DEVELOPMENT of BATTERIES for IMPLANTABLE APPLICATIONS By

DEVELOPMENT OF BATTERIES FOR IMPLANTABLE

APPLICATIONS

BUSHAN KUMAR PURUSHOTHAMAN

Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy

Dissertation Advisor: Prof. Jesse S. Wainright

Department of Chemical Engineering

CASE WESTERN RESERVE UNIVERSITY

August, 2006

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

______

candidate for the Ph.D. degree *.

(signed)______(chair of the committee)

______

(date) ______

*We also certify that written approval has been obtained for any proprietary material contained therein.

I dedicate the dissertation to my family members for their love and steadfast support to me at all times. I am grateful for their many sacrifices to make my life successful.

iii TABLE OF CONTENTS

Page Number List of Tables x List of Figures xiii Acknowledgements xxvi List of Symbols xxvii Abstract xxviii

1.0 Introduction 1 1.1 Background of Networked Neuroprosthetic System 1 1.2 Existing Rechargeable Battery Technologies 2 1.3 Overview of High Pressure Nickel-Hydrogen Batteries 4 1.3.1 Introduction 4 1.3.2 Cell Components and Electrochemistry involved 5 1.3.3 Nickel-Hydrogen Cell Construction 8 1.4 Proposed Research 12 1.5 Prior Work 12 1.6 Objective and Overview of this work 12 1.6.1 Primary Objective 13 1.3.3 Secondary Objective 14 Works Cited 14

2.0 Cycle Life Tests on Lithium Ion Batteries for Implantable Applications 16 2.1 Overview of Lithium Ion Batteries 16 2.2 Description of Li Ion Cell 17 2.3 Implantable Li Ion Batteries 18 2.4 Objective 19 2.5 Charging Protocol for Lithium Ion Batteries 19 2.6 Experimental Design 21

iv 2.6.1 Effect of Charge Rate on Cycle life Performance 23 2.6.2 Effect of Discharge Rate on Cycle Life Performance 32 2.6.3 Effect of Charge Voltage Cutoff on Cycle Life 36 Performance 2.6.4 Effect of DOD on Cycle Life Performance 37 2.6.4.1 Sony Lithium Ion Cells at 100 and 80% DOD 37 2.6.4.2 Wilson Greatbatch Lithium Ion Cells at 80 and 40 60% DOD 2.6.5 Effect of Temperature on Cycle Life Performance 41 2.6.6 Comparison of Sony, Wilson Greatbatch and Quallion 43 2.6.7 Comparison of Sony and WG at 80% DOD 48 2.7 Comparison of Implantable battery technologies 49 2.7.1 Cycle life of Quallion and WG at 60% DOD 49 2.7.2 Cycle Life of WG and Quallion at 100% DOD 52 2.6.3 Long term tests on WG and Quallion 52 2.8 Comparison of Charge Times for New and Used Quallion cells 54 2.9 Conclusions 61 Works Cited 62

3.0 Platinum Negative Electrode – Testing and Activation 65 3.1 Introduction – Platinum Electrode 65 3.2 Characterization of the ESNS Electrode in a Liquid Cell 67 Configuration 3.2.1 Hydrogen Evolution Reaction Kinetics 68 3.2.2 Electrochemical Impedance Spectroscopy 69 3.2.3 Potentiodynamics 73 3.2.4 Conclusions 75 3.3 Characterization of the ESNS Electrode in Double Negative Cell 75 Configuration 3.3.1 Testing of Fresh ESNS Electrodes 76

v 3.3.2 Conditioning of ESNS Electrodes with Cyclic 78 Voltammetry 3.3.3 Effect of Rest 81 3.3.4 Effect of soaking in KOH for days 82 3.4 Testing of Current Collectors 83 3.5 Testing of Separators 85 3.6 Conclusions 85 Works Cited 87

4.0 Testing of ‘D Cell’ Nickel Hydroxide Electrode 89 4.1 Estimation of Specific Capacity 89 4.2 Performance of D cell Electrode in Liquid Cell 89 4.2.1 Description of the Experimental Setup 89 4.2.2 Charge-Discharge Experiments 90 4.2.3 Impedance Analysis 90 4.3 Performance of the D cell Electrode in Nickel-Hydrogen Cell 93 4.4 Performance of the D cell Electrode in Series Cell 97 4.5 Conclusions 102 Works Cited 102

5.0 Fabrication and Formation of Nickel Hydroxide Electrode 103 5.1 Introduction 103 5.2 Existing Methods of Fabrication 103 5.2.1 Pasted Nickel Electrode 104 5.2.2 Plastic Bonded Electrode 104 5.2.3 Sintered Electrode 105 5.3 Fabrication Methodology 105 5.3.1 Substrate 105 5.3.2 Paste Formulation 106 5.3.3 Screen Printing 107 5.4 Formation of the Nickel hydroxide Electrode 109

vi 5.4.1 Overview of Reactions in the Nickel Hydroxide 110 Electrode 5.4.2 Overview of Electrochemical Voltage Spectroscopy 111 5.4.3 Mechanism of Formation 112 5.4.4 Effect of Incomplete Formation 117 5.4.5 Effect of Different Rates of Formation 121 5.4.6 Effect of Overcharging during Formation 123 5.4.6.1 No Overcharge during Formation 123 5.4.6.2 50% Overcharge during Formation 124 5.4.6.3 Comparison of No Overcharge and 50% 125 Overcharge 5.5 Characterization of Nickel Hydroxide Electrode 129 5.5.1 Effect of PVDF Content on Electrode Performance 129 5.5.2 Effect of Loading Density 135 5.5.3 Effect of Fine Ni 139 5.5.3.1 Electronic Conductivity Studies 139 5.5.3.2 Comparison of electrodes Containing 0% and 142 8.8% Ni 5.5.3.3 Comparison of Electrodes Containing 5.2% 147 and 8.7% PVDF in the Presence of Fine Ni 5.5.4 Effect of Pressing the Electrode 148 5.6 Other Methods of Fabrication 148 5.6.1 Mechanical Impregnation Using Spatula 148 5.6.2 Mechanical Impregnation Using Hot Press 150 5.7 Conclusions 151 Works Cited 152

6.0 Assembly and Testing of Low Pressure Nickel Hydrogen Battery 155 6.1 Description of Experimental Setup 155

6.2 Performance of the Ni-H2 Cell in Comparison to the Liquid Cell 157

6.3 Columetric Efficiency of the Ni-H2 Cell 162

vii 6.3.1 Effect of Depth of Charge on Columetric Efficiency 162 6.3.2 Effect of Rest Period on Columetric Efficiency – Self- 165 discharge 6.4 Pressure Data Analysis 169 6.4.1 Pressure Variations During Charge 171 6.4.2 Pressure Variations During the Rest Period 174 6.4.3 Self Discharge at Different Depths of Charge 177

6.4.4 Recombination in the Ni-H2 Cell During Charge and 178 Rest Period 6.4.5 Pressure Contributions – Self discharge and 180 Recombination 6.4.6 Efficiency Loss – Self Discharge and Recombination 180 6.5 Comparison of Efficiency Based on Charge Data and Pressure 183 Data

6.6 Ni-H2 Cell Tests – Below Atmospheric Pressure 183 6.7 Cycle Life Tests 185 6.8 Conclusions 190

7.0 Low Pressure Nickel-Hydrogen Battery with Metal Hydride 192 7.1 Metal hydrides for hydrogen storage 192 7.1.1 Metal Hydride – Advantages and Applications 192 7.1.2 Factors Affecting Cycle Life of Metal Hydride 193 Electrodes 7.1.3 Comparison of Metal Hydride in Ni-MH and Low 195

Pressure Ni-H2 Cell 7.2 Pressure Composition Isotherm of the Metal Hydride 197 7.3 Testing of Metal Hydride Exposed to Air 198 7.4 Testing of Metal Hydride Exposed to KOH Solution 199

7.5 Testing of Ni-H2 Battery with Metal Hydride 200 7.5.1 Columetric efficiency Based on Charge-Discharge Data 203 7.5.2 Pressure Variations During the Charge-Discharge Cycle 206

viii 7.5.3 Pressure Variations During Charge 207 7.5.4 Pressure Variations During Rest after Discharge 208 7.5.5 Pressure Variations During Rest after Charge 210 7.5.6 Correlation of Pressure Variations During Discharge to 212 Pressure Composition Isotherm – Fuel Gauging 7.5.7 Correlation of Pressure Variations During Charge to 216 Pressure Composition Isotherm – Fuel Gauging 7.5.8 Self discharge at Different Depth of Charge 218 7.5.9 Recombination During Charge and Rest 218 7.5.10 Prediction of State of Charge 219

7.6 Problems associated with Ni-H2 Battery 219 7.7 Conclusions 224 Works Cited 225

8.0 Conclusions and Recommendations for Future Work 228 8.1 Conclusions 228 8.2 Recommendations for Future Work 231

9.0 Bibliography 232

ix LIST OF TABLES

Page Number Table 2.1: Comparison of Quallion and Wilson Greatbatch lithium ion cells. 19

Table 2.2: Description of Sony lithium ion cells subjected to different test 24 conditions.

Table 2.3: Comparison of parameters and errors estimated by fitting the 30 =+ discharge capacity as a function of cycle life to equation Ckkcycled 12 for C and C/5 rate of charging.

Table 2.4: Comparison of parameters and errors estimated by fitting the 46 =+ discharge capacity as a function of cycle life to equation Ckkcycled 12 for lithium ion cells of different manufacturers charged at C/5 rate and discharged at C rate.

Table 2.5: Linear fit parameters and their error estimations of the end of 52 discharge voltage data. Wilson Greatbatch and Quallion lithium ion cells were charged and discharged at C rate to 60% DOD.

Table 2.6: Linear fit parameters and their error estimations of the end of 54 discharge voltage data. Wilson Greatbatch and Quallion lithium ion cells were charged and discharged at C/5 and C/4 rate, respectively.

Table 3.1: Impedance analysis in the potential range 0 to -0.5 V. 70

Table 3.2: Impedance analysis at in the potential range 0 to -20V. 72

x Table 3.3: Estimated parameters for fresh electrodes based on the linear fit 77 applied to the CV data.

Table 3.4: Estimated parameters for fresh electrodes based on the linear fit 81 applied to the CV data.

Table 5.1: Peak potentials for different electrochemical reactions. 120

Table 5.2: Utilization of electrode containing about 10% PVDF and 90% 122

Ni(OH)2 charged at different during formation.

Table 5.3: Capacity of electrodes containing different amounts of PVDF. 131

Table 5.4: Peak potentials for different electrochemical reactions of electrodes 133 containing different amounts of PVDF.

Table 5.5: Capacity of electrodes containing different amounts of PVDF. 134

Table 6.1: Parameters associated with the ESNS platinum electrode. 162

Table 6.2: Parameters associated with the nickel hydroxide electrode. 162

Table 6.3: Parameters associated with the complete Ni-H2 cell. 162

Table 6.4: Columetric efficiency of the Ni-H2 cell as a function of depth of 165 charge cycled at C/5 rate.

Table 6.5: Pressure loss contributions of recombination and self discharge for 180 6h rest period.

xi Table 7.1: Change in various parameters following the desorption curve of the 215 pressure composition isotherm for discharge.

Table 7.2: Change in various parameters following the adsorption curve of the 217 pressure composition isotherm for charge.

Table 7.3: Change in various parameters following the desorption curve of the 217 pressure composition isotherm for charge.

Table 7.4: Self discharge rate for different depths of charge. 218

xii LIST OF FIGURES

Page Number Figure 1.1.1: Schematic picture of the networked neuroprosthetic system. The 2 complete system in the body is shown in the left side of the figure. A magnified view of the access port, the network cables and the modules is shown in right side of the figure.

Figure 1.2.1: Comparison of the different battery technologies in terms of 4 volumetric and gravimetric energy density 1. The energy density of the nickel- hydrogen batteries was added to the figure.

Figure 1.3.1: The components of COMSAT nickel-hydrogen cell in stacking 8 arrangement2.

Figure 1.3.2:Cross-sectional view components of Comsat cell in back to back 9 design 8.

Figure 1.3.3: Exploded view showing the components of an Air Force cell in 10 stacking arrangement7.

Figure 2.2.1: Schematic representation of a lithium ion battery during 18 charging. The separator is impregnated with liquid electrolyte.

Figure 2.5.1: Schematic of DC charge characteristics. During the initial stage 20 (stage 1), charging is done at a constant current, 1.4 A for Sony cell 18650 (Capacity – 1.4 Ah, 1 C charge rate) (A) and the voltage is rising (B). tT designates the time at which the voltage reaches a value of about 4.2 V and the charging mode switches from constant current to stage 2, where charging continues at constant voltage while the current is continuously decreasing. About 85 % of the capacity is charged in stage 1 and the rest, 15 %, is charged in stage 2.

Figure 2.6.1: Discharge capacity as a function of cycle index for Sony Li ion 24 cell 1. The cell loses discharge capacity with cycles. The cell was charged and discharged at C rate. The scattered points in the cycle number range 1-100 is due to instrumentation error and does not reflect the battery’s discharge capacity.

xiii Figure 2.6.2: Discharge capacity as a function of cycle index for Sony lithium 25 ion cell 1 fitted to a straight line. The linear fit is poor for discharge capacity in the cycle number range 1-200. The capacity loss is 1.5x10-4 Ah/cycle. The cell was charged and discharged at C rate.

Figure 2.6.3: Discharge capacity as a function of cycle index for Sony Li ion 26 cell 1 fitted to a straight line for the first 200 cycles. The estimated capacity loss per cycle is 4x10-4 Ah. The cell was charged and discharged at C rate.

Figure 2.6.4: Discharge capacity as a function of cycles for Sony Li ion cell 1 28 - fitted to a non-linear curve. The parameters k1 and k2 are 1.47 Ah and 8x10 3 Ah/ cycle . The cell was charged and discharged at C rate.

Figure 2.6.5: Discharge capacity as a function of cycles for Sony Li ion cell 2 28 - fitted to a non-linear curve. The parameters k1 and k2 are 1.47 Ah and 7.2x10 3 Ahcycle/ . The cell was charged and discharged at C rate.

Figure 2.6.6: Discharge capacity as a function of cycles for Sony Li ion cell 3 29 - fitted to a non-linear curve. The parameters k1 and k2 are 1.48 Ah and 8.1x10 3 Ah/ cycle . The cell was charged and discharged at C rate.

Figure 2.6.7: Discharge capacity as a function of cycles for Sony Li ion cell 4 29 - fitted to a non-linear curve. The parameters k1 and k2 are 1.5 Ah and 5.2x10 3 Ah/ cycle . The cell was charged at C/5 rate and discharged at C rate.

Figure 2.6.8: Discharge capacity as a function of cycles comparing Sony Li 31 ion cells 1 and 4. The capacity fade data is fitted to a non-linear curve. Cell 1 was charged at C rate and Cell 4 was charged at C/5 rate. The degradation constant is smaller for the lithium ion cell with C/5 charge in comparison to the lithium ion cell with C charge.

Figure 2.6.9: Energy efficiency as a function of cycles comparing Sony Li ion 32 cells 1 and 4. Cell 1 was charged at C rate and Cell 4 was charged at C/5 rate. The energy efficiency of Cell 1 is always smaller than that of the Cell 4.

Figure 2.6.10: Discharge capacity as a function of cycles for Sony Li ion cell 34 cycled at C charge rate and C/5 discharge rate fitted to a non-linear curve. The -2 parameters k1 and k2 are 1.38 Ah and 1.1x10 Ahcycle/ .

Figure 2.6.11: Discharge capacity as a function of cycles for Sony Li ion cell 34 cycled at C charge rate and 2W discharge rate fitted to a non-linear curve. The -2 parameters k1 and k2 are 1.38 Ah and 1.36x10 Ah/ cycle .

xiv Figure 2.6.12: Discharge capacity as a function of cycles comparing lithium 35 ion cells cycled at different discharge rates. The degradation constant, k2, is larger for the lithium ion cell with C/5 discharge in comparison to the lithium ion cell with C discharge.

Figure 2.6.13: Energy efficiency as a function of cycles comparing lithium ion 35 cells subjected to C and C/5 rate of discharge.

Figure 2.6.14: Discharge capacity as a function of cycles for Sony Li ion cell 37 cycled at C charge rate with 4.2 voltage cut off and 2W discharge rate. The -2 parameters k1 and k2 are 1.66 Ah and 1.25x10 Ah/ cycle for the first 350 cycles.

Figure 2.6.15: Total number of cycles obtained for Sony lithium ion cells 39 cycled to different states of depth of discharge.

Figure 2.6.16: Total number of ampere-hours obtained for Sony lithium ion 39 cells cycled to different states of depth of discharge.

Figure 2.6.17: Total number of cycles obtained for Wilson Greatbatch lithium 40 ion cells cycled to 80 and 60% of depth of discharge.

Figure 2.6.18: End of discharge voltage as a function of cycles comparing 42 Sony Li ion cell cycled at room temperature and 37o C. The cells were charged and discharged at C rate to 80% DOD. The data was fitted to a straight line.

Figure 2.6.19: End of discharge voltage in the cycle number range 0-175 42 comparing Sony Li ion cell cycled at room temperature and 37o C. The cells were charged and discharged at C rate to 80% DOD.

Figure 2.6.20: Discharge capacity as a function of cycles for Wilson 45 Greatbatch Li ion cell 4 fitted to a non-linear curve. The parameters k1 and k2 are 1.4 Ah and 2.7x10-2 Ah/ cycle for a fit to the first 500 cycles. The cell was charged at C/5 rate and discharged at C rate.

Figure 2.6.21: Discharge capacity as a function of cycles for Quallion Li ion 45 cell 4 fitted to a non-linear curve. The parameters k1 and k2 are 0.2 Ah and 2.7x10-3 Ah/ cycle . The cell was charged at C/5 rate and discharged at C rate.

Figure 2.6.22: Number of cycles obtained at 10% capacity loss for different 46 lithium ion cell manufacturers Sony, Wilson Greatbatch and Quallion. The cells were cycled at C/5 charge rate and C discharge rate to 100% DOD.

xv Figure 2.6.23: Number of cycles obtained at 10, 20 and 30% capacity loss for 47 Wilson Greatbatch and Quallion lithium ion cells. The cells were cycled at C/5 charge rate and C discharge rate to 100% DOD at 37o C.

Figure 2.6.24: The degradation constant, k2, for different lithium ion cell 47 manufacturers Sony, Wilson Greatbatch and Quallion. The normalized rate constants were obtained by ratioing the rate constant to the cell’s nominal capacity. The cells were cycled at C/5 charge rate and C discharge rate to 100% DOD.

Figure 2.6.25: Total number of cycles obtained for Sony and Wilson 48 Greatbatch lithium ion cells cycled to 80 % of depth of discharge at C rate charge/ C rate discharge.

Figure 2.7.1: End of discharge voltage as a function of cycles for WG Li ion 51 cell tested at 37o C. The cells were charged and discharged at C rate to 60% DOD. The data was fitted to a straight line over cycles 1 to 1200.

Figure 2.7.2: End of discharge voltage as a function of cycles for Quallion Li 51 ion cell tested at 37o C. The cells were charged and discharged at C rate to 60% DOD. The data was fitted to two straight lines.

Figure 2.7.3: End of discharge voltage as a function of cycles for WG Li ion 53 cell tested at 37o C. The cells were charged and discharged at C/5 rate to 60% DOD. The data was fitted to a straight line.

Figure 2.7.4: End of discharge voltage as a function of cycles for Quallion Li 54 ion cell tested at 37o C. The cells were charged and discharged at C/4 rate to 80% DOD. The data was fitted to a straight line.

Figure 2.8.1: Total charge time as a function of different charging rates for 58 new and used Quallion cells.

Figure 2.8.2: Percentage of constant current and constant voltage charge times 58 as a function of different charging rates for new Quallion cells. The cells during charging spend more time in the constant current phase rather than in the constant voltage phase.

Figure 2.8.3: Percentage of constant current and constant voltage charge times 59 as a function of different charging rates for used Quallion cells.

Figure 2.8.4: End of discharge voltage as a function of different charge rates 59 for new and used Quallion cells. A straight line is fit to the data.

xvi Figure 2.8.5: Watt-hour efficiency of the used and the new Quallion cell at 60 different charging rates.

Figure 2.8.6: Percentage increase in charge time of the used Quallion cell in 60 comparison to the charge time of the new Quallion cell. The percentage charge time is maximum for the 1C charge rate.

Figure 3.2.1: Schematic representation of the experimental setup. 67

Figure 3.2.2: Impedance spectra of the ESNS electrode in the potential range - 71 8 to -154 mV. The charge transfer resistance decreases with increase in the bias potential.

Figure 3.2.3: Impedance spectra of the ESNS electrode in the linear 72 polarization regime. The potential range is -8 to -15 mV. The charge transfer resistance does not change with the bias potential.

Figure 3.2.4: Potentiodynamic scan of the ESNS electrode at a scan rate of 0.1 74 mV/s. Two different regimes are indicated in the figure.

Figure 3.2.5: Linear fit of the potentiodynamic scan in the potential range -6 to 74 -16 mV. The scan rate is 0.1 mV/s. The parameter ‘B’ is the inverse of the charge transfer resistance.

Figure 3.3.1: Schematic representation of the experimental setup. 76

Figure 3.3.2: Charge transfer resistance and double layer capacitance as a 79 function of electrode condition based on impedance measurements before and after every set of cyclic voltammograms. ‘Cy’ in the x-axis text indicates cycles.

Figure 3.3.3: First set of cyclic voltammogramms (3 cycles) at a scan rate of 5 80 mV/s.

Figure 3.3.4: IR corrected cyclic voltammogramms showing the linear regime 81 at a scan rate of 5 mV/s. The data in the voltage range, -2 to -22mV, was linearly fit.

Figure 3.3.5: Charge transfer resistance and double layer capacitance as a 82 function of electrode condition based on impedance measurements before and after different rest periods.

Figure 3.4.1: The high frequency resistance of the cell using nickel chromium 84 mesh current collectors and nickel mesh current collectors is compared.

xvii Figure 3.4.2: The high frequency resistance of the cell, one using nickel 84 chromium mesh as current collectors and other using nickel mesh as current collectors is compared for two different conditions fresh and used.

Figure 4.2.1: Voltage profile during charge-discharge of the D cell nickel 91 hydroxide electrode for cycles 1 to 5. The cell was cycled at C/3 rate to 100% capacity.

Figure 4.2.2: Columetric efficiency of the D cell nickel hydroxide electrode as 92 a function of cycles. The inset shows the small variations in the columetric efficiency. The cell was cycled at C/3 rate to 100% capacity.

Figure 4.2.3: Impedance spectra of the D cell nickel hydroxide electrode as a 92 function of state of charge. The cell was cycled at C/3 rate.

Figure 4.3.1: Voltage profile during the charge-discharge of the Ni-H2 cell 95 using the D cell nickel hydroxide electrode for cycles 1 and 5. There is no discharge cut-off voltage. The cell was cycled at C/3 rate to 80% capacity with no voltage cutoff.

Figure 4.3.2: Impedance spectra of the Ni-H2 cell using the D cell nickel 95 hydroxide electrode measured at the end of charge of cycles 1 and 5. The cell was cycled at C/3 rate to 100% capacity.

Figure 4.3.3: Voltage profile during the charge-discharge of the Ni-H2 cell 96 using the D cell nickel hydroxide electrode for 5 cycles. The discharge cut off voltage is 1.0 V The cell was cycled at C/3 rate to 80% capacity.

Figure 4.3.4: Voltage profile during the charge-discharge of the Ni-H2 cell 96 using the D cell nickel hydroxide electrode comparing overcharge with no overcharge. The cell was cycled at C/6 rate with no voltage cutoff.

Figure 4.4.1: Schematic picture of two nickel-hydrogen cells in series. 99

Figure 4.4.2: Variations in the open circuit voltage of series cell with different 99 bipolar plates. The cell is purged first with nitrogen and then with hydrogen.

Figure 4.4.3: Voltage profile during the charge-discharge of the series cell. The 100 discharge cut off voltage is 2.0 V The cell was cycled at C/3 rate to 80% capacity.

Figure 4.4.4: Impedance spectra of the series cell measured at the end of 100 discharge of cycles 1, 5 and 10. The cell was cycled at C/3 rate to 80% capacity.

xviii Figure 4.4.5: Voltage profile at the end of charge for cycles 4, 7 and 10 of the 101 series cell. The discharge cut off voltage is 2.0 V The cell was cycled at C/3 rate to 80% capacity.

Figure 4.4.6: Voltage profile for the 12th charge of the series cell. The 101 discharge cut off voltage is 2.0 V The cell was cycled at C/3 rate to 80% capacity.

Figure 5.3.1: SEM image of INCO nickel foam of thickness 1.6 mm and cell 107 size 550-700 µm.

Figure 5.3.2: Schematic illustration of screen printing technology to 108 mechanically press the paste into the nickel foam substrate.

Figure 5.4.1: Schematic diagram of the cell setup used for the formation of the 110 nickel hydroxide electrode.

Figure 5.4.2: Schematic diagram for interconversion of active material phases 111 in the nickel hydroxide electrode44.

Figure 5.4.3: Variations in voltage during charge and discharge of cycles 1 and 114 2. The electrode composition is 10% PVDF and 90% Ni(OH)2 by weight.

Figure 5.4.4: Voltage variation in the time period 0 to 0.5h during the first 114 charge. The electrode composition is 10% PVDF and 90% Ni(OH)2 by weight.

Figure 5.4.5: Impedance data of 10% PVDF containing electrode: (A) Before 115 formation, (B) End of discharge after cycles 1 and 2.

Figure 5.4.6: Voltage profile during charge and discharge of cycles 1 and 2 for 116 15% PVDF containing nickel hydroxide electrode.

Figure 5.4.7: Impedance data of 15% PVDF containing electrode measured at 117 the end of discharge after cycles 1, 2 and 3.

Figure 5.4.8: Comparison of EVS scans for completely and incompletely 119 formed nickel electrode containing 16% PVDF and 84% Ni(OH)2.

Figure 5.4.9: Schematic representation of complete and incomplete charging of 120 the nickel hydroxide electrode.

Figure 5.4.10: SEM image of nickel hydroxide electrodes: (A) Fresh 16% 121 PVDF electrode, (B) Used 12.7% PVDF electrode incompletely formed.

xix Figure 5.4.11: Comparison of EVS scans for two different electrodes, one 123 charged at C/10 and another at C/20 rate with 50% overcharge. The electrode composition is 10% PVDF and 90% Ni(OH)2 by weight.

Figure 5.4.12: Impedance spectra of the electrode measured at the end of 126 discharge (A) and charge (B) without overcharge at C/10 rate for the first 5 cycles.

Figure 5.4.13: Impedance spectra of an electrode measured at the end of 127 discharge (A) and charge (B) at C/10 rate with 50% overcharge for the first two cycles.

Figure 5.4.14: Utilization as a function of cycles for electrodes formed with 128 0% and 50% overcharge.

Figure 5.4.15: Impedance spectra measured after discharge for electrodes 128 formed with 0% and 50% overcharge.

Figure 5.5.1: Utilization as a function of cycles for electrodes containing 131 different amounts of PVDF content. The electrodes were cycled at C/10 rate with 50% overcharge.

Figure 5.5.2: Impedance spectra measured at the end of discharge (A) and 132 charge (B) for electrodes containing different PVDF content. The electrodes were formed at C/10 rate with 50% overcharge.

Figure 5.5.3: EVS scans for electrodes containing different PVDF content 133 formed at C/10 rate with 50% overcharge.

Figure 5.5.4: Utilization of electrodes containing different amounts of PVDF 134 content at lower capacities. The electrodes were cycled at C/10 rate with 50% overcharge.

Figure 5.5.5: Utilization of 10% PVDF electrodes with different capacities 136 (different loading levels). The electrodes were cycled at C/10 rate with 50% overcharge.

Figure 5.5.6: Utilization of 15% PVDF electrodes with different capacities 136 (different loading levels). The electrodes were cycled at C/10 rate with 50% overcharge.

Figure 5.5.7: Impedance spectra measured at the end of discharge (A) and 137 charge (B) for electrodes containing 10% PVDF with different capacities. The electrodes were formed at C/10 rate with 50% overcharge.

xx Figure 5.5.8: Impedance spectra measured at the end of discharge (A) and 138 charge (B) for electrodes containing 15% PVDF with different capacities. The electrodes were formed at C/10 rate with 50% overcharge.

Figure 5.5.9: Four-point conductivity ceramic base used for AC impedance 140 spectroscopy.

Figure 5.5.10: Electronic conductivity of the paste containing nickel hydroxide 141 particles (Kansai) as a function of nickel Wt. %.

Figure 5.5.11: Electronic conductivity of the paste containing nickel hydroxide 141 particles (OMG) as a function of nickel Wt. %.

Figure 5.5.12: Comparison of the voltage profile in the first formation cycle 143 for electrodes with and without nickel.

Figure 5.5.13: Impedance spectra measured at the end of discharge (A) and 144 charge (B) for electrodes containing 0% and 8.8% Ni. The electrodes were formed at C/10 rate with 50% overcharge.

Figure 5.5.14: Utilization of electrodes with and without nickel. The electrodes 145 were cycled at C/10 rate with 50% overcharge. The electrodes containing nickel show higher utilization.

Figure 5.5.15: Top view of SEM image of the nickel hydroxide electrode. The 145 spherical structures are the nickel hydroxide particles and the filamentary structures are the nickel 210.

Figure 5.5.16: EVS scans for electrodes containing 0% and 8.8% nickel. The 146 electrodes were formed at C/10 rate with 50% overcharge.

Figure 5.5.17: Utilization of electrodes containing different amounts of PVDF. 147 The nickel in these electrodes is about 8.5 to 8.7 Wt. %. The electrodes were formed at C/10 rate with 50% overcharge.

Figure 5.6.1: Theoretical capacity of electrodes containing 8.8 % Nickel and 149 8.7% PVDF (dry active mass) fabricated by screen printing and pressing using flat spatula. The electrodes were formed at C/10 rate with 50% overcharge.

Figure 5.6.2: Utilization of electrodes containing 8.8 % Nickel and 8.7% 150 PVDF (dry active mass) fabricated by screen printing and pressing using flat spatula. The electrodes were formed at C/10 rate with 50% overcharge.

Figure 6.1.1: Schematic picture of the Ni-H2 battery 156

xxi Figure 6.1.2: Schematic picture of the experimental set up. 156

Figure 6.2.1: Voltage profile of the nickel hydroxide electrode during charge 158 and discharge in the liquid cell and in the nickel-hydrogen cell configuration.

Figure 6.2.2: Impedance spectra measured at the end of discharge (A) and 161 charge (B) for electrodes cycled at C/5 rate in liquid cell and nickel-hydrogen cell configuration.

Figure 6.3.1: Charge-discharge voltage profiles of the nickel-hydrogen cell 164 cycled at C/5 rate to different depth of charge: (A) 50% DOC, (B) 70% DOC and (C) 90% DOC.

Figure 6.3.2: Charge-discharge voltage profiles of the nickel-hydrogen cell 166 cycled at C/5 rate to 90% DOC with different rest periods, 5 min, 1h and 6h.

Figure 6.3.3: Columetric efficiency of the Ni-H2 cell as a function of the rest 167 period at 50% DOC, cycled at C/5 rate.

Figure 6.3.4: Columetric efficiency of the Ni-H2 cell as a function of the rest 167 period at 70% DOC, cycled at C/5 rate.

Figure 6.3.5: Columetric efficiency of the Ni-H2 cell as a function of the rest 168 period at 70% DOC, cycled at C/5 rate.

Figure 6.3.6: Columetric efficiency of the Ni-H2 cell cycled at C/5 rate as a 169 function of the depth of charge for different rest periods.

Figure 6.4.1: Voltage variation during charge, rest and discharge periods of a 170 cycle. The cell was cycled at C/5 rate to 90% depth of charge. The rest period is 6h.

Figure 6.4.2: Changes in pressure during the charge, the rest and the discharge 171 periods of a cycle. The cell was cycled at C/5 rate to 90% depth of charge. The rest period is 6h.

Figure 6.4.3: Changes in pressure during the charge. The cell was cycled at 173 C/5 rate to 90% depth of charge.

Figure 6.4.4: Schematic representation of different pressure variations due to 173 different possible mechanisms during charging.

Figure 6.4.5: Pressure changes during the rest period. The cell was cycled at 175 C/5 rate to 90% depth of charge. A straight line is fit to the pressure data in the 3 to 6h rest period and extrapolated back to time 0.

xxii Figure 6.4.6: Pressure changes due to recombination reaction during the rest 177 period. The cell was cycled at C/5 rate to 90% depth of charge. An exponential decay line is fit to the pressure data.

Figure 6.4.7: Self discharge rate estimated as a function of depth of charge for 178 Dataset 1 and 2.

Figure 6.4.8: Pressure drop due to the recombination reaction as a function of 179 depth of charge during the charge and the rest period.

Figure 6.4.9: Contributions of different processes at different depth of charge: 182 (A) 50% (B) 70% and (C) 90%. The rest period is 6h.

Figure 6.5.1: Comparison of efficiency calculated based on the charge data and 184 pressure data at different depth of charge for data set 1.

Figure 6.5.2: Comparison of efficiency calculated based on the charge data and 184 pressure data at different depth of charge for data set 2.

Figure 6.7.1: Discharge capacity as a function of cycle index for a nickel- 187 hydrogen cell with a capacity of 41 mAh.

Figure 6.7.2: Discharge capacity as a function of cycle index for a nickel- 187 hydrogen cell with a capacity of 62.5 mAh.

Figure 6.7.3: Voltage profiles during charge and discharge for cycles 1, 4 and 188 12. The capacity of the cell was 62.5 mAh.

Figure 6.7.4: Voltage profiles during charge and discharge for cycles 15, 18 188 and 21. The capacity of the cell was 62.5 mAh.

Figure 6.7.5: End of charge voltage as a function of cycle index. The capacity 189 of the cell was 62.5 mAh.

Figure 6.7.6: Pressure measured at the end of every cycle. The capacity of the 189 cell was 62.5 mAh.

Figure 6.7.7: Picture of the failed nickel-hydrogen cell at the end of 190 experiment.

Figure 7.2.1: Pressure-composition isotherm of the 10 wt% Pd modified 198 LaNi4.7Al0.3.

xxiii Figure 7.3.1: Effect of air exposure on the first hydrogen absorption of 199 palladium treated LaNi4.7Al0.3 for periods greater than 2 years. This figure is taken from reference 36.

Figure 7.4.1: Effect of soaking in KOH solution on the hydrogen absorption of 200 palladium treated LaNi4.7Al0.3. The hydrogen pressure at the end of absorption is about 1.1 atm pressure.

Figure 7.5.1: Schematic picture of the Ni-H2 cell with metal hydride. 201

Figure 7.5.2: Schematic picture of the Ni-H2 cell with metal hydride. 202

Figure 7.5.3: Pressure variation inside the nickel hydrogen cell during venting 203 and subsequent hydrogen gas equilibration with the metal hydride.

Figure 7.5.4: Charge-discharge voltage profiles of the nickel-hydrogen cell 204 cycled at C/5 rate to different depth of charge.

Figure 7.5.5: Charge-discharge voltage profiles of the nickel-hydrogen cell 205 cycled at C/5 rate to 50% DOC with different rest periods, 5 min, 1h and 6h.

Figure 7.5.6: Columetric efficiency of the Ni-H2 cell cycled at C/5 rate as a 205 function of the depth of charge for different rest periods.

Figure 7.5.7: Changes in pressure during the charge, the rest and the discharge 207 periods of two successive cycles. The cell was cycled at C/5 rate to 70% depth of charge. The rest period for the 1st cycle and the 2nd cycle is 5 min and 6h, respectively.

Figure 7.5.8: Changes in pressure during charge. The cell was cycled at C/5 208 rate to 70% depth of charge with 5 min rest period.

Figure 7.5.9: Changes in pressure during the rest period at the end of the 2nd 210 cycle shown in Fig. 7.5.7. The cell was cycled at C/5 rate to 70% depth of charge.

Figure 7.5.10: Pressure changes during the 6h rest period after 2nd charge. The 211 cell was cycled at C/5 rate to 70% depth of charge. A straight line is fit to the pressure data in the 4 to 6h rest period.

Figure 7.5.11: Pressure changes due to recombination reaction and hydrogen 212 equilibration with the metal hydride during the rest period after 2nd charge. The cell was cycled at C/5 rate to 70% depth of charge. An exponential decay line is fit to the pressure data.

xxiv Figure 7.5.12: Changes in pressure during the rest period after charge, the 215 discharge period and the subsequent rest period of the 2nd cycle. The cell was cycled at C/5 rate to 70% depth of charge.

Figure 7.6.1: Charge-discharge voltage profiles of the nickel-hydrogen cell 221 cycled at C/5 rate to 50% depth of charge for the case with and without oxygen leak.

Figure 7.6.2: Comparison of impedance spectra for two different cases (i) no 222 oxygen leak and (ii) oxygen leak measured at the end of charge (A) and discharge (B). The cell was cycled at C/5 rate to 50% depth of charge.

Figure 7.6.3: Changes in pressure during the rest period at the end of discharge 223 for the case ‘oxygen leak’. The cell was cycled at C/5 rate to 50% depth of charge.

Figure 7.6.4: Change in pressure inside the nickel hydrogen cell by increasing 223 the pressure suddenly to 31.8 psi for the case ‘oxygen leak’. The cell was cycled at C/5 rate to 50% depth of charge. The pressure variation shows hydrogen gas equilibration with the metal hydride.

xxv Acknowledgements

I acknowledge and greatly appreciate Prof. Jesse S. Wainright for his constant support and feedbacks during my Ph.D. I thank NIH (NINDS R01-NS41809) for supporting this work financially. I also acknowledge the Bio-research group members

Brian Smith, Jim Buckett, Tim Crish, George Blomgren and Prof. Payer for useful discussions in the group meeting.

I acknowledge Prof. Landau for his support and guidance in all my work. I acknowledge Prof. Mark DeGuire and Prof. Savinell for giving intellectual insights and for being part of my committee.

I thank my friends who had supported me in all times. Finally, I thank my family especially my parents and my grandparents for their love and belief in me.

xxvi List of Symbols

Cd discharge capacity of the cell (Ah) -2 -2 Cdl double layer capacitance (mF cm or F cm ) EODV end of discharge voltage (volt) F Faraday constant (96485 C/equivalent) H/M hydrogen concentration, hydrogen atom to metal atom ratio i current density (mA/cm2 or A/cm2) 2 2 io exchange current current density (mA/cm or A/cm ) I current (A) k first order rate constant (s-1) k1 initial capacity of the cell (Ah)

k2 degradation constant ( Ah/ cycle ) l thickness (cm) m voltage drop per cycle (volt/s) n number of electrons transferred in a faradic reaction (equivalents/mole) N number of cycles P oxygen pressure or cell pressure (psia) Po initial oxygen pressure (psia) R Gas constant (J/mole/K) 2 Rct charge transfer resistance (ohm cm ) 2 Rs solution resistance (ohm cm ) t time (s or h) t1 time constant of the recombination reaction (h) T temperature (K) V cell potential (volt) Vo end of discharge voltage for N = 0 (V) Z/ real impedance (ohm cm2) Z// imaginary impedance (ohm cm2)

Greek letters α anodic transfer coefficient β cathodic transfer coefficient ηa activation overpotential (volt) Ω ohmic

xxvii Development of Batteries for Implantable Applications

Abstract

BUSHAN KUMAR PURUSHOTHAMAN

Neuroprosthetic devices that electrically stimulate paralyzed muscles require implantable power sources with exceptional cycle life, safety, and sufficient energy and power density. Of the rechargeable battery technologies, lithium ion batteries have the highest energy density; however, they have limited cycle life of about 1000 cycles.

Nickel-hydrogen batteries, currently used in space applications are remarkable for long cycle life (40,000) and low maintenance; however they utilize high hydrogen pressures (60 atm) making them unsuitable for implantable applications. The present work involves design and development of low pressure nickel-hydrogen batteries (1 atm) by utilizing a metal hydride (MH) to store hydrogen, rather than as a negative electrode in the nickel-metal hydride battery.

A method to increase the exchange current density of the negative platinum electrode using cyclic voltammetry was developed. A nickel mesh was chosen as the current collector because of its low resistance and stability in alkaline solutions. The tested separators, zirconium oxide and polypropylene, were not significantly different from each other.

xxviii A pasted type nickel hydroxide electrode was fabricated by two means: screen printing and spatula pressing. The mechanism of electrode formation, the effect of different formation rates with and without overcharge and the effect of binder and nickel content on utilization were studied. Addition of filamentary nickel to the electrode increases the utilization by 10% by decreasing the oxygen evolution.

A low pressure nickel-hydrogen battery with and without MH was assembled.

Charge and pressure data were analyzed to study the oxygen evolution, the recombination reaction and the self discharge of the cell. Oxygen evolution increases with the depth of charge; however the evolved oxygen recombines completely – 70% during charging and the remainder during the first hour of the rest period. About 40-45% hydrogen from the metal hydride was used as fuel during cycling. The pressure composition isotherm was used to estimate the state of charge and to fuel gauge the cell. The MH is deactivated when exposed to KOH solution, increasing the hydrogen equilibration time.

Of the tested implantable lithium ion cells, Quallion outperformed the Wilson

Greatbatch cells in cycle life. The conditions for maximum cycle life were determined.

xxix 1. Introduction

1.1. Background of Networked Neuroprosthetic System

Neuroprosthetic devices provide electric stimulation to paralyzed muscles thereby

enhancing the functions of patients such as paraplegics, quadriplegics, people with spinal

cord injury and people suffering from other daily activities like standing, steeping, reaching and grasping. Two common methods of stimulation are surface stimulation

(stimulation outside the body) and percutaneous stimulation (stimulation inside the

body). The neuroprosthetic system being developed at Case Western Reserve University

is completely implantable and can be employed for different clinical applications because

of its open architecture. The advantages of an implantable system are repeatability,

cosmetics and flexibility. Instead of using a single large implantable module, a network

based system consisting of smaller modules distributed throughout the body is proposed

as shown in Fig. 1.1.1.

The three main components of the implantable system are the centralized power

system, backbone network cables and the modules. The centralized power system is the access port which holds the power source and provides power to the entire network. The

network cables connect the power system to the modules. The main function of the

modules is to sense and actuate the muscles. This system will have a life of about 30 to

40 yrs and therefore will be typically used in patients in their twenties. The power source

in the access port will be charged inductively across the skin. This system requires a

continuous power of approximately 500 mW, about 50 mW per module.

Figure 1.1.1: Schematic of the networked neuroprosthetic system. The complete system in the body is shown on the left side of the figure. A magnified view of the access port, the network cables and the modules is shown in right side of the figure.

1.2. Existing Rechargeable Battery Technologies

Large and renewable power (about 500 mW) required for the networked neural prosthetic system (NNPS) eliminates the choice of primary batteries. A rechargeable battery was chosen as the power source. The three major requirements of the power source are long cycle life, safety and high energy density. The NNPS has a life about 30 years and therefore the battery cycle life should be comparable, say about 10 to 20 yrs. A power source that is reliable and safe is an important criterion for the patients.

The energy densities of existing battery technologies1 are compared in Fig.1.1.2.

The lithium ion batteries have the highest energy densities both in terms of weight and volume, excluding lithium metal batteries which are unsafe. Because of the high energy density and the ease of minimizing battery size, they are ideal for portable equipments like laptop computers, digital video cameras, cellular phones and other hand held devices.

Companies like Quallion and Greatbatch manufacturer implantable grade lithium ion batteries and are described in detail in Chapter 2. However these are limited by cycle life of about 1000 cycles at complete charge and discharge. Assuming that the battery is cycled once a day, the lithium ion battery will last 3 years necessitating surgery after that.

The nickel-hydrogen batteries have lower energy densities in comparison to lithium ion batteries; however they have an order of magnitude higher cycle life of about

40,000-60,000 cycles at 40% DOD 2 and therefore could be expected to meet or exceed

the design goal of 20-30 yr life time.. They are also referred to as low maintenance batteries because over-charging and over-discharging the system does not affect the cycle life significantly in comparison to the lithium ion batteries. However, the main disadvantage is that they utilize hydrogen at high pressures of about 60 atm and therefore are unsafe for implantable applications. They are currently used in aerospace applications2.

A low pressure nickel-hydrogen battery that is safe is proposed in section 1.4 and

is developed as an alternative technology to high pressure nickel-hydrogen battery.

Ni-H2 Volumetric Volumetric

Gravimetric

Figure 1.2.1: Comparison of the different battery technologies in terms of volumetric and gravimetric energy density 1. The energy density of the nickel-hydrogen batteries was added to the figure.

1.3. Overview of High Pressure Nickel-Hydrogen Batteries

1.3.1. Introduction

Nickel-hydrogen battery is a hybrid combining the battery and fuel cell

technologies. The nickel hydroxide positive electrode is similar to that used in battery

technologies like nickel-cadmium, nickel-metal hydride and nickel-zinc batteries. The

platinum negative electrode is similar to that used in alkaline fuel cells. The remarkable

features of the battery, as discussed earlier, are the long cycle life compared to other

battery systems and their tolerance towards overcharge and reversal (over-discharge)

leading to inherent safety. The state of charge is determined easily by measuring the

hydrogen pressure. All these characteristics have resulted in their use in satellites both

geosynchronous earth-orbit (GEO) and low earth-orbit (LEO) applications for many

years 2, 3.

1.3.2. Cell Components and Electrochemistry involved

The different components in the nickel-hydrogen cell are the positive nickel

electrode, the negative platinum electrode and the separator which is filled with

potassium hydroxide solution. The components of the nickel-hydrogen cell developed by

Communication Satellite Corporation are shown in Fig. 1.3.1

Positive Electrode

The commonly used nickel electrode in high pressure nickel-hydrogen batteries is

the sintered type electrode that consists of a sintered porous nickel plaque impregnated

with nickel hydroxide active material 2, 4. The properties of this electrode are high porosity, large surface area, high electrical conductivity and good mechanical strength.

Other fabrication methods of the nickel hydroxide electrode are discussed in Chapter 5.

The active material is impregnated into the plaque by electrochemical process – precipitation of nickel hydroxide from a solution containing nickel nitrates5, 6.

Negative Electrode

The negative electrode (also called the hydrogen electrode) is made based on

techniques developed for gas electrodes in alkaline fuel cells. They typically consist of

Teflon bonded with platinum particles on a nickel substrate and are described in detail in chapter 3.

Separator

The main function of the separator is to serve as reservoirs for the electrolyte through which ionic conduction occurs. It should also be dimensionally stable on long-

term storage and cycling. The separators typically used in the high pressure Ni-H2 cells are (1) asbestos and (2) Zircar 7, 8. Most of the first nickel-hydrogen cells used asbestos

separators and they were replaced by Zircar because of asbestos’s toxic nature. Zircar is

the trade name for woven zirconium oxide fibers stabilized with yttria.

Electrolyte

The electrolyte used is 26 wt% KOH solution in comparison to the 31 wt% KOH

solution typically used in nickel-cadmium batteries. The use of 26% KOH solution has

7, 8 significantly increased cycle life of the Ni-H2 batteries by reducing electrode swelling .

Reactions Involved

During charge, nickel hydroxide is oxidized to nickel oxyhydroxide at the positive

electrode according to equation [1.1]. The reverse of the reaction occurs during discharge.

−−charg e NiOH()+++ OHZZZZZX NiOOH HO e [1.1] 22YZZZZZdischarg e

At the platinum negative electrode, hydrogen is formed by electrolysis of water during

charge and the hydrogen gas is oxidized during discharge according to equation [1.2].

−−charg e HO++ eZZZZZX 1 H OH [1.2] 22YZZZZZdischarg e 2

The net reaction in the cell during charge and discharge is:

charg e Ni() OHZZZZZX NiOOH+ 1 H [1.3] 22YZZZZZdischarg e 2

Therefore the net effect is the change in the oxidation state of nickel and the formation or

the depletion of hydrogen gas depending on charge or discharge. The hydroxyl ion and

the water concentration remains the same in the cell.

Overcharge

When the cell is overcharged, oxygen is evolved at the nickel hydroxide positive

electrode (equation [1.4]) whereas there is no change in reaction at the platinum negative

electrode (equation [1.5]). The evolved oxygen recombines with hydrogen at the

platinum electrode according to equation [1.6]. Because of this recombination, there is no

change in the hydroxyl ion and water concentration in the cell 2.

−−+ 1 22OH⎯⎯⎯⎯⎯ve, electrode→+ O H O + e [1.4] 2 22

+−− ⎯⎯⎯⎯⎯−ve, electrode→+ 22HO22 e H 2 OH [1.5]

1 HOHOheat+ ⇒ + [1.6] 2222

Over-discharge

When the cell is over-discharged, hydrogen is generated at the positive electrode

(equation [1.7]) and is consumed at the negative platinum electrode (equation [1.8]), without any change in pressure in the nickel-hydrogen cell 2. The net amount of KOH and

water in the cell does not change on over-discharge. Therefore these batteries are called

low maintenance batteries.

+−− ⎯⎯⎯⎯⎯+ve, electrode→+1 HO22 e2 H OH [1.7]

1 +−− ⎯⎯⎯⎯⎯−ve, electrode→+ 2 HOH22 HOe [1.8]

Figure 1.3.1: The components of COMSAT nickel-hydrogen cell in stacking arrangement2.

1.3.3. Nickel-Hydrogen Cell Construction

The nickel-hydrogen cell is assembled inside a cylindrical pressure vessel made of Inconel alloy 718 to withstand high pressures of hydrogen formed during charging, typically 80 atm pressure. The most common type of these pressure vessels are referred to as individual pressure vessels (IPV) because each individual cell is contained within its

own pressure vessel. The electrodes in the cell are stacked together as shown in Figs.

1.3.1 and 1.3.3 and are connected in parallel. Therefore the nominal voltage of these

individual pressure vessels (Fig. 1.3.2) is only 1.25 V. The IPV cells are then arranged in series to get the desired voltage. Two commonly used constructions based on the IPV cells are Comsat and Air Force cell design.

Comsat Ni-H2 Cell

The cross-sectional view of the Comsat nickel-hydrogen cell8 is shown in Fig.

1.3.2. The electrodes are stacked together in a back to back design as shown in Fig. 1.3.1.

The two nickel hydroxide electrodes are placed back to back. A separator soaked in KOH solution separates the nickel hydroxide and the negative platinum electrodes. A gas diffusion screen is positioned on the back of the negative electrodes to facilitate gas diffusion. A complete stack of electrodes arranged in parallel are shown in Fig. 1.3.2. It is noted that the bus bars for the positive and the negative electrodes are located along the outside of the electrode stack. The oxygen evolved during overcharge at the nickel hydroxide electrode diffuses out at the back of these electrodes and then recombines at the negative electrode.

Figure 1.3.2:Cross-sectional view components of Comsat cell in back to back design 8.

Air Force Ni-H2 Cell

The first Air Force nickel-hydrogen cells were based on recirculating electrode

stack design. In these cells, the recombination is carried out in a separate module

containing the hydrogen electrode placed at the end of the stack. The water is then

recirculated back to the electrodes using a wall wick made of zirconium oxide. The newer

Air Force cells utilize the back to back design employed in Comsat cells. The electrode

components in the stack are shaped like a pineapple slice7 as shown in Fig. 1.3.3. A

polysufone central core is used to assemble the cell components. The electrodes are

electrically connected through tabs running through the central core.

Figure 1.3.3: Exploded view showing the components of an Air Force cell in stacking arrangement7.

Advanced Cell Configurations

Other advanced battery designs developed to improve the gravimetric and volumetric

energy densities in comparison to the individual pressure vessel cells are (i) common

pressure vessel Ni-H2 battery and (ii) bipolar Ni-H2 battery.

In the common pressure vessel Ni-H2 battery, the individual cells are connected in

series and are contained within one common pressure vessel. This design results in higher volumetric energy density and lower manufacturing cost. These batteries were designed and developed by Comsat and Johnson Controls, Inc. In bipolar Ni-H2 battery, a bipolar

plate contacts two cells in series and were designed and developed at NASA Lewis

Research Center (NASA Glenn Research Center).

1.4. Proposed Research

For the implantable system, it is necessary to develop a rechargeable battery with

long cycle life that is reliable and safe.

The existing lithium ion rechargeable implantable batteries available form

companies Wilson Greatbatch and Quallion were investigated as near-term solution for

powering the NNPS. Initial tests were conducted on Sony rechargeable batteries to establish the conditions for performance testing. The Wilson Greatbatch and Quallion

implantable batteries were then tested and compared.

In order to meet the goal of long cycle life, a low pressure nickel-hydrogen

battery were designed and developed. The metal hydride, palladium treated LaNi4.3Al0.7,

developed by Prof. Payer and his co-workers 9-11, Material Science Department, Case

Western Reserve University will be used as a hydrogen storage module thereby allowing

the battery to operate at near atmospheric pressures. The cycle life performance of the

high pressure nickel-hydrogen batteries will be duplicated in our low pressure batteries.

The metal hydride is used only as hydrogen storage device and not as an electrode as in

the case of the nickel-metal hydride batteries. This prevents metal hydride corrosion that

limits the performance of the Ni-MH batteries and is discussed in detail in Chapter 7. The

battery developed will operate at atmospheric pressures with long cycle life.

1.5. Prior Work

The use of the metal hydride as a hydrogen storage device in nickel-hydrogen

battery was conceptualized and patented by Dunlop, Martin and Van Ommering12 of

Communications Satellite Corporation in 1976. This led to the subsequent development of nickel-metal hydride batteries which uses metal hydride as the negative electrode instead of the platinum electrode used in high pressure nickel-hydrogen batteries.

However the cycle life of Ni-MH batteries is lower than that of the Ni-H2 batteries

because of the oxide layer formation on the metal hydride and particle pulverization

(chapter 7, section 7.1).

Recently, Ergenics Inc (HERA USA Inc.) have been developing a low pressure

nickel-hydrogen battery using a metal hydride13. However, no literature is available on

the assembly as well as the performance of the low pressure nickel-hydrogen battery.

Additionally, the estimation of the state of charge is no longer straightforward for the low pressure Ni-H2 battery because the pressure change is non-linear in comparison to the

high pressure Ni-H2 battery. The mechanisms like oxygen evolution and the

recombination reaction are critical issues that determine the cycle life of the nickel-

hydrogen battery. It is noted that there is no literature available on these aspects for the low pressure Ni-H2 battery.

1.6. Objective and Overview of this work

1.6.1. Primary Objective

The primary objective of this research is to design and develop a low pressure

nickel-hydrogen battery that is safe and has long cycle life. The important goals under

this objective are

1. Fabrication and selection of cell components.

a. The performance and the activation of the platinum electrode from E-tek,

Inc. are discussed in chapter 3. The different current collectors and the

separators were evaluated in the nickel-hydrogen cell (chapter 3).

b. Nickel hydroxide electrode from a D size Ni-MH battery was used a guide

in the design and development of nickel-hydrogen cell (chapter 4).

c. The fabrication and formation of the pasted type nickel hydroxide

electrode was studied and is discussed in detail in chapter 5. The effect of

electrode components like the binder and the conductive filler content on

utilization and coulometric efficiency were investigated (chapter 5).

2. Assembling single cells and testing them for cycle life. Pressure data were

analyzed to understand the oxygen evolution, the recombination reaction and the

self discharge mechanism of the cell. The coulometric efficiency and the state of

charge of the Ni-H2 cell was predicted by fuel gauging (chapter 6).

3. Incorporating the metal hydride to store hydrogen gas resulting in a low pressure

nickel-hydrogen battery. Pressure data were analyzed to understand the effect of

metal hydride deactivation on the cell performance. The hydrogen fuel was

gauged based on the pressure composition isotherm of the metal hydride and was

used to predict the state of charge of the cell (chapter 7).

The development of the nickel-hydrogen cell was the long term solution to powering the

Networked Neuroprosthetic System.

1.6.2. Secondary Objective

The secondary objective of my dissertation was to maximize the cycle life of the implantable lithium ion cells – Quallion and Wilson Greatbatch and to choose one of them as short term solution for powering the NNPS. The performance of these implantable cells is compared to the Sony lithium ion cells (not implantable grade). The

effect of various factors like depth of discharge, charge cut off voltage, temperature,

charge rate and discharge rate on cycle life were investigated. All of these parameters are

discussed in detail in Chapter 2.

References

1. J. M. Tarascon and M. Armand, Nature, 414, 359 (2001).

2. D. Linden and T. B. Reddy, eds., Handbook of batteries, McGraw-Hill, New York, 2002.

3. J. D. Dunlop, G. M. Rao, and T. Y. Yi, NASA handbook for nickel-hydrogen batteries, National Aeronautics and Space Administration, Scientific and Technical Information Branch, Washington, DC, (1993).

4. A. Fleischer, Journal of the Electrochemical Society, 94, 289 (1948).

5. R. L. Beauchamp, (Bell Telephone Laboratories, Inc.). Application: US 3653967, (1972).

6. D. F. Pickett and J. T. Maloy, Journal of the Electrochemical Society, 125, 1026 (1978).

7. L. H. Thaller and A. H. Zimmerman, Nickel-hydrogen life cycle testing: review and analysis, Aerospace Press, El Segundo, California, (2003).

8. L. H. Thaller and A. H. Zimmerman, Overview of the design, development, and application of nickel-hydrogen batteries, National Aeronautics and Space Administration, Glenn Research Center, Cleveland, Ohio, (2003).

9. X. Shan, Ph.D. dissertation, Material Science Dept., Case Western Reserve University, Cleveland, Ohio., 2004.

10. X. Shan, J. S. Wainright, and J. H. Payer, Journal of Alloys and Compounds, in press.

11. X. Shan, J. S. Wainright, and J. H. Payer, Journal of Alloys and Compounds, in press.

12. J. D. Dunlop, M. W. Earl, and G. Van Ommering, (Communications Satellite Corp., USA). Application: US 3959018, (1976).

13. M. Golben, K. Nechev, D. H. Dacosta, and M. J. Rosso, Annual Battery Conference on Applications and Advances, 12th, Long Beach, Calif., Jan. 14-17, 1997, 307 (1997).

2. Cycle Life Tests on Lithium Ion Batteries for Implantable Applications

2.1. Overview of Lithium Ion Batteries

Sony corporation introduced lithium ion battery technology 1 in 1990 using

graphitic carbons as negative electrodes to intercalate the lithium, replacing lithium metal

and overcoming the problems encountered when lithium metal was present. Lithium

metal cannot be safely used in rechargeable battery applications because of dendritic

deposition 2, shorting and reaction of the electrolyte with metallic lithium resulting in

thermal runaway 3.

The amount of electrical energy, expressed either per unit weight (Wh/kg) or per

unit volume (Wh/l), that a battery is able to deliver is a function of the cell voltage (V) and capacity (Ah/kg). The volumetric and gravimetric energy densities of different battery technologies are compared in Fig. 1.2.1. Lithium ion batteries have high cell voltage (~ 3.5V) and excellent negative electrode capacity (∼ 372 Ah/kg of graphite

mass). Therefore the lithium ion batteries have large gravimetric energy density (~150

Wh/kg) and large volumetric energy density (~300 Wh/l) making it lighter and smaller

compared to other battery technologies. Because of their high energy density and the ease

of minimizing battery size, they are ideal for portable equipment like laptop computers,

digital video cameras, cellular phones and other hand held devices. Li-based batteries

currently account for about 63% of worldwide sales values in portable batteries. Lithium

ion batteries are now being tested for use in aerospace, bio-implantable, and hybrid

electric vehicle applications.

16 2.2. Description of Li Ion Cell

A schematic representation of the lithium ion cell is shown in Fig. 2.2.1. The negative electrode consists of porous graphite connected to a copper current collector and the positive electrode consists of cobalt oxide or manganese oxide in contact with an aluminum current collector. A separator foil impregnated with a liquid electrolyte solution containing lithium ions or a polymer electrolyte separates the electrodes. During the charging process, intercalated lithium is electrochemically oxidized at the cobalt oxide positive electrode to form lithium ions:

→ + + + − LiCoO2 Li1− yCoO2 yLi ye [2.1]

Simultaneously, the lithium ions are electrochemically reduced and intercalated into the

graphite electrode matrix:

+ + + − → Lix− yC6 yLi ye LixC6 [2.2]

During the first charging cycle, solution species next to the graphite electrode are reduced at about 1.5 V versus Li/Li+ to form a film, the “solid electrolyte interface” (SEI) 4, 5,

which, effectively passivates the lithium intercalation electrodes (graphite/carbon)

allowing only lithium ion migration while preventing insertion of electrolyte into the

graphite 6-9. The stability of the SEI film 5 is very important to passivation as it prevents

exfoliation and dendritic growth of lithium. The thickness of this solid electrolyte

interphase film 5 is limited by the tunneling range of the electrons to about 300 nm. The capacity fade of these batteries upon cycling is due to the impedance buildup of this film and is discussed in detail in the later section of this chapter.

17 Charger

electrons current SEI

e- Li Li e- Cu current Al current Li collector Li collector Li e- e- Li Li

Negative Separator Positive electrode: with liquid electrode: electrolyte Graphite LixCoO2

Figure 2.2.1: Schematic representation of a lithium ion battery during charging. The separator is impregnated with liquid electrolyte.

2.3. Implantable Li Ion Batteries

A sample of lithium ion battery manufacturers for portable applications includes

Sony, Sanyo, Panasonic and Matsushita. The potential use of lithium ion batteries for

automotive applications has been explored and these batteries are developed by

specialized manufacturers (e.g. Johnson Controls Inc.) for hybrid electric vehicle

applications.

Quallion and Wilson Greatbatch are manufacturers of lithium ion batteries for

implantable biomedical applications. Lithium ion batteries from these manufacturers

were tested and compared in this chapter. The Wilson Greatbatch Cells, (WG) R10398

are prismatic cells and are hermetically sealed in a stainless steel case. The nominal

capacity of the Wilson Greatbatch Cells is 1.4Ah. The Quallion cells, model QL0200I-A,

are also hermetically sealed in a titanium case. Some of the tests were started by Mike

Haugh10 and were continued by the author. Though the Quallion cells were rated at

18 nominal capacity of 200 mAh, Haugh had tested them based on a capacity of 225 mAh.

Therefore the test results for Quallion are based on a capacity of 225 mAh. Table 2.1

compares the weight, volume and capacity of Quallion and Wilson Greatbatch cells. The

gravimetric capacity density of Quallion is twice that of the Wilson Greatbatch indicating less weight for the same capacity. However, the volumetric capacity density is the same for both the technologies.

Table 2.1: Comparison of Quallion and Wilson Greatbatch lithium ion cells Manufacturer Nominal Weight Volume Capacity Capacity type capacity (Ah) (g) (cc) density (Ah/kg) density (Ah/cc) Wilson 1.4 60 22.6 23.33 6.2 x 10-2 Greatbatch Quallion 0.2 8 3.3 50 6.1 x 10-2

2.4. Objective

The objectives of the lithium ion cell testing are:

1. To identify conditions for increased cycle life. The parameters considered are

the charge rate, the discharge rate, the charge voltage cut-off and the depth of

discharge.

2. To study the effect of temperature on cycle life.

3. To predict the cycle life of the cells based on fitting the experimental data.

4. To compare the implantable technologies – Wilson Greatbatch and Quallion

and to choose one of the two for our implantable applications.

2.5. Charging Protocol for Lithium Ion Batteries

The lithium ion battery charging occurs in two stages as shown in Fig. 2.5.1. The

battery is charged first at constant current until the cell voltage reaches the upper limit of

19 4.1 or 4.2 volts, followed by constant voltage charging until the current drops to the

prescribed value provided by the manufacturer. The constant current charging time (stage

1) and constant voltage charging time (stage 2) depend on the charge rate. For typical 1C

rate of charging, the charge time for stage 1 (constant current charging) is about 0.85 h,

by which time about 85 % of the battery is charged. The 2nd stage (constant voltage

charging) takes about 1 h or less and extends the charging time of the batteries. The total

charge time and the choice of constant voltage in stage 2 depend on the chemistry used in

the lithium ion cell.

STAGE 1 STAGE 2 1.4 (A) Current (A) 0 0 tT Time (h)

4.2

3.5 (B) Cell Voltage (V) Cell Voltage 0 tT Time (h)

100 85 (C) Capacity (%) 0 0 tT Time (h)

Figure 2.5.1: Schematic of DC charge characteristics. During the initial stage (stage 1), charging is done at a constant current, 1.4 A for Sony cell 18650 (Capacity – 1.4 Ah, 1 C charge rate) (A) and the voltage is rising (B). tT designates the time at which the voltage reaches a value of about 4.2 V and the charging mode switches from constant current to stage 2, where charging continues at constant voltage while the current is continuously decreasing. About 85 % of the capacity is charged in stage 1 and the rest, 15 %, is charged in stage 2.

20 2.6. Experimental Design

Lithium ion batteries from three different manufactures Sony, Wilson Greatbatch

LTD., and Quallion were tested. The implantable lithium ion cells Quallion and Wilson

Greatbatch are compared to the standard Sony cell 18650, which is 18 mm in diameter

and 65 mm tall. This cell was used as a reference as it is the most commonly used cell.

The Sony 18650 cells were obtained from Sony NP-F550 battery packs designed for use

in digital cameras. The cells were labeled as 1.5 Ah capacities though in the course of

this research it has been found that the nominal capacity is closer to 1.4Ah. This is fairly

common as the labeled capacities of the cell are typically rated for very slow discharge

currents (around 1/5 of the labeled battery capacity or slower). In this research most of

the discharge currents are much higher than this so the batteries were never able to

provide their rated capacity.

The lithium ion cells were tested using an Arbin BT2000 test station. It is a

general purpose testing equipment for battery testing, life cycling, and material research.

The Arbin test station has 20 channels. Each channel is a fully functional

potentiostat/galvanostat and covers very wide current, voltage and power ranges. A sequence of steps can be written in a program, which is called a schedule, to test the battery. Current and voltage of the battery is monitored throughout the test.

The implantable batteries, Quallion and Wilson Greatbatch were tested at 37o C to

simulate the human body temperature. A heated bath with silicon oil was used to

maintain the required temperature of the batteries. All the Sony cells were tested at room

temperature except one which was tested at 37o C (in the oil bath) to determine the effect

of increased temperature on the battery performance.

21 The lithium ion cells were cycled at different charge and discharge conditions and

other variables were held constant. The effect of various variables on cycle life was

studied:

1. Different charge rates – C, C/5.

2. Different discharge rates – C, C/5, 2W. In addition to constant current discharge,

the effect of constant power discharge on cycle life was monitored.

3. Different charge voltage cut off – 4.1, 4.2 V.

4. Different depths of discharge – 100, 80, 60 %.

The depth of discharge (DOD) of a battery is defined as the amount of the

battery’s capacity used during the discharge phase. The DOD is based on the nominal

capacity of the cells and is given in percentages (for example 80% DOD of 1.4Ah cell is

1.12Ah). In addition to the above variables, the effect of temperature (37o C) on the cycle life of the lithium ion cell was analyzed. The capacity fade data was then fitted to a model to predict cycle life of the lithium ion cells under a given set of conditions. The Quallion and WGB lithium ion chemistries were compared.

The steps involved in a typical test cycle were as follows:

1. Rest at open circuit for 5 minutes.

2. Charge the lithium ion cell.

a. Charge at constant current until the cell voltage reaches 4.1V

b. Charge at constant voltage, 4.1V, until the current drops below 4% of the

cell’s rated current at C rate.

3. Rest at open circuit for 5 minutes

22 4. Discharge at constant current. The discharge is terminated based on a depth of

discharge value or a voltage cutoff of 2.75 V

2.6.1. Effect of Charge Rate on Cycle life Performance

The effect of different charge rates on the capacity fade (capacity degradation) of

Sony lithium ion 18650 cells with cycling was monitored and studied. Three Sony lithium ion cells (denoted as Cells 1, 2, 3) were cycled at C charge and C discharge rate and one Sony lithium ion cell was cycled at C/5 charge and C discharge rate (denoted as

Cell 4) as listed in Table 2.2. All other parameters were held constant (Table 2.2).

The discharge capacity of ‘cell 1’ (charged and discharged at C rate) as a function of cycle number is shown in Fig. 2.6.1. The discharge capacity of this cell averaged over the first 5 cycles is 1.44 Ah and is in agreement with the manufacture’s rated capacity of

1.5 Ah. The discharge capacity of the lithium ion cell decreases with cycling and this phenomenon is commonly referred as the capacity fade. The capacity fade is rapid for the first 180 cycles and then slows down considerably. It is noted that the data points are scattered in the cycle number range of 1-100. These scattered points were caused by instrumentation error and are due to a mis-configured setting in the Arbin unit causing cross talk between the channels. This cross talk occasionally caused early termination of tests. Therefore a charge or a discharge step was stopped before completion leading to either a higher or a lower value recording of discharge capacity. Data point for every cycle is shown in Fig.2.6.1. In all the following figures, only one data point for every 10 cycles will be displayed to avoid congestion.

23 Table 2.2: Description of Sony lithium ion cells subjected to different test conditions.

Cell No. Charge Discharge Charge voltage Temperature Depth of rate rate cutoff, V discharge, % Cell 1 C C 4.1 RT 100 Cell 2 C C 4.1 RT 100 Cell 3 C C 4.1 RT 100 Cell 4 C/5 C 4.1 RT 100 Cell 5 C C/5 4.1 RT 100 Cell 6 C 2W 4.1 RT 100 Cell 7 C 2W 4.2 RT 100 Cell 8 C C 4.1 RT 80 Cell 9 C C 4.1 37o C 80

1.6

1.5

1.4

1.3

1.2

1.1 Discharge capacity (Ah) 1.0 0 500 1000 1500 2000 Cycle index

Figure 2.6.1: Discharge capacity as a function of cycle index for Sony Li ion cell 1. The cell loses discharge capacity with cycles. The cell was charged and discharged at C rate. The scattered points in the cycle number range 1-100 is due to instrumentation error and does not reflect the battery’s discharge capacity.

1.5

C = 1.38 -1.5x10-4(cycle) 1.4 d

1.3

1.2

1.1 Discharge capacity (Ah) 1.0 0 500 1000 1500 2000 Cycle index

Figure 2.6.2: Discharge capacity as a function of cycle index for Sony lithium ion cell 1 fitted to a straight line. The linear fit is poor for discharge capacity in the cycle number range 1-200. The capacity loss is 1.5x10-4 Ah/cycle. The cell was charged and discharged at C rate.

In order to compare the test results of the lithium ion cells subjected to different

conditions, it is necessary to identify parameters relevant to the capacity fade. In the

previous work at Case Western Reserve University by Haugh 10, linear fits were used to

model the capacity fade data. A linear fit to the discharge capacity data following

Haugh’s work is shown in Fig.2.6.2. The equation of the straight line fitted is

=+ CabNd where Cd is the discharge capacity and N is the cycle number. The

parameter ‘a’ is the estimated initial discharge capacity and parameter ‘b’ is the estimated

capacity loss per cycle. It is evident from Fig. 2.6.2 that the fit is poor for cycle numbers

in the range 1 to 200 where the capacity fade rate is high. Additionally the initial discharge capacity, ‘a’ estimated by the linear fit is 1.38 mAh and is less than the nominal capacity of the lithium ion cell. The capacity loss per cycle, ‘b’ is 1.5x10-4 Ah/cycle.

25 A linear fit of the capacity data over the first 200 cycles is shown in Fig. 2.6.3.

The estimated capacity loss per cycle is 4x10-4 Ah and is 2.7 times larger than that

predicted by the linear fit over the entire cycle number range. Based on the capacity loss

estimated, the linear fit model predicts the cell will reach 80% of its capacity (1.12 Ah) in about 775 cycles. In comparison, the experimental data (Fig. 2.6.1) shows that the cell reaches 80% of its capacity (1.12 Ah) in 1735 cycles. This suggests that a linear fit does not predict the cycle life of the lithium ion cells accurately.

1.6

1.5 C = 1.43 -4x10-4N d

1.4

1.3

1.2 Discharge capacity (Ah)

0 50 100 150 200 Cycle index

Figure 2.6.3: Discharge capacity as a function of cycle index for Sony Li ion cell 1 fitted to a straight line for the first 200 cycles. The estimated capacity loss per cycle is 4x10-4 Ah. The cell was charged and discharged at C rate.

The capacity fade of lithium ion batteries has been studied by numerous

researchers11-39. Most of the studies were focused on estimating cycle life of these

batteries. Spotnitz 12 has reviewed the different experimental studies on cycle life. The

capacity fade and the ageing of these batteries have also been modeled12-14, 23, 28, 33.

26 The capacity fade with cycles is non-linear and a model to explain the non-linear behavior was proposed by Broussely et al19. They attributed the capacity loss to the growth of solid electrolyte interphase film (SEI) on the electrodes. The model shows that the capacity loss varies with the square root of float time (idle or shelf time). This capacity loss relates to the calendar life of the battery, i.e., degradation of battery in storage. Inoue et al.25 showed that the capacity loss due to cycling is linear. However, no

attempt is made to separate the capacity losses due to the cycle life and the calendar life.

A nonlinear equation which describes the discharge capacity variation with the square

root of cycle number was fit to the experimental data as shown in Fig. 2.6.4. The equation

=+ 0.5 is CkkNd 12 , where k1 is the estimated initial discharge capacity and k2 is the

degradation constant. The equation fits the experimental data well with R2 (correlation

coefficient) value of 0.95. The estimated initial capacity, k1 is about 1.47 Ah and the

-3 estimated degradation constant, k2 is 8.02x10 Ah/ cycle . The errors in the estimation

of k1 and k2 are small and are listed in Table 2.3. The capacity fade data for cells 2 and 3

(C charge and C discharge) with their non-linear fits are shown in Figs. 2.6.5 and 2.6.6.

The fitted parameters k1 and k2 are listed in Table 2.3. The estimated discharge capacity,

k1, and the degradation constant, k2 are similar for the cells 1, 2 and 3 that were tested in

the same conditions. The degradation constant, k2, for the cells are within 3% of the

average value.

1.5

1.4 C = 1.47 -8.02x10-3(cycle)0.5 d 1.3

1.2

1.1 Discharge capacity (Ah) 1.0 0 500 1000 1500 2000 Cycle index

Figure 2.6.4: Discharge capacity as a function of cycles for Sony Li ion cell 1 fitted to a -3 non-linear curve. The parameters k1 and k2 are 1.47 Ah and 8x10 Ahcycle/ . The cell was charged and discharged at C rate.

1.5

C = 1.47 -7.22x10-3(cycle)0.5 1.4 d

1.3

1.2

Discharge capacity (Ah) 1.1

0 500 1000 1500 2000 Cycle index

Figure 2.6.5: Discharge capacity as a function of cycles for Sony Li ion cell 2 fitted to a -3 non-linear curve. The parameters k1 and k2 are 1.47 Ah and 7.2x10 Ah/ cycle . The cell was charged and discharged at C rate.

1.5

1.4 C = 1.48 -8.13x10-3(cycle)0.5 d 1.3

1.2

Discharge capacity (Ah) 1.1

0 500 1000 1500 2000 Cycle index

Figure 2.6.6: Discharge capacity as a function of cycles for Sony Li ion cell 3 fitted to a -3 non-linear curve. The parameters k1 and k2 are 1.48 Ah and 8.1x10 Ah/ cycle . The cell was charged and discharged at C rate.

1.50 C = 1.5 -5.21x10-3(cycle)0.5 d 1.45

1.40

1.35

1.30

Discharge Capacity (Ah) Discharge Capacity 1.25

0 400 800 1200 1600 Cycle Index

Figure 2.6.7: Discharge capacity as a function of cycles for Sony Li ion cell 4 fitted to a -3 non-linear curve. The parameters k1 and k2 are 1.5 Ah and 5.2x10 Ah/ cycle . The cell was charged at C/5 rate and discharged at C rate.

29 Table 2.3: Comparison of parameters and errors estimated by fitting the discharge =+ capacity as a function of cycle life to equation Ckkcycled 12 for C and C/5 rate of charging. 2 Cell No. Charge rate k1, Ah Error in k1 k2 Ah/ cycle Error in k2 R Cell 1 C 1.47 1.30x10-3 -8.02x10-3 4x10-5 0.95 Cell 2 C 1.47 1.28x10-3 -7.22x10-3 4x10-5 0.94 Cell 3 C 1.48 1.41x10-3 -8.13x10-3 5x10-5 0.94 Cell 4 C/5 1.50 0.97x10-4 -5.21x10-3 5x10-5 0.94

The discharge capacity degradation of the lithium ion cell charged at C/5 rate and

discharged at C rate (cell 4) with cycling is shown in Fig. 2.6.7. The data was fitted using

=+ 0.5 the non-linear equationCkkNd 12 and the fitted parameters are listed in Table 2.3.

-3 The degradation constant k2 for this cell (charged at C/5 rate) is 5.21x10 and is smaller

in comparison to that obtained for the cells 1, 2 and 3 charged at C rate indicating that the

capacity fade is smaller when charged at C/5 rate. The discharge capacity of Cell 1 (C

charge rate) and Cell 4 (C/5 charge rate) with cycling are compared in Fig. 2.6.8. At 1500

cycles, the lithium ion cell cycled at C charge has lost about 17.9% of its capacity and the

lithium ion cell cycled at C/5 charge has lost only about 7.6% of its capacity. This clearly shows that the cycle life of lithium ion cell charged at C/5 rate is longer than that charged at C rate.

The energy efficiency of cell 1 (charged at C rate) and cell 4 (charged at C/5 rate)

are compared in Fig. 2.6.9. The energy efficiency is defined as the ratio of the discharge

energy to the charge energy. It is evident from the figure that the efficiency for the C/5 rate is larger than that of the C rate for all cycles. The difference in the energy efficiency increases with cycling. In the first few cycles the difference primarily results from the C rate charging (1.4 A) of cell 1 which results in large energy consumption and C/5 rate

charging (0.28 A) of cell 4 which results in small energy consumption. This results in 30 lower energy efficiency for the cell 1. However as the cells are cycled, the capacity fade of the cells are different. The capacity fade is larger for the C rate ‘cell 1’ and smaller for

the C/5 rate ‘cell 4’ as discussed earlier. This explains the increase in the difference of the energy efficiency between Cell 1 and Cell 4.

1.5 C C/5 1.4 C = 1.5 -5.21x10-3(cycle)0.5 d

1.3

1.2

C = 1.47 -8.02x10-3(cycle)0.5 d 1.1 Discharge capacity (Ah)

0 500 1000 1500 2000 Cycle index

Figure 2.6.8: Discharge capacity as a function of cycles comparing Sony Li ion cells 1 and 4. The capacity fade data is fitted to a non-linear curve. Cell 1 was charged at C rate and Cell 4 was charged at C/5 rate. The degradation constant is smaller for the lithium ion cell with C/5 charge in comparison to the lithium ion cell with C charge.

100

C 95 C/5

Energy efficiency, % 80

75 0 500 1000 1500 2000 Cycle Index

Figure 2.6.9: Energy efficiency as a function of cycles comparing Sony Li ion cells 1 and 4. Cell 1 was charged at C rate and Cell 4 was charged at C/5 rate. The energy efficiency of Cell 1 is always smaller than that of the Cell 4.

2.6.2. Effect of Discharge Rate on Cycle Life Performance

Three Sony lithium ion cells cycled at C rate charge and C discharge (cells 1, 2

and 3) discussed in the previous section will be compared with Sony lithium ion cells

cycled at C/5 (cell 5) and 2W (cell 6) discharge rates. The 2W discharge is roughly

equivalent to C/2.53 discharge rate. The other relevant parameters were fixed and are

listed in Table 2.2.

Lithium ion cells cycled at C rate charge and C rate discharge were fitted using a

non-linear equation and two fit parameters were estimated (section 2.6.1). The parameters

-3 - k1 and k2 averaged over the three lithium ion cell tests are 1.47 Ah and 7.8x10 Ah cycle

0.5 respectively. The lithium ion cells cycled at C/5 and 2W discharge are shown in Figs.

2.6.10 and 2.6.11, respectively. These cells were tested by Michael Haugh and were stopped after about 300 cycles. The initial discharge capacity of the cells tested at C/5 and 2W discharge rate (over the first few cycles) was about 1.34 Ah and 1.37 Ah

32 respectively. In comparison the initial discharge capacity of cells tested at C rate

discharge (cells 1, 2 and 3) was about 1.44 Ah. This drastic decrease in the initial

capacity before cycling is related to the calendar life of the cells. The lithium ion cells

tested at C rate discharge were fresh and hence their capacity fade (before cycling) is low.

However, the lithium ion cells tested at C/5 and 2W discharge rates were old and hence

their capacity fade (before cycling) was significant. This clearly illustrates the calendar

life issues of Sony 18650 batteries. It is noted here that the exact floating time (idle time)

of these batteries is not known and therefore only qualitative comparisons can be made.

The differences in these lithium ion cells are clearly illustrated in Fig.2.6.12.

The discharge capacity data of the cells cycled at C/5 and 2W discharge rates was

=+ 0.5 fitted to a non-linear equation CkkNd 12 (Figs. 2.6.10 and 2.6.11). The fitted

-2 - parameter k2 for lithium ion cells discharged at C/5 and 2W rate is 1.1x10 and 1.36 x10

2 Ah/ cycle , respectively and is larger that that of the C rate discharge (7.8x10-3

Ah/ cycle ). This indicates that the cells with long idle time undergo larger capacity

degradation during cycling and illustrates the effect of calendar life on the capacity fade

of the Sony lithium ion cells.

The energy efficiency of lithium ion cells discharged at C and C/5 rate is

compared in Fig. 2.6.13. The energy efficiency of the cell discharged at C rate is higher

compared to the cell discharged at the C/5 rate for the first 50 cycles and is due to

calendar life effect on these cells. However, in spite of poor calendar life, the cell cycled

at C/5 discharge rate has better energy efficiency in the range from 50 to 350 cycles. This indicates that cycling at C/5 discharge rate results in higher energy efficiency than cycling at C discharge rate.

33 1.40

1.35 C = 1.38 - 1.1x10-2(cycle)0.5 d,Ah

1.30

1.25

1.20 Discharge capacity (Ah)

1.15 0 100 200 300 400 Cycle index

Figure 2.6.10: Discharge capacity as a function of cycles for Sony Li ion cell cycled at C charge rate and C/5 discharge rate fitted to a non-linear curve. The parameters k1 and k2 are 1.38 Ah and 1.1x10-2 Ah/ cycle .

1.40

C = 1.4 - 1.36x10-2(cycle)0.5 1.35 d,Ah

1.30

1.25

1.20

1.15 Discharge capacity (Ah)

1.10 0 100 200 300 400 Cycle index

Figure 2.6.11: Discharge capacity as a function of cycles for Sony Li ion cell cycled at C charge rate and 2W discharge rate fitted to a non-linear curve. The parameters k1 and k2 are 1.38 Ah and 1.36x10-2 Ah/ cycle .

34 1.5 C 1.4 C/5 2W

1.3

1.2 Discharge capacity (Ah) 1.1 0 100 200 300 400 Cycle index

Figure 2.6.12: Discharge capacity as a function of cycles comparing lithium ion cells cycled at different discharge rates. The degradation constant, k2, is larger for the lithium ion cell with C/5 discharge in comparison to the lithium ion cell with C discharge.

91 C 90 C/5

Wh efficiency (%) 87

86 0 100 200 300 400 Cycle index

Figure 2.6.13: Energy efficiency as a function of cycles comparing lithium ion cells subjected to C and C/5 rate of discharge.

35 2.6.3. Effect of Charge Voltage Cutoff on Cycle Life Performance

In this section, two lithium ion cells cycled at C rate charge and 2W discharge are

considered. The depth of discharge was 100%. The cells were tested at room temperature

(25o C). One lithium ion cell was charged at 4.1V (Cell 6, Table 2.2) and the other

lithium ion cell was charged at 4.2 V (Cell 7, Table 2.2) following the constant current

charge phase. The objective of these two tests was to study the effect of different charge

voltage cut offs on cycle life.

The discharge capacity of the lithium ion cell with 4.1 voltage cut off is shown in

Fig. 2.6.11 and was discussed in detail in the previous section. The conclusion was that

the lithium ion cell tested was an old cell with capacity loss due to idling. The fitted

-2 parameters k1 and k2 are 1.4 Ah and 1.36x10 Ah/ cycle .

The discharge capacity of the cell with 4.2 voltage cutoff is shown in Fig. 2.6.14.

The discharge capacity of this cell (4.2 voltage cutoff) for cycle 1 is 1.63 Ah and in comparison the discharge capacity of fresh cells (Cells 1, 2 and 3) charged at 4.1 voltage cutoff is 1.44 Ah. This indicates that the constant voltage charging at 4.2 V after the constant current charging results in higher charge and discharge capacity, although the charging is stopped when current drops below 4% of the rated current at C rate similar to other lithium ion cell tests. The discharge capacity profile (Fig. 2.6.14) shows two different fades, one in the cycle number range 1-350 and the other in cycle number range

500 to 1200. The capacity fade in the cycle number range 1-350 was fit using the non-

=+ 0.5 linear equation CkkNd 12 as shown in Fig. 2.6.14. This indicates that the capacity

fade decays to the square root of cycle number and supports the mechanism of SEI

growth as the main contributor to capacity fade. The discharge capacity decreases rapidly

36 from 1.37 Ah at 400 cycles to 0.33 Ah at 1700 cycles. The average capacity loss in this

cycle number range is 8x10-4 Ah/cycle.

This test illustrates that charging the Sony lithium ion cell with 4.2V charge cutoff

(C rate charge, 2W ≈ C/2.5 rate discharge) causes rapid capacity fade after 400 cycles. In

comparison, Sony lithium ion cells with 4.1 V cutoff (C rate charge, C rate discharge)

results in long cycle life (1.08 Ah at 1923 cycles). This shows that the charge voltage cutoff should be 4.1 V rather than 4.2 V for good cycle life performance.

C = 1.66 - 1.25x10-2(cycle)0.5 1.6 d,Ah

1.2

0.8

0.4 Discharge capacity (Ah) 0.0 0 200 400 600 800 1000 1200 Cycle index

Figure 2.6.14: Discharge capacity as a function of cycles for Sony Li ion cell cycled at C charge rate with 4.2 voltage cut off and 2W discharge rate. The parameters k1 and k2 are 1.66 Ah and 1.25x10-2 Ah/ cycle for the first 350 cycles.

2.6.4. Effect of DOD on Cycle Life Performance

2.6.4.1. Sony Lithium Ion Cells at 100 and 80% DOD

Three Sony lithium ion cells (Cell 1, Cell 2 and Cell 3) cycled to 100% DOD and

two lithium ion cells (Cell 8, Cell 9) cycled to 80% DOD as listed in Table 2.2 are

37 compared in this section. Cell 9 was cycled at 37o C and other cells were cycled at room

temperature. The voltage cutoff during charging was 4.1 V.

In order to make a fair comparison of the different DOD tests, the cycle life of the

lithium ion cells cycled to 100% DOD was determined when the discharge capacity

dropped to 80% of the nominal capacity. For the 80% DOD lithium ion cells, the cycle

life was determined when the end of discharge voltage reached 2.75 V. The cycle life of

the cells cycled to different depths of discharge is shown in Fig. 2.6.15. The lithium ion

cell cycled to 80% DOD (Cell 8) has a longer cycle life than that of the lithium ion cells

cycled at 100% DOD. The cell cycled at room temperature (Cell 8) has a longer cycle life

than the cell cycled at 37o C. This result is discussed in more detail in the subsequent

section.

There is an important distinction in the comparison of the cells cycled to 100%

DOD and 80% DOD on the basis of cycle life. For the same number of cycles and same

charge and discharge rates, the total hours discharged by the 100% DOD lithium ion cell,

is more than that of the total hours discharged by the 80% DOD lithium ion cell. Taking

this into consideration, the total ampere hours discharged by the cells was calculated and is shown in Fig. 2.6.16. The total ampere-hours for the 80% DOD is larger than that of the 100% DOD. This shows that discharging the cell to 80% of its capacity (80% DOD) not only enhances cycle life but also increases the total ampere-hours obtained in comparison to discharging the cell to 100% of its capacity (100% DOD). It is also evident that the total ampere-hours obtained for the cell cycled at 37o C is less than that

obtained for the cell cycled at room temperature.

38 3500

3000

2500

2000

1500 Cycles

1000

500

0 100%, Cell 1 100%, Cell 2 100%, Cell 3 80%, Cell 8 80% Cell 9, 37 C Depth of discharge (%)

Figure 2.6.15: Total number of cycles obtained for Sony lithium ion cells cycled to different states of depth of discharge.

4000

3500

3000

2500

2000

1500

1000

Total ampere-hours 500

0 100%, Cell 1 100%, Cell 2 100%, Cell 3 80%, Cell 8 80% Cell 9, 37 C Depth of discharge (%)

Figure 2.6.16: Total number of ampere-hours obtained for Sony lithium ion cells cycled to different states of depth of discharge.

39 2.6.4.2. Wilson Greatbatch Lithium Ion Cells at 80 and 60% DOD

Wilson Greatbatch cells cycled to 80% and 60% DOD are compared here. The

cells were charged and discharged at C rate. The voltage cutoff during charge and

discharge was 4.1 and 2.75 V respectively. All the Wilson Greatbatch cells were tested at

37o C.

Cycle life of three cells tested to 80% DOD and one cell tested to 60% DOD is shown in Fig. 2.6.17. The total number of cycles was determined when the end of discharge voltage reached 2.75 V. The 60% DOD lithium ion cells have an order of magnitude higher cycle life than the 80% DOD lithium ion cells.

These tests show that cycle life can be significantly increased by discharging the battery to lower depths of discharge (DOD).

1600

1400

1200

1000

800

600 Total cycles 400

200

0 80%, Cell 1 80%, Cell 2 80%, Cell 3 60%, Cell 1 Depth of discharge (%)

Figure 2.6.17: Total number of cycles obtained for Wilson Greatbatch lithium ion cells cycled to 80 and 60% of depth of discharge.

40 2.6.5. Effect of Temperature on Cycle Life Performance

Sony lithium ion cells were cycled at two different temperatures, one cell (Cell 8)

at room temperature and another cell (Cell 9) at 37o C. These cells were charged and

discharged at C rate with 4.1 V charge cutoff and 2.75 V discharge cutoff. They were

discharged to 80% of their nominal capacity (80% DOD) every cycle. Figs. 2.6.15 and

2.6.16 show that the total cycles and the total-ampere hours obtained at 37o C is less than

that obtained at room temperature.

The end of discharge voltage as a function of cycle index at different temperatures

is shown in Fig. 2.6.18. The cycle life at 37o C is smaller than that at the room

temperature. The data for different temperatures was fitted linearly as shown in Fig.

2.6.18. The slope of the lines fitted to data at the room temperature and 37o C are 2.4x10-4

and 2.6x10-4, respectively. These values are not significantly different. The end of

discharge voltage for the first 175 cycles is shown in Fig.2.6.19. The end of discharge

voltage decreased rapidly in the first 50 cycles for the cell tested at 37o C and in

comparison the end of discharge voltage decreases linearly for the cell tested at room

temperature.

The data points for the cell tested at RT is scattered and therefore a much cleaner

data set would be required to compare the two cells based on their linear fits.

3.6 RT Linear fit (RT) 37 C 3.4 Linear fit (37 C)

EODV = 3.55 -2.4x10-4N 3.2

3.0 EODV (V) EODV = 3.49 -2.6x10-4N

2.8

2.6 0 1000 2000 3000 Cycle index, N

Figure 2.6.18: End of discharge voltage as a function of cycles comparing Sony Li ion cell cycled at room temperature and 37o C. The cells were charged and discharged at C rate to 80% DOD. The data was fitted to a straight line.

3.54 RT 3.52 37 C

3.50

3.48

EODV (V) 3.46

3.44

3.42 0 50 100 150 200 Cycle index

Figure 2.6.19: End of discharge voltage in the cycle number range 0-175 comparing Sony Li ion cell cycled at room temperature and 37o C. The cells were charged and discharged at C rate to 80% DOD.

42 2.6.6. Comparison of Sony, Wilson Greatbatch and Quallion

Lithium ion cells of different manufacturers Sony, Wilson Greatbatch and

Quallion were investigated at C/5 charge rate and C discharge rate. The charge and

discharge voltage cutoff are 4.1 and 2.75 V, respectively. The Sony lithium ion cell was

cycled at room temperature, however, the Wilson Greatbatch and Quallion cells were

cycled 37o C simulating human body temperature. The cells were discharged completely

(100% DOD). The discharge capacity of all the cells was fitted to the non-linear

=+ 0.5 equationCkkNd 12 .

The discharge capacity fade as a function of cycle index for the Sony lithium ion

cell (cell 4) was discussed in section 2.6.1 and is shown in Fig. 2.6.7. The initial discharge capacity of the cell is about 1.46 Ah. The non-linear fit parameters are listed in

Table 2.4. The discharge capacity degradation of the Wilson Greatbatch cell is shown in

Fig. 2.6.20. The discharge capacity of cycle 1 is 1.37 Ah. The discharge capacity data in the cycle number range 1-500 was fit to the non-linear equation. The fit parameters (k1,

k2) are listed in Table 2.4. This fit reinforces the SEI growth mechanism as the primary

factor contributing to capacity loss with cycling. The error in the fit parameters k1 and k2 is small. The discharge capacity drops rapidly after 500 cycles and only 47% capacity remains at 600 cycles. This suggests a different mechanism for capacity loss after 500 cycles relevant to the Wilson Greatbatch cell.

The capacity fade of the Quallion cell on cycling is shown in Fig. 2.6.21. The discharge capacity for the 1st cycle is 0.197 Ah. The capacity fade was fitted to the non-

linear equation for all cycles and the fit parameters are listed in Table 2.4. The correlation

fits the data well and the error in the fit parameters is small.

43 The number of cycles obtained for the lithium ion cells of different manufacturers

at 10% capacity loss was determined and is shown in Fig. 2.6.22. The total capacity loss

for the Sony cell was only 10% over the 1600 cycles tested and therefore the 10% capacity loss was used as the criteria to compare cells of different manufacturers. For the

Sony and Wilson Greatbatch battery a 10% capacity loss corresponds to 0.14 Ah and for the Quallion battery a 10% capacity loss corresponds to 22.5 mAh. The cycles obtained

from the Sony cell (2122 cycles) were an order of magnitude higher than that obtained

from the Wilson Greatbatch (28 cycles) and Quallion cells (99 cycles).

The number of cycles obtained for Wilson Greatbatch and Quallion cells at 10, 20

and 30% losses in capacity are compared in Fig.2.6.23. It is evident that the Quallion

cells provide more cycles than that of Wilson Greatbatch cells at different losses in

capacity. The difference in the cycles provided by Quallion and Wilson Greatbatch cells

also increases with capacity loss as shown in Fig.2.6.23.

The performance of these different cells is described by the fit parameter, k2. This

degradation constant, k2, for different lithium ion cell manufacturers is shown in Fig.

2.6.24 and is listed in Table 2.4. It is evident that the degradation constant is an order of

magnitude smaller for Sony and Quallion than that of Wilson Greatbatch. However the fit parameters are based on different nominal capacities of the lithium ion cell. The nominal

capacity of the Quallion cell is 0.225 Ah and is smaller than that of the Sony and Wilson

Greatbatch (1.4 Ah). Hence, the degradation constant, k2, was normalized by dividing with the nominal capacity of the lithium ion cells. The normalized degradation constant is the smallest for Sony, followed by the Quallion and the Wilson Greatbatch as shown in

44 Fig.2.6.24. This agrees well with the cycle life prediction: Sony > Quallion > Wilson

Greatbatch.

1.4

EODV = 1.4 - 2.7x10-2(cycle)0.5 1.2

1.0

0.8 Discharge capacity (Ah) 0.6 0 100 200 300 400 500 600 700 Cycle index

Figure 2.6.20: Discharge capacity as a function of cycles for Wilson Greatbatch Li ion - cell 4 fitted to a non-linear curve. The parameters k1 and k2 are 1.4 Ah and 2.7x10 2 Ah/ cycle for a fit to the first 500 cycles. The cell was charged at C/5 rate and discharged at C rate.

0.20 EODV = 0.2 - 1.5x10-3(cycle)0.5 0.18

0.16

0.14

Discharge Capacity (Ah) Discharge Capacity 0.12

0 500 1000 1500 2000 2500 Cycle index

Figure 2.6.21: Discharge capacity as a function of cycles for Quallion Li ion cell 4 fitted -3 to a non-linear curve. The parameters k1 and k2 are 0.2 Ah and 2.7x10 Ah/ cycle . The cell was charged at C/5 rate and discharged at C rate.

45 Table 2.4: Comparison of parameters and errors estimated by fitting the discharge =+ capacity as a function of cycle life to equation Ckkcycled 12 for lithium ion cells of different manufacturers charged at C/5 rate and discharged at C rate. 2 Manufacturer k1, Ah Error in k1 k2 Ah/ cycle Error in k2 R Sony 1.5 9.7x10-4 -5.2x10-3 3x10-5 0.94 Wilson Greatbatch 1.4 1.5x10-3 -2.7x10-2 9x10-5 0.99 Quallion 0.2 1.0x10-4 -1.5x10-3 2.7x10-6 0.99

2500 10% Capacity Loss

2000

1500

Cycles 1000

500

0 Sony WG Quallion Manufacturer type

Figure 2.6.22: Number of cycles obtained at 10% capacity loss for different lithium ion cell manufacturers Sony, Wilson Greatbatch and Quallion. The cells were cycled at C/5 charge rate and C discharge rate to 100% DOD.

46 1800 WG Quallion 1500

1200

900 Cycles 600

300

0 10 20 30 Loss in capacity (%)

Figure 2.6.23: Number of cycles obtained at 10, 20 and 30% capacity loss for Wilson Greatbatch and Quallion lithium ion cells. The cells were cycled at C/5 charge rate and C discharge rate to 100% DOD at 37o C.

3.0E-02 Rate constant Normalized rate constant 2.5E-02

2.0E-02 -0.5

1.5E-02 (cycle) 2 1.0E-02 k

5.0E-03

0.0E+00 Sony WG Quallion Manufacturer type

Figure 2.6.24: The degradation constant, k2, for different lithium ion cell manufacturers Sony, Wilson Greatbatch and Quallion. The normalized rate constants were obtained by ratioing the rate constant to the cell’s nominal capacity. The cells were cycled at C/5 charge rate and C discharge rate to 100% DOD.

47 2.6.7. Comparison of Sony and WG at 80% DOD

Two Sony lithium ion cells and three Wilson Greatbatch cells were charged and

discharged at C rates to 80% of their capacity (80% DOD) and are compared in this section. The charge and discharge cutoff voltages were 4.1 and 2.75 V respectively. Of the two lithium ion cells, one was cycled at room temperature and the other at 37o C. Fig.

2.6.25 compares the total number of cycles obtained from the Sony and Wilson

Greatbatch cells before the end of discharge voltage reached the cutoff value of 2.75 V.

As before, the Sony cells delivered many times more cycles.

3500 80% DOD 3000

2500

2000

1500

Total cycles 1000

500

0 Sony cell 1 Sony cell 2, WG cell 1 WG Cell 2 WG Cell 3 37 C Manufacturer type

Figure 2.6.25: Total number of cycles obtained for Sony and Wilson Greatbatch lithium ion cells cycled to 80 % of depth of discharge at C rate charge/ C rate discharge.

48 2.7. Comparison of Implantable battery technologies

In this section, the cycle life performances of the WG and Quallion batteries are compared.

2.7.1. Cycle life of Quallion and WG at 60% DOD

A WG cell and a Quallion cell were cycled at C rate charge and C rate discharge.

The voltage cutoff during charge and discharge were 4.1 and 2.75 V respectively. The cells were cycled to 60% DOD and were tested at 37o C, simulating body temperature.

The end of discharge voltage for every cycle was monitored.

The end of discharge voltage of the WG cell with cycling is shown in Fig.2.7.1.

The end of discharge voltage decreases with cycles linearly up to 1200 cycles and then

=+ drops sharply. The linear part of the data was fitted to a straight line, EODV Vo mN ,

where EODV and N are the end of discharge voltage and cycle number, respectively. Vo is the end of discharge voltage when N is zero and ‘m’ is the voltage drop per cycle. For

-3 the WG cell, the voltage drop per cycle, m, is 3.6x10 V/cycle and Vo is 3.56 V.

In comparison, the end of discharge voltage of the Quallion cell cycled in the same conditions is shown in Fig. 2.7.2. Two linear trends in the end of discharge voltage, one in the cycle number range 1-3500 and another in the cycle number range 4000-6000 cycles are observed. These two linear trends were fitted to straight lines and the fitted parameters are listed in Table 2.12. According to the linear fit in the cycle number range

-5 1-3500, the voltage drop per cycle, m, is 6x10 V/cycle and Vo is 3.32 V (Table 2.5). For

the linear fit in the cycle number range 4000-6000, the voltage drop per cycle is 1.6x10-4

V/cycle and Vo is 3.75 V. The voltage drop per cycle in the cycle number range 4000-

6000 is an order of magnitude higher than the voltage drop per cycle in the cycle number

49 range 1-3500. This indicates that the voltage drop shifts to a higher rate after a large number of cycles. In comparison, the voltage drop per cycle of the WG cell is two orders

of magnitude higher when compared with the voltage drop per cycle of the Quallion cell

in the same cycle number range (Table 2.5). This clearly indicates the superior performance of the Quallion cell at 60% DOD. The cycle life of the Quallion cell is about four times that of the Wilson Greatbatch cell at 60% DOD (Table 2.5).

The fit parameter, Vo, for the WG and Quallion lithium ion cell is 3.55 and 3.32

V. The difference of this value is 0.23 V. The high frequency resistance (solution

resistance) of the WG cell is about 0.12 ohms, whereas the high frequency resistance of

the Quallion cell is about 1 ohm. Therefore at C rate, the ohmic voltage drop of the WG

cell is 1.4 A x 0.12 Ω = 0.168 V and the ohmic voltage drop of the Quallion cell is 0.225

A x 1 = 0.225 V. The difference in these two ohmic voltage drops is (0.225-0.168) =

0.057 V which is considerably smaller than the difference in the Vo value of the two cells

and does not explain the difference in the end of discharge voltage. This suggests that the

value Vo of the two cells is related to the chemistry of the different lithium ion cells.

50 3.6 EODV = 3.56 -3.6x10-3N

3.4

3.2 EODV (V)

3.0

2.8 0 300 600 900 1200 1500 Cycle index

Figure 2.7.1: End of discharge voltage as a function of cycles for WG Li ion cell tested at 37o C. The cells were charged and discharged at C rate to 60% DOD. The data was fitted to a straight line over cycles 1 to 1200.

-5 3.4 EODV = 3.32 - 6x10 N

3.2

3.0 EODV (V)

-4 2.8 EODV = 3.75 - 1.6x10 N

0 1500 3000 4500 6000 Cycle index

Figure 2.7.2: End of discharge voltage as a function of cycles for Quallion Li ion cell tested at 37o C. The cells were charged and discharged at C rate to 60% DOD. The data was fitted to two straight lines.

51 Table 2.5: Linear fit parameters and their error estimations of the end of discharge voltage data. Wilson Greatbatch and Quallion lithium ion cells were charged and discharged at C rate to 60% DOD. Total cycles 2 Manufacturer Vo, V Error in Vo m, V/s Error in m R to EODV = 2.8V Wilson Greatbatch 3.56 5.4x10-3 -3.6x10-3 7.8x10-7 0.99 ~1500 Quallion, 1-3500 cycles 3.32 3.9x10-4 -6x10-5 1.9x10-7 0.96 ~6000 Quallion, 4000-6000 cycles 3.75 2.8x10-3 -1.6x10-4 5.6x10-7 0.97

2.7.2. Cycle Life of WG and Quallion at 100% DOD

The cycle life of Wilson Greatbatch and Quallion lithium ion cell at 100% DOD was tested using the Arbin test stand. The cells were cycled at C/5 charge and C discharge rates. The results of these two cells was analyzed and discussed in section

2.6.6. It was shown that the cycle life of Quallion cell was longer than that of the WG cell at 100% DOD for different capacity losses.

2.7.3. Long term tests on WG and Quallion

In this section, implantable lithium ion cells cycled at low charge and discharge rates (C/4, C/5) are compared and its effect on cycle life is analyzed. Two Wilson

Greatbatch cells were cycled at C/5 charge and discharge rate to 60% depth of discharge.

One Quallion cell was cycled at C/4 charge and discharge rate to 80% depth of discharge.

The charge and discharge cutoff voltage was set at 4.1 and 2.75 V respectively. All the cells were tested at 37o C.

The end of discharge voltage for the WG cell is shown in Fig. 2.7.3. The data was

=+ -5 fitted to the equation EODV Vo mN . The voltage drop per cycle, m, is 8x10 V/cycle.

The voltage drop for another WG cell tested at the same conditions is 7x10-5 V/cycle and is listed in Table 2.6. The Vo values for these two WG cells are 3.71 and 3.72 V respectively.

52 The end of discharge voltage variation for a Quallion cell charged and discharged

at C/4 rate to 80% DOD is shown in Fig. 2.7.4. The data was fitted the equation

=+ -5 EODV Vo mN and the voltage drop per cycle, m, is 9x10 V/cycle (Table 2.6). This

voltage drop per cycle for the Quallion cell at C/4 rate to 80% DOD is about the same as the voltage drop per cycle for the WG cell at C/5 rate to 60% DOD. Although the

Quallion cell was discharged deeper and at a higher rate than that of the WG cell the voltage loss per cycle is the same. This again reinforces the superior performance of the

Quallion cells.

3.8

EODV = 3.71 - 8x10-5(cycle) 3.7

3.6

EODV (V) 3.5

3.4 0 500 1000 1500 2000 2500 3000 Cycle index

Figure 2.7.3: End of discharge voltage as a function of cycles for WG Li ion cell tested at 37o C. The cells were charged and discharged at C/5 rate to 60% DOD. The data was fitted to a straight line.

53 3.40 EODV = 3.39 - 9x10-5(cycle)

3.35 EODV (V) 3.30

3.25 0 300 600 900 1200 Cycle index

Figure 2.7.4: End of discharge voltage as a function of cycles for Quallion Li ion cell tested at 37o C. The cells were charged and discharged at C/4 rate to 80% DOD. The data was fitted to a straight line.

Table 2.6: Linear fit parameters and their error estimations of the end of discharge voltage data. Wilson Greatbatch and Quallion lithium ion cells were charged and discharged at C/5 and C/4 rate, respectively. 2 Manufacturer Vo, V Error in Vo m, V/s Error in m R Wilson Greatbatch, Cell 1, C/5, 60% DOD 3.71 6.1x10-4 -8x10-5 3.9x10-7 0.94 Wilson Greatbatch, Cell 2, C/5, 60% DOD 3.72 3.4x10-4 -7x10-5 2.2x10-7 0.98 Quallion, C/4, 80% DOD 3.39 2.8x10-4 -9x10-5 3.8x10-7 0.98

2.8. Comparison of Charge Times for New and Used Quallion cells

The discussions of the previous section concluded that the Quallion cells are superior to the Wilson Greatbatch cells. In this section, the charge times of new Quallion cells to used Quallion cells are compared. A substantial increase in the charge time can be considered as a mode of failure. Therefore it is essential that the charging time of these cells does not increase significantly on cycling.

The cells (both used and new) were charged and discharged at different rates, C/5,

C/4, C/3, C/2, 3C/5, 3C/4 and C to 80% depth of discharge. The cells were first charged

54 at a constant current followed by a constant voltage charging at 4.1 V. This was followed

by a 5 minute rest period. Subsequently the cells were discharged at constant current. The

test results were analyzed and are discussed in this section. It is noted that the data values reported in this section were averaged over 5 cycles. The new Quallion cell was represented by a cell that was cycled 200 times at C rate charge and discharge to 60% depth of discharge. The used Quallion cell was represented by a cell that was cycled 6000 times at C rate charge and discharge to 60% DOD prior to the charge time tests.

The charge times of new and used Quallion cells at different charge rates are compared in Fig. 2.8.1. It is evident that the total charge time decreases with increase in charge rate (increase in charge current) as expected. For the slowest charging rate, C/5, the total charge time of the cells is less than 5 hours because the cell is only discharged to

80% of its capacity. For rapid charging rate, 1C, the total charge time is about 1.14h for the new cell and 1.45h for the used cell. It is evident that the total charge time of the used cells is always larger than that of the new cells for any given charge rate.

The total charge time is comprised of the constant current charging phase and constant voltage charging phase. It is therefore relevant to look at the constant current charge time and constant voltage charge time to understand the increase in the total charge time with cell use.

The percentage of charge time for the constant current phase and the constant voltage phase during charging for the new Quallion cell is shown in Fig. 2.8.2. The cell spends most of the charging time in the constant current phase for all charging rates except at the C rate. For C/5 rate of charging, the constant current charge time is about

85% and the constant voltage charge time is about 15%. For C rate of charging, the

55 constant current and the constant voltage charge times are nearly equal. At higher

charging rates the overpotentials are larger; the cell reaches 4.1 V sooner, and spends

more time charging in the constant voltage phase.

The percentage charge time spent in the constant current and the constant voltage

charging phase for the used Quallion cell is shown in Fig. 2.8.3. As discussed in the

previous paragraph, similar to the new cell results, the constant current charge time percentage decreases and the constant voltage charge time percentage increases with the

charging rate for the used cell. However for the 3C/5 charging rate, the constant current charging time is equal to the constant voltage charging time. For all charging rates greater than the 3C/5, the constant voltage charge time is greater than the constant current charge time. Comparing Figs. 2.8.2 and 2.8.3 for new and used cells, it is evident that the used cells spend more time in the constant voltage charging mode thereby increasing the total charging time. For example, at 1C charge rate, the new cell spends only 49% of the total charge time in the constant voltage mode, whereas the used cell spends about 73% of the time in the constant voltage mode.

This behavior can be attributed to the increase in the cell impedance on use. The increase in the cell impedance causes the used cells to reach 4.1V earlier in comparison to the new cells and therefore the used cells spend more time in the constant voltage charging phase.

The end of discharge voltage as a function of charge rate for the used and the new cells is shown in Fig.2.8.4. There are two major differences between the used and the new cells. The first difference is that the end of discharge voltage is lower for the used

cell in comparison to the new Quallion cell. This suggests increased cell impedance upon

56 cycling. The second major difference is the slope of the linear fit to the data for the used cell is 278 mV per unit charge rate and in comparison the slope for the new cell is small and is about 22 mV per unit charge rate. Assuming the concentration overpotentials in the electrolyte to be the same in both the used and new cells, the difference in slope can be attributed to the increase in the ohmic and charge transfer resistance due to the increase in thickness of the resistive solid electrolyte interphase layer.

The Wh efficiency of the used and the new cells for different rates of charging is shown in Fig. 2.8.5. It is evident that the Wh efficiency is less than 100% even for the new cell and is due to the various overpotentials which increase the energy consumption during charge and decreases the energy released during discharge. For the new cell, the

Wh efficiency decreases linearly with charge rate. The Wh efficiency of the used cells is lower than that of the new cells again indicating that the cell impedance has increased with cycling for the used cells.

The percentage increase in the total charge time for the used cells in comparison to the new cells at different charging rates is shown in Fig.2.8.6. The first conclusion is that even at large charging rate (1C) the percentage increase in the charging time is only

30% (low). The percentage increase in the charge time increases with the charging rate.

At C/5 rate of charging (0.2C) the increase in total charge time is only about 9.2%. It should be noted that calculations show that at the C/5 rate the energy consumed for charging is less by 3.5% in comparison to the C rate charging.

The conclusions that can be drawn from this section are that the cell impedance increases with cycling, decreasing the Wh efficiency obtained and increasing the total charge time of the Quallion cells. By charging at low rates, the percentage increase in

57 charge time is lowered and a small gain of 3.5% energy is obtained. This is significant

because it is important to efficiently deliver the energy from the RF frequency generator

to the access port (which holds the battery) implanted in the body.

5 New cell Old cell 4

Total charge time (h) 1

0 0 0.2 0.4 0.6 0.8 1 1.2 Charge rate (C)

Figure 2.8.1: Total charge time as a function of different charging rates for new and used Quallion cells.

100 CC time, % CV time, % 80

40 Charge time (%) 20

0 0 0.2 0.4 0.6 0.8 1 1.2 Charge rate (C)

Figure 2.8.2: Percentage of constant current and constant voltage charge times as a function of different charging rates for new Quallion cells. The cells during charging spend more time in the constant current phase rather than in the constant voltage phase. 58

60 CC time, % CV time, % 40

Charge time (%) 20

0 0 0.2 0.4 0.6 0.8 1 1.2 Charge rate (C)

Figure 2.8.3: Percentage of constant current and constant voltage charge times as a function of different charging rates for used Quallion cells.

3.5 y = -0.0215x + 3.3307

3.3

3.1 New Cell Old Cell 2.9

y = -0.2783x + 3.0113 2.7 End of discharge voltage (V) 2.5 0 0.2 0.4 0.6 0.8 1 1.2 Charge rate (C)

Figure 2.8.4: End of discharge voltage as a function of different charge rates for new and used Quallion cells. A straight line is fit to the data.

59 96

90 New Cell Old Cell 88

Watthr efficiency (%) 84

82 0 0.2 0.4 0.6 0.8 1 1.2 Charge rate (C)

Figure 2.8.5: Watt-hour efficiency of the used and the new Quallion cell at different charging rates.

5 Charge time increase (%)

0 0 0.2 0.4 0.6 0.8 1 1.2 Charge rate (C)

Figure 2.8.6: Percentage increase in charge time of the used Quallion cell in comparison to the charge time of the new Quallion cell. The percentage charge time is maximum for the 1C charge rate.

60 2.9. Conclusions

Lithium ion cells from the manufacturers Sony, Wilson Greatbatch and Quallion were investigated. The Wilson Greatbatch and Quallion are implantable batteries. The

effect of different operating conditions on the cycle life of Sony Li ion cell was studied.

Factors favoring increased cycle life were identified. The cycle life can be enhanced by

(i) charging at C/5 rather than C rates, (ii) using 4.1V cutoff rather than 4.2V cutoff, (iii)

low depth of discharge. The effect of discharge rate on the cycle life of the Sony lithium

ion cell is unclear due to the ageing of the tested lithium ion cells. The cycle life of the

Sony Li ion cell was diminished when cycling at 37o C in comparison to cycling at room

temperature. The cycle life of the lithium ion cells was fitted considering a non-linear model rather than a linear fit used in the previous work10 at Case. The non-linear model

accounts for SEI film growth due to cycling and can then be used to predict the cycle life of these batteries. The fit parameters were estimated accurately (very small errors) and were used to compare the different battery technologies.

The cycle life of lithium ion cells from different manufacturers was compared.

The performance of the Sony cells is superior in comparison to the Wilson Greatbatch and Quallion cells. Of the implantable battery technologies, the Quallion cells show longer cycle life in comparison to the Wilson Greatbatch cells and therefore the Quallion cells will be the choice for our implantable power source. The Quallion cells can last about 6000 cycles at 60% DOD, C rate charge, C rate discharge, 4.1V charge cutoff and

37o C. The cycle life of the Wilson Greatbatch cells can be significantly increased by cycling the cells at 60% depth of discharge, however, it is still less than that of the

Quallion cells.

61 The charge time of the new and used Quallion cells was studied. The charge time

of the used cells increases by about 30% and 9% at C rate and C/5 rate, respectively. It is

concluded that charging at low rates results in more efficient energy transfer to the

batteries.

References

1. T. Nagaura and K. Tozawa, Progress in Batteries & Solar Cells, 9, 209 (1990).

2. K. M. Abraham and S. B. Brummer, in Lithium Batteries, J. P. Gabano, Ed., Academic Press Inc., London, p. 371 (1983).

3. J. R. Dahn, A. K. Sleigh, H. Shi, B. M. Way, W. J. Weydanz, J. N. Reimers, Q. Zhong, and U. Sacken, in Lithium Batteries: New Materials, Developments and Perspectives, G. Pistoia, Ed., Elsevier, Amsterdam, p. 1 (1994).

4. D. Aurbach, B. Markovsky, M. D. Levi, E. Levi, A. Schechter, M. Moshkovich, and Y. Cohen, Journal of Power Sources, 82, 95 (1999).

5. E. Peled, J. Electrochem. Soc., 126, 2047 (1979).

6. D. Aurbach, Y. Eineli, O. Chusid, Y. Carmeli, M. Babai, and H. Yamin, Journal of the Electrochemical Society, 141, 603 (1994).

7. D. Aurbach, Y. Eineli, B. Markovsky, A. Zaban, S. Luski, Y. Carmeli, and H. Yamin, Journal of the Electrochemical Society, 142, 2882 (1995).

8. O. Chusid, Y. E. Ely, D. Aurbach, M. Babai, and Y. Carmeli, Journal of Power Sources, 43, 47 (1993).

9. Y. Eineli, B. Markovsky, D. Aurbach, Y. Carmeli, H. Yamin, and S. Luski, Electrochimica Acta, 39, 2559 (1994).

10. M. J. Haugh, M.S. thesis, Dept. of Electrical Engineering and Computer Science, Case Western Reserve University, Cleveland, Ohio., 2003.

11. P. Arora, R. E. White, and M. Doyle, Journal of the Electrochemical Society, 145, 3647 (1998).

12. R. Spotnitz, Journal of Power Sources, 113, 72 (2003).

13. P. Arora, M. Doyle, and R. E. White, Proceedings - Electrochemical Society, 98- 16, 553 (1999).

62 14. P. Arora, M. Doyle, and R. E. White, Journal of the Electrochemical Society, 146, 3543 (1999).

15. D. Aurbach, E. Zinigrad, H. Teller, and P. Dan, Journal of the Electrochemical Society, 147, 1274 (2000).

16. J. R. Belt, C. D. Ho, C. G. Motloch, T. J. Miller, and T. Q. Duong, Journal of Power Sources, 123, 241 (2003).

17. I. Bloom, B. W. Cole, J. J. Sohn, S. A. Jones, E. G. Polzin, V. S. Battaglia, G. L. Henriksen, C. Motloch, R. Richardson, T. Unkelhaeuser, D. Ingersoll, and H. L. Case, Journal of Power Sources, 101, 238 (2001).

18. I. Bloom, S. A. Jones, V. S. Battaglia, G. L. Henriksen, J. P. Christophersen, R. B. Wright, C. D. Ho, J. R. Belt, and C. G. Motloch, Journal of Power Sources, 124, 538 (2003).

19. M. Broussely, S. Herreyre, P. Biensan, P. Kasztejna, K. Nechev, and R. J. Staniewicz, Journal of Power Sources, 97-98, 13 (2001).

20. M. Broussely, Advances in Lithium-Ion Batteries, 393 (2002).

21. M. Broussely, S. Herreyre, F. Bonhomme, P. Biensan, P. Blanchard, K. Nechev, and G. Chagnon, Proceedings - Electrochemical Society, 2001-21, 75 (2003).

22. M. Broussely and G. Archdale, Journal of Power Sources, 136, 386 (2004).

23. G. Ning and B. N. Popov, Journal of the Electrochemical Society, 151, A1584 (2004).

24. M. Broussely, P. Biensan, F. Bonhomme, P. Blanchard, S. Herreyre, K. Nechev, and R. J. Staniewicz, Journal of Power Sources, 146, 90 (2005).

25. T. Inoue, T. Sasaki, N. Imamura, H. Yoshida, M. Mizutani, and M. Goto, NASA Conference Publication, 2002-211466, 452 (2002).

26. T. Kishi, H. Ohta, and Y. Satoh, Electrochemistry (Tokyo, Japan), 71, 1090 (2003).

27. B. Y. Liaw, E. P. Roth, R. G. Jungst, G. Nagasubramanian, H. L. Case, and D. H. Doughty, Journal of Power Sources, 119-121, 874 (2003).

28. B. Y. Liaw, R. G. Jungst, G. Nagasubramanian, H. L. Case, and D. H. Doughty, Journal of Power Sources, 140, 157 (2005).

29. T. Osaka, S. Nakade, M. Rajamaki, and T. Momma, Journal of Power Sources, 119-121, 929 (2003).

63 30. P. Ramadass, B. Haran, R. White, and B. N. Popov, Journal of Power Sources, 112, 614 (2002).

31. P. Ramadass, B. Haran, R. White, and B. N. Popov, Journal of Power Sources, 112, 606 (2002).

32. P. Ramadass, B. Haran, R. White, and B. N. Popov, Journal of Power Sources, 123, 230 (2003).

33. P. Ramadass, B. Haran, P. M. Gomadam, R. White, and B. N. Popov, Journal of the Electrochemical Society, 151, A196 (2004).

34. G. Sarre, P. Blanchard, and M. Broussely, Journal of Power Sources, 127, 65 (2004).

35. G. Sikha, B. N. Popov, and R. E. White, Journal of the Electrochemical Society, 151, A1104 (2004).

36. A. T. Stamps, C. E. Holland, R. E. White, and E. P. Gatzke, Journal of Power Sources, 150, 229 (2005).

37. R. B. Wright, C. G. Motloch, J. R. Belt, J. P. Christophersen, C. D. Ho, R. A. Richardson, I. Bloom, S. A. Jones, V. S. Battaglia, G. L. Henriksen, T. Unkelhaeuser, D. Ingersoll, H. L. Case, S. A. Rogers, and R. A. Sutula, Journal of Power Sources, 110, 445 (2002).

38. R. B. Wright, J. P. Christophersen, C. G. Motloch, J. R. Belt, C. D. Ho, V. S. Battaglia, J. A. Barnes, T. Q. Duong, and R. A. Sutula, Journal of Power Sources, 119-121, 865 (2003).

39. D. Zhang, B. S. Haran, A. Durairajan, R. E. White, Y. Podrazhansky, and B. N. Popov, Journal of Power Sources, 91, 122 (2000).

64 3. Platinum Negative Electrode – Testing and Activation

In this chapter, three major components of the low pressure nickel-hydrogen cell

– the negative electrode, the current collector and the separator are analyzed.

3.1. Introduction – Platinum Electrode

The negative electrode used in the development of our low pressure nickel- hydrogen battery is a platinum catalyzed electrode. When the nickel-hydrogen battery is charged, hydrogen is evolved at the negative electrode and when the cell is used

(discharged), hydrogen is oxidized at the electrode. Therefore the main requirement of the negative electrode is to facilitate the hydrogen evolution and the hydrogen oxidation

(depending on charge or discharge) with fast kinetics. The overpotentials for these reactions should be small at the rated current for the battery.

Platinum with other platinum group metals either as single, binary or ternary combinations 1-12 has been the preferred option for the hydrogen oxidation reaction in alkaline fuel cells. The overpotential of these electrodes is about only 20 mV for current densities less than 400 mA/cm2. Other catalysts like nickel boride, Raney nickel and sintered nickel 9-13 have been investigated for oxidation of hydrogen. However the catalytic activity and the stability of these catalysts are limited. Kiros and Schwartz 14 have shown that the Pt/Pd bimetallic combination catalyst is more stable than the Raney

Ni catalyst at long oxidation times. The main objective of this work is to develop a battery with long cycle life and therefore the Pt catalyst, which is more stable, was

chosen as the negative electrode. Platinum electrodes are also the negative electrodes in

the high pressure nickel-hydrogen cells used in space applications.

The platinum electrodes used in the high pressure nickel-hydrogen cells and in

our low pressure nickel-hydrogen cells are based on electrode designs developed for the

gas diffusion (GDE) electrodes used in the alkaline fuel cells. Typically the platinum particles are supported on carbon (Vulcan XC-72) are mixed in the form of liquid slurry along with polytetrafluoroethylene (PTFE) particles and are then blended together with an emulsifying agent like Triton X100. They are then filtered to the required size and thickness and pressed onto a nickel screen, followed by sintering to hold the PTFE particles together. The electrode structures are typically 50% porous and only one half of

the pores are wetted by the electrolyte. The other half of the electrode is hydrophobic and

facilitates hydrogen diffusion in and out of the electrode filled with a thin layer of

electrolyte.

In the high pressure nickel-hydrogen batteries a platinum loading of 7 mg/cm2 were used in the electrodes. In our cells, a platinum loading of 0.6 mg/cm2 was used and

the results discussed in this chapter shows that the loading levels are sufficient.

The negative electrode used in our nickel-hydrogen batteries is the A-2 silver-

plated nickel screen electrode (ESNS) from E-tek. The support is 80 mesh silver plated

nickel screen and the thickness of the finished electrode is about 0.4 mm. 10% Pt on

Vulcan XC-72 is used as the catalyst and is loaded on one side only. Catalyst containing

10% Pt has a platinum area of about 141 m2/gm (technical information available from E-

tek). Hence, the ESNS electrode with 0.6 mg/cm2 has a platinum area of 846 cm2 per cm2

of electrode.

In this chapter, the performance of the E-tek electrode used in the development of

the low pressure nickel-hydrogen battery is presented. The negative platinum electrode was studied and analyzed in the liquid cell (section 3.2) and bolt nut cell configuration

(section 3.3) simulating the nickel-hydrogen battery conditions. The activation of the platinum electrode required to increase the exchange current density is also discussed. In section 3.4, the effect of two different current collectors, nickel chromium and nickel, on the high frequency resistance will be discussed. In section 3.5, the performance of two separators, zirconium oxide stabilized by yttria and polypropylene, will be discussed briefly.

3.2. Characterization of the ESNS Electrode in a Liquid Cell Configuration

A 3-electrode system with the ESNS electrode as the working electrode, a reversible hydrogen electrode as the reference electrode and a platinum gauze as the counter electrode was set up in the BASi electrochemical cell (Fig. 3.2.1). The electrolyte was 26 wt% potassium hydroxide solution. The cell was purged with nitrogen overnight and cycled at 25 mV/s in the voltage range -0.1 to 0.75 V to remove dissolved oxygen in the electrolyte.

ESNS

Reference H2 electrode electrode Pt mesh

N2 N2 Inlet Outlet

26 wt. % KOH

Figure 3.2.1: Schematic representation of the experimental setup.

3.2.1. Hydrogen Evolution Reaction Kinetics

The hydrogen evolution reaction at the ESNS electrode is

+→+−− 15 22H22Oe H 2 OH. Considering Butler-Volmer kinetics for hydrogen

evolution, the current-potential relationship is given by:

⎛⎞αηnF βη nF =−−⎛⎞⎛⎞aa iio ⎜⎟exp⎜⎟⎜⎟ exp [3.1] ⎝⎠⎝⎠⎝⎠RT RT

where i is the current density and ηa is the activation overpotential for the faradaic

reaction. α and β are the anodic and cathodic transfer coefficients respectively. n is the

number of electrons transferred and is 2 for hydrogen evolution. io is the exchange

current density of the reaction.

The current –potential relation can be separated into two regimes for convenience.

η For small activation overpotentials, a << RT/nF , the kinetics are considered linear and

η for large activation overpotentials, a >> RT/nF , the anodic part in the Butler-Volmer

equation (first part) can be neglected for the hydrogen evolution reaction (cathodic

reaction). The value RT/nF is 12.7 mV for n = 2 .

η In the linear regime, a << 12.7 mV , the Butler-Volmer equation is linearized to

η = RTi give a . The equation indicates that the activation overpotential increases nF io

linearly with current density.

η In the Tafel regime, a >> 12.7 mV , the Butler-Volmer equation is reduced to

RTi⎛⎞ η = ln . The equation indicates that the activation overpotential increases with a β ⎜⎟ nF⎝⎠ io

the current density, however, with a slower rate in comparison to the linear regime.

∂η The charge transfer resistance is defined as R = a . For linear regime, the ct ∂i

= RT charge transfer resistance is given as: Rct . It is evident from the equation that the nFio

charge transfer resistance is independent of the applied current density. The charge

RT transfer resistance in the Tafel regime is given as: R = . The equation indicates ct β nFi

that the charge transfer resistance decreases with increase in the current density.

It is evident from these discussions that the charge transfer resistance measured

using the electrochemical impedance spectroscopy is dependent on the potential range

and could be either in the Tafel or linear regime. This will be discussed in detail in the

next two sections section 3.2.2 and 3.2.3.

3.2.2. Electrochemical Impedance Spectroscopy

The open circuit voltage (OCV) of the negative ESNS electrode in de-aerated

electrolyte was 0.56 V. The ESNS electrode was biased at various negative DC potentials

(reduction reaction), between 0 to -0.5V and the impedance spectra was measured as a

function of frequency. The AC amplitude applied was 10 mV.

The potential at which the impedance spectra were measured was then corrected

for the ohmic potential drop and is labeled as IR corrected potential. Table 3.1 shows that the IR corrected potential range is -0.008 to -0.154 V and covers both the linear and Tafel regimes of the polarization curve. The impedance spectra of the ESNS electrode obtained at different biased potentials is shown in Fig. 3.2.2 in the form of a Nyquist plot. The

diameter of the semicircle gives the charge transfer resistance. The solution resistance,

Rs, charge transfer resistance, Rct, and the double layer capacitance, Cdl, were estimated based on the impedance spectra (Fig. 3.2.2) and are listed in Table 3.1.

As expected, the solution resistance does not change significantly through the entire potential range. The charge transfer resistance decreases as function of the potential indicating that the cell is in Tafel regime (the charge transfer resistance is constant in the linear regime). It is noted that the only data point in the linear regime is -

0.008V. The other data points are scattered over the intermediate and Tafel regimes. It is also evident from Table 3.1 that the solution resistance is larger than the charge transfer resistance and is due to the fact that the reference hydrogen electrode was placed far from the working electrode (negative ESNS electrode). The capacitance decreases with increase in potential as listed in Table 3.1.

Table 3.1: Impedance analysis in the potential range 0 to -0.5 V.

Bias IR corrected Rs, Rct, Cdl, Potential, V potential, V ohm cm2 ohm cm2 F/cm2 -0.025 -0.008 1.06 0.88 0.40 -0.050 -0.029 1.07 0.69 0.39 -0.075 -0.041 1.07 0.57 0.38 -0.100 -0.052 1.07 0.48 0.37 -0.250 -0.102 1.08 0.28 0.28 -0.350 -0.125 1.09 0.21 0.27 -0.500 -0.154 1.12 0.18 0.22

-0.50

-0.25

Decrease in Rct 0 Z'' -8 mV -29 mV -41 mV 0.25 -52 mV -102 mV -125 mV -154 mV 0.50 1.00 1.25 1.50 1.75 2.00 Z'

Figure 3.2.2: Impedance spectra of the ESNS electrode in the potential range -8 to -154 mV. The charge transfer resistance decreases with increase in the bias potential.

Impedance Measurement and Analysis in the Linear Regime

The negative ESNS electrode will typically be in the linear regime when used in the nickel-hydrogen cell because of the low charge and discharge currents required by the nickel hydroxide electrode. Therefore it is necessary to estimate the charge transfer resistance for the hydrogen evolution in the linear polarization regime.

Impedance spectra at different bias potentials in the linear regime were measured and are shown in Fig. 3.2.3. The bias potentials were IR corrected and are listed in Table

3.2. The solution resistance, charge transfer resistance and the double layer capacitance were calculated. It is evident that the charge transfer resistance is constant in the potential range 0 to -15mV and hence, the measurements are in the linear regime. The average

value of the charge transfer resistance in this regime is about 1.6 ohm cm2. The double layer capacitance in the linear regime is also constant as shown in Table 3.2.

The impedance data analysis shows linear polarization in the potential range, 0 to

-20 mV. In the next section, the potentiodynamic data measured in the linear polarization range will be used to calculate the charge transfer resistance and the exchange current density of the electrodes to provide an independent check on the impedance results.

-1.0

-0.5 -8 mV -11 mV -15 mV

0 Z''

0.5

1.0 0.5 1.0 1.5 2.0 2.5 Z'

Figure 3.2.3: Impedance spectra of the ESNS electrode in the linear polarization regime. The potential range is -8 to -15 mV. The charge transfer resistance does not change with the bias potential.

Table 3.2: Impedance analysis in the potential range 0 to -20 mV.

Bias IR corrected Rs, Rct, Cdl, potential, mV potential, mV ohms cm2 ohm cm2 F/cm2 -10 -8 0.96 1.61 0.48 -15 -11 0.96 1.58 0.46 -20 -15 0.96 1.52 0.45

3.2.3. Potentiodynamics

A potentiodynamic scan was also used to estimate the kinetic parameters of the

ESNS electrode. The potentiodynamic scan was performed in the potential range 0 to -

0.15V at a slow scan rate, 0.1 mV/s, and is shown in Fig. 3.2.4. It is assumed that the

electrode is in equilibrium at such slow scan rates. It is evident from the figure that the

current increase is linear in the potential range, 0 to -25 mV, corresponding to linear

polarization regime and is non-linear at potentials larger than -25 mV.

The data in the range, -6 to -16 mV (Fig. 3.2.5) was fitted linearly using Origin

and the error in the parameter estimations is small. The charge transfer resistance is the

reciprocal of the estimated slope and is 1.25 Ω cm2 and is about the same value, 1.6 Ω

cm2, estimated using the impedance spectroscopy. The exchange current density

= () 2 calculated based on the charge transfer resistance ( iRTnFRoct) is 10.3 mA/cm . In comparison, Alcaide et al.16 calculated the exchange current density as 2.03 mA/cm2 for an E-tek electrode containing 0.5 mg/cm2 of platinum loading, five times smaller than the

value for our electrode.

Potentiodynamic scan 0.1mV/s

Non-linear Linear -0.1 Regime ) 2

-0.2 I (Amps/cm

-0.3

-0.4 -0.20 -0.15 -0.10 -0.05 0 0.05 E (Volts)

Figure 3.2.4: Potentiodynamic scan of the ESNS electrode at a scan rate of 0.1 mV/s. Two different regimes are indicated in the figure.

0 y = A + B*x R^2 = 0.99742 -2 A 0.004 ± 0.00011 B 0.8 ± 0.00986

2 -4

-6 i mA/cm

-8

-10

-16 -14 -12 -10 -8 -6 E (mV)

Figure 3.2.5: Linear fit of the potentiodynamic scan in the potential range -6 to -16 mV. The scan rate is 0.1 mV/s. The parameter ‘B’ is the inverse of the charge transfer resistance.

3.2.4. Conclusions

The exchange current density and the charge transfer resistance were calculated

using impedance spectroscopy and potentiodynamic analysis. The estimated exchange

current density is 10.3 mA/cm2 and is larger than that calculated in the work by Alcaide et al16 for a similar E-tek electrode. The charge transfer resistance is in the range 1.3-1.6

ohm cm2.

The Pt surface area in the given electrode area is 846 cm2 /cm2. Based on this

value and our estimated exchange current density for the electrode (10.3 mA/cm2), the exchange current density per unit Pt area is ~ 0.012 mA/cm2 Pt. This is an order of

magnitude smaller in comparison to the experimentally determined exchange current

density, in the range 0.1-0.5 mA/cm2, for pure platinum4, 17. A slightly lower value than

that of pure platinum is expected because not all the platinum in the electrode is exposed

to the electrolyte. However, the calculated exchange current density corresponds to only

about 10% platinum utilization.

Our nickel-hydrogen cells will be operated at about 3 mA/cm2 (C/5 rate) and the

corresponding overpotentials are less than 10 mV.

3.3. Characterization of the ESNS Electrode in Double Negative Cell

Configuration

The ESNS electrodes were tested in a bolt nut cell configuration as shown

schematically in Figure 3.3.1. The main objective, here, is to study the performance of

the negative electrode in a cell setup which will be used for the Ni-H2 cell testing. It is

noted that the main limitation of the negative electrode studies in the liquid cell (previous

section 3.2) is that the results were based only on the hydrogen evolution reaction. Using

the bolt nut cell configuration, the oxidation of hydrogen and the evolution of hydrogen at the two electrodes can be studied, simultaneously. Hydrogen is formed during charge and is oxidized during discharge. Therefore it is important to understand the reaction kinetics of the negative electrode during charge and discharge.

Two ESNS electrodes separated by a polypropylene membrane (Celgard 3400) soaked in 26 wt. % KOH solution were assembled in a stainless steel compartment filled with H2 gas (Fig. 3.3.1). In this setup, one of the ESNS electrodes is the working electrode and the other is the counter and reference electrode. Nickel chromium mesh was used both as current collector and gas diffusion layer. Electrochemical impedance spectroscopy and cyclic voltammetry were used to characterize the electrodes.

Insulator ESNS/Pt ESNS/Pt H2

Nut Bolt

Separator Ni-Cr Current collector Inlet valve

Figure 3.3.1: Schematic representation of the experimental setup.

3.3.1. Testing of Fresh ESNS Electrodes

A set of fresh ESNS electrodes and a fresh separator were soaked in KOH overnight and then assembled in the double negative cell configuration. Impedance

spectra were measured at open circuit voltage (~ 0V) with an AC amplitude of 10 mV.

The charge transfer resistance and the double layer capacitance were estimated based on the impedance data and the values for one electrode are listed in Table 3.3.

The charge transfer resistance, 6.8 Ω cm2, for the fresh electrode is larger than that of the measurement in the liquid cell, 1.3-1.5 Ω cm2. The double layer capacitance, 7 mF/cm2, is almost two orders of magnitude lower than the corresponding measurement in the liquid cell setup.

There are significant differences between the liquid cell test and the bolt nut cell test. In the liquid cell the electrode is flooded with the electrolyte and in comparison the bolt nut cell has electrolyte primarily in the separator. This difference is amplified by the fact that the ESNS electrode as made is hydrophobic so as to facilitate three phase (solid liquid gas) reactions. This affects the wetting of the electrode. The other significant difference is in the liquid cell the charge transfer resistance was measured based on the hydrogen evolution. However, in the bolt nut cell configuration, charge transfer resistance includes both hydrogen evolution at one electrode and hydrogen oxidation at the other electrode. It is noted that the hydrogen oxidation depends on the distribution of hydrogen gas in the ESNS electrode.

Table 3.3: Estimated parameters for fresh electrodes based on the linear fit applied to the CV data. Electrode Parameter Calculated Value Charge transfer resistance (ohm cm2) 6.8 Double layer capacitance (mF cm-2) 7

3.3.2. Conditioning of ESNS Electrodes with Cyclic Voltammetry

The fresh ESNS electrodes were then conditioned by cyclic voltammetry (CV) in

the potential range -0.5 to 0.5 V at scan rates of 5 mV/s to decrease the charge transfer

resistance. The impedance spectra were measured after every set of the cyclic

voltammograms. The calculated charge transfer resistance and the double layer

capacitance from the impedance data are shown in Fig. 3.3.2.

The charge transfer resistance decreases significantly with cycling (Fig. 3.3.2)

from 7 Ω cm2 to 1.75 Ω cm2 and is due to the increase in the wetting of the ESNS

electrode by the electrolyte. This assumption is supported by the fact that the double layer

capacitance increases from 7 to 27 mF/cm2 on cycling (Fig. 3.3.2). It is noted that for the

ESNS electrode most of the capacitance contribution is due to carbon (carbon 5-6

mg/cm2, Pt 0.6 mg/cm2). However, it is assumed that the increase in the wetting of

carbon (increase in capacitance) corresponds to increase in wetting of the Pt in the

electrode. This indicates that cycling results in the activation of the platinum electrodes.

The charge transfer resistance and the double capacitance as shown in Fig. 3.2.2 does not change for the last two sets of cyclic voltammograms indicating that no significant changes in the wetting of the electrode. It is noted that the charge transfer resistance at the end of cycling is 1.75 Ω cm2 is about the same value measured in the liquid cell (1.3-1.6 Ω cm2).

The first set of cyclic voltammograms (used for electrode conditioning) recorded

in the voltage range -0.5 to 0.5 V were IR corrected and are shown in Fig. 3.3.3. Two

distinct regions, linear and Tafel regimes as discussed in the previous section 3.2 are

identified. It is assumed that the overpotential due to mass transport is negligible in the

voltage range -0.2 to 0.2 V. It is noted that the voltage measured is the sum of activation overpotential contributions due to the two electrodes (after IR correction). Therefore the data fit range -2 to -22 mV can be approximated as -1 to -11 mV for one ESNS electrode, assuming that the two electrodes are identical. The linear polarization region corresponds to the voltage range of -22 to 22 mV. The cathodic and anodic Tafel polarization regions are in the voltage range of -0.15 to -0.2V and 0.15 to 0.2V, respectively. It is evident that the current density of the electrode increases with cycling indicating decrease in the charge transfer resistance.

8 32

2 Rct Cdl -2 6 24

4 16

2 8 Resistance, ohm cm Capacitance, mF cm

0 0 Before 5 mV/s, 5 mV/s, 5 mV/s, 5 mV/s, 5 mV/s, 5 mV/s, 5 mV/s, scans 3 cy 5 cy 3 cy 5 cy 10 cy 25 cy 10 cy Electrode condition

Figure 3.3.2: Charge transfer resistance and double layer capacitance as a function of electrode condition based on impedance measurements before and after every set of cyclic voltammograms. ‘Cy’ in the x-axis text indicates cycles.

30 Increase in 20 Current

) 10 Linear 2 Tafel 0 Tafel i (mA/cm -10

-20

-30 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 E (Volts)

Figure 3.3.3: First set of cyclic voltammogramms (3 cycles) at a scan rate of 5 mV/s.

Fig.3.3.4 shows a linear fit to the cyclic voltammogram data in the potential range, -2 to -22 mV, taken at 5 mV/s (cathodic region, cycle 1) without any electrode conditioning. The calculated charge transfer resistance is 7.1 Ω cm2 and is about the same measured using the impedance spectroscopy. The exchange current density is 1.8 mA/cm2 about an order of magnitude lower than that measured in the liquid cell. The charge transfer resistance and the exchange current density for every cycle were determined similarly.

The exchange current density and the charge transfer resistance calculated based on the first and last scan of electrode conditioning are listed in Table 3.4. The charge transfer resistance continued to decrease and correspondingly the exchange current density increased on cycling (other data not shown, Table 3.4). This is similar to the decrease in charge transfer resistance observed for the impedance data (Fig. 3.3.2). The

exchange current density for the last set of scans did not change significantly and is about

5.6 mA/cm2. This shows that cycling is beneficial by increasing the exchange current

density of the electrode.

3 RctRpFit Result 2 Data Range: -2 to -22 mV Rct (Ohms/cm2)= 7.2 io (mA/cm2)= 1.8

) 1 2

0 i (mA/cm -1

-2

-3 -25 -20 -15 -10 -5 0 5 10 15 20 25 E (mV)

Figure 3.3.4: IR corrected cyclic voltammogramms showing the linear regime at a scan rate of 5 mV/s. The data in the voltage range, -2 to -22mV, was linearly fit.

Table 3.4: Estimated parameters for fresh electrodes based on the linear fit applied to the CV data. Electrode condition Exchange current Charge transfer 2 2 density, io (mA/cm ) resistance (ohm cm ) Beginning of cycling 1.8 7.2 End of cycling 5.6 2.3

3.3.3. Effect of Rest

The ESNS electrodes used in the previous experiments were soaked in 26 wt%

KOH solution overnight. Then the cell with the two ESNS electrodes was assembled and

subjected to rest. The impedance spectra were evaluated at different rest periods. The

calculated charge transfer resistances and the double layer capacitance do not change

significantly when the cells are rested (Fig. 3.3.5). This in principle shows that the

conditioned electrode is stable and does not change drastically with the rest periods.

4 37

2 Rct Cdl -2 3 36

2 35

1 34 Resistance, ohm cm Capacitance, mF cm

0 33 Just assembled after 16h rest after 22h rest Electrode condition

Figure 3.3.5: Charge transfer resistance and double layer capacitance as a function of electrode condition based on impedance measurements before and after different rest periods.

3.3.4. Effect of soaking in KOH for days

The activated ESNS electrodes were soaked in liquid KOH for 6 days. The charge

transfer resistance and the double layer capacitance calculated using the impedance

spectra did not change significantly after soaking.

3.4. Testing of Current Collectors

Two types of current collectors, nickel chromium mesh and nickel mesh, were tested using the negative electrodes in the bolt nut cell configuration. The high frequency resistance of the cell was estimated based on the impedance spectra. The high frequency resistance includes contributions from all the contact resistances in the cell, as well as the electrolyte resistance and the resistance of the current collectors.

Fig. 3.4.1 compares the high frequency resistance of two cells one using nickel chromium mesh current collectors and the other using nickel mesh current collectors. It is evident that the high frequency resistance of the cell using nickel mesh current collectors is about 2 times smaller than the cell using nickel chromium mesh current collectors.

The degradation of the current collectors when used in the bolt nut cell is compared in Fig. 3.4.2. It is evident that the high frequency resistance of the cell with

nickel mesh current collector does not degrade significantly when used. However, the

high frequency resistance of the cell using nickel-chromium current collector increases

by a factor of 5. The increase in the high frequency resistance of the Ni-Cr mesh is due to

the corrosion (formation of oxide layer) when two different meals are in contact with

KOH solution. This concludes that the nickel mesh current collectors are better suited to our bolt nut cell configuration.

2 2.5

cm 2 Ω

1.5

Resistance, 0.5

0 Nickel chromium Nickel Type of current collector

Figure 3.4.1: The high frequency resistance of the cell using nickel chromium mesh current collectors and nickel mesh current collectors is compared.

20 Ni Ni-Cr 2 16 cm

Ω 12

4 Resistance,

0 Fresh Used Current collector condition

Figure 3.4.2: The high frequency resistance of the cell, one using nickel chromium mesh as current collectors and other using nickel mesh as current collectors is compared for two different conditions fresh and used.

3.5. Testing of Separators

Two separators (a) Zircar ZYK-15 and (b) Celgard 3400 were tested in the double

negative cell configuration. Zircar is the trade name of zirconium oxide stabilized with

yttria of 380 µm thickness. It is used in nickel-hydrogen batteries because of its

hydrophilicity and excellent solution retention property. The other separator, Celgard

3400 is a micro-porous polypropylene membrane of 25 µm thickness coated with a wetting agent and is used in aqueous battery systems. The high frequency resistance of the cell was not significantly different for the two separators.

The ohmic overpotential of the cell including the current collector resistance and electrolyte resistance is small, 4 mV.

3.6. Conclusions

A platinum electrode, called as ESNS electrode, from E-tek Inc. was tested in the

liquid cell configuration and in the double negative cell configuration. The platinum

loading in the electrode is 0.6 mg/cm2. The electrodes were analyzed by electrochemical impedance spectroscopy and cyclic voltammetry.

In the liquid cell the evolution of hydrogen on the platinum electrode was studied.

Based on the impedance and the potentiodynamic data, the charge transfer resistance was estimated in the range 1.2 to 1.6 Ω cm2. The estimated exchange current density for the

hydrogen evolution reaction is 10.3 mA/cm2 electrode area and is five times larger than

that calculated by Alcaide et al.16 The calculated exchange current density based on

platinum area is 0.018-0.026 mA/cm2 and is small. However, the results also show that

the overpotential of these electrodes, operating at current densities of 15 mA/cm2 (typical

operation of our nickel-hydrogen cells), is about 20 mV. This amount of overpotential is

small and therefore the ESNS electrodes are satisfactory for our application.

In the bolt nut cell configuration two platinum electrodes separated by a

polypropylene membrane soaked in 26 wt. % KOH were tested at simulated conditions of

our nickel-hydrogen battery. Both hydrogen evolution reaction and hydrogen oxidation

reaction on these electrodes were studied. Analysis of the electrodes by impedance

spectroscopy and cyclic voltammetry initially shows a large charge transfer resistance of

7.2 Ω cm2. It is shown that cycling the electrodes in the voltage range -0.5 to 0.5 V lowers the charge transfer resistance by a factor of three to 1.8-2.4 Ω cm2.

Correspondingly the exchange current density increases from 1.8 mA/cm2 before cycling

to 5.6 mA/cm2 after cycling. This increase in exchange current density is attributed to

better wetting of the electrolyte. Once the electrode is activated by cycling, the charge

transfer resistance does not change significantly with further cycling and/or rest periods.

Two different current collectors nickel chromium and nickel were studied.

Analysis shows that the nickel chromium current collectors are easily oxidized and the

high frequency resistance of the cell increases dramatically to 15-20 Ω cm2. However

electrochemical cleaning of the current collectors results in lowered high frequency

resistance of 2 Ω cm2. On the other hand, the high frequency resistance of clean nickel

current collectors is about 1.2 Ω cm2 and is lower than the nickel chromium current collector. The solution resistance of the nickel current collectors also did not significantly

change with time. These results clearly indicate that the nickel current collectors are

better than the nickel chromium current collectors. Two separators, Zirconium oxide

stabilized with yttria and polypropylene membrane were tested and no significant

difference was observed.

References

1. P. N. Ross, K. Kinoshita, A. J. Scarpellino, and P. Stonehart, Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 59, 177 (1975).

2. T. J. Schmidt, P. N. Ross, and N. M. Markovic, Journal of Electroanalytical Chemistry, 524-525, 252 (2002).

3. H. Wendt, New Materials for Fuel Cell Systems I, Proceedings of the International Symposium on New Materials for Fuel Cell Systems, 1st, Montreal, July 9-13, 1995, 532 (1995).

4. G. Couturier, D. W. Kirk, P. J. Hyde, and S. Srinivasan, Electrochimica Acta, 32, 995 (1987).

5. L. Angely and G. Bronoel, Electrochimica Acta, 25, 1541 (1980).

6. N. M. Markovic and P. N. Ross, Surface Science Reports, 45, 117 (2002).

7. J. J. T. t. Vermeijlen, L. J. J. Janssen, and G. J. Visser, Journal of Applied Electrochemistry, 27, 497 (1997).

8. A. Pozio, L. Giorgi, E. Antolini, and E. Passalacqua, Electrochimica Acta, 46, 555 (2000).

9. K. Mund, G. Richter, and F. Von Sturm, Proceedings - Electrochemical Society, 79-2, 47 (1979).

10. A. J. Appleby, Journal of Power Sources, 29, 3 (1990).

11. K. Strasser, Journal of Power Sources, 29, 149 (1990).

12. E. Gulzow, M. Schulze, and G. Steinhilber, Journal of Power Sources, 106, 126 (2002).

13. J. Shim and H. K. Lee, Materials Chemistry and Physics, 69, 72 (2001).

14. Y. Kiros and S. Schwartz, Journal of Power Sources, 87, 101 (2000).

15. A. J. Bard and L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications, p. 119, John Wiley & Sons, Inc., (1980).

16. F. Alcaide, E. Brillas, and P.-L. Cabot, Journal of the Electrochemical Society, 152, E319 (2005).

17. S. Ernst and C. H. Hamann, Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 60, 97 (1975).

4. Testing of ‘D Cell’ Nickel Hydroxide Electrode

A ‘D size’ nickel metal hydride cell was cut open to separate the nickel hydroxide

electrode from other battery components. The characteristics of this nickel hydroxide

electrode: specific capacity, coulometric efficiency, cycle life and impedance were

analyzed in the liquid cell and in the nickel-hydrogen cell configuration. The

performance of the electrode in two cells in series was tested and analyzed. The study of

this electrode served as a guide in the development of the pasted nickel hydroxide

electrode (Chapter 5) and the design of the nickel-hydrogen cell (Chapter 6).

4.1. Estimation of Specific Capacity

Based on the charge-discharge tests, the capacity of the nickel metal hydride cell

was estimated as 6.8 Ah. The D cell was opened and the nickel hydroxide electrode was

removed. The total area of the double sided nickel hydroxide electrode is 465.6 cm2.

Therefore the specific capacity of the electrode is 14.6 mAh/cm2.

4.2. Performance of D cell Electrode in Liquid Cell

4.2.1. Description of the Experimental Setup

The performance of the D cell nickel hydroxide electrode was evaluated in a

liquid cell filled with 26 wt. % KOH solution. A three electrode system, consisting of

working counter and reference electrode was used, similar to the setup used for testing

the platinum electrode (Fig. 3.2.1). The D cell nickel hydroxide electrode was used as the

working electrode. Platinum wire gauze was used as the counter electrode. A reversible

hydrogen electrode was used as the reference electrode. The electrochemical cell was

purged with nitrogen gas.

4.2.2. Charge-Discharge Experiments

The D cell nickel hydroxide electrode was soaked in 26 wt% KOH overnight. The

electrode was cycled at C/3 rate in the liquid cell. A rest period of 1h was applied

between the charge and the discharge period. The voltage profiles during charge and discharge of the electrode is shown in Fig. 4.2.1. It is evident that the charge voltage was larger and the discharge voltage was smaller for the 1st cycle in comparison to other

cycles, indicating larger power consumption during the charge and smaller power output

during discharge. However, the charge voltage decreased and the discharge voltage

increased with cycles as shown in Fig. 4.2.1. For cycles 11 to 25, the voltage profiles

remained the same and no significant changes are observed (Figure not shown). This

indicates good cyclability of the electrode. The coulometric efficiency of the nickel

hydroxide electrode for 26 cycles is shown in Fig. 4.2.3. It is evident that the efficiency is about 99% and does not change significantly with cycles. The inset in the figure shows

small changes in the efficiency of the nickel hydroxide electrode with cycling.

4.2.3. Impedance Analysis

Impedance data of the nickel hydroxide electrode at different states of charge

(SOC) was measured and is shown in Fig.4.2.3. At 100% SOC, most of the active

material is nickel oxyhydroxide and at 0% SOC the active material is nickel hydroxide

(chapter 1). It is noted that nickel oxyhydroxide is a good electronic conductor whereas

the nickel hydroxide is a poorly conducting, semiconductor1. The conductivity of the

electrode is enhanced typically by the addition of cobalt and zinc and will be discussed in

detail in Chapter 5. It is evident that the charge transfer resistance of the electrode is about the same for 100% and 67% SOC. This resistance increases for 33% and 0% SOC and is attributed to the poor electronic conductivity of the nickel hydroxide, formed during discharge in spite of the presence of additives to enhance conductivity. The electrode was not characterized for materials composition and therefore the variation in the impedance cannot be elaborated further.

1.5

1.4 Decrease in charge voltage with cycling

1.3 1 charge 1.2 1 discharge 2 charge 2 discharge Increase in discharge E (Volts) 1.1 3 charge voltage with cycling 3 discharge 4 charge 1.0 4 discharge 5 charge 5 discharge 0.9 0123 Time (Hours)

Figure 4.2.1: Voltage profile during charge-discharge of the D cell nickel hydroxide electrode for cycles 1 to 5. The cell was cycled at C/3 rate to 100% capacity.

120

100

80 100.5 60 100.0 99.5 40 99.0 98.5 20 0 102030 Columetric Efficiency, % 0 0 5 10 15 20 25 30 Cycle Number

Figure 4.2.2: Coulometric efficiency of the D cell nickel hydroxide electrode as a function of cycles. The inset shows the small variations in the coulometric efficiency. The cell was cycled at C/3 rate to 100% capacity.

-1.0 100 % Charge 67 % Charge 33 % Charge -0.5 0 % Charge )

2 0%

0 100 % Z'' (ohm cm Z'' (ohm

0.5

1.0 00.51.01.52.0 Z' (ohm cm2)

Figure 4.2.3: Impedance spectra of the D cell nickel hydroxide electrode as a function of state of charge. The cell was cycled at C/3 rate.

4.3. Performance of the D cell Electrode in Nickel-Hydrogen Cell

The D cell nickel hydroxide electrode was assembled in the bolt-nut type cell

described in Chapter 3 as shown in Fig. 3.3.1. The D cell nickel hydroxide is the positive electrode and the platinum electrode (discussed in chapter 3) is the negative electrode.

The separator is a polypropylene membrane soaked in 26 Wt. % KOH. In this section the significance of fixing the end of discharge voltage and the effect of overcharging will be discussed.

The voltage profile during charge-discharge of cycle 1 is compared to that of

cycle 5 in Fig. 4.3.1. The cell was cycled at C/3 rate to 80% of the nickel hydroxide

electrode capacity. No discharge cutoff voltage was used. During charging, a fraction of

the charge passed to the cell results in oxygen evolution. Therefore when the cell is

discharged to 3h, the cell is discharged to a lower state of charge than the state it

originally started with. Therefore the end of discharge voltage decreases from cycle 1 to

cycle 5 as shown in the Fig. 4.3.1 and the coulometric efficiency is 100%. It is evident

from the figure that the end of charge voltage also increases from cycle 1 to cycle 5,

contrary to the expectation that the charge voltage will decrease. This is due to the

increase in the charge transfer resistance of the nickel hydroxide and the platinum

electrode as shown in Fig. 4.3.2.

It is noted here that the large charge transfer resistance in Fig. 4.3.2 corresponds

to that of the used platinum electrode. The charge transfer resistance is significantly

reduced for fresh platinum electrodes as discussed in chapter 3 and chapter 6. The

pertinence of the two semi-circles to the charge transfer reactions at the different

electrodes is discussed in section 6.2 of chapter 6.

In contrast the voltage profile during charge-discharge of the electrode with a discharge cut-off voltage of 1V for 5 cycles is shown in Fig. 4.3.3. No significant change in voltage profile is observed. By choosing an end of discharge voltage close to the knee, overcharging the electrode is prevented. It is also relatively easy to comprehend the variations in the charge-discharge profile during cycling.

The effect of overcharging the cell at C/6 rate is shown in Fig. 4.3.4. Voltage profiles during two different cycles, one overcharged by 50% and another with no overcharge are compared. The SOC of the cell before the beginning of charge is different for the two cycles and the end of discharge voltage is not fixed. Therefore the coulometric efficiency is 100% for both cycles. It is evident that when the cell is overcharged, the voltage fluctuates at the end of charge as a result of oxygen evolution and in comparison is absent for the cell with no overcharge. The rise in voltage is due to the oxygen bubble formation at the electrode, which cannot diffuse out at a sufficient rate. Finally the bubble breaks out or recombines at the platinum electrode with a large pop, resulting in the fall in voltage1. The voltage fluctuation is also present during the initial discharge of the overcharged cell indicating that the oxygen formed during discharge is still recombining at the platinum electrode. In comparison, for the cycle with no overcharge no voltage fluctuations are observed during the discharge. The charge and the discharge voltage are higher for the overcharged cell in comparison to the cell with no overcharge because the overcharged cell is in a higher state of charge. The consequence of overcharging is dramatic failure of the nickel-hydrogen cell.

1.5 5

1.4 1 charge

1.3

E (Volts) 1 discharge

1.2 5

1.1 0123 Time (Hours)

Figure 4.3.1: Voltage profile during the charge-discharge of the Ni-H2 cell using the D cell nickel hydroxide electrode for cycles 1 and 5. There is no discharge cut-off voltage. The cell was cycled at C/3 rate to 80% capacity with no voltage cutoff.

-10.0 1 5 -7.5

-5.0 Z''

-2.5

2.5 2.5 5.0 7.5 10.0 12.5 15.0 Z'

Figure 4.3.2: Impedance spectra of the Ni-H2 cell using the D cell nickel hydroxide electrode measured at the end of charge of cycles 1 and 5. The cell was cycled at C/3 rate to 100% capacity.

1.5

1.4

1.3

1.2 E (Volts) 1.1

1.0

0.9 0123 Time (Hours)

Figure 4.3.3: Voltage profile during the charge-discharge of the Ni-H2 cell using the D cell nickel hydroxide electrode for 5 cycles. The discharge cut off voltage is 1.0 V The cell was cycled at C/3 rate to 80% capacity.

O2 evolution 1.5

50% Overcharge

1.4 No overcharge

1.3 50% Overcharge E (Volts) E

1.2 No overcharge

1.1 0123456 Time (Hours)

Figure 4.3.4: Voltage profile during the charge-discharge of the Ni-H2 cell using the D cell nickel hydroxide electrode comparing overcharge with no overcharge. The cell was cycled at C/6 rate with no voltage cutoff.

4.4. Performance of the D cell Electrode in Series Cell

Two cells in series were assembled in the bolt-nut cell setup using the ‘D cell’

nickel hydroxide electrode and the platinum electrode as shown in Fig. 4.4.1. A bipolar

plate is used to provide electrical connection between the cells. The requirements of a bipolar plate are (a) good electronic conductivity (b) good gas permeability and (c) no

KOH migration. Three different materials, nickel chromium mesh, carbon cloth,

spectracarb 2050-A (porous carbon paper) were considered as bipolar plates. The

variation of open circuit voltage during the cell assembly for different bipolar plates is

shown in Fig. 4.4.2. The series cells are first purged in nitrogen gas, when the OCV is

about 0.25-0.4 V for different bipolar plates in the time period 0 to 0.1h. The cell voltage

increases rapidly when purged with hydrogen gas. This voltage rises to about 2.52 V for

the series cells using carbon cloth and Ni-Cr mesh. The cell voltage corresponds to

voltage of two cells in series. However for the spectracarb 2050-A, the cell voltage rises

to only about 1.36V, which is more than a single cell voltage in discharged state and less

than ‘two cells in series’ voltage indicating that the spectracarb was shorting the two cells

by KOH migration. It is noted that the D cell nickel hydroxide electrode is double-sided

and results in formation of an opposite cell when there is KOH in the current collector.

This results in a single cell rather than two cells in series.

The series cell utilizing carbon cloth as the bipolar plate was stopped after one

cycle because of poor charging and discharging. The voltage profile during charge is

similar to that shown in Fig. 4.4.6. The series cell with Ni-Cr mesh as current collector

performed better at C/3 rate to 80% capacity. The voltage profile during charge-discharge

for cycles 1, 5 and 10 is shown in Fig. 4.4.2. The discharge cut-off voltage is 2V. The

coulometric efficiency is about 100% for cycle 1. However, the discharge time

(coulometric efficiency) decreases with cycling. The charge voltage increases and the

discharge voltage decreases with cycles clearly indicating an increase in impedance on

cycling. This is evident in the impedance data measured at the end of discharge for cycle

1, 5 and 10, as shown in Fig. 4.4.4. The voltage profile for the 10th cycle (Fig. 4.4.5) shows voltage fluctuations at the end of charge indicating significant oxygen evolution in

the cell.

The voltage fluctuations at the end of charge for cycles 4, 7 and 10 are shown in

Fig. 4.4.5. It is evident that the fluctuations increase on cycling indicating that the oxygen

evolution increases with cycles. This results in poor cell performance as shown by the

poor voltage profile during the 12th charge (Fig. 4.4.6).

The double-sided nickel hydroxide electrode causes uneven charging and results

in overcharging of one of the electrodes. By this, the state of charge of one nickel

hydroxide electrode at the end of discharge is not actually 0% but higher. On further

cycling, the SOC of one electrode increases in comparison to the other resulting in more and more oxygen evolution for the overcharged electrode at the end of charge. This results in cell failure.

Positive Negative electrode electrode Current Ni(OH)2 Platinum collector

Separator

Cell 1 Cell 2 Bipolar plate

Figure 4.4.1: Schematic picture of two nickel-hydrogen cells in series.

Ni-Cr mesh 2 Spectracarb 2050 Carbon cloth E (Volts) 1

0 0 0.25 0.50 0.75 Time (Hours)

Figure 4.4.2: Variations in the open circuit voltage of series cell with different bipolar plates. The cell is purged first with nitrogen and then with hydrogen.

3.2 10 3.0 5 1 2.8

2.6 E (Volts) 2.4 1 2.2 10 5 2.0 0123 Time (Hours)

Figure 4.4.3: Voltage profile during the charge-discharge of the series cell. The discharge cut off voltage is 2.0 V The cell was cycled at C/3 rate to 80% capacity.

-40 1 5 10 -30

Z'' -20

-10

0 0 10203040 Z'

Figure 4.4.4: Impedance spectra of the series cell measured at the end of discharge of cycles 1, 5 and 10. The cell was cycled at C/3 rate to 80% capacity.

100

3.05

10 3.00 7

E (Volts) 4 2.95

2.90 1.5 2.0 2.5 3.0 Time (Hours)

Figure 4.4.5: Voltage profile at the end of charge for cycles 4, 7 and 10 of the series cell. The discharge cut off voltage is 2.0 V The cell was cycled at C/3 rate to 80% capacity.

3.1

3.0 E (Volts) 2.9

2.8 0123 Time (Hours)

Figure 4.4.6: Voltage profile for the 12th charge of the series cell. The discharge cut off voltage is 2.0 V The cell was cycled at C/3 rate to 80% capacity.

101

4.5. Conclusions

The ‘D cell’ nickel hydroxide was tested in liquid and nickel-hydrogen cell

configurations. The importance of fixing the discharge cut-off voltage was studied.

Overcharging the electrode results in voltage fluctuations at the end of discharge due to significant oxygen evolution.

The performance of the electrode in two cells in series was studied. The nickel hydroxide electrode contains active material on both sides which results in shorting of the cell because the potassium hydroxide solution seeps into the current collector. Therefore it is necessary to fabricate a nickel hydroxide electrode with active material on one side only so as to construct series cells.

References

1. A. H. Zimmerman, Proceedings - Electrochemical Society, 94-27, 268 (1994).

102

5. Fabrication and Formation of Nickel Hydroxide Electrode

5.1. Introduction

In chapter 4 the performance of the nickel hydroxide electrode taken from a D size nickel metal hydride battery was discussed in detail. The D cell electrodes are loaded with nickel hydroxide on both sides and as a consequence cannot be used as electrodes in a series cell configuration because they form opposite cells with the platinum electrode in the front and the platinum electrode in the back due to KOH seepage. This results in the drop in voltage of the series cells. A nickel hydroxide electrode with active material on one side was fabricated as an alternative to the D cell electrode.

In this chapter, the fabrication of the nickel hydroxide electrode and its subsequent formation will be discussed in detail. The effect of various electrode components on the electrode performance was studied. The main objective was to develop a nickel hydroxide electrode that can be used in the development of the low pressure nickel hydrogen battery.

5.2. Existing Methods of Fabrication

The nickel hydroxide electrode is the positive electrode in numerous battery

chemistries like the nickel-cadmium batteries, the nickel-zinc batteries, the nickel-metal

hydride batteries and the nickel-hydrogen batteries. The common methods of adding the

nickel hydroxide active material to the substrate are (i) mechanical impregnation1, 2 and

(ii) chemical3 /electrochemical impregnation4-7.

103

In the mechanical impregnation method, the active material is blended with a binder solution. Cobalt and zinc are added to increase the conductivity of the paste and the overvoltage for oxygen evolution. The paste is extruded or rolled or pasted into the substrate. The two major types of electrode fabricated by mechanical means are (i) pasted type nickel electrode1, 2 and (ii) plastic bonded electrode8. Of the two methods, the pasted nickel electrode is more commonly used now.

In the chemical/electrochemical impregnation method, the active material is either chemically or electrochemically precipitated into the substrate3-7. The substrate used for this method has to be sintered and therefore they are called as sintered electrodes. This technique is discussed below in detail.

5.2.1. Pasted Nickel Electrode1, 2

The nickel hydroxide active material formulation is mechanically introduced into a porous nickel substrate. The substrate can be non-woven nickel fiber or a variety of materials. Recently, the nickel foam has been used as the substrate. The main advantage of the pasted type electrode is the simplicity of the fabrication method. The nickel foam is typically produced by coating polyurethane foam with a layer of nickel either by electrodeposition or chemical vapor deposition, followed by a heat treatment process to remove the base polyurethane. The foam so obtained is light in weight and is about 95% porous.

5.2.2. Plastic Bonded Electrode8

The plastic bonded electrode was primarily developed for nickel-zinc batteries. In this method, the active material - nickel hydroxide, and the additives are blended together with polytetrafluoroethylene (PTFE) binder and then roll-bonded into a porous three-

104

dimensional structure by fibrillating with binder. The fibrillated PTFE locks the active

material firmly into the structure. This eliminates shedding of the active material which is

a problem with the pasted nickel electrode. The base structure is graphite eliminating

metallic nickel thereby saving cost and reducing weight. This method is complex in

comparison to the pasted type nickel electrode.

5.2.3. Sintered Electrode3-7

Sintered electrodes involve an expensive and complicated sequence of manufacturing steps and consequently require significant capital investment. The sintered

substrate is obtained by pasting nickel filamentary fiber onto a perforated foil. These

nickel fibres are then sintered by annealing in a high temperature furnace in

hydrogen/nitrogen atmosphere resulting in a conductive skeleton of nickel. Nickel hydroxide is precipitated into the sinter skeleton by chemical or electrochemical impregnation techniques. The main advantage of these electrodes is high rate and power

capability; however, they contain a relatively low ratio of active materials to inactive materials when compared to the pasted type electrode.

Of the different manufacturing techniques discussed, the pasted nickel electrode was chosen because of low cost and ease of fabrication.

5.3. Fabrication Methodology

5.3.1. Substrate

Nickel foam was used as the substrate because of its large porosity (95%).

Precision sized, high purity and uniform nickel foam from INCO was used. The thickness

of the nickel foam is about 1.6 mm and the density is about 420 g/m2. The number of

105

pores per inch is 110. A scanning electron microscope image of the nickel foam is shown

in Fig. 5.3.1.

5.3.2. Paste Formulation

Nickel hydroxide powder from Kansai Catalyst Corporation LTD was used as the active material. The composition of this nickel hydroxide formulation as obtained is 54%

Ni, 3.2% Co and 4.5% Zn. According to the product information, 4.5% Zn and 0.7% Co were mixed as a solid solution with nickel hydroxide. These particles were then coated with 2.5% Co. The cobalt coating is a mixture of cobalt hydroxide and cobalt oxyhydroxide. The addition of cobalt to nickel hydroxide enhances the electrochemical

performance of the electrodes and has been studied by many researchers1, 9-25. The two major beneficial effects of adding cobalt are (i) increase in electrode conductivity1, 9-12

and (ii) increase in the oxygen overvoltage 13-15. This results in increased electrode

capacity. The other beneficial effects are (i) lowering of the β-nickel oxyhydroxide peak

potential16, 17, (ii) minimization of the γ-nickel oxyhydroxide growth18, (iii) increase in

electrode mechanical resistance19, 20, and (iv) increase in proton diffusion17. Addition of

zinc increases the long term stability of the electrodes and delays the formation of γ-

nickel oxyhydroxide phase12, 26. The average size of the nickel hydroxide particles is

about 12 µm. The active material was blended with a binder and a solvent using a

homogenizer. The binder was polyvinylidene fluoride (PVDF, Kynar 2800) and the

solvent was N-methyl pyrrolidone. The binder provides adhesion of particles to one

another and to the substrate.

106

Filamentary nickel powder (Novamet, INCO type 210, 0.5-1 µm) was used as a

filler to enhance the conductivity of the paste. The addition of nickel powder to the

electrode increases the utilization27, 28 and will be discussed in detail in section 5.5.3.

Figure 5.3.1: SEM image of INCO nickel foam of thickness 1.6 mm and cell size 550- 700 µm.

5.3.3. Screen Printing

Screen printing was used to mechanically press the formulated paste into the

nickel foam substrate. Screen printing is one of the several thick film printing technologies used for selective coating of flat surfaces. The technology was originally

developed for the production of miniature, robust, and cheap electronic circuits with high

reliability and versatility 29. In the last decade, thick film printing has been studied to

fabricate batteries 30-32 and fuel cells 33-43. The screen printing technique reproduces the image from the image carrier to the substrate. A mesh or screen is used as the image carrier. The paste is then pressed through the pores in the screen using a squeegee on to

107

the substrate as shown in Fig. 5.3.2. For the fabrication of nickel hydroxide electrode, the nickel foam was used as the substrate. For good performance of the nickel hydroxide electrode, it is desired that all the paste is pushed into the foam as shown in Fig. 5.3.2 rather than forming a thick film on top. This is achieved by adjusting the several process parameters including the squeegee pressure, the screen height from the substrate and squeegee height from the screen. Other parameters like the viscosity of the paste and the screen pore size are critical for efficient impregnation of the paste into the substrate. The squeegee is cycled as many times as needed to get the desired loading level into the substrate.

The nickel foam electrode with the paste is then dried in vacuum at 100o C overnight. This completes the fabrication of the nickel hydroxide electrode. The results discussed in this chapter were based on electrodes fabricated by screen printing unless otherwise specified.

Squeegee Frame Paste Mesh

Ni Foam

Figure 5.3.2: Schematic illustration of screen printing technology to mechanically press the paste into the nickel foam substrate.

108

5.4. Formation of the Nickel hydroxide Electrode

The next step is the formation of the nickel hydroxide electrode. The electrode

obtained at the end of the fabrication process contains nickel hydroxide particles held

together by the binder PVDF. The nickel hydroxide particles themselves have relatively

low electronic conductivity and are coated with cobalt oxide to enhance the conductivity.

During formation the electrodes are charged and discharged successively. The cobalt

oxide is oxidized to cobalt oxyhydroxide forming highly conductive films during this

process1, 2. This decreases the electronic resistance between the nickel hydroxide

particles, which results in good utilization of the active material.

The Ni foam containing the nickel hydroxide paste was soaked in 26 wt% KOH overnight. The binder PVDF is hydrophobic; however upon soaking overnight the pores of the nickel foam are filled with KOH due to capillary effects. The soaked electrode was then formed in a BAS liquid cell shown schematically in Fig. 5.4.1. The nickel

hydroxide electrode was the working electrode. A platinum mesh was used as the counter

electrode. A reversible hydrogen electrode was used as the reference electrode. The cell was purged with nitrogen gas.

The steps involved in the formation of the nickel hydroxide electrode are:

1. The electrode was first charged at C/10 or C/20 rate (with or without

overcharging) based on the theoretical capacity of the nickel foam electrode.

2. This was followed by a rest period for 1 hour.

3. Then the electrode was discharged at C/10 or C/20 rate with a discharge

voltage cutoff of 1.2 V.

4. This was then followed by a 1h rest period.

109

The electrode is cycled continuously by repeating steps 1 to 4 until the electrode

was formed.

Ni Foam based electrode Reference H2 electrode Pt mesh

N2 N2 Inlet Outlet

26 wt. % KOH

Figure 5.4.1: Schematic diagram of the cell setup used for the formation of the nickel hydroxide electrode.

5.4.1. Overview of Reactions in the Nickel Hydroxide Electrode

The electrochemical reactions occurring during charge and discharge and the

reaction pathways under different conditions are represented in Fig. 5.4.2. It is well

44 understood that during charging the β-Ni(OH)2 phase is converted β-NiOOH phase .

This is the principal reaction in the nickel hydroxide electrode and is given as:

charg e Ni() OH+++ OH−−←⎯⎯⎯⎯⎯ ⎯ ⎯⎯→ NiOOH H O e [5.1] 22discharg e

However, on overcharging the electrode, i.e. at sufficiently high overpotentials,

the beta form of nickel oxyhydroxide is converted to the γ-NiOOH as shown in the Fig.

5.4.2. The discharge of the γ-NiOOH phase results in α-Ni(OH)2 phase without the

formation of the β-Ni(OH)2 phase. The α-Ni(OH)2 thermally restructures to the beta form

of nickel hydroxide. The potential for these reactions depends on the electrolyte

concentration, temperature and the type of additives. It is difficult to define a reversible

110

potential in the traditional sense because the electrode always operates at a mixed

potential. In addition to these reactions, there is oxygen evolution occurring in parallel with the formation of both the β and γ-NiOOH phases and is given as:

−−→++ 42OH H22 O O 4 e [5.2]

Charge β-Ni(OH)2 β-NiOOH Discharge

Thermal Overcharge rearrangement

Charge α-Ni(OH)2 γ-NiOOH Discharge

Figure 5.4.2: Schematic diagram for interconversion of active material phases in the nickel hydroxide electrode44.

5.4.2. Overview of Electrochemical Voltage Spectroscopy

Thaller, Zimmerman and To45, 46 used electrochemical voltage spectroscopy to

study the effect of cycling conditions like temperature, electrolyte concentration, number

of cycles, and recharge ratio on the usable capacity of the electrodes and the efficiency of

the charging reaction. In this work, electrochemical voltage spectroscopy was used to

study the effect of incomplete formation, the binder content and the nickel content on the

positive electrode performance.

111

Electrochemical voltage spectroscopy is a form of cyclic voltammetry. A slow

scan rate of 2 µV/s was used for charging and discharging the electrode, in the voltage

range 1.2 to 1.48 V. It is assumed that the electrode kinetics is in equilibrium at such slow

scan rates. The charge and discharge peaks observed were used to compare the

performance of different electrodes.

5.4.3. Mechanism of Formation

The voltage variations during the charge period, the rest period and the discharge

period were monitored. Impedance data was measured at the end of the charge and

discharge periods. The voltage profiles during charge and discharge for the first two

cycles of an electrode containing 10% PVDF and 90% Ni(OH)2 by weight basis are

shown in Fig. 5.4.3. The electrode was charged at C/10 rate with 50% overcharge (15 h)

and discharged at C/10 with a voltage cutoff of 1.2 V. During charging the active

material nickel hydroxide is converted to nickel oxyhydroxide according to equation

[5.1]. During discharge the reverse of the reaction occurs.

It is evident from the figure that the voltage during the 2nd charge is lower than that of the 1st charge. The high voltage during the first charge cycle is due to the poor

electrical conductivity of the particles. However upon charging during cycle 1, in

addition to the oxidation of nickel hydroxide reaction, the cobalt oxide present is oxidized

to highly conductive cobalt oxyhydroxide establishing a highly conductive network in the

active mass within the nickel foam electrode1. This results in a lower charge voltage in

the 2nd cycle. A change in slope is observed in the time period 10-15 h (overcharge) and

this corresponds to the evolution of oxygen after the completion of the nickel

oxyhydroxide reaction.

112

Utilization is defined as the ratio of the discharge capacity obtained to the theoretical capacity of the electrode. The utilization of the 10% PVDF containing

electrode is about 83.1%.

The voltage profile in the beginning of the 1st charge in the time period 0 to 30

min is shown in Fig. 5.4.4. It is evident that the voltage rises initially and then falls

during the first 3 minutes. At the start of charging, the poor conductivity of the particles

causes the potential to rise steeply, however the formation of cobalt oxyhydroxide on

charging enhances the conductivity resulting in the decrease in potential.

The impedance data of the 10% PVDF containing nickel hydroxide electrode

measured before forming is shown in Fig. 5.4.5. It is evident that the impedance profile

indicates capacitive behavior within the frequency range used (20,000 to 0.2 Hz) because

of the large charge transfer resistance, which is due to the poor conductivity of the paste

impregnated into the nickel foam and substantiates the observations noted for the voltage

profile during charging. The impedance of the electrode measured after discharge at the

end of cycles 1 and 2 is shown in Fig. 5.4.6. The data shows a semi-circle corresponding

to the charge transfer resistance for the nickel hydroxide to nickel oxyhydroxide reaction

and the impedance is smaller in comparison to that measured before forming. It is also

evident from the figure that the charge resistance (semi-circle) decreases with cycling.

113

Change of slope

1.5 1 charge

1.4 2 charge 50% overcharging

1.3 1 discharge E (Volts) 2 discharge 1.2 Discharge voltage cutoff

1.1 051015 Time (Hours)

Figure 5.4.3: Variations in voltage during charge and discharge of cycles 1 and 2. The electrode composition is 10% PVDF and 90% Ni(OH)2 by weight.

1.45

1 charge

1.44

Rise and fall

E (Volts) in potential 1.43

1.42 0 0.1 0.2 0.3 0.4 0.5 Time (Hours)

Figure 5.4.4: Voltage variation in the time period 0 to 0.5h during the first charge. The electrode composition is 10% PVDF and 90% Ni(OH)2 by weight.

114

-3000 Before forming

-2000

-1000 Z'' (A)

1000 0 1000 2000 3000 4000 Z'

-10 After cycle 1 After cycle 2

-5

Z'' (B) 0

5 0 5 10 15 Z'

Figure 5.4.5: Impedance data of 10% PVDF containing electrode: (A) Before formation, (B) End of discharge after cycles 1 and 2.

115

The voltage variations during charge and discharge for a nickel hydroxide electrode containing 15% PVDF and 85% Ni(OH)2 by weight basis is shown in Fig.

5.4.6. It is evident that the voltage profiles do not change significantly from cycle 2 to cycle 3. The utilizations for cycle 2 and cycle 3 are 85.2 and 85.1, respectively. The impedance data measured at the end of discharge for cycles 1, 2 and 3 is shown in Fig.

5.4.7. It is evident from the figure that the impedance profile does not change significantly for cycles 2 and 3. Figs. 5.4.6 and 5.4.7 clearly show that the electrode formation is complete at the end of the 2nd cycle.

1.5 2, 3 charge

1.4

1.3 2, 3 discharge E (Volts)

1.2

1.1 0 5 10 15 Time (Hours)

Figure 5.4.6: Voltage profile during charge and discharge of cycles 1 and 2 for 15% PVDF containing nickel hydroxide electrode.

116

-15

-10

-5

Z'' End of 1 discharge

2 3 0

5 0 5 10 15 20 Z'

Figure 5.4.7: Impedance data of 15% PVDF containing electrode measured at the end of discharge after cycles 1, 2 and 3.

5.4.4. Effect of Incomplete Formation

Some of the nickel hydroxide electrodes were charged to only 60% of their theoretical capacity during formation. The effect of such incomplete formation was studied using electrochemical voltage spectroscopy (EVS). It is noted that this mode of incomplete formation is improbable; however it gives useful information about the formation mechanism and the severe consequences of incomplete charging during formation.

The EVS scans of a completely formed and an incompletely formed nickel hydroxide electrode containing 16% PVDF and 84% Ni(OH)2 (weight basis) are compared in Fig. 5.4.8. The voltage is IR corrected for the solution resistance. It is noted that the completely formed electrode was charged to its theoretical capacity (no

117

overcharge) and the incompletely formed electrode was charged to only 60% of its

theoretical capacity. The current measured in the EVS scan is normalized with the

theoretical electrode capacity and the units are h-1. The positive and negative current

corresponds to the charge and the discharge of the electrode. The completely formed

nickel hydroxide electrode shows peaks only for the β-NiOOH phase and β-Ni(OH)2 phase. However, the incompletely formed electrode shows peaks for both forms, β and γ-

NiOOH phases, and the α-Ni(OH)2 phase. The β-Ni(OH)2 peak is absent due to the

formation of the γ-NiOOH which reduces directly to α-Ni(OH)2 as discussed in section

5.4.1.

The peak potentials for the different reactions are listed in Table 5.1. It is evident from the table that the β-NiOOH peak potential is the same for the both the electrodes.

However the discharge potential peaks are different because of the different phases formed on discharge. A lower discharge potential is observed for the incompletely formed electrode than the completely formed electrode and this results in lower watt-hour capacity. It is also evident from the Figure 5.4.8 that the oxygen evolution rate is larger and the oxygen evolution potential is lower for the incompletely formed electrode.

To explain the effects of incomplete formation, the mechanism proposed by

Oshitani et al.1, 2 was extended to this case. When the electrode is charged incompletely,

only the cobalt oxide on the nickel hydroxide particles close to the nickel foam fiber is

oxidized and that further away is not oxidized forming a poorly conductive network of

the nickel hydroxide particles as shown schematically in Fig. 5.4.9. This results in

charging (β-NiOOH phase) and then overcharging of the particles (γ-NiOOH) close to the

118

nickel fiber. These discussions clearly suggest the importance of charging the nickel

hydroxide electrodes completely during formation.

The SEM image of a fresh 16% PVDF electrode is shown in Fig. 5.4.10A. The

nickel foam and the nickel hydroxide particles are evident. In contrast the SEM image of

an incompletely formed 12.7% PVDF electrode taken after the EVS scan is shown in Fig.

5.4.10B. It is evident that nickel hydroxide particles are structurally damaged because of poor formation.

0.6 Incomplete formation Complete formation O2 ) 0.4 -1 γ-NiOOH

0.2 β-NiOOH

0 Current/Capacity (h Current/Capacity -0.2 β-Ni(OH)2

α-Ni(OH)2 -0.4 1.2 1.3 1.4 1.5 E (Volts)

Figure 5.4.8: Comparison of EVS scans for completely and incompletely formed nickel electrode containing 16% PVDF and 84% Ni(OH)2.

119

Table 5.1: Peak potentials for different electrochemical reactions.

Type of β-NiOOH (V) γ-NiOOH (V) β-Ni(OH)2 (V) α-Ni(OH)2 (V) Formation Complete 1.419 - 1.308 - Incomplete 1.418 1.459 - 1.217

Ni Foam CoOH Charging

PVDF Ni(OH)2 Incomplete

Figure 5.4.9: Schematic representation of complete and incomplete charging of the nickel hydroxide electrode.

120

(A)

(B)

Figure 5.4.10: SEM image of nickel hydroxide electrodes: (A) Fresh 16% PVDF electrode, (B) Used 12.7% PVDF electrode incompletely formed.

5.4.5. Effect of Different Rates of Formation

The effect of different charging rates during formation on the electrode

performance was studied and is discussed in this section. The charging rates considered

are C/10 and C/20.

121

Two electrodes, one containing 10% PVDF and the other containing 10.7%

PVDF were cycled at C/10 and C/20 rate with 50% overcharge, respectively. Utilization

is defined as the ratio of the discharge capacity obtained to the theoretical capacity of the

electrode. The utilization during the first two cycles as a function of charge rate is listed

in Table 5.2. It is evident that the utilization is about the same for the two charging rates.

The utilization increases by a small percentage (1-2%) upon cycling. No significant

difference was observed in the impedance spectra measured at the end of discharge for

the electrodes cycled at different rates. However, the impedance spectra measured at the

end of charge showed a smaller charge transfer resistance for the C/20 rate in comparison

to the C/10 rate.

Electrochemical voltage spectroscopy was performed on the formed electrodes

and the scans are compared in Fig. 5.4.11. It is evident that the peak potentials for the β-

NiOOH and the β-Ni(OH)2 are about the same for electrodes charged at different rates.

However for the electrode formed at C/20 rate, the γ-NiOOH peak is also evident. In either case, there is no clear separation of the NiOOH peak from the oxygen evolution peaks, suggesting that at the end of charging the oxygen evolution is significant.

It is concluded that no significant differences were observed for the electrodes formed at two different charging rates.

Table 5.2: Utilization of electrode containing about 10% PVDF and 90% Ni(OH)2 charged at different during formation. Charge rate Utilization Cycle 1 Cycle 2 C/10 82.1 83.1 C/20 81.0 83.1

122

0.2 C/10 C/20

0.1 ) -1

-0.1 Current/Capacity (h

-0.2

-0.3 1.2 1.3 1.4 1.5 E (Volts)

Figure 5.4.11: Comparison of EVS scans for two different electrodes, one charged at C/10 and another at C/20 rate with 50% overcharge. The electrode composition is 10% PVDF and 90% Ni(OH)2 by weight.

5.4.6. Effect of Overcharging during Formation

The effect of overcharging the electrodes during formation is considered in this

section.

5.4.6.1. No Overcharge during Formation

An electrode containing 15% PVDF and 85% Ni(OH)2 was charged and

discharged at C/10 rate without overcharge. The impedance data measured at the end of discharge for cycles 1 to 5 is shown in Fig. 5.4.12A. It is evident that the charge transfer resistance decreases with increasing cycles and does not change significantly for cycles 4 and 5 which indicates that the electrode is completely formed. The decrease in charge transfer resistance corresponds to the enhancement in conductivity between the nickel hydroxide particles due to the formation of the highly conductive cobalt oxyhydroxide

123

from cobalt oxide. When all the conversion is complete, the charge transfer resistance does not change significantly as in cycles 4 and 5. The impedance data measured at the end of charge for cycles 1 to 5 is shown in Fig. 5.4.12B. The charge transfer resistance decreases from cycle 1 to cycle 2 as expected, however surprisingly it increases from cycle 3 to 5.

5.4.6.2. 50% Overcharge during Formation

Another electrode containing 15% PVDF and 85% Ni(OH)2 was charged at C/10

rate with 50% overcharge for the first two cycles. It was established in section 5.4.3 that

the electrode formation is complete by two cycles. For cycle 3, the electrode was not

overcharged. Impedance data measured at the end of discharge and charge are shown in

Figs. 5.4.13A and 5.4.13B, respectively. It is evident that the charge transfer resistance at

the end of discharge decreases from cycle 1 to cycle 2 as expected due to the

enhancement in conductivity due to the formation of cobalt oxyhydroxide. However, for

cycle 3 the charge transfer resistance increases slightly when the electrode is not

overcharged. The charge transfer resistance measured at the end of charge (Fig. 5.4.13B)

is about the same for cycle 1 and 2. For cycle 3, when the electrode is not overcharged

the charge transfer resistance is significantly larger. This is because when the electrode is

overcharged, there is more of nickel oxyhydroxide formation (higher utilization observed,

although with more oxygen evolution) in comparison to the case without overcharge. It is

noted that nickel oxyhydroxide is a good electronic conductor whereas nickel hydroxide

is a poorly conducting, semiconductor47. Therefore the formation of more nickel

oxyhydroxide when the electrode is overcharged (cycles 1 and 2) results in lower charge

transfer resistance.

124

5.4.6.3. Comparison of No Overcharge and 50% Overcharge

The utilization of the two electrodes, one charged with 50% excess and the other without overcharge is compared in Fig. 5.4.14. For the electrode cycled without overcharge, the utilization is about 59.9% for cycle 1 and increases to 81.1% for cycle 2.

The utilization at the end of cycle 5 is about 81.6%. For the electrode cycled with 50% overcharge, the utilization is about 85.6% for the first cycle and increases to 88.3% for the second cycle. The utilization of cycle 3, which is without overcharge, is about 81% and is same as the utilization obtained for the electrode formed without any overcharge.

The impedance data measured at the end of the formation for electrodes, one with overcharge and the other without overcharge is compared in Fig. 5.4.15. It is evident that the charge transfer resistance is the same for both the electrodes. Thus, there is no significant difference between charging with 50% overcharge and charging without any overcharge.

125

-20 Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 -10 (A) Z''

10 0102030 Z'

-1.5 1 2 3 -1.0 4 5

Z'' -0.5 (B)

0.5 0 0.5 1.0 1.5 2.0 Z'

Figure 5.4.12: Impedance spectra of the electrode measured at the end of discharge (A) and charge (B) without overcharge at C/10 rate for the first 5 cycles.

126

-20 1 2 3 -10 (A) Z''

10 0102030 Z'

-2 After charge 1 - 50% overcharge After charge 2 - 50% overcharge After charge 3 - No overcharge

-1

Z'' (B)

1 0123 Z'

Figure 5.4.13: Impedance spectra of an electrode measured at the end of discharge (A) and charge (B) at C/10 rate with 50% overcharge for the first two cycles.

127

No overcharging 50% overcharging

100

40 Utilization, % 20

0 0123456 Cycle No.

Figure 5.4.14: Utilization as a function of cycles for electrodes formed with 0% and 50% overcharge.

-20 No overcharge 50% overcharge

-10 Z''

10 0102030 Z'

Figure 5.4.15: Impedance spectra measured after discharge for electrodes formed with 0% and 50% overcharge.

128

5.5. Characterization of Nickel Hydroxide Electrode

5.5.1. Effect of PVDF Content on Electrode Performance

The effect of PVDF content on the electrode performance was investigated and the results are discussed in this section. Electrodes containing 5, 10 and 15% PVDF were

formed at C/10 rate with 50% overcharge and the utilization of these electrodes for the

first two cycles is shown in Fig. 5.5.1. The theoretical capacity of these electrodes is

similar and is listed in Table 5.3. It is evident from the Fig. 5.5.1 that the utilization

increases with increase in the PVDF content by about 9% from 5% PVDF to 15% PVDF

for the 2nd cycle. This indicates that the electrode with higher PVDF content gives better

performance. It is also evident from the figure that the utilization increases on cycling for

the 10 and 15% PVDF electrodes and decreases with cycling for the 5% PVDF electrode.

It was noted that the 5% PVDF electrode was shedding particles on cycling and this is

probably the reason for the decrease in utilization.

Impedance data measured at the end of end of discharge and charge for electrodes

containing different amounts of PVDF is shown in Fig. 5.5.2. It is evident that the charge

transfer resistance measured at the end of discharge decreases with increasing PVDF

content in the electrodes. This result reinforces the observation that the electrodes with

higher PVDF content yields higher utilization. The charge transfer resistance measured at

the end of charge decreases from 5% to 10% PVDF. However it increases for the 15%

PVDF and this behavior is possibly related to the slightly different capacities of the

electrodes.

The EVS scans of electrodes containing different amounts of PVDF are compared

in Fig. 5.5.3. The peaks for the β-NiOOH phase and the β-Ni(OH)2 phase are evident for

129

all the electrodes. The peak potentials for the β-NiOOH phase and the β-Ni(OH)2 phase shifts towards lower values with increase in the PVDF content and are listed in Table 5.4.

The peak for the γ-NiOOH phase is absent in the 5% PVDF electrode. However, they are present in the 10 and 15% PVDF electrode and the peak potentials decrease with increase in the PVDF content (Table 5.4). This suggests that the electrodes containing larger amounts of PVDF are overcharged in some regions resulting in the formation of the γ-

NiOOH phase. The cobalt oxyhydroxide formed does not provide a sufficiently conductive network resulting in the overcharging of the 10 and 15% PVDF electrodes.

The α-Ni(OH)2 peak is present in all the electrodes and is small, indicating that the γ-

NiOOH phase formed at the end of charging is also small. The peak currents for the β-

NiOOH phase and the β-Ni(OH)2 phase increase with PVDF content. Due to the different

loading levels (difficult to precisely control during fabrication), it is noted that the theoretical capacity of these electrodes decreases with increase in the PVDF content. The

increase in peak currents is possibly related to the normalization of the current with the

electrode capacity.

130

90 5% PVDF 88 10% PVDF 15% PVDF 86

80 Utilization, % 78

74 12 Cycle No.

Figure 5.5.1: Utilization as a function of cycles for electrodes containing different amounts of PVDF content. The electrodes were cycled at C/10 rate with 50% overcharge.

Table 5.3: Capacity of electrodes containing different amounts of PVDF. PVDF content, % Theoretical Capacity (mAh/cm2) 5 15.2 10 14 15 12.8

131

-10 5% PVDF 10% PVDF 15% PVDF

-5 (A) Z''

5 0 5 10 15

-0.25

0 (B) 5 % Z'' 10 % 15 % 0.25

0.50 0.25 0.50 0.75 1.00

Figure 5.5.2: Impedance spectra measured at the end of discharge (A) and charge (B) for electrodes containing different PVDF content. The electrodes were formed at C/10 rate with 50% overcharge.

132

0.2 5% PVDF 10% PVDF 15% PVDF 0.1 ) -1

-0.1 Current/Capacity (h Current/Capacity -0.2

-0.3 1.2 1.3 1.4 1.5 E (Volts)

Figure 5.5.3: EVS scans for electrodes containing different PVDF content formed at C/10 rate with 50% overcharge.

Table 5.4: Peak potentials for different electrochemical reactions of electrodes containing different amounts of PVDF.

PVDF β-NiOOH (V) γ-NiOOH (V) β-Ni(OH)2 (V) α-Ni(OH)2 (V) content, % 5 1.425 - 1.309 1.225 10 1.424 1.450 1.307 1.222 15 1.422 1.436 1.304 1.220

133

The effect of PVDF content on the electrode performance was also studied for a set of electrodes with low theoretical capacities. The utilization of 10% and 15% PVDF containing electrodes for cycles 1 and 2 is shown in Fig. 5.5.4. The theoretical capacity of these electrodes was similar and is listed in Table 5.5. It is evident from the figure that the utilization increases with PVDF content and with cycles and is similar to the results obtained for electrodes with higher Ah capacities.

89.0

88.0 Cycle 1 Cycle 2 87.0

86.0

85.0

84.0 Utilization,% 83.0

82.0

81.0 10% PVDF 15% PVDF

Figure 5.5.4: Utilization of electrodes containing different amounts of PVDF content at lower capacities. The electrodes were cycled at C/10 rate with 50% overcharge.

Table 5.5: Capacity of electrodes containing different amounts of PVDF. PVDF content, % Theoretical Capacity (mAh/cm2) 10 6.5 15 7.5

134

5.5.2. Effect of Loading Density

The loading levels of the active material can be varied depending on the number

of times the squeegee is cycled during screen printing. In this section, the effect of

loading levels on the electrode performance is discussed.

Figs. 5.5.5 and 5.5.6 show the utilization of 10% PVDF and 15% PVDF

electrodes, respectively, having different theoretical capacities because of different

loading levels. It is evident that the electrodes with lower theoretical capacity (lower

loading level) have higher utilization. The impedance data measured at the end of charge

and discharge for 10% PVDF and 15% PVDF electrodes is shown in Figs. 5.5.7 and 5.5.8

respectively. It is evident that the charge transfer resistance measured at the end of charge

and discharge is smaller for the electrodes with the higher loading level. Higher loading

levels correspond to electrodes with larger active area (higher active material) which

means lower charge transfer resistance. It is noted that the charge transfer resistance is the dominant impedance at the end of discharge than at the end of charge and therefore this effect is magnified for the impedance measured at the end of discharge as shown in

Figs. 5.5.7 and 5.5.8.

135

85.5

85.0

84.5

84.0

83.5

Utilization, % Utilization, 83.0

82.5

82.0 6.51 13.95 Theoretical capacity, mAh/cm2

Figure 5.5.5: Utilization of 10% PVDF electrodes with different capacities (different loading levels). The electrodes were cycled at C/10 rate with 50% overcharge. 89.0

88.0

87.0

86.0

85.0 Utilization, % Utilization,

84.0

83.0 7.1 12.8 Theoretical capacity, mAh/cm2

Figure 5.5.6: Utilization of 15% PVDF electrodes with different capacities (different loading levels). The electrodes were cycled at C/10 rate with 50% overcharge.

136

-20 6.5 14

-10 (A) Z''

10 0102030 Z'

-0.50 6.5 14 -0.25

(B) 0 Z''

0.25

0.50 0.50 0.75 1.00 1.25 1.50 Z'

Figure 5.5.7: Impedance spectra measured at the end of discharge (A) and charge (B) for electrodes containing 10% PVDF with different capacities. The electrodes were formed at C/10 rate with 50% overcharge.

137

-20 7.1 12.8

-10 (A) Z''

10 0102030 Z'

-0.50 7.1 12.8 -0.25

0 (B) Z''

0.25

0.50 0.25 0.50 0.75 1.00 1.25 Z'

Figure 5.5.8: Impedance spectra measured at the end of discharge (A) and charge (B) for electrodes containing 15% PVDF with different capacities. The electrodes were formed at C/10 rate with 50% overcharge.

138

5.5.3. Effect of Fine Ni

The utilization of the nickel electrodes without nickel is about 80 to 86%. The main objective is to enhance the utilization of the electrodes by adding filamentary nickel

(INCO 210). In this section, the effect of nickel on the electrode performance will be discussed.

5.5.3.1. Electronic Conductivity Studies

The nickel hydroxide powder from Kansai Catalyst, as described in section 5.3.2, was mixed homogenously with nickel 210 using PVDF binder and NMP solvent. The conductivity of the paste was studied using electrochemical impedance spectroscopy. The paste was printed on a 4 point conductivity base as shown in Fig. 5.5.9. The thickness of the printed paste was estimated using a laser profilometer. The electronic resistance of the paste is given by the high frequency resistance measured. The electronic conductivity of the paste as a function of the nickel weight percent is shown in Fig. 5.5.10 for two hand mixed samples and one homogenized sample. The conductivity of the homogenized sample is always higher than the conductivity of the hand mixed samples. The conductivity of the paste with 0% Ni is about 10-4 Ω-1 cm-1 and increases to 0.94 Ω-1 cm-1 for 8.4% Ni. Therefore the critical weight percent for conductivity is about 8.4. More data points are needed to clearly estimate the critical weight fraction. It is evident that the electronic conductivity of the paste increases with nickel weight up to 12.2%. For nickel weight percentages greater than 12.2% no significant increase in conductivity is observed.

Nickel hydroxide powder with no cobalt and zinc, from OMG, was mixed homogeneously with nickel 210 powder. The electrical conductivity of the paste was

139

monitored using impedance spectroscopy and is compared with the nickel hydroxide powder from Kansai Catalyst. The electrical conductivity of the nickel hydroxide powder from OMG with the addition of nickel is shown in Fig. 5.5.11. It is evident that the electrical conductivity of the paste is poor up to 20 wt. % nickel. The conductivity of he paste increases to about 0.1 to 1 Ω-1 cm-1 at 30 wt. % nickel. This shows that the critical

weight fraction is in the range 20 to 30% Ni. Comparing Figs. 5.5.10 and 5.5.11, it is

observed that the electrical conductivity of the Kansai powder is always higher than that

of the OMG powder. This indicates that the cobalt oxide coated on the nickel hydroxide

particles (Kansai) does play a role in enhancing the conductivity of the paste, even before

the electrode is formed.

Figure 5.5.9: Four-point conductivity ceramic base used for AC impedance spectroscopy.

140

100 Handmixed 1 -1 -1 10 Handmixed 2 Homogenizer 1 cm cm -1 -1 1

0.1

0.01

1E-3

1E-4 Conductivity, ohm Conductivity, ohm

1E-5 0 5 10 15 20 25 Nickel, Wt %

Figure 5.5.10: Electronic conductivity of the paste containing nickel hydroxide particles (Kansai) as a function of nickel Wt. %.

1 -1 Sample 1 Sample 2 cm

-1 0.1

0.01

1E-3

1E-4 Conductivity, ohm Conductivity,

1E-5 0 5 10 15 20 25 30 35 Nickel, Wt %

Figure 5.5.11: Electronic conductivity of the paste containing nickel hydroxide particles (OMG) as a function of nickel Wt. %.

141

5.5.3.2. Comparison of Electrodes Containing 0% and 8.8% Ni

In this section, a set of nickel electrodes containing 10% PVDF and 0% nickel

will be compared to another set of nickel electrodes containing 8.7% PVDF and 8.8%

nickel. The first formation cycle of two such electrodes is compared in Fig. 5.5.12. The

charge voltage for the electrode containing nickel is lower in comparison to the other

electrode. The voltage rise and fall in the 0-0.5h charge period in the electrode without

nickel (Fig. 5.4.4) is absent in the electrode containing nickel. It is also evident that the

discharge time for the electrode containing nickel is longer indicating better utilization of the active material. The impedance spectra obtained at the end of discharge and charge for the formed electrodes is shown in Figs. 5.5.13A and 5.5.13B respectively. The impedance at the end of discharge (Fig. 5.5.13A) of the electrode containing nickel is significantly lower from the increase in conductivity of the electrode by addition of nickel. This is supported by the fact that the electrodes containing nickel charges at lower voltage. The impedance at the end of charge for the electrode containing 8.8% Ni shows only a small decrease in the charge transfer resistance in comparison to the electrode without Ni. This is because the nickel oxyhydroxide formed at the end of charging is conductive and the addition of nickel makes little difference to conductivity of the electrode paste.

The utilization of electrodes with and without nickel is shown in Fig. 5.5.14. It is evident that the utilization for the electrodes containing nickel is about 10% higher than the utilization obtained in electrodes without nickel. A model based on three level

conductive networks in the pasted electrode was proposed to explain the effect of addition of nickel by Yang et al27. The same model can be used to explain our results.

142

The nickel foam substrate forms the primary network or first level network via its nickel strands with a cell size of about 400 to 700 µm as shown in Fig. 5.3.1. The filamentary nickel powder (INCO nickel 210) provides the secondary network by spreading among the nickel hydroxide particles of size 10 µm as shown in Fig. 5.5.15. The cobalt oxyhydroxide formed on the surface of nickel hydroxide forms the tertiary network and enhances the conductivity among the nickel hydroxide particles. Therefore, when 8.8

Wt% Ni is added, the utilization of the electrode is increased as shown in Fig. 5.5.14. It is noted that the effect of cobalt enhancing the conductivity of the nickel hydroxide powder along with nickel was not taken into account in the work done by Yang et al.27

1.5 0% Ni

8.8% Ni 1.4

1.3 E (Volts)

1.2 0% Ni 8.8% Ni

1.1 0 5 10 15 Time (Hours)

Figure 5.5.12: Comparison of the voltage profile in the first formation cycle for electrodes with and without nickel.

143

-20 0% Ni 8.8% Ni

-10

Z'' (A)

10 0102030 Z'

-0.5 0% Ni 8.8% Ni

Z'' (B)

0.5

1.0 00.51.01.5 Z'

Figure 5.5.13: Impedance spectra measured at the end of discharge (A) and charge (B) for electrodes containing 0% and 8.8% Ni. The electrodes were formed at C/10 rate with 50% overcharge.

144

100 90 80 70 60 50 40

Utilization, % 30 20 10 0 0% Ni - 1 0% Ni - 2 0% Ni - 3 8.8% Ni - 1 8.8% Ni - 2

Figure 5.5.14: Utilization of electrodes with and without nickel. The electrodes were cycled at C/10 rate with 50% overcharge. The electrodes containing nickel show higher utilization.

Figure 5.5.15: Top view of SEM image of the nickel hydroxide electrode. The spherical structures are the nickel hydroxide particles and the filamentary structures are the nickel 210.

145

The EVS scans of electrode with and without nickel are compared in Fig. 5.5.16.

The electrodes without nickel and with nickel were fabricated by screen printing and spatula pressing, respectively. The β-NiOOH and β-Ni(OH)2 peaks are present for both

the electrodes and they shift towards lower potentials for the electrode containing nickel.

The γ-Ni(OH)2 phase is present for electrode without nickel and is absent for electrode

with nickel. This indicates that the addition of nickel enhances the conductivity of the

secondary network, thereby preventing overcharging of the electrode in certain regions.

The other benefit of adding nickel is the separation of the β-NiOOH peak from the

oxygen evolution as shown in Fig. 5.5.16. This clearly suggests that addition of nickel

enhances the conductivity of the electrode and decreases the side reaction – oxygen

evolution.

0.2 0% Ni, 10% PVDF 8.8% Ni, 8.7% PVDF

0.1 ) -1

-0.1 Current/Capacity (h

-0.2

-0.3 1.2 1.3 1.4 1.5 E (Volts)

Figure 5.5.16: EVS scans for electrodes containing 0% and 8.8% nickel. The electrodes were formed at C/10 rate with 50% overcharge.

146

5.5.3.3.Comparison of Electrodes Containing 5.2% and 8.7% PVDF in the

Presence of Fine Ni

In section 5.5.1, it was shown that the utilization increases with increase in PVDF

content. The effect of PVDF in the presence of fine nickel is discussed here. One set of

electrodes containing 5.2% PVDF and the other set of electrodes containing 8.7% PVDF

were formed at C/10 rate with 50% overcharge. The nickel content in these electrodes is

about 8.5-8.7% Nickel. The utilization of these electrodes is shown in Fig. 5.5.17. The

utilization for electrodes containing 5.2% PVDF is less than the electrodes containing

8.7% PVDF. This result agrees with the previous conclusion in section 5.5.1 that the

utilization increases with increase in the PVDF.

Cycle 1 Cycle 2 100

60 Utilization, %

40 5.2% PVDF 5.2% PVDF 5.2% PVDF 5.2% PVDF 8.7% PVDF 8.7% PVDF 1 2 3 4 1 2

Figure 5.5.17: Utilization of electrodes containing different amounts of PVDF. The nickel in these electrodes is about 8.5 to 8.7 Wt. %. The electrodes were formed at C/10 rate with 50% overcharge.

147

5.5.4. Effect of Pressing the Electrode

Electrodes containing 8.5% Ni and 5.2% PVDF were reduced in thickness using a

hot press before the electrodes were formed. Two electrodes were reduced to 50 and 75%

of their original thickness. The electrodes were then characterized using impedance

spectroscopy and electrochemical voltage spectroscopy and no significant changes were observed. The utilization of these electrodes also did not change significantly. By reducing the thickness of the electrodes, volumetric energy density of the battery can be

increased.

5.6. Other Methods of Fabrication

When thick inks are used the solvent drying during screen printing results in poor

penetration of the paste into the foam and formation of a thin layer of paste on top of the

foam. This causes particle shedding from the surface of the nickel foam during cycling.

When thin inks are used, they flow through the nickel foam because of the squeegee

pressure resulting in low electrode capacities. These problems are associated with the

screen printed nickel hydroxide electrodes. By impregnating the paste using a spatula

(section 5.6.1) or hot press (section 5.6.2) these problems can be alleviated.

5.6.1. Mechanical Impregnation Using Spatula

A spatula was used to mechanically press the paste into the nickel foam. This

method is simpler than screen printing and is advantageous when thin inks that flow

easily are used. The composition of the paste in dry active mass was 8.8% PVDF, 8.7%

nickel and the rest nickel hydroxide. The capacities of the electrodes fabricated by screen

printing and by pressing using the spatula are shown in Fig.5.6.1. The theoretical capacity

148

of the electrodes fabricated by spatula was in the range 16-20 mAh/cm2 and was twice

that fabricated by screen printing (8 mAh/cm2).

The utilization of the electrodes fabricated using screen printing and spatula is compared in Fig.5.6.2. The average value of the utilization of the electrodes made by spatula pressing is larger than that of electrodes made by screen printing. This is in spite of the fact that the theoretical capacity of the electrodes made by pressing using spatula being large (Fig. 5.6.1). It is noted that the nickel hydroxide electrodes tested in the low pressure nickel-hydrogen cell (Chapters 6 and 7) were fabricated by mechanical impregnation using spatula.

20 2

4 Theoretical Capacity, mAh cm 0 Screen Screen Pressed by Pressed by Pressed by Pressed by Pressed by printing, printing, spatula, spatula, spatula, spatula, spatula, Sample 1 Sample 2 Sample 1 Sample 2 Sample 3 Sample 4 Sample 5

Figure 5.6.1: Theoretical capacity of electrodes containing 8.8 % Nickel and 8.7% PVDF (dry active mass) fabricated by screen printing and pressing using flat spatula. The electrodes were formed at C/10 rate with 50% overcharge.

149

100

88 Utilization, % Utilization,

80 Screen Screen Pressed by Pressed by Pressed by Pressed by Pressed by printing, printing, spatula, spatula, spatula, spatula, spatula, Sample 1 Sample 2 Sample 1 Sample 2 Sample 3 Sample 4 Sample 5

Figure 5.6.2: Utilization of electrodes containing 8.8 % Nickel and 8.7% PVDF (dry active mass) fabricated by screen printing and pressing using flat spatula. The electrodes were formed at C/10 rate with 50% overcharge.

5.6.2. Mechanical Impregnation Using Hot Press

The existing fabrication methodology for pasted electrodes involves two steps, (a)

mechanical impregnation of the paste and (b) reducing the electrode thickness by

compression. By using a hot press both the steps can be accomplished in a single stroke.

By adjusting the load of the press and the paste composition, the desired electrode

thickness and electrode capacity can be obtained. The hot press was operated at room

temperature. An electrode with a loading as high 30.22 mAh/cm2 with a utilization of

92% was obtained. The existing hot press in the lab operates at heavy loads and is difficult to press the paste without crushing the nickel foam. The main disadvantage of this technique is the formation of a thick electrode layer on top or bottom of the nickel

foam due to excess paste and load. These problems can be solved by trial and error

method.

150

5.7. Conclusions

A pasted nickel hydroxide electrode was fabricated as an alternative to the D cell nickel hydroxide electrode. Nickel foam from INCO was used as the substrate. The paste was the pressed into the foam by two different techniques: (a) screen printing and (b) using spatula. Electrodes with higher capacity were obtained by mechanical impregnation using the spatula.

The parameters for the electrode formation were varied and their effect on electrode performance was studied. No significant differences were observed in utilization, impedance spectra and electrochemical voltage spectroscopy data when the electrode was formed at C/10 or C/20 rate. The electrode is completely formed in two cycles when overcharged by 50% and in 5 cycles without any overcharge. However, no differences in utilization, impedance or EVS data are evident with or without overcharge during formation.

Electrochemical voltage spectroscopy was used to study the effect of incomplete formation, the PVDF content and the nickel content on the electrode performance. In the case of incompletely formed electrodes, the oxidation of cobalt oxide to cobalt oxyhydroxide is incomplete resulting in a poorly conductive network. This results in overcharging in some regions of the electrode and is evident by the γ-NiOOH phase formation.

The utilization of the electrodes increases with the PVDF content in the electrodes, however higher PVDF content also results in the formation of the γ-NiOOH phase. The electronic conductivity of the Kansai and OMG nickel hydroxide powder was enhanced by the addition of INCO nickel 210. The electrode paste made from Kansai

151

(with Co and Zn) and OMG nickel hydroxide powder were conductive at 8.4 and 30 wt.

% nickel respectively. The utilization of the electrodes containing fine nickel was

increased by 10% to about 95%. The peak potentials of the β-NiOOH and β-Ni(OH)2

phase decreased with the addition of nickel. The peak of the γ-NiOOH is completely

absent indicating that the overcharging of the electrode is prevented by addition of nickel.

The other beneficial effect is the increase in the separation between the β-NiOOH peak

and the oxygen evolution rise. These clearly show the advantages of the adding of fine

nickel to the electrode paste.

References

1. M. Oshitani, H. Yufu, K. Takashima, S. Tsuji, and Y. Matsumaru, Journal of the Electrochemical Society, 136, 1590 (1989).

2. M. Oshitani, M. Watada, T. Tanaka, and T. Iida, Proceedings - Electrochemical Society, 94-27, 303 (1994).

3. A. Fleischer, Journal of the Electrochemical Society, 94, 289 (1948).

4. L. Kandler, Application: DE 1207358, (1965).

5. R. L. Beauchamp, (Bell Telephone Laboratories, Inc.). Application: US 3653967, (1972).

6. D. F. Pickett, (United States Dept. of the Air Force). Application: US 3827911, (1974).

7. D. F. Pickett and J. T. Maloy, Journal of the Electrochemical Society, 125, 1026 (1978).

8. B. S. Baker and M. G. Klein, (Energy Research Corp., USA). Application: US 3898099, (1975).

9. M. Butel, L. Gautier, and C. Delmas, Solid State Ionics, 122, 271 (1999).

10. V. Pralong, A. Delahaye-Vidal, B. Beaudoin, J. B. Leriche, and J. M. Tarascon, Journal of the Electrochemical Society, 147, 1306 (2000).

11. W. Taucher-Mautner and K. Kordesch, Journal of Power Sources, 132, 275 (2004).

152

12. M. E. Unates, M. E. Folquer, J. R. Vilche, and A. J. Arvia, Journal of the Electrochemical Society, 139, 2697 (1992).

13. R. D. Armstrong, G. W. D. Briggs, and E. A. Charles, Journal of Applied Electrochemistry, 18, 215 (1988).

14. G. Bronoel and J. Reby, Electrochimica Acta, 25, 973 (1980).

15. X. Wang, J. Yan, H. Yuan, Z. Zhou, D. Song, Y. Zhang, and L. Zhu, Journal of Power Sources, 72, 221 (1998).

16. M. Oshitani, Y. Sasaki, and K. Takashima, Journal of Power Sources, 12, 219 (1984).

17. K. Watanabe, M. Koseki, and N. Kumagai, Journal of Power Sources, 58, 23 (1996).

18. B. B. Ezhov and O. G. Malandin, Journal of the Electrochemical Society, 138, 885 (1991).

19. D. H. Fritts, Journal of the Electrochemical Society, 129, 118 (1982).

20. H. H. Law and J. Sapjeta, Journal of the Electrochemical Society, 136, 1603 (1989).

21. V. Pralong, Y. Chabre, A. Delahaye-Vidal, and J. M. Tarascon, Solid State Ionics, 147, 73 (2002).

22. R. S. Jayashree and P. V. Kamath, Journal of the Electrochemical Society, 149, A761 (2002).

23. F. Lichtenberg and K. Kleinsorgen, Journal of Power Sources, 62, 207 (1996).

24. H. S. Lim and R. E. Doty, Journal of Power Sources, 50, 57 (1994).

25. J. McBreen, W. E. O'Grady, G. Tourillon, E. Dartyge, A. Fontaine, and K. I. Pandya, Journal of Physical Chemistry, 93, 6308 (1989).

26. C. Tessier, C. Faure, L. Guerlou-Demourgues, C. Denage, G. Nabias, and C. Delmas, Journal of the Electrochemical Society, 149, A1136 (2002).

27. Q. M. Yang, V. A. Ettel, J. Babjak, D. K. Charles, and M. A. Mosoiu, Journal of the Electrochemical Society, 150, A543 (2003).

28. M. S. Wu, C. M. Huang, Y. Y. Wang, and C. C. Wan, Electrochimica Acta, 44, 4007 (1999).

29. M. J. Madou, Fundamentals of Microfabrication: The science of Miniaturization, CRC Press LLC, Boca Raton, (1994).

153

30. G. A. Ghiurcan, C.-C. Liu, A. Webber, and F. H. Feddrix, Journal of the Electrochemical Society, 150, A922 (2003).

31. J.-S. Do, S.-H. Yu, and S.-F. Cheng, Journal of Power Sources, 117, 203 (2003).

32. Y. Zhao, Q. Zhou, L. Liu, J. Xu, M. Yan, and Z. Jiang, Electrochimica Acta, 51, 2639 (2006).

33. J. S. Wainright, R. F. Savinell, C. C. Liu, and M. Litt, Electrochimica Acta, 48, 2869 (2003).

34. A. Morata, A. Tarancon, G. Dezanneau, F. Peiro, and J. R. Morante, Materials Research Society Symposium Proceedings, 822, 109 (2004).

35. V. V. Srdic, R. P. Omorjan, and J. Seydel, Materials Science & Engineering, B: Solid-State Materials for Advanced Technology, B116, 119 (2005).

36. M. Hrovat, J. Holc, and D. Kolar, Solid State Ionics, 68, 99 (1994).

37. M. Hrovat, N. Katsarakis, K. Reichmann, S. Bernik, D. Kuscer, and J. Holc, Solid State Ionics, 83, 99 (1996).

38. R. N. Basu, S. K. Pratihar, M. Saha, and H. S. Maiti, Materials Letters, 32, 217 (1997).

39. N. T. Hart, N. P. Brandon, M. J. Day, and J. E. Shemilt, Journal of Materials Science, 36, 1077 (2001).

40. L. J. Hobson, Y. Nakano, H. Ozu, and S. Hayase, Journal of Power Sources, 104, 79 (2002).

41. Y. Matsuzaki and I. Yasuda, Solid State Ionics, 152-153, 463 (2002).

42. A. C. Muller, D. Herbstritt, and E. Ivers-Tiffee, Solid State Ionics, 152-153, 537 (2002).

43. R. Peng, C. Xia, X. Liu, D. Peng, and G. Meng, Solid State Ionics, 152-153, 561 (2002).

44. H. Bode, K. Dehmelt, and J. Witte, Electrochimica Acta, 11, 1079 (1966).

45. L. H. Thaller, A. H. Zimmerman, and G. A. To, Journal of Power Sources, 122, 85 (2003).

46. L. H. Thaller, A. H. Zimmerman, and G. A. To, Journal of Power Sources, 114, 309 (2003).

47. A. H. Zimmerman, Proceedings - Electrochemical Society, 94-27, 268 (1994).

154

6. Assembly and Testing of Low Pressure Nickel Hydrogen Battery

The main objective in this chapter was to assemble and test a low pressure nickel

hydrogen cell. The cell components studied and developed independently were put

together. The factors that affect the cell performance, (a) self discharge, (b) oxygen

evolution and (c) recombination will be discussed in this chapter.

6.1. Description of Experimental Setup

A schematic picture of the nickel hydrogen battery is shown in Fig. 6.1.1. The

formed nickel hydroxide electrode and the platinum electrode (from E-tek) separated by a

polypropylene membrane (Celgard 3400) filled with 26 wt. % KOH were assembled in a

bolt-nut cell. The nickel mesh serves as both current collector and gas diffusion layer.

Two chlorinated polyvinyl chloride discs, 3.2 cm in diameter, at the ends of the cell are

for insulation. A nylon bolt, 1/8″ diameter, was used to assemble the cell (Fig. 6.1.1). The

bolt-nut cell is hosted in a stainless steel cylinder filled with hydrogen gas. The inlet and outlet valves are used for purging and filling the cell to the required hydrogen pressure.

The Ni-H2 cell hosted in the SS cylinder (Fig. 6.1.1) is connected to a

potentiostat/galvanostat as shown in Fig. 6.1.2. The pressure inside the cell is monitored

using a pressure gauge, which is a strain meter from Omega Corporation and is connected

to the outlet of the Ni-H2 cell. The accuracy of the pressure gauge according to manufacturer’s specifications is 0.025 psi. A National Instrument data acquisition card

155

was used to convert the analog pressure signal to digital and the data is then logged into a computer. As the cell is charged or discharged the pressure is recorded simultaneously.

Positive Negative electrode electrode Insulator Ni(OH)2 Platinum

3.2 cm Outlet valve Bolt Nut

Separator Current collector

Inlet Stainless steel compartment valve 1cm

Figure 6.1.1: Schematic picture of the Ni-H2 battery

Outlet Inlet Valve Valve

Pressure Ni H2 Cell Gauge

Potentiostat/ Galvanostat

Figure 6.1.2: Schematic picture of the experimental set up.

156

6.2. Performance of the Ni-H2 Cell in Comparison to the Liquid Cell

A nickel hydroxide electrode of theoretical capacity 17.7 mAh/cm2 was formed in the liquid cell (described in detail in chapter 5). The electrode contained 8.7% PVDF,

8.8% nickel and the rest Ni(OH)2 in dry weight basis. The utilization of these electrodes at 50% overcharge during formation was discussed in chapter 5. The coulometric efficiency of this electrode when charged completely at C/5 rate without overcharge was

89.3%. The coulometric efficiency of the electrodes increases to 93.2-96.3%, when

charged to a capacity of 14.3 mAh/cm2 due to the decrease in the oxygen evolution at the

end of charging. The experimental capacity of the electrode was therefore taken as 14.3

2 mAh/cm . The performance of this electrode in the Ni-H2 cell will be discussed in this

chapter unless otherwise specified.

The voltage profiles during charge and discharge of the nickel hydroxide

electrode at C/5 rate in the liquid cell and in the nickel-hydrogen cell configuration are

compared in Fig. 6.2.1. It is noted that the nickel hydroxide electrode in the nickel-

hydrogen cell was charged to only 90% of the experimental capacity (14.3 mAh/cm2). It

is evident that the charge voltage is larger and the discharge voltage smaller for the

nickel-hydrogen cell in comparison to the liquid cell measurement. This indicates larger

overpotentials for the nickel-hydrogen cell than observed in the liquid cell. It is noted that

the voltage measured in the liquid cell includes only the overpotentials for the solution iR

loss and the nickel hydroxide electrode. However for the Ni-H2 cell, in addition to the above overpotentials, the voltage measured also includes the overpotential for the

platinum negative electrode.

157

The coulometric efficiency of the nickel-hydrogen cell was about 90.9-91.2 %

(charged to 90% experimental capacity) and is lower than the coulometric efficiency of

the nickel hydroxide electrode in the liquid cell, 93.2-96.3% (charged to 100%

experimental capacity). Fig. 6.2.1 shows a change in slope at the end of charging both in

the liquid cell and nickel-hydrogen cell tests suggesting that the oxygen evolution is the

reason for the decrease in coulometric efficiency. This will be discussed in more detail in

sections 6.3 and 6.4.

1.5 Ni-H2 cell

1.4

Ni(OH)2 electrode

1.3 Ni-H2 cell E (Volts)

1.2

1.1 012345 Time (Hours)

Figure 6.2.1: Voltage profile of the nickel hydroxide electrode during charge and discharge in the liquid cell and in the nickel-hydrogen cell configuration.

Impedance spectra measured in the frequency range 20000 to 0.2 Hz, at the end of discharge and charge in the liquid cell and nickel-hydrogen cell configuration are

compared in Figs. 6.2.2A and 6.2.2B. It is evident from the figures, that the impedance at

the end of charge and discharge is larger for the nickel hydrogen cell in comparison to the

158

liquid cell. This explains the higher charge voltage and the lower discharge voltage

observed in Fig. 6.2.1. The major difference between the two measurements is that the

impedance measured in the case of the nickel-hydrogen cell configuration includes the

impedance of both the nickel hydroxide and the platinum electrode (two electrode cell).

However, in the case of the liquid cell measurements a reference electrode is used and the

impedance measured corresponds to only that of the nickel hydroxide electrode.

The impedance spectra measured at the end of charge and discharge in the nickel- hydrogen cell configuration show two semi-circles, whereas only one semi-circle is

evident in the liquid cell configuration. The two semi-circles present in the impedance

spectra are related to the charge transfer resistances associated with the nickel hydroxide

electrode and the ESNS platinum electrode. This was confirmed by comparing time

constants associated with electrodes in liquid cell and nickel-hydrogen cell

configurations.

The electrochemical parameters and the time constant associated with the charge

transfer reaction (hydrogen evolution, hydrogen oxidation) and double layer charging of

the ESNS platinum electrode are listed in Table 6.1. The time constant is estimated as the

product of charge transfer resistance and double layer capacitance, which are measured

by a semi-circle fit to the impedance data. The time constant for the platinum electrode

measured in the liquid cell and in the nickel hydrogen cell configuration (two negative

electrodes in bolt nut cell as described in detail in Chapter 3) are in the same range, 160-

350 ms, and the corresponding frequency is about 0.4-1 Hz.

The time constant for the nickel hydroxide electrode relevant to the charge

transfer reaction and double layer capacitance in the liquid cell configuration is listed in

159

Table 6.2. The resistance and the capacitance measured are different for the electrode in

the charged and in the discharged states, however, the time constant is about 10 ms, an

order of magnitude lower in comparison to the time constant measured for the platinum

electrode (Table 6.1).

The time constants of the two semicircles evident in the impedance spectra

measured in the nickel-hydrogen cell configuration (Fig. 6.2.2) were calculated and are

listed in Table 6.3. It is extremely difficult to calculate the charge transfer resistance and

the double layer capacitance of the 2nd semi-circle accurately using the data available.

The frequency corresponding to the last data point of the 2nd semi-circle is 0.2 Hz and the

relevant frequency of the semi-circle is less than 0.2 Hz. The frequency of first semi-

circle is 14-28 Hz, about the same as that measured for the nickel hydroxide electrode in

the liquid cell configuration (Table 6.2). The frequency of the 2nd semi-circle is about 0.2

Hz and is close to the frequency estimated for the platinum electrode (Table 6.1). This strongly suggests that the negative electrode is the main contributor to the large impedance (2nd semi-circle) observed in the nickel-hydrogen cell configuration.

It is noted that in other nickel-hydrogen cells, a lower charge transfer resistance for the platinum electrode, about 2.6 Ω cm2, has been obtained. The solution resistance is

about 0.6 Ω cm2. The corresponding overpotentials including the solution and the charge

transfer resistances are small (about 10 mV). Unused platinum electrodes have low

charge transfer resistance resulting in low charge voltage and high discharge voltage.

However, the charge transfer resistance of the platinum electrode is always larger than that of the nickel hydroxide electrode.

160

-10 Liquid cell Ni-H2 cell

-5 (A) Z''

5 0 5 10 15 Z'

-5 Liquid cell at 100% DOC Ni-H2 cell at 90% DOC -4

-3 (B) Z'' -2

-1

1 0123456 Z'

Figure 6.2.2: Impedance spectra measured at the end of discharge (A) and charge (B) for electrodes cycled at C/5 rate in liquid cell and nickel-hydrogen cell configuration.

161

Table 6.1: Parameters associated with the ESNS platinum electrode.

Mode of cell test Rct, Cdl, Time constant, Frequency, Ω cm2 F/cm2 mS Hz Liquid cell 0.88 0.4 350 0.45 Ni-H2 cell, Best condition 0.74 0.222 160 0.97 Ni-H2 cell, Used condition 3.5 0.058 200 0.78

Table 6.2: Parameters associated with the nickel hydroxide electrode.

Mode of cell test Rct, Cdl, Time constant, Frequency, Ω cm2 F/cm2 mS Hz Liquid cell, charged 0.28 0.034 10 16.71 Liquid cell, discharged 1.97 0.005 10 16.15

Table 6.3: Parameters associated with the complete Ni-H2 cell.

Mode of cell test Rct, Cdl, Time constant, Frequency, Ω cm2 F/cm2 mS Hz Ni-H2 cell, charged, First Loop 1.88 0.003 6 28.21 Ni-H2 cell, charged, Second Loop - - - < 0.20 Ni-H2 cell, discharged, First Loop 6.88 0.0016 11 14.45 Ni-H2 cell, discharged, Second Loop - - - < 0.20

6.3. Coulometric Efficiency of the Ni-H2 Cell

The coulometric efficiency is an important parameter for all battery technologies,

especially for the nickel hydrogen cells that evolve oxygen as a side reaction at the end of

charging.

6.3.1. Effect of Depth of Charge on Coulometric Efficiency

The coulometric efficiency was investigated at different depths of charge. Depth

of charge is analogous to the depth of discharge typically used in lithium ion batteries.

Depth of charge is defined as the charge applied to the cell from a state of complete

discharge. This parameter is more relevant to the nickel hydrogen cells, whose efficiency and cycle life, as will be shown in this chapter, is highly dependent on the depth of charge and will be used very frequently in this chapter.

162

The voltage profile during charge and discharge of the nickel-hydrogen cell cycled at C/5 rate to different depth of charge is shown in Fig. 6.3.1. The charge cut off is based on the depth for charge, e.g. for 50% DOC the cell is charged for 2.5 h (at C/5 rate). The end of discharge is based on the cut-off voltage of 1.15 V, which is determined by the knee at the end of the discharge profile. It is evident from Fig. 6.3.1 that the cell cycles well at different depth of charges. For 90% DOC, a change in slope at the end of charging due to oxygen evolution is observed. The coulometric efficiency decreases with increase in the different depth of charge as listed in Table 6.4. This supports the discussions in chapter 5 that the side reaction, oxygen evolution, increases at higher charge potentials (which translates to longer charging times, here) leading to a decrease in efficiency.

163

1.5

Charge 1.4

1.3 End of (A) discharge E (Volts) Discharge cutoff 1.2

1.1 0123 Time (Hours)

1.5

1.4

1.3 (B) E (Volts) 1.2

1.1 01234 Time (Hours)

1.5

1.4 O2 Evolution 1.3 (C) E (Volts) 1.2

1.1 012345 Time (Hours)

Figure 6.3.1: Charge-discharge voltage profiles of the nickel-hydrogen cell cycled at C/5 rate to different depth of charge: (A) 50% DOC, (B) 70% DOC and (C) 90% DOC.

164

Table 6.4: Coulometric efficiency of the Ni-H2 cell as a function of depth of charge cycled at C/5 rate. Depth of charge, % Efficiency 50 95.5 70 94.2 90 91.2

6.3.2. Effect of Rest Period on Coulometric Efficiency – Self-discharge

It is well known that the nickel hydrogen cells suffer from self discharge due to

reduction of nickel oxyhydroxide to nickel hydroxide in the presence of hydrogen gas and KOH electrolyte. The self discharge rate is directly proportional to the hydrogen gas pressure inside the cell. Monitoring and understanding the self discharge is therefore critical to the development of a practical battery.

The self discharge in our low pressure nickel hydrogen battery should be lower than in the high pressure nickel hydrogen batteries used in space applications. The self discharge phenomenon was studied by introducing a rest period between charge and discharge. The rest period was varied and the coulometric efficiency was estimated. The voltage profiles during charge and discharge of the nickel-hydrogen cell with different rest periods, 5 min, 1h and 6h are shown in Fig. 6.3.2. It is evident that the discharge time decreases with increase in the rest period. This indicates that the low pressure nickel hydrogen cell does undergo self discharge.

165

1.5

1.4 Discharge time 1.3 decreases E (Volts) 5 min 1.2 1 h 6 h

1.1 012345 Time (Hours)

Figure 6.3.2: Charge-discharge voltage profiles of the nickel-hydrogen cell cycled at C/5 rate to 90% DOC with different rest periods, 5 min, 1h and 6h.

Coulometric efficiency of the cell was determined for different rest periods at 50,

70 and 90% depth of charge and is shown in Figures 6.3.3, 6.3.4 and 6.3.5 respectively.

The coulometric efficiency decreases with rest period for all depths of charge and indicates self discharge of the battery. A straight line was used to fit the coulometric efficiency as a function of the rest period. The slope of the linear fit (Figs. 6.3.3, 6.3.4, and 6.3.5) shows that the rate of self discharge is about 1% per hour. This indicates that the battery will discharge completely in about 100 hours. No definite difference in the self-discharge rate is observed at different depths of charge.

166

96 Equation: y = A + B*x R^2 = 0.99 95 A 95.28 ±0.37 94 B -1.02 ±0.11 93

92 91 90 Columetric efficiency (%) 89 0123456 Rest period after charge, h

Figure 6.3.3: Coulometric efficiency of the Ni-H2 cell as a function of the rest period at 50% DOC, cycled at C/5 rate.

94 Equation: y = A + B*x R^2 = 0.996 A 93.67 ±0.29 B -1.29 ±0.08 92

86 Columetric efficiency (%)

0123456 Rest period after charge, h

Figure 6.3.4: Coulometric efficiency of the Ni-H2 cell as a function of the rest period at 70% DOC, cycled at C/5 rate.

167

92 Equation: y = A + B*x R^2 = 0.998 A 90.86 ±0.15 90 B -1.01 ±0.04

86 Columetric efficiency (%) 84 0123456 Rest period after charge, h

Figure 6.3.5: Coulometric efficiency of the Ni-H2 cell as a function of the rest period at 70% DOC, cycled at C/5 rate.

The coulometric efficiency of the cell for different depth of discharges and different rest periods was compiled and is shown in the Fig. 6.3.6. The efficiency decreases with increasing depth of charge because of the oxygen evolution reaction at the end of charge. This conclusion is supported by the electrochemical voltage spectroscopy

(EVS) study of the nickel hydroxide electrodes. The efficiency at 90% depth of charge is dramatically lower than that at 50 and 70% depth of discharge for 5 min and 1h rest periods when the efficiency is primarily determined by the amount of oxygen evolution during charge. This result will be substantiated by analyzing the pressure variations at the end of charge and during the rest period.

For a given depth of discharge, the efficiency decreases with increase in the rest period and is due to the self discharge of the battery. This will be verified below by analyzing the pressure variations during the rest period. The decrease in efficiency from 5

168

min rest period to 6h rest period is about the same for all depths of charge, indicating that the self discharge rate is not significantly different.

96 0.083h 1h 94 6h

Columetric efficiency 86

50 60 70 80 90 Depth of charge (%)

Figure 6.3.6: Coulometric efficiency of the Ni-H2 cell cycled at C/5 rate as a function of the depth of charge for different rest periods.

6.4. Pressure Data Analysis

The voltage variations during the charge, the rest, and the discharge period of the

cell cycled at C/5 rate to 90% depth of charge are shown in Fig. 6.4.1. The rest period in

this case is about 6h. It is evident from the figure that the voltage increases during charge,

followed by relaxation of the voltage during the rest period. Subsequently, during

discharge the voltage drops until the cut-off voltage of 1.15V is reached followed by

relaxation of the voltage during the rest period for 6h. The change in slope at the end of

charge is clearly seen in Fig. 6.4.1.

The cell pressure was measured simultaneously during the charge-discharge cycle

and is shown in Fig. 6.4.2. The pressure inside the cell increases during charge due to the

formation of hydrogen at the platinum electrode. The rate of pressure increase is linear

169

and corresponds to the current applied to the cell. Based on the pressure change and charge passed, the gaseous volume inside the cell was estimated as 40 cc. During the rest period, the pressure is not constant as one would expect; instead a decrease in the hydrogen pressure is seen. This decrease in pressure can be accounted for by two mechanisms: a) recombination of evolved oxygen with hydrogen at the platinum electrode causing a decrease in pressure and b) self discharge resulting in pressure decrease. In reality, the pressure decrease is a combination of the above two mechanisms and will discussed in detail, below. The pressure decreases linearly during discharge because of the hydrogen oxidation at the platinum electrode. During the rest period after discharge, no significant change in pressure is observed as shown in Fig. 6.4.2.

1.5

Discharge 1.4 Rest Charge Rest 1.3 E (Volts) 1.2

1.1 05101520 Time (Hours)

Figure 6.4.1: Voltage variation during charge, rest and discharge periods of a cycle. The cell was cycled at C/5 rate to 90% depth of charge. The rest period is 6h.

170

28 Rest 26 Charge Rest Absolute Pressure, psi Discharge 24 0.0 4.0 8.0 12.0 16.0 20.0 Time, h

Figure 6.4.2: Changes in pressure during the charge, the rest and the discharge periods of a cycle. The cell was cycled at C/5 rate to 90% depth of charge. The rest period is 6h.

6.4.1. Pressure Variations During Charge

The pressure inside the cell as a function of the charge time is shown in the Fig.

6.4.3. The pressure variation during the initial charging time (0 to 1h) was fitted to a

straight line. The slope of the linear fit (parameter B), 2.52 psi/h, in Fig. 6.4.3

corresponds to the rate of formation of hydrogen. It is evident that pressure variation towards the end of the charging phase is non-linear when compared to the straight line fit.

A schematic representation of possible variations in pressure during charge is shown in Fig. 6.4.4. At initial charge times, it is assumed there is no significant oxygen evolution and that all current goes to the main reaction which is oxidation of nickel

hydroxide to nickel oxyhydroxide. Therefore the pressure increases linearly

171

corresponding to the rate of hydrogen formation as shown by a straight line marked as

‘No O2’ in Fig. 6.4.4, which assumes that there is no oxygen evolution.

In reality, as evident in the voltage profile change at the end of charging (Fig.

6.4.1) part of the current to the nickel hydroxide electrode goes to oxygen evolution,

which is a side reaction. Therefore in addition to hydrogen evolution in the platinum electrode, there is also oxygen evolution in the nickel hydroxide electrode causing a steep rise in pressure as shown by the curve marked as ‘Yes O2 Poor recombination’ in Fig.

6.4.4. It is assumed here that the recombination of oxygen with hydrogen is poor and

oxygen builds up in the gas phase.

However, if the oxygen evolved at the nickel hydroxide electrode recombines

rapidly with the hydrogen evolved at the platinum electrode a decrease in the pressure

rate will be evident as shown by the curve marked as ‘Yes O2 Good recombination’ in

Fig. 6.4.4. It is evident by comparing Figs. 6.4.3 and 6.4.4 that the evolved oxygen does

recombine with hydrogen during charging. This is evident from the difference in the

predicted pressure based on the linear fit and the measured pressure at the end of 4.5 h,

which is about -0.84 psi. This indicates that the measured pressure at the end of the

charge period (4.5 h) is mostly hydrogen plus a small amount of the oxygen that has not

recombined.

172

36 0.84 y = A + B*x 34 A 25.42 32 B2.52

30 Non-linear

Pressure, psia 28

0.0 1.0 2.0 3.0 4.0 5.0 Time, h

Figure 6.4.3: Changes in pressure during the charge. The cell was cycled at C/5 rate to 90% depth of charge.

Yes O2 Poor recombination

No O2

Yes O2 Pressure Good recombination

Charge time

Figure 6.4.4: Schematic representation of different pressure variations due to different possible mechanisms during charging.

173

6.4.2. Pressure Variations During the Rest Period

The pressure in the cell decreases monotonously during the rest period is shown in Fig. 6.4.5. This pressure drop when no charge or discharge current is applied to the cell is due to (a) the recombination of oxygen with hydrogen gas and (b) the self discharge of the cell. The leftover oxygen gas that was not recombined during the charge period recombines at the Pt electrode. The self discharge occurs due to the oxidation of the hydrogen gas at the nickel electrode and reduction of nickel oxyhydroxide to nickel hydroxide. Both these processes results in the decrease of hydrogen pressure in the cell. It is noted that the noise in the pressure data is due to 60 Hz electrical noise introduced during the analog to digital conversion by the data acquisition card.

The self discharge rate is assumed constant at a given state of charge of the battery. Therefore the hydrogen pressure in the cell will decrease linearly with rest time as a result of self discharge. It is also evident from the Fig. 6.4.5 that the pressure change is linear in the rest period from 3 to 6h and therefore it is assumed that all oxygen recombination is complete in the first 3h of the rest period. A linear fit to the pressure data in the time period, 3 to 6h is shown in Fig. 6.4.5. The error calculated for the linear fit is very small. The slope of the line or the rate of the pressure change is 0.108 psi/h.

This self discharge rate is about 23 times slower than the charge rate (C/5). At this rate, the cell will discharge completely in about 117h and the self discharge rate can be expressed as C/117. In principle, the self discharge rate is not constant and decreases with decrease in state of charge of the cell and therefore the cell will last significantly longer than 117h. It is noted that the linear fit based on the coulometric efficiency data estimated the self discharge rate as C/100 (section 6.3.2). This agrees well with the result based on

174

the pressure data analysis. Based on the self discharge rate, the drop in pressure due to self discharge is about 0.648 psi. The effect of state of charge of the nickel-hydrogen cell on the self discharge rate will be discussed in detail in section 6.4.3.

36.0 Equation: y = A + B*x R^2 = 0.90743 35.8 A 35.45 ±0.002 B -0.108 ±3.4x10-4 35.6

35.4

35.2

Pressure, psia 35.0

34.8

34.6 0.0 1.4 2.8 4.2 5.6 6.9 Rest period, h

Figure 6.4.5: Pressure changes during the rest period. The cell was cycled at C/5 rate to 90% depth of charge. A straight line is fit to the pressure data in the 3 to 6h rest period and extrapolated back to time 0.

The pressure contribution due to the self discharge of the cell during the rest

period based on the linear fit was subtracted from the total pressure data and the

difference, the pressure contribution due to the recombination reaction, is plotted in Fig.

6.4.6. The pressure variation due to the recombination reaction follows an exponential

decay and the pressure changes are nearly complete in 2h. The recombination of oxygen

with hydrogen gas is considered to be a first order reaction in oxygen and zero order in

175

hydrogen due to hydrogen excess in the cell. The rate equation for the recombination

reaction is:

dP =−kP dt

where P is the pressure of oxygen and k is the rate constant.

The solution to this rate equation is:

=− P Pkt0 exp( )

where P0 is the initial pressure of oxygen. The above solution is in the form of an exponential decay. Therefore the pressure drop due to the recombination reaction was fitted to an exponential decay equation as shown in Fig. 6.4.6. An excellent fit with low error values is obtained. The parameter P0 represents the total pressure drop due to the

recombination reaction during the rest period and is 0.445 psi. The pressure drop due to

the self discharge (0.648 psi) and that due to the recombination are of the same order and

therefore neither one can be neglected. The time constant t1 represents the total time it

takes to for all oxygen to recombine completely and is 1.02 h. In comparison, the time

constant for the recombination reaction obtained for just the platinum electrode in a

separate experiment, is about 15 min. This difference in time indicates that the

recombination process is limited by the diffusion of oxygen from the nickel hydroxide electrode to the platinum electrode. The reciprocal of time constant t1 gives the rate

constant of the recombination reaction. The rate constant is 0.98 h-1.

176

0.5

Equation: P = P exp(-t/t ) 0 1 0.4 R^2 = 0.93 P 0.445 ±0.001 0 t 1.02 ±4.4x10-3 0.3 1

0.2

Pressure, psia 0.1

0.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0 Rest period, h

Figure 6.4.6: Pressure changes due to recombination reaction during the rest period. The cell was cycled at C/5 rate to 90% depth of charge. An exponential decay line is fit to the pressure data.

6.4.3. Self Discharge at Different Depths of Charge

The self discharge rate determined by the linear fit described in section 6.4.2 is plotted as a function of depth of charge for two datasets from the same cell in Fig. 6.4.7.

It is evident from the figures that the self discharge rate is lower for 50% depth of charge

in comparison to the 70% and 90% DOC. The self discharge rate at 50% DOC is 0.068

psi/h and corresponds to a rate of C/180. Dataset 1 shows that the self discharge rate

increases with depth of charge; however no definite trend is noted for dataset 2. For the

70% and 90% DOC, the self discharge rate varies between 0.11 and 0.14 psi/h. This

corresponds to rates of C/117 and C/90. The decrease in hydrogen pressure in the rest

period due to self discharge can be calculated by multiplying the self discharge rate with

the rest time.

177

0.16 Data set 1 Data set 2 0.12

0.08

0.04 Self discharge rate, psi/h rate, discharge Self 0.00 50 70 90 Depth of charge, %

Figure 6.4.7: Self discharge rate estimated as a function of depth of charge for Dataset 1 and 2.

6.4.4. Recombination in the Ni-H2 Cell During Charge and Rest Period

The difference in the predicted pressure based on the linear fit and the measured pressure at the end of the charge as described in section 6.4.1 was calculated for different depth of charge. This decrease in pressure is due to the recombination of hydrogen gas with oxygen gas and will be referred to as the pressure drop due to recombination

(∆P(H2/02)) during the charge period. It is noted here that only a fraction of oxygen formed is recombined during the charging period and the rest recombines during the rest period. The pressure drop due to the recombination reaction in the rest period was calculated by fitting an exponential decay equation as described in section 6.4.2.

The pressure drop due to recombination as a function of depth of charge during the charge and the rest period is shown in Fig. 6.4.8. The pressure drop due to the

178

recombination increases with increase in the depth of charge. Assuming that the recombination rate is the same at all depth of charges, the figure indicates that the oxygen evolution increases and therefore the recombination increases with the depth of charge.

This increase in oxygen evolution agrees well with the coulometric efficiency results,

(section 6.3.1) which show a decrease in efficiency with increase in depth of charge.

Comparing the pressure drops in the charge and the rest period, it is evident that about 70% of the recombination is complete during the charge period for the 70 and 90%

DOC and about 50% of the recombination is complete for the 50% DOC. This clearly shows good recombination on the platinum electrode.

The rate constant obtained for the recombination reaction is about the same, 0.94 h-1, for all depths of charge. This indicates that all the recombination is complete during the first 1h of the rest period.

1.0 Charge Rest

0.8 ), psi 2 0.6 / O 2 0.4 P( H ∆ 0.2

0.0 50 70 90 Depth of charge, %

Figure 6.4.8: Pressure drop due to the recombination reaction as a function of depth of charge during the charge and the rest period.

179

6.4.5. Pressure Contributions – Self discharge and Recombination

The pressure loss contribution due to self discharge and recombination at different

depths of charge for the nickel-hydrogen cell is listed in Table 6.5. The rest period between the charge and the discharge is 6h. It is evident that the pressure loss due to the total recombination (including charge and rest) and self discharge increases with the depth of charge. In terms of percentage loss, the self discharge contribution decreases with depth of charge. This indicates that the oxygen evolution is the major mechanism for pressure loss at deep depth of charges and the self discharge is the major mechanism for pressure loss at short depth of charges for the 6h rest periods.

For the 5 min and 1h rest periods, oxygen evolution is the predominant factor for all DOC as the time for self-discharge is small.

Table 6.5: Pressure loss contributions of recombination and self discharge for 6h rest period. Pressure drop 50% DOC 70% DOC 90% DOC Recombination, psi 0.22 0.62 1.48 Self discharge, psi 0.41 0.68 0.77 Recombination, % 34.6 47.9 65.7 Self discharge, % 65.4 52.1 34.3

6.4.6. Efficiency Loss – Self Discharge and Recombination

The contributions of different mechanisms to the coulometric efficiency of the

nickel-hydrogen cell for different depth of charge was calculated based on the pressure

data analysis and are shown in the Figs. 6.4.9 (A), (B) and (C). The rest period between

charge and discharge is 6h.

The coulometric efficiency for the 50% depth of charge is 89.9%. The efficiency

loss is due to self discharge and oxygen evolution. The efficiency loss due to self

180

discharge is higher than the loss due to the O2 evolution. As expected, the coulometric efficiency drops to 84.9% and 83% for 70% and 90% depth of charge. For the 70% depth of charge, the efficiency loss due to self discharge and oxygen evolution are about the same. At 90% depth of charge, the efficiency loss due oxygen evolution is two times that of the self discharge and is the main contributor to lower efficiency.

In comparing the Figs 6.4.9(A), (B) and (C), it is evident that the efficiency loss due to O2 evolution increases with increase in the depth of charge (as observed in the pressure data measurements) from 3.5% for the 50% depth of charge to 11.3% for the

90% depth of charge. This indicates that the oxygen evolution is the main contributor to efficiency loss at high depth of charge. The efficiency loss due to self discharge for 50,

70 and 90% depth of charge is about the same.

For the 5 min and 1h rest periods, the efficiency loss is mainly due to the oxygen evolution for all depths of charge.

181

Efficiency 89.9%

O evolution 2 (A) 3.5% Self discharge 6.6%

Efficiency 84.9%

O evolution 2 (B)

7.2% Self discharge 7.9%

Efficiency 83%

O evolution 2 (C) 11.3% Self discharge 5.7%

Figure 6.4.9: Contributions of different processes at different depth of charge: (A) 50% (B) 70% and (C) 90%. The rest period is 6h.

182

6.5. Comparison of Efficiency Based on Charge Data and Pressure Data

The coulometric efficiency calculated based on charge passed and charge obtained during discharge is compared to the coulometric efficiency obtained by pressure data analysis for two different data sets in Figs. 6.5.1 and 6.5.2. The estimated efficiency based on the pressure data matches the efficiency based on the charge data for the 50% depth of charge very well. However, small differences of about 2% are observed for the

70 and 90% depth of charge. The pressure data analysis under predicts the actual coulometric efficiency observed. For a practical cell, under prediction would be preferred as opposed to over prediction.

These two figures clearly show that the efficiencies obtained from pressure data analysis are correct and supports the analysis of extracting the contributions of different mechanisms during charge and discharge of the nickel hydrogen cell. With the clear understanding of the pressure variations during charge and discharge, it is possible to use the pressure data, in addition to other methods, to reliably estimate the state of charge of the nickel-hydrogen cells. In fact, the pressure data will be more reliable as it directly accounts for the self discharge and the oxygen evolution occurring in the cell.

6.6. Ni-H2 Cell Tests – Below Atmospheric Pressure

The performance of the nickel-hydrogen cell was also studied in the pressure range 0.4-1 atm. The cell was cycled to different depth of charge with different rest periods between charge and discharge. The pressure data and the charge-discharge data were analyzed. The results obtained regarding the different mechanisms – self-discharge, oxygen evolution and recombination were the same as discussed earlier for the nickel hydrogen cells tested at 2-3 atm.

183

Data set 1

Charge data 90 Pressure data

20 Efficiency, % 10

0 50 70 90 Depth of charge, %

Figure 6.5.1: Comparison of efficiency calculated based on the charge data and pressure data at different depth of charge for data set 1.

Data set 2 100 Charge data Pressure data 90

Efficiency, % 20

0 40 50 60 70 80 90 100 Depth of charge, %

Figure 6.5.2: Comparison of efficiency calculated based on the charge data and pressure data at different depth of charge for data set 2.

184

6.7. Cycle Life Tests

The cycle life of the nickel-hydrogen cells were investigated using the Arbin test station. Two cells of different experimental capacities, one 41 mAh and another 62.5 mAh were cycled at C/5 rate to 90% of their capacity. The discharge capacity as a function of cycles for the two nickel-hydrogen cells is shown in Figs. 6.7.1 and 6.7.2. The

41 mAh cell (Fig.6.7.1) degrades with cycling and loses about 50% of its capacity by 40 cycles. The 62.5 mAh cell (Fig. 6.7.2) provides constant discharge capacity for 10 cycles and then degrades significantly losing 50% of its capacity by 20 cycles. The failure of the

62.5 mAh cell will be considered in detail.

The voltage profiles during the charge-discharge of cycles 1, 4 and 12 are shown in Fig. 6.7.3. It is evident that the discharge time increases from cycle 1 to 4 and then

decreases to cycle 12. The voltage profile for the 12th charge shows a larger voltage at

the end of charging in comparison to 1st and 2nd charge. The voltage profiles of cycles 15,

18 and 21 are shown in Fig. 6.7.4. The discharge time (coulometric efficiency) of the

nickel-hydrogen cell decreases rapidly from cycle 15 to 21. The voltage profile during

charge for cycles 15, 18 and 21 shows a large change in voltage at the end of the

charging. This change to higher voltage corresponds to oxygen evolution during charge.

The timing of this transition to larger voltage decreases with cycles. This clearly shows

large amounts of oxygen evolution in the cell, which results in the decrease in

coulometric efficiency (discharge time) of the cell.

The end of charge voltage as a function of cycle number is shown in Fig. 6.7.5. It

is evident that the voltage increases at a slow rate below 11 cycles and then increases

rapidly from cycle 12 to 22. This corresponds well with cycle life data shown in

185

Fig.6.7.2. The pressure inside the cell at the end of every cycle is shown in Fig.6.7.6. The

pressure increases with cycles suggesting that the oxygen evolved does not recombine

completely during the rest period or during the subsequent discharge. The pressure

increases at a slow rate until cycle 14, and then increases rapidly. This again agrees with

the fact that the discharge capacity is constant up to 10 cycles and then decreases rapidly

on further cycling (Fig. 6.7.2).

Based on the results, the failure of the battery can be explained as follows:

• Oxygen evolves as a side reaction. This evolution causes a decrease in the

water content in the electrodes.

• If the oxygen recombines with hydrogen at the platinum electrode, the total

water content in the cell does not change and should not affect the cell

performance. However this recombination occurs at the platinum electrode

and the water formed has to equilibrate throughout the cell.

• Poor recombination or poor water equilibration will cause increased charging

voltages at the nickel hydroxide electrode and therefore increases oxygen

evolution. This creates a loop resulting in more and more oxygen evolution

and causes battery failure.

The picture of the failed cell is shown in Fig. 6.7.7. It is noted that nickel

hydroxide material is extruded out of the nickel foam and can be seen as a lump in the

picture. This is the typical mode of failure of these nickel hydrogen cells.

The cycle life performance of the cell needs to be improved by facilitating easy recombination at the negative electrode and preventing material extrusion from the nickel

186

hydroxide electrode. These two modes of failure need to studied in detail to enhance the cycle life.

5 Discharge capacity, mAh 0 0 1020304050 Cycle Index

Figure 6.7.1: Discharge capacity as a function of cycle index for a nickel-hydrogen cell with a capacity of 41 mAh. The cell was charged at C/5 rate to 90% depth of charge.

10 Discharge capacity, mAh 0 0 5 10 15 20 25 Cycle index

Figure 6.7.2: Discharge capacity as a function of cycle index for a nickel-hydrogen cell with a capacity of 62.5 mAh. The cell was charged at C/5 rate to 90% depth of charge.

187

1.55

1.5 12 4 1.45 1

1.4

1.35

1.3

Voltage(V) 1.25

1.2 12 4 1 1.15

1.1 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Time (s)

Figure 6.7.3: Voltage profiles during charge and discharge for cycles 1, 4 and 12. The capacity of the cell was 62.5 mAh.

1.55 21 18 1.5 15 1.45

1.4

1.35

1.3

Voltage (V) Voltage 1.25

1.2 21 15 1.15 18 1.1 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Time (s)

Figure 6.7.4: Voltage profiles during charge and discharge for cycles 15, 18 and 21. The capacity of the cell was 62.5 mAh.

188

1.54

1.52

1.5

1.48

1.46 End of charge voltage, V

1.44 0 5 10 15 20 25 Cycle Index

Figure 6.7.5: End of charge voltage as a function of cycle index. The capacity of the cell was 62.5 mAh.

8 Pressure (psi) 4

0 0 5 10 15 20 25 Cycle index

Figure 6.7.6: Pressure measured at the end of every cycle. The capacity of the cell was 62.5 mAh.

189

Figure 6.7.7: Picture of the failed nickel-hydrogen cell at the end of experiment.

6.8. Conclusions

A low pressure nickel hydrogen battery of 2-3 atm. pressure was successfully assembled. The performance of the nickel-hydrogen cell was compared to the performance of the individual electrodes in the liquid cell configuration. It was shown that the increase in charge voltage and the decrease in the discharge voltage of the Ni-H2 cell in comparison to the liquid cell are due to the large charge transfer resistance of platinum electrode. It is noted that the charge transfer resistance is significantly reduced for fresh negative electrodes and the total overpotential drop in these cells is small, about

10 to 20 mV.

Coulometric efficiency calculated from charge-discharge data shows that the oxygen evolution increases with depth of charge. The efficiency also decreases with increase in the rest period after charge indicating self-discharge of the nickel-hydrogen cell.

190

The pressure data measured simultaneously during charge and discharge were

analyzed. The results and discussions show that the oxygen evolution is observed at the

end of charging and increases with depth of charge supporting the charge data

measurement. The evolved oxygen recombines with the hydrogen gas and about 70% of

recombination occurs during the charge period. The rest of the recombination (30%) is

complete in the first 1h of the rest period. The recombination time constant is 1h in the

nickel-hydrogen cell in comparison to 15 min measured for a cell containing only the

platinum electrode. The time constant is large because the oxygen formed at the nickel

hydroxide electrode has to diffuse to the platinum electrode. The self discharge rate was

measured and is about C/117 at 90% DOC indicating that battery will discharge

completely in 117h.

Coulometric efficiency was estimated based on the pressure data analysis and

compares well with coulometric efficiency calculated based on the charge-discharge data.

SOC of the Ni-H2 cell can be estimated by monitoring the cell pressure in addition to the

existing methods like voltage measurements and coulomb counting.

The cycle life of the Ni-H2 cells was limited by the large amount of oxygen evolution and poor recombination. More studies need to be done to enhance the cycle life

performance by focusing on facilitating recombination and preventing material extrusion

from the nickel hydroxide electrode.

191

7. Low Pressure Nickel-Hydrogen Battery with Metal Hydride

In this chapter, the performance of the nickel-hydrogen cell using a metal hydride

for hydrogen storage will be discussed. The aspects of fuel gauging and state of charge

estimation and the problems associated with the nickel-hydrogen battery will be

introduced. The metal hydride technology for hydrogen storage will be briefly reviewed.

The differences between the nickel-metal hydride battery and our low pressure nickel-

hydrogen battery will be discussed.

7.1. Metal hydrides for hydrogen storage

7.1.1. Metal Hydride – Advantages and Applications

Metal hydride technology was developed in the 1970’s for hydrogen storage

applications. Subsequently they gained particular interest owing to their potential

application as a negative electrode to replace cadmium in nickel-cadmium batteries1-13.

The nickel metal hydride cell has higher gravimetric and volumetric energy densities than the conventional nickel cadmium cell by approximately 30-40% for cells of the same size. They are also environmentally benign in comparison to the toxic cadmium present in the nickel cadmium cell.

Metal hydrides are intermetallic compounds generally consisting of one strong hydride forming element (A) and one weak hydride forming element (B). Commonly used hydrogen storage alloys are (1) a wide set of alloys composed of rare earth elements with nickel (AB5 type systems) and (2) alloys of zirconium, vanadium, magnesium and/or

titanium with nickel (A2B, AB or AB2 type systems). Various hydrogen storage alloy

192

systems used as metal hydride negative electrodes are reviewed by Klepris et al.12 and

Feng et al13.

7.1.2. Factors Affecting Cycle Life of Metal Hydride Electrodes

The performance of the metal hydride electrodes are measured in terms of their

capacity, cycle life and power density. These capabilities are determined by the kinetics

of the processes occurring at the electrode/electrolyte interface and the rate of hydrogen

diffusion in the bulk of the alloy. For good cycle life, the metal hydride electrodes should

exhibit good oxidation resistance in alkaline electrolyte and when exposed to oxygen gas.

The nickel metal hydride battery operates in a strongly oxidizing environment containing

highly concentrated alkaline electrolyte. Therefore, the chemical elements often react to

form oxides (e.g. Lanthanum oxide13-16, Magnesium oxide17, 18) and fail to store hydrogen

reversibly. In some metal hydride systems, the metal (aluminum) dissolves19 and

migrates to the positive electrode. The oxidation of the chemical elements and dissolution

of aluminum in the metal hydrides can be minimized by adding Pd to the metal hydride17,

18, 20-34. Palladium is added to the metal hydride by either electroless coating20-23, ball milling along with the hydrogen storage alloy18 and electrodeposition17. High resolution transmission electron microscope (HRTEM) and X-ray diffraction (XRD) observations

18 have shown that the addition of Pd minimizes the formation of Mg(OH)2 and La(OH)2 .

These studies show that addition of Pd by coating or by bulk mixing (ball milling) to metal hydrides enhances the cycle life of these electrodes, significantly.

The other factor that increases the corrosion of the metal hydrides is commonly referred as particle pulverization (breaking of particles) or decrepitation13, 19, 35. An

193

electrode which has high storage capacity and large molar volume of hydrogen in the

hydride phase will undergo large volume changes upon cycling and pulverize rapidly.

This results in increased particle to particle contact resistance and polarization resistance

of the adsorption/desorption (hydriding/dehydriding) reaction. Therefore it is necessary to

formulate electrodes with low molar volume of hydrogen or with a narrow plateau region

in the pressure-composition isotherm. Zheng et al.23 indicated that an electrode made from Pd-coated alloy has much less ohmic resistance (particle to particle contact resistance and current collector to electrode pellet contact resistance) when compared to bare alloy. Researchers have also shown that the addition of palladium also enhances the kinetics of the hydrogen adsorption/desorption reactions due to the electrocatalytic effect of palladium20-22, 34, 36. Electrochemical techniques like cyclic voltammetry, impedance

spectroscopy and linear polarization have shown that the polarization resistance for the

hydrogen desorption/adsorption reaction is smaller when the metal hydride is coated with

37 Pd. Other studies show that the addition of Ce to the AB5 alloy retards corrosion

regardless of the lattice expansion/contraction. The explanation proposed is that Ce forms

a protective oxide film on the metal surface37. These studies indicate that the cycle life of

the metal hydrides can be increased by the addition of palladium or cerium.

The two factors discussed above, (a) surface passivation due to the presence of

surface oxide or hydroxide and (b) pulverization due to the large molar volume of

hydrogen in the hydride phase, primarily determine the cycle life of the battery. No other

significant factors were found in the literature search.

194

7.1.3. Comparison of Metal Hydride in Ni-MH and Low Pressure Ni-H2 Cell

The low pressure nickel hydrogen battery developed in this work uses a metal

hydride to store the hydrogen gas. The major difference between the low pressure nickel

hydrogen battery and the Ni-MH battery is that the metal hydride in the Ni-H2 cell is no

longer an electrode but just a hydrogen storage alloy. The metal hydride is physically and

electrically separated from the electrodes and the highly concentrated electrolyte in the

Ni-H2 cell. This should significantly reduce the oxidation of the chemical elements in the

metal hydride and enhance the cycle life of the metal hydride. The metal hydride used in

our Ni-H2 cell is LaAl0.3Ni4.7 surface modified by Pd and was developed in Prof. Payer’s

lab34, 36, 38 (Material Science dept.). Test results showed that the metal hydride does not

deactivate over 55 weeks in humid air38.

The pulverization or decrepitation phenomena occurs irrespective of whether the

metal hydride is used as a negative electrode (in a Ni-MH cell) or as a hydrogen storage

alloy (in a low pressure Ni-H2 cell). The consequence of pulverization of particles is

electrical isolation. In the case of Ni-MH cells, pulverization leads to increased particle to

particle contact resistance, metal hydride to current collector resistance and polarization

resistance for the hydrogen adsorption/desorption (hydriding/dehydriding) reactions23.

The increase in the polarization resistance is due to the decrease in the electrochemically active area of metal hydride as a result of increased particle to particle contact resistance.

All these factors contribute to the decrease in the capacity and cycle life of the metal hydride. The hydrogen adsorption/desorption occurs through an electrochemical reaction as follows:

++−− + M xH2 O xeU MHx OH [7.1]

195

where M is the hydrogen storage alloy that forms metal hydride MHx. The water in the electrolyte is reduced to hydroxyl ions at the metal hydride electrode.

In comparison, the pulverization of the metal hydride particles in the low pressure

Ni-H2 cell does not affect the capacity or cycle life of the metal hydride. The broken particles as a result of pulverization are still active to hydrogen adsorption/desorption reaction which occurs through the interaction of hydrogen gas with the hydrogen storage material M as follows:

1 M ()sxHgMHs+ ( )U () [7.2] 2 2 x

Therefore the contact resistance and the polarization resistance for the electrochemical mechanism in the case of the Ni-MH cell becomes irrelevant for the low pressure Ni-H2 cell. The pulverization phenomenon will not affect the cycle life of the Ni-H2 cell. In addition, the metal hydride, LaAl0.3Ni4.7 surface modified by Pd, used in our low pressure

Ni-H2 cell has a fairly narrow plateau width, ∆x = 0.62 approximately. This compares

35 well with the plateau width, ∆x = 0.54 of the stable metal hydride La0.7Sm0.3Ni4.0Fe1.0 .

The narrow plateau width also significantly reduces the decrepitation phenomena of the metal hydride.

In summary, addition of palladium to metal hydride increases the cycle life by minimizing the formation of metallic oxides and the pulverization of the particles. This is an advantage common to both applications in a) metal hydride batteries and b) low pressure nickel hydrogen batteries. However, the cycle life of the metal hydride in the low pressure Ni-H2 cell is enhanced (in comparison to Ni-MH cell) due to two reasons.

1. In the low pressure nickel hydrogen cells, the formation of oxide is significantly

lower because the metal hydride is physically separated from the electrolyte.

196

2. The pulverization phenomena does not decrease the capacity of the metal hydride

in Ni-H2 cell. It decreases the electrochemically active area for hydrogen

adsorption/desorption thereby affecting only the Ni-MH battery.

7.2. Pressure Composition Isotherm of the Metal Hydride

As discussed in the previous section, the surface modified intermetallic alloy

LaNi4.7Al0.3 with palladium developed by Shan et al (Prof. Payers Lab, Material Science

Dept., Case Western Reserve University) was used in our nickel hydrogen cell setup. The alloy is easily activated at sub-atmospheric pressures34. The pressure composition isotherm (PCI) of the surface modified alloy is shown in Fig. 7.2.1. The hydrogen concentration in the alloy is expressed in terms of the atom ratio of hydrogen to metal,

H/M. The equilibrium pressure, pounds per square inch, at different hydrogen compositions in the alloy is shown in the figure. As the hydrogen pressure is increased during absorption, the hydrogen concentration increases in the alloy. It is noted that one atm. pressure corresponds to 14.7 psia and most of the absorption is below 1 atm. pressure. The pressure composition curve is lower during the desorption indicating hysteresis.

The pressure composition curves for adsorption and desorption were fitted using a polynomial to predict the hydrogen concentration for a given pressure. For adsorption, the polynomial fit is:

H/M = 3.1376 x10-5P4 – 1.3261x10-3P3 + 1.7142x10-2P2 – 1.8507x10-2P +

5.7693x10-2, for 2.7≤P≤ 16.3 psi. [7.3]

For desorption, the polynomial fits over two ranges are:

H/M = -5.8105x10-4P2 + 1.9344x10-2P + 6.2049x10-1 for 8.75 ≤P≤16.3 psi

197

[7.4]

H/M = 3.1712x10-4P4 - 9.8569x10-3P3 + 9.4420x10-2P2 - 2.3430x10-1P +

3.1063x10-1, for 2.4≤P<8.75 psi. [7.5]

18 16 LaNi4.7Al0.3 with 10wt% Pd 14 12 10 Absorption 8 6 Pressure, psia Pressure, 4 Desorption 2 0 0 0.2 0.4 0.6 0.8 1 H/M

Figure 7.2.1: Pressure-composition isotherm of the 10 wt% Pd modified LaNi4.7Al0.3.

7.3. Testing of Metal Hydride Exposed to Air

The effect of air exposure on the hydrogen absorption of the metal hydride was studied by Shan et al. 36 and is shown in Fig. 7.3.1. It is evident that the no significant degradation is observed for 110 weeks (2 years) of exposure. A small decrease in the initial reaction rate and a small increase in the time constant for absorption are evident for the hydrogen absorption for 60 and 110 weeks, however these changes are negligible

198

This indicates that the metal hydride, Pd treated LaNi4.7Al0.3, is suitable for the application as a hydrogen storage device in our low pressure nickel-hydrogen battery.

0.8 LaNi4.7Al0.3 treated with 10wt% Pd 0.7

0.6

0.5

0.4 1st absorption

0.3 Freshly ground 5 weeks 0.2 11 weeks

Hydrogen Concetration, H/M Concetration, Hydrogen 60 weeks 0.1 110 weeks 0.0 0 5 10 15 20 25 30 Time, min

Figure 7.3.1: Effect of air exposure on the first hydrogen absorption of palladium treated 36 LaNi4.7Al0.3 for periods greater than 2 years. This figure is taken from reference .

7.4. Testing of Metal Hydride Exposed to KOH Solution

The effect of potassium hydroxide solution on the metal hydride pasted in nickel foam was studied by soaking the metal hydride in KOH solution for a day. The hydrogen absorption, in moles, before and after soaking in KOH solution is shown in Fig. 7.4.1.

The time constant of the hydrogen absorption reaction before soaking in KOH solution is about 5 min, similar to that shown in Fig. 7.3.1. It is evident that the rate of hydrogen

199

absorption is slower for the metal hydride soaked in KOH in comparison to that of the fresh metal hydride. The soaked metal hydride absorbs only about 50% of its original capacity at the end of 2.5h. This indicates that the metal hydride is deactivated, probably by the formation of an oxide layer when exposed to KOH solution. Therefore it is necessary to physically isolate the metal hydride from the KOH solution in the nickel- hydrogen cell. It is known that the metal hydride can be activated again by raising the pressure.

2.0x10-4 Fresh

1.5x10-4

1.0x10-4

1 day 5.0x10-5 abosrbed,moles 2 H 0.0

0 30 60 90 120 150 Time, min

Figure 7.4.1: Effect of soaking in KOH solution on the hydrogen absorption of palladium treated LaNi4.7Al0.3. The hydrogen pressure at the end of absorption is about 1.1 atm pressure.

7.5. Testing of Ni-H2 Battery with Metal Hydride

A nickel-hydrogen cell with metal hydride for hydrogen storage was assembled as shown schematically in Fig. 7.5.1. It is a bolt nut cell as described in section 6.1. Nickel

200

foam pasted with the metal hydride, Pd treated LaNi4.7Al0.3, was placed at the end of the bolt nut cell and was physically separated from the platinum electrode by a chlorinated polyvinyl chloride disc. This minimizes contact of the KOH solution with the metal hydride. The metal hydride was placed close to the platinum electrode to facilitate easy hydrogen oxidation during discharge. The hydrogen absorption of the metal hydride was tested in a Sievert’s Apparatus38 before use in the cell. The stainless steel compartment was first purged with N2. This was followed by purging with hydrogen gas, and the metal hydride absorbed hydrogen according to the pressure composition isotherm described in section 7.2. The final pressure of the cell at the end of purging was about 1-2 atm pressure.

Positive Negative Metal electrode electrode Hydride Ni(OH)2 Platinum Insulator

3.2 cm Outlet valve Bolt Nut

Separator Current collector

Inlet Stainless steel compartment valve 1.4 cm

Figure 7.5.1: Schematic picture of the Ni-H2 cell with metal hydride.

201

The nickel-hydrogen cell was reduced to lower pressure as shown schematically in Fig. 7.5.2 by opening the cell to an evacuated chamber. This method eliminates the use of a vacuum pump which affects the potassium hydroxide electrolyte during pumping.

The corresponding pressure change measured during venting is shown in Fig. 7.5.3. The time constant for the hydrogen gas to equilibrate with the metal hydride in the nickel hydrogen cell setup is about 0.1 to 0.2h and is about the same time as that measured for the metal hydride separately (Fig. 7.3.1) indicating that there is no problem associated with the hydrogen access to the metal hydride.

The results of a nickel-hydrogen cell with experimental capacity of 13.24 mAh/cm2 (positive electrode) will be discussed in the section. The experimental capacity was determined in a liquid cell for coulometric efficiency greater than 90%. The absolute capacity of the positive electrode is 72 mAh. The capacity of the metal hydride used is

121 mAh (1 atm pressure). The capacity of the gaseous volume (excluding the cell volume) at 1 atm pressure is 86 mAh. Therefore the total capacity of the negative electrode for hydrogen oxidation (including the gas capacity and metal hydride capacity) is 207 mAh. This shows that the capacity of the positive electrode is limiting in comparison to that of the negative electrode.

Inlet Evacuated Valve Ni-H2 Cell chamber

Outlet Valve

Figure 7.5.2: Schematic picture of the Ni-H2 cell with metal hydride.

202

Venting of H 2 16

10 Pressure change by venting Pressure Equilibration Equlibration time 8 Absolute Pressure, psi 0.00 0.25 0.50 0.75 1.00 Time, h

Figure 7.5.3: Pressure variation inside the nickel hydrogen cell during venting and subsequent hydrogen gas equilibration with the metal hydride.

7.5.1. Coulometric efficiency Based on Charge-Discharge Data

The coulometric efficiency of the cell containing the metal hydride was investigated at different depths of charge and different rest periods to evaluate oxygen evolution during charge and the self discharge mechanism during rest period of the cell.

The cell was cycled at C/5 rate to different depths of charge, 50, 70 and 90%. The voltage variations during charge and discharge are shown in Fig. 7.5.4. The discharge cut-off voltage is 1.15 V. The rest period between charge and discharge was about 5 min. The voltage profile during the charge is about the same for different depths of charge indicating the cyclability of the cell.

The rest period between the charge and the discharge was varied from 5 min to 1h to 6h. The voltage profile during charge-discharge of the cell for different rest periods is

203

shown in Fig. 7.5.5. The depth of charge is 50%. It is evident that the time of discharge decreases with increase in the rest period indicating self discharge. The voltage plateau during the discharge also decreases with increase in the rest period because the state of charge of the cell at the end of rest period is lower with increase in the rest period.

The coulometric efficiency of the cell was calculated for different depths of charge and different rest periods and is shown in Fig. 7.5.6. It is evident that the coulometric efficiency decreases with increasing depth of charge for all rest periods indicating that the oxygen evolution increases with depth of charge, similar to the results obtained for the nickel-hydrogen cell without metal hydride. The coulometric efficiency also decreases with increase in the rest period for different depths of charge indicating that self discharge is occurring in the nickel-hydrogen cell. Self discharge in the nickel hydrogen cell with metal hydride will be compared to the cell without metal hydride in section 7.5.8.

1.5

1.4

1.3 E (Volts)

1.2

50% DOC 70% 90% 1.1 012345 Time (Hours)

Figure 7.5.4: Charge-discharge voltage profiles of the nickel-hydrogen cell cycled at C/5 rate to different depth of charge.

204

1.5

1.4

1.3 E (Volts)

1.2 5 min 6h 1h 1.1 0123 Time (Hours)

Figure 7.5.5: Charge-discharge voltage profiles of the nickel-hydrogen cell cycled at C/5 rate to 50% DOC with different rest periods, 5 min, 1h and 6h.

96 0.083h 1h 94 6h

Columetric efficiency, % 86

50 60 70 80 90 Depth of charge (%)

Figure 7.5.6: Coulometric efficiency of the Ni-H2 cell cycled at C/5 rate as a function of the depth of charge for different rest periods.

205

7.5.2. Pressure Variations During the Charge-Discharge Cycle

The pressure inside the nickel hydrogen cell was monitored using a pressure gauge as described in section 6.4. The pressure variation during two successive cycles of the nickel-hydrogen cell with metal hydride is shown in Fig. 7.5.7. The cell was cycled at

C/5 rate to 70% depth of charge. The rest period for the first cycle and the second cycle is

5 min and 6h, respectively. The pressure variation during charge and discharge is similar to that observed for the nickel-hydrogen cell without the metal hydride discussed in chapter 6. The pressure increases during the charge due to hydrogen evolution and the pressure decreases during discharge due to hydrogen oxidation at the platinum electrode.

However, three major differences are observed in the cell with metal hydride in comparison to the cell without metal hydride. The first difference is the net drop in pressure measured at the end of the first cycle with 5 min rest period as shown in Fig.

7.5.7. For the cell without metal hydride no such drop in pressure was observed. The second difference is the pressure rise during the 6h rest period after the 2nd discharge. The pressure at the end of two cycles is equal to the pressure measured at the beginning of the two cycles. This suggests that the 6h rest periods results in good hydrogen equilibration with the metal hydride. No pressure rise was observed during the rest period after discharge of the cell without metal hydride (Fig. 6.4.2). It is noted that the pressure variation during discharge is non-linear in comparison to the linear variation observed for the cell without metal hydride (third difference). The non-linear pressure variation during the charge and the discharge are related to hydrogen gas equilibration with the metal hydride and will be discussed in detail in section 7.5.3. The pressure drop evident during the 6h rest period after charge is related to the recombination reaction, the hydrogen

206

equilibration with the MH and the self discharge of the cell and will be discussed in section 7.5.5.

5 min rest cycle 6h rest cycle 12

8 Pressure, psia

∆P ∆P 6

04812162024 Time, h

Figure 7.5.7: Changes in pressure during the charge, the rest and the discharge periods of two successive cycles. The cell was cycled at C/5 rate to 70% depth of charge. The rest period for the 1st cycle and the 2nd cycle is 5 min and 6h, respectively.

7.5.3. Pressure Variations During Charge

The change in pressure during the charge corresponding to the cycle with 5 min rest period is shown in Fig. 7.5.8. The capacity of the nickel hydrogen cell is 72 mAh

(13.24 mAh/cm2). The volume of the cell is about 40 cc. For C/5 charge rate, the rate of pressure rise corresponding to the hydrogen formation at the platinum electrode calculated based on Faraday’s law and ideal gas law is 2.33 psi/h (when there is no MH).

However, part of the hydrogen formed is absorbed into the metal hydride and therefore the rate of pressure change should be lower than 2.33 psi/h. Two different slopes corresponding to two rates of pressure rise are shown in Fig. 7.5.8. The pressure rate in

207

the time period, 0 to 0.5h is 1.73 psi/h and that in the 1 to 2h is 1.51 psi/h. As expected they are lower than the pressure rate of 2.33 psi/h. This indicates that the hydrogen formed is absorbed into the MH. The two different slopes are related to the pressure plateau and the pressure rise observed in the pressure composition isotherm.

11 1.51 psi/h 10

8 1.73 psi/h Pressure, psia

6 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Time, h

Figure 7.5.8: Changes in pressure during charge. The cell was cycled at C/5 rate to 70% depth of charge with 5 min rest period.

7.5.4. Pressure Variations During Rest after Discharge

The pressure during the 6h rest period at the end of the 2nd cycle rises to a higher value as shown in Fig. 7.5.9. It is noted that there is no oxygen evolution and therefore no recombination at the end of discharge to cause a change in pressure. It was shown in section 7.4 that the time constant for metal hydride equilibration increases (metal hydride is deactivated) when soaked in KOH solution. Therefore it is likely that this pressure rise during the rest period is related to slow hydrogen equilibration (slow hydrogen desorption/absorption) with the metal hydride. This is also supported by the fact that the

208

pressure measured at the end of the rest period, after the pressure rise, is equal to the pressure measured at the beginning of the 1st charge (Fig. 7.5.7). The slow hydrogen desorption was also observed when the cell pressure was lowered by venting the hydrogen gas in the cell into an empty chamber (Fig.7.5.2).

The time constant for the hydrogen to equilibrate with the metal hydride was estimated using an exponential fit as shown in Fig. 7.5.8. Oh et al. 39 have developed a model for the hydrogen absorption based on different rate controlling steps: (a) dissociative chemisorption of the hydrogen molecule (b) diffusion of the hydrogen atoms through the metal hydride and (c) chemical reaction of the hydrogen atom with the hydride. Here we do not analyze which of them is a rate determining step and instead focus our attention on estimating the time constant for hydrogen absorption. Shan et al. 36 have briefly reviewed the literature on the kinetics of hydrogen adsorption in metal hydride. The exponential fit to the pressure data is good and estimates the time constant as 1.37h. In comparison, it is noted that the time constant of a fresh metal hydride to equilibrate with hydrogen gas is 0.1-0.2h. The increase in time constant is attributed to the formation of oxide layer on the metal hydride particles. The oxide layer is formed due

KOH solution seeping into the metal hydride probably through the bolt hole in the cell.

The increase in pressure during the rest period, given by parameter ‘A1’ is 0.59 psi.

209

6.8

6.6

6.4 Equation: y = A1*exp(-x/t1) + y0 R^2 = 0.99 -4 6.2 y0 6.6 ±2.4x10 A1 -0.59 ±4.5x10-4 t1 1.37 ±2.5x10-3 Pressure, psia 6.0

5.8 0123456 Rest period, h

Figure 7.5.9: Changes in pressure during the rest period at the end of the 2nd cycle shown in Fig. 7.5.7. The cell was cycled at C/5 rate to 70% depth of charge.

7.5.5. Pressure Variations During Rest after Charge

The change in pressure during the rest period after the 2nd charge (Fig. 7.5.7) is shown in Fig. 7.5.10. As discussed in chapter 6, the pressure decrease during the rest period after charge is due to the recombination of oxygen with hydrogen and self discharge of the nickel hydroxide electrode consuming hydrogen gas. The self discharge occurs throughout the rest period. However, it was shown that the recombination reaction is complete in the first 1h of the rest period. In addition to the self discharge and the recombination reaction during the rest period, there is also the hydrogen equilibration with the metal hydride as discussed in the previous section. The time constant for the hydrogen equilibration is about 1.4h. Therefore any changes in pressure observed for rest periods longer than 1.5h correspond to the self discharge of the cell.

210

A straight line was fit to the pressure data in the time period of 4 to 6h as shown in Fig. 7.5.10. The straight line fit corresponds to the self discharge of the battery. The error for the fit is small. The self discharge rate given by the slope of line or parameter

‘B’ of the fit is 0.063 psi/h. This self discharge rate is about 37 times slower than the C/5 rate. That is, the battery will discharge completely in about 185h at this rate and the self discharge rate is expressed as C/185. The effect of state of charge on the discharge rate will be discussed in section 7.5.8.

11.8 Equation: y = A + B*x R^2 = 0.90 A 11.31 ±1.3x10-3 11.6 B -6.3x10-2 ±2.5x10-4

11.4

11.2 Pressure, psia 11.0

10.8 01234567 Rest period, h

Figure 7.5.10: Pressure changes during the 6h rest period after 2nd charge. The cell was cycled at C/5 rate to 70% depth of charge. A straight line is fit to the pressure data in the 4 to 6h rest period.

The pressure drop due to the self discharge mechanism was subtracted out of the pressured data based on the linear fit parameters and the difference, the pressure contributions due to the recombination reaction and the hydrogen equilibration with the

MH is plotted in Fig. 7.5.11. An exponential decay equation was fit to the pressured data as shown in the figure. The estimated time constant is about 1.29h. It is noted that the

211

time constant for the recombination reaction (chapter 6) and that of the hydrogen equilibration with the metal hydride are about 1h (section 7.5.4) and therefore it is difficult to separate the pressure contributions due to these two mechanisms. However, it is noted that in the pressure range during the rest period, 11.8 -11 psi, the hydrogen absorption/desorption is insignificant as shown by the pressure composition isotherm

(Fig.7.2.1). Therefore the pressure variation shown in Fig. 7.5.11 is mostly due to the oxygen-hydrogen recombination reaction.

0.5 Equation: y = A1*exp(-x/t1) + y0 R^2 = 0.99 0.4 -4 y0 -0.009 ±1.9x10 A1 0.47 ±3.9x10-4 0.3 t1 1.29 ±2.4x10-3

0.2

Pressure, psia 0.1

0.0

01234567 Rest period, h

Figure 7.5.11: Pressure changes due to recombination reaction and hydrogen equilibration with the metal hydride during the rest period after 2nd charge. The cell was cycled at C/5 rate to 70% depth of charge. An exponential decay line is fit to the pressure data.

7.5.6. Correlation of Pressure Variations During Discharge to Pressure

Composition Isotherm – Fuel Gauging

The change in pressure during the rest period after 2nd charge, the subsequent discharge and the following rest period is shown in Fig. 7.5.12. As discussed earlier, it is

212

noted that the decrease in pressure during the discharge is due to hydrogen oxidation at the platinum electrode. Though the hydrogen oxidizes at a constant rate (pressure change should be linear), the pressure measured during discharge is non-linear because the hydrogen equilibration with the metal hydride is non-linear as shown in the pressure composition isotherm (Fig. 7.2.1). The pressure variation and other parameters as a function of the state of charge (SOC) of the cell were calculated from the pressure composition isotherm and are listed in Table 7.1. It is noted that the state of charge of cell at the beginning of discharge is 60.5% instead of 70% because of significant self discharge during the rest period after charge (coulometric efficiency for this cycle is

86.4%).

The pressure at the start of discharge is 10.93 psi as shown in Fig. 7.5.12 and listed in Table 7.1. As the pressure decreases during discharge, hydrogen from the gaseous phase and from the metal hydride is oxidized at the platinum electrode. For example, when the cell is discharged from the 60.5% SOC to 31.4% SOC (Table 7.1), the contribution of hydrogen atom from the gaseous phase is 6.2x10-4 moles and that from the metal hydride is 1.6x10-4 moles. The method to calculate these contributions is discussed below.

The contribution of the hydrogen from the gaseous phase is based on ideal gas law: nPVRT=∆ , where ∆P is the pressure drop, V is the volume gaseous space in the cell, R is the gas constant and T is the temperature. From 60.5% to 31.4% SOC, the pressure drop is (10.93-8.12) = 2.81 psi. The volume of the gaseous space is 40 cc. The moles of hydrogen gas oxidized from the gaseous phase is

56− 2.81 14.696×××× 1.013 10 40 10 − n ==×3.1 10 4 8.314× 298

213

The number of moles of hydrogen atom = 23.110××−4 = 6.2× 10−4

The contribution of the hydrogen from metal hydride phase is obtained from the pressure composition isotherm. For 8.12 psia, the polynomial fit given by equation [7.5] was used to estimate the H/M ratio as 0.735. The number of hydrogen atoms consumed from the metal hydride is given as

=∆() × × nHMNMH 6 where M/H is the number of hydrogen atoms per metal atom, NMH is the number moles of the metal hydride, 9.64x10-4, and 6 represents that there are 6 metal atoms per mole of metal hydride. When the cell discharges from 60.5% to 31.4% SOC, the change in the

M/H ratio is (0.762-0.735) = 0.027. The moles of hydrogen atom oxidized from the metal hydride is n =×××=×0.027 9.64 10−−44 6 1.6 10 .

The rate of pressure change in the 0-1 h discharge (16.85-17.85h,) is 1.94 psi/h as shown in the Fig. 7.5.12. For this period, 0-1 h, the corresponding pressure change is

10.93 to 9 psia. It is noted for this pressure range, there is no significant change in the

H/M ratio and as listed in Table 7.1 all the hydrogen consumed is from the gaseous phase rather than from the metal hydride. This is explains the fact that the actual pressure rate measured, 1.94 psi/h, is close to that of 2.33 psi/h (without MH). It is evident from the figure that the rate of pressure change decreases towards the end of discharge agreeing with the fact that the hydrogen oxidized is both from gaseous and the metal hydride phase as listed in Table 7.1. It is noted that at the end of discharge, about 41% of hydrogen oxidized at the platinum electrode is from the metal hydride and the rest, 59%, is from the gaseous phase.

214

It can be seen in Table 7.1 that the pressure estimated at 0% SOC is 6.62 psi. This agrees well with the pressure measured at the end of the rest period after discharge, 6.58 psi. This clearly indicates that the pressure inside the nickel hydrogen cell during discharge follows the desorption curve of the pressure composition isotherm as expected and can be used to gauge the fuel (hydrogen available) in the metal hydride.

11 10.93 Discharge 10 Rest 1.94 psi/h

8 Rest

Pressure, psia 7 6.58 6 5.94

12 14 16 18 20 22 24 26 Time, h

Figure 7.5.12: Changes in pressure during the rest period after charge, the discharge period and the subsequent rest period of the 2nd cycle. The cell was cycled at C/5 rate to 70% depth of charge.

Table 7.1: Change in various parameters following the desorption curve of the pressure composition isotherm for discharge. Pressure, H moles, M/H ratio H moles, Total H SOC,% psia from air from MH moles 10.93 0 0.762 0 0 60.5 10.06 1.9E-04 0.756 3.6E-05 2.3E-04 52.0 9.67 2.8E-04 0.753 5.4E-05 3.3E-04 48.1 8.89 4.5E-04 0.745 9.9E-05 5.5E-04 40.0 8.12 6.2E-04 0.735 1.6E-04 7.8E-04 31.4 7.35 8.0E-04 0.701 3.6E-04 1.2E-03 17.6 6.96 8.8E-04 0.675 5.1E-04 1.4E-03 8.8 6.62 9.6E-04 0.647 6.7E-04 1.6E-03 0.0

215

7.5.7. Correlation of Pressure Variations During Charge to Pressure

Composition Isotherm – Fuel Gauging

The change in pressure in the nickel-hydrogen cell during charging was estimated using the adsorption curve of the pressure composition isotherm and is listed in Table 7.2 as a function of state of charge. The estimation is similar to that described in section

7.5.6. It is noted that the coulometric efficiency to 70% depth of charge is 92.9%, based on the 5 min rest cycle. Therefore the SOC of the cell is actually 65.1% at the end of charge instead of 70%. Based on the pressure composition isotherm, the pressure at the end of discharge is estimated as 9.63 psi (Table 7.2). The pressure measured at the end of

2nd charge and at the end of the subsequent rest period is about 11.81 and 10.93 psi, respectively as shown in Fig. 7.5.7. It is evident that the estimated pressure based on the adsorption curve of the isotherm does not agree and is smaller than the pressure measured at the end of charge or that measured at the end of the subsequent rest period.

The pressure variation for different states of charge during charging was then estimated based on the desorption curve of the pressure composition isotherm and is listed in Table 7.3. The pressure estimated by the PCI at 65.1% SOC (end of charge) is

11.49 psi and is below the pressure measured at the end of charge, 11.81 psi, and is above the pressure measured at the end of the rest period, 10.93 psi. This suggests that the pressure change during charge follows the desorption curve rather than the adsorption curve of the pressure composition isotherm. It should be noted that the desorption curve of the pressure composition isotherm as shown in Fig. 7.2.1 starts from complete desorption and in our experiments the metal hydride is not desorbed completely before the start of charging.

216

The pressure analysis during charge and discharge (sections 7.5.6 and 7.5.7) shows that the desorption curve can be used to gauge the fuel (hydrogen available) of the nickel-hydrogen cell. As discussed in page 11, the positive electrode capacity is about 2.9 times smaller than the hydrogen capacity in the tested cell. In the case of the hydrogen capacity being limiting, the pressure variations will be significant and can be followed using the desorption curve. In either case, a voltage cutoff should be used terminate the charging. Otherwise, overcharging will cause enormous amounts of oxygen evolution at the positive electrode and hydrogen evolution at the negative electrode resulting in possible separator dry out.

Table 7.2: Change in various parameters following the adsorption curve of the pressure composition isotherm for charge. Pressure, H moles, M/H ratio H moles, Total H SOC,% psia from air from MH moles 6.67 0 0.366 0 0 0.0 6.96 6.5E-05 0.386 1.2E-04 1.8E-04 6.8 7.35 1.5E-04 0.413 2.7E-04 4.2E-04 15.7 7.73 2.4E-04 0.439 4.2E-04 6.6E-04 24.6 8.12 3.2E-04 0.464 5.7E-04 8.9E-04 33.2 8.51 4.1E-04 0.489 7.1E-04 1.1E-03 41.7 8.89 5.0E-04 0.512 8.5E-04 1.3E-03 50.1 9.28 5.8E-04 0.535 9.8E-04 1.6E-03 58.1 9.63 6.6E-04 0.555 1.1E-03 1.8E-03 65.2

Table 7.3: Change in various parameters following the desorption curve of the pressure composition isotherm for charge. Pressure, H moles, M/H ratio H moles, Total H SOC,% psia from air from MH moles 6.65 0 0.650 0 0 0.0 6.96 6.9E-05 0.675 1.5E-04 2.1E-04 8.0 7.35 1.5E-04 0.701 3.0E-04 4.5E-04 16.8 7.73 2.4E-04 0.721 4.1E-04 6.5E-04 24.3 8.12 3.3E-04 0.735 4.9E-04 8.2E-04 30.6 8.89 5.0E-04 0.745 5.5E-04 1.1E-03 39.2 9.67 6.7E-04 0.753 6.0E-04 1.3E-03 47.3 10.44 8.4E-04 0.759 6.3E-04 1.5E-03 55.0 11.49 1.1E-03 0.766 6.7E-04 1.7E-03 65.1

217

7.5.8. Self discharge at Different Depth of Charge

The self discharge rate was estimated based on the linear fit on the pressure data in the 4 to 6h rest period after charge as discussed in section 7.5.5 and is listed in Table

7.4. It is evident that the self discharges rate at 70 and 90% depth of charge are the same and are greater than that at the 50% depth of charge. The self discharge rate of cell with

MH is C/185 at 90% DOC. In comparison, the self discharge rate of the nickel hydrogen cell without the MH is C/117 indicating that the self discharge rate is lower for the cell with MH. However, it is noted that these tests were based on two different experiments with different set of electrodes.

Table 7.4: Self discharge rate for different depths of charge. % DOC Self discharge rate, psi/h 50 0.048 70 0.063 90 0.063

7.5.9. Recombination During Charge and Rest

The pressure contributions due to the recombination of oxygen with hydrogen during the charge period and the rest period are difficult to estimate from the pressure data for this cell with the metal hydride because of the slow equilibration of hydrogen with the MH. However, in the case of a fresh metal hydride, the kinetics of hydrogen absorption/desorption are fast, and the amount of oxygen evolution can be estimated considering the pressure composition isotherm. Therefore it is important to prevent deactivation of the metal hydride which makes it difficult to estimate the pressure drop due to the recombination reaction.

218

7.5.10. Prediction of State of Charge

As discussed in section 7.5.4, the time constant for the hydrogen equilibration with the metal hydride is about 1.37h and indicates deactivation of the metal hydride.

This slow equilibration causes difficulty in the estimation of oxygen evolution and the coulometric efficiency. Therefore the state of charge of the nickel hydrogen cells cannot be predicted accurately based on pressure data analysis. However for the fresh metal hydride, the state of charge can be predicated accurately.

7.6. Problems associated with Ni-H2 Battery

The main requirement of the low pressure nickel-hydrogen cell is that it should be hermetically sealed to prevent oxygen diffusion into the cell and hydrogen loss out of the cell. Loss of hydrogen results in low fuel and will affect the cell performance. The effect of oxygen diffusion into the cell is discussed in this section. The diffused oxygen recombines with hydrogen at the platinum electrode to form water. Excess oxygen diffusion leads to flooding at the platinum electrode due to excess water formation. This results in poor hydrogen distribution to the platinum electrode. The voltage profile during charge and discharge of a nickel hydrogen cell with and without any oxygen leak is shown in Fig. 7.6.1. The cell was cycled at C/5 rate to 50% depth of charge. It is evident from the figure that the voltage profile during charge is the same for the two cases because hydrogen is formed as a result of charging and there are no diffusion limitations.

However, the voltage profile during discharge is lower for the case with oxygen leak because of poor hydrogen diffusion to the platinum electrode. The impedance measured at the end of charge and discharge for the two cases is shown in Fig.7.6.2. It is evident

219

that the charge transfer loop corresponding to the platinum electrode (2nd loop) is larger for the case of oxygen leak in comparison to the case with no oxygen leak. This supports the idea that excess flooding prevents good hydrogen distribution.

The pressure variation during the rest period after discharge for the case ‘oxygen leak’ is shown in Fig. 7.6.3. It is evident that pressure rises during the rest period and could be due to (a) poor hydrogen equilibration with the metal hydride as discussed in section 7.5.4 or (b) poor hydrogen diffusion in the platinum electrode. An exponential fit to the pressure data measured during the rest period shows that the time constant is 1.85h.

The time constant for hydrogen equilibration with the metal hydride was determined by increasing the pressure inside the cell from about 6.34 psi to about 31.8 psi as shown in Fig. 7.6.4. During this time the no current was applied to the cell. The time constant estimated from the exponential fit as shown in the figure is about 0.29h (17.4 min) and corresponds to the time constant for a fresh metal hydride. This indicates that the hydrogen absorption/desorption is fast.

Therefore the time constant measured at the end of discharge (Fig. 7.6.3) corresponds to poor hydrogen diffusion in the platinum electrode. This analysis illustrates the importance of hermetically sealing the cell to prevent oxygen leaking or otherwise results in flooding of the platinum electrode.

The other main problem is the deactivation of the metal hydride when exposed to the potassium hydroxide solution as discussed earlier in the chapter. This deactivation leads to poor hydrogen adsorption/desorption and is difficult to estimate the amount of oxygen evolution. It is necessary to utilize a metal hydride that does not deactivate on contact with KOH solution.

220

1.50

1.25 No oxygen leak E (Volts) 1.00

Oxygen leak

0.75 0123 Time (Hours)

Figure 7.6.1: Charge-discharge voltage profiles of the nickel-hydrogen cell cycled at C/5 rate to 50% depth of charge for the case with and without oxygen leak.

221

-4 FloodingOxygen leak -3

-2 No flooding No Oxygen (A) Z''

-1

1 012345 Z'

-7.5

OxygenFlooding -5.0 leak

NNoo floodingoxygen Z'' -2.5 (B)

2.5 0 2.5 5.0 7.5 10.0 Z'

Figure 7.6.2: Comparison of impedance spectra for two different cases (i) no oxygen leak and (ii) oxygen leak measured at the end of charge (A) and discharge (B). The cell was cycled at C/5 rate to 50% depth of charge.

222

6.4

6.3

6.2 Equation: y = A1*exp(-x/t1) + y0 6.1 R^2 = 0.97 y0 6.34 ±3.3x10-4 -4 6.0 A1 -0.42 ±5x10 t1 1.85 ±5.7x10-3 Pressure, psia 5.9

5.8

0123456 Rest period, h

Figure 7.6.3: Changes in pressure during the rest period at the end of discharge for the case ‘oxygen leak’. The cell was cycled at C/5 rate to 50% depth of charge.

32 Equation: y = A1*exp(-x/t1) + y0 30 R^2 = 0.97 y0 19.70 ±3.6x10-3 28 A1 8.20 ±1.8x10-2 t1 0.29 ±1x10-3 26

22 Pressure, psia 20

0.0 0.5 1.0 1.5 2.0 2.5 3.0 Rest period, h

Figure 7.6.4: Change in pressure inside the nickel hydrogen cell by increasing the pressure suddenly to 31.8 psi for the case ‘oxygen leak’. The cell was cycled at C/5 rate to 50% depth of charge. The pressure variation shows hydrogen gas equilibration with the metal hydride.

223

7.7. Conclusions

A nickel-hydrogen cell with metal hydride for hydrogen storage was assembled and tested successfully. The coulometric efficiency of the cell was studied for different depth of charge and different rest periods. The coulometric efficiency decreases with increase in depth of charge because of oxygen evolution as side reaction. The coulometric efficiency also decreases with increasing rest period because of self discharge of the cell.

These results are similar to the observations in the cell without the metal hydride.

Pressure data measured simultaneously during cycling was analyzed. In addition to the variables: oxygen evolution, recombination reaction and self discharge for the cell, the use of metal hydride for hydrogen storage introduces another variable the equilibration of hydrogen with the metal hydride. It was shown that metal hydride is deactivated when soaked in potassium hydroxide solution for one day. For the nickel hydrogen cell discussed in detail, it was shown that the hydrogen equilibration with the metal hydride is slow based on the exponential fit for the pressure data in the rest period after discharge. The time constant for the slow equilibration and for the recombination reaction (chapter 6) is about one hour making it difficult to separate the pressure contributions due to these mechanisms. The self discharge rate, however, was obtained using a linear fit at longer time scales.

It was shown that the discharge of the battery follows the desorption part of the pressure composition isotherm very well. However, the pressure variation during charge does not follow the adsorption part of the pressure composition isotherm and in fact follows the desorption part of the isotherm. About 40-45 % hydrogen from the metal

224

hydride was used as fuel for cycling the battery. Based on the pressure composition isotherm, the fuel in the cell can be easily gauged.

For a well behaved metal hydride (fast hydrogen equilibration) and with the knowledge of the pressure composition isotherm, the oxygen evolution and the coulometric efficiency can be estimated based on the pressure data analysis, similar to that shown in chapter 6.

The problems associated with nickel hydrogen battery as constructed were briefly discussed. It was shown that oxygen diffusion into the cell caused flooding at the negative electrode leading to poor hydrogen access at the platinum electrode. The cell should be hermetically sealed to prevent oxygen diffusion into the cell. The other problem, the deactivation of metal hydride has to be eliminated either by improving the metal hydride or by preventing the seeping of potassium hydroxide solution into the metal hydride.

References

1. P. Dantzer, Topics in Applied Physics, 73, 279 (1997).

2. S. K. Dhar, S. R. Ovshinsky, P. R. Gifford, D. A. Corrigan, M. A. Fetcenko, and S. Venkatesan, Journal of Power Sources, 65, 1 (1997).

3. O. A. Petrii, S. Y. Vasina, and I. I. Korobov, Uspekhi Khimii, 65, 195 (1996).

4. G. Wojcik and M. Kopczyk, Wiadomosci Chemiczne, 49, 285 (1995).

5. A. Anani, A. Visintin, K. Petrov, S. Srinivasan, J. J. Reilly, J. R. Johnson, R. B. Schwarz, and P. B. Desch, Journal of Power Sources, 47, 261 (1994).

6. M. A. Fetcenko, S. Venkatesan, and S. R. Ovshinsky, Proceedings - Electrochemical Society, 92-5, 141 (1992).

7. T. Sakai, K. Muta, H. Miyamura, N. Kuriyama, and H. Ishikawa, Proceedings - Electrochemical Society, 92-5, 59 (1992).

225

8. C. Iwakura and M. Matsuoka, Progress in Batteries & Solar Cells, 10, 81 (1991).

9. H. F. Bittner and C. C. Badcock, Journal of the Electrochemical Society, 130, 193C (1983).

10. K. Hong, Journal of Power Sources, 96, 85 (2001).

11. S. R. Ovshinsky, M. A. Fetcenko, and J. Ross, Science (Washington, DC, United States), 260, 176 (1993).

12. J. Kleperis, G. Wojcik, A. Czerwinski, J. Skowronski, M. Kopczyk, and M. Beltowska-Brzezinska, Journal of Solid State Electrochemistry, 5, 229 (2001).

13. F. Feng, M. Geng, and D. O. Northwood, International Journal of Hydrogen Energy, 26, 725 (2001).

14. Z. Zhou, J. Yan, Y. Li, D. Song, and Y. Zhang, Journal of Power Sources, 72, 236 (1998).

15. P. Ruetschi, F. Meli, and J. Desilvestro, Journal of Power Sources, 57, 85 (1995).

16. J. J. G. Willems and K. H. J. Buschow, Journal of the Less-Common Metals, 129, 13 (1987).

17. H. J. Park, N. H. Goo, and K. S. Lee, Journal of the Electrochemical Society, 150, A1328 (2003).

18. T. Ma, Y. Hatano, T. Abe, and K. Watanabe, Journal of Alloys and Compounds, 391, 313 (2005).

19. P. Leblanc, C. Jordy, B. Knosp, and P. Blanchard, Journal of the Electrochemical Society, 145, 860 (1998).

20. R. C. Ambrosio and E. A. Ticianelli, Surface and Coatings Technology, 197, 215 (2005).

21. D. Barsellini, A. Visintin, W. E. Triaca, and M. P. Soriaga, Journal of Power Sources, 124, 309 (2003).

22. A. Visintin, C. A. Tori, G. Garaventta, and W. E. Triaca, Journal of the Electrochemical Society, 145, 4169 (1998).

23. G. Zheng, B. N. Popov, and R. E. White, Journal of Applied Electrochemistry, 28, 381 (1998).

24. J.-W. Lee and S.-I. Pyun, Electrochimica Acta, 50, 1777 (2005).

25. T. Ma, Y. Hatano, T. Abe, and K. Watanabe, Toyama Daigaku Suiso Doitai Kagaku Kenkyu Senta Kenkyu Hokoku, 22, 35 (2002).

226

26. T. Ma, Y. Hatano, T. Abe, and K. Watanabe, Journal of Alloys and Compounds, 372, 251 (2004).

27. V. Badri and A. M. Hermann, International Journal of Hydrogen Energy, 25, 249 (1999).

28. D. B. Willey, I. R. Harris, and A. S. Pratt, Journal of Alloys and Compounds, 293- 295, 613 (1999).

29. J. Han, F. Feng, M. Geng, R. Buxbaum, and D. O. Northwood, Journal of Power Sources, 80, 39 (1999).

30. D. Smith, A. Visintin, W. Zhang, S. Srinivasan, A. Anani, and H. Wu, Proceedings - Electrochemical Society, 97-13, 247 (1997).

31. A. Visintin, C. A. Tori, G. Garaventta, and W. E. Triaca, Proceedings - Electrochemical Society, 97-18, 660 (1997).

32. A. Visintin, C. A. Tori, G. Garaventta, and W. E. Triaca, Journal of the Brazilian Chemical Society, 8, 125 (1997).

33. M. Geng, Journal of Alloys and Compounds, 217, 90 (1995).

34. X. Shan, J. S. Wainright, and J. H. Payer, Journal of Alloys and Compounds, in press.

35. Y. Nakamura, K. Sato, S. Fujitani, K. Nishio, K. Oguro, and I. Uehara, Journal of Alloys and Compounds, 267, 205 (1998).

36. X. Shan, J. S. Wainright, and J. H. Payer, Journal of Alloys and Compounds, in press.

37. G. D. Adzic, J. R. Johnson, J. J. Reilly, J. McBreen, S. Murkerjee, M. P. S. Kumar, W. Zhang, and S. Srinivasan, Journal of the Electrochemical Society, 142, 3429 (1995).

38. X. Shan, Ph.D. dissertation, Material Science Dept., Case Western Reserve University, Cleveland, Ohio., 2004.

39. J. W. Oh, C. Y. Kim, K. S. Nahm, and K. S. Sim, Journal of Alloys and Compounds, 278, 270 (1998).

227

8. Conclusions and Recommendations for Future Work

8.1. Conclusions

The main objective of this work was to develop batteries with long cycle life and safety for implantable applications. Lithium ion batteries, although limited by cycle life, were chosen as the short term option for powering networked neuroprosthetic system

(NNPS). The potential of the low pressure nickel-hydrogen batteries using metal hydride for hydrogen storage was identified as having an order higher magnitude cycle life in comparison to the lithium ion batteries.

Implantable grade lithium ion cells from Quallion and Wilson Greatbatch (WG) were tested and compared to Sony Li ion cells (chapter 2). The conditions for maximizing cycle life were determined: lower charge and discharge rates, lower charge voltage cutoff, lower depth of discharge and lower temperature. The ‘discharge capacity’ and the ‘end of discharge voltage’ data were modeled to predict the cycle life and to compare the different lithium ion cells. Based on the fit parameters, it was concluded that the cycle life of Sony cells was better than that of the Quallion and WG cells. Of the implantable batteries, the Quallion cells outperformed the WG cells and were chosen as the short term powering option for NNPS. The cycle life of the Quallion cells is about

6000 cycles at C rate charge, C rate discharge and 60% depth of discharge.

A low pressure nickel-hydrogen battery was designed and developed in this work

(chapters 3, 4, 5, 6 and 7). The main components of this battery are the positive nickel hydroxide electrode, the negative platinum electrode, the separator, the current collector

228

and the metal hydride. A platinum catalyzed electrode from E-tek Inc. was used as the negative electrode. The charge transfer resistance of this electrode was decreased using cyclic voltammetry (chapter 3). Nickel mesh was chosen as the current collector because of its low resistance and stability in alkaline solutions (chapter 3). Two separators, zirconium oxide stabilized by yttria and polypropylene were tested and no significant difference was observed (chapter 3).

Nickel hydroxide electrode extracted from a ‘D size’ nickel-metal hydride battery was tested in liquid and nickel-hydrogen cell configurations. The electrode contained active material on both sides. The problems associated with using it in a series cell were discussed in chapter 4. This electrode served as a guide in the development of the pasted nickel hydroxide electrode (chapter 5) and the design of the nickel-hydrogen cell.

The positive nickel hydroxide electrode was fabricated by impregnating a paste using either screen printing or spatula pressing (chapter 5). Higher capacity and higher utilization were obtained in the electrodes fabricated by spatula pressing in comparison to that of screen printing. The mechanism of nickel hydroxide electrode formation was studied and discussed in detail in chapter 5. The effect of different formation rates with and without overcharge and the effect of PVDF and nickel content on the electrode performance were studied using electrochemical voltage spectroscopy and impedance spectroscopy. The electrode utilization increases with PVDF content. The electrode performance was enhanced by the addition of nickel, which increases the utilization by

10% and lowers the oxygen evolution.

A low pressure nickel-hydrogen battery without metal hydride was assembled successfully and the results are discussed in chapter 6. The decrease in the coulometric

229

efficiency due to oxygen evolution and self discharge was investigated. The recombination of the evolved oxygen with hydrogen was studied. Pressure data analysis shows that 70% of the recombination is complete during the charge and the remaining during the 1st h of the rest period following the charge. Pressure variations due to oxygen evolution, the recombination reaction and self discharge were completely understood.

Coulometric efficiency calculated based on the pressure data analysis agreed well with the measured coulometric efficiency. It was shown that the state of charge of the cell can be estimated by monitoring the cell pressure. The cycle life of the nickel-hydrogen cells was limited because of significant oxygen evolution and material extrusion from the nickel hydroxide electrode.

A low pressure nickel-hydrogen battery with metal hydride (MH) was successfully developed (chapter 7). The metal hydride, palladium treated LaNi4.7Al0.3, was stable in humid air. However, it deactivates on soaking in KOH solution and therefore was physically separated from the electrodes. In spite of this separation, the metal hydride deactivated with time resulting in increased hydrogen equilibration time with the MH. The time constant for this equilibration increased from 0.1h to about 1h. It was shown that the pressure variation during charge and discharge of the cell followed the desorption curve of the pressure composition isotherm. The pressure variation can be used to estimate the state of charge and to fuel gauge the cell after accounting for contributions due to oxygen evolution, recombination, and self discharge. About 40-45% hydrogen from the metal hydride was used as fuel during the cycling of the cell.

230

8.2. Recommendations for Future Work

In the future, there are three major issues that need to be investigated:

1. The cycle life of the nickel-hydrogen cells needs to be improved. This can be

achieved by preventing flooding at the back of the platinum electrode thereby

facilitating recombination. Heat is produced as a result of the recombination

reaction at the platinum electrode. Therefore future nickel hydrogen cells

should be designed to facilitate heat transfer from the cell.

2. The metal hydride, though physically separated from the KOH solution, was

deactivated due to KOH solution seeping along the nylon bolt used to

assemble the cell. In future, cells should be designed to prevent migration of

KOH solution to the metal hydride.

3. After solving the cycle life issues, a series cell can be developed using a

bipolar plate. The nickel hydroxide electrode fabricated in chapter 5 contains

active material on only one side and can be used in the series cell

development.

231

9. Bibliography

Abraham, K. M. and S. B. Brummer (1983). Lithium Batteries. J. P. Gabano. London, Academic Press Inc.: 371-406.

Adzic, G. D., J. R. Johnson, et al. (1995). "Cerium content and cycle life of multicomponent AB5 hydride electrodes." Journal of the Electrochemical Society 142(10): 3429-33.

Alcaide, F., E. Brillas, et al. (2005). "Hydrogen Oxidation Reaction in a Pt-Catalyzed Gas Diffusion Electrode in Alkaline Medium." Journal of the Electrochemical Society 152(10): E319-E327.

Ambrosio, R. C. and E. A. Ticianelli (2005). "Studies on the influence of palladium coatings on the electrochemical and structural properties of a metal hydride alloy." Surface and Coatings Technology 197(2-3): 215-222.

Anani, A., A. Visintin, et al. (1994). "Alloys for hydrogen storage in nickel/hydrogen and nickel/metal hydride batteries." Journal of Power Sources 47(3): 261-75.

Angely, L. and G. Bronoel (1980). "Effect of pH in basic medium on the kinetics of oxidation of hydrogen at a platinum electrode." Electrochimica Acta 25(12): 1541-5.

Appleby, A. J. (1990). "From Sir William Grove to today: fuel cells and the future." Journal of Power Sources 29(1-2): 3-11.

Armstrong, R. D., G. W. D. Briggs, et al. (1988). "Some effects of the addition of cobalt to the nickel hydroxide electrode." Journal of Applied Electrochemistry 18(2): 215-19.

Arora, P., M. Doyle, et al. (1999). "Mathematical modeling of capacity fade mechanisms in lithium-ion batteries: lithium deposition overcharge reaction on the negative electrode." Proceedings - Electrochemical Society 98-16(Lithium Batteries): 553- 572.

Arora, P., M. Doyle, et al. (1999). "Mathematical modeling of the lithium deposition overcharge reaction in lithium-ion batteries using carbon-based negative electrodes." Journal of the Electrochemical Society 146(10): 3543-3553.

Arora, P., R. E. White, et al. (1998). "Capacity fade mechanisms and side reactions in lithium-ion batteries." Journal of the Electrochemical Society 145(10): 3647- 3667.

Aurbach, D., Y. Eineli, et al. (1994). "The Correlation between the Surface-Chemistry and the Performance of Li-Carbon Intercalation Anodes for Rechargeable

232

Rocking-Chair Type Batteries." Journal of the Electrochemical Society 141(3): 603-611.

Aurbach, D., Y. Eineli, et al. (1995). "The Study of Electrolyte-Solutions Based on Ethylene and Diethyl Carbonates for Rechargeable Li Batteries .2. Graphite- Electrodes." Journal of the Electrochemical Society 142(9): 2882-2890.

Aurbach, D., B. Markovsky, et al. (1999). "New insights into the interactions between electrode materials and electrolyte solutions for advanced nonaqueous batteries." Journal of Power Sources 82: 95-111.

Aurbach, D., E. Zinigrad, et al. (2000). "Factors which limit the cycle life of rechargeable lithium (metal) batteries." Journal of the Electrochemical Society 147(4): 1274- 1279.

Badri, V. and A. M. Hermann (1999). "Metal hydride batteries: Pd nanotube incorporation into the negative electrode." International Journal of Hydrogen Energy 25(3): 249-252.

Baker, B. S. and M. G. Klein (1975). Hydrophilic electrode. Application: US 3898099, (Energy Research Corp., USA).

Bard, A. J. and L. R. Faulkner (1980). Electrochemical Methods: Fundamentals and Applications, John Wiley & Sons, Inc.

Barsellini, D., A. Visintin, et al. (2003). "Electrochemical characterization of a hydride- forming metal alloy surface-modified with palladium." Journal of Power Sources 124(1): 309-313.

Basu, R. N., S. K. Pratihar, et al. (1997). "Preparation of Sr-substituted LaMnO3 thick films as cathode for solid oxide fuel cell." Materials Letters 32(4): 217-222.

Beauchamp, R. L. (1972). Positive electrode for use in nickel cadmium cells. Application: US 3653967, (Bell Telephone Laboratories, Inc.).

Belt, J. R., C. D. Ho, et al. (2003). "A capacity and power fade study of Li-ion cells during life cycle testing." Journal of Power Sources 123(2): 241-246.

Bittner, H. F. and C. C. Badcock (1983). "Electrochemical utilization of metal hydrides." Journal of the Electrochemical Society 130(5): 193C-198C.

Bloom, I., B. W. Cole, et al. (2001). "An accelerated calendar and cycle life study of Li- ion cells." Journal of Power Sources 101(2): 238-247.

Bloom, I., S. A. Jones, et al. (2003). "Effect of cathode composition on capacity fade, impedance rise and power fade in high-power, lithium-ion cells." Journal of Power Sources 124(2): 538-550.

233

Bode, H., K. Dehmelt, et al. (1966). "Nickel hydroxide electrodes. I. Nickel(II) hydroxide hydrate." Electrochimica Acta 11(8): 1079-87.

Bronoel, G. and J. Reby (1980). "Mechanism of oxygen evolution in basic medium at a nickel electrode." Electrochimica Acta 25(7): 973-6.

Broussely, M. (2002). "Aging mechanisms and calendar-life predictions in lithium-ion batteries." Advances in Lithium-Ion Batteries: 393-432.

Broussely, M. and G. Archdale (2004). "Li-ion batteries and portable power source prospects for the next 5-10 years." Journal of Power Sources 136(2): 386-394.

Broussely, M., P. Biensan, et al. (2005). "Main aging mechanisms in Li-ion batteries." Journal of Power Sources 146(1-2): 90-96.

Broussely, M., S. Herreyre, et al. (2001). "Aging mechanism in Li ion cells and calendar life predictions." Journal of Power Sources 97-98: 13-21.

Broussely, M., S. Herreyre, et al. (2003). "Aging mechanisms in Li-ion batteries." Proceedings - Electrochemical Society 2001-21(Batteries and Supercapacitors): 75-84.

Butel, M., L. Gautier, et al. (1999). "Cobalt oxyhydroxides obtained by 'chimie douce' reactions: structure and electronic conductivity properties." Solid State Ionics 122(1-4): 271-284.

Chusid, O., Y. E. Ely, et al. (1993). "Electrochemical and Spectroscopic Studies of Carbon Electrodes in Lithium Battery Electrolyte Systems." Journal of Power Sources 43(1-3): 47-64.

Couturier, G., D. W. Kirk, et al. (1987). "Electrocatalysis of the hydrogen oxidation and of the oxygen reduction reactions on platinum and some alloys in alkaline medium." Electrochimica Acta 32(7): 995-1005.

Dahn, J. R., A. K. Sleigh, et al. (1994). Lithium Batteries: New Materials, Developments and Perspectives. G. Pistoia. Amsterdam, Elsevier: 1.

Dantzer, P. (1997). "Metal-hydride technology: A critical review." Topics in Applied Physics 73(Hydrogen in Metals III): 279-340.

Dhar, S. K., S. R. Ovshinsky, et al. (1997). "Nickel/metal hydride technology for consumer and electric vehicle batteries - a review and up-date." Journal of Power Sources 65(1-2): 1-7.

Do, J.-S., S.-H. Yu, et al. (2003). "Thick-film nickel-metal-hydride battery based on porous ceramic substrates." Journal of Power Sources 117(1-2): 203-211.

234

Dunlop, J. D., M. W. Earl, et al. (1976). Low pressure nickel hydrogen cell. Application: US 3959018, (Communications Satellite Corp., USA).

Dunlop, J. D., G. M. Rao, et al. (1993). NASA handbook for nickel-hydrogen batteries. Washington, DC, National Aeronautics and Space Administration, Scientific and Technical Information Branch.

Eineli, Y., B. Markovsky, et al. (1994). "The Dependence of the Performance of Li-C Intercalation Anodes for Li-Ion Secondary Batteries on the Electrolyte Solution Composition." Electrochimica Acta 39(17): 2559-2569.

Ernst, S. and C. H. Hamann (1975). "pH-Dependence of the hydrogen exchange current density at smooth platinum in alkaline solution (potassium hydroxide)." Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 60(1): 97-100.

Ezhov, B. B. and O. G. Malandin (1991). "Structure modification and change of electrochemical activity of nickel hydroxides." Journal of the Electrochemical Society 138(4): 885-9.

Feng, F., M. Geng, et al. (2001). "Electrochemical behavior of intermetallic-based metal hydrides used in Ni/metal hydride (MH) batteries: a review." International Journal of Hydrogen Energy 26(7): 725-734.

Fetcenko, M. A., S. Venkatesan, et al. (1992). "Selection of metal hydride alloys for electrochemical applications." Proceedings - Electrochemical Society 92-5(Proc. Symp. Hydrogen Storage Mater., Batteries, Electrochem., 1991): 141-67.

Fleischer, A. (1948). "Sintered plates for nickel-cadmium batteries." Journal of the Electrochemical Society 94: 289-99.

Fritts, D. H. (1982). "The mechanics of electrochemically coprecipitated cobalt hydroxide in nickel hydroxide electrodes." Journal of the Electrochemical Society 129(1): 118-22.

Geng, M. (1995). "Electrochemical characteristics of Ni-Pd-coated MmNi5-based alloy powder for nickel-metal hydride batteries." Journal of Alloys and Compounds 217(1): 90-3.

Ghiurcan, G. A., C.-C. Liu, et al. (2003). "Development and Characterization of a Thick- Film Printed Zinc-Alkaline Battery." Journal of the Electrochemical Society 150(7): A922-A927.

Golben, M., K. Nechev, et al. (1997). "A low pressure bipolar nickel-hydrogen battery." Annual Battery Conference on Applications and Advances, 12th, Long Beach, Calif., Jan. 14-17, 1997: 307-312.

235

Gulzow, E., M. Schulze, et al. (2002). "Investigation of the degradation of different nickel anode types for alkaline fuel cells (AFCs)." Journal of Power Sources 106(1-2): 126-135.

Han, J., F. Feng, et al. (1999). "Electrochemical characteristics of the interface between the metal hydride electrode and electrolyte for an advanced nickel/metal hydride battery." Journal of Power Sources 80(1-2): 39-45.

Hart, N. T., N. P. Brandon, et al. (2001). "Functionally graded cathodes for solid oxide fuel cells." Journal of Materials Science 36(5): 1077-1085.

Haugh, M. J. (2003). Cycle life, state of charge identification, and failure detection in lithium ion batteries. M.S. thesis, Dept. of Electrical Engineering and Computer Science. Cleveland, Ohio., Case Western Reserve University.

Hobson, L. J., Y. Nakano, et al. (2002). "Targeting improved DMFC performance." Journal of Power Sources 104(1): 79-84.

Hong, K. (2001). "The development of hydrogen storage electrode alloys for nickel hydride batteries." Journal of Power Sources 96(1): 85-89.

Hrovat, M., J. Holc, et al. (1994). "Thick film ruthenium oxide/yttria-stabilized zirconia- based cathode material for solid oxide fuel cells." Solid State Ionics 68(1-2): 99- 103.

Hrovat, M., N. Katsarakis, et al. (1996). "Characterization of LaNi1-xCoxO3 as a possible SOFC cathode material." Solid State Ionics 83(1,2): 99-105.

Inoue, T., T. Sasaki, et al. (2002). "Calendar and cycle life prediction of 100 Ah lithium- ion cells for space applications." NASA Conference Publication 2002- 211466(NASA Aerospace Battery Workshop, 2001): 452-459.

Iwakura, C. and M. Matsuoka (1991). "Application of hydrogen storage alloys to battery- related fields: nickel-hydrogen batteries." Progress in Batteries & Solar Cells 10: 81-114.

Jayashree, R. S. and P. V. Kamath (2002). "Modified nickel hydroxide electrodes effect of cobalt metal on the different polymorphic modifications." Journal of the Electrochemical Society 149(6): A761-A764.

Kandler, L. (1965). Cathode for an alkali chloride electrolysis cell. Application: DE 1207358.

Kiros, Y. and S. Schwartz (2000). "Long-term hydrogen oxidation catalysts in alkaline fuel cells." Journal of Power Sources 87(1-2): 101-105.

Kishi, T., H. Ohta, et al. (2003). "Development and pack cycle simulation of high power Li-ion batteries." Electrochemistry (Tokyo, Japan) 71(12): 1090-1092.

236

Kleperis, J., G. Wojcik, et al. (2001). "Electrochemical behavior of metal hydrides." Journal of Solid State Electrochemistry 5(4): 229-249.

Law, H. H. and J. Sapjeta (1989). "Effect of cobalt on fibrous nickel hydroxide electrodes." Journal of the Electrochemical Society 136(6): 1603-6.

Leblanc, P., C. Jordy, et al. (1998). "Mechanism of alloy corrosion and consequences on sealed nickel-metal hydride battery performance." Journal of the Electrochemical Society 145(3): 860-863.

Lee, J.-W. and S.-I. Pyun (2005). "Anomalous behaviour of hydrogen extraction from hydride-forming metals and alloys under impermeable boundary conditions." Electrochimica Acta 50(9): 1777-1805.

Liaw, B. Y., R. G. Jungst, et al. (2005). "Modeling capacity fade in lithium-ion cells." Journal of Power Sources 140(1): 157-161.

Liaw, B. Y., E. P. Roth, et al. (2003). "Correlation of Arrhenius behaviors in power and capacity fades with cell impedance and heat generation in cylindrical lithium-ion cells." Journal of Power Sources 119-121: 874-886.

Lichtenberg, F. and K. Kleinsorgen (1996). "Stability enhancement of the CoOOH conductive network of nickel hydroxide electrodes." Journal of Power Sources 62(2): 207-211.

Lim, H. S. and R. E. Doty (1994). "Migration of cobalt in nickel oxide/hydroxide active material of a nickel electrode in a Ni/H2 cell." Journal of Power Sources 50(1-2): 57-66.

Linden, D. and T. B. Reddy, Eds. (2002). Handbook of batteries. New York, McGraw- Hill.

Ma, T., Y. Hatano, et al. (2002). "Effects of Pd addition on the electrochemical properties of MgNi." Toyama Daigaku Suiso Doitai Kagaku Kenkyu Senta Kenkyu Hokoku 22: 35-46.

Ma, T., Y. Hatano, et al. (2004). "Effects of Pd addition on electrochemical properties of MgNi." Journal of Alloys and Compounds 372(1-2): 251-258.

Ma, T., Y. Hatano, et al. (2005). "Effects of bulk modification by Pd on electrochemical properties of MgNi." Journal of Alloys and Compounds 391(1-2): 313-317.

Madou, M. J. (1994). Fundamentals of Microfabrication: The science of Miniaturization. Boca Raton, CRC Press LLC.

Markovic, N. M. and P. N. Ross (2002). "Surface science studies of model fuel cell electrocatalysts." Surface Science Reports 45(4-6): 117-229.

237

Matsuzaki, Y. and I. Yasuda (2002). "Electrochemical properties of reduced-temperature SOFCs with mixed ionic-electronic conductors in electrodes and/or interlayers." Solid State Ionics 152-153: 463-468.

McBreen, J., W. E. O'Grady, et al. (1989). "In situ time-resolved x-ray absorption near edge structure study of the nickel oxide electrode." Journal of Physical Chemistry 93(17): 6308-11.

Morata, A., A. Tarancon, et al. (2004). "Optimized screen-printing and SEM-FIB characterization of YSZ thin films for solid oxide fuel cells and gas sensors." Materials Research Society Symposium Proceedings 822(Nanostructured Materials in Alternative Energy Devices): 109-114.

Muller, A. C., D. Herbstritt, et al. (2002). "Development of a multilayer anode for solid oxide fuel cells." Solid State Ionics 152-153: 537-542.

Mund, K., G. Richter, et al. (1979). "Electrocatalysts for the reactions in alkaline fuel cells." Proceedings - Electrochemical Society 79-2(Proc. Workshop Electrocatal. Fuel Cell React., 1978): 47-66.

Nagaura, T. and K. Tozawa (1990). "Lithium ion rechargeable battery." Progress in Batteries & Solar Cells 9: 209-17.

Nakamura, Y., K. Sato, et al. (1998). "Lattice expanding behavior and degradation of LaNi5-based alloys." Journal of Alloys and Compounds 267(1-2): 205-210.

Ning, G. and B. N. Popov (2004). "Cycle Life Modeling of Lithium-Ion Batteries." Journal of the Electrochemical Society 151(10): A1584-A1591.

Oh, J. W., C. Y. Kim, et al. (1998). "The hydriding kinetics of LaNi4.5Al0.5 with hydrogen." Journal of Alloys and Compounds 278(1-2): 270-276.

Osaka, T., S. Nakade, et al. (2003). "Influence of capacity fading on commercial lithium- ion battery impedance." Journal of Power Sources 119-121: 929-933.

Oshitani, M., Y. Sasaki, et al. (1984). "Development of a nickel electrode having stable performance at various charge and discharge rates over a wide temperature range." Journal of Power Sources 12(3-4): 219-31.

Oshitani, M., M. Watada, et al. (1994). "Development of a high-energy-density cadmium-free nickel metal hydride battery." Proceedings - Electrochemical Society 94-27(Hydrogen and Metal Hydride Batteries): 303-28.

Oshitani, M., H. Yufu, et al. (1989). "Development of a pasted nickel electrode with high active material utilization." Journal of the Electrochemical Society 136(6): 1590- 3.

238

Ovshinsky, S. R., M. A. Fetcenko, et al. (1993). "A nickel metal hydride battery for electric vehicles." Science (Washington, DC, United States) 260(5105): 176-81.

Park, H. J., N. H. Goo, et al. (2003). "In Situ Pd Deposition on Mg2Ni Electrodes for Ni/MH Secondary Batteries during Charge Cycles." Journal of the Electrochemical Society 150(10): A1328-A1332.

Peled, E. (1979). "The electrochemical behavior of alkali and alkaline earth metals in nonaqueous battery systems - the solid electrolyte interphase model." J. Electrochem. Soc. 126(12): 2047-51.

Peng, R., C. Xia, et al. (2002). "Intermediate-temperature SOFCs with thin Ce0.8Y0.2O1.9 films prepared by screen-printing." Solid State Ionics 152-153: 561-565.

Petrii, O. A., S. Y. Vasina, et al. (1996). "Electrochemistry of hydride-forming intermetallic compounds and alloys." Uspekhi Khimii 65(3): 195-210.

Pickett, D. F. (1974). Nickel electrodes. Application: US 3827911, (United States Dept. of the Air Force).

Pickett, D. F. and J. T. Maloy (1978). "Microelectrode studies of electrochemically coprecipitated cobalt hydroxide in nickel hydroxide electrodes." Journal of the Electrochemical Society 125(7): 1026-32.

Pozio, A., L. Giorgi, et al. (2000). "Electrooxidation of H2 on Pt/C Pt-Ru/C and Pt-Mo/C anodes for polymer electrolyte fuel cell." Electrochimica Acta 46(4): 555-561.

Pralong, V., Y. Chabre, et al. (2002). "Study of the contribution of cobalt additive to the behavior of the nickel oxy-hydroxide electrode by potentiodynamic techniques." Solid State Ionics 147(1,2): 73-84.

Pralong, V., A. Delahaye-Vidal, et al. (2000). "Electrochemical behavior of cobalt hydroxide used as additive in the nickel hydroxide electrode." Journal of the Electrochemical Society 147(4): 1306-1313.

Ramadass, P., B. Haran, et al. (2004). "Development of First Principles Capacity Fade Model for Li-Ion Cells." Journal of the Electrochemical Society 151(2): A196- A203.

Ramadass, P., B. Haran, et al. (2002). "Capacity fade of Sony 18650 cells cycled at elevated temperatures. Part I. Cycling performance." Journal of Power Sources 112(2): 606-613.

Ramadass, P., B. Haran, et al. (2002). "Capacity fade of Sony 18650 cells cycled at elevated temperatures. Part II. Capacity fade analysis." Journal of Power Sources 112(2): 614-620.

239

Ramadass, P., B. Haran, et al. (2003). "Mathematical modeling of the capacity fade of Li- ion cells." Journal of Power Sources 123(2): 230-240.

Ross, P. N., K. Kinoshita, et al. (1975). "Electrocatalysis on binary alloys. I. Oxidation of molecular hydrogen on supported platinum-rhodium alloys." Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 59(2): 177-89.

Ruetschi, P., F. Meli, et al. (1995). "Nickel-metal hydride batteries. The preferred batteries of the future?" Journal of Power Sources 57(1-2): 85-91.

Sakai, T., K. Muta, et al. (1992). "Nickel-metal hydride batteries using rare-earth based hydrogen storage alloys." Proceedings - Electrochemical Society 92-5(Proc. Symp. Hydrogen Storage Mater., Batteries, Electrochem., 1991): 59-91.

Sarre, G., P. Blanchard, et al. (2004). "Aging of lithium-ion batteries." Journal of Power Sources 127(1-2): 65-71.

Schmidt, T. J., P. N. Ross, et al. (2002). "Temperature dependent surface electrochemistry on Pt single crystals in alkaline electrolytes Part 2. The hydrogen evolution/oxidation reaction." Journal of Electroanalytical Chemistry 524-525: 252-260.

Shan, X. (2004). Hydrogen storage for micro-fabricated electrochemical devices. Ph.D. dissertation, Material Science Dept. Cleveland, Ohio., Case Western Reserve University.

Shan, X., J. S. Wainright, et al. "Improved Durability of Hydrogen Storage Alloys." Journal of Alloys and Compounds, in press.

Shan, X., J. S. Wainright, et al. "Increased performance of hydrogen storage by Pd- treated LaNi4.7Al0.3, CaNi5 and Mg2Ni." Journal of Alloys and Compounds, in press.

Shim, J. and H. K. Lee (2001). "Improved performance of Raney nickel electrode by the addition of electrically conductive materials for hydrogen oxidation reaction." Materials Chemistry and Physics 69(1-3): 72-76.

Sikha, G., B. N. Popov, et al. (2004). "Effect of porosity on the capacity fade of a lithium-ion battery. Theory." Journal of the Electrochemical Society 151(7): A1104-A1114.

Smith, D., A. Visintin, et al. (1997). "Enhancement of rate capability of a MHx electrode by Pd coating of active material." Proceedings - Electrochemical Society 97- 13(Electrode Materials and Processes for Energy Conversion and Storage IV): 247-256.

Spotnitz, R. (2003). "Simulation of capacity fade in lithium-ion batteries." Journal of Power Sources 113(1): 72-80.

240

Srdic, V. V., R. P. Omorjan, et al. (2005). "Electrochemical performance of (La,Sr)CoO3 cathode for zirconia-based solid oxide fuel cells." Materials Science & Engineering, B: Solid-State Materials for Advanced Technology B116(2): 119- 124.

Stamps, A. T., C. E. Holland, et al. (2005). "Analysis of capacity fade in a lithium ion battery." Journal of Power Sources 150: 229-239.

Strasser, K. (1990). "The design of alkaline fuel cells." Journal of Power Sources 29(1-2): 149-66.

Tarascon, J. M. and M. Armand (2001). "Issues and challenges facing rechargeable lithium batteries." Nature 414(6861): 359-367.

Taucher-Mautner, W. and K. Kordesch (2004). "Studies of pasted nickel electrodes to improve cylindrical nickel-zinc cells." Journal of Power Sources 132(1-2): 275- 281.

Tessier, C., C. Faure, et al. (2002). "Electrochemical Study of Zinc-Substituted Nickel Hydroxide." Journal of the Electrochemical Society 149(9): A1136-A1145.

Thaller, L. H. and A. H. Zimmerman (2003). Nickel-hydrogen life cycle testing: review and analysis. El Segundo, California, Aerospace Press.

Thaller, L. H. and A. H. Zimmerman (2003). Overview of the design, development, and application of nickel-hydrogen batteries. Cleveland, Ohio, National Aeronautics and Space Administration, Glenn Research Center.

Thaller, L. H., A. H. Zimmerman, et al. (2003). "Flooded utilization and electrochemical voltage spectroscopy studies on nickel electrodes." Journal of Power Sources 122(1): 85-94.

Thaller, L. H., A. H. Zimmerman, et al. (2003). "Techniques to improve the usability of nickel-hydrogen cells." Journal of Power Sources 114(2): 309-319.

Unates, M. E., M. E. Folquer, et al. (1992). "The influence of foreign cations on the electrochemical behavior of the nickel hydroxide electrode." Journal of the Electrochemical Society 139(10): 2697-704.

Vermeijlen, J. J. T. t., L. J. J. Janssen, et al. (1997). "Mechanism of hydrogen oxidation on a platinum-loaded gas diffusion electrode." Journal of Applied Electrochemistry 27(5): 497-506.

Visintin, A., C. A. Tori, et al. (1997). "The effect of palladium coating on hydrogen storage alloy electrodes for nickel/metal hydride batteries." Journal of the Brazilian Chemical Society 8(2): 125-129.

241

Visintin, A., C. A. Tori, et al. (1997). "Electrochemical characterization of Pd-coated metal hydride electrodes for battery applications." Proceedings - Electrochemical Society 97-18(Batteries for Portable Applications and Electric Vehicles): 660- 668.

Visintin, A., C. A. Tori, et al. (1998). "The electrochemical performance of Pd-coated metal hydride electrodes with different binding additives in alkaline solution." Journal of the Electrochemical Society 145(12): 4169-4172.

Wainright, J. S., R. F. Savinell, et al. (2003). "Microfabricated fuel cells." Electrochimica Acta 48(20-22): 2869-2877.

Wang, X., J. Yan, et al. (1998). "Surface modification and electrochemical studies of spherical nickel hydroxide." Journal of Power Sources 72(2): 221-225.

Watanabe, K., M. Koseki, et al. (1996). "Effect of cobalt addition to nickel hydroxide as a positive material for rechargeable alkaline batteries." Journal of Power Sources 58(1): 23-28.

Wendt, H. (1995). "New electrocatalysts for fuel cells." New Materials for Fuel Cell Systems I, Proceedings of the International Symposium on New Materials for Fuel Cell Systems, 1st, Montreal, July 9-13, 1995: 532-43.

Willems, J. J. G. and K. H. J. Buschow (1987). "From permanent magnets to rechargeable hydride electrodes." Journal of the Less-Common Metals 129: 13- 30.

Willey, D. B., I. R. Harris, et al. (1999). "The improvement of the hydrogenation properties of nickel-metal hydride battery alloy by surface modification with platinum group metals (PGMs)." Journal of Alloys and Compounds 293-295: 613-620.

Wojcik, G. and M. Kopczyk (1995). "Multi-component metallic systems absorbing hydrogen for rechargeable batteries." Wiadomosci Chemiczne 49(5-6): 285-99.

Wright, R. B., J. P. Christophersen, et al. (2003). "Power fade and capacity fade resulting from cycle-life testing of Advanced Technology Development Program lithium- ion batteries." Journal of Power Sources 119-121: 865-869.

Wright, R. B., C. G. Motloch, et al. (2002). "Calendar- and cycle-life studies of advanced technology development program generation 1 lithium-ion batteries." Journal of Power Sources 110(2): 445-470.

Wu, M. S., C. M. Huang, et al. (1999). "Effects of surface modification of nickel hydroxide powder on the electrode performance of nickel/metal hydride batteries." Electrochimica Acta 44(23): 4007-4016.

242

Yang, Q. M., V. A. Ettel, et al. (2003). "Pasted Ni(OH)2 Electrodes Using Ni Powders for High-Drain-Rate, Ni-Based Batteries." Journal of the Electrochemical Society 150(4): A543-A550.

Zhang, D., B. S. Haran, et al. (2000). "Studies on capacity fade of lithium-ion batteries." Journal of Power Sources 91(2): 122-129.

Zhao, Y., Q. Zhou, et al. (2006). "A novel and facile route of ink-jet printing to thin film SnO2 anode for rechargeable lithium ion batteries." Electrochimica Acta 51(13): 2639-2645.

Zheng, G., B. N. Popov, et al. (1998). "Electrochemical investigations of bare and Pd- coated LaNi4.25Al0.75 electrodes in alkaline solution." Journal of Applied Electrochemistry 28(4): 381-385.

Zhou, Z., J. Yan, et al. (1998). "Changes in the properties and structure of hydrogen- storage electrodes after long-term charge/discharge cycling." Journal of Power Sources 72(2): 236-238.

Zimmerman, A. H. (1994). "Introduction to the nickel hydroxide electrode." Proceedings - Electrochemical Society 94-27(Hydrogen and Metal Hydride Batteries): 268-83.

243