions), and possibly increase the total heat release rate of the fire. Other combustible components in a lithium-ion cell include a polymeric separator, various binders used in the electrodes, and the graphite of the anode. Some of these components will degrade if a cell undergoes thermal runaway and produce flammable gases that will vent from the cell. Lithium-ion cells do not contain metallic lithium in any significant quantity to affect fire suppression; in lithium-ion cells, Li+ ions function simply as carriers of electric charge. In contrast, lithium primary (lithium metal) batteries contain a significant mass of metallic lithium as their anode material. There is no publically available data from large-scale lithium-ion cell or battery pack fire/fire suppression tests. There are a number of reasons for the lack of large-scale test data. The lithium-ion cell industry has been evolving rapidly, so there has been an inherent difficulty in defining an “average” cell, battery pack, or device. Thus, if testing were to be conducted and considered reasonably comprehensive, it would require testing of multiple models of cells, packs, or devices from multiple suppliers, and even so might quickly become obsolete as cell chemistries and mechanical designs evolved. Lithium-ion batteries must pass a number of safety tests before they can be certified for use by a consumer. The tests include electrical tests such as external short circuit, mechanical tests such as nail penetration, crushing, dropping to the ground, and environmental tests such as heating in a microwave oven, throwing into a hot liquid, and leak tests in a vacuum. Several techniques have been devised to improve safety. They include use of safety vents, positive temperature coefficient (PTC) elements, shutdown separators, more oxidation-tolerant or less flammable electrolyte constituents and redox shuttle mechanisms.

1.3. Lithium batteries

A quick glance at Li-ion batteries The growth in the use of portable electronic devices such as mobile phones and laptops in the past two decades created strong expectation in the market for prototypes of compact and lightweight batteries that offer higher energy density and higher charge capacity. Furthermore, increasingly stringent environmental regulations and a more rational use of energy resources is driving to the development of batteries for electric vehicles. Initially, the Ni-Cd batteries took the majority share of the market for mobile devices, but environmental problems associated with the use of Cd and the low capacity of this type of cells stimulated the research and the replacement of these devices with Nickel Metal hydride (Ni-MH) batteries, which use hydrogen storage alloys as anodes. The production of this type of battery began around 1990 and their demand for consumer electronics dramatically grew over the years. However, both the Ni-Cd that the Ni-MH

4

batteries have a low potential of the cell (~1.35V) and heavy reagents and reaction products. In parallel to the Ni-MH batteries also lithium-ion cells came out from the R&D laboratories and found increasing application areas along the 90s: in the first decade of the new century they fully replaced Ni-MH cells in the field of portable electronics and mobiles. In fact the high voltage of the cell (~3.7V for commercial prototypes) and the low molecular weight of lithium cause a high gravimetric and volumetric energy density. The energy density of various systems of secondary batteries are shown in Fig. 1.1 [13], which clearly shows that the lithium batteries are smaller and lighter when compared with other systems.

Fig. 1.1 - Comparison of gravimetric and volumetric energy density of different kind of commercial batteries.

Development of Li-ion batteries Lithium is characterized by a low atomic weight (6.94 gmol-1), low density (0.534 gcm-3 at 20°C), a melting temperature of 180.5°C, a boiling temperature di 1347 °C and electrical resistivity of 9.446 · 10-6 Ω cm at 20 °C. It also turns out to be a ductile and malleable material, also able to adhere to different support materials through simple application of a pressure [14]. But the characteristic that makes it so attractive as electrode material is definitely its ability to combine a high theoretical specific capacity equal to 3860 mAh g-1 at a very low electrochemical potential (-3.01V vs. SHE). Research on lithium batteries began in the 50s when it was discovered that lithium metal was stable in a large number of non-aqueous electrolytes. This stability is due to the formation of a passivation layer that prevents direct chemical reaction between metallic lithium and electrolyte, but allows the transport of lithium ions. The research and development of these devices led to the marketing of primary lithium cells in the early 70s. We define "primary cells" all those electrochemical systems able to transform the free energy associated with a redox process, ∆G, in electricity; this process can take place

5

only as long as there are reactants. In primary lithium cells, the main electrochemical processes (i.e. the involved reactions) are irreversible and are: the oxidation of metallic lithium, which acts as an anode, Li+ ion, which is followed by the transfer of electron to the other electrode through an external circuit; the reduction reaction at the cathode; the migration of Li+ in the electrolytic solution. The device consists of: a negative electrode which is precisely the metallic lithium; a positive electrode which usually consists of a composite material, that is an active material; a polymeric binder and optionally an electron-conductor additive; an electrolytic solution consisting of a lithium salt dissolved in a solvent or in a mixture of aprotic solvents. The main features that make this device suitable for use are: a high potential of cell 3-4 V, long periods of storage, due to the formation of a passivation layer on the lithium that minimizes the self-discharge, high energy weight over the 250 Wh / kg (twice the conventional systems). The interest into the above listed characteristics led researchers to a new class of compounds named “intercalation compounds”, usable as cathodes, that could ensure the reversibility of the electrochemical process, leading to secondary lithium cells. The secondary cells are systems able to alternately perform the transformation of chemical energy into electrical energy and vice versa; these devices are also said accumulators because they are able to store electric energy in the form of chemical energy in the charge phase and to return it in discharge phase. They consist of a pattern similar to the one of the primary cells, having a metallic lithium negative electrode and a positive electrode that also in this case is a composite material, made of an active material, a polymeric binder and optionally an additive electron conductor.

Fig. 1.2 - Discharge process in a secondary lithium cell: xLi + A zBw Li xAzBw

Such devices are schematized as: Lithium metal / electrolyte / intercalating material. In the discharging process the lithium metal is oxidized, producing lithium ions, while the electrons flow in the external circuit to the cathode and occupy the lower energy

6

electronic band of the transition metal. At the same time an equal number of lithium ions migrate into the electrolyte and diffuses in the lattice of the cathode material to restore the electroneutrality. During the intercalation-deintercalation, the cathode material undergoes modifications, both structural and electronical, which place limits on the reversibility of the process [15]. The identification of those limits is very important. A serious limitation to the use of systems based on lithium metal is due to the low reversibility of the electrochemical process involving lithium [16]. This is due to the passivation phenomena: while the formation of a passivating film on the surface of the lithium may be beneficial for the stability, it is detrimental to the reversibility of the anodic process. Furthermore, during the process of oxidation of the metallic lithium, a partial destruction of the passivation layer occurs; during the lithium ions reduction process, instead, the electro-crystallization occurs in areas devoid of the passivation layer. The accumulation of lithium leads to the formation of dendritic structures, visible in Fig. 1.3, which grow until they reach the other electrode, generating an internal short circuit of the electrochemical cell. Occasionally, it can also happen that, even before touching the electrode, these structures generate local increases of the current density which, by Joule effect, cause internal overheating capable of decomposing the organic solvents of the electrolyte also generating gases such as CH4, H2, CO2, which may produce dangerous deflagrations.

Fig. 1.3 - Phenomenon of internal short circuit in a secondary lithium cell

Given the lack of stability of the lithium-electrolyte interphase, researchers focused on finding systems with a higher electrochemical stability. It was in the early 90s that Sony Energy Tech. announced the production of what we now call a rechargeable lithium-ion batteries [17]. The real innovation was the nature of the anode material. The lithium-ion accumulators were characterized by the fact that the negative electrode of metallic lithium is replaced with a material that is less reactive to the electrolyte [18]; the “lithium ion” name is due to the fact that lithium is present solely as ion. This device is also called "rocking-chair-battery" due to the way the lithium ions move between the two electrodes. Theoretically, the substitution of lithium with other materials should result in a loss of the specific energy due to the dead weight and volume of the intercalation of anodic

7

material; actually, this loss is less than expected, because in the lithium cells it was never used in stoichiometric amount, but large excess were neded due to the poor lithium plating reversibility. On the other hand also in the lithium-ion cells the first time that the lithium enters the anode, during the charging phase, the process of passivation takes place, irreversibly consuming part of lithium. Therefore the needed amount of lithium is much higher than stoichiometric value in both lithium and lithium-ion cells [19]. State of the art of lithium-ion accumulators The lithium-ion batteries are lightweight, compact and work with a potential of 3.7V with a specific energy that goes between 100 Wh kg-1 and 150 Wh kg-1. Their more conventional configuration consists of a graphite anode, (e.g. MCMB, mesocarbon microbeads), a cathode formed by lithium metal oxide (LiMO2) (e.g. LiCoO2) and an electrolytic solution like the conventional LP30 that is an 1M solution of lithium hexafluorophosphate (LiPF6) in a 1 to 1 mixture by weight of ethylene and dimethyl carbonate (EC: DMC 1:1 by weight) [20].

Fig. 1.4 - Schematic of a common lithium-ion accumulator

Fig. 1.4 [2] shows a typical configuration of a lithium-ion cell. In most cases, these kind of cells are schematized as C/LiPF6 EC-DMC/LiMO2 and operate according to the process:

yC + LiMO2 LixCy + Li(1-x) MO2 with x ~ 0.5, y = 6, cell voltage ~3.7V

which is based on extraction and reversible insertion of lithium ions between the two electrodes with concomitant removal and addition of electrons. At first sight, the electrochemical process that guides the lithium-ion looks very simple and appears to consist of a reversible exchange of lithium ions between two electrodes. Actually, there some key points to consider. In Fig. 1.5 (A) we can see that the redox process at the carbonaceous anode (MCMB) occurs at about 0.05V vs. Li+/Li, while the one of LiCoO2 at cathode takes place at approximately 4V vs. Li+/Li. The current density increase in the electrolyte reveals phenomena of reduction and oxidation that define its domain of stability. The figure shows that this domain extends from about 0.8 to about 4.5 V vs. Li+/Li and that the anode works well outside of this electrolyte window of stability, while the cathode operates on the upper limit. This is clearly visible in Fig. 1.5 (B), which shows the

8

anode and cathode voltage ranges compared to the window of stability of the electrolyte. It can be concluded that the C/LiCoO2 battery is thermodynamically unstable in these conventional electrolytes [2].

Fig. 1.5 -(A) Profile of cyclic voltammetry of the lithium-ion battery components: anode and cathode (green), electrolyte (blue). (B) Voltage range of operation of the C/LiCoO2 electrodes

compared with the domain of stability of the most common liquid electrolytes However it happens that the battery operates under kinetic control but not under thermodynamic control. In fact, the initial decomposition of the electrolyte results in the formation of a protective film to the surface of the anode, which is able to ensure the continuity of operations of the battery in its charge/discharge cycles. This film is known as solid-electrolyte interphase (SEI). Instead, the oxidative processes that occur at the positive electrode are very dangerous. In standard operating conditions the battery operates below the limit of oxidation of the electrolyte; however, in some conditions, like accidental phenomena of overcharging, it could happen that this limit is exceeded, and, since no protective films have formed on the cathode, the oxidation of the electrolyte will continue up to determine the failure of the cell. Both anode and cathode decomposition processes imply a consumption of active masses of electrolyte and also the generation and evolution of gaseous substances, as shown in Fig. 1.6 [2-21].

9

Fig. 1.6 - Operating principle of SEI formation in a C/LiCoO2 cell That causes a loss of capacity (initial irreversible capacity) and a not negligible security risk. In order to control and reduce those effects, improved manufacturing processes have been developed. Improvements of Li-ion technology The adopted strategy for improving the technology of Li-ion devices in terms of safety, costs, materials availability, lifetime and operating temperature interval has basically been focusing on modifying the chemical nature of its main components. Two main approaches have been pursued in literature: substituting graphite and LiCoO2 with alternatives anodic and cathodic materials, having a greater capacity and lower cost; substituting or modifying conventional electrolytes by the addition of additives, in order to reduce their safety issues.

1.3.1. Electrolytes The electrolyte is one of the three main components of an electrochemical cell; it is the medium through which occurs the transfer of ions between the electrodes. The electrolytes being used within rechargeable Li-ion batteries fall in three main categories:

non-aqueous liquid electrolytes polymer electrolytes solid electrolytes

Due to the low conductivity of solid and polymer electrolytes, liquid electrolytes are used in most devices and are usually made of a lithium salt dissolved in a mixture of non-aqueous solvents. The most important requirements that the electrolyte in a battery has to meet are:

10

1. High ionic conductivity (i.e. > 10-4 S cm-1) and very low electronic conductivity, in order to minimize the internal resistance of the medium and therefore reduce any heat generation;

2. Wide electrochemical stability window, making the medium capable of resisting to high potential difference (3.7V);

3. High chemical and thermal stability, in order to prevent the decomposition of the electrolyte itself;

4. Low reactivity to the other cell components, like electrodes, separators and current collectors;

5. Non-toxicity, ease of handling and recycling; 6. Low melting point in order to ensure a good conductivity even at low

temperatures; 7. High boiling point to prevent explosions resulting from potential overpressures

within the cell; 8. Low-cost, in order to compete with the already existing energy storage devices

that use aqueous electrolytes.

Solvents An ideal electrolyte solvent has the following properties: 1. High dielectric constant (ε) in order to dissolve the salt present in it; 2. It must be aprotic and stable to negative potentials; 3. Low viscosity (η); 4. Must be inert towards the other components of the cell; 5. Low melting point and high boiling point; 6. Must be safe, non-toxic and inexpensive. The most important features that affect the ionic conductivity of the solvent are definitely the viscosity and the dielectric constant. Many types of organic solvents for electrochemical applications were investigated and most of them can be classified into three categories: alkyl carbonates, esters and ethers.

11

Tab. 1-1 - Carbonates and organic esters used as electrolyte solvents

Tab. 1-2 - Organic ethers used as electrolyte solvents

Tab. 1-1 and

Tab. 1-2 [22] show a list of the various solvents used in lithium batteries and their physical properties. The values in the tables show that no solvent, if taken absolute, is able to meet the necessary requirements for lithium-ion devices applications. For example, all acyclic ethers and esters have low values of viscosity (η ≈ 0.3-0,7 cP) and at the same time are characterized by low values of dielectric constant (ε ≈ 2-7), which makes them inadequate means for the dissolution of the salts; the cyclic esters, instead, have very high dielectric constants, because of their cyclic structure that helps to maintain a more ordered alignment of the molecule dipoles. On the other hand, solvents with high polarity (ε ≈ 40-90) such as ethylene carbonate (EC) have high values of viscosity (η ≈ 1.7-2.0 cP) which obviously goes to affect the ionic conductivity of the solution as it restricts the mobility ions.

12

To overcome this problem due to the concomitant presence of two phenomena which act in opposite directions, mixtures of solvents are often used. For example, two different solvents may be used in a mixture, one with a high dielectric constant and one with a low viscosity.

1.3.2. Electrodes

Anodes Carbon is the material currently used as an anode in the current generation of lithium battery. In fact, its light weight and its electrochemical potential close to that of lithium metal have attracted interest in it as an anode material [23]. It has a theoretical capacity of 372 mAhg-1, which corresponds to the insertion of a lithium ion every six carbon atoms (x = 1 in LixC6). One of the great disadvantages in using this material as the anode is the occurrence of a significant irreversible capacity during the first charge-discharge cycle. During the first electrochemical intercalation of lithium in the carbon, a some lithium is irreversibly consumed for the formation of the solid-electrolyte interface (SEI) and therefore can not be recovered in the discharge phase with consequent loss of capacity. The characteristics of the SEI of course depend on the type of electrolyte and the type of carbonaceous material used in the anodes. When the film turns out to be thick enough to prevent the tunneling of the electrons, the reduction of the electrolyte is suppressed and the electrode can reversibly cycle. The capacity of the second cycle and the subsequent ones is approximately the same and the intercalation during the step of charging and discharging is approximately 100% reversible. The carbonaceous materials used in the cells are generally sorted into three categories: graphite, soft carbon and hard carbon. Their structure is shown in Fig. 1.7 [24].

Fig. 1.7 - 3 types of carbon used in Li-ion batteries

The graphite structure is basically regular and is appear as a series of thin carbon layers. The carbon atoms forming these graphene layers are interconnected by covalent bonds between the layers, while Van der Waals forces keep layers linked to each other. In the charging phase, the lithium ions enter between the layers and form lithiated carbon, as shown in Fig. 1.8 [24]. The soft carbon is a material that has a discrete crystallinity (not perfect as the on of graphite) and a preferred orientation similar to graphite. The hard carbon, instead, is not very crystalline and consists of small randomly oriented aggregates and amorphous areas, containing small irregular

13

spaces. Each crystal has a layered structure, like as small broken pieces of graphite, which is able to hold lithium ions.

Fig. 1.8 - Model of intercalation of lithium ions in carbon materials

The distance between adjacent layers in the hard carbon is greater than in the graphite. Thus, hard carbon has the advantage of being able to have a greater capacity than the graphite. This is thought to be due to the fact that the hard carbon has the capability of holding lithium ions into both sides of the graphene sheet as well as in extra cavities and edges and faces with high surface area. Anyway, in order to improve the specific energy of the new lithium-ion batteries, the strategy today is to replace the common anodes made of carbonaceous material with metal and lithium alloys, such as alloys with silicon (Li-Si) or tin (Li-Sn). These alloys have a specific energy that far exceeds that of the Li-C (370 mAhg-1) and are respectively 4000 mAhg-1 for (Li-Si) and 990 mAhg-1 (Li-Sn) [2]. Unfortunately, however, these alloys can not be used as they are in batteries because of their strong deformation during the cycles of charging and discharging due to the process of deintercalation and intercalation of lithium in a short time, causing the disintegration of the electrodes [25]. The problem has been circumvented by optimizing the morphology of the electrode through the development of nanostructured configurations that support deformation and mechanical stress, thus ensuring to the battery as well as a high capacity also excellent stability in terms of cyclability. A good example of this type of anode material is an electrode structure based on tin and coal nanocomposites [26-27]. In this nanostructure, the carbonaceous matrix plays two roles: it provides sufficient free volume to contain the movement of tin expansion-constriction and acts as protection against electrode nanopowder whose handling is dangerous. In Fig. 1.9 (A) the average distribution of the tin particles within the carbonaceous matrix is shown, while Fig. 1.9 (B) describes the excellent stability of this anode, which shows no significant loss of capacity even after 200 cycles [2].

14

Fig. 1.9 - (A) TEM image of the Sn-C electrode morphology; (B) Galvanostatic cycling of the

Li/LP30/Sn-C cell

Research for new anode materials is also addressed to titanium oxides. In this range of materials, anatase titanium oxide TiO2 (TO) [28] and lithium titanium oxide, Li 4TI5O12 (LTO) [29] are attractive negative electrodes for advanced lithium ion batteries. The lithium insertion potential of these oxides is between 1.2V and 2.0V vs. Li, i.e. within the stability window of common organic electrolytes. LTO has a lithium-rich, spinel-framework structure. This electrode material is characterized by a two-phase electrochemical process evolving with a flat voltage profile. The theoretical capacity is lower, and the voltage level higher, than those of conventional graphite, i.e. 170mAhg−1 versus 370mAhg−1 and 1.5V vs. Li versus 0.05V vs. Li, respectively. Both differences may result in lower specific energy; however, the interest in Li 4TI5O12 remains high because of its specific properties that include: (i) a very low volume change (<1%) during cycling, which leads to high cycling stability; (ii) no electrolyte decomposition and thus, no SEI formation, (iii) high rate and very low temperature charge/discharge capability and (iv) high thermal stability in both the charged and discharged state. Indeed, LTO is presently practically exploited to develop batteries for PHEVs [30]. Cathodes The usage of olivine structure systems (LiMePO4) as cathode materials in lithium-ion electrochemical cells is currently being evaluated with great interest; they would replace the conventional LiCoO2, that has very high costs and also presents safety issues related to the possible release of oxygen. This class of compounds offers a theoretical specific capacity reasonably high (see Tab. 1-3), associated with a high working potential (see Tab. 1-4) [31-32], mainly due to the presence of polyanions (PO4)

3, which is kept constant during the entire electrochemical redox process. During the first oxidation reaction of these compounds (device charge), one lithium ion per formula unit is transferred to the anode and the oxidation of LiMePO4 to MePO4 induces a reduction of volume, which compensates for the volumetric

15

expansion of the carbon-based anode, contributing to an efficient use of the volume of the cell within a real device.

LiMePO4 , Me Theoretical specific capacity/mAhg-1

Fe 169.9

Mn 170.9

Co 166.7

Tab. 1-3 - Theoretical specific capacity for some LiMePO4

Redox pair Potential vs Li+/Li/V

Fe3+/Fe2+ 3.4

Mn3+/Mn2+ 4.1

Co3+/Co2+ 4.8

Tab. 1-4 - Working potential for some LiMePO4

Some researches focus on non-lithiated materials, which are present in nature in the form of various minerals and whose synthesis is extremely simple. Amorphous and crystalline materials have been studied, both anhydrous and hydrated. The FePO4 is a promising compound, since it has a high theoretical specific capacity, greater when anhydrous, since the molecular weight is lower in that case [33]. The LiFePO4 (used as a cathode material in the present work) is not toxic, is chemically stable in most organic solvents typically used in the field of lithium batteries and is able to operate in a wide temperature range. It shows high capacity (169.9 mAhg-1) and theoretical specific energy (≈ 580 WhKg-1) and a constant potential during the charge-discharge process (3.45 V vs. Li+/Li). The electrochemical activity of the Fe3+/Fe2+ pair is based on the biphasic reaction LiFePO4 ↔ FePO4 + Li+ + e -, which a phase transition of the first order is associated to [34].

Fig. 1.10 - Crystal structure of LiFePO4

16

The crystalline structure is olivine-type (see Fig. 1.10), with an orthorhombic unit cell, able to accommodate 4 LiFePO4 units. The olivine contains a compact hexagonal packing of anions slightly distorted, with half of the octahedral sites and one eighth of the tetrahedral ones occupied by cations. Two types of octahedral sites can be identified from the energetic point of view, usually occupied by different cations. The lithium atoms occupy chains of octahedra, which share edges and are parallel to the axis c, while the iron atoms are arranged in a zig-zag along chains of octahedra, which share corners and are parallel to the axis c, along the other a-c floors. The a-c plans containing the lithium atoms are connected to the PO4 tetrahedra, which form a three-dimensional structure with strong bonds and which reduce the free volume available for the lithium ions. All of this allows the diffusion of lithium along bidimensional paths. The electrochemical extraction of lithium from such material allows to obtain FePO4 with the same structure, with only a small change in the reticular parameters. Stability under operating conditions at medium to high temperatures and low cost conclude the set of the valuable features of the proposed material. From a practical point of view the safety factor of a rechargeable system being put on the market cannot be overlooked. In fact, an important problem in the context of the batteries resides in the possibility of short-circuits in the lithium systems, which cause a local increase of temperature and induce release of oxygen by the cathode material. As a result the oxygen reacts with the organic electrolyte, leading to the development of gaseous substances. Such event is extremely dangerous and may be accompanied by small explosions. In LiFePO4 oxygen constitutes strong covalent bonds with phosphorus (P5

+) to achieve the tetrahedral (PO4)3-, from which the oxygen extraction

is extremely difficult. Currently the limitations to a practical use of the LiFePO4 compound are due to the fact that its preparation must be carried out under conditions which aim at obtaining a final product in which the Fe is located only in its oxidation state +2 (synthesis in an inert or reducing atmosphere) and, mainly, to its low electronic conductivity (10-9

÷ 10-10 Scm-1 at room temperature [35]). A low value of electronic conductivity greatly reduces the fraction of lithium ions which can be extracted and inserted in a reversible way during operation; that directly affects the value of the practical specific capacity obtainable. Anyway, in order to allow a good real capacity close to the theoretical one, methods of synthesis have been developed, which are based on a process of coating with coal that is able to increase the electronic conductivity. Another promising example in the manganese family is the lithium nickel manganese oxide, LiNi0.5Mn1.5O4, which adopts a spinel structure [36]. This material is characterized by a two-phase electrochemical process reflecting in a flat voltage profile evolving around 4.5V vs. Li. The theoretical specific capacity is 146mAhg−1, i.e. of the same order as the conventional lithium cobalt oxide, LiCoO2. However, the key difference is in the high operational voltage, which makes LiNi0.5Mn1.5O4 a very interesting material due to its potentiality of assuring substantial increase in energy density, such as 30% more than that associated with conventional lithium manganese spinel. On the other hand, the practical use of this cathode material may be prevented by the lack of suitable electrolyte media, since the presently available organic carbonate electrolyte solution are not totally compatible with the high voltage of the lithium nickel manganese oxide, especially in the course of its charge process. R&D

17

projects are under way to develop more stable electrolytes, so that to make this interesting cathode material viable for industrial use.

18

References

[1] B. Scrosati, Chem. Rec., 5 (2005) 286. [2] B. Scrosati, J. Garche, Journal of Power Sources, 195 (2010) 2419. [3] S. Brutti, V. Gentili, H. Menard, B. Scrosati, P.G. Bruce. Advanced Energy Materials, 2 (2012) 322. [4] S. Brutti, V. Gentili, P. Reale, L. Carbone, S. Panero, Journal of Power Sources, 196 (2011) 9792. [5] P.G. Balakrishnan, R. Ramesh, T.P. Kumar, Journal of Power Sources, 155 (2006) 401. [6] Macworld, 36 (1995) 12. [7] Los Angeles Times, September 15, 1995, p. 1D. [8] Dagens Industry, January 12, 1996, p. 10. [9] Standard for Lithium Batteries, UL 1642, third ed., Underwriters Laboratories, Northbrook, IL, 1995. [10] National Research Council, Review of the Research Program of the Partnership for a New Generation of Vehicles, National Academy Press, Washington, DC, 2000. [11] D. MacArthur, G.E. Blomgren, in: R. Powers (Ed.), Lithium Batteries: A Review and Analysis, 2000. [12] www.bccresearch.com, September 2004. [13] J.M. Tarascon & M. Armand, Nature, 414, (2001), 359. [14] J.D. Lee, Concise Inorganic Chemistry, 5th edition, 1996, Chapman & Hall, London. [15] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Applications, Wiley editor, 2000. [16] M. Lazzari, F. Bonino, B. Scrosati, La chimica e l’industria, 66 (1984) 406. [17] T. Nagaura and T. Tozawa, Progress in Batteries & Solar Cells, 9, 20 (1990). [18] E. Peled, Journal of Electrochemical Society, 126 (1979) 2047-2051. [19] J.R. Dahn, A. K. Sleight, Hang Shi, B. M. Way, W. J. Weydanz, J. N. Reimes, Q. Zhong and U. von Sacken, Lithium batteries, ed G. Pistoia, Elsevier Sequoia, London (1995) 97. [20] W. van Schalkwijk, B. Scrosati (Eds.), Advances in Lithium-ion Batteries, Kluwer Academic/Plenum, Boston, 2004. [21] Report of the Basic Energy Science Workshop on Electric Energy Storage, Department of Energy, USA ,http://www.sc.doe.gov/bes/reports/files/EES rpt. pdf. [22] K.Xu, Chem. Rev (2004), 104. [23] N.Imanishi, Y. Takeda, and O. Yamamoto, “Lithium-Ion Batteries: Fundamentals and Performance” (M. Wakihara and O. Yamamoto, Eds.), (1998), P. 98, Wiley-VCH, Weinheim. [24] T. Osaka, and M. Datta, Energy Storage System for Electronics, Gordon and Breach scince publishers, (2000). [25] M. Winter, J.O. Besenhard, Electrochim. Acta 45 (1999) 31. [26] G. Derrien, J. Hassoun, S. Panero, B. Scrosati, Adv. Mater. 19 (2007) 2336. [27] G. Derrien, J. Hassoun, S. Panero, B. Scrosati, Adv. Mater. 20 (2008) 3169. [28] A. Ohzuko, N. Ueda, N. Yamamoto, J. Electrochem. Soc. 142 (1995) 1431. [29] P. Kubiak, J. Geserick, N. Hüsing, M. Wohlfahrt-Mehrens, J. Power Sources 175

19

(2008) 510. [30] www.altairnano.com ; www.enerl.com [31] A. Yamada, S. C. Chung, Journal of The Electrochemical Society, 148 (8) (2001) A960-A967. [32] A. Yamada, M. Hosoya, S. Chung, Y. Kudo, K. Hinokuma, K. Liu, Y. Nishi, Journal of Power Sources 119-121 (2003) 232-238. [33] F. Croce, A. D’Epifanio, P. Reale, L. Settimi, and B. Scrosati, Journal of Electrochemical Society 150 (5) (2003) A576-A581. [34] A.K. Padhi, N.S. Nanjundaswamy, and J.B. Goodenough, Journal of Electrochemical Society, 144 (4) (1997) 1188-1193. [35] S.Y. Chung, J.T. Blocking, and Y.M. Chiang, Nat. Mater. 1 (2002) 123. [36] J. Hassoun, P. Reale, B. Scrosati, J. Mater. Chem. 178 (2007) 3668.

20

2. Innovative electrolytes – State of the art 2.1. Ionic Liquids (ILs)

ILs are salts with melting points below 100°C [1], that are liquid over a wide temperature range. Although Seddon et al. [2] demonstrated that some ILs can evaporate under high vacuum at high temperature, the vapour pressure of ILs is usually negligible under ambient conditions and they are liquid up to their decomposition temperature) some can exceed 300°C). These unique physicochemical properties makes ILs very promising materials to replace traditional volatile organic solvents in several applications, such as synthesis [3-8], catalysis [9-11] and purification [12-13]. Begin composed of only ions, the ILs have high ion density, which provides high ionic conductivity [14]. This property, coupled with the wide electrochemical stability window, have led to the use of ILs as solvents in electrodeposition [15], and electro-polymerization [16], and as electrolytes in electrochemical devices such as batteries [17-21], capacitors [22-26], fuel cells [27-30], and solar cells [31-33]. The most attractive feature of ILs is the unlimited possibility of designing the component ions, which allows for fine-tuning the physical and chemical properties of the resulting IL. It should be recognized that the relative toxicity of ILs depends on their exact ionic composition. Full toxicological investigations have yet to be carried out, however the generation of potential hazardous decomposition products (such as HF) has been reported during the purification of PF6-based ILs [34]. The thorough comprehension of the effects of the combination of cation and anion used, or of the modification of the alkyl chains on the cation is therefore a fundamental tool for fine tuning the IL final properties. Due to the high variety of ILs that have been developed, a complete classification doesn’t exist yet; one of the most used categorization is based on the ILs cationic structure, as shown in Fig. 2.1.

Fig. 2.1 – ILs classification based on cationic structure [61]

21

The non-flammability and non-volatility of ILs are appealing properties in view of their use as high safety electrolyte materials. Moreover, recent studies have shown that the electrochemical stability window can be up to 5V [35]. ILs have thus been suggested to be suitable for electrochemical applications such as Li-ion batteries [36,37,17-21,38,39], solar cells [31,32,33], supercapacitors [40,41] and, more recently, fuel cells [27-30]. ILs for lithium batteries applications Thanks to their unique properties, which include low vapour pressure, non-flammability, high ion conductivity as well as high thermal and electrochemical stability, room temperature ionic liquids (ILs) are considered as very promising electrolyte media for a variety of electrochemical devices [42–44]. For instance, great attention is currently directed to various derivatives of ILs formed by bulky organic cations and highly delocalized-charge inorganic anions as new electrolytes for lithium batteries, since these materials offer promises of reducing the safety hazards which still limit the range of application of these batteries [45–46]. the challenge is to widen the cathodic limit such as to allow stability towards the lithium metal electrode. Indeed, the 1-alkyl-3-alkyl-imidazolium-based ILs, in addition to a high ionic conductivity, have other favorable properties, which in principle make them very appealing electrolyte materials [47]. However, the use of these ILs is still prevented by a major drawback, namely a poor cathodic electrochemical stability, caused by attacks to hydrogen at C(2) carbon site. This issue can in part be controlled by alkyl substitutions at the C(2) carbon site; however, this results in a serious decay in the ionic conductivity which rules the ILs out from battery interest [47]. Thus, the challenge is to widen the cathodic limit such as to allow stability towards the lithium metal electrode without sacrificing the ionic conductivity. Success has been recently obtained by using IL systems based on Py cations, e.g. 1-alkyl-1-methylpyrrolidinium cations [45,48-49]. Indeed, recent electrochemical studies have demonstrated that solutions of lithium N,N-bis(trifluoromethanesulfonyl)imide in a N-n-butyl-N-ethylpyrrolidinium N,N-bis(trifluoromethanesulfonyl)imide (Py24TFSI-LiTFSI) have a good compatibility with the lithium metal electrode still maintaining a high thermal stability and an acceptable ionic conductivity [46]. In view of the progress of the lithium battery technology, it is highly desirable to pass from a liquid to a polymer structure, since this design offers expectance of improvements in safety and reliability [50–51]. Consequently, it is of interest to extend this strategy also to the IL electrolyte cases by developing IL-containing membranes.

2.2. “Fluorine-free” salt-based electrolytes

Another valid approach to enhance safety in lithium ion batteries is to partially or completely substitute the common fluorine salt as LiPF6, with safer alternative “fluorine-free” lithium salt. Among new electrolyte salts, a very promising candidate is lithium bis(oxalato)borate (LiBOB) [52]. This material has been proposed both as exclusive electrolyte salt and as electrolyte additive and its compatibility with a large number of different electrode materials has widely been studied (e.g. [53]). It has been shown that LiBOB-based electrolytes have higher stability and higher safety with most

22

electrode materials (except for Co-rich cathodes) but also slightly lower rate capability and lower temperature performance than LiPF6-based electrolytes [54,55]. One important difference to LiPF6 is that LiBOB takes part in the formation of the solid electrolyte interphase (SEI) on the anode [56]. The electrode filming and SEI formation occur in several stages. Most striking is a process occurring at 1.75 V vs. Li+/Li which gives rise to additional irreversible capacity. There have been discussions whether this process is related with LiBOB itself or with an intrinsic impurity present in LiBOB electrolytes [54,57]. What is clear is that the irreversible capacity associated with the formation of the SEI, especially with the reduction process occurring at around 1.75 V vs. Li+/Li, is strongly dependent on the surface area and on the type of the carbon material involved in the electrode mass [54,58] and also on the nature of the electrolyte solvents [59].

2.3. Gel polymer electrolytes (GPEs)

The rapid growth of the miniature electronic and computer-related industries has led to great demand for smaller and lighter batteries with improved safety, energy and power characteristics. Lithium polymer batteries are expected to meet the above requirements and are thus considered as next-generation rechargeable batteries. The conventional lithium-ion batteries, which contain a large amount of liquid electrolyte, emit an appreciable amount of gas and this is attributed to the decomposition of a protective layer at the carbon surface. This phenomenon eventually leads the battery system to safety hazards. By virtue of their advantages such as high theoretical capacity, improved safety, lower material costs, ease of fabrication into flexible geometries, and the absence of electrolyte leakage, lithium polymer batteries have placed an unprecedented demand for battery researchers. Polymer electrolyte can be defined as a membrane that possesses transport properties comparable with that of common liquid ionic solutions. In principle, a polymer electrolyte battery can be formed by sandwiching the electrolyte between a lithium metal (or a composite carbon) anode and a composite cathode. Because of its rigid structure, the electrolyte can also serve as a separator. The prerequisites for a polymer electrolyte to be used in any battery system are: high ionic conductivity at ambient and subambient temperatures, good mechanical strength, appreciable transference number, good thermal and electrochemical stabilities, better compatibility with electrodes. Although the ionic conduction in polymer electrolyte was discovered by D. E. Fenton and coworkers in 1973, its technological importance was recognized only after a decade. Polymer electrolytes used in lithium batteries can be classified into three categories: dry polymer electrolytes, gel polymer electrolytes, and composite polymer electrolytes. The dry polymer electrolyte generally contains an alkali metal salt complexed with the polymer matrix. The very first example of “dry solid” polymer electrolyte is the poly(ethylene oxide) (PEO)-based system that showed very low ambient temperature conductivities on the order of 10-8 Scm–1. This system does not posses any organic liquid and thus the polymer host is used as a solid solvent. However, the cycling performance of this dry solid polymer electrolyte with lithium metal electrode was not satisfactory as the usage was as low as 200-300 cycles.

23

The second category of polymer electrolyte is called “gel polymer electrolyte” or “plasticized polymer electrolyte”, which is neither liquid nor solid. Gels posses both cohesive properties of solids and the diffusive property of liquids. This unique characteristic makes the gel to find various important applications including polymer electrolytes. One of the best approaches to increase the ionic conductivity is the addition of low-molecular-weight aprotic solvents such as ethylene carbonate (EC), propylene carbonate (PC), dimethyl carbonate (DMC), diethyl carbonate (DEC), and γ-butyrolactone, along with the lithium salt. The idea behind this is to increase the ionic mobility and concentrations of the charge carriers in the solid polymer electrolyte by enhancing the dissociation of the lithium salt. Plasticizers are low-molecular-weight non-volatile substances that, when added to a polymer, improve its flexibility and processability. The plasticizer significantly reduces the brittleness of many amorphous polymers glass transition temperature (Tg) of the polymer, which in turn reduces the cohesive forces of attraction between polymer chains. The relatively small plasticizer molecules penetrate into the polymer matrix, establish polar attractive forces between the polymer chains, and increase the segmental mobility, thereby reducing the Tg value. It is believed that the lithium ion is also solvated by the plasticizing molecule, and the gel polymer electrolyte approaches high room temperature ionic conductivities similar to those of the typical non-aqueous liquid electrolytes. Another approach in designing polymer electrolytes with improved electrical and mechanical properties at ambient temperatures is the incorporation of high surface area particulate inorganic fillers such as ZrO2, TiO2, Al2O3, and hydrophobic fumed silica and these are called “composite polymer electrolytes” or “composite ceramic electrolytes”. This composite electrolyte, on the contrary, is a subset of polymer electrolytes. The advantages of incorporating the fillers are twofold: one is the enhancement in ionic conductivity at low temperatures and the other is the improved stability at the interface with electrodes [60].

24

References [1] K.R. Seddon, A. Stark, M.-J. Torres, Pure Appl.Chem., 72 (2000), 2275 [2] M.J. Earle, J.M.S.S. Esperança, M.A. Gilea, J.N.C Lopes, L.P.N. Rebelo, J.W. Magee, K.R. Seddon, J.A. Widegren, Nature, 439 (2006), 831 [3] T. Welton, Chem Rev. 99 (1999), 2071 [4] H.Zhao, S.V. Malhotra, Aldrichim. Acta, 35 (2002), 75 [5] L.J. Xu, W.P. Chen, J.L. Xiao, Organometallics, 19 (2000), 1123 [6] A.J. Carmichael, M.J. Earle, J.D. Holbrey, P.B. McCormac, K.R. Seddon, Org. Lett., 1 (1999), 997 [7] R.A. Brown, P. Pollet, E. McKoon, C.A. Eckert, C.L. Liotta, P.G. Jessop, J. Am. Chem. Soc., 123 (2001) 1254 [8] J. Dupont, P.A.Z. Suarez, A.P. Umpierre, R.F. de Souza, J. Braz. Chem. Soc., 11 (2000) 293 [9] H. Olivier, J. Mol. Catal. A.: Chem., 146 (1999) 285 [10] T. Welton, Coord. Chem. Rev., 248 (2004) 2459 [11] P. Wassescheid, W. Keim, Angew. Chem. Int. Ed., 39 (2000) 3773 [12] S.G. Cull, J.D. Hlbrey, V. Vargas-Mora, K.R. Seddon, G.J. Lye, Biotechnol. Bioeng., 69 (2000) 227 [13] D.W. Armstrong, L.F. He, Y.S. Liu, Anal. Chem. 71 (1999) 3873 [14] H. Ohno, K. Fukumoto, Electrochem. 76 (2008) 16 [15] F. Endres, Chem. Phys. Chem. 139 (2002) 3103 [16] G.W. Nyce, T. Glauser, E.F. Connor, A. Mock, R.M. Waymouth, J.L. Hedrick, J. Am. Chem. Soc. 125 (2003) 3046 [17] H. Sakaebe, H. Matsumoto, K. Tatsumi, J. Power Sources 146 (2005) 693 [18] K. Hayashi, Y. Nemoto, K. Akuto, Y. Sakurai, J. Power Sources 146 (2005) 689 [19] N. Byrne, P.C. Howlett, D.R. MacFarlane, M. Forsyth, Adv. Mater. 17 (2005) 2497 [20] H. Sakaebe, H. Matsumoto, Electrochem. Commun. 5 (2003) 594 [21] D.R. MacFarlane, J.H. Huang, M. Forsyth, Nature 402 (1999) 792 [22] T. Sato, G. Masuda, K. Takagi, Electrochem. Commun. 49 (2004) 3603 [23] A. Lewandowski, A. Swiderska, Solid State Ionics 161 (2003) 243 [24] M. Ue, M. Takeda, A. Toriumi, A. Kominato, R. Hagiwara, Y. Ito, J. Electrochem. Soc. 150 (2003) A499 [25] A.B. McEwen, E.L. Ngo, K. LeCompte, J.L. Goldman, J. Electrochem Soc. 146 (1999) 1687 [26] C. Nanjundiah, S.F. McDevitt,, V.R. Koch, J. Electrochem. Soc. 144 (1997) 3392 [27] R.F. de Souza, J.C. Padilha, R.S. Goncalves, J.Dupont, Electrochem. Commun. 5 (2003) 728 [28] M. Yoshizawa, W. Xu, C.A. Angell, J. Am. Chem. Soc. 125 (2003) 15411 [29] J.H. Shin, W.A. Henderson, G.B. Appetecchi, F. Alessandrini, S. Passerini, Electrochem. Commun. 50 (2005) 3859 [30] M. Doyle, S.K. Choi, G. Proulx, J. Electrochem. Soc. 147 (2000) 34 [31] W. Kubo, S. Kambe, S. Nakade, T. Kitamura, K. Hanabusa, Y. Wada, S. Yanagida, J. Phys. Chem. B. 107 (2003) 4374 [32] P.Wang, S.M. Zakeeruddin, P. Comte, I. Exnar, M. Gratzel, J. Am. Chem. Soc. 125 (2003) 1166

25

[33] H. Matsumoto, T. Matsuda, T. Tsuda, R. Hagiwara, Y. Ito, Y. Miyazaki, Chem. Lett. 26 (2001) [34] R. P. Swatloski, J.D. Holbrey, R.D. Rogers, Green Chem. 5 (2003) 361 [35] K. Matsumoto, R. Hagiwara, Y.Ito, Electrochem. Solid-State Lett. 7 (2004) E41 [36] T. Sato, T. Maruo, S. Marukane, K. Takagi, J. Power Sources, 138 (2004) 253 [37] P.C. Howlett, D.R. MacFarlane, A.F. Hollenkamp, Electrochem. Solid-State Lett. 7 (2004) A97 [38] V.R. Koch, C. Nanjundiah, G.B. Appetecchi, B. Scrosati, J. Electrochem. Soc. 142 (1995) L116 [39] H. Nakagawa, S. Izuchi, K. Kuwana, T. Nukuda, Y. Aihara, J. Electrochem Soc. 150 (2003) A65 [40] T. Sato, S. Marukane, T. Narutomi, T. Akao, J. Power Sources 164 (2007) 390 [41] U. Makoto, T. Masayuki, T. Takako, T. Masahiro, Electrochem. Solid-State Lett. 5 (2002) A119 [42] H. Ohno, Electrochemical Aspects of Ionic Liquids, Wiley, New York, 2005. [43] A. Fernicola, B. Scrosati, H. Ohno, Ionics 12 (2006) 95. [44] M. Galinski, A. Lewandowski, I. Stepniak, Electrochim. Acta 51 (2006) 5567. [45] G.B. Appetecchi, S. Scacia, C. Tizzani, F. Alessandrini, S. Passerini, Electrochem. Soc. 153 (2006) A1685. [46] A. Fernicola, F. Croce, B. Scrosati, T. Watanabe, H. Ohno, J. Power Sources 41 (2007) 348. [47] S. Seki, Y. Kobayashi, H. Miyashiro, Y. Ohno, A. Usami, Y. Mita, N. Kihira, M. Watanabe, N. Terada, J. Phys. Chem. B 110 (2006) 10228. [48] J. Salminen, N. Papiconomou, R.A. Kumar, J.-M. Lee, J. Kerr, J. Newman, J.M. Prausnitz, Fluid Phase Equilibria 261 (2007) 421. [49] D.R. MacFarlane, P. Meakin, J. Sun, N. Amini, M. Forsyth, J. Phys. Chem. B 103 (1999) 4164. [50] D.R. MacFarlane, J. Sun, J. Golding, P. Meakin, M. Forsyth, Electrochimica. Acta. 45 (2000) 1271. [51] J.Y. Song, Y.Y. Wang, C.C. Wan, J. Power Sources 77 (1999) 183. [52] J. Hassoun, M. Wachtler, M. Wohlfahrt-Mehrens, B. Scrosati, Journal of Power Sources, 196 (2011) 349-354 [53] Z.H. Chen, W.Q. Lu, J. Liu, K. Amine, Electrochim. Acta 51 (2006) 3322. [54] M. Wachtler, M. Wohlfahrt-Mehrens, S. Ströbele, J.-C. Panitz, U. Wietelmann, J. Appl. Electrochem. 36 (2006) 1199. [55] K. Xu, B. Deveney, K. Nechev, Y.F. Lam, R.T. Jow, J. Electrochem. Soc. 155 (2008) A959. [56] K. Xu, S.S. Zhang, T.R. Jow, Electrochem. Solid-State Lett. 6 (2003) A117. [57] K. Xu, S.S. Zhang, R. Jow, J. Power Sources 143 (2005) 197. [58] J.-C. Panitz, U. Wietelmann, M. Wachtler, S. Ströbele, M. Wohlfahrt-Mehrens, J. Power Sources 153 (2006) 396. [59] K. Xu, J. Electrochem. Soc. 155 (2008) A733. [60] A.M. Stephan, S. Thomas, Enciclopedia of Electrochemical Power Sources, vol. II, 140 [61] T. Tsuda, C.L. Hussey, Electrochemical Society Interface, (2007) 42-49

26

3. ILs electrolytes 3.1. Pyrrolidinium-based ILs

The most attractive feature of ILs is the unlimited possibility of designing the component ions, which allows for fine-tuning the physical and chemical properties of the resulting IL. The deep comprehension of the effects of the combination of cation and anion used, or of the modification of the alkyl chain on the cation is therefore a fundamental tool for fine-tuning the IL final properties. Provided a fixed anion, the cation structure influences properties such as the viscosity and the ion association degree, the latter controlling the low vapour pressure typical of the ILs. It has been found that the longer the aliphatic chains attached to the cation, the higher the viscosity is, due to the increase of Van der Waals interactions among larger aliphatic moieties [1,2]. Apart from the dimensions of the cation and the anion moieties, the symmetry of the ion structures is also fundamental in determining the glass transition temperature, the melting point and the conductivity of the ILs. Depending on the synthesis, it is also possible to obtain protic or aprotic molten salts. This flexibility in the molecular design of ILs, associated to the large number of combinations of available cations and ions, may allow the development of ILs displaying electrochemical stability windows suitable for applications in lithium batteries. The quaternary ammonium ions produce low melting point salts compared with many of the organic salt of the same anions. In this research project IL based on a quaternary ammonium cation, i.e. pyrrolidinium, has been synthesized. In particular the N-butyl-N-ethylpyrrolidinium cation (Py24

+) was selected due to its asymmetric big structure (two alkyl chain with different chain length attached on the N atom) and localized charge on the N atom stabilized by an inductive effect of the alkyl chain and protected by their steric hindrance. The selected anion is a low Lewis base, bis(trifluoromethansulfonyl)imide (TFSI-), chosen due to its good properties in terms of lithium ion transfer, but also to the flexibility (many conformational and rotational degrees of freedom) and to the delocalized charge (protected in the ion inner part). A commercial pyrrolidinium-based IL was also characterized in this research project i.e. N-butyl-N-methylpyrrolidinium bis(trifluoromethansulfonyl)imide (Py14TFSI) from Solvionic. Due to assure lithium conduction, in the ILs was dissolved a lithium salt, i.e. lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) having the same anion of the ILs choosing 0.2 mol Kg-1 as the preferred concentration [3]. This salt was chosen because of its good properties in terms of safety, thermal stability, conductivity, stability to the oxidation (until 5V ca. vs. Li). Thanks to these advantageous properties LiTFSI can substitute the other conventional lithium salt like Li-triflate (low conductive), LiClO4 (hazardous), LiBF4 and LiPF6 (termically instable), and LiAsF6 (toxic).

27

3.2. Py24TFSI: synthesis

In order to obtain an ionic liquid suitable for application as electrolyte in lithium batteries, we synthesized a non commercial ionic liquid, N-butyl-N-ethylpyrrolidinium bis(trifluoromethansulfonyl)imide (Py24TFSI). The ionic liquid here developed i.e. N-butyl-N-ethylpyrrolidinium bis(trifluoromethanesulfonyl)imide was synthesized in two steps according to the scheme in figures Fig. 3.1 and Fig. 3.2. First step is the alchilation: 1-butylpirrolidine (Py4) and 1-bromoethane (EtBr) are mixed in ethyilacetate (EtAc) and stirred at 50°C for 2 days. The solvent is evaporated, leaving the solid bromide salt that, after recrystallization in a suitable mixture of ethylacetate and ethanol, appears as a white solid. This solid is filtered and dried under vacuum at 70°C for 4 hours.

Br+

N

N

Br -

Br+

N

N

Br -

Br+

N

N

Br -

N

Br - Fig. 3.1- First synthesis step and a picture of the obtained bromide

Second step is the anionic exchange: equimolar amounts of the purified dried bromide and of LiTFSI are than dissolved and mixed in distilled water and stirred at room temperature for 1 hour, during which the Py24TFSI is formed. The IL phase spontaneously separates from the aqueous phase, and repeatedly (7-8 times) rinsed with fresh water to remove LiBr impurities. The total removal of the bromides is checked by silver nitrate test (AgNO3 2M). The Py24TFSI is then purified by consecutive treatments with activated carbon (Activated Charcoal NoritTM type DarcoTM G60, Fluka) at 70°C for 10 hours, and with Al2O3 (Aluminum oxide, activated, acidic, Brockmann I, Sigma-Aldrich) at room T for 5 hours in order to remove organic and water residues, respectively. The final product is dried under vacuum at 80°C for 12 hours and finally stored in a glove box [4,5]

Py24TFSI

N

Br -

+ CF3

S SCF3N

O

O

O

OLi

N

CF3S S

CF3N

O

O

O

O-

+ LiBr Py24TFSI

N

Br -

+ CF3

S SCF3N

O

O

O

OLi

N

Br -

N

Br -

+ CF3

S SCF3N

O

O

O

OLiCF3

S SCF3N

O

O

O

OCF3

S SCF3N

O

O

O

OLi

N

CF3S S

CF3N

O

O

O

OCF3

S SCF3N

O

O

O

O-

+ LiBr Py24TFSI

N

Br -

+ CF3

S SCF3N

O

O

O

OLi

N

CF3S S

CF3N

O

O

O

O-

+ LiBr Py24TFSI

N

Br -

+ CF3

S SCF3N

O

O

O

OLi

N

Br -

N

Br -

+ CF3

S SCF3N

O

O

O

OLiCF3

S SCF3N

O

O

O

OCF3

S SCF3N

O

O

O

OLi

N

CF3S S

CF3N

O

O

O

OCF3

S SCF3N

O

O

O

O-

+ LiBr

Fig. 3.2 – Second synthesis step and a picture of the final product (IL)

28

3.3. Py24TFSI and Py14TFSI: characterization

The two selected ILs, i.e. the synthesized Py24TFSI and the commercial Py14TFSI, have been studied both pure and in solution with the LiTFSI salt. In order to understand the potentiality of these ILs for lithium batteries applications, the samples have been characterized from a thermal and an electrochemical point of view and finally their performances in lithium cells have been investigated.

3.3.1. Thermal characterization: DSC, TGA Before any characterization the water content of both the ILs was measured by a standard Karl Fischer titration method (Metrohm KF 831 Coulometer), resulting to be about 20 ppm for the commercial Py14TFSI and even lower ppm for the synthesized Py24TFSI. These very low values are acceptable for lithium battery grade solution [6]. Thermal properties were detected by using differential scanning calorimetry (DSC 821 Mettler-Toledo) and thermal gravimetric analysis (TGA/SDTA 851 Mettler-Toledo), at 5 °C/min. The DSC measurements were performed by quenching the samples and then slowly heating them up to 80°C. Two consecutive temperature scans were performed, in order to reveal peaks relative to structural rearrangements. The results reported below have been evaluated by analyzing the second scan traces.

-80 -60 -40 -20 0 20 40 60 80-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5

Hea

t flo

w,

Wg

-1

Temperature, °C

Py24

TFSI

-80 -60 -40 -20 0 20 40 60 80-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5 Py14TFSI (commercial product)

Temperature, °C

Hea

t flo

w,

Wg

-1

Fig. 3.3 - DSC heating trace of the

synthesized Py24TFSI Fig. 3.4 - DSC heating trace of the

commercial Py14TFSI The DSC trace of the Py24TFSI, reported in the Fig. 3.3, clearly evidences an exothermic peak related to the crystallization at – 48 °C and the melting transition at -40°C. The DSC response of the Py14TFSI, reported in the Fig. 3.4, is very similar to the response obtained with the Py24TFSI, showing a crystallization at -50°C (the peak is higher and smaller than the crystallization peak of the Py24TFSI, but the area is very similar) and the melting point at -15°C. The only difference is in the small endothermic peak at about -30°C, related to the melting transition, that is probably due to the different steric hindrance. In any case the melting points are below room temperature, thus they may be classified as room temperature ionic liquids (RTILs). In the case of the solutions the addition of an amount of salt, corresponding to a 0.2 mol kg-1 solution, strongly modifies the DSC response. The ionic solution Py24TFSI-

29

LiTFSI 0.2m, as reported in Fig. 3.5, does not show the transitions detected for the neat IL. Very similar response, here not reported, is showed in the case of the neat Py14TFSI compared to the Py14TFSI- LiTFSI 0.2m.

Pes

o, %

-100 -80 -60 -40 -20 0 20 40 60 80-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

Flu

sso

di c

alor

e, W

g-1

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

EN

DO

5°C/min

Temperature, °C

Hea

t flo

w, W

g-1

Pes

o, %

-100 -80 -60 -40 -20 0 20 40 60 80-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

Flu

sso

di c

alor

e, W

g-1

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

EN

DO Pes

o, %

-100 -80 -60 -40 -20 0 20 40 60 80-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

Flu

sso

di c

alor

e, W

g-1

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

Pes

o, %

-100 -80 -60 -40 -20 0 20 40 60 80-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

Flu

sso

di c

alor

e, W

g-1

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

EN

DO

5°C/min

Temperature, °C

Hea

t flo

w, W

g-1

Fig. 3.5 - DSC heating traces of the neat Py24TFSI and of the Py24TFSI-LiTFSI 0.2m solution

The change in thermal behaviour between the neat IL and its LiTFSI solution is due to the ion aggregation caused by the relatively strong coordination of the anion by the Li ion. Based on previous studies the 0.2 mol kg-1 concentration was selected for all the IL solutions [7]. The TGA responses, reported in Fig. 3.6, show a comparable behaviour for the two ILs under investigation, with the temperature corresponding to the starting point of the decomposition process a little bit higher for the commercial Py14TFSI (330°C for the Py14TFSI and 300°C for the Py24TFSI).

100 200 300 400 500 6000

20

40

60

80

10024

Wei

ght,

%

Temperature, °C

Py14TFSI

Py24TFSI

100 200 300 400 500 6000

20

40

60

80

10024

Wei

ght,

%

Temperature, °C

Py14TFSI

Py24TFSI

Fig. 3.6 – TGA traces of the synthesized Py24TFSI and the commercial Py14TFSI

The TGA traces of the ionic solutions are compared to the traces obtained by testing the neat ionic liquids. In Fig. 3.7 are reported TGA traces of Py24TFSI and of Py24TFSI-

30

LiTFSI and is possible to note that the temperature corresponding to the starting point of the decomposition process is a little bit higher for the solution due to the occurrence of strong stabilizing interactions, but in both cases the values are very high (300-350°C).

50 100 150 200 250 300 350 400 450 5000

20

40

60

80

100

Pes

o, %

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

5°C/min

Temperature, °C

Wei

ght,

%

50 100 150 200 250 300 350 400 450 5000

20

40

60

80

100

Pes

o, %

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

5°C/min

50 100 150 200 250 300 350 400 450 5000

20

40

60

80

100

Pes

o, %

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

50 100 150 200 250 300 350 400 450 5000

20

40

60

80

100

Pes

o, %

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

50 100 150 200 250 300 350 400 450 5000

20

40

60

80

100

Pes

o, %

Temperatura, °C

Py24

TFSI

Py24

TFSI - LiTFSI 0.2m

5°C/min

Temperature, °C

Wei

ght,

%

Fig. 3.7 - TGA traces traces of the neat Py24TFSI and

of the Py24TFSI-LiTFSI 0.2m solution

3.3.2. Electrochemical characterization: conductivi ty test, cathodic cyclic voltammetry and anodic linear scan The ionic conductivity was evaluated by impedance spectroscopy, performed using FRA 1255 Solartron on a two Pt-electrode cell (Amel product). The ionic conductivity values reported in the Fig. 3.8 are all of the same order of magnitude. The two IL-salt solution showed values lower than the corresponding neat ILs due because of their higher viscosity.

0,0

0,5

1,0

1,5

2,0

2,5

3,03.17

1.541.53

2.09

Py14TFSI Py14

TFSILiTFSI 0.2m

Py24

TFSILiTFSI 0.2m

Py24

TFSI

Ioni

c C

ondu

ctiv

ity, m

Scm

-1

Fig. 3.8 – Ionic Conductivity at room temperature

of all the samples

Cyclic cathodic voltammetry measurements and linear anodic scans were performed to evaluate the electrochemical stability window of all the solutions. Measurements were carried out by using home-made, “T-type” polyethylene cells with “SuperP”

31

carbon casted on copper or aluminum as working electrode and two lithium foils as counter and reference electrodes. Cell was scanned in the 3.0-0 V (cathodic scan) and 3.0-6.0 V (anodic scan) voltage ranges at a scan rate of 0.2 mVs-1. In the Fig. 3.9 are reported the ESWs of the commercial Py14TFSI and the solution Py14TFSI-LiTFSI 0.2m (similar behaviour, here not reported, was obtained in the case of Py24TFI and the respective IL-salt solution). The figure clearly shows the wide electrochemical stability of both the samples, that is a good properties for lithium batteries electrolytes.

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5 6,0-1,00

-0,75

-0,50

-0,25

0,00

0,25

0,50

0,75

1,00Li / Electrolyte / C

Potential, V vs. Li/Li +

Cur

rent

, m

A c

m-2

Py14

TFSI

Py14

TFSI LiTFSI 0.2 m 0.2mVs-1

Fig. 3.9 – Electrochemical Stability Window of the commercial

Py14TFSI and the solution Py14TFSI-LiTFSI 0.2m

3.3.3. Conclusions The ILs were studied pure and in solution with a lithium salt showing good performances from an electrochemical and a thermal point of view: that make them suitable for application in lithium batteries. After these preliminary characterization both the ionic liquids were studied in mixtures with alkylcarbonates: the commercial Py14TFSI was studied as liquid in alkylcarbonates mixtures (see cap 4), the synthesized Py24TFSI was used and characterized in gel polymer membranes (see cap 6).

32

References [1] H. Ohno, Electrochemical aspects of ionic liquid, wiley & Sons Inc., 2005 [2] H. Tokuda, K. Hayamizu, K. Ishii, Md.A.B.H. Susan, M. Watanabe, J. Phys. Chem. B. 108 (2004) 16593 [3] A. Fernicola, F. Croce, B. Scrosati, T. Watanabe, H. Ohno, J. Power Sources 41 (2007) 348 [4] J. Hassoun, A.Fernicola, M.A. Navarra, S. Panero, B. Scrosati, J. Power Sources, 195 (2010) 574 [5] G.B. Appetecchi, S. Scaccia, C. Tizzani, F. Alessandrini, S. Passerini, J. Electrochem. Soc., 153 (2006) A1685 [6] D. Aurbach, I. Weissman, A. Zaban, P. Dan, Electrochim Acta 45 (1999) 1135 [7] A. Fernicola, F. Croce, B. Scrosati, T. Watanabe, H. Ohno, J. Power Sources 41 (2007) 348

33

4. IL-alkylcarbonates mixtures 4.1. IL-alkylcarbonates mixtures

The substitution of the conventional flammable and volatile organic solutions with ILs is expected to greatly reduce the risk of thermal runaways and, eventually, fire accidents, thus drastically improving the overall safety of the battery. Four new mixed liquid electrolytes made by large addition of ionic liquid to commercial battery grade electrolytes were investigated, as well as the performances of lithium and Li-ion cells made with these new electrolytes. The commercial battery grade electrolytes here used (Merck products) are constituted by a lithium salt, (lithium-hexafluorophosphate), dissolved in carbonate-based (ethylene carbonate, dimethyl carbonate and diethyl carbonate) solutions: 1M LiPF6 in EC:DMC 1:1 wt/wt and 1M LiPF6 in EC:DMC:DEC 1:1:1 wt/wt named LP30 and LP71 respectively. The electrolyte solutions were prepared by mixing different amounts (30% wt. and 50% wt.) of the commercial Py14TFSI IL and the commercial electrolytes LP30 and LP71 that were used as provided. Details on the composition of the studied samples, i.e. the four mixture electrolytes as well as the pure LP30 and LP71, are summarized in the Tab. 4-1.

Sample Sample name % of ionic liquid% of carbonate

solution

Py14TFSI IL 100 0

LP30 LP30 0 100

LP30/Py14TFSI 70/30 wt/wt LP30/IL-30 30 70

LP30/Py14TFSI 50/50 wt/wt LP30/IL-50 50 50

LP71 LP71 0 100

LP71/Py14TFSI 70/30 wt/wt LP71/IL-30 30 70

LP71/Py14TFSI 50/50 wt/wt LP71/IL-50 50 50

Tab. 4-1 - The composition of the electrolyte samples studied in this work

To avoid any contamination with external ambient, all the materials handlings were carried out in a controlled argon atmosphere dry box having an humidity content below 1 ppm.

34

4.1.1. Thermal characterization: flammability test The thermal stability of the solutions was investigated by flammability tests. Few drops of the samples were placed on a watch glass and exposed to a burner for 3 seconds to allow ignition. The time required to extinguish the flame was recorded and normalized against liquid mass to evaluate the self-extinguish time (SET) in s g-1 [1]. The addition of the IL to the standard electrolytes results in large improvements of the response to flammability tests (see Fig. 4.1; results for the LP71-cases are very similar and are therefore omitted to avoid redundancy). When exposed to a free flame the LP30 electrolyte still burns after 20 seconds whereas the flame on samples containing 30% in weight of the ionic liquid extinguished within 20 seconds (Fig. 4.1 (h)) and within 15 seconds in the case of samples containing 50% in weight of ionic liquid. Details on flammability test and the estimated values of the self extinguish time for the LP30-based solutions are summarized Tab. 4-2. Briefly, by increasing the amount of ionic liquid in the sample, the self-extinguishing time (SET, i.e. the flame extinguish time normalized to the weight of the samples) strongly decreases, thus highly improving safety. This effect is expected as pure Py14TFSI is non-flammable.

At Start Time After 10s After 15s After 20s

LP30

LP30/IL-30

LP30/IL-50

Fig. 4.1 - Pictures of the flammability performances of LP30 at starting time (a), after 10 seconds (b), after 15 seconds (c) and after 20 seconds (d), LP30/Py14TFSI 70/30 wt/wt at starting time (e), after 10 seconds (f), after 15 seconds (g) and after 20 seconds (h) and LP30/Py14TFSI

50/50 wt/wt at starting time (i), after 10 seconds (j), after 15 seconds (k) and after 20 seconds (l)

35

LP30 LP30/IL-30 LP30/IL-50

Sample initial weight 0.508 g 0.4826 g 0.4844 g

Flame exposure time 3 s 3 s 3 s

Flame extinguish time 22.86 s 17.11 s 12.94 s

Self-extinguishing time (SET) 45.00 s g-1 35.45 s g-1 26.71 s g-1

Tab. 4-2 - Flammability test: sample initial weight, flame exposure time, flame extinguish time and self extinguishing time (SET) of LP30, LP30/Py14TFSI 70/30 wt/wt, LP30/Py14TFSI 50/50

wt/wt

4.1.2. Electrochemical characterization: conductivi ty test, cathodic cyclic voltammetry and anodic linear scan The ionic conductivity measurements were carried out by AC impedance spectroscopy, dipping in the solutions a test cell with platinized platinum blocking electrodes and known cell constant. All the measurements were performed in the temperature range from 10°C to 60°C. Before each measurement cells were kept at constant temperature for 1 day ca. to reach thermal equilibration. The Arrhenius plots of the LP30-based and the LP71-based solutions are shown in the Fig. 4.2 (a) and (b) compared to the pure Py14TFSI ionic liquid. The conductivity of the ionic liquid ranges between 10-3 and 10-2 Scm-1 in the entire temperature range i.e. from 15°C to 60°C: this value is suitable for lithium battery applications [2]. The addition of the ionic liquid to the commercial electrolyte solutions does not reduce the overall ionic conductivity neither for LP30 nor for LP71. In fact, although the large added amounts of ionic liquid (30 to 50%), the conductivity values of the mixtures are comparable with the pristine IL-free carbonate-based electrolyte, i.e. LP30 or LP71, and are approximately 5 times larger in comparison to the IL. Slightly higher conductivity values are obtained in the case of LP30-based solutions, with respect to LP71-based electrolytes.

3,0 3,2 3,4 3,610-4

10-3

10-2

10-1

Con

duct

ivity

, S c

m-1

1000/T, K-1

IL LP71 LP71/IL-30 LP71/IL-50

52040

T, °C

60

3,0 3,2 3,4 3,610-4

10-3

10-2

10-1

Con

duct

ivity

, S c

m-1

1000/T, K-1

52060 40 T, °C

IL LP30 LP30/IL-30 LP30/IL-50

Fig. 4.2 - Arrhenius plot of Lp30-based solutions and Py14TFSI (a) and Lp71-based solutions

and Py14TFSI (b)

36

Cyclic cathodic voltammetry measurements and linear anodic scans were performed to evaluate the electrochemical stability window of all the solutions. Measurements were carried out by using T-type polyethylene cells with “SuperP” carbon casted on copper or aluminum as working electrode and two lithium foils as counter and reference electrodes. Cell was scanned in the 3.0-0 V (cathodic scan) and 3.0-6.0 V (anodic scan) voltage ranges at a scan rate of 0.2 mVs-1 [3,4,5]. The electrochemical stability windows of all the investigated solutions are reported in the Fig. 4.3 (a) and (b). These plots were built by adding the experimental results obtained in cathodic cyclic voltammetry and anodic linear sweep voltammetry tests [6]. Only the first anodic and cathodic scans are here reported. The steady current-voltage response of the SuperP carbon electrodes shows that the addition of the ionic liquid slightly improves the stability window of all the mixed solutions, in particular in the anodic region. To be noticed that pure IL shows the highest stability at high voltage. The pre-decomposition peaks that are observed in pure alkylcarbonates solutions at about 4.6 V for LP71 and 4.8 V for LP30 (see Fig. 4.3 (a)), are absent in all the mixed electrolytes (see Fig. 4.3 (b)) and the current drift is slightly shifted towards higher voltage.

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5 6,0-0,5

-0,4

-0,3

-0,2

-0,1

0,0

0,1

0,2

0,3

0,4

0,5

Potential vs. Li +/Li, V

Cur

rent

, mA

cm-2

LP30/IL-30 LP30/IL-50 LP71/IL-30 LP71/IL-50

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5 6,0-0,5

-0,4

-0,3

-0,2

-0,1

0,0

0,1

0,2

0,3

0,4

0,5

Potential vs. Li +/Li, V

Cur

rent

, mA

cm-2

IL LP30 LP71

Fig. 4.3 - Electrochemical stability window of Py14TFSI, LP30 and LP71 (a) and mixtures of

Py14TFSI, LP30 and LP71

It is interesting to observe that the typical poor cathodic stability of conventional imidazolium-based ILs [7] are partially overcome by the choice of a suitable pyrrolidinium cation having no highly acidic protons. Indeed, the small current drift observed around 1.5 V vs. Li+/Li for pure Py14TFSI decreases in subsequent cathodic cycles (not here reported): it is likely due to a multi-step decomposition process, resulting in the formation of a protective passivation film [1,3-5,8-15]. Therefore, due to the occurrence of a solid-electrolyte interface film (SEI), we conclude that the cathodic stability of Py14TFSI can be kinetically extended well below 1.5 V. This is expected to occur also on the IL-carbonate mixture electrolytes (see Fig. 4.3 (b)) due to the formation of the stable film typically observed on EC-based electrolytes. In fact the stable passivation of the electrodes upon reduction is nicely confirmed by the

37

CV cycles besides the first discharges (here omitted to avoid redundancies and save space). In fact the charge irreversibly consumed in the first discharges below 1.5 V (but mainly at 0.7-0.8 V) is not observed in the subsequent cycles for all the studied electrolytes. It is to be noted that Appetecchi et al. [16] observed that the pure Py14TFSI ionic liquid is more stable cathodically than our results. In particular the small drift current observed at a potential of approximately 1.5 V vs. Li+/Li is not observed by Appetecchi and co-workers [16]. However the experimental conditions are different and therefore the corresponding experimental results are not easily comparable [7,16]. We determined the stability windows by using carbonaceous working electrodes (superP carbon versus lithium metal) and a third Li reference. These working electrodes are made with high surface area non graphitic carbon nanoparticles in order to enhance any parasitic decomposition due to electrolyte degradation. The parasitic decomposition reactions were also enhanced by using a slow scan rates (0.2 mVs-1). In literature similar tests are usually done using platinum or glassy carbon flat working electrodes versus carbon paper (and Ag inert reference electrodes) with fast scan rates (e.g. 5-10 mVs-1) thus minimizing any electroactive process [7,16]. In this view our tests are much severe with respect to the literature.

4.1.3. Cell performances: galvanostatic cyclation The cells for galvanostatic tests were assembled by coupling a lithium metal foil anode with a LiFePO4 cathode (LFP), using the selected solution soaked in a WhatmanTM separator as electrolyte. The same electrolytes were also tested in a different cell configuration by coupling a lithium foil and a Li 4Ti5O12 electrode (LTO). The LFP or LTO lithium cells were tested by galvanostatic cycling in a 2.2 – 4.5 V and in a 1.0 – 2.5 V voltage ranges, respectively. All cells were cycled at C/5 rate (1C current being 0.223 A g-1 cm-2 with respect to Li4Ti5O12 active mass and 0.217 A g-1 cm-2 with respect to LiFePO4 active mass). The performance of each cell was evaluated in terms of specific capacity, charge/discharge efficiency and cycle life. A lithium-ion cell was assembled by coupling a Li4Ti5O12 anode with a LiFePO4 cathode with a selected electrolyte solution, i.e. LP30/Py14TFSI 70/30 wt/wt., soaked in a WhatmanTM separator. The complete Li-ion cell was tested by galvanostatic cycling in a 0.2-2.3 V voltage range at C/5 rate (1C current being 0.223 A g-1 cm-2 with respect to Li4Ti5O12 active mass). The cell was anode-limited and the positive to negative electrodes weights ratio was P/N=1.5. The four mixed electrolytes have been galvanostatically tested in lithium cells vs. LiFePO4 and compared with benchmark cells assembled with pure LP30 and LP71 electrolytes. The charge/discharge voltage profiles (1st and 5th cycles) are reported in Fig. 4.4 (a) and (b) for the LP30-based and LP71-based electrolytes, respectively. The specific capacity plots vs. cycle number are reported for the first forty cycles in Fig. 4.5 (a) and (b) for the LP30-based and LP71-based electrolytes, respectively. The addition of the ionic liquid to both the standard electrolytes, i.e. LP30 and LP71, apparently does not affect the performances of the electrode active material in terms of specific capacity, reversibility and cyclability. In particular the electrolytes LP30/Py14TFSI 70/30 wt/wt and LP71/Py14TFSI 70/30 wt/wt show negligible differences in performances with respect to the pure alkylcarbonate electrolytes. Only in the case of LP71/Py14TFSI 50/50 wt/wt a clear decrease of the cycled

38

specific capacity is observed. This experimental evidence, that has been confirmed in triplicate, suggests that at such high concentration of IL the nature of the alkylcarbonate mixture plays a key role in the stabilization/destabilization of the electrode-electrolyte interphase, even at potentials well within the stability window of the electrolyte mixture.

0 20 40 60 80 100 120 140 160

2,75

3,00

3,25

3,50

3,75

4,00

4,25

4,50

Specific Capacity, mAh/g

Vol

tage

, V

LP71 - 1st cycle LP71/IL-30 - 1st cycle LP71/IL-50 - 1st cycle LP71 - 5th cycle LP71/IL-30 - 5th cycle LP71/IL-50 - 5th cycle

0 20 40 60 80 100 120 140 160

2,75

3,00

3,25

3,50

3,75

4,00

4,25

4,50

Specific Capacity, mAh/gV

olta

ge, V

LP30 - 1st cycle LP30/IL-30 - 1st cycle LP30/IL-50 - 1st cycle LP30 - 5th cycle LP30/IL-30 - 5th cycle LP30/IL-50 - 5th cycle

Fig. 4.4 - Cell voltage vs. specific capacity for the Li / LP30-based electrolyte solution / LiFePO4

(a) and Li / LP71-based electrolyte solution / LiFePO4 (b) cells

0 5 10 15 20 25 30 35 400

20

40

60

80

100

120

140

160

180

200

Spe

cific

cap

acity

, mA

h/g

Cycle number

LP71 charge dischargeLP71/IL-30 charge dischargeLP71/IL-50 charge discharge

0 5 10 15 20 25 30 35 400

20

40

60

80

100

120

140

160

180

200

Spe

cific

cap

acity

, mA

h/g

Cycle number

LP30 charge dischargeLP30/IL-30 charge dischargeLP30/IL-50 charge discharge

Fig. 4.5 - Specific capacity vs. cycle number for the Li / LP30-based electrolyte solution /

LiFePO4 (a) and Li / LP71-based electrolyte solution / LiFePO4 (b) cells

A similar study has been also performed for all the four mixed electrolytes in lithium cells vs. Li4Ti5O12. Galvanostatic cycling have been performed, in comparison with benchmark cells assembled with pure LP30 and LP71 electrolytes. The charge/discharge voltage profiles (1st and 5th cycles) are reported in Fig. 4.6 (a) and (b) for the LP30-based and LP71-based electrolytes, respectively. The specific capacity vs. cycle number is reported, for the first forty cycles, in Fig. 4.7 (a) and (b) for the LP30-based and LP71-based electrolytes, respectively. Also the galvanostatic tests in lithium cells vs. lithium titanate confirm that the addition of the ionic liquid to both the standard solutions, i.e. LP30 and LP71 does not affect the performances in terms of specific capacity, reversibility and cyclability. Similarly to the LiFePO4

39

case, a remarkable worsening of the performances is observed in the case of LP71/Py14TFSI 50/50 wt/wt electrolyte. This effect has been confirmed by testing three different batches of the same electrolyte. This evidence supports the above mentioned hypothesis: the nature of the alkylcarbonate mixture, and in particular the presence of DEC, strongly affect the interphase stability at high concentration of IL.

0 20 40 60 80 100 120 140 1600,5

1,0

1,5

2,0

2,5

3,0

Vol

tage

, V

Specific Capacity, mAh/g

LP71 - 1st cycle LP71/IL-30 - 1st cycle LP71/IL-50 - 1st cycle LP71 - 5th cycle LP71/IL-30 - 5th cycle LP71/IL-50 - 5th cycle

0 20 40 60 80 100 120 140 1600,5

1,0

1,5

2,0

2,5

3,0

Specific Capacity, mAh/gV

olta

ge, V

LP30 - 1st cycle LP30/IL-30 - 1st cycle LP30/IL-50 - 1st cycle LP71 - 5th cycle LP71/IL-30 - 5th cycle LP71/IL-50 - 5th cycle

Fig. 4.6 - Cell voltage vs. specific capacity for the Li / LP30-based electrolyte solution / Li4Ti 5O12

(a) and Li / LP71-based electrolyte solution / Li4Ti 5O12 (b) cells

0 5 10 15 20 25 30 35 400

20

40

60

80

100

120

140

160

180

200

Spe

cific

cap

acity

, mA

h/g

Cycle number

LP71 Charge DischargeLP71/IL-30 Charge DischargeLP71/IL-50 Charge Discharge

0 5 10 15 20 25 30 35 400

20

40

60

80

100

120

140

160

180

200

Spe

cific

cap

acity

, mA

h/g

Cycle number

LP30 Charge DischargeLP30/IL-30 Charge DischargeLP30/IL-50 Charge Discharge

Fig. 4.7 - Specific capacity vs. cycle number for the Li / LP30-based electrolyte solution /

Li 4Ti 5O12 (a) and Li / LP71-based electrolyte solution / Li4Ti 5O12 (b) cells

For a deeper understanding of this aspect, further investigations such as in-situ XRD analysis and ex-situ Raman IR spectroscopic characterization are currently in progress in order to elucidate structural features of the solid electrolyte interphases under working conditions, as well as to define the molecular interactions among the various components of the mixture electrolytes. The results of this study will be reported in a future publication. With reference to Fig. 4.4 (a) and (b) and Fig. 4.6 (a) and (b), it’s remarkable that the voltage hysteresis between charge and discharge is only marginally affected for both LFP and LTO lithium cells when passing from 30 to 50 wt.% of the ionic liquid. To be noticed that specific capacities achieved for the investigated cells are always slightly lower than the theoretical values (assumed to be 170 mAhg-1 for LiFePO4 and 175 mAhg-1 for Li4Ti5O12), including the case of cells adopting pure LP30 and

40

LP71 reference electrolytes. Being the scope of the present paper a comparative investigation of different electrolyte compositions, the issue of absolute cell performance is beyond the scope of this work. Applicability of IL-added solutions was also confirmed by cycling a complete lithium-ion battery. The studied Li-ion cell has been assembled by coupling the LiFePO4 cathode and the Li4Ti5O12 anode and one of the mixed electrolyte. Among those investigated, mixtures containing 30 wt% of IL appeared the best choice, showing capacity values in lithium metal cells comparable, or even better, than the pure, IL-free electrolyte. Thus, LP30/Py14TFSI 70/30 wt/wt was selected, based on its enhanced conductivity with respect to LP71/Py14TFSI 70/30 wt/wt. The origin of this superior behavior is rooted in the fundamental chemistry that drives the changes of the conductivities, variation of the stability windows and ability to sustain galvanostatic cycling, and it is beyond the scope of this paper. The voltage profiles vs. the specific capacity of first and tenth cycles of the assembled lithium-ion cell is shown in Fig. 4.8.

0 2 4 6 8 100

50

100

150

200

250

Cycle number

Spe

cific

cap

acity

, mA

h/g

Charge Discharge Efficiency

0

20

40

60

80

100

Efficiency, %

0 20 40 60 80 100 120 140 1600,0

0,5

1,0

1,5

2,0

2,5

3,0

Specific Capacity, mAh/g

Vol

tage

, V

1st cycle 10th cycle

Fig. 4.8 - Cell voltage vs. specific capacity (a) and specific capacity vs. cycle number (b) for the

LiFePO4 / LP30/Py14TFSI 30/70 wt/wt / Li4Ti 5O12 lithium-ion cell.

This final test has been carried out in order to demonstrate the ability of these electrolytes to be successfully used in complete Li-ion cells thus drastically increasing the safety of the overall device.

4.2. Conclusions

The mixed electrolytes show conductivity values comparable to the pure carbonate-based electrolyte solutions, improved electrochemical stability windows and a large reduction of the SET when exposed to a free flame. With the exception of the LP71/Py14TFSI 50/50 wt/wt mixture, the mixed IL-carbonate electrolytes show good performances in lithium cells vs. LTO and LFP. A LFP-LTO Li-ion cell has been assembled and tested thus demonstrating the use of one of these mixed electrolytes in a complete device with a drastically improved safety profile.

41

References [1] C. Arbizzani, G. Gabrielli, M. Mastragostino, J. Power Sources 196 (2011) 4801. [2] J. Barthel, H.J. Gores, in: J.O. Besenhard (Ed.), Handbook for Battery Materials, Wiley-VCH, Weinheim, 1999, p. 457. [3] S. Megahed, B. Scrosati, Interface 4 (1995) 34. [4] J.L. Nowinski, P. Lightfoot, P.G. Bruce, J. Mater. Chem. 4 (1994) 1579. [5] P.C. Howlett, N. Brack, A.F. Hollenkamp, M. Forsyth, D.R. MacFarlane, J. Electrochem.Soc. 153 (2006) A595. [6] C. Sirisopanaporn, A. Fernicola, B. Scrosati, J. Power Sources 19 (2009) 490. [7] M. Galinski, A. Lewandowski, I. Stepniak, Electrochim. Acta 51 (2006) 5567 [8] J. Hassoun, A. Fernicola, M. A. Navarra, S. Panero, B. Scrosati, J. Power Sources 195 (2010) 574. [9] P. Reale, A. Fernicola, B. Scrosati, J. Power Sources 194 (2009) 182. [10] M. A. Navarra, J. Manzi, L. Lombardo, S. Panero, and B. Scrosati ChemSusChem 4 (2011) 125. [11] A. Guerfi, M. Dontigny, P. Charest, M. Petitclerc, M. Lagacé, A. Vijh, K. Zaghib, J. Power Sources 195 (2010) 845. [12] P.M. Bayley, G.H. Lane, N.M. Rocher, B.R. Clare, A.S. Best, D.M. MacFarlane, M. Forsyth, Phys Chem Chem Phys 11 (2009) 7202. [13] T. Sato, T. Maruo, S. Marukane, K. Takagi, J. Power Sources 138 (2004) 253. [14] L. Larush, V. Borgel, E. Markevich, O. Haik, E. Zinigrad, D. Aurbach, G. Semrau, M. Shmidt, J. Power Sources 189 (2009) 217. [15] D. Aurbach, I. Weissman, A. Zaban, P. Dan, Electrochim. Acta 45 (1999) 1135. [16] G.B. Appetecchi, S. Scacia, C. Tizzani, F. Alessandrini, S. Passerini, Electrochem. Soc. 153 (2006) A1685.

42

5. Fluorine-free salt-based electrolytes Non conventional electrolyte solution constituted by non-fluorinated lithium salt were developed in the last years in order to enhance safety and stability of the lithium batteries. Among new electrolyte salts, a very promising candidate is lithium bis(oxalato)borate (LiBOB). The use of LiBOB is aimed to decrease fluorinated component and to improve film forming capability. Also LiBOB salts shows enhanced thermal stability with respect to conventional LiPF6 in virtue of its decomposition temperature approaching 300°C.

5.1. Fluorine-free salt-based electrolytes: charact erization

In this work the effect of the addition of the LiBOB salt to conventional electrolytes as EC:DMC and EC:PC:DMC was investigated: in the first case the LiBOB salt gradually substitute LiPF6 salt, in the second case LiBOB salt is the only salt in the solution. In the Tab. 5-1 are reported compositions of all the investigated solutions based on EC:DMC.

Solvent LiPF6, M LiBOB-1, M

Sol 0 (LP30) EC:DMC 1:1 w/w 1 --

Sol 1 EC:DMC 1:0.87 w/w 1 0.332

Sol 2 EC:DMC 1:0.87 w/w 1 0.135

Sol 3 EC:DMC 1:0.87 w/w 0.53 0.134

Sol 4 EC:DMC 1:0.87 w/w 0.6 0.4

Sol 5 EC:DMC 1:0.87 w/w -- 0.33

Tab. 5-1 - All the EC:DMC-based solutions investigated The solution named “Sol 0” is the commercial battery electrolyte LP30 from Merck and was used as reference. The others solutions contain both LiBOB an LiPF6 salts, except the solution “Sol 5” in which LiBOB is the only salt. The study of the addition of LiBOB salt to carbonate-based solutions were performed using EC:PC:DMC 1:1:3 wt/wt ternary solution, obtaining a solution 0.7M (Sol 6). The selected composition is the same reported in the Hassoun et al. work [1].

5.1.1. Electrochemical characterization: conductivi ty test, cathodic cyclic voltammetry and anodic linear scan Conductivity measurements on all the electrolyte solutions were performed by dipping in them a standard test cell with platinized platinum blocking electrodes and known cell constant. Tab. 5-2 reports the conductivity values at room temperature of these solutions:

43

Solvent LiPF 6, M LiBOB-1, M σσσσ (rT), S cm -1

Sol 0 (LP30) EC:DMC 1:1 w/w 1 -- 1.1 · 10-2

Sol 1 EC:DMC 1:0.87 w/w 1 0.332 1.1 · 10-2

Sol 2 EC:DMC 1:0.87 w/w 1 0.135 1.0 · 10-2

Sol 3 EC:DMC 1:0.87 w/w 0.53 0.134 8.8 · 10-3

Sol 4 EC:DMC 1:0.87 w/w 0.6 0.4 8.6 · 10-3

Sol 5 EC:DMC 1:0.87 w/w -- 0.33 5.3 · 10-3

Tab. 5-2 - Composition and conductivity values of the electrolyte solution investigated

It’s possible to note that the conductivity values are good for all the solution: that confirms that LiBOB salt can be used for batteries electrochemical solutions. Because of the good conductivity value and the high amount of the LiBOB salt dissolved, the most interesting solution seemed to be the Solution 4. In the Tab. 5-3 is reported the composition and the conductivity value at room temperature of the ternary solution investigated.

Solvent

Sol 6 EC:PC:DMC 1:1:3 w/w

LiBOB, M σσσσ (rT), Scm-1

0.7 5.5 · 10-3Solvent

Sol 6 EC:PC:DMC 1:1:3 w/w

LiBOB, M σσσσ (rT), Scm-1

0.7 5.5 · 10-3

Tab. 5-3 - Composition and conductivity values of the ternary solutions investigated The conductivity values are good and, in terms of order of magnitude, comparable each other and also to the values obtained with binary solutions, that are a little bit higher because of the presence of the LiPF6 salt. The electrochemical stability window of the solution 4 was investigated by cyclic cathodic voltammetry and linear anodic scan and reported in the Fig. 5.1 compared to the Sol 0 ESW as reference.

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5 6,0-0,6

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0

1,2

1,4

1,6

cyclic cathodic

voltammetry

C Super P / Electrolyte / Li

0.2 mV s-1

Cur

rent

, m

A c

m

-2

Potential, V vs Li/Li +

Sol 0 Sol 4

linear

anodic scan

Sol 4 ESW 1.7 ÷ 5.3 V

Fig. 5.1 - Electrochemical stability windows of

two of the electrolyte solution investigated

44

Is possible to note that the in red curve there is a peak at 1.7V, typical of the LiBOB decomposition, and at high potential the solution is stable until 5.2V: the ESW of the sol 4 is between 1.7V and 5.2V. To be noticed that the peak around 1,7 V disappears at subsequent cathodic cycles (not reported here). In Fig. 5.2 is reported the electrochemical stability window of the solution EC:PC:DMC 1:1:3 by wt, 0.7M LiBOB.

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5 6,0

-1,2-1,0-0,8-0,6-0,4-0,20,00,20,40,60,81,01,2 C Super P / Electrolyte / Li

0.2 mV s-1

ESW 1.7 ÷ 5.3 V

cyclic cathodic

voltammetry

linear

anodic scan

Cur

rent

, m

A c

m-2

Potential, V vs. Li/Li +

1st 2nd Anodic

Fig. 5.2 - Electrochemical stability window of the solution

EC:PC:DMC 1:1:3 by wt, 0.7M LiBOB

5.1.2. Thermal characterization: DSC, TGA The improved thermal stability of LiBOB-added solution was demonstrated by DSC measurements, performed under nitrogen atmosphere at a scan rate of 3°C/min. The Fig. 5.3 reports the DSC response of Sol 4 (LiBOB/LiPF6 0.4/0.6 M/M) compared to those of Sol 0 (1M LiPF6, commercial LP30) and pure EC:DMC solvent mixture. The effect of the salts and, mainly, of LiBOB with respect of LiPF6 is evident in suppressing thermal transitions related to the solvents.

45

-80 -60 -40 -20 0 20 40 60 80

0

4

8-80 -60 -40 -20 0 20 40 60 80

0

4

8-80 -60 -40 -20 0 20 40 60 80

0

4

8

Temperature / °C

EC:DMC, LiPF6/LiBOB 0.6/0.4 M/M

Hea

t flo

w /

W g

-1

EC:DMC

EC:DMC, LiPF6 1M

Fig. 5.3 - DSC responses of Sol 0, Sol 4 and pure EC:DMC solvent mixture.

5.1.3. Cell performances: galvanostatic cyclation Applicability of Sol 4 electrolyte was tested in a lithium-metal cell, adopting SnC as working electrode. Figure 4 A and B reports specific capacity versus cycle number for cells containing Sol 4 and Sol 0 electrolyte, respectively.

0 10 20 30 40 50 600

150

300

450

600

750

900

1050

Cap

acity

/ m

Ahg

-1

Cycle number

Charge Discharge

0 10 20 30 40 50 600

150

300

450

600

750

900

1050 Charge Discharge

Cap

acity

/ m

Ah

g-1

Cycle number

A BA

0 10 20 30 40 50 600

150

300

450

600

750

900

1050

Cap

acity

/ m

Ahg

-1

Cycle number

Charge Discharge

0 10 20 30 40 50 600

150

300

450

600

750

900

1050 Charge Discharge

Cap

acity

/ m

Ah

g-1

Cycle number

A BA

Fig. 5.4 - Specific capacity versus cycle number of A) Li / Sol 4/ SnC and B) Li / Sol 0 / SnC

cells. Current density: 100 mA cm2 g-1

Even though acceptable, the performances of the cell using sol 4 i.e. EC:DMC, LiPF6/LiBOB 0.6/0.4 M/M electrolyte are lower than those of conventional LP30 electrolyte, Sol 0-based cell. Performances of the conventional and the ternary solution named sol 6 (LP30 and EC:PC:DMC 0.7M LiBOB) were finally investigated in metal lithium cells by galvanostatic cyclation tests using SnC (Fig. 5.5 and Fig. 5.6) or LiNi0.5Mn1.5O4 (Fig. 5.7 and Fig. 5.8) as working electrodes.

46

0 10 20 30 40 50 60 70 80 90 1000

100

200

300

400

500

600

700

800

0 100 200 300 400 500 600 700 800

0,0

0,5

1,0

1,5

2,0

2,5

Cycle number

Spe

cific

cap

acity

, mA

h/g

Charge Discharge

i = 400 mA/g

Vol

tage

, V

Specific Capacity, mAh/g

Cycle 1 Cycle 5 Cycle 10 Cycle 20 Cycle 50

Fig. 5.5 - Cell performances of the EC:PC:DMC 1:1:3 by wt, 0.7M LiBOB solution in terms of voltage vs. specific capacity and specific capacity vs. cycle number in a two electrode cell with

lithium and SnC (as prepared).

0 10 20 30 40 50 60 70 80 90 1000

100

200

300

400

500

600

700

800

0 100 200 300 400 500 600 700 800

0,0

0,5

1,0

1,5

2,0

2,5

Cycle number

Spe

cific

cap

acity

, mA

h/g

Charge Discharge

i = 400 mA/g

Vol

tage

, V

Specific Capacity, mAh/g

Cycle 1 Cycle 5 Cycle 10 Cycle 20 Cycle 50

Li / LP30 / SnC

Fig. 5.6 - Cell performances of the LP30 solution in terms of voltage vs. specific capacity and specific capacity vs. cycle number in a two electrode cell with lithium and SnC (as prepared).

Is possible to note that the irreversibility showed in the first cycle in presence of a carbon electrode is increased by the LiBOB salt. Also, the stability during the cyclation looks to be better for LP30 electrolyte. From previous studies is well known that a pre-activation treatment on the SnC electrode can reduce significantly the irreversibility obtained in the first cycle [1].

47

0 25 50 75 100 125 150 175 200 225 250 275 3003,0

3,5

4,0

4,5

5,0

5,5

6,0

0 10 20 30 40 50 60 70 80 90 1000

50

100

150

200

250

300

350

400

Vol

tage

, V

Specific Capacity, mAh/g

Cycle 1 Cycle 5 Cycle 10 Cycle 20 Cycle 50

i = 50 mA/g

Cycle number

Spe

cific

cap

acity

, mA

h/g

Charge Discharge

Fig. 5.7 -Cell performances of the EC:PC:DMC 1:1:3 by wt, 0.7M LiBOB solution in terms of voltage vs. specific capacity and specific capacity vs. cycle number in a two electrode cell with

lithium and LiNi 0.5Mn 1.5O4 as electrodes

0 25 50 75 100 125 150 175 200 225 250 275 3003,0

3,5

4,0

4,5

5,0

5,5

6,0

0 10 20 30 40 500

50

100

150

200

250

300

350

400i = 147 mA/g

Vol

tage

, V

Specific Capacity, mAh/g

Cycle 1 Cycle 5 Cycle 10 Cycle 20 Cycle 50

Cycle number

Spe

cific

cap

acity

, mA

h/g

Charge Discharge

Fig. 5.8 - Cell performances of the LP30 solution in terms of voltage vs. specific capacity and specific capacity vs. cycle number in a two electrode cell with lithium and LiNi 0.5Mn 1.5O4 as

electrodes In the case of LNMO electrodes, the presence of LiBOB salt reduces the irreversibility at first cycle. Also, capacity retention looks to be good.

5.2. Conclusions

From the results obtained in this part of the work is clear that non conventional electrolyte solution constituted by non-fluorinated lithium salt like LiBOB con be considered as a promising alternative to the conventional electrolyte solutions. Based on the electrochemical tests discussed above, EC:PC:DMC 1:1:3 by wt, 0.7M LiBOB solution appears the best choice as electrolyte coupled with SnC and LNMO electrodes. References [1] J. Hassoun, M. Wachtler, M. Wohlfahrt-Mehrens, B. Scrosati, Journal of Power Sources, 196 (2011) 349-354

48

6. Gel Polymer Electrolytes (GPEs) Very important improvements, in terms of safety and reliability, can be achieved by moving from liquid solutions to polymer electrolytes (PEs).[1–5] Despite many favourable properties related to the nature of PEs, including negligible electrolyte leakage, practical PE-based Li-batteries based are confined to a small number of niche applications. Problems delaying their development include the low conductivity of most solid PEs at ambient temperature and reactivity with the lithium metal electrode in solvent plasticized polymer systems. In this regard, polymer electrolytes, resulting from a polymer matrix together with an ionic liquid solution, represent an attractive strategy since they combine the mechanical and chemical stability of the polymer component with the intrinsic good conductivity, non-flammable nature, and high thermal stability of the ionic liquid component. Few examples of this class of materials for application in lithium batteries have been reported. Shin et al. reported on a lithium-metal battery formed by a LiFePO4 cathode and a polymer electrolyte consisting of poly(ethylene oxide), lithium N,N-bis(trifluoromethanesulfonyl)imide (LiTFSI), and a room-temperature ionic liquid, with a high and stable performance at 40°C.[6] Improved cycling performances were reported by Chew et al. by replacing conventional liquid electrolytes with an ionic liquid polymer composite in a lithium-metal cell using a polypyrrole-coated lithium vanadium oxide cathode at room temperature. [7] J.-K. Kim et al. developed two PVDF-based polymer membranes based on room temperature ionic liquid and organic carbonate electrolytes and successfully tested them in a lithium-battery with a carbon coated LiMn0.4Fe0.6PO4cathode.[8] In this work two different kind of membranes have been prepared and characterized: IL-based GPEs and carbonate-based GPEs that differ both in the composition and in the preparation procedure.

6.1. IL-based and carbonate-based GPEs: preparation procedures

The “IL-based gel polymer membranes” were prepared by immobilizing the solution of LiTFSI in Py24TFSI ionic liquid with added mixtures of organic solvents, such as ethylene, propylene and dimethyl carbonates (EC, PC, and DMC), into a poly(vinylidenefluoride-co-hexafluoropropylene) (PVDF-HFP) matrix. This first type of polymer gel electrolyte membranes were prepared by a solution casting procedure. First, PVDF-HFP (Kynar Flex 2801) was dissolved in acetonitrile. Thereafter, LiTFSI–Py24TFSI, or, alternative, the mixture of IL-salt solution and alkylcarbonates, was added to the PVDF-HFP/acetonitrile solution. The EC:PC:DMC weight ratio was selected as 1:1:2. The resulting solution was vigorously stirred overnight at room temperature and then casted in a Petri dish by heating at 70°C and rapidly cooling to room temperature. The heating–quenching process was repeated until freestanding membranes were obtained; the membranes were finally dried under

49

reduced pressure at 60°C. All the procedures and material handlings were carried out in an argon-filled dry box. The “carbonate-based gel polymer membranes” were prepared by an optimized casting – swelling procedure. The PVDF (Solvay Solef® UHMW 6020) powder was dissolved in a EC:PC (1:1 w:w) or EC:DMC (1:1 w:w) electrolyte solution (the ratio is 20:80 in weight) by mixing them for 2 hours at room temperature. The obtained solution were casted in a Petri dish by heating at 80°C for 20 minutes and then quenched at room temperature until gelification. Final step was the swelling the obtained membranes in the selected Li-ion conducting solution (usually LP30). By the procedure here described homogeneous, mechanically stable and transparent membranes were obtained.

6.2. IL-based GPE: characterization

In this work were prepared and tested four types of membranes by varying the reciprocal composition of the PVDF-HFP, LiTFSI–Py24TFSI, and EC:PC:DMC components. Tab. 6-1 lists the four prepared samples, denoted as M1, M2, M3, and M4.

Membrane PVDF-HFP

content [wt%] LiTFSI–Py24TFSI

content [wt%] EC–PC–DMC content [wt%]

M1 30 70 -- M2 30 56 14 M3 22 50 28 M4 22 60 18

Tab. 6-1 - Compositions and acronyms of the ionic-liquid membranes investigated in this work.

6.2.1. Electrochemical characterization: conductivi ty test, cathodic cyclic voltammetry and anodic linear scan Impedance spectroscopy measurements were performed to all the samples due to determine their ionic conductivity. Fig. 6.1 shows the impedance response of four independent cells formed by sandwiching the given membrane sample between two blocking stainless-steel electrodes.

50

Fig. 6.1 - Impedance response of the four membranes developed in this study. M1 ; M2 ; M3 ; M4 ×

The Fig. 6.1 shows the impedance spectra, in the form of Nyquist plots. The intercepts with the real axis allow calculation of the conductivity values of the membranes. Notably, in the whole investigated frequency range, no signs of charge transfer or passivating layer formation are detectable. A linear response, as typically expected for blocking electrodes, indicates that our membranes do not undergo unexpected collateral reactions or undesired phenomena when placed in contact with the stainless steel electrodes. Fig. 6.2 shows the temperature dependence of the ionic conductivity of the membranes in the form of Arrhenius plots.

51

Fig. 6.2 - Conductivity Arrhenius plots of the four membranes M1: heating , cooling ; M2: heating , cooling ;

M3: heating , cooling ; M4: heating , cooling .

The M3 sample displays the highest conductivity in the whole range of temperatures investigated in this study, attaining values in the order of 10-2 Scm-1 at 100°C. The regular increase in conductivity upon heating for all the membranes confirms that neither physical transitions nor segregation phenomena occurred in the course of the test. Small hysteresis were detected at low temperatures, possibly indicating a thermal memory effect of the polymer matrix. Membrane M3 was selected as the most promising membrane and concentrated our attention on this membrane for further characterization. The electrochemical stability of the M3 membrane was evaluated by running a sweep voltammetry of a cell using a Super P carbon working electrode and a lithium metal counter electrode.

52

Fig. 6.3 - Electrochemical stability window of the M3 membrane in the cathodic voltage range. Super P working electrode. Room temperature.

The resulting current–voltage traces (Fig. 6.3) show in the first sweeping cycle a large current flow starting at around 1.0 V, associated with a multistep decomposition process, very likely due to the reduction of the carbonate solution component, with the consequent formation of a passivating film on the testing electrode. This interpretation is supported by a series of findings, such as: a) the irreversibility of the peaks in the 1.0–0.5 V region and b) the trend of the second sweep cycle where no trace of events down to approximately 0.25 V is detected. The current flow in the 0.25 V voltage range occurring in both first and second sweep cycles, can be ascribed to lithium deposition at the working Super P carbon electrode. The low peak current in the following oxidation scan is accounted for by considering the poor reversibility of the lithium deposit-stripping process on a SuperP carbon electrode.[9] In general, it’s possible to say that the results reported in Fig. 6.3 confirm that the passivating film, once formed on the electrode surface, is stable with no further growth. Fig. 6.4 shows the steady current-voltage response of the Super P carbon electrode extended over the entire anodic and cathodic range (0–5 V).

53

Fig. 6.4 - Electrochemical stability window of the M3 membrane in the overall voltage range. Super P working electrode. Room temperature

Even though weak current drifts occur at around 1.0 V and above 4.5 V, their low current values allow us to assume that the M3 membrane is electrochemically stable in the examined voltage range; hence, this membrane is expected to be efficiently used as safe electrolyte separator in lithium batteries operating in the range 3.0-4.0V.

6.2.2. Thermal characterization: TGA and flammabili ty test Membrane M3 was selected as the most promising membrane and concentrated our attention on this membrane for further characterization. Fig. 6.5 shows the thermogravimetric analysis (TGA) trace of membrane M3.

Fig. 6.5 - Thermogravimetric analysis of the M3 membrane

A weight loss, due to partial removal of the alkyl carbonate solution, starts approaching 100°C. No other effects are noticed up to about 350°C, at which temperature a large loss, due firstly to IL-solution decomposition and then to

54

polymer decomposition, is detected. This result is important in that it demonstrates the high thermal stability of the membrane that is expected to be reflected in safe operation when used as the electrolyte in a lithium battery. This expectation was qualitatively evaluated by performing flammability test, consisting of igniting a pre-weighed electrolyte sample followed by recording the time needed for the extinguishing of the flame. On applying a flame to the M3 membrane sample, no distinguishable ignition took place and, after tens of seconds, only the melting of the polymer under the flame heat was detected. Thus, the polymer electrolyte proposed herein can be defined as non-flammable. It’s possible to explain this positive result by assuming that, even though volatile organic components are present, their partial vapour pressure is strongly reduced when dispersed in the matrices thanks to interactions with the polymer and, especially, IL components, thus avoiding undesired ignition under sudden thermal runaways.

6.2.3. Cell performances: galvanostatic cyclation Accordingly to the results obtained, in particular to the electrochemical stability window of the M3 membrane, was assembled and tested a battery using a lithium metal anode and an iron phosphate cathode, separated by a M3 membrane electrolyte. Considering the electrode used, the battery was expected to operate at around 3.5 V [10], well within the stability domain of the M3 electrolyte. Fig. 6.6 illustrates typical charge (lithium removal from LiFePO4 to form FePO4) and discharge (lithium acceptance by FePO4 to reconvert into LiFePO4) cycles run at room temperature and at a C/10 rate (1C current=0.217 Acm-2g-1 with respect to the LiFePO4 active mass).

Fig. 6.6 - Charge-discharge cycle of the Li/M3/LiFePO4 polymer battery.

Reproducible voltage profiles with a delivered specific capacity approaching the theoretical value of 170 mAhg-1[10] were obtained. This result confirms the feasibility of the M3 membrane as a new electrolyte for advanced lithium polymer batteries.

55

6.3. Carbonate-based GPE: characterization

The fundamental advantage in the use of this second type of membrane studied in this is that is possible to prepare a “precursor” membrane and obtain a conductive one after swelling it in a selected conductive solution. Preliminary efforts have been devoted to the characterization of a membrane composed by using EC:DMC 1:1 w:w and PVDF (in a weight ratio of 80:20) and swelled in the conventional LP30 electrolyte. [11]. This membrane can be considered our reference membrane. Afterwards we tried to optimize the compositions of the membranes by varying both the ratios and the nature of the components.

The compositions of the precursor membrane were chosen very similar to the composition of the swelling solution. All the compositions are reported in

Tab. 6-2.

PRECURSOR MEMBRANE

COMPOSITION

SOLUTIONS FOR SWELLING

A (EC:DMC 1:1 wt/wt) : PVDF

80:20 wt/wt

EC:DMC 1:1 wt/wt, 1M LiPF6

(LP30)

B (EC:DMC 1:1 wt/wt) : PVDF

80:20 wt/wt

EC:DMC 1:1 wt/wt, 0.4M LiBOB, 0.6M LiPF6

C (EC:DMC 1:1 wt/wt) : PVDF

80:20 wt/wt

EC:DMC 1:1 wt/wt, 0.7M LiBOB

Tab. 6-2 - Composition of the membranes To be noticed that, from the amount of starting materials reported in table above, we can obtain round-shaped membranes of 8 cm diameter and ca. 700µm thickness.

6.3.1. Thermal characterization: DSC, TGA Thermal characterization were performed only to the reference membrane i.e. (EC:DMC 1:1 w:w):PVDF 80:20 swollen in LP30. The Fig. 6.7 clearly shows two endothermic transition: for the first one is not possible to define the temperature, but only a temperature range (5-25°C) in which the transition occurs.

56

-100 -50 0 50 100 150 200-1,5

-1,0

-0,5

0,0

0,5

1,0

Hea

t flo

w /

W/g

Temperature / °C

PVDF EC:DMC LiPF6

endo

-100 -50 0 50 100 150 200-1,5

-1,0

-0,5

0,0

0,5

1,0

Hea

t flo

w /

W/g

Temperature / °C

PVDF EC:DMC LiPF6

endo

Fig. 6.7 - DSC of the membrane (EC:DMC 1:1 w:w):PVDF 80:20 w:w in LP30. Heating rate: 5°C/min

This transition is related to the carbonate-based components and the temperature range can be related to the not uniform interactions and/or segregations that these components can have within the polymeric matrix. The other transition, that occurs at about 125°C is related to the fusion of the polymeric matrix. From this DSC profile is possible to conclude that the operational range of this membrane is 25-125°C. The two-step TGA response, reported in the Fig. 6.8, shows a first weight reduction at about 50°C related to the DMC removal. The second step at 15°C is attributable to both the EC removal and LiPF6 decomposition process.

0 50 100 150 200 250 30020

30

40

50

60

70

80

90

100

110

Temperature / °C

% w

eigh

t

TGA dTGA

Fig. 6.8 – TGA and dTGA of the membrane (EC:DMC 1:1 w:w):PVDF 80:20 w:w in LP30

Air atmosphere, heating rate: 5°C/min

At the end of the measure the residual weight percentage is 30% ca. and not 20% as expected. This difference is due to the volatile nature of the carbonate components that probably evaporate during the storage of the membrane. Also by trapping the carbonates in a polymeric matrix, is not possible to avoid a modification of composition of the membrane. For this reason is necessary to use low volatile components, like ILs, or salt with high decomposition temperature, like LiBOB.

57

6.3.2. Electrochemical characterization: cathodic c yclic voltammetry, anodic linear scan and conductivity test The electrochemical stability window of the reference membrane was evaluated by performing cathodic cyclic voltammetry and anodic linear scan. In … are reported only first cycles and is possible to note that the membrane is stable between a wide potential range (1.5-4.5 V vs. Li).

0 1 2 3 4 5-0,10

-0,05

0,00

0,05

0,10

Cur

rent

/ m

A*c

m-2

Potential / V vs Li

cathodic anodic

Fig. 6.9 - ESW of the membrane (EC:DMC 1:1 w:w):PVDF 80:20 w:w in LP30

The peak at 0.7V is related to the carbonates solvents (EC and DMC) and probably to a SEI film formation that allows the lithium intercalation at 0V.

Electrochemical characterization of selected swollen-type membranes has been performed in terms of room temperature conductivity. Results are reported in

Tab. 6-3.

MEMBRANE

COMPOSITIONS

SOLUTIONS

FOR SWELLING

CONDUCTIVITY OF

THE SOLUTIONS

CONDUCTIVITY OF

THE SWOLLEN

MEMBRANES

A

(EC:DMC 1:1

wt/wt) : PVDF

80:20 wt/wt

EC:DMC 1:1 wt/wt,

1M LiPF6

(LP30)

σ = 1.1 · 10-2

Scm-1

σ = 3.8 · 10-3

Scm-1

B

(EC:DMC 1:1

wt/wt) : PVDF

80:20 wt/wt

EC:DMC 1:1 wt/wt,

0.4M LiBOB,

0.6M LiPF6

σ = 8.6 · 10-3

Scm-1

σ = 2.0 · 10-3

Scm-1

C

(EC:DMC 1:1

wt/wt) : PVDF

80:20 wt/wt

EC:DMC 1:1 wt/wt,

0.7M LiBOB

σ = 7.2 · 10-3

Scm-1

σ = 1.5 · 10-3

Scm-1

Tab. 6-3 - Composition and conductivity values of the membranes and of the starting swelling

solutions The complete characterization of these membranes is planned in future works in virtue of the very good conductive values obtained.

58

6.4. New “in situ” GPE: preparation procedure

In order to reduce time and cost of the GPEs preparation route a different preparation procedure here described were successfully proposed. The gel polymer electrolyte was obtained by mixing two low-molecular-weight aprotic solvent such as ethylene carbonate (EC) and dimethyl carbonate (DMC) in a weight ratio of 1:1, with polyvinylidene fluoride (PVDF Solvay Solef® 6020) as plasticizer and swelling the obtained membrane in a solution that contains the conductive salt lithium exafluorophosphate (LiPF6) i.e. ethylene carbonate : dimethyl carbonate, lithium-hexafluorophosphate (EC:DMC 1:1 wt/wt, 1M LiPF6) from Merck. The composition of the membrane was: EC:DMC/PVDF 80/20 wt/wt. The preparation procedure of the gel polymer electrolyte is reported in the table below:

Temperature, °C Time

Mixing Room T 2h

Pre-heating 70°C 15min

Casting 70°C 20min

Quenching Room T 20min

Tab. 6-4 - Scheme of the preparation procedure

6.5. New “in situ” GPE: characterization

The ionic conductivity measurement was carried by AC impedance by placing in a “T-cell” a disk of the obtained membrane having a cell constant of k = 0.065 cm-1: the diameter was 10 mm and the thickness 0.5mm. After the first acquisition, the solution containing the conductive salt LiPF6 were added to the membrane and the measurements were performed during the time. The Fig. 6.10 and Fig. 6.11 reports respectively the Nyquist impedance plots in the 100 kHz - 100 Hz frequency range and the corresponding time evolution of the conductivity of the performed experiment in which it can be clearly observed the initial increasing and the final conductivity steady state upon LiPF6 diffusion in the membrane.

59

0 2 4 6 8 10 12 14 16 180

2

4

6

8

10

12

14

16

18

0 100 200 300 4000

100

200

300

400

EC-DMC-LiPF6 swelling EC-DMC-PVdF gel

Zre / kΩ

-Zim

/ kΩ

5 h 30 sec

-Zim

/ kΩ

Zre / Ω

Fig. 6.10 - Nyquist impedance plots during LiPF6 diffusion in the membrane

0 100 200 3001x10-6

1x10-5

1x10-3

1x10-2

Time / min

Con

duct

ivity

/ S

cm

-1

Fig. 6.11 - initial increasing and final conductivity steady state upon LiPF6 diffusion in the membrane

The electrochemical characteristics of the LNMO/EC-DMC-PVDF-LiPF6 in-situ gel /Li cell were investigated by using a membrane similar to the previous one prepared in situ in a T-type lithium cell using LNMO (i.e. LiNi0.5Mn1.5O4) standard cathode and Whatman separators. The in situ gelification procedure involved the preparation the solution above described (EC:DMC 1:1 / PVDF, 80/20 wt ) at 70°C and the subsequent infiltration in the cell hold at the same temperature, followed by a room temperature quenching. Following the LiPF6 was exchanged as previously described for a total duration of 200 min, i.e. until the conductivity steady state is reached. Then, the cell has been galvanostatically cycled at a C/5 rate (1C current being 0.187 Ag-1cm-2 with respect to LiNi0.5Mn1.5O4 active mass), within a 3.5 – 5.0 V voltage range. The performance of the cell was evaluated in terms of specific capacity and cycle

60

life. The figures reported below, showing the voltage profile (Fig. 6.12) and the cycling behaviour (Fig. 6.13) of the gel-type cell, evidences the expected voltage signature of the LNMO spinel and a satisfactory capacity retention, this confirming the validity of the in-situ gelification approach.

0 10 20 30 40 50 60 700

25

50

75

100

125

150

175

Spe

cific

cap

acity

/ m

Ah

g-1

Cycle number

Charge Discharge

0 40 80 120 1603.0

3.5

4.0

4.5

5.0

5.5

Vol

tage

/ V

Specific capacity / mAh g-1

Fig. 6.12 - Specific capacity vs. cycle number

for the Li / gel-type electrolyte / LiNi 0.5Mn 1.5O4 cell

Fig. 6.13 - Voltage vs. specific capacity for the Li / gel-type electrolyte /

LiNi 0.5Mn 1.5O4 cell

6.6. Conclusions

The results reported in this work demonstrate that membranes formed by combining an ionic liquid solution with a polymer matrix and with the appropriate addition of a carbonate solution have a series of favorable properties, including a high lithium ion conductivity, a wide electrochemical stability a high thermal stability and non-flammability, that make them suitable for application in advanced lithium polymer batteries. Was shown that in fact this type of membranes may be successfully used as electrolyte separators in batteries using a lithium metal anode and a lithium iron phosphate cathode. Efforts are now addressed to modifying and optimizing preparation procedure to obtain gel polymer membranes for safer lithium batteries.

61

References [1] J. Fuller, A.C. Breda, R.T. Carlin, J. Electrochem. Soc. 144 (1997) L67. [2] S. Megahed, B. Scrosati, Interface 24 (1995) 34. [3] T.E. Sutto, J. Electrochem. Soc. 154 (2007) P101. [4] A. Fernicola, F.C. Weise, S.G. Greenbaum, J. Kagimoto, B. Scrosati, A. Soleto, J. Electrochem. Soc. 156 (2009) A514. [5] C. Zhu, H. Cheng, Y. Yang, J. Electrochem. Soc. 155 (2008) A569. [6] J.H. Shin, W.A. Henderson, S. Passerini, J. Electrochem. Soc. 152 (2005) A978. [7] S.Y. Chew, J. Sun, J. Wang, H. Liu, M. Forsyth, D.R. MacFarlane, Electrochim. Acta 53 (2008) 6460. [8] J.-K. Kim, J. Manuel, G.S. Chauhan, J.-H. Ahn, H.-S. Ryu, Electrochim. Acta 55 (2010) 1366. [9] P. Reale, A. Fernicola, B. Scrosati, J. Power Sources 194 (2009) 182. [10] A. K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc. 144 (1997) 1188. [11] V. Gentili, S. Panero, P. Reale, B. Scrosati, J. Power Sources 170 (2007) 185

62

7. Technological approach to safety 7.1. Getters

The gases evolution within devices like electrolytic supercapacitors and lithium batteries is a serious issue that can lead to permanent failures in their functioning. Getters are used to prevent those kind of issues; they act as chemical absorber, capable of reducing or completely removing the presence of those gases. In this work, two different kind of getters produced by SAES Getters S.p.A. company have been used: metallic getters, based on titanium and composite polymeric getters. The getters from the first category above are used as hydrogen absorbers in electrolytic supercapacitors; the others are used as getters for lithium batteries. The electrochemical analysis that has been performed on the metallic getters allowed to evaluate the behaviour of getters in presence of currents/voltage applied to them. Those measurements were part of a more complete characterization that the vendor did previously, which was aimed to investigate the response of getters to many kind of stresses (e.g. temperature stress or mechanical stress). Furthermore, comparative analysis were performed on the behaviour of three different getters: the first and the second one had same chemical composition but different active material thickness, the third one with same chemical composition and thickness of the first one, but covered by a polymeric film on both sides. About the polymeric getters, instead, an analysis of the impact of the getter on a cell has been evaluated, focusing on overall impedance and specific capacity. In order to do such analysis, electrochemical systems have been developed in order to simulate the typical working conditions of lithium batteries.

7.2. Titanium-based metallic getters

7.2.1. Electrochemical characterization Three different hydrogen getters produced by SAES have been analyzed; two of them (DSC 25_150T and DSC 120_150T) were made of a titanium sheet (body) thick 150 µm, covered by a layer of palladium (active material), thick 25nm (for the first sample) or 120 nm (for the second sample); the third one has the same body, but the 25nm palladium layer itself is covered by a polymeric film on both sides (DSCM 25). The electrolytic solution is also provided by SAES S.p.A. and is composed of ammonium borate, with the addition of a small quantity of water which promote the hydrogen development: (NH4)3BO3 in CH3OHCH3OH. In some preliminary tests the ferricyanide/ferrocyanide has been added to the solution as internal reference: K4[Fe(CN)6] / K3[Fe(CN)6]. Before starting the analysis on the titanium-based getters, preliminary electrochemical analysis have been performed on the electrolytic solution: the

63

conductivity of the solution has been evaluated by an electrochemical impedance measurement and then cyclic voltammetry measurements have been performed in order to obtain an estimate of the potential of the decomposition reactions. The impedance measurement has been performed in the 5·105÷1 Hz frequency range, with a ±10 mV amplitude signal, using a conductivity cell for liquid samples, which is made of symmetrical platinized platinum electrodes. Distilled water was also added to the electrolytic solution. The voltammetry measurements, aimed to evaluate the potential of platinum as reference pseudo-electrode within the ethylenglycol solution, were performed at 5 mV/s in the -0.5÷0.5 V range. The cells used for that analysis were configured in the following way: working electrode and counter electrode made of platinum; reference electrode was made of platinum at first, then silver/silver chloride couple. The ferricyanide/ferrocyanide has been added to the electrolytic solution as internal reference. Afterwards, electrochemical measurements were made on getter by performing cyclic voltammetry measurements and electrochemical impedance measurements. Voltammetry measurements were performed using platinum as reference electrode and counter electrode and the getter as working electrode. These measurements have been repeated at different scan rate (5, 10, 20, 50 mV/sec) for each titanium/palladium getter, also varying the state of the solution from in agitation to steady. Here, the “pure” electrolytic solution has been used, which means without the addition of ferrocyanide and ferricyanide. Those measurement conditions allowed to get data about the process of decomposition of the electrolyte. Further measures of cyclic voltammetry were subsequently conducted by using the getter coated with polymer as working electrode and platinum as a the reference electrode and counter electrode. The measurements were made at a scan rate of 5 mV/s and keeping the solution in quiet. Even in this case the “pure” electrolytic solution was used, without the addition of ferrocyanide and ferricyanide. Finally various electrochemical impedance spectroscopy measurements were conducted on the three titanium based samples, according to the following procedure: 1) Open circuit; 2) Impedance (1·105 to 1·10-2 Hz, ∆v = ± 10 mV); 3) Relaxation (limit of the potential variation dV/dt < 1 mV/h); 4) Impedance (1·105 to 1·10-2 Hz, ∆v = ± 10 mV); 5) Polarization to-0.5V vs.Pt (linear scan at a rate of 5 mV / s); 6) Impedance (1·105 to 1·10-2 Hz, ∆v = ± 10 mV); 7) Relaxation (limit of the potential variation: dV/dt < 1 mV/h); 8) Impedance (1·105 to 1·10-2 Hz, ∆v = ± 10 mV). Through this procedure, the behaviour of the getter was evaluated in open circuit conditions, polarization and subsequent relaxation, which means simulating all possible operating conditions. Experimental investigations were also conducted in order to confirm the attribution of the peak between -0.3 V and -0.5 V vs. Pt to the decomposition of water: cyclic voltammetry measurements at different concentrations of water (1.2%, 5%, 10%, 20%, 50% by weight), made keeping the solution in quiet; cyclic voltammetry measurements with the solution diluted with ethylene glycol, also in quiet.

64

For both measurements the cells were assembled in this way: reference electrode and counter electrode were made of platinum and the getter DSC 25_150T and, just for comparison, DSC 120_150T was acting as working electrode. All the tests were executed using a 3-electrode polarographic glassy cell in which the getter was included as a working electrode, with an exposed area of 2 cm2. Also the getter coated with polymer was subjected to measurements of cyclic voltammetry, using the same solution of ammonium borate in ethylene glycol. The measure of impedance spectroscopy on the getter coated with polymer, DSCM 25, was performed according to a procedure designed to verify the stability of the interphase in different conditions: 1. At open circuit, 2. After slight perturbation (induced by the measurement itself), 3. In conditions of polarization to -0.5V vs. Pt, 4. In conditions of relaxation (after the reopening of the circuit).

7.2.2. Results and discussion The measurements of the conductivity of the electrolytic solution were carried out using a conductivity cell with platinized platinum electrodes at a temperature of 25°C. The obtained impedance spectrum, reported in Fig. 7.1, shows a complex behaviour because the trend is not straight as it typically is for an ionic conductor between blocking electrodes, but a semicircle at high frequencies, followed by a straight section. Considering the capacity of the semicircle obtained (i.e. 10-10 F), such a behaviour may be linked to the adsorption of the solvent and/or the electrolyte on the surface of platinum.

0 5000 10000 15000 200000

5000

10000

10Hz

(5 ⋅102Hz)

Zim

m /

Ω

Zreal / Ω

5 ⋅105Hz

2.3 ⋅103Ω 16 ⋅103Ω

Fig. 7.1 - Spectrum of impedance for the electrolytic solution

In any case, it was decided to extrapolate the intercept with the real axis at high frequency (i.e. 1.2·103 Ω) as resistance of the electrolyte, obtaining a conductivity of the order of 4·10-4 S/cm. This quite low value for the conductivity was expected, due to the viscosity and the low concentration of the electrolyte. The comparison between the voltammetry of ferricyanide/ferrocyanide systems with reference electrode made of platinum and silver / silver chloride couple has allowed

65

to estimate the potential of the pseudo-reference platinum in the solution of ammonium borate in ethylene glycol. This potential, as shown in Fig. 7.2, is around -150 mV vs AgCl/Ag, or about +50 mV vs. SHE.

-1.0 -0.5 0.0 0.5 1.0-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

Cur

rent

/ m

A

Potential / V

Fe3+/Fe2+ in Ethylene glycol (NH4)3BO3 vs AgCl/Ag Fe3+/Fe2+ in Ethylene glycol (NH4)3BO3 vs Pt

Fig. 7.2 - Comparison between the ferri / ferrocyanide systems voltammetry with Pt and AgCl / Ag

reference electrode

The voltammetric study of the getters in the solution of ammonium borate in ethylene glycol showed the presence of a faradic reaction at potential lower than 0.2V vs. Pt (Fig. 7.3): this reaction is attributable to the decomposition of the water present in the electrolyte solution.

DSC 25_150T DSC 120_150T

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Cor

rent

e / m

A

Potenziale / V vs Pt

soluzione in agitazione soluzione in quiete

Potenziale / V vs SHE

5mV/s

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Potenziale / V vs SHE

Cor

rent

e / m

A

Potenziale / V vs Pt

soluzione in agitazione soluzione in quiete

DSC 25_150T DSC 120_150T

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Cor

rent

e / m

A

Potenziale / V vs Pt

soluzione in agitazione soluzione in quiete

Potenziale / V vs SHE

5mV/s

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Potenziale / V vs SHE

Cor

rent

e / m

A

Potenziale / V vs Pt

soluzione in agitazione soluzione in quiete

Fig. 7.3 - Cyclic voltammetry measurements on getters in solution of (NH4)3BO3 in etilenglicole at 5

mV/s. Stirred solution and in quiet

This study was repeated under the same conditions, but varying the state of agitation of the solution and the scanning rate of the measurements; the results (Fig. 7.4) revealed limitations in the kinetics, which can only be ascribed to the slow diffusion of the reagent (water) in the ethylene glycol solution, quite viscose, but no substantial difference due to the different thickness of the layer of palladium.

66

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Potenziale / V vs SHE

Cor

rent

e / m

A

Potenziale / V vs Pt

5mV/s 10mV/s 20mV/s 50mV/s

Soluzione in quiete

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Potenziale / V vs SHE

Cor

rent

e /

mA

Potenziale / V vs Pt

5mV/s 10mV/s 20mV/s 50mV/s

Soluzione in agitazione

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Potenziale / V vs SHE

Cor

rent

e /

mA

Potenziale / V vs Pt

5mV/s 10mV/s 20mV/s 50mV/s

Soluzione in agitazione

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2-0.16

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Potenziale / V vs SHE

Cor

rent

e / m

A

Potenziale / V vs Pt

5mV/s 10mV/s 20mV/s 50mV/s

Soluzione in quiete

Fig. 7.4 - Cyclic voltammetry measurements on getters in a solution of (NH4)3BO3 in ethylene glycol

at various scan speeds. Stirred solution and in quiet. The decomposition of water on the surface of the getter is enhanced by the agitation of the solution; in quiet conditions, instead, the reaction ends after the first scan of potential, as a result of the surface depletion. The scanning rate does not significantly affect the current intensity, which demonstrates that the kinetic limitation is due to the concentration and not to the electrodic kinetic (e.g. charge transfer), which would be affected by the speed of polarization of the electrode. The impedance measurements were conducted on each electrode according to a protocol aimed to verify the stability of the interphase under different conditions: 1. At open circuit, 2. Following slight perturbation (induced by the measurement itself), 3. In conditions of polarization at the water decomposition potential, 4. In conditions of relaxation (after reopening the circuit). The open circuit potential of the getter within the cells resulted to be around 70mV vs Pt, constant in time and not affected by slight perturbations. Even after the polarization at -500mV, the potential quickly returns to a value close to the initial one. The impedance spectroscopy measurements on both samples showed that the getters have the same behavior, both in solution and under polarization (Fig. 7.5 and Fig. 7.6). In both cases it is possible to confirm the open circuit stability and a comparable capability of recovering the initial balance after the polarization. A fitting procedure of the obtained impedence spectra was performed by using an equivalent circuit that is shown in Fig. 7.7, using a resistor (Re, resistance of the

67

electrolyte) in series with a capacitor (QdS), in parallel to a resistance (Rt) which represents the getter / electrolyte interphase. Looking at the trend of the spectra it can be observed that the mesh on the processes at the interphase is significantly reduced due to the polarization, which confirms that redox processes were occurring, as already demonstrated by voltammetric measurements. At the polarized conditions, the spectrum is also quite noisy, probably due to the complexity of the process which includes the formation of a gaseous species and its adsorption in the bulk of the electrode.

0 20000 40000 60000 80000 100000 120000

0

20000

40000

60000

80000

100000

120000

-Zim

m /

Ω

Zreal / Ω

Pristine Pristine 2 Polarized Relaxed

DSC 120_150T

0 20000 40000 60000 80000 100000 120000

0

20000

40000

60000

80000

100000

120000 DSC 120_150T

-Zim

m /

Ω

Z real / Ω

pristine 70mV pristine 2 70mV polarized -500mV relaxed 60mV day after 80mV

Fig. 7.5 – Nyquist plot of two different getter DSC120_150T in solution of (NH4)3BO3 in ethylene

glycol. Reference electrode of platinum.

0 10000 20000 30000 40000 50000 600000

10000

20000

30000

40000

50000

60000

-Zim

m /

Ω

Zreal / Ω

Pristine Pristine 2 Polarized Relaxed

DSC 25_150T

0 10000 20000 30000 40000 50000 600000

10000

20000

30000

40000

50000

60000 DSC 25_150T

-Zim

m /

Ω

Zreal / Ω

Pristine Pristine 2 polarized relaxed day after

Fig. 7.6 - Nyquist plot of two different getter DSC25_150T in solution of (NH4)3BO3 in ethylene

glycol. Reference electrode of platinum

Fig. 7.7- Equivalent circuit used for the fit of the spectra of impedance

68

0.917.2E-55.6E35.4E3Polarized

0.803.9E-54.5E41.2E4Relaxed

0.844.7E-52.6E45E3Relaxed

0.857.5E-56E35E3Polarized

0.855.1E-55.8E45E3Pristine 2

0.855.1E-55.7E45E3Pristine

DSC25_150T_04

0.874.7E-56.2E45.4E3Day after

0.834.8E-52.0E45.2E3Relaxed

0.865.3E-54.3E45.4E3Pristine 2

0.865.3E-54.2E45.4E3Pristine

DSC25_150T_03

0.823.6E-54.4E47.9E3Relaxed

0.716.5E-51.2E47.9E3Polarized

0.843.56E-51.4E58E3Pristine 2

0.843.5E-51.4E58E3Pristine

DSC120_150T_03

0.843.6E-51.3E51.1E4Day after

…………Polarized

0.843.7E-51.1E51.2E4Pristine 2

0.843.7E-51.1E51.2E4Pristine

DSC120_150T_02

nQ / FRt / ΩΩΩΩRe / ΩΩΩΩSample

0.917.2E-55.6E35.4E3Polarized

0.803.9E-54.5E41.2E4Relaxed

0.844.7E-52.6E45E3Relaxed

0.857.5E-56E35E3Polarized

0.855.1E-55.8E45E3Pristine 2

0.855.1E-55.7E45E3Pristine

DSC25_150T_04

0.874.7E-56.2E45.4E3Day after

0.834.8E-52.0E45.2E3Relaxed

0.865.3E-54.3E45.4E3Pristine 2

0.865.3E-54.2E45.4E3Pristine

DSC25_150T_03

0.823.6E-54.4E47.9E3Relaxed

0.716.5E-51.2E47.9E3Polarized

0.843.56E-51.4E58E3Pristine 2

0.843.5E-51.4E58E3Pristine

DSC120_150T_03

0.843.6E-51.3E51.1E4Day after

…………Polarized

0.843.7E-51.1E51.2E4Pristine 2

0.843.7E-51.1E51.2E4Pristine

DSC120_150T_02

nQ / FRt / ΩΩΩΩRe / ΩΩΩΩSample

Tab. 7-1 - Result of the fit of the impedance spectra

The voltammetric study of the getter DSC25_150T in the solution of ammonium borate in ethylene glycol and water at various concentrations showed that the peak that is observed at about -0.4 V is not attributable to the decomposition of the water present in the electrolyte solution, and that’s because it is present in each scan, regardless of its content (Fig. 7.8). Instead the increase of the water content increases the intensity of current up to -1V vs. Pt; thus, this is the potential for water decomposition.

69

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e / m

A

Potenziale / V

Soluz. pura 5%

W H

2O

10%W

H2O

20%W

H2O

50%W

H2O

DSC 25_150T

Fig. 7.8 – Cyclic voltammetric response of DSC 25_150T getter in a solution of ammonium borate

in ethylene glycol with increasing water content

Further voltammetric scans were conducted using DSC25_150T getter; in this case the water content was reduced or more electrolytic solution was added in an equal amount of ethylene glycol, in order to obtain 1:1 solution in volume (Fig. 7.9).

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e/ m

A

Potenziale / V

Soluzione pura50%wEt-glicole

DSC 25_150T

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e/ m

A

Potenziale / V

Soluzione pura50%wEt-glicole

DSC 25_150T

Fig. 7.9 - Cyclic voltammetric response of DSC25_150T getter in a solution of ammonium borate in

ethylene glycol with decreasing water content

Those plots show that the peak at -0.4 V is still there, even diluting the electrolyte solution with ethylene glycol. This change in the composition seems to be influencing the shape of the diagrams at low potential region, where it is clearly shown that halving the water content also halves the value of measured current.

70

To get further confirmation of the results the same voltammetric measurements were performed also on the DSC 120_150T getter, by varying the water content; trends similar to the DSC 25_150T getter ones were observed (Fig. 7.10).

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e / m

A

Potenziale / V

Soluz. pura

20%w H

2O

50%w H

2O

DSC 120_150T

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e / m

A

Potenziale / V

Soluzione pura 50%

W Et-glicole

DSC 120_150T

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e / m

A

Potenziale / V

Soluz. pura

20%w H

2O

50%w H

2O

DSC 120_150T

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e / m

A

Potenziale / V

Soluzione pura 50%

W Et-glicole

DSC 120_150T

Fig. 7.10 - Cyclic voltammetric response of DSC 120_150T getter in solution of ammonium

borate in ethylene glycol and increases with decreasing water content

The voltammetric analysis on the DSCM 25 getter in the solution of ammonium borate in ethylene glycol showed that no significant electrochemical reaction takes place (Fig. 7.11). The behaviour that is observed, in fact, is typical of a non-conducting material.

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.020

-0.015

-0.010

-0.005

0.000

0.005

0.010

Cor

rent

e / m

A

Potenziale / V

I ciclo II ciclo

DSCM25

Fig. 7.11- Cyclic voltammetric response of DSCM 25 getter getter in solution of ammonium

borate in ethylene glycol The comparison of these results with those obtained under the same operating conditions with the getter DSC 25_150T, which is identical but the polymer coating, allows to appreciate even more the non-conducting effect of the polymer layer (Fig. 7.12).

71

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-0.45

-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

Cor

rent

e / m

A

Potenziale / V

DSC 25_150T DSCM 25

Fig. 7.12 - Comparison between cyclic voltammetric responces of DSCM 25 and DSC 25_150T

getters in solution of ammonium borate in ethylene glycol

The impedance spectroscopy measure on the getter coated with polymer, DSCM 25, showed that the getter in the relaxed state has a capacitive behaviour (Fig. 7.13) due to the presence of the non-conducting polymeric layer which makes it “blocking” of any redox reaction.

0 1x106 2x106 3x106 4x106 5x106 6x106 7x106

0

1x106

2x106

3x106

4x106

5x106

6x106

7x106

-Zim

m /

Ω

Zreal / Ω

Pristine1 Pristine2 Polarized Relaxed

DSCM 25

Fig. 7.13 - Nyquist plot of DSCM 25 getter in a solution of ammonium borate in ethylene glycol. Reference electrode of platinum

72

An equivalent circuit fitting is shown in Fig. 7.14, using a resistor (Re, resistance of the electrolyte) in series with a parallel of a capacitor (QdS) and a resistance (Rt) which represents the getter/electrolyte interphase.

Fig. 7.14 - Equivalent circuit used to fit impedance spectra.

It is possible to appreciate some charge transfer at the polarized state in the order of MΩ, negligible compared to that of the order of kΩ measured in the case of non-polymeric getter.

7.3. Polymeric composite getters

7.3.1. Electrochemical characterization The second set of tests was performed on samples of composite SAES getter constituted by a polymer matrix, wherein the getter material is dispersed. A two electrodes C/LP30/Li cell has been used:

Working electrode: graphite Electrolyte solution EC:DMC (1:1)/LiPF6 1M, Counter electrode: lithium.

The getter under investigation has been inserted inside of electrochemical cells with two electrodes in which the materials used as electrodes and the electrolyte solution were appropriately chosen to simulate conditions typical of lithium batteries. In particular graphite working electrodes were chosen, since they are known to be suffering the problem of development of CO2 more than others; also, those graphite electrodes are made from thin industrial films, therefore they are expected to be homogeneous and with repeatable behaviour. The electrolytic solution is based on carbonates (known as LP30) typical of lithium batteries. The experimental investigations have been carried out on the getter samples in order to assess the effect of the presence / absence of the getter on the performance of the cell in terms of overall impedance and specific capacity. 8 cells were assembled, identical each other from the electrochemical point of view; a 10mm diameter disk of the composite under investigation has been housed in 4 of them. Each measurement was carried out simultaneously on a cell without the getter and one containing the getter and then duplicated. Following measurements have been performed: Electrochemical Impedance Spectroscopy (open circuit)

Frequency range: 1·105÷1·10-1Hz

73

Potential variation range: ∆V=±5mV Galvanostatic cycling measurements

C rate set to C/5, Potential range 1·10-2÷2.5 V

All the tests were executed using a "T" cell made of polypropylene and steel, in which the getter occupies the position of the third electrode. All the cells were assembled in a controlled atmosphere (Argon) within a Glovebox.

7.3.2. Results and discussion The spectroscopic study was performed on two cells not containing the getter and two cells containing the getter, in order to evaluate the reactivity of the interphase. The measurements were made at open circuit, 24 hours apart each other, starting from the day of assembly for the next 7-5 days. As far as a two electrodes measurement allows to see, all the spectra follow the same trend due to the presence of a surface film which is not completely stable, but apparently not influenced by the presence/absence of the getter (Fig. 7.15 and Fig. 7.16).

0 50 100 150 200 250 300 3500

50

100

150

200

250

300

350

-Zim

m /

ΩΩ ΩΩ

Zreal / ΩΩΩΩ

Day 0 Day 1 Day 2 Day 5 Day 6 Day 7

Cell "01" without getter

C/LP30/Li

0 50 100 150 200 250 300 3500

50

100

150

200

250

300

350Cell "03" without getter

C/LP30/Li

-Zim

m /

ΩΩ ΩΩ

Zreal / ΩΩΩΩ

Day 0 Day 1 Day 3 Day 4 Day 5

Fig. 7.15 - Electrochemical impedance spectra of the two cells without getters

0 50 100 150 200 250 300 3500

50

100

150

200

250

300

350

- Z

imm

/ ΩΩ ΩΩ

C/LP30/Li

Cell "02" with getter 01

Zreal / ΩΩΩΩ

Day 0 Day 1 Day 2 Day 5 Day 6 Day 7

0 50 100 150 200 250 300 3500

50

100

150

200

250

300

350

C/LP30/Li

Cell "04" with getter 01

-Zim

m /

ΩΩ ΩΩ

Zreal / ΩΩΩΩ

Day 0 Day 1 Day 3 Day 4 Day 5

Fig. 7.16 - Electrochemical impedance spectra of the two cells with getters

74

The constant current cycling were conducted on four cells assembled ex-novo (two without getter and two with the getter). Those measurements results in both the potential and the specific capacity following trends almost identical both in the presence and absence of the getter object of measurement (Fig. 7.17 and Fig. 7.18).

0 2 4 6 8 10 120

50

100

150

200

250

300

350

400

C/LP30/Li

Cap

acità

spe

cific

a / m

Ah/

g

05 Senza Getter - charge 05 Senza Getter - discharge 06 Con Getter - charge 06 Con Getter - discharge

Misure di ciclazione galvanostatica

Cicli0 20 40 60 80 100

0

50

100

150

200

250

300

350

400Misure di ciclazione galvanostatica

C/5

C/LP30/Li

CicliC

apac

ità s

peci

fica,

mA

h/g SENZA GETTER carica scarica

CON GETTER carica scarica

Fig. 7.17 - Behavior in charging and discharging in terms of specific capacity as a function of the number of cycles of cells with and without getters

0 50 100 150 200 250 300 350 400

0.0

0.5

1.0

1.5

2.0

2.5

3.0

C/LP30/Li

Misure di ciclazione galvanostatica

Ten

sion

e / V

Capacità specifica / mAh/g

05 Senza getter 06 Con getter

Fig. 7.18 - Potential profile of cells with and without getters

7.4. Conclusions

The electrochemical study showed no significant differences in the behaviour of the getter in different thicknesses. Both the getters have an open circuit potential of around 70 mV. From cyclic voltammetry measurements a water decomposition process is shown at a potential lower than 0.2 V. This process is influenced by kinetic limitations arising from the slow diffusion of water in the solution, but not from the intrinsic properties of the getter. The impedance measurements indicate a similar behaviour for the two getters. The effect of the polarization on the impedance is the reduction of the interphase resistance, due to the reduced resistance to charge transfer in the presence of hydrogen evolution redox process.

75

Based on the voltammetric analysis on the electrolytic solution at different concentrations of water and ethylene glycol, it can be concluded that the water decomposition potential is around -1 V. Thus, the peak observed at -0.4 V seems to be caused by the salts dissolved in the solution. The study conducted on electrochemical getter coated with polymeric film did not reveal significant reactions, therefore it can be concluded that the behaviour is typical of an insulating interphase. From the impedance measurements it can be confirmed that the DSCM 25 getter, is a non-conductive interphase, since it shows the typical behaviour of a capacitor. From the investigation on the polymeric composite getter it can be concluded that the presence/absence of the getter does not affect performance of the cells during the galvanostatic cycling neither positively nor negatively.

76

8. General conclusions Lithium-ion (Li-ion) has become the dominant rechargeable battery chemistry for consumer electronics devices and is poised to become commonplace for industrial, transportation, and power-storage applications. From a safety standpoint, a high energy density coupled with a flammable organic, rather than aqueous, electrolyte has created a number of new challenges with regard to the design of batteries containing lithium-ion cells, and with regard to the storage and handling of these batteries. Note that energy storage is an area of rapidly evolving technology. There are a number of efforts underway to commercialize cells with different chemistries than lithium-ion including rechargeable lithium metal cells, ultracapacitors, and fuel cells. Electrolyte chemistry is an active area of research. A number of groups have conducted research to produce non-flammable, or reduced flammability electrolytes either through the addition of additives to typical organic solvent mixtures, or through the development of non-organic ionic liquids. The present PhD thesis showed a multiplicity of the possible approaches for solving the safety issue in lithium batteries. In each chapter has a single approach has been described. Ionic Liquids resulted to be very good candidates for completely or partially substituting traditional alkylcarbonate-based liquid electrolytes, due to their good electrochemical and thermal performances. This work explored the behaviour of Py24TFSI IL, which was synthesized in our lab, and Py14TFSI, which is a commercial product. The usage of non-fluorinated lithium salt as additive in an electrolyte solution has been studied as alternative approach to safety. LiBOB salt was studied in this work as a substitute for LiPF6, in order to eliminate the production of dangerous HF gases. Electrochemical analyses confirmed that a alkylcarbonate and LiBOB salts is suitable for battery application. As additional approach being studied in this work, it was also demonstrated that membranes formed by combining an ionic liquid solution with a polymer matrix and with the appropriate addition of a carbonate solution have a series of favorable properties, including a high lithium ion conductivity, a wide electrochemical stability a high thermal stability and non-flammability, that make them suitable for application in advanced lithium polymer batteries. A technological approach to safety was also investigated in this work; in order to evaluate if the usage of getters is a viable solution to the safety issue, the electrochemical performances of some commercial getters has been studied. From that investigation it can be concluded that the presence of the polymeric composite getter does not affect the cell performances during the galvanostatic cycling.

77

The effectiveness of the explored approaches against the safety issue has been evaluated in different ways: flammability tests on some IL-solutions being used have confirmed their non-flammability, which was already know from literature; the substitution of LiPF6 with LiBOB itself makes the electrolyte solution safer, since it excludes the HF gas release; the usage of a membrane reduces the volatility of the substances within the electrolyte since they are physically trapped in a matrix; the reduction of gases developed within the cell, reduces the risk of flame development or explosion. The most appealing solution to the safety issue can be a combination of the approaches studied in this PhD thesis. The most promising way seems to be a combination of the modularity and reliability of the polymer membranes with the non-volatilty and non-flammability properties of the ionic liquids, but also with the high stability of the LiBOB salt.

78

Appendix – Experimental methods Methods of thermal analysis

Thermal analysis is defined by the International Confederation of Thermal Analysis and Calorimetry (ICTAC) [1] as “a group of techniques in which a property of a sample is monitored against time or temperature while the temperature of the sample, in a specified atmosphere, is programmed.” In practice, the temperature of the oven that contains the sample actually is programmed, while the temperature of the sample in some cases may differ from the programmed temperature. Exothermic or endothermic reactions or phase transitions in the sample subjected to the programmed temperature variation may cause variations in the temperature between the sample and oven up to several degrees. The experimental work of this thesis has required the use of differential scanning calorimetry (DSC) and thermogravimetry (TGA) techniques such as thermal characterization of materials under investigation.

Differential Scanning Calorimetry (DSC) The differential scanning calorimetry technique, commonly indicated by the acronym DSC (Differential Scanning Calorimetry), is the most technique for characterizing the thermal parameters of a material in a quantitative manner. Using this technique, the difference in the amount of heat required to increase the temperature of a sample and a reference is measured as a function of temperature [2]. The DSC, in fact, measures the deviation of the heat flow that passes through the sample compared to an inert reference material, while both are subjected to the same temperature variation. The difference between the flows of heat is constant up to when chemical-physical phenomena (e.g. state transitions, decompositions, other chemical reactions), occur in the sample, determining emission or absorption of heat. Two different kind of DSC instrumentation exist: heat flux DSC instrumentation, in which both the sample and the reference are in the same compartment; power compensation DSC, in which the sample and the reference are in two separate compartments. In this work, a heat flux DSC has been used (i.e. METTLER TOLEDO DSC 821e). Two crucibles of aluminium are housed in the same compartment, one containing the test sample and the other empty used as reference); each of them is in contact with a thermocouple which allows the monitoring of the temperature. This compartment is immersed in the thermal field of a furnace, kept in a nitrogen atmosphere, whose temperature is set by means of a linear programmer, capable of maintaining a speed of heating (or cooling) constant. Initially, the sample and the reference are in thermal equilibrium. Instead, when the sample undergoes an endothermic or exothermic change in physical-chemical nature, a temperature variation between them occurs. As soon as the temperature variation is measured by the temperature sensors, a control circuit restores the thermal equilibrium between the sample and the reference. There is thus a current signal (differential)

79

proportional to the flow of heat absorbed or evolved from the sample, which is displayed as a function of temperature in a thermogram. Fig. A.1 shows the typical trend of a DSC thermogram obtained from analysis of a semi-crystalline polymer that has been subjected to rapid cooling (quenching).

Fig. A.1 - Trend of a typical thermogram obtained from a semi-crystalline polymer that has been

subjected to quenching. Tg is the glass transition temperature; Tc is the crystallization temperature of the crystallizable phase; Tm is the melting temperature.

The processing of the data provides information on the physical-chemical properties of the sample under test. Through differential scanning calorimetry, phenomena like melting, crystallization and glass transition can be observed, and also the specific heat, latent heat of phase transition, and the degree of crystallinity can be quantitatively determined.

Thermal Gravimetric Analysis (TGA) The thermogravimetric analysis, also known with the acronym of TGA, is defined as a technique in which "the weight of the sample is monitored as a function of temperature or time, while the sample is subjected to a controlled temperature program" [3]. The thermogravimetric analysis can be carried out in two different modes: isothermal thermogravimetry and scanned thermogravimetry. The first method is to monitor the changes over time in mass of a test sample at a constant temperature in a controlled atmosphere. With the thermogravimetry scan, instead, it records the weight change of the sample as a function of temperature, which varies linearly in time with a heating rate fixed. In this work, second operating mode has been used. The result of thermogravimetric analysis is displayed in a thermogram in which the loss in weight of the sample is reported as a function of time, in the case of isothermal thermogravimetry, or as function of temperature, in the case of scanned thermogravimetry. Steps can be observed on the obtained plot, in correspondence of mass losses. Also the speed of occurring processes can be evaluated from such kind of plot; looking at the plot of the derivative function (DTA, Derivative Thermogravimetric Analyses), in fact, peaks indicating the temperature at which the process takes place at with the maximum speed can be observed. Moreover, the derivative function allows to

80

distinguish reactions occurring at temperatures very close to each other, since distinguishing two peaks is easier than deconvoluting two overlapping steps. By definition, the thermogravimetric methods are limited to the study of phenomena involving a variation in weight of the sample, as the reactions of decomposition or oxidation, or physical processes of evaporation, sublimation or desorption. In this experimental work, the METTLER TOLEDO TGA / SDTA 851st instrument has been used. The material to be analyzed is inserted into a sample holder of alumina, resistant to high temperatures, and set on a thermobalance which is located inside furnace from which, however, is thermally isolated. The oven can reach temperatures above 1000 ° C, and during its operation it is subjected to a gas flow that acts as a purging system from the products of the sample degradation, which could otherwise saturate the environment and limit the decomposition process in place. In general, inert purge gases (nitrogen or argon) are used, but it is also possible to supply air or oxygen in case you want to study the oxidation phenomena. The flow rate of the purge gas determines the speed of removal of the volatile products, and therefore affects the kinetics of the process. Another parameter that can affect the analytical result is the heating rate. For kinetic reasons, in fact, high heating rates can easily determine temperature gradients within the samples, which cause a delay in the process under investigation. At low heating rate, instead, the step representing the loss in weight of the material under test is wider.

Techniques for electrochemical analysis

The electrochemical characterization of the materials studied in the present work has required the measurement of their ionic conductivity and the determination of the window of electrochemical stability.

Electrochemical impedance spectroscopy According to Ohm’s law, the current (I) through a conductor between two points is directly proportional to the potential difference (V) across the two points and inversely proportional to the conductor resistance (R):

In the case of conductors of the second kind, the relation above is not verified, since the electrolyte resistance is itself function of the applied potential:

In fact, the resistance of ionic conductors depends on many variables, like the state of charge of the mobile species, the concentration gradients and the viscosity of the medium. Since the application of a DC voltage (V) causes phenomena of polarization at the electrodes and possible variation of composition of the electrolyte as results of

81

electrolysis reactions, it is useful to measure the resistance of the electrolyte when applying AC voltage. Conductivity measurements are then conducted on the electrochemical cell by applying a periodic voltage V given by the following equation

, which generates the periodic current I as effect of the disturbance given by

. Vmax and Imax are, respectively, the input voltage signal amplitude and the resulting current amplitude; is the oscillation frequency, is the phase difference between voltage and current. The electrochemical impedance (Z) is defined by the generalized Ohm’s law as the ratio between the input voltage (V) and the resulting current (I):

A commonly used representation of the impedance is a vector on the complex plane, whose real part and imaginary part are:

where,

The spectrum of impedance is the result of applying a periodic perturbation of variable frequency to an electrochemical cell, and can be viewed through the Nyquist graphs, in which the abscissa is the real part of the impedance Zr, and the ordinate the imaginary part Zi (changed in sign). At each point, the distance from the origin represents the modulus of the complex impedance at a given frequency. For small perturbations, the impedance depends only on the frequency ω, and not by the applied voltage. In other words, the electrochemical system has a linear response to small perturbations. Thanks to this property, the behaviour of the system can be simulated as an equivalent electric circuit composed of simple elements (resistors and capacitors) connected in series or in parallel. Each circuit component can be associated to phenomena of migration, diffusion and charge transfer occurring within the electrochemical cell. Typically, the resistances represent processes which limit the phenomena of conduction, while the capacities represent phenomena of charge accumulation. The lithium-electrolyte interphase stability measurements were conducted through measurement of electrochemical impedance, made using VersaSTAT MC multi-channel

82

potentiostat / galvanostat Princeton Applied Research. Instead, the impedance spectra for ionic conductivity measurements were obtained using a frequency response analyzer Solartron FRA 1255. Both instruments are interfaced to a computer in order to capture and store the data, which were then processed using the "Equivalent Circuit". Such data processing software, which is a version of that described by Boukamp [7-8], allows to interpret the experimental results according to the model of the equivalent circuits. Of course, the equivalent circuit that describes a particular system is assumed a priori, taking into account both the impedance spectrum trend and the physical processes potentially observable in the experimental conditions in which the measurement was made. The SW application allows to subdivide the overall circuit into its constituent elements, and thus to distinguish the various contributions to the total impedance. The interpolation of the experimental data was done through a non-linear regression using the least squares method. The goodness of the regression is assessed through the test of Kramers-Kronig, that binds the real part and the imaginary part of the complex impedance, and provides a value of the random variable χ2, which is the smaller, the greater is the agreement between the experimental data and the interpolated curve, ie the more the chosen equivalent circuit is suitable to describe the real electrochemical system.

Voltammetric techniques The term voltammetry refers to a collection of potentiodynamic analysis techniques, based on measurement of the current I which passes through an electrochemical cell subjected to a scan of potential V. The resulting diagram I (V), or voltammogram, allows to observe peaks associated to redox reactions involving the electrolyte in the range of potential investigated. This voltammogram data allows to retrieve both qualitative (peak potential) and quantitative (peak area) information about the redox processes involved [8]. These techniques are applied on three electrodes cells composed of a working electrode (W), a reference electrode (A) and a counter electrode (C). A potential ramp is applied between the working electrode and the reference one, while measuring the current flowing between the working electrode and the counter electrode. In this way, the overvoltage due to the current flow and the ohmic drops caused by the resistance of the electrolyte are excluded. Consider a solution containing a generic electroactive species Ox. Applying a voltage scan to an electrochemical cell composed of the above mentioned electrolytic system interposed between a working electrode and a counter electrode, it is possible to record the current generated by the following reduction:

Ox + n e– Red If the reaction is reversible, the concentrations of the oxidized state [Ox] and the reduced one [Red] of the electroactive redox pair at the electrode surface are related to the potential of electrode (E) at every instant, by the Nernst law:

83

[ ][ ]d

Ox

nF

RTEE

Relog0 += ,

where E0 is the default potential of the Ox/Red pair, R is the universal gas constant, T is the absolute temperature, n is the number of exchanged electrons, F is the Faraday constant. Suppose to make a cathodic scan (in reduction) in a potential range that includes the standard potential of the Ox / Red pair. During the scan, therefore, the applied potential will increase from a value higher than the standard potential of the Ox / Red pair to a lower value, linearly varying with time. During the first section of the scan, a very low residual current is measured, due to many causes, such as the resistance of the cell and the background noise of the circuit. While the potential decreases, the concentration [Ox] decreases near to the electrode, due to Ox reduction. Thus a concentration gradient is created and other Ox species diffuse from the body of the electrolyte towards the electrode; a cathodic current proportional to the concentration gradient flows in the external circuit. When the reduction potential of Ox is reached, the [Ox] / [Red] ratio at the electrode / solution interface becomes unitary, and the current passing through the cell is maximum. A further decrease of the applied potential results in a progressive decrease of the cathodic current. In fact, the concentration gradient tends to zero, and thus the diffusion of Ox species towards the electrode surface is gradually slower. In other words, even if all Ox species reaching the electrode surface immediately reduces, the current decreases since the number of Ox species reaching the electrode per unit time is very low. This process shows up as a peak. The peak potential is characteristic for each substance, and is thus a parameter of interest in the qualitative analysis of a solution, because it allows the identification of any present oxidant or reductant species. The current density that corresponds to the height of the peak is instead proportional to the concentration of the electroactive species in solution and therefore constitutes the voltammetric parameter of interest for a quantitative analysis. The electrochemical process is fully reversible if the rate of charge transfer is much greater than the diffusion rate, which, in this case, regulates the kinetics of the electrodic process. In that case a perfectly symmetrical peak on the voltammogram is obtained. Conversely, in a completely irreversible system, the speed of charge transfer is much lower than the rate of diffusion and therefore determines the kinetics of the process; the redox system is characterized by strong overvoltage phenomena, which means that the redox potential lags behind the one indicated by the Nernst law. In intermediate cases, the partial irreversibility of the system is reflected in the voltammogram determining the asymmetry and the enlargement of the peak. Based on these considerations, it is evident that the scanning speed of the potential is a parameter that can significantly affect the voltammetric analysis. In fact, the lower is the scanning speed, the more the kinetics of the process is governed by the diffusion of the electroactive species, and so the system tends to be reversible [9]. The voltammetric techniques used in the present work are linear sweep voltammetry and cyclic voltammetry. In the linear sweep voltammetry (LSV) a voltage linearly varying with time is applied to the electrochemical cell. As previously explained, the presence of peaks in the

84

voltammogram indicates the presence of electroactive species in solution that undergo redox reactions in the range of explored potential. The LSV technique was used during the experimental work both in order to identify any impurities present in the samples under examination and to determine the range of electrochemical stability of the electrolyte, defining the potential of decomposition (anodic and cathodic limit). In the cyclic voltammetry (CV) a triangular potential sweep is applied on the electrochemical cell, in order to induce oxidation and then reduction (or vice versa) of an electroactive species. The voltammogram has a closed loop shape, which significantly varies depending on the degree of reversibility of the redox system being studied. A reversible system gives rise to a path in which both the anodic and cathodic peak are visible, at a distance of 59.6 / n mV from each other (at 25° C), regardless of the speed of sweeping, the explored interval of potential and the concentration of the electroactive species. Through the CV, therefore, it is possible to determine the number n of electrons involved in a sufficiently reversible redox reaction. The irreversibility of the reaction can instead be immediately displayed, not only by the distance between the anodic and cathodic peaks, but also by the decrease in the peaks area as the number of cycles increases. If the reaction is irreversible, in fact, the electroactive species are progressively consumed, so the amount of charge used by the system decreases with the number of cycles [9]. The voltammetric measurements have been performed using the VersaSTAT MC multi channel potentiostat/galvanostat by Princeton Applied Research. In the experimental work, the voltammetric analyses were performed with the PAR 362 Scanning Potentiostat instrument. The potentiostat applies a potential difference between the working electrode (WE) and the counter electrode (CE), by varying the potential at the WE in a controlled manner with respect to a reference electrode (RE). In the circuit that includes WE and RE there is no current flow, and the measured current is only the one flowing between WE and CE. The instrument is interfaced with a computer that enables the acquisition and storage of data through a dedicated software.

Galvanostatic cycling Galvanostatic techniques are based on the application of a constant direct current to an electrochemical cell, and on the measurement of the voltage over time. Let’s assume a Li/electrolyte/LiAxBy electrochemical cell. The application of a constant current to that system determines the following reaction:

LiA xBy Li+ + AxBy + e–

The above written reaction goes from the left side to right side when the current is positive, i.e. during the device charge, until the anodic inversion limit is reached, and from the right side to the left side during the discharge phase, up to the cathodic inversion limit. The analysis of cell voltage profile during a charge and discharge cycle provides information on the phenomena of overvoltage and ohmic drop affecting the electrochemical cell. The ohmic drops are mainly caused by the resistivity of the

85

electrolyte medium, while the overvoltage may be attributable to both the process of charge transfer and the diffusion of the Li+ ions in the electrolyte and in the electrode body. The charge and discharge cycles are typically repeated hundreds of times, and from the data analysis many information are extracted, such as, for example, the specific capacity, the cycling Coulombic efficiency and the lifecycles. The specific capacity is the amount of charge flowing through the electrochemical cell during charge or discharge per unit of mass (specific capacity weight, expressed in mA h g-1) or volume (volumetric specific capacity, expressed in mA h dm-3) of the electrochemical cell. The cycling Coulombic efficiency (η) indicates the level of reversibility of the electrochemical process and is given by:

red

ox

Q

Q=η

Qox is the amount of charge that flows through the cell during the charge phase and Qred is the amount of charge that flows during the discharge phase. The difference between Qox and Qred are often due to undesired reactions happening within the electrochemical cell, mainly involving the electrolyte. The system lifecycles (i.e. the number of cycles that the system can undergo prior to fall below 75% of original capacity) can be obtained by plotting the specific capacity over the performed cycles.

86

References [1] J ICTA, International Confederation for thermal analysis, ASTM E 474 [2] J ICTA, International Confederation for thermal analysis, ASTM E 914 [3] J.R. MacCallum, C.A. Vincent, Polymer Electrolyte Reviews - 1, Elsevier Applied Science [4] Impedance Spectroscopy, Ed. J. Ross MacDonald, John Wiley & Sons, 1987 [5] P. Mazzoldi, M. Nigro, C. Voci, Fisica - vol II, 2nd ed., 2000, EdiSES srl [6] B.A. Boukamp, Solid State Ionics, 18-19 (1986), 136 [7] B.A. Boukamp, Solid State Ionics, 20 (1986), 31 [8] Instrumental Method in Electrochemistry, 1985, Ed. Ellis Horwood Limited [9] W.A. Henderson, V.G. Joung Jr., W. Pearson, S. Passerini, H.C. De Long, P.C. Trulove, Journal of Physics: Condensed Matter, 18 (2006), 10377.

87

Acknowledgements This work has been performed with the financial support of:

- SAES Getters S.p.A. – Milan – Italy - European Community within the Seventh Framework Programme APPLES

(Advanced, High Performance, Polymer Lithium Batteries for Electrochemical Storage) Project (contract number 265644).

I would like to thank my tutor Prof. Stefania Panero for helping me in my professional growth and Prof. Bruno Scrosati for giving me the opportunity to work in his group. My sincere gratitude goes to Dr. Maria Assunta Navarra, Dr. Priscilla Reale, Dr. Sergio Brutti, Dr. Jusef Hassoun, Dr. Judith Serra Moreno.

Finally, I also thanks Prof. Fausto Croce, Dr. Alessandra Fernicola, Sonia Cirinnà and everyone who have supported me and encouraged during my work.