E NABLING HIGHER ENERGY AND POWER DENSITY LITHIUM ION BATTERIES THROUGH ELECTRODE DESIGN AND THE INTEGRATION OF SOLID - STATE ELECTROLYTES By Yunsung Kim A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirement s for the degree of Materials Science and Engineering - D octor of P hilosophy 2015 ABSTRACT E NABLING HIGHER ENERGY AND POWER DENSITY LITHIUM ION BATTERIES THROUGH ELECTRODE DESIGN AND THE INTEGRATION OF SOLID - STATE ELECTROLYTES By Yunsung Kim At present , Li - ion technology is the leading battery chemistry to enable the large - scale adoption of electric vehicles. However, meeting the demands of hybrid and plug - in hybrid electric vehicles requires higher specific and volumetric energy density, faster charge rates, longer cycle life, and improved safety. A particular focus is on achieving high power density without compromising energy density. This dissertation seeks to determine the phenomena that couple energy and power density and to develop solutions to simultaneously increase both. An engineered electrode design is proposed that improves Li - ion transport in thick high energy density electrodes, while suppressing the deleterious formation of Li metal dendrites during charging. Furthermore, a novel hybr id cell design is proposed employing Li 7 La 3 Zr 2 O 12 (LLZO) ceramic electrolyte membrane technology, which acts as a physical barrier to prevent Li metal dendrite propagation. The overarching goal of this dissertation is to develop materials and materials pr ocessing technology to improve the performance and safety of Li - ion batteries. Copyright by YUNSUNG KIM 2015 iv ACKNOWLEDGEMENTS First, I would like to express my sincere appreciation to my advisor Dr. Jeff Sakamoto for givin g me this opportunity to work in the field of energy storage. I would like to sincerely express my gratitude for his advice and his support throughout my Ph. D. degree. I would like to thank him for encouraging me to challenge myself and always do better . Shahriari, Maria Regina Garcia Mendez, Asma Sharafi, Dan Lynam, Isabel David, Dr. Youngsam Park, Dr. Ezhilyl Rangasamy, Dr. Travis Thompson, Dr. Heechul Lee, Dr. Robert Sch midt, Dr. Eric Jianfeng Cheng, and others who always support me in intelligently and emotionally. I would like to thank my committee members Dr. Carl Boehlert, Dr. Scott Calabrese Barton, Dr. Donald Morelli , and Dr. Viktor Poltavets for agreeing to be part of my committee . I would also like to thank them for their precious time and for their great personality to help me throughout the course of this Ph . D. I would like to express my sincere appreciation to my former advisor Dr. Heeman Choe and his gr oup: Eunji Hong , Dr. Hyelim Choi, Taehoon You, Minchol Hyeon, Yoonsook Noh, Dongjun Shin, Hyoungjoo Lee, Myounggeun Choi, Hyungyung Jo, Hyeji Park, Jooyoung Kim and Jina Kim who giving brains to me whenever I am having a tough time. I have to also thank Dr . Jeff Wolfenstine, Dr. Andy Drews, Dr. Raji Chandrasekaran, Dr. Ted Miller , Eongyu Yi , and Per Askeland for their assistance in interpretation and characterization in my research. v Finally I would like to thank my parents Guansuck Ki m and Yoosoon Jang, m y sister Jinyoung Kim, and my brother Hyunsung Kim for their unconditional support and love during my Ph. D. studies. I would like to express my sincere appreciation to them respecting this endeavor. vi TABLE OF CONTENTS . i x . . x 1 Introductio n ................................ ................................ ................................ ............................... 1 1.1 Energy demand and storage technology need ................................ ............................. 1 1.2 Energy storage technologies ................................ ................................ ....................... 1 1.3 Li - ion batter ies ................................ ................................ ................................ ............ 3 1.3.1 Li - ion battery operation principles ................................ ................................ .............. 4 1.3.2 Negative electrodes (anodes) ................................ ................................ ...................... 6 ................................ ................................ ......... ................................ ................................ .............. 1.3.3 Positive electrodes (cathodes) ................................ ................................ ................... 11 ................................ ....................... ................................ .......................... ................................ ........................ 1.3.4 Electrolytes ................................ ................................ ................................ ................ 14 ................................ ................................ ........................ - ................................ ................................ ........ 1.4 Challenges for electric vehicle systems ................................ ................................ .... 17 1.4.1 Power and energy ................................ ................................ ................................ ...... 18 1.4.2 Safety ................................ ................................ ................................ ......................... 22 1.5 Need for cell design ................................ ................................ ................................ .. 23 1.5.1 Electrode design approaches ................................ ................................ ..................... 23 1.5.2 Cell design with garnet - like solid electrolyte ................................ ............................ 24 1.6 Scope of present work ................................ ................................ ............................... 26 2 Experimental methodology ................................ ................................ ................................ ..... 28 2.1 Electrode preparation ................................ ................................ ................................ 28 2.2 Ceramic electrolyte processing ................................ ................................ ................. 29 2.2.1 Powder preparation ................................ ................................ ................................ ... 29 2.2.2 Consolidation ................................ ................................ ................................ ............ 30 2.3 Electrochemical measurement methods ................................ ................................ .... 32 2.3.1 Galvanostatic rate mapping test ................................ ................................ ................ 32 2.3.2 Intentional overcharge test ................................ ................................ ........................ 36 2.3.3 Galvanostatic intermittent titration technique (GITT) ................................ .............. 37 ................................ ................................ ............ 2.3.4 Electrochemical impedance spectroscopy (EIS) ................................ ....................... 40 - ................................ .................. 2.4 Mechanical property characterization ................................ ................................ ....... 46 2.4.1 Vickers hardness ................................ ................................ ................................ ....... 46 vii 2.4.2 Nano indentation ................................ ................................ ................................ ....... 48 2.5 Other methods ................................ ................................ ................................ ........... 49 2.5.1 Laser patterning ................................ ................................ ................................ ......... 49 2.5.2 X - ray power diffraction (XRD) ................................ ................................ ................ 50 2.5.3 Scanning electron microscope (SEM) ................................ ................................ ....... 51 2.5.4 Raman ................................ ................................ ................................ ....................... 51 3 Determining power limiting process and understanding cell failure mechanisms ................. 53 3.1 Intercalation vs d eintercalation rate ................................ ................................ .......... 53 3.2 Rate limiting process es ................................ ................................ ............................. 55 3.2.1 Electron injection and extraction resistances ................................ ............................ 57 3.2.2 Ion insertion and extraction resistances at the int erface between electrode and electrolyte ................................ ................................ ................................ ................................ 57 ................................ ........................... ................................ ................................ ................................ .... - ................................ .................. 3.3 Understanding cell failure mechanisms ................................ ................................ .... 67 3.4 Summary ................................ ................................ ................................ ................... 72 4 Laser patterned electrodes ................................ ................................ ................................ ....... 73 4.1 Background: three dimensional (3D) e lectrode designs ................................ ........... 73 4.2 Highly ordered hierarchical (HOH) graphite electrode ................................ ............ 79 4.2.1 Laser patterning technique ................................ ................................ ........................ 80 4.2.2 HOH electrode design ................................ ................................ ............................... 80 4.2.3 HOH electrode design optimization ................................ ................................ .......... 82 4.2.4 HOH electrode characterization ................................ ................................ ................ 87 ................................ .......... ................................ ................................ ............... 4.3 Summary ................................ ................................ ................................ ................... 91 5 Electrochemical characterization of HOH electrodes ................................ ............................. 92 5.1 Solid - state Li diffusivity in graphite electrode ................................ .......................... 92 5.2 Rate mapping ................................ ................................ ................................ ............ 95 5.2.1 Effects of loading ................................ ................................ ................................ ...... 95 5.2.2 HOH graphite electrode vs c onventional graphite electrode ................................ ..... 97 ................................ ................................ .................. 5.3 Cell impedance characterization ................................ ................................ ............. 107 5.3.1 Polarization interrupt test ................................ ................................ ........................ 107 5.3.2 Transmission l ine m ethod (TLM) and EIS characterization ................................ ... 108 - ................................ ............ ................................ ................................ ................................ ................ 5.4 Summary ................................ ................................ ................................ ................. 116 6 T he effect of microstructure on the mechanical properties of hot - pressed cubic Li 7 La 3 Zr 2 O 12 ... 118 viii 6.1 LLZO ceramic electrolyte characterization ................................ ............................ 119 6.1.1 Density of LLZO ................................ ................................ ................................ ..... 119 6.1.2 Phase characterization ................................ ................................ ............................. 119 6.1.3 Micro structure of LLZO ................................ ................................ ........................ 121 6.2 Mechanical properties of LLZO ................................ ................................ .............. 127 6.2.1 Hardness of LLZO ................................ ................................ ................................ .. 127 6.2.2 Fracture toughness of LLZO ................................ ................................ ................... 131 6.3 Ionic conductivity of LLZO ................................ ................................ .................... 134 6.4 Summary ................................ ................................ ................................ ................. 136 7 Summary and f uture work ................................ ................................ ................................ ..... 138 7.1 Summary ................................ ................................ ................................ ................. 138 7.2 Future work ................................ ................................ ................................ ............. 139 7.2.1 HOH charge abuse testing ................................ ................................ ....................... 139 7.2.2 Rate mapping at low temperature ................................ ................................ ........... 142 7 .2.3 Realizing a novel hybrid cell design with ceramic electrolytes .............................. 142 REFERENCES ................................ ................................ ................................ ............................. 147 ix LIST OF TABLES Table 1 - 1: USABC target of obtaining a high - energy storage and low - cost electric vehicle battery applications [ 5 7 ] . Depth of discharge (DOD), State of charge (SOC). ................................ ......... 20 - E is the Youn H v is the Vickers hardness, c is the crack length, a is the length of half diagonal, and P is the applied load .) ................................ ................................ .............................. 47 Table 6 - 1: It presents the information of hot - pressed LLZO pellets as changing hot - pressing time . ... 119 x LIST OF FIGURES Figure 1 - 1: Comparison of the different battery systems in terms of gravimetric power and energy density [ 7 ]. ................................ ................................ ................................ ................................ ........ 3 Figure 1 - 2: Schematic of the principle operation of a Li - ion battery [ 7]. (Modified from [16 ] ) ..... 5 Figure 1 - 3 : Schematic of (a) Li - ion intercalation process between graphene layers, (b) staging of graphite during Li intercalation process [ 22 ]. (Modified from [24]) ................................ .............. 8 Figure 1 - 4: (a) Voltage vs capacity of various electrode materials [28] , (b) Volume change effects associated to the charge and discharge process es of me tal Li - alloy ing electrodes in Li - ion battery (left lower) [27] . SEM images of the discharged Sn at different cycle numbers [ 26 ]. ................. 10 Figure 1 - 5: The crystal structure of cubic LLZO. ................................ ................................ .......... 17 Figure 1 - 6: Correlation between specific energy, electrode loading, and open porosity assuming LiMO 2 and no packaging. ................................ .............. 21 Figure 1 - 7: Dependence of the power capability of SFG44 graphite electr odes in the electrode loading [19] . ................................ ................................ ................................ ................................ ... 21 Figure 1 - 8: Schematic of hybrid cell design composed of graphite negative electrode, LiCoO 2 positive electrode , liquid electrolyte, and LLZO ceramic electrolyte where between electrodes. . 25 Figure 1 - 9: Ragone plot for various energy devices [72 ]. ................................ .............................. 27 Figure 2 - 1: Processes flow diagram describing the graphite electrode fabrication process. ......... 29 - image of the rapid induction hot - pressing. (b) Schematic of the cross - section of a graphite die with LLZO power for hot - press. ................................ ................................ .............. 31 Figure 2 - 3: (a) An image of a Swagelok ® cell with 1/2 inch diameter stainless stee l 304 pins. (b) Schematic and the image of stepped stainless steel 304 pins (Outer diameter: 1/2 inch, inner diameter: 3/8 inch), and (c) Swagelok ® cell under pressure (45 N) by a force gage. ................... 34 - ................................ ................................ ................................ ....... 38 Figure 2 - 5: Schematic of a symmetric cell for a free - standing electrode [80]. .............................. 40 Figure 2 - 6: Schematics of (a) a typical equivalent cir cuit for a solid electrolyte and (b) a cell preparation with solid electrolyte between blocking electrodes for EIS. ................................ ....... 42 xi Figure 2 - 7: Schematics of (a) simple equivalent circuit model and (b) a typical Nyquist behavior for porou s electrodes. ................................ ................................ ................................ ..................... 43 Figure 2 - 8: Nyquist plots for symmetric cells using two positive electrodes. (a) SOC = 0 % (squares) and (b) SOC = 50 % (circles). The solid lines are the best - fitted results with the equivalent circuits us ing Eq. 2 - 6 and Eq. 2 - 7 for (a) and (b), respectively [83]. ........................... 45 Figure 2 - 9: The laser patterning equipment fabricated by integrating a computer numerical control (CNC) 3D positioning system with the laser beam. ................................ ........................... 50 Figure 3 - 1: Charge and discharge rate mapping of the low loading (1.15 mAh cm - 2 and 50 % total open porosity) graphite electrodes. Black data: Intercalation (charge) rate capability, Red data: deintercalation (discharge) rate capability . ................................ ................................ .................... 54 Figure 3 - 2: (a) Schematic representation of possible Li - ion diffusion paths in electrolyte - filled pores in a graphite electrode, (b) Li intercalation process at a particle scale. ................................ 56 Figure 3 - 3: Two mechanisms for the electrochemically induced reduction of carbonate - based ................................ ........................... 59 Figure 3 - 4: Complex impedance plot of Li/graphite half - cell in the delithiated state (SOC=0%). 60 Figure 3 - 5: (a) 3D reconstruction of a graphite electrode (2.8 mAh cm - 1 and 40 %) by FIB - SEM technique using MIMICS® software. (b) SEM image of a graphite electrode by FIB - SEM. ....... 64 Figure 3 - 6: Schematic of spherical graphite particle. This figure shows solid state diffusion time according to SOC level. L is diffusion length, t is time for Li diffusion, and D is Li diffusion coefficient. ................................ ................................ ................................ ................................ ...... 67 Fig ure 3 - 7: A schematic of thermal runaway causes fires by improper charging in Li - ion batteries. (a) normal state battery, (b) Li dendrite formation due to improper charge such as fast charging, (c) short - circuiting by Li dendrite growth and short circuit on th e positive electrode causing instantaneous discharge, (d) cell temperature goes up (>70 ° C) by Joule heating and electrolyte start to decompose, then flammable hydrocarbon gases are released, (e) Joule heating and exothermic reactions further increase temp erature, and the metal oxide positive electrode starts to decompose (>1 5 0 ° C), then releasing oxygen. These steps can cause cell failure and explosion. (Cell swelling f igure [ 6 5 ]). ................................ ................................ ................................ ............. 69 Figure 3 - 8: (a) Optical image of the sur face and (b) fracture surface of over charged SFG6 graphite electrode at 1 C - rate for 1 h, and (c) optical images represent color change in SFG6 graphite electrode by SOC. ................................ ................................ ................................ ............ 71 Figure 4 - 1: Previously reported 3D architecture el ectrode designs and fabrication methods. (a) Process for fabricating the hierarchical V 2 O 5 electrode [6 8 ] . (b) O utline of the Ni foam fabrication by template based method. Lower image is MnO 2 electrode fabricated by electrodeposition on Ni foam [6 9 ]. (c) A schematic of 3D image of pillars by Super ink jet xii printing [1 20 ]. (d) Outline of the electrode fabrication process. Left lower shows the surface of a patterned electrode and right lower shows cross - section of a patterned electrode [ 70 ]. ................ 77 Figure 4 - 2: Schematic representation of possible Li - ion diffusion paths (a) in a convention al porous electrode, (b) in a HOH electrode , and schematic of top view of HOH electrode and short Li - ion diffusion length induce d by hexagonal close - packed linear channels. ............................... 82 Figure 4 - 3: Secondary SEM images of laser patterned electrode (Timcal, SFG6, 4.0 mAh cm - 2 , 50 % porosity) (a) top view of fabricated HOH electrode , (b) cross - section of a conica l shaped pattern, (c) c ollapsed walls between laser - ablated channels , and (d) a laser cut HOH electrode after laser patterning (3/8 inch diameter) . ................................ ................................ ...................... 85 Fig ure 4 - 4: Raman spot analysis of an HOH graphite electrode at various spots (1 to 4 and cross - section) . ................................ ................................ ................................ ................................ .......... 88 Figure 4 - 5: SEM images of laser patterned graphite electrode (SFG6 graphite electrode with 5.5 mAh cm - 2 and 50 % total open porosity) . (a) T op view of HOH electrode before and (b) after rate mapping. ................................ ................................ ................................ ................................ ......... 90 Figure 5 - 1: Typical potential vs x in Li x C 6 plot with 1.2 mAh cm - 2 and 63 % SFG6 graphite electrode. ................................ ................................ ................................ ................................ ........ 93 Figure 5 - 2: The GITT plot of the graphite electrode with 1.2 mAh cm - 2 and 63 % porosity. The measured potential range was 80 mV to 0.75 V. ................................ ................................ ............ 95 Figure 5 - 3: Results of rate mapping as a function of graphi te electrodes with various loading from 1.15 mAh cm - 2 to 5.5 mAh cm - 2 with the same total open porosity (50 %) . N=4. ....................... 97 Fig ure 5 - 4: Charge rate mapping as a function of SFG6 graphite electrodes with conventional and HOH elec trode s with 50 % total open porosity . (a) C apacity (%) vs intercalation rate with 4 mAh cm - 2 , (b) with 5.5 mAh cm - 2 , and (c) specific capacity (mAh g - 1 ) vs intercalation rate with 5.5 mAh cm - 2 . N=4. ................................ ................................ ................................ ............................ 100 Figure 5 - 5: Schematic diagram showing the Li concentration and diffusivity profiles in a graphite electrode [126]. ................................ ................................ ................................ ............................. 102 Figure 5 - 6: SEM images of (a) Celgard 2400 ® and (b) Zeus ® separators . ................................ . 105 Figure 5 - 7: Typical pr e conditioning cycles with Zeus ® separators and with different graphite electrode loadings, which were (a) 1.2 mAh cm - 2 and (b) 5.5 mAh cm - 2 , respectively ............... 106 Figure 5 - 8: Galvanostatic polarization, followed by interrupt and r elaxation test (HOH vs Conventional electrode with 5.5 mAh cm - 2 and 50 %), and the schematic of symmetric cells for polarization interrupt [ 80 ] . ................................ ................................ ................................ ........... 108 xiii Figure 5 - 9: Simulated Nyquist plots for a cylindrical pore in an electrode w ith different models. (a) Non - faradaic, (b) faradaic with low charge transfer resistance, and (c) is faradaic with high charge transfer resistance [ 83 ]. ................................ ................................ ................................ ..... 110 Figure 5 - 10: (a) Schematic representation symmetric cell (SC)[ 83 ], N yquist plots after TLM - EIS - SC test s with (b) 1.2 mAh cm - 2 and 50 % and (c) 5.5 mAh cm - 2 and 50 % SFG6 symmetric cells. ... 112 Figure 5 - 11: Nyquist plots for symmetric cells with two graphite electrodes at SOC 0 %. T he loading of 5.5 mAh cm - 2 and porosity of 50 % c onventional graphite electrode s were used. ..... 114 Figure 5 - 12: Nyquist plots after TLM - EIS - SC test with HOH symmetric cell (5.5 mAh cm - 2 and 45 + 5 %). ................................ ................................ ................................ ................................ ..... 115 Figure 6 - 1: X - ray diffraction patterns of Li 6.19 Al 0.27 La 3 Zr 2 O 12 calcined powder and hot - pressed pellets pressed for 30, 60, 90, and 240 min at 1050 o C. * Pyrochlore (La 2 Zr 2 O 7 ) ....................... 120 Figure 6 - 2: Fracture surface of Li 6.19 Al 0.27 La 3 Zr 2 O 12 hot - pressed for : (a) 30 min , (b) 60 min , (c) 90 min , and (d) 240 min . The relative densities are indicated in top right of each image. ......... 122 Figure 6 - 3: Li 6.19 Al 0.27 La 3 Zr 2 O 12 hot - pressed pellets after th ermal etching at 700 o C for 30 min in air. The Li 6.19 Al 0.27 La 3 Zr 2 O 12 pellets were hot - pressed at 1050 o C for: (a) 30 min, (b) 60 min, (c) 90 min, and (d) 240 min. The relative densities are indicated in top right of each image. ......... 124 Figure 6 - 4: Grain size distributions of hot - pressed Li 6.19 Al 0.27 La 3 Zr 2 O 12 . ................................ ... 126 Figure 6 - 5: H v and H n of Li 6.19 Al 0.27 La 3 Zr 2 O 12 as a function of relative density. ........................ 129 Fig ure 6 - 6: H v vs lattice parameter for single crystalline garnets from the literature (open squares)[139] and the value for Li 6.19 Al 0.27 La 3 Zr 2 O 12 from this work (closed square) . ............... 130 Fig ure 6 - 7 : Fracture toughness of Li 6.19 Al 0.2 7 La 3 Zr 2 O 12 as a function of relative density. .......... 132 Fig ure 6 - 8 : T he Vickers indentation crack propagation path trajectories for (a) relative density of 85 % and (b) relative density of 98 %. Arrows point to crack the propagati on path in each grain. ... 133 Figure 6 - 9: Total ionic conductivity of Li 6.19 Al 0.27 La 3 Zr 2 O 12 as a function of relative density . .. 135 Figure 7 - 1: (a) Optical top image and (b) fracture surface images of electrode with 5.5 mAh cm - 2 and 35 total porosity (30 % intrinsic porosity and 5 % laser ablated porosity) after overcharging at 1 C - rate for 1 h. ................................ ................................ ................................ ............................ 141 Figure 7 - 2: Schematic of (a) cell configuration for asymmetric DC test, (b) the results of the DC test, and (c) DC cycling test. The DC test was conducted after conditioning cycles at 0.01 mA xiv cm - 2 for 10 symmetric cycles (each cycle takes 2h). Then DC cycling test was conducted at 1 mA cm - 2 for 20 cycles. The each cycle takes 2h. ................................ ................................ ......... 144 Figure 7 - 3: Schematic of novel hybrid design of Li - ion batteries with combining HOH concept and LLZO electrolyte for the higher performance and safety. ................................ ..................... 146 1 1 Introductio n 1.1 Energy demand and storage tec hnology need Fossil fuels are the primary source of anthropogenic energy [1 - 2]. However, fossil fuels are not only non - renewable energy sources, their combustion results in air pollution such as carbon dioxide, sulfur dioxide, and nitrous oxide, the latte r of which converts into ozone in the presence of sunlight [1,3]. To reduce and eventually eliminate the dependency on fossil fuels, renewable energy resources and technology have been investigated [4]. Most of the more mature renewable energy sources, s uch as thermal, wind, and solar energies, do not produce greenhouse gases, but they do not continuously produce energy [1,4]. To facilitate renewable energy generation technologies, compl e mentary energy storage technology is needed. 1.2 Energy storage techno logies The large - scale stationary energy storage technologies enable to use intermittent renewable energy along the energy demand curve. Therefore, energy storage technology is a key enabler for the implementation of electric vehicles and the smart grid c oncept. However, developing large scale energy storage systems is not trivial. Energy storage systems can be divided into several different categories such as mechanical, electrical, chemical, and electrochemical [5]. Examples of mechanical energy storag e systems include flywheels, compressed air energy storage, and pumped - storage hydroelectricity (pumped hydro). Examples of electrical storage systems are capacitors and superconductive electromagnetic storage. An example of chemical storage is the energ y stored in the form of 2 hydrogen. Electrochemical energy storage systems are those such as rechargeable battery, fuel cells, and redox flow batteries. Each energy storage systems typically have distinguishing performance characteristics, i.e. supercapaci tors have high specific power (>10 3 W kg - 1 ), but low energy density (<10 Wh kg - 1 )[6]. Electrochemical energy storage systems are typically considered as one of the most promising energy storage technologies because they generally possess a number of desir able characteristics such as long cycle life, moderate power and energy, high efficiency, and eco - friendly chemistry [7]. To date, several different types of batteries have been developed and utilized. A rechargeable battery is an electrochemical energy storage device that is able to store electrical energy, in the form of chemical potential, and convert the chemical energy into electricity, reversibly. Batteries are typically composed of negative electrode , positive electrode , and liquid electrolyte. During charging and discharging, ions move through the electrolyte, and electrons transport via an external circuit to maintain charge neutrality in the cell. The cell potential is determined by Nernst equation. Nickel - metal hybrid batteries are still us ed for some portable devices, but are being replaced by Li - ion. At present, l ead - acid batteries are widely adopted as the battery of choice for vehicle starting and back up grid storage. In comparison, Li - ion batteries have the highest specific energy an d power compared to other battery types (Figure1 - 1)[ 7 ]. More detailed discussion about Li - ion batteries is presented below . 3 Figure 1 - 1: Comparison of the different battery systems in terms of gravimetric power and energy density [ 7 ]. 1.3 Li - ion batter ies S ince Li has not only low redox potential to generate high cell voltage ( - 3.04 V vs H/H + ), it is also light weight (0.53 g cm - 3 ), making it a promising candidate electrode for batteries [ 8 ]. However, the use of metallic Li as a negative electrode is hinder ed by the formation of Li dendrites, which can cause short - circuiting leading to ignition [ 8 ]. To mitigate the Li electrode instability, yet take advantage of the low redox potential, alternative carbon - based negative electrodes were developed [ 9 ]. Rathe r than depositing Li on the surface, Li - ions are inserted into carbon - based negative electrodes , thus enabling the invention of Li - ion batter ies . In 1991, the Sony® Corporation commercialized Li - ion batteries and since has dominated the market for 4 portabl e electronic devices such as cellular phones, computers, and digital cameras due to their high energy density compared to other batteries [4, 8 - 11 ]. 1.3.1 Li - ion battery operation principles Li - ion batteries consist of three primary components: (i) a graphite ne gative electrode (anode), (ii) a non - aqueous liquid electrolyte permeating a porous polymeric membrane (separator) to transport Li - ions between electrodes, and (iii) a transition metal oxide, such as LiCoO 2 , LiMn 2 O 4 , or LiNi 0.33 Mn 0.33 Co 0.33 O 2 positive elec trode (cathode)(Figure 1 - 2)[ 7,12 - 14 ]. During operation, Li - ions are inserted or extracted from the electrodes and diffuse through the liquid electrolyte while electrons are transported through an external circuit to maintain charge neutrality in the cell (Figure 1 - 2)[ 7 ]. The graphite negative electrode and transition - metal oxide positive electrode get reduced during the charge and discharge processes, respectively. The half - reactions and overall cell reaction can be written as The chemical driving forc e for charge and discharge is caused by the difference of the chemical potentials between the electrode materials. The driving force for the redox reactions during charge and discharge processes is given by 5 ( Eq. 1 - 1) where is free energy change for the reaction, z is the charge number of the mobile ionic species, F is Faraday constant (96,500 C), and E is cell potential between electrodes [ 1 5]. Figure 1 - 2: Schematic of the principle operation of a Li - ion battery [ 7 ]. (Modified from [ 16 ] ) 6 1.3.2 Negative electrodes (anodes) Based on convention, the negative electrode is the electrode where oxidation occurs during discharge [ 17 ]. To maximize battery performance, the negative electrode material s should have several attribu tes [ 15 ]. First, it should have a low redox potential to provide high cell potential when coupled with a relatively high redox potential positive electrode . Second, the volume change should be minimized during cycling to reduce fatigue and decrepitation. Third, the negative electrode material s ideally should be a mixed conductor with equally high ionic and electronic conductivity. The ionic and electrical conductivity limit how quickly a Li - ion battery can be charged and discharged or also referred to a s power. Lastly, it should have a high specific and/or volumetric capacity to maximize the quantity of Li stored per unit mass or volume, respectively. 1.3.2.1 Graphite negative electrode s Yazami et al .[ 1 8 ] was the first to propose the use of a graphite negativ e electrode in 1983. Today, graphite negative electrode s are almost exclusively used in state - of - the - art commercial Li - ion batteries owing to their relatively long cycle life, low discharge potential, low cost, and abundance of precursors [ 9 , 19 - 20 ]. Grap hite exhibits sp 2 - hybridized bonding, and consists of stacked layers of graphene. The layers are bonded by weak Van der Waals force caused by the - orbitals [ 2 1 ]. Since the electrons can transport between the graphite layers relatively freely, graphite has a high electrical conductivity. During charging, Li - ions are electrochemically inserted vs Li/Li + )(Figure 1 - 3a)[ 12 , 22 ]. 7 Several Li - ion staging phenomena comprise distinct Li - C ordering when x varies between 0 and 1 in Li x C 6 , which has a theoretical specific capacity of 372 mAh g - 1 [1 9 ]. As shown in Figure 1 - 3b, the intercalation of Li - ions int o graphite shows several plateau regimes ( staging ) . This staging is a thermodynamic phenomenon, and indicates that graphite undergoes phase transition from ABAB stacking to AAAA stacking filled with Li - ion between every graphite layers (Figure 1 - 2b)[ 22 - 23 ]. The plateau regime s exhibit coexistence of two phases resulted from a difference in the energy required to expand the graphene layers and the repulsive force between Li - ions [ 22 - 23 ]. 8 (a) (b) Figure 1 - 3 : Schematic of (a) Li - ion intercal ation process between graphene layers, (b) staging of graphite during Li intercalation process [ 22 ]. (Modified from [24]) 9 1.3.2.2 Alloy negative electrodes I n recent years, Li - alloy ing materials , such as Si and Sn , have been considered as attractive negative ele ctrode materials due to their significantly high theoretical energy capacities [ 25 - 27 ]. During charging, alloys store Li by forming Li compounds [ 12,2 5 - 2 7 ]. The Li - alloying process can be presented by the following reactions [ 12 ]: where M is Si, Sn, Pb , Sb, Al, and Bi. As a result, theoretical energy capacities of Si and Sn are 4200 mAh g - 1 and 992 mAh g - 1 , respectively [ 2 5 - 2 7 ]. Although the maximum energy capacity of these materials is 10 times higher than that of a conventional graphite electrode (3 72 mAh g - 1 ), they have relatively high operating potentials resulting in lower cell potentials (Figure 1 - 4a) [ 28 ] . Also, alloys are notorious for undergoing severe volume change during cycling (Figure 1 - 4b)[ 8 ,26 - 27 ]. For example, the process es of alloying and dealloying cause a volume change up to 400 % in a Si negative electrode [ 2 5 ]. The mechanical stress related to expansion and contraction leads to decrepitation of the electrode and capacity rapid ly fade (Figure 1 - 4 b)[ 26 - 27 ]. To alleviate these probl ems, various approaches such as reducing particle size, designing stress - reducing structures, and selecting intermetallic alloys are suggested [ 8,2 5 - 2 7 ]. Despite these efforts to reduce volume expansion, the short cycle life of Si and /or Sn has still not been solved [ 8 ]. 10 (a) (b) Figure 1 - 4: (a) Voltage vs capacity of various electrode materials [28 ] , (b) Volume change effects associated to the charge and discharge process es of metal Li - alloy ing electrodes in Li - ion battery (left lower) [2 7 ] . SEM imag es of the discharged Sn at different cycle numbers [ 2 6 ]. 11 1.3.3 Positive electrodes ( c athodes) Based on convention, the positive electrode is the electrode where reduction occurs during discharge [ 15,17 ]. LiCoO 2 was the positive electrode material originally pa ired with graphite electrode [ 2 9 ]. LiCoO 2 is a good example of a positive electrode because it has relatively high redox potential, is a chemically and thermally stable structure, and is a good mixed conductor. In general, positive electrode materials ca n be classified based on their atomic structure such as layered, spinel, and olivine compounds [ 30 ]. Spinel and olivine refer to general mineral names for families of transition metal oxides. The requirements for positive electrode materials are high spe cific and volumetric capacity, power, cycle life, and safety [ 30 ]. To maximize the quantity of Li stored per unit mass or volume, a high specific and/ or volumetric capacity is required, respectively. In addition, the irreversible phase transition should not occur for long cycle life, and the chemical and electrochemical stabilities are required for safety, respectively. 1.3.3.1 Layered positive electrode compounds The layered structure compounds with LiMO 2 (M=Co, Ni, and Mn) consist of the oxygen anions forming a close - packed structure with cations located in the 6 - fold coordinated octahedral sites. The LiMO 2 compounds exhibit the O - Li - O - M - O - Li - O - M - O (MO 2 - Li - MO 2 - Li) repeating structure. Since the MO 2 layer forms strong ionic bonds and there is Coulombic repulsi on between MO 2 layers, Li - ion s de / intercalation between MO 2 layers is possible [ 31 ]. LiCoO 2 has been widely used as a positive electrode material for 20 years since LiCoO 2 was firstly commercialized in the early 1990s [ 30 ]. However, only ~50 % of the theo retical capacity 12 of LiCoO 2 (274 mAh g - 1 ) is available because LiCoO 2 is unstable and the phase transition occur s when more than 50 % of the Li - ions are extracted [ 32 - 33 ]. In addition, concerns regarding cost and environmental problems related to cobalt ha ve driven research to focus on alternative transition metal positive electrodes which are more abundant and environmental friendly [ 30 ]. Subsequently other positive electrode material compounds, such as LiNiO 2 and LiMnO 2 , have been developed [ 30,34 ]. How ever, the capacity rapidly decreases as a function of cycles due to crystallographic instability. In the case of LiNiO 2 , Ni 2+ migrates into Li sites which can hinder Li diffusion [ 30, 3 4 ]. LiMnO 2 can also be unstable due to Jahn - Teller distortions causin g a sliding of the basal planes at higher de intercalation states [ 3 5 ]. These problems have limited the use alternative positive electrodes in a commercial Li - ion battery [ 3 5 ]. 1.3.3.2 Spinel positive electrode compounds The LiM 2 O 4 (M=Ti, V, and Mn) compounds have the spinel structure. The oxygen framework of LiM 2 O 4 is the same with that of the layered structure, but 1/4 of M ions are located in the Li layer result s in leaving Li vacancies in transition metal layer [ 30 ]. These vacancies create empty octahedral si tes that share faces with the tetrahedral sites occupied with Li in the Li layer. This three - dimensional (3D) Li diffusion path allows fast de/intercalation rates [ 3 6 ]. LiMn 2 O 4 [ 3 6 ] is a common cathode owing to the fact that Mn is abundant and eco - friendl y. However, this positive electrode compound has a relatively low theoretical specific capacity (~148 mAh g - 1 ) and undergoes capacity fade due to the following reasons [ 30 ]: 1) Since the surface of the positive electrode has the higher Li concentration at the beginning of 13 de/intercalation owing to concentration polarization, it undergoes phase transition from the cubic phase to the tetragonal phase that leads to micro - strain and result s in severe capacity loss through Jahn - Teller distortion. The concentra tion polarization will be described in detail in C hapter 3. 2) Disproportionation of Mn ions during discharge process causes 2Mn 3+ = Mn 2+ + Mn 4+ reaction and Mn 2+ is dissolved in the liquid electrolyte. Consequently, the amount of LiMn 2 O 4 active material s is reduced and the dissolved Mn 2+ can be electrochemically deposited on a negative electrode which caused the decomposition of the liquid electrolyte by acting as catalyst [ 30 ]. 1.3.3.3 Olivine positive electrode compounds Iron has been commercially used in va rious industries due to its low cost, non - toxicity, and abundant characters. Therefore the positive electrode materials including iron have been investigated and the most attractive positive electrode compound is olivine structure of LiFePO 4 which was fir st developed by Padhi et al . in 1997 [ 3 7 ]. It has theoretical capacity of ~170 mAh g - 1 with ~3.4 V operating potential [ 3 7 ]. In addition, it is well known that LiFePO 4 has the high structural and chemical stability related with enhanced cycle performance . However, it suffers low electronic conductivity issue [ 34 ], and Nishimura et al . [ 38 ] experimentally demonstrated Li transport path is one dimensional channel along the (101) which slows Li diffusion. In spite of its drawbacks, it is widely adapted as p ositive electrode material s today by improving its limitations through nanodimensional processing/effects and/or doping methods [ 30 ]. 14 1.3.4 Electrolytes An electrolyte provides Li - ions transport, but no t electron transport, between electrodes during charging a nd discharging (Figure 1 - 2). Since high ion ic conductivity is required to obtain high or adequate power, the role of electrolyte is important in Li - ion batteries . The ideal electrolyte should have: i) high ionic conductivity, ii) high electrochemical and thermal stability, iii) low cost, and iv) negligible electronic conductivity. 1.3.4.1 Liquid electrolyte Since the charge and discharge potential of Li - ion batteries (>3 V) are beyond the decomposition potential of an aqueous electrolyte (~1 V), the aqueous elec trolyte cannot be utilized for Li - ion batteries. Hence, electrolytes consisting of inorganic salts (e.g. LiClO 4 and LiPF 6 ) dissolved in a mixtures of alkyl carbonates (non - aqueous organic solvents) including ethylene carbonate (EC), dimethyl carbonate (DM C), diethyl carbonate (DEC), propylene carbonate (PC) and ethyl - methyl carbonates (EMC) have been developed [ 39 ]. Alkyl carbonates are accepted as an electrolyte solvent owing to their stability for the 4 V positive electrodes . Also, the performance of l iquid electrolytes significantly depends on the mixed solvent compositions. Since the operating temperature range of Li - ion batter ies is between - 20 and 60 °C, the electrolyte must be stable with high ionic conductivity in this temperature range. The ion ic conductivity is proportional to mobility of solvent and the concentration of mobile ions [ 39 ]. In general, EC has a high dielectric constant, but its viscosity and melting point are high (~36.4 °C ) [ 40 - 42 ]. On the other hand, linear carbonates such as DMC and DEC have low viscosity but relatively low 15 dielectric constant compared to EC [ 40 - 42 ]. Additionally, the electrolyte compositions affect the formation of the solid electrolyte interphase (SEI) layer on the surface of graphite electrode during the initial charge cycles due to its thermodynamical ly instability at 0.4 - 0.9 V vs Li + /Li [ 43 ]. The SEI formation will be discussed in C hapter 3. Due to the above stated reasons, the solvent compositions have been optimized and widely used in Li - ion batterie s. However, the flammability of liquid electrolytes causes safety concerns such as fires and explosions [ 44 - 46 ]. 1.3.4.2 Garnet - like solid electrolyte Solid - state electrolytes are an alternative solution to mitigate the risk of combustion in Li - ion batteries [ 47 ]. Since liquid electrolytes are flammable, solid - state electrolytes have garnered significant attention as a next generation electrolyte due to 1) non - flammability, 2) possible simplified cell fabrication, 3) reduced packaging mass, and 4) low cost [ 4 7 ]. Therefore, various types of solid - state electrolytes have been investigated [ 48 - 50 ], but few simultaneously meet the selection criteria. To be used as a n electrolyte in a Li - ion batter y , a solid - state electrolyte should fulfill the following criteria [51 ] : 1) >0.2 mS cm - 1 at room temperature, 2) negligible electronic conductivity, 3) a wide potential window, 4) stability in air, 5) stability against Li , and 6) low grain boundary resistance. In this respect, Thangadurai et al. [ 52 ] reported that garnet - typ e compounds with chemical formulas of Li 5 La 3 M 2 O 12 (M=Ta, Nb) are promising candidates as a solid state electrolyte. These materials have high ion ic conductivity of 0.04 mS cm - 1 and a wide potential window of >6 V vs Li/Li + [ 52 ]. In addition to Ta and Nb, Murugan et al. [ 53 ] discovered a higher conductivity formulation Li 7 La 3 Zr 2 O 12 (LLZO). The LLZO structure is composed of ZrO 6 octahedra and LaO 8 dodecahedra forming a rigid framework with Li located 16 in two types of site: 24d sites in tetrahedral and 96h si tes in distorted octahedral (Figure 1 - 5). It is reported that LLZO has two polymorphs: cubic and tetragonal , where the former has a higher conductivity compared to the later [ 54 ]. In general, LLZO forms a tetragonal crystal structure with an ordered Li n etwork at room temperature [ 55 ]. However, this crystal structure of LLZO can be changed by adding super valent dopants [ 54 ]. The dopants introduce some disordered Li arrangement by expelling Li - ions, and the disordered Li leads to a change in the crystal structure from tetragonal to cubic [ 54 ]. This cubic structure has approximately two orders of magnitude higher ionic conductivity (0.4 - 1 mS cm - 1 ) compared to the tetragonal structure of LLZO (0.16 x 10 - 2 mS cm - 1 ) [ 54 ]. Since the ionic conductivity can be improved by optimizing the lattice parameter, which affects the energy barriers for Li - ion transport, and Li vacancies concentration, the effect of dopants on ionic conductivities of LLZO ha s been intensively studied [ 56 ]. Another important role of a LLZ O ceramic electrolyte is to act as a physical barrier between electrodes. To utilize LLZO as an electrolyte and a separator simultaneously, the mechanical properties of LLZO are important. However, there are few studies that characterize the mechanical p roperties of LLZO ceramic electrolyte. 17 Figure 1 - 5: The crystal structure of cubic LLZO. 1.4 Challenges for electric vehicle systems As discussed in previous sections, Li - ion batteries are appealing owing to their high energy and power compared to other typ es of batteries. However, using Li - ion batteries in transportation such as hybrid (HEV), plug - in hybrid electric (PHEV), and battery electric vehicles (BEV) requires higher energy density, faster charging and discharging, longer cycle life, and improved s afety [1, 9 - 12 ]. 18 1.4.1 Power and energy The United States Advanced Battery Council (USABC) proposed specific requirements for high performance batteries (Table 1 - 1)[ 5 7 ]. A particular focus is on achieving high power density (600 W L - 1 ) without compromising ener gy density. Energy density represents how much energy can be stored in a given mass and/ or volume whereas power density indicates how fast the energy can be released from/to electrodes. Graphite electrodes are typically used in commercial Li - ion batterie s as a negative electrode due to its long cycle life, low discharge potential, low cost, and moderate theoretical specific capacity (372 mA g - 1 )[ 8 - 9 , 19 - 20 ]. To improve energy density, while using the same electrodes and electrolyte, the challenge turns to the peripheral mass in a batter y pack; in other words, minimize peripheral mass and volume to maximize performance. Electrodes are composed of electro - active storage materials (such as graphite and LiCoO 2 particles), a metal current collector, and a bind er (Figure 1 - 2)[ 5 8 ]. The current collectors offer the homogeneous distribution of electrons into and out of the electrode active materials as well as mechanical support. In general, Cu foil is used for a negative electrode due to its stability against Li at low potentials, and Al foil is used for a positive electrode owing to its low cost, low density, and stability results from the passive Al 2 O 3 layer [ 59 ]. A polymeric binder is used to help adhesion between active particles and current collector, and i t can also help to maintain mechanical stability during charg e and discharg e process es [ 60 ]. In general, the total specific energy (Wh kg - 1 ) of a Li - ion battery can be presented by the following equation [61 ] : Total cell (mAh g - 1 ) = ( Eq. 1 - 2) 19 where C A and C C are the theoretical specific capacities of the negative and positive electrodes , respectively, and 1/Q M is the specific mass of other battery components such as current collectors, separator, and e lectrolyte in g mAh 1 . Therefore, to increase energy storage per unit mass, the weight fraction of active mass ( negative and positive electrode materials ) should be maximized and the peripheral mass such as a metal foil current collector, an electrolyte , and a separator should be minimized. One approach is to design thick and low porosity electrodes, to minimize the mass fraction of metal foil current collectors and electrolyte, respectively. Based on the calculations to determine the individual battery component (active and inactive) mass fractions, Figure 1 - 6 shows that the specific energy is increased by increasing the active electrode mass/loading (thick electrode) and decreasing porosity (dense electrode). However, it is generally known that dischar ge and charge rates are inversely related to thicker and denser electrodes (Figure 1 - 7)[ 1 9,62 ]. In addition, less porosity in the electrode also reduced Li - ion transport resulting in low charge and discharge rates [ 1 9 ]. 20 Table 1 - 1: USABC targe t of obtaining a high - energy storage and low - cost electric vehicle battery applications [ 5 7 ] . Depth of discharge (DOD), State of charge (SOC). Parameter (Units) of fully burdened system Minimum goals for long term commercialization Long term goal Power d ensity (W L - 1 ) 460 600 Energy density (Wh L - 1 ) 230 (C/3 discharge rate) 300 (C/3 discharge rate) Cycle life (Cycles) 1,000 (80% DOD) 1,000 (80% DOD) Operating environment ( ° C ) - 40 to 50 (20% performance loss) - 40 to 85 Normal recharge time (h) 6 3 to 6 High rate charge (min) 20 - 70% SOC in <30 min at 150 W kg - 1 ) 40 - 80% SOC in 15 min Total pack size (kWh) 40 40 21 Figure 1 - 6: Correlation between specific energy, electrode loading, and open porosity assuming LiMO 2 and no packaging. Figure 1 - 7: Dependence of the power capability of SFG44 graphite electrodes in the electrode loading [19] . 22 1.4.2 Safety In addition to performance, the safety of Li - ion batteries is also important f or vehicle applications [ 44 - 46,63 - 64 ]. There are two phenomena that can result in Li - ion battery fires. First, improper charging (too fast at < - 20 °C) of a Li - ion battery could result in metallic Li deposition on the negative electrode causing Li dendrit es to grow and short - circuit to the positive electrode causing instantaneous discharge [ 4 4 - 45 ]. The instantaneous discharge results in rapid Joule heating to the point that the organic solvent - based electrolyte ignite s , causing combustion [1, 44,46,64 ]. S econd, the penetration of Li - ion batteries by a metallic object can cause short - circuiting resulting in a phenomenon similar to when Li dendrites cause short - circuiting. The Li - a nail is driven through a Li - ion battery to cause short circuiting [ 65 ]. The nail penetration test is considered to simulate an internal short - circuit in a cell, and this test is important to demonstrate short - circuit is caused by a battery itself or ot her aspects like a manufacturing defects [ 6 5 ]. During charging, Li - ions diffuse through electrolyte - filled pores in a porous electrode. Since the Li diffusion rate is not uniform and relatively slow in the porous electrodes , the distribution of Li - ions is not uniform under severe charge and/or discharge conditions. This mechanism causes high Li - ion concentration gradients within the electrode s , which is called as concentration polarization, and results in Li dendrite formation and growth. This phenomenon occurs more frequently at fast charge rates due to higher concentration polarization [ 66 ]. In addition, a thick and dense electrode (high energy density) exacerbates concentration polarization because of the longer and more tortuous Li - ion diffusion path s [ 67 ]. Therefore, concentration polarization can 23 be mitigated by reducing the diffusion path length. Further, Li dendrite penetration into a positive electrode has to be prevented for safety. 1.5 Need for cell design 1.5.1 Electrode design approaches Achieving h igh power without sacrificing high energy density in Li - ion batteries is most important for electric vehicle applications. Conventionally, microstructural features of an electrode , such as porosity and electrode thickness , were considered as important fac tors to minimize the cell - level power density limitations [ 19 ]. By improving electrode microstructural properties, advances in the Li - ion batteries have been made [ 14 ]. Nevertheless, technological challenges still remain such as relatively low power dens ity and safety issues. In general, it was believed that the low power performances are caused by material limitations and slow kinetics [ 9 , 14 ]. In recent years, however, it has been demonstrated that significant power losses, which arise from slow transp ort of ions, can be improved by cell design and electrode architecture [ 19, 68 - 70 ]. To facilitate Li - ion transport in electrodes, Sakamoto et al. [ 68 ] designed V 2 O 5 electrodes with hierarchical ordered pores. Due to the highly order ed pores, they obtained high rate performance compared to random porosity electrodes. Zhang et al .[ 69 ] also suggested novel electrode architectures by electroplating the positive electrode (MnO 2 ) materials in an opal - like porous nickel framework. This group has demonstrated the high power performance of electrodes can be achieved by mesopores network (76 % reversible capacity at 185 C - rate in ~30 nm thick electrode). Bae et al .[ 70 ] developed improved kinetics in the positive electrode (LiCoO 2 ) by providing homogeneous linear ch - scale 24 porosity electrode was created by a co - extrusion process. Through the increased thickness and by creating linear channels, they obtained both high energy density and power. Although previous work s on electrode architecture demonstrated that the high performance of Li - ion batteries can be achieved by reducing internal resistances [ 6 8 - 70 ], those techniques have not been adopted by the commercial Li - ion battery industry owing to their complex and expensive manufac turing processes that limit scale - ability. In addition, those methods could not precisely control the porosity on electrodes to provide uniform patterning, therefore, uniform current density over relatively large electrode areas. In other words, it is de sired that simple and fast patterning process on high energy density (thick) electrode and the ability to control precise pattern ed porosity for electric vehicle applications. Consequently, a laser patterning process was developed in this research. This technique not only enables a fast and simple patterning process on thick electrodes (high energy), but precisely controls the engineered porosity ( uniform current density over a patterned electrode ). The laser patterning technique and optimizing electrode design will be discussed in C hapter 4, and the electrochemical performance of patterned electrode will be compared with conventional electrodes in C hapter 5. 1.5.2 Cell design with garnet - like solid electrolyte As discussed previously, the safety issue is one of the most important requirements for electric vehicle applications. Although the electrode design approach can suppress the metallic Li deposition on the negative electrode by mitigating concentration polarization, it is not the ultimate solution to pre vent cell explosion caused by Li dendrite growth. Hence, a new hybrid cell design is proposed in this study (Figure 1 - 8). Garnet - like ceramic electrolyte may be able to 25 provide both a physical barrier and Li - ion transport paths between electrodes [ 71 ]. Consequently, it is important to characterize and optimize the mechanical properties of the ceramic electrolyte membrane to effectively suppress Li dendrite growth on electrodes during fast charge and/or discharge processes . Since mechanical properties of ceramics are highly sensitive to microstructure s , the mechanical properties of LLZO can be optimized by controlling features such as the grain size and relative density. Therefore, the mechanical properties of LLZO ceramic electrolyte were characterized in C hapter 6. Figure 1 - 8: Schematic of hybrid cell design composed of graphite negative electrode, LiCoO 2 positive electrode , liquid electrolyte, and LLZO ceramic electrolyte where between electrodes. 26 1.6 Scope of present work A s discussed in s ection 1.2. 2.1, a graphite electrode is one of the most widely used negative electrode materials due to its outstanding properties [ 9 , 19 - 20 ]. However, the low practical power remains a major disadvantage. It is known that the power capability of graphite electrodes is affected by microstructur al properties such as its influence on the thickness, the porosity, the tortuosity, and the electronic conductivity of the graphite electrode network [ 19 ]. Hence, this study focuses on determining and understanding the main ra te limiting process of a graphite electrode as well as the Li dendrite formation and growth related to safety concerns in Li - ion batteries. Based on the understanding the mechanisms, the power (rate capability) and safety of graphite electrode can be impr oved using cell design approaches. Furthermore, novel hybrid cell designs including a garnet - like ceramic electrolyte are proposed by optimizing the microstructure of LLZO ceramic electrolyte based on characterization of mechanical properties. The ultima te goal of this study is to enable safe and high performance batteries for electric vehicle applications (Figure 1 - 9). 27 Figure 1 - 9: Ragone plot for various energy devices [72 ]. 28 2 Experimental methodology 2.1 Electrode preparation In this study , graphite electrodes (TIMREX SFG6: TIMCAL, Bodio, Swizerland) were prepared using a standard tape casting technique ( - ) . The SFG6 graphite was mixed with 10 wt.% polyvinylidene difluoride (PVdF, Alfa Aesar, Johnson Matthey GmbH) binder and N - met hyl - 2 - p yrrolidone solvent (NMP, Alfa Aesar, Johnson Matthey GmbH). The mixture was ball - milled us ing a planetary ball mill (PM 100, Retsch, Germany) to make a homogeneous slurry. A n 80 ml a gate jar with 6 agate balls (10 mm diameter, Retsch, Germany) was used. The ball - milling was c onducted for 20 min using 350 rpm. The resulting graphite electrode ick copper foil (MTI Corporation, USA) using a doctor blade (MTI Corporation, USA), travellin g at a 24 mm s - 1 . The graphite electrode loading (thickness or areal capacity in mAh cm - 2 ) was c ontrolled by changing doctor blade hei ght. Three areal loadings were studied: 1.15, 4.0, and 5.5 mAh cm - 2 . The cast electrode sheets were dried using a 250 W infrared light bulb (Philips, Neth erlands) approximately 25 cm from the electrode sheet overnight to evaporate solvent and residu al wa ter. The desired porosity was controlled by calendaring (Mini F100, Durston, UK) (Figure 2 - 1) . 29 - 2.2 Ceramic electrolyte processing 2.2.1 Powder preparation - - - 30 2.2.2 Consolidation The LLZO powders were hot - pressed at 1050 ° C under a constant pressure of 62 MPa using rapid induction hot - pressing (RIHP) in flowing argon , as described previously (Figure 2 - 2) [5 6 ,7 3 ] . The hot - pressing temperature of 1050 o C was selected to maximize density based on previous work [7 4 ] . The relative density was cont rolled by changing the hot - pressing time (30, 60, 90, and 240 min). The cooling rate was about 6 ° C per min. After hot - pressing, each pellet was mounted in Crystalbond® wax and sliced into 3 discs using a diamond saw. The discs were ground on sand paper ( Black ice ; Norton Corporation , USA) from 400 to 1500 grit then polished using 1 and 0.5 m diamond paste (Diamond polishing compound; LECO Corporation , USA) on a polishing pad ( White technotron; LECO Corporation , USA) with a lapping fixture (Model 900; Southbay Technologies, San Clemente, CA) and an oil - based lubricant (Micro diamond compound extender; LECO Corporation , USA) . The discs were stored in a glove box (<1 ppm O 2 , <1 ppm H 2 O) before testing to minimize surface contamination [ 75 - 76 ] . 31 - image of the rapid induction hot - pressing. (b) Schematic of the cross - section of a graphite die with LLZO power for hot - press. 32 2.3 Electrochemical measurement methods 2.3.1 Galvanostatic rate mapping test - - - - ® - perfluoroalkoxy, Swagelok®, USA - The Swagelok® - type cell boreholes were machined with a 12.827 mm diameter reamer to allow for the insertion of 12.70 mm diameter s. alf - cell was constructed consisting of a 3/8 inch graphite electrode as the working electrode, 25 µ m thick Celgard® 2400 (Celgard®, USA) separator, 0.75 mm thick Li counter electrode (Alfa Aesar, Johnson Matthey GmbH), and 1.0 M LiPF 6 in ethylene carbonate/pr opylene carbonate/ethyl - methyl carbonate (EC/PC/EMC, 2:1:7, in vol.%) electrolyte (Soulbrain, Korea) (Figure 2 - 3b) . - - - 33 - - - - To assure th e Li surface was clean, Li ribbon was scraped with a stainless steel tool in the glovebox before cell assembly. To assure electrolyte permeation, the Celgard® 2400 separators were soaked in the electrolyte overnight. To assure consistent cell impedance, 45 N of force was applied to the cell, confirmed by a force gage (PS100, Imada, USA). The cell fabrication was conducted in an argon - filled glove box (<1 ppm O 2 , <1 ppm H 2 O) to prevent moisture and air contamination. 34 - 35 - 36 was conducted using a potentiostat (VMP300, Bio - Logic, Knoxville , TN ) at room temperature. All the potential values reported are hereafter referenced to Li/Li + . Prior to the rate map ping test, preconditioning cycles were conducted to form a stable SEI layer . The precondi tioning cycles were conducted for 3 cycles at 1/5 C - rate for 1.15 mA cm - 2 and 1/10 C - rate for 4 and 5.5 mAh cm - 2 loadings of electrodes , respectively. In addition, s ince the SEI typically forms between the as - assembled open circuit voltage (~3) and 0.24 V, the first intercalation was conducted at 1/20 C - rate until 0.24 V . The cycling potential window was fixed between 0 and 0.75 V. - After preconditioning, intercalation was carried ou t at 1/5, 1/3, 1/2, 1, 2, 3, 5, and 10 C - rates. To limit kinetic e ffects during Li extraction, de intercalation was conducted at 1/5 C - rate after each intercalation half cycle. To confirm that capacity decline with increasing intercalation C - rate was not a result of irreversible capacity loss, the last de/ intercalation cycles were conducted at a relatively slow C - rate ( 1/5 C - rate ). Four identical cells were cycled for each electrode loading . 2.3.2 I ntentional overcharge test - 37 - - - - , Knoxville , TN 2.3.3 Galvanostatic intermittent titration technique (GITT) - - - - - - 38 - - - - - - 39 - - , Knoxville , TN - 2.3.3.1 Polarization interrupt test - - - - - 40 - - - 2.3.4 Electroche mical impedance spectroscopy (EIS) - - 41 - - - - - - - - - - - - - 42 - 43 - 44 - - - - - 2.3.4.1 Transmission line model (TLM) and e lectrochemical impedance spectroscopy using symmetric electrode (EIS - SC) - - - - - - - - 45 - - - - - - - 46 2.4 Mechanical property characterization 2.4.1 Vickers hardness - - 47 - - E is the H v is the Vickers hardness, c is the crack length, a is the length of half diagonal, and P is the applied load .) - - - - he H v was determined using a Vickers hardness tester (Vickers/ K noop hardness tester; Mitutoyo Corporation, Japan, HM122 V/K series 810 micro). Before indentation, the Vickers hardness tester was calibrated using a steel hardness block (Vickers hardness test block; Mitutoyo Corporation, Japan, Hardness Test Block HMV 700HV). The Vickers hardness tests were conducted at a load of 0.294 N for an indentation time of 10 s. This load was chosen to minimize micro cracking. For each hot - pressed sample, 10 hardness measurements at 150 m spacing were performed. The H v was determined by Eq . 2 - 8 . The fracture toughness ( K IC ) was estimated using the indentation technique (Vickers indenter). For the samples with relative densities above 95 % the applied load was 0.686 N. For the sample w ith the lowest relative density of 85 % in this study , the load was increased to 4.9 N to produce measureable crack leng t hs. In all cases , the dwell time was 10 s. 10 indents were made per sample. Crack lengths were determined from SEM images. - Eq. 2 - 11 [ 90 ] The H value is from the H v . For both the hardness and K IC , the 48 indentation diagonal and crack length measurements were made immediately after testing to minimize the reaction with ambient air. The radius of the surface crack length was about 2 - 3 times larger than the half diagonal for all the measurements. 2.4.2 Nano indentation - - - - - 49 - The nano hardness ( H n ) E ) of the hot - pressed discs w ere determined using a nanoindenter (Nanoindenter; MTS system Corporation, G200) with a Berkovich three - sided pyramidal diamond tip employing a 20 nm radius. The test parameters were 0.05 s - 1 strain rate, 2 nm harmonic displacement, and a frequenc y of 45 Hz. The maximum depth limit was 1 . The average maximum load value used for a total of four samples was 124.9 ± 8.3 µN. 10 hardness measurements were taken for each relative density . was calculated from the load - displacement curve during unloading using the Oliver - Pharr method [ 93 ]. Fused silica (Corning 7980; MTS system Corporation, USA) was used as the standard reference material to calibrate the instrument. 2.5 Other methods 2.5.1 Laser patterning The laser patterning appar atus was custom fabricated by combining a computer numerical control (CNC, Mach3, Newfangled solutions, USA) system, laser light source (IPG Corporation, USA), and optics. A 5 W green fiber laser (532 nm, 1 nm pulse length) focused from 5 mm down to a 20 cm - 2 ) was chosen for electrode ablation. The laser patterning conditions were optimized for each electrode conditions such as loading and porosity. The power and frequency ranged between 80 and 87 % and between 80 and 87 k Hz, 50 respectively. A close - packed hexagonal array of channels was made by synchronizing the CNC with the laser beam. - 2.5.2 X - ray power diffraction (XRD) - - 51 - - 2.5.3 Scanning electron microscope (SEM) - - - - 2.5.4 Raman 52 - - - he patterned electrodes were characterized by Raman spectroscopy (LabRAM, Horiba Scien tific, Japan) to determine if laser ablation changed the graphite chemistry. Raman spectroscopy was conducted on the patterned electrodes using a 532 nm wave length green laser and 2400 lines per mm holographic grating to identify the phase characteristic s. 53 3 Determining power limiting process and understanding cell failure mechanisms - - - 3.1 Intercalation vs deintercalation rate - - - - - - - - - 54 - - - - - - - - - - - - 55 3.2 Rate limiting process es - - - - - - - - - - - - - - - - - 56 - - - 57 3.2.1 Electron injection and extraction resistances - 3.2.2 Ion insertion and extraction resistances at the interface between electrode and electrolyte 3.2.2.1 SEI and charge transport resistances - - - - - 58 - - - - - - - - - - - 59 - - - - - - - - - - 60 - - - - 61 3.2.2.2 Tortuosity - - - - - - - - - 62 - - - - - - - - - - - - - - - 63 - - 64 - - - - 65 - - - - - - - 66 3.2.2.3 Solid - state diffusion in graphite electrode - - - - - - - - - - - - - - - - - - - - - - - - - - 67 - 3.3 Understanding cell failure mechanisms - - - - 68 - - - - - - - - - --- --- 69 - - - - - - - Fig ure 3 - 7: A schematic of thermal runaway causes fires by improper charging in Li - ion batteries. (a) normal state batt ery, (b) Li dendrite formation due to improper charge such as fast charging, (c) short - circuiting by Li dendrite growth and short circuit on the positive electrode causing instantaneous discharge, (d) cell temperature goes up (>70 ° C) by Joule heating and electrolyte start to decompose, then flammable hydrocarbon gases are released, (e) Joule heating and exothermic reactions further increase temperature, and the metal oxide positive electrode starts to decompose (>1 5 0 ° C), then releasing oxygen. These step s can cause cell failure and explosion. (Cell swelling f igure [ 6 5 ]). 70 - - - - - - - - - - - 71 - - 72 3.4 Summary - - - - - - 73 4 Laser patterned ele ctrodes In principle the current Li - ion batteries can achieve higher specific energy than what is found in state - of - the - art technology [118 ]. The general approach to achieve higher specific energy, while using the same electrode materials, is to maximize electrochemical active mass per unit mass of the cell and battery pack. One specific approach is to increase the mass of electrode material (graphite and metal oxide positive electrodes) per unit area, thus reducing the relative contribution of the metal foil current collectors. The negative electrode, in particular, can benefit significantly from this approach since Cu foil (density = 8.96 g cm - 3 ) is used as the metal foil current collector. While this approach is viable in theory, increasing the loadin g per unit area result in thicker electrodes that hinder rate performance. In other words, from a thermodynamic perspective, Li - ions batteries can deliver higher specific energy, but the approach to employ thicker electrodes is not practical due to kineti c limitations. The goal of this work is to engineer thick electrodes to decouple the relationship between thermodynamics and kinetics, thus enabling higher specific energy Li - ion batteries. As will be discussed, the same approach can also suppress Li den drite deposition, which arises as a result of kinetic limitations during high rate charging. 4.1 Background: three dimensional (3D) electrode designs The electrode kinetics and mass transport are the most important factors in the performance of electrochemica l energy storage technology. The simplest solution for these limitations can be to reduce the electron and ion diffusion distances during the charge and discharge process. The 74 electron diffusion length can be reduced by reducing the particle size. This can also reduce solid - state Li - ion diffusion distance in the same manner. Regarding ionic transport in the electrolyte, the ion diffusion distance is related to the open porosity and pore size distribution. For example, although electrodes have the same total open porosity, if relatively small pores are present , - ion transport in the electrode. Therefore, achieving uniform Li - ion transpor t in the pores of the electrode requires control over al l the pore size, morphology and orientation. One strategy is to design and manufacture 3D electrode architectures. In 2002, Sakamoto et al. [6 8 ] fabricated a hierarchically ordered electrode with the i nverted opal structure (Figure 4 - 1a). They used a se lf - assembled templating method with V 2 O 5 as the positive electrode . Since this structure allows homogeneous Li - ion transport , they observed improved performance compared to conventional V 2 O 5 electrode. Zhang et al. [6 9 ] constructed 3D MnO 2 positive electr ode by electrodeposition on Ni foam (Figure 4 - 1b). The Ni foam was prepared by electrodeposition on self - assembled opal template from polystyrene spheres. Prior to electrochemical active material plating on the Ni foam, the porosity of the resulting Ni f oam was increased using electro - polishing technique to prevent pore closing by deposition of active materials. Subsequently, a MnO 2 positive electrode with 150 - 200 nm thick was obtained, and its capacity remained 60 % at 62 C - rate due to its uniform pores . Ji et al .[ 119 ] also investigated 3D LiFePO 4 positive electrodes using a template - based technique. Since graphite is not only lightweight compared with metals but also has acceptable electrical conductivity, this group proposed to use a hollow graphite as a current collector for positive electrode instead of metals to reduce mass portion of inactive components in electrodes. The graphite foam current collector was fabricated by depositing graphite on Ni foam followed by the removal of Ni foam as a 75 selec tive etching technique . LiFePO 4 was then deposited on the hollow graphite foam. These positive electrodes showed high power capabilities delivering capacities of 158 mAh g - 1 , 70 mAh g - 1 , and 36 mAh g - 1 at current densities of 15 mA g - 1 , 1280 mA g - 1 , and 2560 mA g - 1 , respectively. The theoretical capacity of LiFePO 4 is 170 mAh g - 1 . While these are examples that demonstrate the efficacy of 3D structure electrodes, template - based techniques not only require repeating complex energy consuming processes, but it is also difficult to scale - up. Another example of engineered electrodes involved 3D printing [ 120 ] to fabricate 3D electrodes for Zn - Ag alkaline micro - batteries. These electrodes consisted of pillar - like structures that were of - 1c). This approach achieved a ~60 % increase in areal capacity compared to conventional planar b atteries. While these achievements are promising, at present it is not known if 3D printed electrodes can be scaled to meet the capacity and cost constraints for electric vehicle batteries. Bae et al .[ 70 ], designed LiCoO 2 electrodes with periodic linea r channels made by co - extrusion (Figure 4 - 1d). A feedrod was used and composed of a mixture of LiCoO 2 powder and polymeric binder (Figure 4 - 1d). The mixture was consolidated with carbon mandril or rod that would eventually be removed through oxidation to create linear channels. After co - extrusion, the resulting fibers were assembled into arrays, followed by a binder and graphite burnout to form linear channels by heat treatment (Figure 4 - 2 positive electrodes were made with pores in ~ 5 diameter range (Figure 4 - 1d). Since the linear channels permeated the thickness of the electrode, to some degree, the Li - ion diffusion path length was decreased. As a result, the relatively thick LiCoO 2 electrode exhibited ~2 times higher specif ic capacity at 1 C - rate compared to that of state - of - the - art LiCoO 2 electrodes. Despite this improvement in power 76 capability, the co - extrusion technique using sacrificial graphite porogen may not be amenable to large - scale fabrication (Figure 4 - 1d). Essen tially, the aforementioned examples of 3D electrode architectures have demonstrated that the slow kinetics of Li - ion batteries can be improved through electrode design and engineering. However, not all of these examples meet processing criteria such as lo w cost, rapid fabrication, precise pore position control, and scale up. 77 (a) Figure 4 - 1: Previously reported 3D architecture electrode designs and fabrication methods. (a) Process for fabricating the hierarchical V 2 O 5 electrode [6 8 ] . (b) O ut line of the Ni foam fabrication by template based method. Lower image is MnO 2 electrode fabricated by electrodeposition on Ni foam [6 9 ]. (c) A schematic of 3D image of pillars by Super ink jet printing [1 20 ]. (d) Outline of the electrode fabrication proce ss. Left lower shows the surface of a patterned electrode and right lower shows cross - section of a patterned electrode [ 70 ]. 78 Figure 4 - 1 (cont d) . (b) (c) 79 Figure 4 - 1 (cont d) (d) 4.2 Highly ordered hierarchical (HOH) graphite electrode I n the present work, a highly - ordered and hierarchical (HOH) graphite electrode is proposed to achieve high specific energy and power density . Based on our experimental results, the intercalation process was intensively focused. 80 4.2.1 Laser patterning technique Keepi ng in mind the desired electrode design and scale - able processing, laser patterning investigated as a technique to fabricate HOH electrodes. The basic principle is to introduce cylindrically - shaped and through - thickness pores by laser ablation. The laser ablation provides several advantages: 1) Laser patterning is a post calendaring process, thus the laser ablated porosity will not be affected by further electrode fabrication; 2) Laser patterning enables precise control over the cylindrical pore position and geometry; 3) L aser patterning is a non - toxic process. Unlike previous template - based techniques, no chemicals required for laser patterning. C O 2 is the product of graphite ablation/oxidation; 4) Because state - of - the - art lasers can achieve adequate in tensity, the ability to penetrate several hundred micron s thick graphite electrodes is expected; 5) The laser patterning process can be fast if commercial galvo - focusing heads or equivalent are employed. It takes a few seconds to make a pattern on electro de by laser. In addition, solid - state lasers typically do not require extensive maintenance. 4.2.2 HOH electrode design As discussed in Chapter 3, the internal resistance of electrolyte - filled pores inside a porous electrode can be a rate limiting process in a thick electrode (Figure 4 - 2a). Thus, the electrode design must facilitate Li - ion transport through the electrode thickness to provide a homogeneous ionic current, especially at high charge rates. In this regard, the HOH electrode was designed. The HO H electrode is the electrode consisting of an array of highly ordered or close - packed linear channels that direct transport of Li - ions into smaller intrinsic pores ( Figure 4 - 2b ). First, 81 macro - scale linear channels are produced in a conventional graphite e lectrode (Figure 4 - 2b) to allow for uniform Li - ion transport. Second, close - packed hexagonal patterning was employed to improve enable a uniform Li - ion current through the entire electrode (Figure 4 - 2b). Since macro - scale linear channels can facilitate L i - ions diffusion, Li - ions will transport through the larger and linear channels first then diffuse from the walls of the channels to the intrinsic pores between active particles. Thus, the spacing between channels should be minimized to reduce the diffusi on length in micro - scale intrinsic pores. The close - packed hexagonal patterning provides not only the shortest spacing between channels, but the same spacing between channels, which contributes homogeneous Li - ions distribution in a porous electrode by off ering the same diffusion length. Though it seems that the lateral distance between linear channels should be minimized, the cumulative electrode porosity should be less than ~50 % to maximize energy density. Thus, the optimum engineered porosity should b e determined to maximize energy and power density simultaneously. In addition, the patterning process should be able to make sufficiently small featu channels, assuming the total volume fraction of patterned porosity is fixed. However, the linear channel size is typically proportional to the laser power. Therefore, it is required to optimize a HOH patterning conditions for the electrode with high energy and power density. In the next section, the HOH fabrication conditions were optimized. 82 (a) (b) Figure 4 - 2: Schematic representation of possible Li - ion diffusion paths (a) in a convention al porous electrode , (b) in a HOH electrode , and schematic of top view of HOH electrode and short Li - ion diffusion length induced by hexagonal close - packed li near channels. 4.2.3 HOH electrode design optimization To demonstrate the feasibility of the laser patterning process on thick graphite electrodes, the introduction of 10 % laser patterned porosity was attempted on conventional graphite electrodes consisting of 4 mAh cm - 2 83 to Zheng et al . [ 121 ], relationships between areal capacity and maximum C - rate with variable - ion selected for laser patterning. The HOH electrode was successfully fabricated as intended (Figure 4 - 3a). It indicates that the laser technique enables pre cise patterning on a thick electrode. The minimum linear channel size was approximately ~53 ± 1.2 µ m (Figure 4 - 3a), but in the initial stages of this investigation, the holes were clearly tapered (Figure 4 - 3b). Since the laser is designed to focus from 5 mm down to 20 m diameter , the nature of the laser patterning causes conical shaped holes resulting in approximately 60 % less ablated porosity compared to if perfect cylinders were made. In addition, there was evidence of channels collapse likely due to inadequate sp acing (Figure 4 - 3c). This indicated that the electrode walls between channels could be collapsing during the laser patterning process when the laser patterned porosity exceeds a critical volume fraction . The maximum laser patterned porosity in the graphi te electrode with 4 mAh cm - 2 and 40 % intrinsic porosity was approximately 9 %. It is assumed that overlapping heat affected zones caused the channels collapse. Thus, total volume fraction of linear channels was reduced to 5 % for the thicker 5.5 mAh cm - 2 To meet the accurate target volume fraction of linear channels in a HOH electrode, the total number of channels was increased by about 6 0 % to compensate for the effect of conical shaped channels ( i.e. conical: 1.1 x 10 5 m 3 , cylindrical 2.9 x 10 5 m 3 in HOH electrode with 5.5 mAh cm - 2 loading and total 50 % open porosity) . In HOH electrodes , the sum of the Li - ion transport resistance consists of the transport resistance in the laser ablated linear patterned channel s (macro - scale ; M ) and in the intrinsic pores (micro - scale; m ). Macro - scale laser ablated linear channels ideally reduce the total mass transport 84 resistance by offering wide, shorter and line - of - site Li - ion diffusion path s . However, incorporating macro - scale linear channels into graphite electrodes causes the lower intrinsic micro - scale open porosity ( total = M + m ) compared to the average intrinsic open porosity of a conventional graphite electrode ( total = m ) of the same total open porosity. Thu s the cumulative open porosity ( total ) was fixed at 50% in all cases to minimize the impedance associated with Li - ion mass transport caused by the intrinsic porosity. It was determined that the effects of intrinsic porosity are negligible between 40 % an d 50 %. The rate capability of the electrodes with 5.5 mAh cm - 2 showed similar values at various C - rates, regardless of the open porosity between 40 % and 50 %. After the laser patterning, a precise circular shape electrode was obtained by laser cutting without a mechanical damage generated by cutting tools (Fig ure 4 - 3d). 85 (a) (b) Figure 4 - 3: Secondary SEM images of laser patterned electrode (Timcal, SFG6, 4.0 mAh cm - 2 , 50 % porosity) (a) top view of fabricated HOH electrode , (b) cross - sectio n of a conical shaped pattern, (c) c ollapsed walls between laser - ablated channels , and (d) a laser cut HOH electrode after laser patterning (3/8 inch diameter) . 86 Figure 4 - 3 (cont d) . (c) (d) 87 4.2.4 HOH electrode characterization 4.2.4.1 Phase characterization by Raman spectroscopy Since laser patterning produces intense heat, it was possible that the graphite could have been affected. Thus graphite in an HOH electrode was characterized using Raman spectroscopy. Raman spectroscopy was conducted from spot 1 to 4 (Fig ure 4 - 4 ) on surface and on fractured HOH graphite electrode surface. Although the Raman spectrum of the cross - section showed a weak peak at ~1340 cm - 1 , all Raman spectrum peaks of the HOH electrode are consistent with the graphite in non - laser ablat ed electrodes. In addition, there is no observable secondary phase peaks. The resulting Raman spectrum is in agreement with typical Raman spectrum of graphite [ 9 5 ], and it indicates that laser patterning does not cause any phase changes in a graphite ele ctrode. 88 Fig ure 4 - 4: Raman spot analysis of an HOH graphite electrode at various spots (1 to 4 and cross - section) . 89 4.2.4.2 Morphological analysis After rate mapping tests, SEM morphological analysis of HOH electrodes was conducted to compare HOH electrodes before and after rate mapping (Fig ure 4 - 5 ). The purpose of the test was to determine if the mechanical integrity of the HOH electrode was compromised by introducing the laser ablated channels . Since there is volume change in the graphite electrode that occurs during charge and discharge processes , it is possible that the laser ablated channels could cause particle erosion . For the SEM analysis, t he cycled HOH electrode s were rinsed in dimethyl carbonate (DMC) to remove the Li sa lt (LiPF 6 ) that precipitate on the surface thereby covering the electrode topography. Based on the SEM analysis (Fig ure 4 - 5 ), there was no observable morphological change before and after rate mapping. This indicated that HOH electrodes maintained their integrity during cycling. I t is in good agreement with the results of Raman spectroscopy analysis 90 (a) (b) Figure 4 - 5: SEM images of laser patterned graphite electrode (SFG6 graphite electrode with 5.5 mAh cm - 2 and 50 % total open porosity) . (a) T op view of HOH electrode before and (b) after rate mapping. 91 4.3 Summary In this Chapter, the novel HOH graphite electrode design and fabrication were presented, which include linear laser ablated channels in closed - packed hexagonal arrays, to improve the rate capability in thick graphite electrodes. Previous electrode architecture approaches have demonstrated that controlling the electrode microstructure can enhance the rate capability of Li - ion batteries. However, it is still a challenge to use these techni ques for practical cell manufacturing due to their complexity and high production cost. Therefore, the laser patterning technique was employed due to its ease of integration with state - of - the - art production and potentially low production cost. The thick linear - macro channels was successfully obtained by laser patterning. The minimum pattern pore - ablated porosity was fixed as 5 % to avoid the effects of heat affected z one by laser beam. The Raman spectroscopy and SEM analysis have proved that there is no phase change and mechanical degradation during rate mapping tests after laser patterning. To date, this is the first report of the homogeneous patterning of an electr ode with a - to - roll process. 92 5 Electrochemical characterization of HOH electrodes 5.1 Solid - state Li diffusivity in graphite electrode The rate capability of Li - ion batteries can be significantly influenced by the diffusivity of Li inside active materials. Therefore, solid - state Li diffusivity ( D Li ) has to be accurately measured to determine the rate limiting step. There are different electrochemical techniques for measuring D Li such as the galvanostatic intermittent titration technique (GITT)[7 9 ], potential intermitte nt titration technique (PITT)[122 ], electrochemical impedance spectroscopy (EIS)[12 3 ], and cyclic voltammetry (CV)[ 124 ]. However, in the literature, t here is discrepancy of D Li values for the same materials. For example, orders of magnitude differences in D Li values were reported in the previous study by Shen et al .[1 25 ]. This inconsistency can be attributed to the different electrode preparation and construction such as phase, porosity, size, and shape of electrochemical active materials. It is well known that the solid - state diffusion rate is strongly dependent on the SOC, which causes phase transitions during charging and discharging. Figure 5 - 1 s hows the potential of the graphite electrode with a 1.2 mAh cm - 2 loading and 63 % porosity as function of x in Li x C 6 . A sequence of constant potential plateaus is clearly observed in the potential vs composition plot. This phenomenon is called staging wh ereby a constant potential plateau indicates that two distinct Li x C 6 phases are in thermodynamic equilibrium. According to Gibbs phase rule, when two phases are in equilibrium, there are two degrees of freedom necessitating a constant potential, assuming pressure and temperature are fixed [15]. A stage refers to Li occupying a specific stacking configuration. For example, in stage 4, Li - ions are intercalated in every forth basal layer in a stack of basal planes found in graphite. Subsequently, Li - ions a re intercalated in every third and second basal plane in stages 3 and 2, respectively. When a graphite electrode is fully charged (x=1 in Li x C 6 ), each graphite layer is filled with Li. 93 This staging is a thermodynamic phenomenon, thus distinct voltages ar e associated with each stage transition. Moreover, the solid - state Li - ions diffusion rate is affected by the SOC. Originally, GITT and PITT methods were developed for dense planar electrodes such as single crystal highly - oriented pyrolytic graphite (HOPG ) [ 15 ]. However, conventional graphite electrodes are porous and composed of an assembly of graphite particles bound together by a polymer binder. Thus, the actual interface area is dependent on the electrode preparation and construction such as materials (size and shape), inactive components (binder), and porosity. Figure 5 - 1: Typical potential vs x in Li x C 6 plot with 1.2 mAh cm - 2 and 63 % SFG6 graphite electrode. 94 In the present work, the GITT technique was used to determine D Li of the graphite electro des. For GITT, the conventional graphite electrodes with 1.2 mAh cm - 2 and 63 % were fabricated. SFG6 graphite ( TIMCAL, Bodio, Swizerland) was used for all experiments. The surface area of SFG6 particles is reported as approximately 17.1 m 2 g - 1 . However , as was discussed, the surface area can be reduced by the electrode conditions. Since the conventional graphite electrode was composed of 90 % SFG6 and 10 % PVdF binder, BET analysis was conducted to obtain the actual surface area of the same composition and open porosity (63 %) of the graphite electrode. As a result, the measured surface area was 6.9 m 2 g - 1 . This value is approximately 60 % lower than the sum of the surface area of particles. This difference can result from occluded porosity between p articles resulting from the presence of the PVdF binder. Additional experimental details are described in Chapter 2. Figure 5 - 2 presents the GITT plot of the graphite electrode with 1.2 mAh cm - 2 and 63 % porosity. The measured potential range was 80 mV t o 0.75 V and the staging phenomena are clearly observable (Figure 5 - 2). The D Li values of the graphite electrode were calculated by Eq. 2 - 2 based on the GITT plot. Essentially, the D Li values increasing as th e SOC decreased, as expected [111 ]. The D Li v alues were 1.8 x 10 - 8 cm 2 s - 1 at SOC 60 %, 3.8 x 10 - 8 cm 2 s - 1 at SOC 40 %, and 1.1 x 10 - 7 cm 2 s - 1 at SOC 20 %, respectively. These values are consistent with the D Li range (10 - 7.5 - 10 - 9.5 cm 2 s - 1 ) of artificial graphite electrode, which is calculated by Ta kami et al. [111 ]. 95 Figure 5 - 2: The GITT plot of the graphite electrode with 1.2 mAh cm - 2 and 63 % porosity. The measured potential range was 80 mV to 0.75 V. 5.2 Rate mapping 5.2.1 Effects of loading To understand the effects of electrode loading on rate performa nce, rate mapping was conducted with three different loadings (1.15 mAh cm - 2 , 4 mAh cm - 2 , and 5.5 mAh cm - 2 with 50 % open porosity) of conventional graphite electrodes as a function of intercalation rate from 1/5 to 10 C - rate after conditioning cycles (Fig ure 5 - 3) . From Fig ure 5 - 3 , several important points are noted. First, SFG6 graphite electrodes, the active material mainly used in this study, satisfies the high reversible capacity and stable cycleability demands in Li - ion batteries. In general, graphi te electrodes undergo irreversible capacity loss during de/ intercalation cycles due to side reactions 96 such as a SEI formation by a reductive decomposition reaction of the electrolyte composed of organic solvents and Li salt [ 43 ]. Thus, it is important for active materials to have low irreversible capacity loss for the higher practical reversible capacity. In this regard, 1.15 mAh cm - 2 graphite electrode showed high reversible specific capacity of 353±8 mAh g - 1 (theoretical capacity (372 mAh g - 1 )) at 1/5 C - rate, and there was no observable irreversible capacity loss during rate mapping up to 10 C - rate (Fig ure 5). This behavior is in good agreement with previous work [ 19 ]. It indicates that a stable SEI layer is formed during the preconditioning protocol, thus preventing additional irreversible capacity loss by inhibiting further . In addition, the capacity retention of the electrodes reached approximately 99±1 % of the capacity in the first cycle capacity at the same 1/5 C - rate af ter extreme cycles regardless of the loadings . It confirms that the performance of SFG6 graphite electrode is not degenerated by drastic cycling conditions and the capacity diminishes with increasing intercalation rate is not a result of irreversible capa city loss. Second, the capacity retention decrease s as a function of increasing electrode loading. This correlation between loading and capacity retention is comparable to previous studies [ 1 9,62 ]. The 4 and 5.5 mAh cm - 2 graphite electrodes achieved app roximately 30 % and 70 % lower capacity retention a t 1/3 C - rate, and 60 % and 87 % lower capacity retention at 1/2 C - rate compared to those of 1.15 mAh cm - 2 . Th is behavior can be attributed to IR polarization potential. 5.5 mAh cm - 2 electrode s require ap proximately 5 times higher current density compared to that of 1.15 mAh cm - 2 electrode s to charge at the same C - rate. Since, consequently, the IR resistance is proportional to current density, the 5.5 mAh cm - 2 electrode undergoes about 5 times higher IR p olarization which can cause premature potential cut - off. To support these assumptions, rate mapping was conducted with a high loading of 5.5 mAh cm - 2 electrode at slow intercalation rate (1/10 C - rate) and, as expected, showed reasonable 97 specific capacity (336 mAh g - 1 ) (Figure 5 - 3) . Third , the specific capacity retention diminished as increase the intercalation rate regardless of the electrode loadings (Figure 5 - 3) . It can be attributed to the polarization in the same manner with the correlation between ca pacity retentions and loadings of electrodes because the current related polarization potential drop increases as a function of the intercalation rate. Figure 5 - 3: Results of rate mapping as a function of graphite electrodes with various loading from 1.1 5 mAh cm - 2 to 5.5 mAh cm - 2 with the same total open porosity (50 %) . N=4. 5.2.2 HOH graphite electrode vs c onventional graphite electrode The rate performance of conventional electrodes and HOH electrodes were compared at various C - rates (Figure 5 - 4) . The per cent capacity retention vs intercalation C - rate is shown in Fig ure 5 - 4ab . The percent capacity is calculated assuming the capacity at relatively low rate (1/5 C - rate) is 100 %. Both types of electrode s were prepared with 1.15 mAh cm - 2 , 4.0 mAh cm - 2 , and 5.5 98 mAh cm - 2 loadings , respectively, and their total open porosity was fixed as 50 % as discussed in Chapter 4 . Unlike conventional electrodes, HOH electrodes consist of 5 % laser ablated linear porosity and 45 % intrinsic open porosity. It is seen that the average capacity retention as a function of increasing rate for each types of electrode is similar up to the 4 mAh cm - 2 loading (Fig ure 5 - 4a ). On the other hand, the HOH electrodes, which have a 5.5 mAh cm - 2 loading , exhibited 65 % and 120 % higher ca pability in percentage compared to those of a conventional graphite electrode at 1/3 C and 1/2 C - rate , respectively (Figure 5 - 4b ). In addition, it is interesting to note that the specific capacity retention also improved despite the fact that laser ablate d channels decreasing the amount of electrochemical active material per unit area (capacity)(Figure 5 - 4c). Typically, i t is believed that the relatively high loading (thick) electrode ha s a more tortuous and longer Li - ion transport paths compared to those of the lower loading electrodes. The longer and more tortuous Li - ion diffusion paths cause more likely concentration polarization. This can result in higher local IR resistance in electrolyte - filled pores inside a porous electrode by the non - uniform cur rent density inside of a porous electrode [6 8 ] . Since only the 5.5 mAh cm - 2 loading HOH electrodes showed significant improvement in rate capabilities compared to conventional electrodes, the local resistance of electrolyte - filled pores inside a porous el ectrodes seem s to rapidly increase between 4.0 mAh cm - 2 and 5.5 mAh cm - 2 electrode loadings with 50 % open porosity. T h ese result s indicate that mass transport resistance , which results in concentration polarization, inside the porous electrode can be dominant when the electrode is sufficie the HOH electrodes can reduce the mass transport resistance. As predicted, the uniformly patterned macro - scale linear channels can provide improved Li - ion diffusion paths through the linear channels and reduce d diffusion distance in micro - scale intrinsic pores. The improved mass 99 transport properties enable more uniform Li - ion distribution at high C - rate, which leads higher capacity retention by suppressing concentration polarization. Concentration polarizatio n leads to significant cell polarization potential drop and also causing Li deposition. Therefore, the design of linear channels can mitigate safety concerns by suppressing the Li deposition possibility. Additionally , there are no advantages observed in HOH electrodes at high C - rates (>1 C - rate) regardless of electrode loadings . This can be attributed to solid - state diffusivity limitations, which leads to particle scale concentration polarization resulting in a rapid reduction in cell potential (Figure 5 - 5) [12 6 ] . Since a number of Li - ions simultaneously intercalated into a graphite particle in short time scale under fast charge rate, the Li concentration at the edge of the particle is alike SOC 100 % even the center of the particle is empty (concentratio n polarization) . As a result, the capacity retention significantly drops because the concentration polarization leads to reach the premature cell cut - off potential before each particle is saturated by Li [1 2 6 ]. Consequently, the solid - state diffusivity c an dominate the rate capability above 1 C - rate instead of the resistance of Li - ion diffusion through electrolyte - filled pores inside a porous electrode ( tortuosity ) . 100 (a) Fig ure 5 - 4: Charge rate mapping as a function of SFG6 graphite electrodes w ith conventional and HOH electrode s with 50 % total open porosity . (a) C apacity (%) vs intercalation rate with 4 mAh cm - 2 , (b) with 5.5 mAh cm - 2 , and (c) specific capacity (mAh g - 1 ) vs intercalation rate with 5.5 mAh cm - 2 . N=4. 101 Figure 5 - 4 (cont d) . ( b) (c) 102 Figure 5 - 5: Schematic diagram showing the Li concentration and diffusivity profiles in a graphite electrode [126 ]. 103 5.2.2.1 Effects of separators The main role of the separator is to prevent physical contact of the electrodes while providing a n ionic transport path and preventing electronic transport. Ideally, separators should have sufficient porosity (>40 %) to allow for the facile diffusion of ionic species, while the pore size conducting additives penetration [ 12 7 ]. Furthermore, in the event of Li dendrite formation and exfoliation , the separator should block the exfoliated Li (also referred to as Li moss) transport , which can cause cell failure. In addition, the separator sho uld be thin enough to minimize the diffusion distance between electrodes. Therefore, the structure of a porous separator is typically highly tortuous to meet the required properties. This tortuous structure of separator can induce high IR resistance at h igh current densities to the degree that power density is affected. To demonstrate the effects of separator on rate capability of electrodes , a highly porous separator (Zeus®) was employed to compare with the most common separator; Celgard ® 2400. In the p resent work, Celgard ® 2400 was used in the most experiments. Celgard ® 2400 consists of analysis (Figure 5 - porosity was >50 %. The average pore size is ~400 nm based on SEM observation (Figure 5 - 6b). It was believed that the different separator structural characteristics would affect cycling behavior, especially at high C - rate. However, the cells employing Zeus ® separators frequently failed during cycling tests, especi ally at high current densities. The Figure 5 - 7 shows the typical conditioning cycles with Zeus ® separators and with different electrode loadings, which were 1.2 mAh cm - 2 and 5.5 mAh cm - 2 , respectively. The highly porous Zeus ® separator works well with 104 1. 2 mAh cm - 2 electrode while it causes cell failure during the conditioning cycles with the 5.5 mAh cm - 2 electrodes, especially when current density was increased (>0.139 mA cm - 2 ). This may indicate that the Zeus ® separator could not block the Li moss when the current density was higher than ~0.139 mA cm - 2 . Consequently, it might be true that an alternative separator with superior mechanical properties and high ionic conductivity, such as a ceramic electrolyte, is ultimately necessary to endure high current flow for long cycle life. The ceramic electrolyte, LLZO, is discussed in Chapter 6. 105 (a) (b) Figure 5 - 6: SEM images of (a) Celgard 2400 ® and (b) Zeus ® separators . 106 (a) (b) Figure 5 - 7: Typical pre conditioning cycles with Zeus ® separat ors and with different graphite electrode loadings, which were (a) 1.2 mAh cm - 2 and (b) 5.5 mAh cm - 2 , respectively 107 5.3 Cell impedance characterization 5.3.1 Polarization interrupt test To characterize the effects of tortuosity on graphite electrodes , the polarizati on interrupt tests [ 80 ] was conducted with conventional graphite electrodes (5.5 mAh cm - 2 , 50 % intrinsic open porosity) and HOH graphite electrodes (5.5 mAh cm - 2 , 45 % intrinsic open porosity and 5 % laser ablated porosity). The polarization interrupt te sts were carried out on a free standing both conventional and HOH electrodes between 2 layers of Celgard® 2400 using a symmetric cell (Figure 5 - 8) . The concentration gradient was produced by applying a constant current. Subsequently, the current was stop ped, and the potential change was recorded. This potential change is produced by a redistribution of Li + and PF 6 - through the porous electrodes and separators . The results of both cell s with 5.5 mAh cm - 2 loading showed a linear slope line after stopping the applied current, and the potential slope (~ - 1.57 x 10 - 4 V s - 1 ) of HOH electrode was lower than that of the conventional electrode (~ - 1.24 x 10 - 4 V s - 1 )(Fig ure 5 - 8 ). Since the slope depends on the transportation of ions back to the equilibrium state vi a porous medium , these data indicate that the internal resistance of the HOH electrode is lower than conventional electrode. 108 Figure 5 - 8: Galvanostatic polarization, followed by interrupt and relaxation test (HOH vs Conventional electrode with 5.5 m Ah cm - 2 and 50 %), and the schematic of symmetric cells for polarization interrupt [ 80 ] . 5.3.2 Transmission l ine m ethod (TLM) and EIS characterization Ogihara et al .[ 8 3 ] estimated the resistance of Li - ion inside the porous electrodes using the TLM based EIS - sym metric cell (SC) technique. Based on TLM model for cylindrical pores, the overall impedance is expressed in Eq. 2 - 6 for a non - faradaic and Eq. 2 - 7 for faradaic process [8 3 ] . When each numerical parameter is provided, the Nyquist plots exhibit linear beha vior in the high frequency range, regardless of faradaic and non - faradaic models (Figure 5 - 9). Thus, the 109 linear slope region can be interpreted as a mass transport resistance without the effects of charge transfer resistance. Mass transport resistance va lues can be calculated based on the TLM model. According to the non - farad a ic ( Eq. 2 - 6 ) and faradaic ( Eq . 2 - 7 ) TLM model s , the impedance of the real part ( Z re ) and impedance of the imaginary parts ( Z im ) can be shown with the following relationships when go to 0 in a non - faradaic process ( Eq. 5 - 1 and 5 - 2), and in faradaic process ( Eq. 5 - 3 and 5 - 4) [ 7 9,83 ]. - - where R ion is resistance of electrolyte - filled pores in porous e lectrode (mobility of Li - ion), C dl is total electric double layer capacitance. - - where R ct is charge transfer resistance. 110 Figure 5 - 9: Simulated Nyquist plots for a cylindr ical pore in an electrode with different models. (a) Non - faradaic, (b) faradaic with low charge transfer resistance, and (c) is faradaic with high charge transfer resistance [ 83 ]. 5.3.2.1 Reliability of TLM based EIS - SC technique To demonstrate the reliability of TLM based EIS - SC method, EIS test was conducted with using electrodes with various loadings (Figure 5 - 10). The symmetric cell was fabricated with SOC 0 % electrodes after pre conditioning cycles to form a stable SEI layer (Figure 5 - 10a). The two differen t loading electrodes were selected to compare the effects of loading. The loadings for graphite electrodes were 1.2 mAh cm - 2 and 5.5 mAh cm - 2 , respectively , and the open porosity was fixed as 50 % . As seen in Figure 5 - 10bc, both SC with different loading s show a linearly 111 sloped region at the higher frequencies and a linear tale at lower frequencies. This behavior is consistent with non - faradaic model of Ogihara et al [ 8 3 ]. Although graphite is a non - blocking electrode, the charge transfer reaction was no t observed in the frequency range used. Therefore, the ionic resistance in porous graphite electrode can be interpreted by the non - faradaic model. The SC with 1.2 mAh cm - 2 electrode had less resistance by ~65 % compared to 5.5 mAh cm - 2 of the same porosi ty (Figure 5 - 10bc) . Since the 5.5 mAh cm - 2 electrode consists of longer and more tortuous Li - ion diffusion paths, the internal resistance should be higher than that of the thin electrode s (1.2 mAh cm - 2 ). In addition, it should be pointed out that their d ifferent bulk resistances in the high frequency region. The bulk resistances are mainly from the electrolyte - filled porous separator. Therefore, the bulk resistances ideally have the same values due to use the same separator and electrolyte. However, th is resistance also includes peripheral componentry such as the cables and stainless steel 304 electrode s , which can account for the 1 ohm variance. 112 (a) (b) Figure 5 - 10: (a) Schematic representation symmetric cell (SC)[ 83 ], Nyquist plots after TLM - EIS - SC test s with (b) 1.2 mAh cm - 2 and 50 % and (c) 5.5 mAh cm - 2 and 50 % SFG6 symmetric cells. 113 Figure 5 - 10 (cont d) . (c) Figure 5 - 11 shows the temperature dependence of the Nyquist plots of fully delithiated conventional graphite electrodes (SOC 0 %) with 5.5 mAh cm - 2 and 50 %. The temperature range was from RT to 55 °C. In all temperature ranges, the Nyquist plots showed the same behavior. Figure 5 - 11 clearly shows the decreasing trend for both the bulk resistance ( R b ) and the resistance of the ele ctrolyte - filled pores ( R ion ) as a function of temperature increase. Since the Li - ion mobility is proportional to the temperature increase, this trend is well matched with predicted behavior. Based on the loading and temperature dependence tests, EIS - SC b ased on the TLM technique can be adopted to measure the resistance of electrolyte - filled pores. 114 Figure 5 - 11: Nyquist plots for symmetric cells with two graphite electrodes at SOC 0 %. T he loading of 5.5 mAh cm - 2 and porosity of 50 % c onventional graphi te electrode s were used. 5.3.2.2 Comparison the internal resistance of HOH electrode s vs c onventional electrode s Figure 5 - 12 shows Nyquist plots of both HOH and conventional porous electrodes with 5.5 mAh cm - 2 and 50 % using the EIS - SC technique. The HOH electro de also exhibits a linearly sloped region at the high frequencies. The imaginary impedance increases at low frequency and the plot is nearly a vertical line. T he tale at low frequencies is ideally a vertical line , which indicates electrical blocking beha vior but they showed a low angle slope (Figure 5 - 11) . The slope of this region can be attributed to leakage current effect. Since the HOH electrodes also exhibited non - faradaic behavior, the Li - ions transport resistance in porous electrode can be estimat ed by Eq. 5 - 1. The resulting resistance showed ~35 % lower resistance compared to conventional electrodes 115 with the same loading and open porosity. This trend is consistent with the rate mapping and polarization interrupt tests. Figure 5 - 12: Nyquist p lots after TLM - EIS - SC test with HOH symmetric cell (5.5 mAh cm - 2 and 45 + 5 %). 116 5.4 Summary In C hapter 5, the solid - state Li diffusivity of SFG6 grade electrode was characterized by GITT. The measured Li diffusivity range was consistent with previous work (10 - 7.5 - 10 - 9.5 cm 2 s - 1 )[ 1 11 ]. In addition, the surface area of fabricated electrode showed ~60 % lower values compared to that of particles due to pore occlusion likely from the polymer binder and calendaring. The rate mapping tests were conducted to m easure the effects of loading. As was expected, high loading electrode (5.5 mAh cm - 2 ) showed ~ 55 % lower capacity compared to that of low loading electrode (1.15 mAh cm - 2 ) a t 1/3 C - rate . It can be attributed to different current density requirements depe ndent on different loadings at the same C - rate. The rate capability of an HOH electrode was compared with that of a conventional electrode. Based on rate mapping tests with different loading electrodes, the internal resistance of electrolyte - filled pores seems to be significantly increased between 4 mAh cm - 2 and 5.5 mAh cm - 2 . When the electrode loading was 5.5 mAh cm - 2 , an HOH electrode , which has 45 % intrinsic open porosity and 5 % laser ablated open porosity, showed 65 % and 120 % higher rate capabilit y compared to the conventional electrodes of the same loading and porosity at 1/3 C and 1/2 C - rate, respectively. The polarization interrupt test and TLM - EIS - SC tests demonstrated that HOH electrodes have lower internal resistance compared to conventional electrodes. However, the capacities retention of both types of electrodes approaches zero at high C - rate (>1 C - rate). Also, it was shown that solid - state Li diffusivity causes concentration polarization, which leads premature potential cut - off when the rate is >1 C - rate . In conclusion, HOH electrodes can improve the rate capability and ameliorate safety concerns by suppressing concentration 117 polarization. However, when the C - rate is beyond a critical level (~1 C - rate ), the relatively slow solid - state Li diffusion rate can dominate the internal resistance. 118 6 The effect of microstructure on the mechanical properties of hot - pressed cubic Li 7 La 3 Zr 2 O 12 As a result of increased demands for higher energy and safety the novel hybrid cell design incl uding LLZO ceramic electrolyte was proposed in this study . To be used in this situation the solid electrolyte must meet several important requirements [ 4 7 ]. These include: 1 ) high ionic conductivity with low electronic conductivity, 2 ) chemical stability against the Li negative electrode and positive electrode and 3 ) good mechanical properties. Of the possible Li - ion conducting solid electrolytes cubic , Li 7 La 3 Zr 2 O 12 (LLZO) is a potential candidate as a result of its high ionic conductivity ( 10 - 4 to 10 - 3 S cm - 1 [ 5 4 ]) and stability with Li [ 5 2 ]. There have been many investigations on the ionic conductivity of LLZO [ 50 - 52 ]. The effects of microstructure (e.g., porosity and grain size) on the ionic conductivity of LLZO are well documented [12 8 ] . In contr ast, there have been very few investigations focusing on the mechanical properties of LLZO and none reported on the effects of microstructure on mechanical behavior [12 9 ] . As consequence , it is the purpose of this Chapter to investigate and relate the mec hanical properties such as hardness and fracture toughness of hot - pressed Al - substituted LLZO to the microstructure. In addition, the ionic conductivity will be measured. The correlation between ionic conductivity, hardness, fracture toughness and micros tructure (porosity and grain size) will be reported. This information is needed if cubic LLZO is to be used as a Li - ion conducting electrolyte in a solid - state battery and/or hybrid cell design which was proposed in this study . 119 6.1 LLZO ceramic electrolyte characterization 6.1.1 Density of LLZO From Table 6 - 1, it is observed that as the hot - pressing time increased, the relative density increased, reaching the maximum value of 98 % at 240 min. In addition , it can be seen that a minimum hot - pressing time of about 60 min is needed to achieve relative densities above 95 %, where the porosity typically transitions from open to closed [1 30 - 1 32 ] . Table 6 - 1: It presents the information of hot - pressed LLZO pellets as changing hot - pressing time. Hot - press time (min) 30 60 90 240 Relative density (%) 85 95 96 98 Grain size ( m) 2.7 ± 1. 68 3. 2 ±1.87 3.5 ±1.83 3.7 ±1.84 6.1.2 Phase characterization The X - ray diffraction patterns of the LLZO calcined powder, the hot - pressed LLZO as a function of hot - pressing time , and the reference patte rn for cubic LLZO are shown in Fig ure 6 - 1. A comparison of the X - ray diffraction patterns for the calcined and hot - pressed samples with that for the reference pattern suggests that the pellets were predominantly cubic LLZO with no observable second phases except a small amount ( 0.5 wt.%) of pyrochlore (La 2 Zr 2 O 7 ) that was present only in the sample hot - pressed for 60 min. However, closer inspection of Fig ure 6 - 1 reveals that the triple peaks between 50 - 53° two - theta of the calcined powder and the hot - pres sed sample for 30 min show a slightly right - skewed shape. According to previous studies, a skewed peak shape can result from the presence of some tetragonal phase due to a Li content above that 120 needed to form the pure cubic phase [ 5 6 ]. It is likely that presence of the tetragonal phase in these calcined and sample hot - pressed for 30 min samples is a due to the incomplete evaporation of the 10 wt.% excess Li precursor that was intentionally added to compensate for Li loss during high temperature processing . At longer hot - pressing times when the excess Li has evaporated only the pure cubic phase is exhibited (Fig ure 6 - 1). Figure 6 - 1: X - ray diffraction patterns of Li 6.19 Al 0.27 La 3 Zr 2 O 12 calcined powder and hot - pressed pellets pressed for 30, 60, 90, and 240 min at 1050 o C. * Pyrochlore (La 2 Zr 2 O 7 ) [133] 121 6.1.3 Micro structure of LLZO Fracture surfaces of the hot - pressed LLZO samples as a function of relative density are shown in Fig ure 6 - 2. From Fig ure 6 - 2 , several points are noted. First, in agreement with d ensity measurements , it is seen that the relative density increased (porosity decreased) with increasing hot - pressing time. Second, the dominant fracture mode changed from inter to intragranular with increased relative density . I t can be seen that the 85 % relative density (hot - pressed for 30 min ) sample exhibited almost 100 % intergranular fracture (Fig ure 6 - 2 a ) whereas intragranular fracture was the primary fracture mode at above 95 % relative density ( hot - press ed f or 60 min, Fig ure 6 - 2 b ). At the highe st relative density of 98 % (hot - pressed for 240 min, Fig ure 6 - 2d) , the fracture mode was almost entirely intragranular. 122 Figure 6 - 2: Fracture surface of Li 6.19 Al 0.27 La 3 Zr 2 O 12 hot - pressed for : (a) 30 min , (b) 60 min , (c) 90 min , and (d) 240 min . The re lative densities are indicated in top right of each image [133] . 123 Microstructures of the thermally etched hot - pressed LLZO samples as a function of relative density are shown in Fig ure 6 - 3. From Fig ure 6 - 3 , several points are noted. First, the majority of the porosity is located at the grain boundaries and decreases with increasing pressing time, which is in agreement with the density measurements and fracture surface micrographs (Fig ure 6 - 2). Second, the average grain size, determined using the average length between the major axis and minor axis is listed in Table 6 - 1 . From Table 6 - 1 , it is seen that the grain size increased with increasing pressing time (increased density). Because each hot - pressed sample contained a ure 6 - 3 ) , a large standard deviation resulted (Table 6 - 1 ). Consequently, t he grain size distribution was determined using Eq. 6 - 1 and plotted in Fig ure 6 - 4 for as a function of relative density ( hot - pressing times ) . - where GSD is the grain size distribution ( i = 0 - 2, j = 2 - 4, and k = 4 - n is the number of grains in a grain size ra nge. From Fig ure 6 - 4 , it is observed that the fraction of 0 - 0. 42 to 0. 17, and the fraction of 4 - 0. 16 to 0. 36 for the 85 % and 98 % relative density samples, respectively . This result is in good agre ement with the average grain size measurements (Table 6 - 1 ), confirming that the grain size increased with increasing relative density (or hot - pressing time ) . The average grain size after hot - pressing ( ~ 2 - 4 µm) is much smaller than those typically observed in conventionally sintered LLZO, which exhibit grain sizes in the range of ~ 20 - 200 µm [ 12 8 ] . 124 (a) (b) Figure 6 - 3: Li 6.19 Al 0.27 La 3 Zr 2 O 12 hot - pressed pellets after thermal etching at 700 o C for 30 min in air. The Li 6.19 Al 0.27 La 3 Zr 2 O 12 pellets were hot - pressed at 1050 o C for: (a) 30 min, (b) 60 min, (c) 90 min, and (d) 240 min. The relative densities are indicated in top right of each image [133] . 125 Figure 6 - 3 (cont d) . (c) (d) 126 Figure 6 - 4: Grain size distributions of hot - pressed Li 6.19 Al 0.27 La 3 Zr 2 O 12 [133] . One important aspect of the microstructural characterization is that a change in fracture mode from inter to intragranular with increasing hot - pressing time was observed. This could be associated with microstructural variables such as the grain size and/or porosity at the grain boundaries or the grain boundary composition/cohesion. In general, an increase in grain size and/or decrease in porosity at the grain boundaries can lead to an increase in the percentage of intragranular fracture [1 3 4 - 1 3 6 ] . W hile a difference in grain size is observed, the maximum disparity is ~25%; the average grain size was 2.7 and 3.7 for the 85 % vs 98 % relative density samples, respectively. Thus, we do not believe the difference in grain size is responsible for the dramatic transition from inter to intragranular fracture when comparing the 85 % and > 95 % relative dens ity samples. In a first approximation , we believe that the pores could act as stress 127 intensifiers, thus initiating cracks at the grain boundaries. Because the 85 % relative density sample had the highest fraction of intergranular porosity, intergranular fracture was the primary fracture mode. At 98 % relative density where the volume fraction of porosity was the lowest, intragranular fracture was dominant. It is also possible that the grain boundary composition/strength could increase with increasing ho t - pressing time, though this cannot be verified at this time. In summary , it was observed that hot - pressing times of at least 60 min is required to obtain high density (>95 %) LLZO with relatively strong grain boundaries compared to the 85 % relative dens ity sample. 6.2 Mechanical properties of LLZO 6.2.1 Hardness of LLZO The hardness of hot - pressed LLZO as a function of relative density is shown in Fig ure 6 - 5. The Vikcers hardness ( H v ) is shown by the open symbols while the nanoindentation hardness ( H n ) is shown by the closed symbols. From Fig ure 6 - 5 , several important points are noted. First, both the H v and H n values increase with increasing relative density and gradually level off at high relative densities. For the nanoindentation, the increase in H n with relative density is not as pronounced compared to the increase in H v for the reasons explained below. Second, at the lowest relative density of 85%, the H v is 4. 7 ± 0.2 GPa, which is about half the value for the H n of 8. 1 ± 0.8 GPa at 95 % relative densit y. This difference diminishes ( H v is 7.4 ± 0.4 GPa vs 9. 3 ± 0.5 GPa for H n ) above a relative density of 96 % for both the H v and H n , which are nearly equal at 9.1 GPa. 128 The results of Fi gure 6 - 5 can be explained by the difference in microstructural variab les (porosity and grain size) and measurement techniques ( H v and H n ). In general hardness can be affected by porosity and grain size. Typically hardness decreases with increasing porosity and increasing grain size [ 13 7 - 13 9 ]. Since the average grain size was nearly the same ( within 25 % ) among all relative densities, we believe the main microstructural variable that influences the hardness is the porosity. In nanoindentation , the indent impression size was ~1 µm or less. This is smaller than the average grain size ( ~ 2 - 4 µm), thus each valid nanoindentation measurement was essentially in a LLZO single crystal. However , the 85 % relative density LLZO exhibited a lower average H n compared to the > 9 5 % relative density samples. T his could result from nanoi ndentations in the proximity of pores. Since the 85 % relative density LLZO consisted of more porosity than the other relative densities, it was more likely that there were more nanoindentations in the proximity of pores, which lowered the average hardnes s value. The large standard deviation in the H n (± 0.8 GPa) among all the relative densities measured for the 85 % samples most likely results from a non - uniform pore distribution within this sample. For the case of H v , the indent impression size was betw een 10 - This is bigger than the average grain size, thus the H v can be affected by the intergranular porosity. It is expected that as the porosity decreases, the H v should increase. According to the data in Fig ure s 6 - 3 and 6 - 5, the H v indeed incre ases as the relative density increases. Furthermore, additional proof that the H v is influenced by the intergranular porosity is the increase in H v with increasing density follows the change in fracture mode from inter to intragranular (Fig ure 6 - 2). From Fig ure 6 - 2 , it is observed that at low relative density, the fracture mode is intergranular, implying relatively weak grain boundaries that decrease the hardness compared to the higher relative density samples. Additionally, as the relative density incre ases the fracture mode changes to 129 intra granular implying relatively stronger grain boundaries exhibiting relatively higher hardness compared to the 85 % relative density sample. At the highest relative density (98 %), where the fracture mode is almost ent irely intragranular, the H v should equal the H n value since the effects of porosity and grain size are negligible. From Fig ure 6 - 5 , it can be observed that indeed both values are about equal ( 9. 1 GPa ). Figure 6 - 5: H v and H n of Li 6.19 Al 0.27 La 3 Zr 2 O 12 as a function of relative density [133] . It was shown by Sirdeshmukh et al . [ 1 40 ] that the H v (measured on the (111) face) of several oxide based garnet single crystals can be correlated with lattice parameter. It was observed that H v decreased linearl y with an increase in lattice parameter. Sirdeshmukh et al . [ 1 40 ] suggested 130 that a smaller lattice parameter results in stronger interatomic binding and hence, higher hardness. The H v values vs lattice parameter for the single crystal oxide based garnets from Sirdeshmukh et al . [ 1 40 ] are plotted in Fig ure 6 - 6. Also Fig ure 6 - 6 includes the H v for the highest relative density (98 %) LLZO. It should be noted that from Fig ure 6 - 5 that this value is equivalent to the H n . The lattice parameter for this materia l determined by Rietveld refinement is 12.964 Å. From Fig ure 6 - 6 , it is observed that the measured H v of LLZO is in good agreement with the predicted value. These results suggest that the correlation between hardness and lattice parameter is similar to o ther garnets, thus the single crystal hardness of LLZO was estimated to be 9.1 GPa. Fig ure 6 - 6: H v vs lattice parameter for single crystalline garnets from the literature (open squares)[1 40 ] and the value for Li 6.19 Al 0.27 La 3 Zr 2 O 12 from this work (close d square) [133] . 131 6.2.2 Fracture toughness of LLZO The K IC of hot - pressed LLZO as a function of relative density is shown in Fig ure 6 - 7. From Eq. 2 - 11 [ 90 ] , it is seen that a value of E is required to determine K IC . It was observed that E varied from 135 GPa for the 85% relative density sample to The E values for the higher relative density samples are in agreement with the experimental value of E 150 GPa determined using resonant ultrasonic spectroscopy for the 97 % relat ive density Li 6.24 Al 0.24 La 3 Zr 2 O 11.98 sample [12 9 ] . From Fig ure 6 - 7 , it can see that the K IC decreased with increasing relative density. The K IC values are 2.37 ± 0.1 for the 85 % and 98 % relative density samples , respectively . These values are within the range typically exhibited by polycrystalline ceramics 2 - [ 12 9 , 1 4 1 ] . The predicted K IC value for the 97 % relative density sample was 1. 11 ure 6 - 7) . This value is in good agreement with the K IC value of 1 relative density (97 %), similar composition ( Li 6.24 Al 0.24 La 3 Zr 2 O 11.98 ) , but slightly larger grain size of 5 µm [1 4 2 ] . The decrease in K IC with increasing density could be a result of the change in grain size and/or the amount of porosity at the grain boundaries. It has been observed that the K IC is independent of grain size for cubic oxides over the grain size range investigated in this study [1 4 3 - 14 4 ] . Thus, the difference in grain size of LLZO cannot explain the dec rease in K IC with increasing density. 132 Fig ure 6 - 7 : Fracture toughness of Li 6.19 Al 0.27 La 3 Zr 2 O 12 as a function of relative density [133] . Examination of the crack propagation path from the corner of the Vickers indents (Fig ure 6 - 8 ) in the 85 % and 98 % relative density samples clearly shows different fracture modes. For the sample with the 85 % relative density , the crack propagation path is primarily intergranular, whereas for the sample with 98 % relative density , the crack propagation path is mainly intragranular . Similar trends were observed on the fracture surfaces shown in Fig ure 6 - 2. At 85 % relative density, the fracture mode was intergranular whereas at 98 % relative density, it was predominately intragranular . We believe, the relatively high volume fraction of intergranular porosity (Fig ure 6 - 3) in the 85 % relative density sample can explain why the primary fracture mode is intergranular. Typically, intergranular porosity is correlated with relatively weak grain boundaries [13 5 ] . Thus, the weak grain boundaries deflect the cracks out of the plane of 133 maximum driving force and hence, require more energy to propagate compared to the samples exhibiting strong grain boundaries where intragranular fracture was observed [ 1 4 5 ]. (a) (b) Fig ure 6 - 8 : T he Vickers indentation crack propagation path trajectories for (a) relative density of 85 % and (b) relative density of 98 %. Arrows point to crack the propagation path in each grain [133] . From the K IC values for LLZO , the fracture surface energy ( ) can be determined using Eq. 6 - 2 [ 13 5 , 13 7 - 13 9 ] : - Using values of K IC 0.97 and E 140 GPa into Eq. 6 - 2 yields a 3 J m - 2 . This value is in very good agreement with values of 0.5 to 3 J m - 2 commonly exhibited by single crystal ceramics [1 3 5 ] . This resu lt suggests that the K IC values for high relative density ( 98 %) LLZO 134 sample measured in this study is likely approaching the single crystal K IC values. Using values of K IC 2.37 and E 135 GPa for LLZO with 85% relative density in Eq. 6 - 2 yields a 21 J m - 2 . This value is within the range typically exhibited by polycrystalline ceramics (10 - 50 J m - 2 ) [ 13 5 ] . 6.3 Ionic conductivity of LLZO The logarithm of total ionic conductivity of hot - pressed LLZO as a function of relative density is shown in Fig ure 6 - 9 . From Fig ure 6 - 9, it is observed that the total ionic conductivity increases with increasing relative density. This trend is typically observed in LLZO and is usually associated with a decrease in the grain boundary resistance [ 14 6 ] . The decrease i n the grain boundary resistance component could be a result of a change in the nature of the grain boundary as suggested by David et al [7 3 ]. At 85 % relative density , the total ionic conductivity is 0.0094 mS cm - 1 and increases to 0.3 4 mS cm - 1 for the 98 % relative density sample . The value of 0.3 4 mS cm - 1 is in good agreement with the upper values of total conductivity for Al - substituted LLZO of similar composition (0.02 to 0.5 mS cm - 1 ) [7 3 , 14 7 ] . Ex trapolation of the curve in Fig ure 6 - 9 yields a total io nic conductivity of 0.4 mS cm - 1 , which is in excellent agreement with previously reported bulk conductivity values [ 56 ] . 135 Figure 6 - 9 : Total ionic conductivity of Li 6.19 Al 0.27 La 3 Zr 2 O 12 as a function of relative density [133] . One of the important resul ts of this study is the opposite trend in K IC (decrease) and total ionic conductivity (increase) with increasing relative density. This is likely a result of the nature of the grain boundaries that vary with relative density. At low relative density, wea k grain boundaries result as evidenced by intergranular fracture whereas, at high relative density strong grain boundaries result as evidenced by intragranular fracture. It is suggested that that the nature of the grain boundaries is mainly controlled by the volume fraction of intergranular porosity, which is known to correlate with grain boundary strength [ 13 5 ] . These results suggest that if high ionic conductivity LLZO is the goal, a sacrifice in K IC will occur. One possible solution to this dilemma is to engineer a toughening mechanism that acts within the grains yet, leaves strong grain boundaries to yield high ionic conductivity. A method that may 136 increase K IC without drastically decreasing the total ionic conductivity would be the addition of parti ally stabilized ZrO 2 particles within the LLZO matrix, as has been used for the case of beta Al 2 O 3 [ 14 8 - 14 9 ] . This could lead to increased K IC due to transformation toughening within the matrix and hopefully, would not diminish the high conductivity grain boundaries that result from hot - pressing to high relative density. Another possible solution could be the addition of a second phase (e.g., glass) along the LLZO grain boundaries that allows for high Li - ion conductivity across the grain boundaries but, w hen subjected to a mechanical stress would preferentially fracture along the grain boundaries giving improved toughness . 6.4 Summary The effect of relative density (porosity) on the hardness, K IC and total ionic conductivity of hot - pressed Al - substituted cub ic LLZO was investigated. It was observed that hot - pressing for 30 min , 60 min , 90 min , and 240 min at 1050 o C, resulted in 85 % , 95 % , 96 % , and 98 % relative densities, respectively. The average grain size varied from about 2.7 µm to 3.7 µm, while the primary fracture mode changed from inter to intragranular as the hot - pressing time increased from 30 min to 240 min. The H v increased with relative density up to approximately 96 %, above which the H v was constant. The increase in H v was correlated with a change in fracture mode from inter to intragranular as a result of reduced porosity at the grain boundaries leading to stronger boundaries as the relative density increased. At 98 % relative density , where almost 100 % intragranular fracture was exhibite d , the H v was equal to the H n . This hardness value is in good 137 agreement with the predicted value based on the behavior of single crystalline oxide garnets , suggesting that the single crystal hardness of LLZO is approximately 9.1 GPa. The K IC values decre ased linearly with increased relative density. The K IC values were 2.37 and relative density, respectively . Microstructural analysis suggests that the reason s for the increased K IC values at low density are a result of increased intergranular porosity at the grain boundaries. The intergranular porosity results in weak grain boundaries, which deflects cracks out of the plane of maximum driving force and hence, increasing K IC . The total ionic conductivity increa sed with increasing relative density. This increase is associated with an increase in the grain boundary conductivity as a result of the change in the nature of the grain boundaries with the increasing relative density. At a relative density of 85%, the total ionic conductivity wa s 0.0094 mS cm - 1 and increase d to 0.3 4 mS cm - 1 for the sample with a relative density of 98 %. An interesting correlation between ionic conductivity and K IC was observed. As the relative density increased, the ionic conductivit y increased while the K IC decreased. This correlation suggests that if one desires a LLZO material with high ionic conductivity a sacrifice in K IC will occur. One possible solution to this dilemma is the addition of partially stabilized ZrO 2 particles wit hin the LLZO matrix which results in a toughening mechanism that acts with in the grains yet, leaves strong grain boundaries to yield high ionic conductivity. Another possibility is the addition of a low K IC phase along the grain boundaries that exhibits good Li - ion transport, but promotes inter rather than intragranular fracture. 138 7 Summary and f uture work 7.1 Summary - - - - - - - 139 7.2 Future work 7.2.1 HOH charge abuse testing - - - - - - - - 140 - - - - - - - - - - - - - - - 141 - - - 142 7.2.2 Rate mapping at low temperature - - - - - 7.2.3 Realizing a novel hybrid cell design with ceramic electrolytes (0.4 - 1 mS cm - 1 ), integrating it into all solid - state batteries may be challenging due to high interfacial impedance. However, the proposed hybrid solid - liquid cell design can reduce the interfacial contact resistance. To support this argument, a symmetric hybrid cells employing LLZO separators were fabricated and characterized (Figure 7 - 143 2). The cell was composed of an LLZO electrolyte membrane placed between 2 electrolyte - saturated separators, placed between two Li metal electrodes (Figure 7 - 2). The symmetric cell exhibited stable and ohmic behavior with negligible contact resistance up to 1 mA cm - 2 current density (Figure 7 - 2b). Moreover, the results of DC cycling test of the hybrid symmetric cell also showed stable behavior at 1 mA cm - 2 current density for 20 cycles (Figure 7 - 2c). These demonstrate that a fast ion conducting ceramic electrolyte may allow for the facile transport of Li - ions while acting as a physical barrier to stop Li dendrite propagation. Therefore, the same hybrid cell should be investigated as an alternative approach to mitigate safety concerns related Li metal dendrites. Furthermore, the combining HOH electrode and the ceramic electr olyte is expected to result in higher performance and safety (Figure 7 - 3). This combined approach may enable the development of high energy and power density Li - ion batteries with improved safety. 144 (a) - - - 145 Figure 7 - 2 (cont d) . 146 - - 147 REFERENCES 148 REFERENCES [1] Q. Wang , P. Ping, and X. Zhao et al. "Thermal runaway caused fire and explosion of lithium ion battery," Journal of P ower S ources 208 (2012): 210 - 224. [2] L.X. Yuan , Z.H. Wang, and W.X. Zhang et al. PO 4 cathode material for lithium - Energy & Environmental Science 4 (2011): 269 - 284. [3] G. Majeau - of lithium - ion and nickel metal hydride batteries for plug - in hy Environmental S cience & T echnology 45 (2011): 4548 - 4554. Materials 6 (2013): 156 - 183. [ 5 ge - Annual R eviews of C hemical and B imolecular E ngineering 2 (2011) : 503 - 27. [6] Science 334 (2011) : 935 - 939. [7] B. Dunn, H. Kamath, and J Science 18 (2011) : 928 - 935. [8] J.M. Tarascon and M. Armand. "Issues and challenges facing rechargeable lithium batteries." Nature 414 (2001): 359 - 367. [9] Y.P. Wu, E. Rahm, an d Journal of Power S ources 114 (2003): 228 - 236. [10] R. Moshtev and B. Johnson. "State of the art of commercial Li ion batteries." Journal of Power Sources 91 (2000): 86 - 91. [11] T. Nagaura an Prog. Batteries Solar Cells 9 (1990): 209 - 217. [12] C. Jiang, E. Hosono, and H. Zhou. "Nanomaterials for lithium ion batteries." Nano Today 1 (2006): 28 - 33. [13] P. Arora, R.E. White, and M. Doyle. "Capaci ty fade mechanisms and side reactions in lithium ion Batteries." Journal of the Electrochemical Society 145 (1998): 3647 - 3667. 149 [14] A.S. Arico, P. Bruce, and B. Scrosati et al. "Nanostructured materials for advanced energy conversion and storage devices." Nature M aterials 4 (2005): 366 - 377. [15] R.A. Huggins. Advanced batteries , Springer Science+Business Media, LLC (2009). Department of Energy, Washington, DC, July (2007). [1 7 ] W.A. van Schalkwijk, B. Scros ati. Advances in lithium ion batteries, Kluwer Academic Publishers (2002) [1 8 - lithium negative electrode for Journal of P ower S ources 9 (1983) : 365 - 371. [1 9 ] H. Buqa, D. Goers, and M. Holzapfel et al. "High rate capability of graphite negative electrodes for lithium - ion batteries." Journal of the Electrochemical Society 152 (2005): A474 - A481. [ 20 ] X. Zhu, Y. Zhu, and S. Murali et al. "Nanostructured reduced graphene oxid e/Fe 2 O 3 composite as a high - performance anode material for lithium ion batteries." A CS Nano 5 (2011): 3333 - 3338. [2 1 Journal of P hysi cs C: Solid S tate P hysic 10 (1977) : 3235 3242. [2 2 ] Storylines in intercalation chemistry Dalton T ransactions 43 (2014) : 10276 - 102 91. rechargeable li - 763. Angewandte Chemie International Edition 22 (1983): 950 - 975. [2 5 ] C.K. Chan, H. Peng, and G. Liu et al. "High - performa nce lithium battery anodes using silicon nanowires , " Nature N anotechnology 3.1 (2008): 31 - 35. [2 6 ] M. Winter and J.O. Besenhard. "Electrochemical lithiation of tin and tin - based intermetallics and composites , " Electrochimica Acta 45.1 (1999): 31 - 50. [2 7 ] Journal of P ower S ources 195 (2010) : 2419 - 24 30. [28] A. Manthiram. "Materials challenges and opportunities of lithium ion batteries," The Journal of Physical Chemistry Letters 2.3 (2011): 176 - 184. 150 [2 9 ] K. Mizushima, P.C. Jones, P.J. Wiseman, J.B. Goodenough x CoO 2 (0