EMULSION-TEMPLATED SILICON/CARBON ANODES WITH REDUCED GRAPHENE OXIDE

...................................................................................................... ii ACKNOWLEDGMENTS ................................................................................. iv TABLE OF CONTENTS ................................................................................... v LIST OF TABLES .......................................................................................... viii LIST OF FIGURES ........................................................................................... ix CHAPTER


Motivation
Still representing the highest performing secondary battery systems [1], [6], [7], the lithium-ion batteries (Li-ion batteries), based on the intercalation concept proposed by M. Whittingham [8] in the '70s, nowadays play an essential role in modern technologies, being in particular the best battery technology now available for vehicles. In portable electronics and mobile communication devices they are already commonly used and with the growing market in electrical mobility they are now entering the markets of hybrid and electrical vehicles. Many other "green" technologies, such as the solar cells, benefit from the possibility to produce less expensive batteries with extremely reduced sizes and long cycle life. Expectations of convenience and long-living portable power urged to develop technological strategies that resulted in a net improvement of the batteries performances. These advances can be better appreciated if one considers that in the last decade the energy density has been improved two times. But obviously, a great breakthrough is needed in order to increase the energy density from the current 210 Wh kg -1 of Li-ion batteries to the ambitious target of 500-700 Wh kg -1 to satisfy application in electrical vehicles before 2030 [9]. The energy density per unit area is a critical figure of merit for power modules, whereas for other applications, such as electrical vehicles, the density per unit weight is the key parameter. To reach high energy density the anode materials must combine high specific storage capacity and Coulombic efficiency. [10], [11] Although the good performances of the graphite materials in terms of electronic conductivity, low electrochemical potential and Coulombic efficiency 2 (>95%), the low specific lithiation capacity (372mAhg -1 ) [12] limit the possibility to force up the efficiency and to meet the ever increasing requirements of our society. In order to enhance the performances of such batteries, in the place of graphite several different anode materials with higher specific capacity of lithium (Li) accommodation are thinkable. Without any doubt, the most promising element is silicon (Si), characterized by a high theoretical specific capacity that is an order of magnitude beyond that offered by conventional graphite anodes [4], [13]. In this work, a method for the production of silicon/carbon (Si/C) anodes is treated and improved.

1.2 Thesis outline
This study has two main objectives; the increase of the Si/C ratio in an anode produced by emulsion-templated direct assembly and the introduction of reduced graphene oxide (r-GO). Previous work of my research group by Chen et al. (2014) [5] has shown, that it is possible to prepare high capacity, stable Si/C Anodes for Li-ion batteries using this method. This study is about improving this approach.
A porous anode Si/C alloy material was formed using CB particles to stabilize a Pickering emulsion which is then simply dried on a current collector. The anode material was physically characterized using light microscopy (LM), field emission and cryo-scanning electron microscopy (FE-SEM and cryo-SEM), transmission electron microscopy (TEM) and energy-dispersive x-ray spectroscopy (EDS). Furthermore the assembled batteries were electrically tested as half cells to determine capacity and cycle stability. The batteries were tested under conditions designed to evaluate different aspects of their performance.
In order to provide the reader with the necessary knowledge base to utilize the content of this thesis, a review of current status of the research and development of Si anodes will be given first. Due to the broad spectrum of the main topics, this review is rather extensive, and contains two main parts; the Li-ion battery in general and the Si anode. This is followed by the theory necessary to understand the experimental work which is illustrated briefly.  [14]. A comparison of the specific power and energy of different battery technologies can be seen in Figure 2-1, showing the very good performance of two of the most common lithium-ion technologies. The difference between these lies in the use of different cathode materials, a topic that will be described in detail later.
This section will first introduce the field of Li-ion batteries, beginning with a brief explanation of the working principle of the lithium-ion battery. An overview of important characteristics of batteries will then provide, an explanation and definition of some expressions that are used later in the thesis. An important aspect of the Li-ion battery, especially with regard to anode materials; the solid electrolyte interface (SEI), will then be explained. In addition, a review of materials used for other components and purposes in Li-ion battery is also deemed while the research performed in this study mainly concerns the Si anode.

Working Principle
Secondary battery systems in their simplest form consist of two electrodes immersed in an ionically conducting electrolyte and electronically connected through an external circuit. The structure of the Li-ion battery does not differ notably from this. Through a redox reaction the battery converts chemical energy to electrical energy during discharge. The reduction and oxidation reactions happen on different electrodes, and as these are separated by a non-electronically conducting electrolyte, the electrons are forced through the external circuit, where a load can be applied.
During charging, by driving this redox reaction in the opposite direction chemical energy gets generated from electrical energy. The cathode and anode are by definition the electrodes where the reduction and oxidation reactions happen, respectively, and which electrode is which therefore changes depending on whether the battery is 7 charged or discharged. In battery science the convention is to designate the electrodes by their function during discharge. [16] In addition to the many similarities between the Li-ion battery and other battery chemistries, there are some important differences. The most obvious being the use of lithium ions, which is one of the most reductive elements. The Li/Li + couple has a reduction potential of -3.04 volts vs. the standard hydrogen electrode [17].
Combined with a low atomic weight of only 6.94 u [18], Li meta is ideal for energy storage, yielding both a high number of electrons per mass and high energy per electron. Ideally, to obtain the highest possible capacity, the anode should be pure lithium, as is the case for the lithium metal primary battery. However, there are obvious problems when attempting to charge a Li metal battery. Not only is the high chemical reactivity of metallic lithium afflicted with inherent risk, one also encountered issues with Li being deposited on the anode in the form of sharp dendrites able to pierce the separator and cause a short circuit. These problems are severely reducing the reliability of the batteries, and also exhibit a risk for thermal runaway reactions. In cases where lithium is involved these can pose a serious safety hazard because of being most violent [14].
This was the motivation behind the research that led to the Li-ion battery, which solves these problems by using so called intercalation electrodes as both anode and cathode, meaning that Li ions can be reversibly inserted and extracted from the electrodes, rather than deposited on their surfaces. A spontaneously movement from the anode to the cathode is caused by a potential-difference between the electrodes causing Li ions to through the electrolyte. As the electrolyte is electronically isolating ,   8   the electrons are left to travel through the external cycle, where their electrical energy can be used.
Schematic of a lithium-ion battery. In this case, graphitic carbon is used as an anode and a relevant transition metal oxide as cathode. On charging, Li ions are removed or deintercalated from the layered oxide compound and intercalated into the graphite layers. The process is reversed on discharge. [15] An applied voltage forces the electrons and Li ions to move in the opposite direction, during charging. A Li-ion battery is represented in Figure 2-2. This battery is using a common combination of intercalation electrodes; a graphite anode and a layered transition metal oxide cathode. The schematic structures for the charged and discharged anode and cathode can be seen in Figure 2-3, along with the half-cell reactions and full cell reaction for this example. The factor in these reactions vary depending on which transition metal oxide is used, but is usually between 0.5 and 1.
At open circuit conditions the electrons are hindered from moving, establishing an electric field nullifying the potential difference between the electrodes and halting the reaction by the transport of ions. As soon as the external circuit is closed, the reaction is free to continue and hence, energy can be released.

Important Characteristics
• Specific capacity is the capacity of the whole cell or the electrode capacity relative to its weight and is given in mAh/g. The relationship between the capacity of the entire cell and the capacities of its components is commonly expressed Where " is the anode capacity, # is the cathode capacity and 1 $ % ⁄ is the capacity specific mass of the remaining cell components; electrolyte, casing etc.
• Charge density is the capacity of the whole cell or the electrode capacity, but relative to volume rather than weight, and is usually given in mAh/cm 3 .
• Capacity retention is a measure of how well the battery copes with being charged and discharged. It is representing the cycle stability of a cell, and is represented by a certain percentage of its initial capacity.
• C-rate is a normalized charge/discharge rate that depends on the electrode or battery capacity. It is defined as the rate at which the entire capacity of the electrode is charged or discharged in one hour.
Charge/discharge rate during cycling is usually given as a fraction or multiple of the C-rate. E.g. for a 4 gram electrode with specific capacity of 1000mAh/g the C-rate is 4000mA, and if cycled at for instance C/8 the current is set to 500mA.
• Coulombic efficiency is defined as the ratio between the amount of charge that can be extracted from the battery and the amount of charge that was put in during charging, typically given in percent.
• Charge and discharge in battery science, are defined as a forced or spontaneous reaction respectively in the battery. For an anode in a full cell, this means that Li intercalation would be denoted charging.
However, when coupled with Li metal counter electrodes in half cells, anode materials will in fact act as cathode, hence lithiation would constitute discharging. To prevent ambiguity, in this work, lithiation of anode materials will always be denoted charging and delithiation discharging, regardless of its actual role in the cell.

SEI Formation
There is only a certain range of redox potentials where all compounds are stable, below this range the compound is reduced and above it is oxidized.
Considering that e.g. for a standard graphite anode most of the intercalation happens <0.1 volt relative to the Li/Li + redox couple [18 ], intercalation of Li into the anode of the Li-ion battery happens at a very low potential, and decomposition of the electrolyte solution constituents is thus virtually inevitable and will happen to at least some degree. By using an electrolyte that, together with the electrode material and Li, decomposes into stable solid compounds and therefore form a coating around the electrode, this problem is circumvented. The coating, commonly referred to as the solid electrolyte interphase (SEI), should ideally prevent further decomposition of solvent by forming an impermeable as well as electronically isolating layer, but still be ionically conducting to allow Li ions to pass from the electrolyte to the electrode.
The SEI forms through a number of parallel and competing reactions, resulting in a coating which is complicate to characterize and to evaluate due to its inhomogeneous composition and varying thickness. representation of SEI of a standard carbon anode formed on a graphite particle. It stands to reason, that the composition and efficiency of the SEI is greatly influenced by the composition of the electrolyte. The complexity of the formation makes it sensitive to external influence as well, e.g. varying operating conditions in different laboratories. However, for carbon anodes, which are the most extensively researched, most groups agree that the SEI is composed of multiple layers; a layer consisting mostly of inorganic salt reduction products closest to the electrode followed by a layer of organic reaction products from the decomposition of the solvents [20].  [20] While being essential for the successful operation of the Li-ion battery, the SEI also has some adverse effects on the battery performance. The most notable is the capacity loss caused by lithium being irreversibly bound in compounds during the formation of the interphase, observed as a reduced Coulombic efficiency for the first few charge/discharge cycles. The magnitude of this reduction is related to the thickness and composition of the SEI, which in turn can be manipulated by tuning the composition of the electrolyte [21].

Electrolyte solvents
The primary role of the electrolyte in any electrochemical cell is to conduct properties [23]. Advanteous of Li polymer batteries are cheaper and simpler construction, more rugged and resistant to short circuiting, while on the other hand generally being regarded as having a smaller charge capacity than conventional Li-ion batteries. In addition, the absence of a free liquid removes the need for a metal housing and increases flexibility when it comes to battery shape and packaging [24]. Liquid electrolytes are made by dissolution of a Li salt (e.g. ) ) in a nonaqueous solvent.

Solutes
Two main requirements have to be satisfied in order for a Li salt to be applicable as electrolyte solute for Li-ion: 1) the salt must have a high solubility in non-aqueous solvents, to meet the primary purpose of the solute is to provide a sufficient concentration of lithium ions and 2) the anion must be stable in relation to the other components of the battery, either by being inherently so or by passivation.
Examples of salts that fulfill these requirements are ' + , ,-) , ) and ( ( . Apart from these requirements, there are numerous other properties that should be considered, e.g. toxicity ( ,-) ), fire or explosion hazard ( ' + ), low conductivity ( ( ( ) and cost, which narrows down the number of practical candidates considerably [26]. For a fact, the vast majority of Li-ion batteries both in production and research use the same salt as primary solute, namely lithium hexafluorophosphate ( ) ). This is not because it is a perfect solute, but because 15 ) has a good compromise of properties. However, for applications where demands are more extreme, ) may not fulfill all the requirements. For instance, at high temperatures, the decomposition of ) into and . becomes a major problem, as . readily hydrolyses to form / , which is devastating for the electrode performance [26]. This has led to research into alternative solutes, like lithium tris(pentafluoroethyl)trifluorophosphate and lithium bis(oxalato)borate, commonly known as , [27] and 0 0 [28], respectively. These have shown great potential, but are not yet commercialized on a large scale.

Cathodes
Although all tested cells in this research are half cells, a short overview over   [29].While this applies for a large number of compounds, other essential criteria reduces the number of candidates notably. The large changes in composition of the electrode during lithiation often lead to changes in the crystal structure of the material that may compromise the structural integrity of the electrode or otherwise have energetically unfavorable effects. Therefore, it is to use a material with a crystal structure that is stable over a wide range of compositions. In addition to this, the Li diffusivity, electronic conductivity, cost and environmental compatibility of the material must be taken into account. Since the first development of the Li-ion battery, three main groups of materials have been found to exhibit these properties; layered oxides, spinels and olivines and tavorites.
Layered chalcogenides were discovered first, starting with the sulfides and selenides like 2 1 and 3 1 , before moving on to the oxides, e.g. 1 and 4 1 . Closely packed anions provide the main structure of these materials, with the transition metal cations occupying the space between every other layer. The Li can then be freely intercalated into the remaining unoccupied ones, resulting in the layered structure that gives this group its name. The structure itself is named 5 − 47 1 and can be seen in Figure 2-5a [8].
Being quite similar to the first group, the second consists of materials having the spinel structure. However, rather than having the alternating layer structure that was described above, in the spinel structure both of the cation species are ordered throughout all of the atomic layers, as seen in Figure 2-5b. This ordered spinel structure provides an interconnected matrix of interstitial positions that allows for the storage and diffusion of Li ions, again while the transition metal atoms accommodate the electrons by changing oxidation state [30].
The last group, the olivine materials, were developed with an extensive research effort into making electrodes based on iron, being the best available transition metal by far. The effords resulted in + [8], whose structure, seen in Figure   2-5c, consists of edge sharing octahedra of oxygen coordinated iron atoms forming layers which are bound together by tetrahedra of oxygen coordinated phosphorous atoms, between which lithium atoms can reside. Made only of materials of abundant availability and having capacities comparable to that of 1 , as well as very good cycle life, low toxicity and being environmental compatibility, it has attracted great interest. Its defining limitation has been an extremely low electronic conductivity, for which making composites with carbon has been the most common solution [31].
Corresponding compounds of other transition metals like manganese, nickel and cobalt have also been investigated. None of these have shown any superior electrochemical performance, however, and the materials are far more costly, hence the interest in these compounds has been limited [8]. Tavorite phase materials like + F have been found to exhibit similar properties as the olivine phases. Its structure, which can be seen in Figure 2-5d, also consists of layers of octahedra, but unlike in the olivine structure, these share corners and are coordinated by both oxygen and fluorine atoms, while the layers themselves are separated by tetrahedra of oxygen and fluorine coordinated sulfur atoms [32]. Li-ion batterie anodes. While the use of carbonaceous materials as intercalation electrode material was proposed as early as in the patent describing the first ever lithium-ion secondary battery by Yoshino et al. (1987) [33], it has yet to be succeeded as the standard anode material. Excellent cycle life, good material availability and low cost are the attributes leading to this favored position. The vast majority of the carbonaceous electrodes are made from graphite or graphitable carbons, for which the intercalation mechanism has been widely studied and is well known [34]. However, lithiation into graphitic electrodes is limited by the composition of the most Li rich phase in the Li-graphite system. With an upper limit of ) , this results in a relatively low theoretical maximum capacity of ~372 mAh/g, above which Li metal is formed.
This has prompted research into alternative anode materials, including both more advanced carbon based electrodes, alloys and composites [35], [36].
Hard carbons prepared by pyrolysis of organic material are widely seen as a promising alternative carbon material and therefore acquired some research intrest.
Only a short range ordering and various surface termination groups can be attained using this process, depending on the precursor and pyrolization temperature. The process of Li intercalation into such materials is not easily determined because of the lack of long range order and varying composition, hence the theoretical capacity limit is not known. However, it has been established that that at least some of these materials exhibit far higher capacities than the limit for graphitic carbon. Yet, while the lithium insertion capacity may be very high, these materials suffer from very large irreversible capacities, severely limiting their practical applicability. There are other carbon polymorphs like kish graphites and multi-walled carbon nanotubes (MWCNTs) which are possible alternative carbon materials [35]. While strictly being a member of graphitable carbons, kish graphite have proven to have capacities reaching ~20% above the theoretical limit of perfect graphite, indicating a certain degree of disorder [37].
MWCNTs, on the other hand, are rather expensive, but have still attracted a fair share of attention due to the many potential mechanisms of lithium intercalation, namely in the graphene sheets constituting the tubes, in the space between the tubes 20 formed due to imperfect packing, and inside the tubes themselves. However, as with the hard carbons, the irreversible capacity is too high for this kind of electrodes to be commercially viable [35].

Alloying anodes
In order to replace carbonaceous anodes with anodes having a higher capacity, experimental work on anodes using chemical elements which form alloys with lithium was started in early 1960s. [38] In 1971, Dey [39] found that Li can be electrochemically alloyed with a number of metals at room temperature, including Si, Sn, Pb, Al, Au, Pt, Zn, Cd, Ag, and Mg. Before, Silicon, the theme of this thesis, will be covered in further detail in the following section, a brief summary of the common features of the alloy anode materials is given here. Because of their availability, price and environmental sustainability, Si, Sn, Sb, Al, and Mg have been most extensively examined for this purpose [36]. As following from the key electrode properties of these alloy anode materials and graphite, shown in Table 2-2, the potential capacities of alloy forming electrodes are far superior to that of the graphite anode. The specific capacities of intercalation anodes come close to that of the Li metal anode. As a matter of fact, the amount of Li some alloy electrodes can accommodate per volume exceed that of metallic Li itself, e.g. 11 . contains 88.56 mol Li/L, while lithium metals only contain 76.36 mol Li/L [40]. This is partly caused by the fact, that Li is stored in ionic rather than atomic form in the alloy electrodes and partly because metallic Li is not in a close packed crystal structure, and therefore have a potential for densification.
In addition, from a safety-perspective, alloy anodes are also superior to carbon anodes. Lithiation of the alloys happen at potentials ranging from 0.1V to 0.9V (Table   21 2-2), compared to the carbon anode where lithiation typically happen around 0.05V.
On the one hand, this reduces the voltage of the cell, and consequently the specific energy, on the other hand it also reduces the probability of accidental deposition of metallic Li on the electrode surface during fast charging [36]. In many industries safety benefits like this are significant, e.g. in the automotive industry [41].
Apart from excellent capacity and safety benefits, there are several other characteristics that are necessary to make a suitable anode material. Especially in cycling stability and Coulombic efficiency, these alloy anodes have been found to be seriously lacking. Due to the high electropositivity of Li, many of the intermetallic phases formed by Li and group 13-16 metals or metalloids are of highly ionic character. These are generally known as Zintl-phases and are usually brittle.
An increase in volume of the material is caused by the reduction of the otherwise neutral host atoms which results in a substantial diameter increase. In contrast, the Li-ions can occupy interstitial positions in the lattice, and hence cause only a limited volume expansion [40]. This is obvious from Table 2-2, where expansions of up to 420% are observed for the alloy materials, compared to 12% for the graphite electrode. During cycling, this volume change introduces large stresses into the electrodes. In combination with the brittleness of many of these materials, these stresses can eventually lead to cracking of the electrode [42].This is associated with partially contact loss of the electrode material or the electrode itself detaching from the current collector. The result of these effects is an irreversible capacity fade by causing electrode material to become essentially dead weight. The pulverization leads to a continuous formation of new surface area, which must be stabilized by forming an SEI, as described in Section 2.1.3. This process involves irreversible loss of Li in anodes made with these materials, finally yielding to the observed low Coulombic efficiency.
All there effects are challenges to overcome in the research on the way to a successful alloy anode. However, as the potential benefits are of such magnitudes, these are challenges for which there has been proposed numerous solutions. Concrete examples of this will be discussed in more detail in Section 2.2.2.  In addition to its outstanding capacity, silicon is the second most abundant element on earth. Because of these attributes, a great proportion of research has focused on using silicon as Li-ion cell anode material. However, at room temperature the alloying process of Li with Si was found to be less reversible [46]- [48].  Table 2-3 shows the data for crystal structure, unit cell volume, and volume per Si atom for each alloy formed during the alloying process.
Since the number of Si atoms is constant, the relative volume increase of the anode can be established by calculating the volume of electrode per Si atom in each of these phases. This can be expressed by With 3 # and 4 PQ being the unit cell volume and the number of silicon atoms in the unit cell, respectively. It shows that the volume per silicon atom for Li 22 Si . alloy is four times higher than that of the parent silicon atom, i.e., a 420% volume expansion of the silicon lattice occurs. This result in cracking and disintegration of the electrode, with active material loss via reduced electronic contact, giving severe capacity fade.
According to the equilibrium phase diagram, when lithium is inserted into Si,

The Breakdown Mechanism
As you can see in Table 2-3, the volume expansion versus the concentration of Li is nearly linear. Various methods [52]- [55], including X-ray diffraction and nuclear magnetic resonance, have been used to study the amorphization process; these studies have primarily shown experimental evidence for the crystalline-to-amorphous phase transition and have also provided information on the local atomic structure of the 27 amorphous phase. In addition, atomic force microscopy has been used to show that volume expansion occurs during lithiation and contraction occurs during delithiation of Si thin films [56]- [60]. Based on the studies, three fundamental materials challenges to using Si as a viable battery electrode, as illustrated in Figure 2     Li-ion battery electrodes with Si nanoparticles could be produced with the current manufacturing process. [76], [77] Finally, reducing the size of silicon particles will help to release the stress and prevent the cracking of silicon during the Li insertion, which will significantly improve the cycling performance of the electrodes. [75], [78] However, Si anodes with nanoparticles suffer from rapid capacity changing due to the breakdown mechanisms described in Section 2.2.1. Strategies to improve the nanoparticles approach have also been investigated. These include the utilization advanced binders like PAA (2500 mAh/g for 100 cycles [65]) or porous Si nanoparticles (1400 mAh/g for 200 cycles [66]).

1-dimensional structures: nanowires and nanotubes
In 2007, Chan et al. [67] reported a Si anode design concept utilizing Si nanowires prepared by the vapor-liquid-solid synthesis method. By growing these thin wires directly on a stainless steel current collector, a significant improvement of the electrochemical performance, compared to thick films and large particles. [67] The better performance of 3500 mAh/g for 20 cycles at C/5, is due to the sufficient empty space between adjacent NWs, hence the lithiation and delithiation process does not cause damage to the anode. However the material will undergo large volume changes but these do not lead to a disconnection from the current collector. Since the NWs form a one dimensional electronic pathway to the current collector, the addition of 33 conductive carbon and polymer binder becomes unnecessary. The basic concept was adopted by many other research groups [51], but soon reached its limitations in terms of cycling stability.

2-dimensional structures: thin film
Besides the composite electrodes, another main group of electrodes; thin-films, have received a fair share of attention. The motivation for using thin-films compared to composites is that they contain no inactive components, and thus have a higher effective specific capacity [38]. Being essentially bulk silicon, the performance deteriorate quickly when film thickness approaches 1µm [12]. However, thin films have demonstrated extremely good performance even at high charge and discharge rates, e.g. 50 nm films cycled at 30C that retain a capacity of over 2000 mAh/g even after 3000 cycles [71]. However, as a 50 nm pure silicon film is equivalent to a material loading of only 13.25 µg/cm 2 . (for a silicon density of 2,65 g/cm 3 ). The active area would have to be immense to obtain a usable capacity and the mass of the associated current collectors would be considerable. On the other hand, far thicker films have been made with acceptable results, and much effort is still being put to the 35 task. Approaches include making binary and ternary alloy thin films, and while not yet commercialized, they show much promise [38].

3-dimensional structures: various designs
A large potential in fabricating Si anodes lies in the formation of threedimensional structures. These provide the possibility of high loadings, and potentially the control of SEI formation while providing enough void spaces or being elastic enough to compensate the volume changes of Si during cycling. Various designs have already been tested, and the possibilities are almost unlimited, but to give a general introduction in this topic, three examples will be further described in this section.

Ternary silicon nanoparticles/conducting polymer/carbon nanotubes hybrid anode
This anode design basically combines the results of studies of carbon nanotubes [69], [70] with more elastic and conductive binders [76], [80]. Si    The free energy of adsorption (the free energy input required for desorption of one particle, either into the aqueous or the oil phase) is related to the interfacial tensions and the size of the solid particles. Since spontaneous entry of particles into liquid-fluid interfaces is slow, [86] so mixing is used to accelerate this process. The energy ∆E required to detach a spherical particle from an oil−water interface into either bulk phase is given by [84] The maximum stability of the emulsion is reached at a contact angle of 90° in most cases [87]. Obviously, corresponding to the influence of the radius in Eqs. (2.3-6) and (2.3-7), large particles experience a larger adsorption free energy due to a larger area contacting oil and water. However, nanoparticles strongly adsorb to oil-water interface as well. Regarding the wetting-based theory, there is no lower limit of particle size (Eqs. (2.3-6) and (2.3-7)). Indeed, the adsorption free energy is always much larger than the thermal energy, even when the solid particles are very small. As example, the adsorption free energy of spherical solid nanoparticles of 10 nm diameter at the hydrocarbon-water interface ( ZXY = 50t4t X9 ) having a contact angle of 90° is Δ n\W [ = 1.6 × 10 X9w x, which is much larger than y2 (4 × 10 X19 x 7 293 z). [84]

Particles that stabilize Pickering Emulsions
The partial wetting condition is satisfied by many types of particles, either inorganic or organic, for most common oils. A range of particles including carbon black, silica, polystyrene, iron oxide, bentonite clay, and graphene oxide [88]- [95] have been employed for emulsion formation. Determining the final characteristics of the emulsion, the particle properties play useful roles [85]. Functional particles bring about supplementary properties; as examples, thermosensitive poly(Nisopropylacrylamide) particles give temperature-sensitivity to the emulsions [89]- [93], and pH-sensitive emulsions have been prepared with the help of pH-responsive particles [85], [94], [95]. In this study we are using surface modified para-amino Only a small interfacial area can be stabilized, when only a low amount of solid particles is used. This should result in very large droplets. A mild emulsification process (vortex mixing) allows the formation of stable emulsions [101], [102].
Depending on the introduced shear to the emulsion, the droplets become smaller and smaller. These droplets undergo coalescence once the mixing process is stopped.
Coalescence may be terminated when the interfacial area reaches the area 45 corresponding to full coverage by the particles [101], [103]. Full emulsification can only be reached, if the emulsion remains stable during this process. Otherwise free oil is released and the dispersion process ends with a partial emulsification of the oil [104]. Theoretically, the higher the amount of solid particles is, the smaller droplets should be able to be prepared. However, an efficient emulsification process is Upon mixing two types of particles that undergo heterocoagulation, Abend and Lagaly built a percolating network of flocculated solid particles that were partly adsorbed to the oil droplets of an o/w Pickering emulsion [96], [97], [99], [100].
Gelation of the aqueous phase prevented the coagulation and most other events leading to de-stabilization quite efficiently. That thickening increases the stability of emulsions is a well-known fact; it holds for Pickering emulsions as well [98].
Emulsion gels can be prepared on the same grounds [107].  After a steady shearing flow has been imposed on a fluid for a suitable period of time, the shear stress often comes to a steady state, ( | ), which depends on the imposed shear rate | . The ratio of the steady shear stress the shear rate | is then the steady-state shear viscosity ( | ). The creep test is a related way of obtaining time-depending rheological information. In it, a constant shear stress is imposed on the material, and the shear rate is measured as a function of time until a steady shear rate is obtained. The creep test is especially useful for measuring the yield stress ‚ , since if the imposed stress is below ‚ the steady-state shear rate will be zero.

Storage and Loss Moduli
Another way to explore rates of structural rearrangement within a complex fluid, one that does not significantly deform the fluid's microstructure, is to impose small-amplitude oscillatory shearing. This kind of deformation can be achieved in a cone-plate geometry by rotating the cone about its axis with an angular velocity that oscillates sinusoidally, Ω( ) = Ω ƒ cos (" ), where " is the frequency of oscillation, in units of radians per second. The shear rate is then also a sinusoidal function of time, | ( ) = Ω tan 5 ⁄ = Ω ƒ cos(" ) / tan 5, and so is the shear strain , which is the time integral of the shear rate, = (Ω ƒ " ⁄ ) sin(" ) / tan 5. The ratio (Ω ƒ " ⁄ ) is the amplitude of the angular deflection of the cone, and ƒ = (Ω ƒ ") ⁄ / tan 5 is the strain amplitude of the displacement of the sliding plate, divided by the gap.
If the strain amplitude ƒ is small enough (typically ƒ ≪ 1) that the fluid structure is not much distributed by the deformation, then the stress measured during The term proportional to ‡ (") is in phase with the strain and is called the storage modulus, while the term containing ‡ ‡ (") is in phase with the rate of strain | and is called the loss modulus. The storage modulus represents storage of elastic energy, while the loss modulus represents the viscous dissipation of that energy. The ratio ′′ ′ ⁄ , which is called the loss tangent tan ‰, is high (≫ 1) for materials that are liquid-like, but is low (≪ 1) for materials that are solid-like. The regime of small amplitude straining, in which the stress can be represented by Equation (2.4-3), is called the linear viscoelastic regime.

Types of Rheological Response
Differences

Rheology of Particulate Gels
In 'Sol-Gel' processes a liquid suspension, or 'sol', of colloidal particles is 'gelled', or flocculated, into a quasi-solid mass by addition of a chemical agent [111].
For an initially stable sol composed of colloidal particles, the gelling agent, which is usually a pH modifier, an electrolyte, or a polymer, produces gelation by reducing repulsive particle-particle interactions, so that attractive van der Waals forces can draw particles into near contact. If the particle concentration is high enough, a samplespanning network of such contacts forms, producing a solid-like gel phase.
A way to produce a sol phase is to use particles whose surfaces in solution are charged, resulting in electrostatic stabilization of the sol. Addition of acid or an acid former then tends to neutralize these groups, like hydroxylic or carboxylic groups, producing gelation. The first three of these difficulties arise because of the nonequilibrium structure of strongly flocculated gels. Particles bound strongly together in their primary van der Waals potential minima are unable to rearrange within laboratory time scales; hence the structures cannot relax to achieve thermodynamic equilibrium. Therefore, the gel structure depends on preparation history, including any deformation experienced by the gel prior to the rheological measurement.

The Rheology of Strongly Flocculated Gels
A strongly flocculated dispersion can be considered when the particle concentration is high enough to produce a rigid gel network. Once particle-particel contacts are formed, they are released so infrequently that particle rearrangements are strongly suppressed. Thus, the time for equilibration of the structure, given by a relaxation time OE, is too long to occur within the experimental time frame, which is usually no more than several hours. The 'Newtonian' zero-shear viscosity is only attained at shear rates | ≤ OE X9 , and these rates are too low to be accessed. Thus, strongly flocculated gels are characterized by a yield stress, rather than a zero-shear viscosity. Other rheological quantities that are important for strongly flocculated gels 56 include compactive strength, linear (and nonlinear) elastic moduli at high and low frequencies, and the shear-rate-dependent viscosity. The dependencies of these on particle size, particle concentration, and particle-particle interaction are important for the processing of colloidal gels.
Rheological measurements are difficult on strongly gelled colloids, and data often do not reproduce well. Multiple runs often must be averaged together to reduce data scatter. Strongly flocculated suspensions are by definition not at equilibrium, and so their properties are sensitive to preparation technique and deformation history. In addition, in large-deformation or continuous shearing, slipping of the gel against the surfaces of the rheometer tools is always a danger. [113], [115], [116] The volume fraction dependence of the high-frequency modulus Ž above the minimum particle concentration is expected to behave like Ž ∝ R• − • • T ' [117]- [119]. However, over most of the range of •, Ž increase with • roughly as [114], [120] Ž ∝ • ' (2.4-4) A similar result is obtained for the yield stress in shear ‚ , except the exponent is smaller, ‚ ∝ Φ " . Such power laws have been derived from theories that model the gel as a network of interconnected fractal clusters [121]- [123].
The modulus of strongly flocculated gels tends to be highly strain-dependent, with linear behavior confined to very low strain amplitudes [121]. As flocculation becomes stronger, the strain sensitivity increases. In addition, strongly flocculated gels are likely to be more brittle than weakly flocculated ones, in that they are harder than weakly flocculated gels, they cannot be deformed as much without fracturing [124].
This finding is if considerable importance for the processing of gel bodies.
Thus, the mechanical properties of a gel depend not only on particle radius 7 and concentration •, but also on the flocculation strength. For gels in aqueous media, the mechanical properties therefore depend on the charge on the particle surfaces, as well as on the type of ions that might be bound to them. By adding acids or bases the pH is adjusted, which change the surface charge of the particles, presumably through surface binding of / • or / X . The yield stress ‚ is maximized at the isoelectric point.

Yield Stress and Elastic Modulus
Developing accurate theories for strongly flocculated gels is challenging, since the structures of such gels are not at thermodynamic equilibrium. It might be able to assume that such gels are in a state of static equilibrium in which the forces acting on each particle are in balance. Since the interaction potential between particles in a strongly flocculated gel has a minimum • "Q' = •(ƒ ) that is deep compared to y -2, gaps between neighboring particle surfaces in such gels will presumably almost be close toƒ , unless the gel is subjected to a mechanical strain. Therefore, the shape of the potential •(-) nearƒ is important in determining the gel's mechanical properties.
Sensitivity to the shape of •(-) differentiates weakly from strongly interacting particles. For weakly flocculated gels, the precise shape of the potential is not important. But insensitivity to the shape of the potential can only be expected when the particles are only weakly bound by that potential, so that rapid, thermally 58 driven changes in particle-particle separation average out the details of the shape of the potential. For strongly flocculated gels, the particle-particle separation remain trapped near the minimum in the potential well, and the shape of the well near this minimum matters much more.
In a static equilibrium one can attempt to estimate the dependence of the yield stress ‚ and the modulus on the shape and depth of the interparticle potential. For a gel subjected to a shear strain that homogeneously displaces particles from their position of static equilibrium. Pairs of particles are pulled apart by this strain, and separation between particle centers of mass should increase roughly by an amount q ƒ , where q ƒ ≡ 27 +ƒ is the separation between the centers of mass in the absence of strain. Hence, the imposition of a strain increases the gap between particle surfaces fromƒ to -≈ƒ + (27 +ƒ ) (2.4-5) If is small, the increased separation of particles is already large relative to the initial gapƒ . Thus the ratio the gap between particles after the strain to that before the strain is Since the ratio 27ƒ ⁄ is usually large (≳ 100), even a strain of only 1% multiplies the gap between neighboring particles by a factor of two or more. From this, it is obvious why strongly flocculated gels, with particle-particle gaps as low as 1nm, are so strain-sensitive [120].

59
A force = −• ‡ (-) with -=ƒ + (27 +ƒ ) is produced by this increased separation between the particles, where •′ is the derivative of • with respect to D. This force would restore the original interparticle spacing if the shearing stress were removed. The macroscopic stress is this force times the number of interparticle bonds that cross a unit area of the sample; this latter factor should scale as 125]. As long as the local applied force increases with increased strain, increases with increasing strain, and the gel maintains its mechanical stability. But once the strain reaches the point that the slope • ‡ of the potential is a maximum, any further strain produces a decreasing force, and the interparticle structure breaks apart.
This corresponds to the point of yield. Thus, the yield strain ‚ is given by the Larson (1990) [126] suggested that the maximum of • ‡ might to be expected when separation -= -‚ is on the order of twiceƒ , the value ofat static equilibrium.
The yield stress is proportional to "nV = • "nV ‡ = • ‡ (-‚ ) times the number ot interparticle bonds that cross a unit area of the sample, • 1 7 1 ⁄ thus ‚~• 1 Larson (1990) [126] suggested that • ‡ can be estimated as -• "Q'ƒ • at the yield point. Considering only van der Waals interactions and electrostatic interactions with a constant surface charge, the equation for • "Q' is the following Hence from these equations we obtain The linear modulus can be estimated by analogously [125]. is defined as the stress divided by the strain , where is small enough that is independent of .
As argued above, the stress is given by roughly (• 1 7 1 ⁄ )• ‡ (-), withgiven by Equation (2.4-5)). If the quantity ⁄ is to be independent of strain, then must be small enough that •′ can be linearized in -.
Thus, the linear modulus is controlled by the curvature of the particle-particle potential • at its minimum. This local curvature is extremely sensitive to the details of the particle-particle interactions at close separations [127]. Larson (1990) [126] estimates The process of drying of a porous material can be divided into several stages. In this section an overview about occurring drying stresses and appearing fractures and how to avoid them will be given. The reasons for drying stress to occur in the studied emulsion are a nonuniform pressure distribution and stress induced by the liquid.

Pressure distribution
The course of drying during the constant rate period is illustrated schematically in Figure 2 x P«¬-n®¯= -° ± ∇ | P«¬-n®¯= 3 | ª (2.  whereis the permeability of the network and ° is the viscosity of the liquid. The lower the permeability, the greater the pressure gradient must be to support a given evaporation rate. The reason that gels are more difficult to dry than ordinary ceramics is that the permeability of gels is very low (because of the small pore size), so modest drying rates produce very steep pressure gradients. The higher tension in the liquid near the outer surface make that portion of the network shrink faster than the body as a whole, so it tends to crack. The steeper ∇ , the greater the difference in shrinkage rate between the exterior and interior, and the more likely the gel is to fracture.
As the gel shrinks, the viscosity of the network increases and this causes an increase in . The syneresis pressure is negligible compared to the capillary stress near the end of the constant rate period, so the tension in the liquid at the surface of the plate is where z´ is the bulk viscosity of the network. The constant rate phase ends when the increasing viscosity causes ( ) to rise to © , the maximum value of tension, when the radius of the meniscus becomes equal to that of the pore, and the liquid-vapor interface recedes into the gel.

Stress
If the pressure in the liquid were uniform, the network would be uniformly compressed and there would be no tendency to crack. However, the low permeability of the gel gives rise to a pressure gradient, so the tension in the liquid is greater near the drying surface, and the contraction of the network is consequently greater. The difference in shrinkage rate between the inside and outside of the body is the cause of drying stress.

Avoiding Fracture
Drying produces a pressure gradient in the liquid phase of a gel, which leads to differential shrinkage of the network. When the exterior of the gel tries to shrink faster than the interior, tensile stresses arise that tend to fracture the network at the exterior.
As shown in Figure 2-22 the material on either side of the crack can contract more freely, so it is favorable for the crack to grow into the drying surface. If the pressure in the liquid were uniform, the whole network would be isotropically compressed and the gel would shrink without risk of cracking. However the higher tension in the liquid at the exterior causes greater contraction of the network in that region. Since that contraction is inhibited by the slower-contracting interior, the network at the exterior is effectively stretched, and this promotes cracking. Thus, it is the differential contraction that produces macroscopic tension in the network and causes cracking.
[111] where z¸# is a material property called the critical stress intensity factor and V represents the applied stress. Crack growth is called 'catastrophic', because the stress intensity increases with the size of the crack, so the bigger the crack gets, the faster it goes until it reaches the speed of sound.
The theory of LEFM applies for brittle elastic materials, whereas gels are viscoelastic, so the theory would be expected to apply only when the strain rate is too fast for significant relaxation to occur. It can be shown [132] that there is an elastic region near the tip of a moving crack whose dimension, ¹, is given by ¹~º ® OE »ª (2.5-4) where º ® is the velocity of the crack and OE »ª is the viscoelastic relaxation time.
There is a zone of plastic deformation near the crack tip with a characteristic dimension of where ‚ is the yield stress. The concept of critical stress intensity is meaningful as long as ¹ ¼ ≪ ¹ [132] Cracking is sometimes attributed to the existence of a pore size distribution in the gel [133], [134]. As indicated in Figure 2-23, when larger pores are emptied by evaporation, the wall between adjoining pores is subjected to uneven stress that can cause cracking. This provides a simple explanation for the observation that cracking often occurs at the critical point, as the pores begin to empty. However this does not explain, why cracking is prevented by slower evaporation. Slower drying makes the drying front more irregular on the scale of the pore size [49], so fracture would seem to be more probable at slower drying rates if uneven draining were the problem.
Another difficulty with the model is that the resulting flaw would be a 'point defect' similar in size to the pore diameter, which is too small to produce a catastrophic failure. However, one could argue that these small cracks could link together as the

CHAPTER 3 PRELIMINARY WORK
This work is based on the preliminary work of Chen et al. (2014) [5], who studied high capacity Si/C anodes prepared using emulsion-templated direct assembly.
This simple, inexpensive processing strategy is designed to overcome many of the limitations (see Section 2.  after drying, still leaving adequate space for Si expansion during lithiation. We note that this method of forming emulsions cannot guarantee that each oil droplet will have the same size or contain the same number of nanoparticles. This is a stochastic process that depends upon the mixing conditions. After drying at 50 °C in an oven and examining this sample using FE-SEM, the droplet morphology is visible ( Figure   3-2C). Elemental mapping using EDS, shown in Figure 3-2D, confirms that the Si nanoparticles are confined to the 'oil' regions, and the CB particles surround these patches of Si nanoparticles. The surfaces of the anodes before and after cycling are shown in Figure 3 Figure 5 shows the morphology of Si NPs after cycling. As shown in Figure   3-4A, the Si NP in the Si/Super P anodes suffered severe pulverization, and lost their distinctive spherical morphology. EDS data showed a significant amount of elemental Si in the SEI layer, comparable to the Si content away from it, suggesting that some small pieces of Si broke away and remained in the SEI when Si NPs shrank during delithiation. For the emulsion-templated Si/CB anodes, the spherical shape of Si NPs was retained, as shown in Figure 3-4B. Compared to the center of Si NP, very little Si was detected around the particles, implying that Si NPs did not suffer severe pulverization.  about 55% and ∼1540 mAh/g, respectively. The low first cycle efficiency is associated with SEI formation around the CB particles, promoted by their high specific surface area, as well as SEI formation around the Si NP.31 The delithiation capacity increased to a maximum 1940 mAh/g after five cycles, indicating that most of the Si NPs have been activated by the fifth cycle. The delithiation capacity is ∼1300 mAh/g after 50 cycles. The Coulombic efficiency varied between 95% and 99% and was at 97.4% at the end of the 50th cycle. We note that at lower cycle rates, Si lithiation can be much more complete at the end of the first charging cycle. The capacity will then show a drop over the following cycles.  The difference in performance between these anodes is statistically significant. [5] 78 CHAPTER 4 MATERIALS

Silicon
Silicon nanoparticles with an average 50nm diameter (purchased by Alfa Aesar) are used as active material in the anodes. Silicon nanoparticles possess high purity, smaller particle size distribution, larger specific surface area, and lower bulk density which makes it to be a very active surfactant. Si nanoparticles are non-toxic, odorless, and with active features. The main function of the Si is to provide capacity to the anode material. As mentioned in Section 2.2 due to its high specific capacity of 4200 mAh/g in this work Si is seen as the most promising material to achive high capacity Li-ion batteries. Nanoparticles are used, because it was already shown in previous studies [64] that already the transition from micro-to nano-size particles leads to superior electrical behavior (see Section 2.2.2).

Carbon Black
Driven by their easy availability, range of surface chemistry, biocompatibility, high specific surface area, their ability to absorb organics, their classification as GRAS (generally regarded as safe) materials, and their fractal nature, the carbon particles used in this study are a commercially available grade of surface modified paraaminobenzoic acid-terminated carbon black (CB) particles suspended in water at pH 7.5. These are able to create particle-stabilized octane-in-water emulsions (see Pickering Emulsions, Section 2.3). The presence of the emulsifier particles in the aqueous phase promotes the formation of oil-in-water emulsions. The CB particles are aggregates of [107], [138] "primary" particles, each of diameter ∼20 nm, fused together in a flame process. The resulting fractal particle is about 100 nm-200 nm in 79 nominal size and has a specific surface area of approximately 200m²/g. In this work, we take advantage of covalently linked surface groups that can be used to tune the CB hydrophilicity, to consistently form oil-in-water emulsions. The pKa of the acid is ∼6.5. Thus, the carboxyl groups are deprotonated at pH 7.5  [85] found that these particles can form multiple layers of CB consisting of closely packed particle aggregates at the interface. Furthermore the formed emulsions were found to be stable, as coalescence is surpressed by an increased interfacial shear viscosity produced by the presence of a connected network of particles at the surface. [85] All these properties are beneficial for the production of the anodes. They function as the conductive backbone by forming a cross-linking network, this network gives a certain mechanical stability and due to the multiple layers of around the oil droplets, the Si gets partially protected. This should lead to a more stable SEI. 80

Reduced Graphene Oxide
Reduction of graphene oxide (GO) is a promising low-cost synthetic approach to bulk graphene, which offers an accessible route to transparent conducting films and flexible electronics. [139] The graphene oxide used in this study has already been reduced and was purchased from Graphene Supermarked. The specific surface area of these sheets is ~833 m 2 /g and the Oxigen/Carbon ratio is 10.5. The average flake thickness is supposed to be 1 monolayer and the average size 3-5µm, however SEM pictures of the sample showed significantly larger sheets and also multiple layers (see   [139] In this study, r-GO is used, because it can eventually further increase the conductivity of the anode material. This could lead to higher capacities and better capacity retention due to the availability of better conductive pathways.  [146], and CMC pKa was 3.5 [147].
In this work a 1:1 mixture of CMC and PVA is used to improve the adhesion between the particles and to the current collector while also providing some elasticity 83 to overcome the volume changes of the silicon. By dropping the pH initially to a value of 2, we provide an acidic environment, where the carboxylic groups are getting protonated and can crosslink with the CB.

Processing
The samples were made by using the approach of Chen et al. (2014) [148]. 0.1 N hydrochloric acid is added to a 1.5% w/w CB suspension to a final concentration of 0.01 N. In order to achieve the goal of reducing the amount of used CB other samples made with a 0.75% w/w CB suspension and less were tested.  Because the reduction of concentration showed significant limitations, the oilwater ratio was also changed. Without any silicon, the ability of a 1.5% w/w CB suspension to emulsify higher amounts of oil was evaluated at pH 2. Emulsions with a ratio oil-water-ratio of 1:1, 2:1, 3:1, 4:1, 5:1, 6:1 and 7:1 by volume were made by vortexmixing for 1 min. The results were evaluated using optical microscopy and the type was identified by trying to dissolve the emulsion in water and oil.
After evaluating the results, emulsions with a higher Si/C ratio were formed by using less CB suspension with the former concentration of 1.5% relative to the oil where the Si concentration was even slightly reduced.
Emulsions with a 4:1, 5:1 and 6:1 oil-water ratio by volume were produced with a 2.1% w/w Si concentration in the oil phase. This corresponds to a final weight of Si in the final anode of 72%, 75%, and 78% of the total mass (Si + CB + binder).
However the procedure remained the same. 0.1 N hydrochloric acid is added to a 1.5% w/w CB suspension to a final concentration of 0.01 N. The 2.1% w/w suspension of Si 86 nanoparticles in octane is mixed with the CB suspension at the different ratios for 5 min. An aqueous solution of the binders CMC and PVA [65], [135] at a mass ratio of 1:1 is then added to the emulsion. The volume of the CMC/PVA solution is adjusted so that the final weight of binder after drying is 10% of the total anode mass (Si + CB + binder). In addition the PAA binder was used in the same way as the CMC/PVA binder to evaluate the expected better adhesion to the current collector. The rheological measurements are all done for CB concentrations of 3%, 6%, , 9% and 12% after the addition of acid to an initial pH of 1, to make the CB particles hydrophobic and form a network.

Light Microscopy
The Si/CB emulsion was observed by bright-field optical microscopy in a Nikon Eclipse E 600 microscope (the sample for optical microscopy was diluted to allow adequate transmittance)

Scanning Electron Microscopy
An uncycled emulsion-templated Si/CB anode was observed with the Zeiss Sigma FE-SEM microscope.

Transmission Electron Microscopy
Delithiated electrodes after 50 cycles were also characterized using a JEOL 2100 transmission electron microscope. For observation using the TEM, a small piece was taken of the electrode and dispersed it in dimethyl carbonate. A drop of the dispersion was then placed on a TEM grid and dried in a vacuum oven. The grid was stored in an argon-filled vial until it was observed. Samples were loaded rapidly into the TEM to minimize ambient exposure.

Energy-dispersive X-ray Spectroscopy
An uncycled emulsion-templated Si/CB anode was observed with the Zeiss Sigma FE-SEM microscope and selected area energy dispersive spectroscopy (EDS) in order to examine the spatial distribution of Si and C. 89

Cell Fabrication and Anode Electrochemical Characterization
CR2032 type coin cells were assembled with specially designed current collector anodes and a lithium metal working electrode (see Figure 5  An Arbin BT2000 is used for the cycling test. All half cells are subjected to galvanostatic (constant current) charge/discharge cycles with a cycling time of 10 h (C/10, current density of 1 mA/cm 2 ) from 0.05 V to 1.5 V versus Li/Li+ at ~25 °C, starting with the fourth cycle. The first tree cycles are used to gently establish the SEI at a cycling rate of C/25. All channels can be run completely independent allowing to test multiple independent tests simultaneously. An accuracy of up to 0.02% for low power and 0.05% for high power applications allows precise measurements of the electrical properties. After a total number of 58 cycles, the testing is stopped and the specific discharge capacity, Coulombic efficiency, and capacity retention is calculated.

Processing
The formation of the emulsions was possible for concentrations of the CB suspension from 1.5% w/w down to 0.1% w/w for a 3:5 oil-water ratio without any problems to report. However a brightening of color could be observed.  The average droplet diameter was evaluated using the image processing software ImageJ analyzing more than 200 drops, and found to be ~40µm [137]. This diameter does not change for higher CB concentrations, leading to the conclusion, that the formation of the emulsion is done in the third regime (see Figure 2-16 in Section 2.3.5). That means that the oil-water interface is completely covered with CB particles and any increase in concentration only leads to more excess particles in the continuous phase. These excess particles were observed under the microscope. In Figure 6-2 agglomerates of CB and r-GO can be seen.  Despite the fact, that lower amounts of CB can emulsify the octane, it was only possible to dry emulsions with an initial CB concentration of at least 0.75% w/w. All other batteries showed cracking on the surface. Different drying temperatures were tried to prevent this phenomena, and as suggested in Section 2.5.2, lower temperature increased the quality of the results significantly. Finally 35°C were found to be optimal for drying, providing a slow enough evaporation speed. Nevertheless, 0.75% w/w of CB was found to be the lower limit. The surface of an intact anode can be descriped as rough but uniform (see Replacing larger amounts of C by r-GO turned out to be very difficult. Still the formation of the emulsions was not causing any difficulties, but the drying performance seemed to be reduced. This lead to a maximal replacement of only 5% w/w of C. As a result, the battery with a 3% w/w replacement of C with r-GO finally consist of 1.35% r-GO by total weight, and a 5% w/w replacement of C leads to 2.25% of r-GO in the total battery mass. Cracking during the drying process prohibited the further increase of the r-GO concentration. The explanation for this might be a weaker network structure due to the lower amount of functional surface groups on the r-GO sheets. As described in Section 4.3, in order to achieve good carrier transport within the carbon plane, functional epoxy and hydroxyl groups on the plane have to be eliminated. The r-GO used in this study has already been reduced. As a consequence, the CB particles cannot interact with the r-GO sheets in the same way they interact with each other. The attractive forces between the r-GO sheets and the CB particles are lower, and as a result the network is weakened, which leads to more cracking.
With the goal to increase the Si/C ratio more than the 2:1 that was achieved by reducing the CB concentration in the suspension, the ability of CB so stabilize emulsions with an inverse oil water ratio was evaluated. It was found that an oil-water ratio of up to 6:1, could be achieved without any excess oil or phase inversion. The Evaluating these emulsions with light microscopy does not show any differences in the droplet size. On the other hand, the excess CB particles could not be identified either, leading to the assumption, that because of the high oil-water ratio, most CB particles are at the oil-water interface. When forming the emulsions with Si, the coloring, described earlier in this section, was darker compared to the samples with the 3:5 o/w ratio. The reason for this might be that a less concentrated Si oil suspension, 2.1% w/w Si compared to 3.6% w/w Si, was used. I assume that this leads to less excess Si in the aqueous phase.
Drying these emulsions caused less cracking than, even with a replacement of CB by r-GO by 20%. The placement of the emulsion on the current collector was done very gently, so the formed structure could be preserved. Form this it follows that the CB network is already present instead of being formed during the drying process during the former process.
A total of 10 different batteries were prepared for cycling. four with an oilwater ratio of 3:5, splitted in two with a Si/C ratio of 1:1 and 3% w/w and 5% w/w r-GO for comparison with the result of Chen et al. (2014) [5] and two with a Si/C ratio of 2:1. For the 2:1 Si/C ratio, one battery was prepared without any r-GO and in the other 5% w/w of CB were replaced by r-GO. Furthermore, six batteries, divided in respectively two with a 4:1, 5:1, and 6:1 Si/C ratio were produced. In one of those 20% w/w CB was replaced by r-GO. Unfortunately, it was impossible to test these batteries because of the low adhesion to the current collector. The material peeled of when the electrolyte was added. To overcome this also the PAA binder was evaluated because of the superior adhesion strength reported by   [76], but when adding the electrolyte the dried emulsion always came off. 99

Rheology
It is assumed that the CB suspension, which forms the aqueous, continuous phase in the emulsion haves like a complex fluid. Rheological measurements of the viscosity, the yield stress, the yield strain and the modulus are performed to further identify the behavior of this fluid.
Performing a flow ramp with CB suspensions of 3% w/w, 6% w/w and 12% w/w, the rheological response showed shear-thinning behavior. A zero-shear viscosity could not be attained within the measurable boundaries of the instrument. this means that the relaxation time OE of the CB network is too long because measuring the zeroshear viscosity is possible for | ≤ OE X9 . Since the lowest shear rate was 10 -3 s -1 the relaxation time must be higher than 1000 s. As you can see in Figure 2-19 in Section 2.4.1.4, the behavior of the suspension is typical solid-like, since ∝ | X9 . However, the response of the 3% w/w CB suspension shows an slight increase of shear stress with an increasing shear rate. On the other hand this increase is not constant and seems to fluctuate. One assumption would be that the 3% w/w CB suspenstion is just below the gel point, so neither solid-like nor liquid-like behavior is fully developed. For a solid-like complex fluid, the steady-state shear stress is constant, which is the case for the 6% w/w and 12% w/w suspension for shear rates below 2 s -1 . From the results , shown in Figure 6-10, it can be concluded that the particle concentration that forms a sample-spanning network, and produces a solid-like gel phase is between 3% w/w and 6% w/w. Since a zero-shear viscosity cannot be measured, the CB network can be characterized by the yield stress. Performing creep tests, Figure 6-11 was developed. A yield stress of 47 Pa was measured for the 6% w/w CB suspension, and the twofold, 95 Pa, for the 9% w/w suspension. 101 Figure 6-11. Yield stress of 6% w/w and 9% w/w CB suspensions Performing amplitude sweeps, where the strain is successively increased, and the moduli are measured, the linear viscoelastic region was identified. The modulus of the CB gel is highly strain-dependent, with a linear behavior confined to very low strain amplitudes, as seen in Figure 6-12. All amplitude sweeps are performed with a frequency of 1 Hz. The critical strain was identified to be 0.07%. This is typical for a electrostatically stabilized system [151]. Because a linear response of the system is necessary for the frequency sweeps, the strain for these is set to 0.04% for the following tests. 16% v/v CB for the three tested concentrations. Because the gelling point is between 3% w/w and 6% w/w, it is assumed to be in between at 4.5% w/w CB. This corresponds to a volume fraction of Φ g =2.5% v/v CB. Using the values of ′ of the first two curves, the power exponent n is found to be 2.2. With a proportionality factor of 2650, the modulus of the 9% w/w CB suspension is predicted to be 61000 Pa which is very close to the real value of 65268 Pa. . Frequency sweep at a amplitude of 0.04 % for a 6% w/w, 9% w/w and 12% w/w CB suspension Because of their characteristics, strongly flocculated gels are likely to be more brittle than weakly flocculated ones, in that they are harder than weakly flocculated gels, they cannot be deformed as much without fracturing. [124] This could be the reason for the cracking which was observed during the drying process. 104

Electrical Characterization
The measured performance of the emulsion templated Si/C composite anodes with a 1:1 Si/C ratio are shown in Figure 6-14. Starting with a discharge capacity of 1827 mAh/g for the anode with 3% w/w of C replaced by r-GO and 1887 mAh/g for the one with 5% w/w. The first three cycles were used to build up the SEI very gently at a cycling rate of C/25. After the third cycle the cycling rate is change to C/10 and the capacity is much lower. It is still higher than the theoretical capacity, which might explain the rapid capacity fading till the 10 th cycle. There the capacity reaches a value of about ~ 1200 mAh/g and remains very stable. During the whole cycling period, this battery shows a high Coulombic efficiency above 97% with a final value of 99.5%.
The only exception are the first three cycles where the SEI is build up, the Coulombic efficiency increases from 92% to 98%. The high values show evidence, that a stable SEI is formed. This is because the CB particles partially protect the Si. The former multiple layers of CB around the oil droplets are preserved over the drying process and do now cover the Si particles. After 58 cycles, the capacity is still at ~1000 mAh/g which corresponds to a capacity retention of 75% after the fourth cycle. What is remarkable about the performance is, that after the 10 th cycle, the capacity fading is only 85 mAh to the 58 th cycle. That means for these 48 cycles over 90% of the capacity is conserved. The 5% w/w r-GO anode performed slightly worse. Part of the reason might have been, that this is an average value of two cycled batteries. One of those performed worse, the other was almost identical to the 3% r-GO ones. However the characteristic is the same. The Coulombic efficiency is high, above 97% after the third cycle, and 98.5% after the 58 th cycle. In addition the capacity retention is almost 105 identical. After 58 cycles 74.6% of the capacity is maintained from the fourth cycle.
Comparing the final capacity to those of the 10 th cycle the retention is 84.8%. The usage of r-GO improved the performance of the batteries significantly compared to the precious work of Chen et al. (2014) [5]. They reported a delithiation capacity of ~1300 mAh/g after 50 Cycles. In this approach after 50 cycles the capacity is ~1000 mAh/g which is a decrease by more than 20%. Increasing the Si/C ration to 2:1 lead to the cycling behavior shown in Figure   6-16. One battery with r-GO and one without was cycled and compared. The gentle cycling of these anodes exhibit a really high delithiation capacity of 2252 mAh/g for the battery with r-GO, which is slightly lower than the theoretical of 2630 mAh/g.
Without the r-GO the capacity was worse, only 1600 mAh/g were obtained. The Coulombic efficiency for both is almost identical, after the initial thee cycles the values only vary between 99% and 100%. As a consequence the SEI seems to be very stable, even with the reduced amount of CB. Changing the cycling rate to C/10 resulted in 1505 mAh/g for the battery with r-GO and 1150 mAh/g for the one without. After 58 cycles the capacity of the battery without r-GO is as low as 737 mAh/g, the other maintained 915 mAh/g. The capacity retention after the fourth cycle is consequently 64.1% and 60.8%.  the same in both anodes (5% of carbon was replaced). It is obvious, that the anode with the higher Si loading initially exhibit a much higher capacity. However, the 108 surperior cycling stability of the anode with a 1:1 Si/C ratio leads to a crossing point.
After about 35 cycles, the capacities become equal, and after this point the capacity of the 1:1 Si/C anode exceeds those of the 2:1 Si/C anode. The capacity retention of the 1:1 Si/C battery with r-GO is much better than those of the 2:1 Si/C battery.
Nonetheless, both batteries performed worse than the ones from the previous work of

CHAPTER 7 CONCLUSION
For the simple emulsion-templated directed assembly technique for forming silicon carbon composite anodes for lithium ion batteries, r-GO was introduced up to 2.25% of the total anode mass, and the Si/C ratio was increased to 2:1. The introduction of r-GO induced a better cycling stability for the 1:1 Si/C ratio. For the 2:1 Si/C ratio, also the capacity was increased significantly. However, for all the batteries the capacity was even worse than the ones with the 1:1 Si/C ratio or the that the CB nanoparticles in aqueous suspension behave as a strongly flocculated gel, which cannot deform much without fracturing. Which lead to processing problems during the drying process.
However, the network maintained electronic contact with the current collector throughout the lithiation and delithiation cycles. A stable solid electrolyte interphase appears to form around the high surface area CB particles, which is indicated by a high Coulombic efficiency. The use of Si nanoparticles in a configuration that partially protects it from the electrolyte and the sustained electronic contact of the anode 111 material with the current collector allows all anode to have a high capacity and good cycling performance after 58 cycles.