Performance Analysis and Morphology of Lithium Ion Battery Anodes Prepared by Colloidal Processing

Lithium ion batteries (LIBs) with high power and energy density are desirable for use in portable electronics and electric vehicles. Silicon (Si) and tin (Sn) are among the most promising candidates for LIB anodes owing to their high theoretical specific capacity. However, both Si and Sn suffer from a dramatic volume change during lithiation/delithiation. Extensive efforts have been made using nanostructures to overcome this issue and improve the electrochemical performance of Si and Sn anodes. Carbon black (CB) is usually added to either electrode in LIBs to provide electrical conductivity. Because CB does not contribute to capacity, minimizing its use can lower the mass of battery to further increase the energy density. Due to the high aspect ratio and excellent electrical conductivity, sheet-like material reduced graphene oxide (RGO) is able to form a conducting network at much lower volume fractions than CB based on the percolation theory. In this work, Si and Sn based anodes were prepared respectively using an emulsiontemplating strategy where active material nanoparticles were confined in the oil phase of the formed emulsions. CB stabilized these emulsions and formed a conductive network. We reduced the total carbon loading of Si anode by replacing a small amount of CB with RGO. In Si anode with a lower total carbon content, the formation of a conducting network consisting of CB and RGO contributed to a good cycle performance, which is comparable to the anode with double carbon loading but no RGO. We also studied the influence of the oil on the structure and electrochemical behavior of emulsion-templated Sn based anode. Density and vapor pressure of the oils affected the creaming and drying rates of the emulsions, which in turn affected the structures of dried emulsions and anode performance. Emulsion droplet size also had an impact on the drying process. Sn anode prepared with hexadecane that had a smaller density than water and a low vapor pressure along with smaller oil droplet showed the highest capacity and capacity retention which were attributed to the smooth and dense morphology with no cracking. Aqueous suspensions of CB and mixtures of CB and RGO were also examined. It was found that the concentrations of CB and pH conditions played major roles in determining the rheological properties and microstructures of the suspensions. The combinations of optical microscopy, cryogenic and conventional scanning electron microscopy, transmission electron microscopy were used to characterize the structures of fresh emulsions, dried emulsions (anodes) and suspensions throughout this dissertation.

also necessary in the battery which prevents the short circuit between two electrodes and also allows Li ions to transport during charging/discharging.
The schematic description of a lithium ion cell is shown in Figure 1.1. 2 The ongoing electrochemical process is the deintercalation of Li ions from the graphitic structure of anode and the simultaneous intercalation into the layered structure of cathode. This is a discharge process of the cell; the reversed reaction occurs in the charge process. Si or Sn is less prone to fracture or pulverize. 3,9 These configurations mainly include nanotubes 8 , nanowires 10 , core-shell structures [11][12] , thin films [13][14] , porous structures [15][16] as well as composites with graphene [17][18] . Their electrochemical results show high capacity and good cycle stability.
The techniques reported for preparing Si or Sn anodes usually include complex processing with a long period of fabrication and high cost, such as aerosol spray pyrolysis, 19 forcespinning, 20 freeze-drying and thermal reduction 21 . Emulsification is a process where one liquid is dispersed into another with the assistance of surface-active agents such as amphiphilic surfactants or partially wettable particles. The two liquids to form the emulsions are normally immiscible. Emulsification is a simple colloidal concept and very easy to process. With careful design and implementation, emulsiontemplating method can be used for anode preparation which requires drying of the formed emulsions. The properties of two involved liquids in the emulsions such as density and vapor pressure are correlated with the drying procedure. [22][23] Therefore, studying the influence of these properties can help control and improve the cycling performance for the emulsion-templated anodes.
In most LIBs, carbon black (CB) is used in either electrode to improve electrical conductivity between the active particles. With increasing amount of conductive carbon, the electrical electrode resistivity decreases until the critical volume fraction is reached where the electrical resistivity of the electrode sharply drops to a lower level. 24 This critical volume fraction is termed percolation threshold. Since conductive carbon does not contribute to capacity, therefore, to further improve the specific mass capacity of the battery, the amount of CB used in anodes should be reduced without sacrificing the electrical conductivity. Besides the good electrical conductivity, CB shows other advantages such as the properties of reinforcement and coloring, leading to a wide range of commercial applications in tire, ink and plastics. [25][26][27] Reduced graphene oxide (RGO) is also an important member in the carbon group besides CB. It is usually obtained through partial chemical or thermal reduction of graphene oxide platelets (GO) by removing the oxygen-containing groups with the recovery of a conjugated structure. Hence the reduced GO (RGO) sheets are usually considered as one class of chemically derived graphene. The graphene structure equips RGO with excellent electrical conductivity, mechanical and optical properties, which give rise to a number of applications in electronics and energy storage. [28][29] Generally, as a sheet-like material, RGO has a large surface area and a high aspect ratio that is in the order of 10 3 -10 4 . This gives a much lower percolation threshold of RGO compared to that of spherical particles. 30 This feature can be exploited towards reducing carbon content for electrically conductive polymer composites and LIB electrodes where only a small amount of RGO is required to provide good electrical conductivity.
Given the outstanding properties of CB and RGO, investigating their suspensions in terms of rheological behavior and microstructure is important, which can provide insights on how they change with varying conditions such as concentrations and pH.
This will promote a better understanding and handling in the manufacturing process.
In this study, we utilized emulsion-templating strategy for preparing Si and Sn based anodes with carboxyl-terminated CB as the stabilizer and the conductive agent. In Si based anodes, we replaced a small amount of CB with RGO and reduced the total carbon content of the anode without sacrificing the cycle performance and electrical conductivity. In Sn based anodes, the influence of the oil on the electrochemical performance was investigated and correlated with anode morphology. We also examined the aqueous suspensions of CB as well as the aqueous mixtures of CB and RGO to study their rheology and microstructure at different concentrations and pH.
The combinations of optical microscopy, cryogenic and conventional scanning electron microscopy, transmission electron microscopy were used to characterize the structures of fresh emulsions, dried emulsions (anodes) and suspensions.

3.
Chen, J., Recent progress in advanced materials for lithium ion batteries.

Abstract
Conducting carbon is added to electrodes of lithium ion batteries (LIBs) to provide electrical conductivity. Because this carbon does not contribute to capacity, there is a significant drive towards decreasing its content with a goal of lowering the mass of the electrode. Reduced graphene oxide (RGO) has a high electrical conductivity, and is a potential alternative to traditionally used conductive carbon black (CB) in anodes for LIBs. Importantly, because of its high aspect ratio, RGO is expected to form a conducting network at much lower volume loadings than CB. We report the use of this concept to significantly reduce carbon loading in silicon-carbon (Si-C) anodes for LIBs that are formed by emulsion-templating. An anode with 1 wt% RGO and 14 wt% CB (15 wt% total carbon) showed specific capacity and capacity retention that was comparable to an anode with 30 wt% CB with or without RGO. The capacity retention was significantly lower for the anode with 15 wt% total carbon that had no RGO.
Cryo-SEM imaging of the emulsions, SEM imaging of the dried emulsion, and electrochemical impedance spectroscopy confirmed the formation of a conducting carbon network at 15 wt% total carbon loading when 1 wt% of the CB was replaced with RGO, and the lack of a well-connected network without the RGO. While this concept has been demonstrated for anodes for LIBs, the framework is relevant for other applications where electrical conductivity at minimum carbon content is desired.

Introduction
Lithium ion batteries (LIBs) with high energy density are desirable for use in advanced electronic devices and electric vehicles. 1-6 Because high electron conductivity is required for the anode and cathode, conductive carbon black (CB) is usually added to either electrode. The conductive carbon additive does not contribute to electrode capacity, and one important goal in LIBs electrode design is to minimize its use, thus increasing the specific mass capacity of the battery. [7][8][9][10] The volume fraction loading at the percolation threshold indicates the minimum amount of conducting filler needed to impart conductivity to an insulating matrix. [11][12] For monodispersed solid spheres, the percolation threshold ø sphere = 0.29. 13 When fractal particles, such as CB are deployed, their effective volume fraction in a suspension can be estimated using the following equation, [14][15] where ø eff and ø o are the effective and actual volume fractions of the CB particles in the suspension, D is the hydrodynamic diameter of the fractal CB particle (~120 nm), D 0 is diameter of the primary particle in the CB (~20 nm) and f is the mass fractal dimension of the particle. D/D 0 = 6, and the fractal dimension f for the CB particles is 2.2. 16 Eq.
(1) shows that the effective volume fraction occupied by these CB particles would be ~4 times the volume fraction occupied by the same mass of spherical particles. This lower material requirement is one of the key reasons behind the use of fractal CB rather than dense spherical carbon particles to enhance electrode conductivity. For sheet-like materials, the percolation threshold ø sheet ~ ø sphere /AR,. 13,17 The aspect ratio AR = L/t, where L is the lateral dimension, and t is the sheet thickness.
AR for reduced graphene oxide (RGO) is typically of the order of 10 3 -10 4 . Given the much lower percolation threshold for sheet-like materials of high aspect ratio and the high electrical conductivity of RGO, we hypothesized that replacing some of the CB with RGO would reduce the total carbon used in an electrode without sacrificing electrical conductivity. We examined the hypothesis in this work.
In previous work, we used emulsion-templating for forming silicon-based anodes that showed good electrochemical performance. 18 CB particles were used to stabilize emulsions as well as form a conducting carbon network. The organization of silicon nanoparticles (SiNPs) into porous carbon cages formed by the CB allowed expansion and contraction of the SiNPs during lithiation and delithiation without pulverization and without disrupting the conducting carbon network. In this paper, we show that when a small fraction of the conducting carbon comes from RGO, the total carbon content of an anode prepared by emulsion-templating can be lowered significantly without sacrificing electrochemical performance.

Preparation of anodes
SiNPs (average diameter ~50 nm) were purchased from Alfa Aesar. A para-amino benzoic acid-terminated CB (specific surface area of CB ~200 m 2 /g) suspension in water at pH 7.5 was obtained from Cabot Corporation. While the surface treatment for this CB introduces electrical resistance, the surface functionality is important for stabilizing emulsions. At neutral pH, the carboxyl groups on the CB particles are deprotonated. These CB particles are highly hydrophilic and form a stable suspension in water. Reduced graphene oxide particles (RGO) were obtained from Graphene Supermarket. The RGO particles have a specific surface area of ~833 m 2 /g and a carbon/oxygen ratio of 10.5. The average lateral dimension of the RGO particles is 4 µm.
The SiNP/CB/RGO anodes were prepared by an emulsion-templating method shown in Figure 2.1. 1M HCl was added to a mixed CB and RGO suspension, or only a CBcontaining suspension, until the pH reached 3.3. The addition of the acid protonated some of the surface carboxylate groups on the CB, and made the particles partially hydrophobic. These CB particles favor oil-water interfaces, and stabilize oil-in-water emulsions. 19 In addition, the partially hydrophobic CB particles formed a network in the aqueous phase. 19 When RGO was also present in the suspension, the network contained both entities. The emulsion was transferred on to a CR2032 stainless steel coin cell and dried overnight at 50 ºC under a vacuum, thus removing all the octane and water. The areal loading on the coin cell was ~0.54 mg/cm 2 and the final dried thickness was ~100-150 µm. We examined samples at 30 wt% and 15 wt% total carbon, measured for the dry state. Two classes of samples were prepared at each total carbon loading, one with 1 wt% RGO and the other without RGO. Because the binder concentration was fixed at 10 wt%, reducing the total carbon content meant that the SiNP wt% was increased.

Figure 2.1.
Method for preparing the emulsion-templated SiNP/CB/RGO anode. A carboxyl-terminated CB/RGO suspension was formed in water at neutral pH. 1M HCl was added to this suspension, until the pH dropped to 3.3. SiNPs suspended in octane were then added to the CB/RGO suspension and the mixture was vortexed to form octane-in-water emulsions. After a binder solution was added, the emulsion was placed on a coin cell and dried. The SiNP/CB anode (no RGO) was prepared in the same way.

Electrochemical characterization
CR2032 coin cells were assembled in an argon-filled glovebox that had a water vapor concentration of less than 1 ppm. The SiNP-based anodes served as the working electrodes. Lithium foil was employed as the counter electrode.

Cryogenic scanning electron microscopy (Cryo-SEM)
The microstructures present in the SiNP/CB/RGO and SiNP/CB emulsions were imaged by cryo-SEM using a Gatan Alto 2500 cryo preparation system attached to a Zeiss Sigma VP field emission scanning electron microscope. A small drop containing the emulsion was placed on the sample holder. The sample was frozen rapidly by plunging into liquid nitrogen, then transferred into the preparation chamber maintained at -130 ºC. A flat-edge cold knife in the chamber was utilized to fracture the frozen sample. The chamber was then warmed to -100 ºC for 2 minutes to enhance the surface topology of the sample by differential sublimation. After cooling the chamber back to -130 ºC, the sample was sputter coated with platinum for 60 s, then transferred onto a liquid nitrogen cooled cold stage for imaging at 2 kV.

Scanning electron microscopy (SEM)
Secondary electron imaging on the Zeiss Sigma VP field emission scanning electron microscope was used at 10kV to visualize the dried emulsion samples.

Results and discussion
We confirmed that the emulsions we prepared were all of the oil-in-water type by adding drops of water or octane to the emulsion. 22 The water drop blended in readily, but the octane drop did not. between the performance of anodes with and without RGO. However, for 15 wt% total carbon, the performance of the anode with RGO was significantly better than the one without -the delithiation capacities at 50 cycles were 1370 mAh/g for the SiNP/CB/RGO and 1044 mAh/g for the SiNP/CB anode.
We allowed for initial SEI formation by using charge/discharge rates of 0.1C for the wt% carbon with and without RGO had capacity retentions of 83% and 80%, respectively. At this total carbon loading, the replacement of some of the CB with RGO did not provide any advantage. At 15 wt% total carbon with 1 wt% RGO, the capacity retention after 50 cycles was 77%, comparable to the retention for the 30 wt% carbon loading. However, with no RGO, the 15 wt% carbon anode showed a capacity retention of only 61% after 50 cycles. Thus, the total carbon loading can be reduced by a factor of 2, from 30 wt% to 15 wt%, without paying a capacity stability penalty, if only a small amount of RGO is used in combination with CB.  the first 3 cycles is associated with SEI formation. 23 In the following cycles, the efficiencies varied between 94.5% and 98.9%, indicating that changing the carbon content in our anodes did not lead to modification of lithium ion transport in the cell during charge/discharge cycles.

Conclusions
We demonstrate an effective and simple way to reduce the total carbon content from 30 wt% to 15 wt% in SiNP-based anodes prepared by emulsion-templating. We introduced 1 wt% RGO to the electron-conducting phase along with conductive CB during anode fabrication. For 30 wt% total carbon loading, the replacement of some of the CB with RGO did not impact the electrochemical performance of the anode. The anode with 15 wt% total carbon but with 1 wt% RGO showed electrochemical performance that was comparable to an anode with twice that total carbon loading. Its capacity retention after 50 cycles was much higher than the anode without RGO. The performance difference for the anodes with 15 wt% total carbon is a consequence of the formation of a carbon network in the samples with RGO contributing to good electronic conductivity, and the lack of a well-connected network in the samples without RGO, leading to poor conductivity as well as a large capacity decay.

Acknowledgements
We gratefully acknowledge funding from Department of Energy, Office of Basic

CHAPTER 3 The Influence of the Oil on the Structure and Electrochemical Performance of Emulsion-Templated Tin/Carbon Anodes for Lithium Ion Batteries
In

Abstract
Tin (Sn) is a useful anode material for lithium ion batteries (LIBs) owing to its high theoretical capacity. Emulsion-templating is a simple and scalable technique for preparing anodes for LIBs by forming oil-in-water emulsions with confined active material nanoparticles inside the oil. Carbon black (CB) stabilizes the emulsions as well as forms a conductive network. We fabricated emulsion-templated Sn/CB anodes with octane, hexadecane, 1-chlorohexadecane and 1-bromohexadecane as the oil phase. The vapor pressure and density of the oil affect the drying rates and creaming in the emulsion. The oil droplet size distribution of formed emulsion also has impact on the drying process. All these elements affect the morphology of the dried emulsion, and the electrochemical performance of the anode. Sn/CB anodes prepared with hexadecane (Sn/CB(Hx)) showed a smooth morphology with no cracks, and had the highest capacities and capacity retention. The vapor pressure of hexadecane is low, thus lowering the drying rate and reducing differential stresses on the sample, leading to reduced cracking. Because the density of hexadecane is lower than that of water, creaming forces the emulsion droplets into a close packed arrangement on the surface of the continuous water phase. Smaller oil droplets at the surface also allowed for rapid water evaporation from underneath without damaging the droplets. Upon drying on the current collector, the areal capacity for anodes with creaming are higher than those where the droplets are uniformly dispersed in the emulsions. Therefore, choosing an appropriate oil is helpful in obtaining the best cell performance for emulsion-templated anodes for use in LIBs.

Introduction
Lithium ion batteries (LIBs) with high energy density and long cycle life are important for advanced electronic devices and electric vehicles. 1-6 Among a variety of anode materials for LIBs, tin (Sn) has been regarded as one of the most promising candidates due to its high theoretical specific capacity of 994 mAh/g, and the highest volumetric energy density of 7262 mAh/cm 3 . 7-8 However, a major drawback is that Sn undergoes a dramatic volume change of around 300% during lithiation/delithiation, resulting in fracture and pulverization, contact loss with conductive carbon agents and the current collector, as well as the continuous formation of a solid electrolyte interphase (SEI) that consumes lithium ions. All these mechanisms contribute to capacity fading and low Coulombic efficiencies. [9][10][11] Sn nanoparticles, 12-13 porous Sn nanostructures, 14 based anodes using various oils. We investigate the effect of the oil on the morphology, and the electrochemical performance for the emulsion-templated Snbased anodes.

Preparation of anodes
SnNPs (average diameter ~60-80 nm) were purchased from US Research Nanomaterials Incorporation. The para-amino benzoic acid-terminated CB suspension in water at pH 7.5 was provided by Cabot Corporation. The CB particles are fractal and have a specific surface area of ~200 m 2 /g. At neutral pH, the carboxyl groups on CB are deprotonated. These CB particles are highly hydrophilic and are stably suspended in water. The oils used in this work were octane (99%), hexadecane (99%), 1-chlorohexadecane (95%) and 1-bromohexadecane (97%), purchased from Sigma Aldrich. Their densities at 25°C and vapor pressures at 50°C of all the oils are shown in Table I. All oils were used as received. Sn has a density of 7.3 gm/cc, making it challenging to disperse in the oil. Our calculations reveal that Brownian motion will dominate over sedimentation so long as the particle diameter is less than 360 nm.
These particles may however, aggregate during drying. Deionized water was obtained from a Millipore Milli Q system. The Sn/CB anode was prepared by the emulsion-templated method shown in  Galvanostatic (constant current) charge/discharge cycling tests were conducted on a BST8-WA battery cycler (5 V/1 mA) at room temperature, with a cycle rate of 0.05C (current density of ~ 30 µA/cm 2 ) for the first 4 cycles followed by 0.2C (current density of ~120 µA/cm 2 ) for the remaining cycles, for a total of 50 cycles, over the potential range of 0.05-3 V vs. Li + /Li.

Morphological characterization
Fresh Sn/CB emulsions prepared with various oils were observed by bright-field optical microscopy in a Nikon Eclipse E 600 microscope. Image-J was used to process the emulsion images for ~200 droplets to obtain average droplet sizes and droplet size distributions. The emulsions were also imaged using cryogenic scanning electron microscopy (cryo-SEM) using a Gatan Alto 2500 cryo preparation system attached to a Zeiss Sigma VP field emission scanning electron microscope, operating at 2 kV. The

Results and discussion
A drop of water or oil was added to the emulsion. The water drop spread out immediately, but the oil drop did not, confirming that we always had oil-in-water emulsions.      Sample surface is smooth with few small flaws. (C) Sn/CB(ClHx). Large voids are visible in the sample, but the rest of the sample is smooth. In A, B and C, the dried droplets regions always remained at the surface due to the lower densities of these oils in oil-in-water emulsions. (D) Sn/CB(BrHx). Sample surface has large voids. Red arrows in C and D indicate the occurrence of large voids in the samples. (E) A demonstration of oil-in-water emulsions shows creaming in the samples labeled as 1, 2, 3, corresponding to octane, hexadecane, and 1-chlorohexadecane as the oil phases, and non-creaming for the sample labeled as 4, where the oil is 1-bromohexadecane.
The vapor pressure and density of the oils, as well as the droplet size distributions of the prepared emulsions, all play important roles in determining the morphology of the dried emulsions. At 50°C, octane has a vapor pressure that is slightly lower than water, hence the evaporation rates of octane and water are similar. The consequence after drying was a porous structure of the dried emulsion due to the large induced stress resulting from the rapid evaporation of oil and water. For hexadecane, 1chlorohexadecane and 1-bromohexadecane, their vapor pressures are significantly smaller than water in a few orders of magnitude. Therefore, in the drying process, all of the three oils underwent much slower evaporation comparably. Creaming of the emulsions prepared with hexadecane and 1-chlorohexadecane caused the oil droplets to rise to the water surface. As mentioned above, Sn/CB(Hx) emulsions had smaller droplets than the others. In the drying step, these small oil droplets remained at the water surface with slow evaporation. The packing of small droplets led to a large number of short connected 'channels' in between, which allowed the water underneath to evaporate rapidly without disrupting the oil droplets. The volume of the aqueous phase shrank after drying. This resulted in a smooth and dense morphology of the dried emulsion. For 1-chlorohexadecane, the oil droplets with larger size packed at the water surface and evaporated slowly. However, the space between the packed droplets was less compared to that of hexadecane, which hindered the evaporation of water.
Meanwhile, some droplets formed aggregation. Upon drying, all these factors contributed to a rough structure with significant voids, in the previous aggregated droplet regions as well as some cracks. For 1-bromohexadecane, the oil droplets are suspended throughout the emulsion. With water evaporation around the droplets, the sample underwent some shrinkage causing partial droplet flocculation. These flocculated oil drops then became big voids after drying, and showed an unevenly porous structure in the dried emulsion. We believe these distinct morphologies of the dried Sn/CB emulsions influence their cycle performance.
The electrochemical performance of the Sn/CB containing anodes prepared with the four oils are shown in Figure 3.5. They represent an average from 3 different cells, and the error bars indicate the maximum spread in the data. Figure 3.5A shows the specific delithiation capacities of the Sn/CB anodes. All capacities were calculated based on the total mass of Sn, CB and binder. Sn/CB(Hx) anodes showed the highest capacities, with an initial capacity of 484 mAh/g and a capacity of 327 mAh/g after cycling.
Sn/CB(Oct) anodes had a similar initial capacity of 469 mAh/g to that of Sn/CB(Hx), however, as cycling continued, the capacity dropped rapidly, ending up with only 169 mAh/g after 50 cycles. The initial capacities of Sn/CB(ClHx) and Sn/CB(BrHx) anodes were 297 mAh/g and 218 mAh/g, respectively, indicating that the number of SnNPs involved in the first delithiation process was significantly reduced compared to those of Sn/CB(Hx) and Sn/CB(ClHx) anodes. The Sn/CB(BrHx) anodes showed the lowest capacities at all times during cycling. The low capacities of Sn/CB(ClHx) and Sn/CB(BrHx) anodes had a strong dependence on their rough structures. The possible reason could be that the pre-existing significant voids and cracks in these anodes impaired the electrical contact between some of the active materials and the conductive agent as well as the current collector. Therefore, a significant number of SnNPs were not available for Li ion insertion/extraction even at the beginning of the cycling.
We allowed for initial SEI formation by using charge/discharge rates of 0.05C for the However, a small reduction peak at 1.0 V was observed in the 1 st cycle but disappeared afterwards, we proposed that it was representative of the SEI formation. A very broad anodic peak occurred at 1.0 V in the 1 st and 2 nd cycle, which can be attributed to the Li extraction from the carbon. 14,17 Although Sn/CB(ClHx) and Sn/CB(BrHx) anodes both showed very low capacities, we did not see any side reactions. The difference in peak intensities on the dQ/dV curves for all Sn/CB anodes, was consistent with their cycle performance. All of these results were correlated with the unique structures of Sn/CB anodes prepared using various oils.

Conclusions
We fabricated Sn-based anodes for LIBs utilizing a simple emulsion-templating strategy with different oils. We investigated the effect of oil on the morphology and the electrochemical performance for the Sn/CB anodes. As hexadecane has a smaller density and a significantly lower vapor pressure than water, the evaporation rate of hexadecane is much reduced during drying process. Smaller oil droplets of Sn/CB(Hx) emulsions remaining at the water surface allowed water underneath to evaporate rapidly without disrupting the oil droplets. All of these elements contributed to a dense These results indicate that the choice of the oil is very important in the emulsiontemplating technique for forming anodes for LIBs.

Acknowledgements
We gratefully acknowledge funding from Department of Energy, Office of Basic Energy Sciences, EPSCoR Implementation award DE-SC0007074.

Introduction
Carbon black (CB) and reduced graphene oxide (RGO) both belong to the carbon family with shared properties such as high surface area and excellent electrical conductivity, resulting in wide-spread laboratory research study and industrial applications. As a basic material with a long history, CB has been applied into many commercial products including tires, 1-3 ink and paints, 4-6 and plastics, 7-9 working as a reinforcing, coloring, or conductive agent, respectively. Newly developed fields such as electronic equipment and portable devices are also incorporating CB to make use of its outstanding electrical conductivity. [10][11] Typically, RGO is obtained by partial chemical or thermal reduction of graphene oxide platelets (GO) prepared by a modified Hummers method. [12][13][14] In this way, the graphitic network structure can be restored in RGO exhibiting excellent electrical conductivity, mechanical and optical properties. These benefits bring RGO a number of applications in electronics and energy storage. [15][16][17][18][19][20] Generally, as a sheet-like material, the aspect ratio for RGO is very high, in the order of 10 3 -10 4 . This led to a much lower percolation threshold for RGO that makes RGO very promising towards reducing carbon content applications such as conductive composites and battery electrodes. 21 Given the exceptional characteristics of CB and RGO, investigating the rheological behavior and microstructure of their suspensions is also important for better understanding and handling in the manufacturing processes. In terms of imaging, cryogenic scanning electron microscopy (cryo-SEM) shows its excellence in preserving structures at frozen and nearly natural hydrated states, allowing us to examine the "real" morphology of studied specimens straightforwardly. The distortion of sample resulting from the drying process typically for conventional SEM can be avoided by this method. [22][23][24] The objective of this study is to investigate the rheological properties and  between CB particles. Addition of the acid protonated some of these groups and increased the surface charge, the repulsion of CB particles was reduced. Some CB started to form aggregates which immobilized part of the matrix inside them and therefore increased the effective solids volume fraction in the suspension and its viscosity. 29 As shear thinning occurred in much lower concentrations of CB at pH 3, the degree of CB aggregation at pH 3 was much larger than pH 5.

A B C
Towards reducing carbon content in the suspensions for future applications such as electrode slurries for lithium ion batteries, pH 3 was more effective with a lower CB concentration to form aggregation as well as network. Oscillatory frequency sweep experiments were performed at an oscillatory stress of 1Pa within the linear viscoelastic region to investigate the structures of the CB suspensions at 0.05-10wt% at pH 3. In Figure 4.2, we only present the valid data points where raw phase angle is below 175 degree and the system inertia is not dominant in this case. at 7.5% CB with frequency below 0.1 rad/s, a plateau region where G' equals to G" can be observed, indicating a gel formation in the sample which correlated with an elastic network structure. However, this structure broke down as the frequency increased showing splitting of G' and G". Afterwards, G" was always above G' in the tested frequency range, showing a viscous behavior. For G', it first slowly dropped with increasing frequency till 2 rad/s, then bounced back and gradually went up close to G" again.
This possibly suggested a reorganization of the broken structure in the suspension. For the other samples in the range of 1.5-10wt% CB, all suspensions showed a more fluidlike behavior with G">G' at low frequencies. Their crossover frequencies could not be accessed under current testing conditions. But a potentially solid-like behavior with G'>G" after crossover at medium-high frequencies can be predicted from the trends of G' and G". The moduli curves for samples below 1.5wt% CB are not shown here as they pretty much overlapped with the one at 1.5wt%.   At CB concentrations below 7.5wt% where no network was observed, we examined if addition of RGO influenced the rheological behavior and microstructure of the mixed suspensions at pH 3. As 4wt% CB, as shown in Figure 4.4, the viscosities of the mixed suspensions with RGO at both mass ratios 0.01wt% and 0.05wt% were nearly identical to that with no RGO (Figure 4.4A). Storage and loss moduli remained the same values and trend for all of the three samples (Figure 4.4B). Cryo-SEM imaging showed that RGO was mostly embedded in the CB aggregates in both suspensions ( Figure 4.4C). This was due to the attraction between CB and RGO with certain hydrophobicity at acidic condition. As CB concentration was much higher than that of RGO, addition of RGO was not sufficient to change the microstructures. Therefore, RGO addition was not helpful in forming network at medium CB concentrations.  5A and B). However, for both samples, the storage and loss moduli with added RGO did not cause any change, indicating only slight or no structural transorfomation (Figures 4.5C and D). Cryo-SEM images (Figures 4.5E and F) showed at low CB concentrations, with 0.05wt% RGO, RGO was dominant in forming the connections with some CB particles binding on RGO and it was beneficial for the partial network formation as backbones. This led to a viscosity increase in the mixed suspensions. But this effect was negligible in terms of the whole sample structure. For 0.05wt% CB, due to the total low concentration of CB and RGO, the mixture suspensions with lower or higher RGO concentrations showed the same viscosities as the one with no RGO with a Newtonian behavior, and there was no change in moduli as well. Therefore, the results were not shown in this work. CB+0.05wt% RGO (F) 0.5wt% CB+0.05wt% RGO. In E and F, RGO sheets marked by red arrows were dominant in forming some connections with CB particles binding on RGO. Scale Bars = 10µm.

Conclusions
We studied the rheological behavior and microstructures of the suspensions of CB and the mixture of CB and RGO in water. It turned out that pH value and CB suspensions. An obvious viscosity increase showing shear thinning behavior for CB above 0.5wt% was observed at pH 3 compared to pH 5 and 7.5. Cryo-SEM images showed the microstructure evolution of CB suspensions from forming aggregates to transient network at pH 3. Addition of high wt% RGO in suspensions containing 0.5wt% and 1.5wt% CB at pH 3 led to a viscosity increase corresponding to a microstructure change with RGO as the dominant component for forming connections.
However, the storage and loss moduli with added RGO did not show much difference from the one with no RGO, indicating no significant structural transition in the whole sample.

Acknowledgements
We gratefully acknowledge funding from Department of Energy, Office of Basic Energy Sciences, EPSCoR Implementation award DE-SC0007074.