A Sensor-Less Method for Detecting the Methanol Concentration in an Open-Cathode Direct Methanol Fuel Cell

MISSING IMAGE

Material Information

Title:
A Sensor-Less Method for Detecting the Methanol Concentration in an Open-Cathode Direct Methanol Fuel Cell
Physical Description:
1 online resource (93 p.)
Language:
english
Creator:
Shoraka, Amir R
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Mechanical Engineering, Mechanical and Aerospace Engineering
Committee Chair:
Lear, William E
Committee Members:
Crisalle, Oscar D
Fletcher, James

Subjects

Subjects / Keywords:
concentration -- dmfc
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre:
Mechanical Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Methanol concentration in the solution fuel playsan important role in maintaining the power output of a direct methanol fuelcell (DMFC) system. Hence, in most of the DMFC systems, in order to monitor andcontrol the methanol concentration a concentration sensing method is used. Inthis research, a sensor-less methanol concentration sensing method for anopen-cathode direct DMFC is discussed.  In this method, the methanol concentration isdetected by measuring the current density of the stack during the air-breakperiod. Air-break is a common strategy to recover the temporary cathodeperformance loss during which the cathode air delivery is stopped and the stackoperates at a constant voltage. In comparison with a methanol concentration sensoran average difference of 0.068 M was reported over an operating temperaturerange of 45-55 ºCand methanol concentration range of 0.3-1.5 M. Also, the effect of the stacklife time, and the set voltage during the air-break was studied. One of thegreatest features of this method is that it does not require any auxiliarydevice and since it is using the readily available data during the air-breakperiod it does not require any change in the normal load profile of the system.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Amir R Shoraka.
Thesis:
Thesis (M.S.)--University of Florida, 2012.
Local:
Adviser: Lear, William E.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-08-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044723:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 A SENSOR LESS METHOD FOR DETECTING THE METHANOL CONCENTRATION IN AN OPEN CATHODE DIRECT METHANOL FUEL CELL By AMIR REZA SHORAKA A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Amir Reza Shoraka

PAGE 3

3 To my family and fri ends for their love and support

PAGE 4

4 ACKNOWLEDGMENTS I would like to take this opportunity to express my thankful gratitude to my committee, Dr. William Lear, Dr. James Fletcher, and Dr. Oscar Crisalle for all their guidance, support and enco uragement. I owe my deepest gratitude to Dr. Joseph Campbell, Dr. Philip Cox, and my dear friend and col league Jason Harrington. Without their help and advice this thesis would not have been possible. Also, I would like to thank all my friends and collea gues in the University of North Florida Fuel Cell Laboratory.

PAGE 5

5 TABLE OF CONTENT S page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 ABSTRAC T ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ...... 13 Background ................................ ................................ ................................ ............... 14 DMFC Performance ................................ ................................ ............................... 14 Polarization Curve and Performance Losses ................................ ......................... 14 Membrane Electrode Assembly (MEA) ................................ ................................ .. 15 DMFC systems ................................ ................................ ................................ ....... 17 Motivations and Objectives ................................ ................................ ........................ 17 Motivations ................................ ................................ ................................ ............. 17 Objectives ................................ ................................ ................................ .............. 19 2 LITERATURE REV IEW ................................ ................................ ............................ 25 Electrochemical Sensing Methods ................................ ................................ ............. 25 Physical Sensing Methods ................................ ................................ ......................... 28 Sens or less Methods ................................ ................................ ................................ 31 3 EXPERIMENTAL METHODO LOGY ................................ ................................ ........ 34 Recovering the Cathode Performance Loss by Air break ................................ .......... 34 Fuel Cell Testing ................................ ................................ ................................ ........ 36 Data Acquisition ................................ ................................ ................................ ......... 37 4 RESULTS AND DISCUSSI ON ................................ ................................ ................. 48 The Effect of Voltage during the Air Break ................................ ................................ 48 The Effect of the Stack Operating Temperature ................................ ........................ 50 The Effect of the Stack Operating Hours ................................ ................................ ... 50 Statistical Analysis ................................ ................................ ................................ ..... 51 Co mparing the Method with a Methanol Concentration Sensor ................................ 53

PAGE 6

6 5 CONCLUSIONS ................................ ................................ ................................ ....... 74 APPENDIX A A DISCUSSION ABOUT T HE SHAPE OF THE CURV ES ................................ ....... 76 Factors Affecting the Shape of the Curves ................................ ................................ 76 Methanol Crossover Modeling ................................ ................................ ................... 77 Theoretical Air Flow in the Cathode Channels ................................ ........................... 78 Measuring the Oxygen Flow Rate in the Cathode Channels ................................ ..... 79 B STATISTICAL ANALYSIS ................................ ................................ ........................ 88 LIST OF REFERENCES ................................ ................................ ............................... 89 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 93

PAGE 7

7 LIST OF TABLES Table page 4 1 The average value and the standard deviation at the 15 th 30 th and 45 th second of the air break period for various solution concentrations at 50 C and 0.4 V/cell rest. ................................ ................................ ................................ .................... 72 4 2 The average difference of the calculated methanol concentration in comparison with the sensor readings. ................................ ................................ .................... 73 A 1 Calculated methanol diffusion coefficient at 50 C. ................................ ............. 87 B 1 Stati stical analysis of the data shown in Figure 4 8. ................................ ........... 88

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Basic scheme of the DMFC. ................................ ................................ ............... 20 1 2 Polarization curve. ................................ ................................ .............................. 21 1 3 Membrane electrode assembly (MEA). ................................ .............................. 22 1 4 A conventional (closed cathode) DMFC system. ................................ .............. 23 1 5 An open c athode DMFC system. ................................ ................................ ........ 24 3 1 An open cathode DMFC and its components. ................................ .................... 40 3 2 A typical load profile for a DMFC. ................................ ................................ ....... 41 3 3 A) Test rig components B) Test station. ................................ ............................. 42 3 4 Test experiment flow diagram (Flow through mode). ................................ .......... 43 3 5 Test experiment flow diagram (Recirculating mode). ................................ .......... 44 3 6 A sample data taken at 0.9M fuel solution, 50C, and a constant voltage of 0.4V during the air break. ................................ ................................ ................... 45 3 7 Sample data for various fuel solutions concentrations at 50 C and 0.4 V during the air break. ................................ ................................ ................................ ...... 46 3 8 Current density during the air break vs. solution fuel concentration at 50 C and 0.4V rest. ................................ ................................ ................................ ............ 47 4 1 Current density during the air break vs. solution concentration at 50 C and 0.4V rest. ................................ ................................ ................................ .................... 56 4 2 Comparing the current density during the air break vs. solution concentration at 50 C, for 0.4V rest and 0.3V rest. ................................ ................................ ...... 57 4 3 Comparing the current density during the air break vs. solution concentration at 50 C, for 0.4, 0.3, 0.2, 0.1, and 0.05 V/cell during the air break. ....................... 58

PAGE 9

9 4 4 Effect of the stack operating temperature on the current density during the air break at 0.4V/cell rest. ................................ ................................ ........................ 59 4 5 The current density during the air break vs. concentration at the operating hours of 200, 650, 1150, and 1650 hour. ................................ ................................ ...... 60 4 6 Sample statistical experiment conducted for 0.5 M solution. .............................. 61 4 7 Comparison of the performance of the stack during the air break over 50 consecutive load cycle. ................................ ................................ ....................... 62 4 8 Comparison of the statistical experiment for the premixed solution range of 0.3 1.5 M, 50 C stack operation temperature, 0.4 V/cell rest (50 cycles. .................. 63 4 9 The average values (black dots) and the range of the data (error bars) for a range of concentration of 0.3 1.5M and 50 C at 15 th second of the air break period. ................................ ................................ ................................ ................ 64 4 10 The average values (black dots) and the range of the data (error bars) for a range of concentration of 0.3 1.5M and 50 C at 30 th second of the air break period. ................................ ................................ ................................ ................ 65 4 11 The average values (black dots) and the range of the data (error bars) for a range of concentration of 0.3 1.5M and 50 C at 45 th second of the air break period. ................................ ................................ ................................ ................ 66 4 12 ISSYS FC 10 methanol se nsor. ................................ ................................ .......... 67 4 13 Measured (sensor reading) and calculated methanol concentration at 15 th second of the air break period and 50 C. ................................ ................................ ....... 68 4 14 Measured and calculated solution concentration at various operating temperatures, 0.4V rest, 15 th second of the air bre ak period. ............................. 69 4 15 Measured and calculated solution concentration at various operating temperatures, 0.4V rest, 30 th second of the air break period. ............................. 70 4 16 Measured and calculated solution concentration at various operating temperatures, 0.4V rest, 45 th second of the air break period. ............................. 71 A 1 Diffusion limited methanol oxidation current experiment. ................................ ... 81

PAGE 10

10 A 3 Diffusion limited oxidation current at various anode solution concentrations and stack temperatures. ................................ ................................ ............................ 82 A 4 The structure of MEA. ................................ ................................ ......................... 83 A 5 Calculated required oxygen flow rate in the cathode channels during the air break periods as a function solution concentration at 50 C and 0.4 V rest. ........ 84 A 6 Test set up for measuring the air flow in the cathode channels during the air break. ................................ ................................ ................................ ................. 85 A 7 The comparison between the theoretical calculate curve (green curve) and the measured values (blue curve) of the oxygen flow in the cathode channels during the air break, at 50 C, and 0.4 V rest. ................................ ................................ 86

PAGE 11

11 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science A SENSOR LESS METHOD FOR DETECTING THE METHANOL CONCENTRATION IN AN OPEN CATHODE DIRECT METHANOL FUEL CELL B y Amir Reza Shoraka August 2012 Chair: William E. Lear, Jr. Major: Mechanical Engineering Methanol concentration in the solution fuel plays an important role in maintaining the power output of a direct methanol fuel cell (DMFC) system. Hence, in most of the DMFC systems, in order to monitor and control the methanol concentration a concentrat ion sensing method is used. In this resear ch a sensor less methanol concentration sensing method for an open cathode direct DMFC is discussed. In this method, the methanol concentration is detected by measuring the current density of the stack during the air break period. Air break is a common s trategy to recover the temporary cathode performance loss during which the cathode air delivery is stopped and the stack operates at a constant voltage. I n comparison with a methanol conce ntration sensor an average difference of 0.068 M was reported over a n operating temperature range of 45 55 C and methanol concentration range of 0.3 1.5 M. Also, the effect of the stack life time and the set voltage d uring the air break was studied. One of the greatest features of this method is that it does not require any auxiliary device and since it is using the readily available

PAGE 12

12 data during the air break period it does not require any change in the normal load profile of the system

PAGE 13

13 CHAPTER 1 INTRODUCTION Increas ing demand for power sources in portable application s such as cell phones and laptop s has revealed the necessity for developing technologies other than the conventional batteries. Currently, primary (disposable) and secondary (rechargeable) batteries are widely used and well established as the power source for portable applications. However, the demand for portable power has been increasing faster than the battery technologies and capabilities. This has attracted a global attention to fuel cells which potentially offer a high er power de nsity, and longer lifespan [ 1 ] Among the differe nt types of the fuel cells, direct methanol fuel cell s (DMFC s) are considered to be the most promising fuel cell system for the commercialization purposes in the short term particularly for small to mid size portable applications such as laptops One of t he main reasons for this is attributed to the fuel used in DM FCs, liquid methanol. Methanol is a high energy density fuel ( 15700 kJ/l), easy to handle, transport, and store and can be produced from various sources such as fossil fuels and biomass In addition, low operating temperature (typically between 30 C to 120 C), lack of separate fuel reformer, ease of re fueling and compactnes s are among other features of DMFCs [ 2 ] Even in comparison with the Li I on batteries, in terms of the gravimetric and volumetric efficiency DMFCs show better performance in long term operations [ 2 ] However, DMFCs suffer from several drawbacks including the following : the expensive platinum based catalyst required for both the anode and cathode sides, and methanol crossover from anode to cathode which wastes the fuel and lower the system fuel efficie ncy [ 2 ].

PAGE 14

14 Background DMFC Performance In a DMFC, the electrochemical reaction between methanol and water in the anode produces hydrogen ions, electrons, and carbon dioxide (CO 2 ). The electrons flow through an external circuit while the hydrogen ions diffuse through the electrolyte membrane to the cathod e side, Eq uation 1 1 At the cathode, the electrons and the hydrogen ions complete the reduction reaction at the presence of oxygen from air, Eq uation 1 2 The overall reaction is shown in Equation 1 3. (Anode) ( 1 1 ) (Cathode) ( 1 2 ) (Overall cell ) ( 1 3 ) A single cell DMFC in operation is depicted in Figure 1 1 DMFC single cell s can be combined in series (or parallel) configurations to form stacks to achieve a desired power output. Polarization Curve and Performance Losses The performance of a fue l cell is typically depicted with a graph of its current voltage characteristics, called the polarization curve [ 3 ] The polarization curve shows the output voltage for any given current density imposed on the fuel cell. A typical polarization curve is shown in Figure 1 2 The theoretical thermodynamic open circuit voltage (OCV) for a fuel cell can be calculated by the following equation: ( 1 4 )

PAGE 15

15 where E is the theoretical thermodynamic OCV, is the change in the free energy for the reaction, n is the number of transferred electrons per molecule of fuel reacted (6 in case of DMFC), and F is the Fa raday constant (96486 coulomb /mol e ). An ideal fuel cell would sustain the theoretical OCV at any current density (as long as fuel is supplied). However, due to kinetic issues which cause losses, the actual voltage output of a real fuel cell is lower than the theoretical thermodynamic va lue. Three regions can be distinguished in Figure 1 2 and in each region a certain type of loss is dominant. Activation losses are caused by the slowness of the electr ochemical reactions on the surface of the electrodes and a portion of the theoretical OCV is lost to drive these reactions. Ohmic losses are the drop in the voltage due to resistance to the flow of electrons and ions through the electrode materials and the membrane. Mass transport or concentration losses are the result of the reduction of the fuel concentration or a failure to transport sufficient fuel to the electrodes [ 4 ] Membrane Electrode Assembly (MEA) The overall performa nce of a DMFC is highly dependent on the membrane electrode assembly (MEA) and its structure [ 5 ] The MEA structure ( Figure 1 3 ) is constructed from several layers: Gas diffusion layer (GDL) is a layer that provides paths for the reactants (methanol, water, and oxygen) to reach to the electrode surfaces and also for the products ( CO 2 and water vapor) to leave the electrode surfaces. The GDL m ust be conductive and porous and the most common mat erial used for the GDL is carbon [ 6 ]. The electrolyte membrane used polymer electrolyte membrane (PEM). The PEM provides an ionic conductive medium for the protons, separates the

PAGE 16

16 reactants, and acts as an ele ctrical insulator for conduction of electrons A desirable electrolyte membrane must have high proton c onductivity, low methanol cross over (MCO), and must be chemically and electrochemically stable under operating conditions [ 7 ]. Nafion a poly(tetrafluoroethylene) based ionomer [ 8 ], is known as the standard electrolyte for DMFC because of its great proton conductivity, electrochem ical stability, and durability; h owever it suffers from high methanol permeability and water drag which can cause cathode flooding [ 7 ]. Neburchilov [ 9 ] discussed the recent progress in the new membranes for DMFC and concluded that the h ydrocarbon membrane s seem to have advantages in comparison wi th Nafion with red uced MCO, increased methanol utilization smaller water drag number (number of the water molecules dragged with each proton), and possibly improved dimensional stability [ 7 ]. The anode catalyst layer is where the methanol electro oxidation reaction (MOR) occurs. The catalytic electro oxidation reaction between methanol and water that results in electron and proton liberation and CO 2 formation occurs at the presence of platinum (Pt) cat alyst T he methanol ele ctro oxidation reaction is a complex sequence of reactions [ 10 ] and carbon monoxide ( CO ) is the most stable intermediate formed. CO is considered a catalyst since it is quickly absorbed to the catalyst, and drastically reduces the active catalysts sites I n order to overcome this problem a second ca talyst must be used. A combination of platinum and ruthenium (Ru) is the most active combination for DMFC [ 5 11 ] The cathode catalyst layer is where the oxygen reduction reaction (ORR) occurs. The most common catalyst for the cathode is platinum (Pt). Protons, liberated in the

PAGE 17

17 anodic reaction, transfer through the membrane and reach the cathode to complete the ORR with the electrons and oxygen, Eq uation 1 2 DMFC s ystem s In addition to the DMFC stack, a complete DMFC system requires several peripheral components such as pumps, fan/blower, filter, etc. These periphery components are referred as the balance of plant (BOP). Depending on the cathode design, two types of DMFC systems can be identified: closed cathode or conventional ( Figure 1 4 ) and open cathode ( Figure 1 5 ) [ 12 ] Water management is a critical issue in a DMFC system particul arly for portable applications In order t o keep the balance between the water consumption in the anode and the water formation in the cathode, a portion of the water formed in the cathode reaction must be recovered. In a conventional closed cathode DMFC system, in order to recover the water in the cathode exhaust air, auxiliary water recovery component s are required. However, these components are large and cumbersome and also do not directly aid the electrochemical process. In an open cathode DMFC system, by addition of a hydrophobic liquid barrier layer between the cathode catalyst layer and the cathode gas diffusion layer in the MEA structure, the water recovery is taken place within the MEA. This is a great advantage of an open cathode configuratio n whi ch eliminates the bulky water recovery components required for the conventional configuration. Motivations and Objectives Motivations The methanol concentration in the anodic fuel solution plays an important role in the performance and power output of DMFCs [ 13 14 ] In order to maintain the optimum power output of a DMFC, maintaining the methanol concentration in the fuel solution

PAGE 18

18 within a predetermined range is required. If the methanol concentration falls lower than the optimum range, it can cause f uel starvation in the anode which causes damage to the anodic catalyst layer. However, if the methanol concentration exceeds the optimum range, it increases the methanol crossover to the cathode side, where it can directly react with oxygen within the cathode catalyst layer. Since methanol crossover is a waste of fuel, it adversely affect s the fue l system efficiency. Also, methanol crossover decrease s the cathode potential because of the mixed potential In the DMFC, cathode potential deteriorates as a result o f the depolarization effect caused by methanol crossover [ 15 ]. Hence, in order to maintain the power output of a DMFC system within an optimum range monitoring and controlling methanol concentration in the fuel is required. In general, the existing methan ol concentration sensors are hard to miniaturize, might require auxiliary driving devices which add complexity to the system, and costly. These would significantly limit their application particularly for portable DMFC systems. For instance, the methanol c oncentration sensor used in UNF DMFC [ 12 ] system accounts for roughly 10% of the system total volume and one third of the system cost. However, s ince the DMFC performance is dependent on methanol concentration, there is an opportunity to determine the methanol concentration by analyzing the DMFC operating characteristics such a s voltage or current. However the DMFC performance is also dependent on other variabl es such as the operating temperature and operating time. This could complicate establishing a relationship between methanol concentration and DMFC performance characteristics particularly during normal operation. Another opportunity to determine methanol c oncentration is during the rejuvenation or air break

PAGE 19

19 period [ 16 17 ] Air break is a strategy to partially recover the performance loss in DMFCs and is performed by stopping the cathode air delivery while the voltage is set to a fixed value. Objectives The p urpose of this thesis is to explore the dependence, in an open cathode DMFC, of the current density during the rejuvenation period at a constant voltage on the methanol concentration. Based on this, the feasibility of a sensor less method to detect the met hanol concentration is to be explored, in which the methanol concentration in the fuel is determined by measuring the current density during the rejuvenation period. In addition, the effect of the rejuvenation voltage set point, the operating temperature, and the stack life time on this method is to be determined.

PAGE 20

20 Figure 1 1 Basic scheme of the DMFC

PAGE 21

21 Figure 1 2 Polarization curve

PAGE 22

22 Figure 1 3 Membrane electrode assembly (MEA)

PAGE 23

23 Figure 1 4 A conventional (closed cathode) DMFC system. Reprinted by permission from Dr. James Fletcher, 2012, Advanced direct methanol fuel cell for mobile computing, University of North Florida.

PAGE 24

24 Figure 1 5 An open cathode DMFC system. Reprinted by permission from Dr. James Fletcher, 2012, Advanced direct methanol fuel cell for mobile computing, University of North Florida.

PAGE 25

25 CHAPTER 2 LITERATURE REVIEW In this chapter the literature review for different types of m ethanol concentration sensing methods is pre sented. The methanol concentration techniques can be categorized into three groups: electrochemical sensing methods, physical sensing methods, and s ensor less sensing methods The requirements for a suitable m ethanol concentration sensor can be summarized as [ 13 ] : wide methanol sensing range of up to 5 M, high resolution and accuracy around the operating point, in line operation and rapid response, tolerance to metallic ion impurities such as Fe 3+ and Cu 2+ ideally tolerance to the dissolved CO 2 and CO 2 bubbles, long term stability, wide range of operating temperature, amenable to miniaturization for tight packaging and system integration. Electrochemical Sensing M ethods In principle an electrochemi cal sensor is a small DMFC that its performance is use d as the sensing measurement Depending on the reactio n mec hanism, there are two types of electrochemical sensors: oxidation current (electro lytic type) and fuel cell based (galvanic type) [ 13 ] The relationship between methanol concentration and me thanol oxidation current is described in [ 18 19 ] by Naray anan. In this sensor the methanol oxidation current under a consta nt anodic potential is used as the sensing signal. The oxidat ion current is dependent on methanol concentration in the fuel solut ion. The sensor is suitable for methanol concentration range of 0.25 2.0 M with an average error of 0.02 M during an

PAGE 26

26 entire 70 hour test for a circulating solution stream of 0.5 M. The temperature dependency of this method is not clearly discussed. An alternating pulse electrochemical methanol concentration sensor is described in [ 20 ]. The operation of this sensor is based on measuring the methanol oxidation current. In this sensor, the structure of both electrodes is the same and methanol solution is fed into both electrodes simultaneously. Alternative pulse signal changes the potentials at the working electrode and counter electrode of the sensor in a certain frequency. Hence, methanol oxidation and hydrogen evolution occur on the same electrode alternately. This method has effective measurement in the range of 0 2. 0 M. It is reported that t he required power for this sensor is 1.5 W and the time response is 20 seconds. A miniaturized electrochemical methanol sensor is discussed in [ 21 ]. The sensing principle is based on measuring the methanol oxidation current. In thi s sensor, reduction of size is made possible by eliminating both the gas diffusion backings and any housing part needed for physical integrity of the unit. It is reported that this sensor is suitable for methanol concentration range of 0.5 3.0 M and the te mperature range of room temperature to 80 C The total power consumption by the sensor is 1 mW and the sensor responds within few seconds to the concentration changes. A fuel cell base (galvanic type) methanol concentration sensor is discussed in [ 22 ]. Th is sensor is essentially a passive DMFC. The MEA for this sensor is uniquely designed to have the anode side much smaller than that of the cathode. This is to ensure that all the methanol diffuses to the anode/membrane interface can be completely reacted, thus the measured sensor current reflects the true concentration of

PAGE 27

27 the solution. It is claimed that this sensor is able to operate over a 20 80 C temperature range and a methanol concentration range of 0 4 M. This sensor has a linear relationship in cali bration under a constant temperature and also a first order correlation with respect to the temperature change. A fuel cell based methanol c oncentration sensor is explained in US patent 4,810,597 [ 23 ]. The operation principle of this particular sensor is based on measuring the open circuit voltage of the cell. The methanol crossover is directly related to the solution fuel concentration in the anode side. So, the mixed potential, which reduces the cat hode potential and is caused by the meth anol crossover, is larger for the case of higher concentration in the anode side. This means the methanol concentration at the anode can be determined by measuring the open circuit voltage of the cell. However, the c hange in the open circuit voltage is not significant over a range of methanol concentration of 0.25 5.0 M Instead of measuring the open circuit voltage of the sensing cell, in US patent 5,624,538 [ 24 ] the cell voltage at a constant load of 1 ohm is used as the sensing measurement to determine the methanol concentration. The method is suitable for concentration s only up to 1 M. To extend the measuring range of this method, a new method was introduced in US patent 6 527 943 [ 25 ] in which the methanol concent ration range was extended to 4 M. This sensor is claimed to have an operating temperature range of 40 100 C. The fuel cell based methanol concentration sensor described in [ 26 ] is a small passive DMFC. In this method the methanol concentration is determine d by measuring the cell temperature. It is shown in [ 26 ] that there is a linear relationship between the

PAGE 28

28 maxim um temperature rise of the sensor and methanol concen tration in the range of 0.5 10.0 M. The feasibility of using single walled carbon nanotube (SWCNT)/Nafion composites as methanol sensors is discussed in [ 27 ]. In this method, methanol concentration is determined by measuring the resistance of the sensor under a constant voltage. The result for a range of methanol concentration of 0.5 5.0 M is presented in [ 27 ]. Physical Sensing M ethods The physical properties of aqueous methanol solution have b een investigated and measured using sev eral different approaches [ 13 ]. T he physical properties of aqueous methanol solution such as density, viscosity, relative dielect ric constant heat capacity, and speed of sound are dependent on the methanol concentration in the solution A density t ype sensor is discussed in [ 28 ] In this sensor the resonant frequency of a solution sample is used to determine the sample d ensity. T he resonant frequency of vibrati on of the sample changes as the result of the change in t he mass of the sample s This sensor operates over the ent ire range of concentrations. This sensor is not specifically designed for DMFC systems A micro electro mechanical system (MEMS) based methanol concentration sensor designed for DMFC systems is discussed in [ 29 ]. This senso r is based on a vacuum packaged resonant microtube and an on chip platinum temperature sensor to measure the density of the solution sample. For a methanol concentration range of 0.6 3.0 M and a temperature range of 10 50 C, this sensor is found to have a linear relationship. Also, formic acid contaminants under concentration of 100 ppm do not

PAGE 29

29 affect the accuracy of this sensor [ 29 ]. This sensor is used in a DMFC s ystem developed by the University of North Florida (UNF ) [ 12 ]. The dielectric property of the aqueous methanol solution was in vestigated in [ 30 ] and it is shown t hat the static dielectric constant changes with variation in the methanol concentration. By measuring the frequency dependent capacitance, the dielectric constant can be determined. This type of sensor consists of a capacitor that comprises a pair of space d electrode plates. By measuring the capacitance as the fuel passes through the electrodes, the dielectric constant is determ ined The dielectric spectra for a metha nol concentration range of 0 5 %wt (about 0 1.6 M) were measured by Doerner [ 31 ] A viscosity type sensor was introduced in US Patent 6 536 26 2 [ 32 ] The viscosity of aqueous methanol solution is dependent on the methanol concentration of the solution and in the range of 0.0 9 .0 M a li near relationship can be seen. In this sensor, t he visc osity of the sample is determined by measuring the pressu re across the constriction with a constant laminar flow r ate. D ue to the shape of the viscosity concentration curve, double value could be a prob lem. But this may not be an issue for the DMFCs since the recommended methanol concentration in the fuel is less than 1.5 2.0 M for DMFCs [ 33 ]. The speed of sound in the methanol solution is related to its density and elastic properties [ 34 ]. In the US Patent 6 748 793 [ 35 ] an ultrasound method for determining the methanol concentration in an aqueous solution is discussed that can determine the methanol concentrat ion in the range of 0.1 5.0 wt % (about 0 1.6 M) In this sensor a piezoelectric transducer was used to emit and receive ultrasound waves. Using similar

PAGE 30

30 reference and sample chambers, the differential propagation time of the ultrasound pulses was measured in response to the change of methanol concentration. Evaluation of ultrasonic sensing of methanol concentration for DMFC over a methanol concentration r ange of 0 15 wt% (about 0 5.0 M) and a temperature range of 25 78 C is discussed in [ 36 ]. The results show that the propagation distance and equipment sampling rate affect the accuracy of sound speed measuring. An infrared spectrum type sensor is discusse d in US Patent 6,815,682 [ 37 ] The transmittance of infrared radiation is a function of the methanol concentration in the solution. This sensor is reported to have effective measurement over a range of methanol concentration of 0.0 5.0 wt% (about 0 1.6 M) w ith an accuracy of 0.1 wt% (about 0.03 M). This sensor has high accuracy particularly for methanol concentration less than 1 .0 M. There are several methanol concentration sensing methods that work based on measuring the refractive index [ 38 39 ]. The NeHe laser based technique proposed in [ 38 ] can determine the methanol concentration in the solution over the entire range of concentration with a precision on the or der of 0.01 wt% [ 13 ]. The specific heat capacity of an aqueous methanol solution is dependent on methanol concentration of the solution. Hence, the technique intr oduced in [ 40 ] determines the methanol concentration by measuring the specific heat capacity of the solution. A temperature difference of approximately 2 C was reported for a change of methanol concentration from 0.5 M to 1,0 M with a flow rate of 100 mL /m in to the sensor [ 13 ].

PAGE 31

31 A methanol sensor using a shear horizontal surface acoustic wave device is explained in [ 41 ]. A linear relationship between the sensor resp onse and concentration was obtained over a temperature range of 25 60 C and concentration range of 0 50 wt% (about 0 16 M). The reported accuracy was 0.1 %wt. The nonlinear frequency response of a DMFC is studied by analyzing the total harmonic distortio n (THD) spectra in [ 42 ]. The results show that there is a monotonic correlation between THD values and methanol concentration at certain frequencies. The sensitivity of this correlation is improved with increased current amplitude. This correlation for meth anol concentration range of 0.3 1.0 M is presented in [ 42 ]. Sensor less M ethods Other than the conventional methods of employing a methanol concentration sensor as discussed i n the previous sections, there are several works that focus on sensor less methods in order to control the ope ration of the fuel cell. In a sensor less method, the methanol concent ration is determined based on a known relationship b etween the fuel cell performance characteristics and the methanol concentration [ 13 ]. A strategy of estimating methanol concentration in a DMFC system is explained in [ 43 ] Th e algorithm o f estimating methanol concentration is developed by using a three dimensi onal measurement space and constant c oncentration surface (CCS). The three indices are chosen to be current, voltage, and temperature. This method can estimate the methanol concentration without interrupting the operation of the system. The reported error over a concentration range of 3.5 5.5 vol.% (about 0.85 1.40 M) and temperature range of 30 55 C is 0.3 vol.% (about 0.075 M) A modified and more advanced form of the method discussed in [ 43 ] is explained in [ 44 ] where a compensation scheme is proposed and validated to overcome the

PAGE 32

32 voltage decay limitation in the method described in [ 43 ]. Also, an approach to evaluate the consumption rate of water and methanol is provided in [ 44 ]. A sensor less method of controlling the methanol concentration in DMFC, which is based on estimating the methanol consumption rate in the system, is explained in [ 45 ] The total rate of methanol consumption is calculated based on the fuel cell current and th e estimated rate of methanol crossover. The average error at operating temperature of 60 C was within 20% of the set points (0.4 M and 0.8 M). In the proposed method in [ 46 ], the methanol concentration is regulated by using feedback from voltage measuremen ts. This paper concluded that at a constant current density, steadily increasing the concentration would result in a rapid increase in voltage from an initial value of zero until a maximum point. After this maximum point, further increase in the concentrat ion would cause a reduction in voltage which is mainly due to mixed potential losses under conditions of high crossover. A portable DMFC system which utilizes a methanol sensor less c ontrol method is presented in [ 47 ]. The method is discussed in detail in [ 48 characteristics are measured to regulate the fuel concentration and to optimize the performance. It is claimed that the fuel sensor less control algorithm is effective and is able to control the solution concentration within the range 0.5 1.3 wt% (about 0.015 0.45 M). The reliability of this method under dynamic load ing conditions is experimentally verified in [ 49 ] and it is shown that the control algorithm of this method can survive during the stack continuous degradation. This control algorithm is further improved in [ 50 ] A new load current integral function with a fixed domain is validated,

PAGE 33

33 which calculates the amount of fuel consumed during the last monitoring cycle. Also, the response time is reduced in [ 50 ]. In a sensor less method discussed in [ 51 ] it is shown that the methanol concentration in the solution fuel is a function of the open circuit voltage transient response in the methanol concentration range of 0.6 1.60 M. This method was successfully applied to a brassboard system over 20 hours of operation.

PAGE 34

34 CHAPTER 3 EXPERIMENTAL METHODOLOGY T he sensor less method is considered as one of the most desirable option to determine and control the methanol concentration in a DMFC system particularly for portable application s In this chapter, a new sensor less method to determine the methanol concentration in a DMFC system, including the empi rical data and analysis, is explained. Recovering the Cathode Performance Loss by Air break L ong term durability is one of the main challenges with regard to DMFCs [ 16 ] In general two types of performance los ses can be identified: non recoverable performance lo ss or degradation and temporary (recoverable) performance degradation. G radual loss of active el ectrocatalyst surface area (anode and cathode), loss of cathode hydrophobic properties, or ruthenium crossover from the anode to the cathode through the electrolyte membrane would cause non recoverable performance loss over long term operation [ 16 ] I n short term operation, loss of cathode activity due to surface o xide formation causes temporary performance degradation. T he platinum surface oxide (and/or hydroxide) leads to a decrease in the number of the free platinum sites which can inhibit the oxygen reduction reaction (ORR) on the cathode surface and reduce the cathode activity. Bringing the cathode potenti al down to a value that results in comp lete reduction of the surface platinum oxide can completely recover the cathode activity [ 16 ]. However, in a DMFC, if the cell voltage is lowered, the sluggish nature of the methanol oxidation process can lead to an increase in the anode potential which is not desirable and may induce further permanent performance degradation. An alternative method to reduce the cathode potential in DMFCs is

PAGE 35

35 interrupting the air (oxygen) delivery to the cathode (air break). During the air break the oxygen remaining in the cathode plenum i s quickly consumed by the cross over methanol which lead s to lowering the cathode potential [ 16 ] In a closed cathode DMFC which was the subject of study in [ 16 ], during the air break no more air can flow through the cathode plenum. In contrary, i n an open cathode DMFC like the one shown in Figure 3 1 the cathode channels are open to the atmosphere. So, even during the air break, a limited amount of air can flow thro ugh the cathode channels due to diffusion and/ or natural convection due to the temperature gradient between the fue l cell stack and atmosphere The air break method descri bed for a closed cathode DMFC in [ 16 ] included four steps: a) stopping air flow to the cell, b) immediately switching cell operation to constant current mode using the same current as the current generated by the fuel cell at the time of the switch, c) restarting airflow as soon as the cell reaches a preprogrammed low voltage limit, d) i mmediately running the fuel cell in constant voltage mode. Step ( c ) is fair ly quick since the remaining air in the cathode pl enum is quickly consumed and the cell voltage drops quickly. By h olding the cell voltage at a specified level, in case of a closed cathode DMFC, the current would quickly drop to zero since all the remaini ng oxygen would be quickly co nsumed and no more reactions could happen in the cathode. However, this would not be the case for an open cathode DMFC where even during the air break periods a limited amount of oxygen can diffuse through the cathode c hannels Th is can be clarified by discussing a typical load profile for an open cathode DMFC Figure 3 2 shows a typical load profile for a DMFC running in a constant load (current density) mode The red line indicates the load (current density) on the system

PAGE 36

36 and the blue line shows the aver age cell voltage. Prior to the air break period, the load is removed for a short period of time (10 sec in Figure 3 2 ) and the voltage reach es open circuit voltage (OCV). Following this step, the air flow is stopped and the cell voltage is set to a specified value (0.4 V in Figure 3 2 ). During the air break, the current density initially drops rapidly and finally reaches a steady state value Following the air break period, the load is removed, the cathode air flow is resumed, and the cell voltage reaches the OCV value and finally the cell continue s running under a co nstant load Preliminary experimental data show that the magnitude of the current density during the air break at a constant voltage is related to the concentration of the fuel in the anode side. Fuel Cell Testing The fuel cell testing was co nducted on an 8 cell stack built in the Fuel Cell lab at the University of North Florida. The 8 cell stack shown in Figure 3 1 was assembled using bipolar plates such as the one s shown i n Figure 3 1 A bipolar plate is a structure for distributing the reactant s and collecting and distributing current The membrane electrode assembly (MEA) was fabrica ted using a piece of membrane made by the UNF DMFC research group. This is a 20 m coating formulation applied and laser drilled. The anode was a Johnson Matt h ey electrode ELE147 which is designed to work for methanol fue l cells, with a Pt/Ru cata lyst. The membrane was hot pressed with cathode and anode The area of each MEA was 15.5 cm and a total of 8 MEAs were used to build the stack As shown in Figure 3 3 the stack is placed in a special duct wh ere the cathode air is delivered to the duct and flowed through the cathode channels from the bottom The two thermocouples used to monitor the fuel s olution temperature before the stack and

PAGE 37

37 right after the stack are shown in Figure 3 3 The stack temperature, measured by the second thermocouple, is controlled by t he amount air deliver ed to the duct. The cathode air mass flow controller, the heater surface and solution temperature controllers are assembled in a test station shown in Figure 3 3 In addition, the load on the stack (current or voltage mode ) was controlled by a load bank integrated in to the test station. The test station parameters such as the cathode flow rate, the anode solution flow rate, anode solution temperatu re, and the load o n the system are controlled by the LabVi ew program. The flow diagram of the test experiment configuration is shown in Figure 3 4 Note that in Figure 3 4 th e test experiment co nfiguration was such that the solution fuel was not re circulated. This configuration was used to develop a relationship between methanol concentration and the current density during the air break periods at various temperatures by using premix methanol solutions with various concentrations. After the relation ship was established, i n order to evaluate the re liability and robustness of the relationship between the current density during the air break and the solution fuel concentration a set of test experiments was performed in a recirculating mode shown in Figure 3 5 In the recirculatin g mode, the methanol solution was recirculated back to the solution tank where pure methanol was added to the solution in order to maintain the concentratio n at a specified concentration while the solution concentration was monitored by a density type sensor (ISSYS FC 10) Data Acquisition T he preliminary data showed that the current density during the air break was a function of the methanol concentration in the solution fuel. Figure 3 6 shows sample data taken with 0.9 M fuel solution, stack temperature of 50 C, and a constan t voltage of 0.4 V during the air break. The air break period is clearly shown and defined in Figure

PAGE 38

38 3 6 Note that prior to and right after the air break, the stack period with the air pres ent. After an initial rise the current density falls down and reaches a steady state value. This steady state value is dependent on the methanol concentration in the fuel as shown in Figure 3 7 Figure 3 7 shows the sample data for a range of solution concentration s of 0.3 1.5 M at a stack operating temperature of 50 C and 0.4 V/cell during the air break Clearly, t he current density during the air break at a constant voltage is in versely related to the methanol concentration in the fuel solution. This relationship could be a fun ction of the operating temperature, the voltage during the air break, or the operating time of the stack The data shown Figure 3 7 can be presented in the form shown in Figure 3 8 where current densities at three particular moments of the air break period are graphed versus the solution fuel concentration. The conducted experiment s for this work were performed in two categories. In the first category, the experiments were focused on studying the effective parameters on the relationship between the current density during the air break and the solution fuel concentration. These param eters were including the following: the set voltage during the air break, the operating temperature, and the operating time of the stack. Studying the effect of the set voltage during the air break helped to determine which set voltage resulted in better r esolution. A series of test experiments was conducted for a range of premixed methanol solution of 0.3 1.5 M and air break voltage of 0.4, 0.3, 0.2, 0.1, and 0.05 V/cell. In all experiments, the stack was operated on a no load mode (OCV) for two periods, o ne prior to the air break and one right after the air break ends, with each period lasting 15 seconds. By i nvestigating the effect of the operat ing temperature, an

PAGE 39

39 empirical model was developed which can determine the methanol concentration for an operatin g temperature range of 45 55 C and methanol concentration range of 0.3 1.5 M Also, since the performance of DMFCs is subjected to degradation over long term operation, the validity of this method was examined over 1700 hours of operation. The second category of the experiments was dedicated to study the re liability of this method through s tatistical analysis and also in comparison with a methanol concentration sensor. The repeatability of the data was studied over 50 consecutive load cycles for a give n operating temperature and solution concentration. Finally, the empirical model was validated in comparison with a methanol concentration sensor over an operating temperature range of 45 55 C and methanol concentration range of 0.3 1.5 M.

PAGE 40

40 Figure 3 1 An open cathode DMFC and its components Courtesy of Amir Shoraka.

PAGE 41

41 Figure 3 2 A typical load profile for a DMFC

PAGE 42

42 Figure 3 3 A) Test rig components B) Test station Courtesy of Amir Shoraka.

PAGE 43

43 Figure 3 4 Test experiment flow diagram (Flow through mode)

PAGE 44

44 Figure 3 5 Test experiment flow diagram (Recirculating mode)

PAGE 45

45 Figure 3 6 A sample data taken at 0.9M fuel solution, 50C, and a constant voltage of 0. 4V during the air break

PAGE 46

46 Figure 3 7 Sample data for various fuel solutions concentrations at 50 C and 0.4 V during the air break.

PAGE 47

47 Figure 3 8 Current density during the air break vs. solution fuel concentration at 50 C and 0.4V res t

PAGE 48

48 CHAPTER 4 RESULTS AND DISCUSSI ON In this chapter, the tests results are shown and the relation between the fuel concentration and the current dur ing the air break is discussed. In addition, the effect s of the voltage during the air break, the stack temperature, the operating hours of the stack are discussed. The Effect of Vol tage during the Air Break Figure 4 1 shows the data taken for various premixed solution concentrations at a stack operation temperature of 50 C and set voltage of 0.4 V/cell For the purpose of comparison in this section, the curves are constructed at 45 th of a 60 second air break (45 seconds after the air delivery was stopped and a constant voltage load of 0.4 V/cell was imposed on the stack). The current density during the air break is in versely related to the fuel solution concentration. The reason for s uch a behavior can be attributed the higher rate of methanol crossover at higher solution concentration s Note that the methanol crossover is driven by diffusion phenomenon and hence higher concentration in the anode side would increase the rate of methan ol crossover. During the air break a limited amount of oxygen can diffuse to the cathode channel s during the air break which can be consumed for either the cathode oxygen reduction reaction or can react with the crossover methanol within the cathode catal yst layer The higher the solution fuel concentration results in a higher rate of methanol crossover and consequently more oxygen would be consumed for the m ethanol oxidation reaction. This means less oxygen would be available for the cathode oxygen reduct ion reaction and as the result,

PAGE 49

49 l ower current density would be observed. This is explained in detail and proven in Appendix A Figure 4 2 shows the effect of reducing the voltage from 0.4 V/cell to 0.3 V/c ell during the air break period where the curve is shifted upward This can be explained by considering the polarization curve for a DMFC ( Figure 1 2 ), where generally decreasing the voltage results in a higher current density. In Figure 4 3 the resu lt of the experiments for a range of air break voltages is shown. In general, the current density during the air break increases as the voltage is set to a lower value during the air break which can be explained by considering the polarization curve. Wit hin the first portion of the polarization curve ( Figure 1 2 ), where the activation losses are the dominant type of performance losses, decreasing the voltage does not significantly increase the current density, so the increase in t he current density is not significant as the air break voltage is decreased from 0.4V to 0.3V or even 0.2 V. However, within the second region of the polarization curve, where the ohmic losses are dominant, a small decrease in the voltage can significantly increase the current density and this could explain the significant increase in the current density from 0.2 V to 0.1 V or 0.05 V during the air break. In case of the very low solution concen tration such as 0.3 M, the current density does not increase significantly by reducing t he voltage lower than 0.1 V during the air break which can be caused by the concentration losses which is dominant in third region of the polarization curve ( Figure 1 2 ). Considering the curves for various voltages during the air break, the curves related to the lo w voltages such as 0.1 and 0.05 V/cell during the a ir break have less

PAGE 50

50 resolution when compare d with the curves related to the higher voltages during the rest such as 0.4, 0.3, or 0.2V/cell during the air break. The Effect of the Stack Operating Temperature The effect of the stack operating temperature on the current density during the air break was studied by performing a set of experiment s over a stack operating temperature range of 45 55 C. Figure 4 4 shows the result of a set of ex perime nts conducted at 0.4 V/ cell during the air break Generally, as the operating temperature increases the current density during the air break decreases for a particular solution concentration. The reason for this could be mainly attributed to higher r ates of methanol cross over at higher temperatures. Since methanol crossover is a diffusion phenomenon, it is highly temperature dependent. The effect of the temperature inc rease is less pronounced at very low concentrations such as 0.3 M due to the extreme ly low methanol concentration at the anode catalyst and membrane interface (concentration loss part of the polarization curve). The Effect of the Stack Operating Hours T he performance degradation is one of the challenges about of DMFCs. This wor k is not intended to discuss the method intro duced in this thesis is dependent on the performance of DMFC during the air break, it is necessarily to investigate the effect of the operating hour s of the DMFC on t his me thod. During the time that the 8 cell stac k was not being used in any specific experiment, it was operating at a typical constant load of 120 mA/cm and operating temperature of 50 C with 60 second air break periods every 10 minutes of operation. By implementing this strategy, investigation of the effect of the operating time was possible. The performance of the 8 cell stack was measured at several specified

PAGE 51

51 times over a range of solution conc entrations of 0.5 1.5 M. The result is shown in Figure 4 9 There is a significant change between the data taken aft er 200 hours operation hours compare d with the dat a taken af ter 650 hours of operation. It is possible that the DMFC have an initial degradation during the first few hundred hours of operation followed by a much slower rate of degradation. This could explain why there is not a significant difference betwe en the data taken after 6 50 hours of operation in comparison with the data taken after 1150 hours of operation or the data taken after 1650 hours of operation. Since the perf ormance degradation is a complicated issue affected by various factors including the MEA structure, catalyst, and even the load profile a comprehens ive set of test experiments including several DMFC stacks must be conducted prior to express ing any certain conclusion s regarding the effect of the operating hours of the fuel cell on this method T his is beyond the scope of this work However, Figure 4 9 shows that the effect of the operating hours is not significant after the first few hundred hours o f the stack operation hours for this type of DMFC and this stack in particular Statistical Analysis B efore this method can be used as a methanol concentration measurement technique, it is important to know the repeat ability A comprehensive set of statistical experiments was performed at stack operating temperature of 50 C and the air break voltage of 0.4 V/cell for a range of premixed methanol solutions of 0.3 1.5 M. The statistical experiment for each solution concentrati on included 50 repetitive cycles and each cycle was comprised of four steps: 1 ) 300 seconds running at a constant load of 120 mA/cm at a stack temperature of 50 C that is contr olled by the

PAGE 52

52 cathode air flow, 2 ) 15 seconds running at no load (OCV) with a c athode flo w rate of 4 SLPM/cell, 3 ) 60 seconds of running at a constant voltage of 0.4 V/cell (air break), and 4 ) 15 seconds running at no load (OCV) with a cathode flow rate of 4 SLPM/cell. T he purpose of this experiment was to compare the performance of the stack during the air break periods of 50 consecutive cycles A sample of the statistical experiment conducted for a premixed fuel solution of 0.5 M is shown in Figure 4 6 The circles in Figure 4 6 show the current density during the air break periods. In Figure 4 7 the performance of the stack during 50 consecutive air break periods is shown. Each cross sign indicates one cycle and the black circle s represent the average of the performance values at any g iven second of a 60 second air break. T he result of the statistical experiment, conducted for a range of premixed solution concentrations of 0.3 1.5 M, is shown in Figure 4 8 Table 4 1 shows the average (m) and standard deviation ( values at 15 th 30 th and 45 th second of th e air break periods It is shown in Table B 1 that t he data has a normal distribution for 50 samples at a ny given second. Statistically, it means there is no systematic variation in the data and the probability that a random measurement falls in the range of [m m+ ] is roughly 0.65 and the probability that a random measurement fall in the range of [m m+ ] is roughly 0.96. The data shown in Table 4 1 is presented in the form of three plots which each shows the current density as a function of solution concentration at a particular time after the start of the air break period. The error bars in Figure 4 9 Figure 4 10 and Figure 4 11 show the range of the data for a particular conc entration Considering the curves sho wn in Figure 4 9 Figure 4 10 and Figure 4 11 it could be concluded that,

PAGE 53

53 statistically there is no significant advantage for choosing any particular curve over the others Comparing the Method with a Methanol Concentration Sensor The reliab ility of this method was examined by comparing the result of this method with the measurements of a methanol concentration sensor. In ord er to do so, a test set up shown in Figure 3 5 was prepared. The methanol sensor used for this experiment was an ISSYS FC 10, shown in Figure 4 12 which is suitable for methanol concentration range of 0 30%wt and temperature range of 0 70 C The methanol sensor was able to continuously measure and monitors the solution concentration, however, the m ethod introduced in this work was on ly able to determine the solution concentration during the air break period s So, in order to compare the results from the sensor a nd this method, the calculated methanol concentrations based on the current density at three particular moments during the ai r break (the 15 th the 30 th and the 45 th second of a 60 second air break) were compared to the solution c oncentration measured by the sensor at the beginning of the air break period The experiment wa s conducted at a range of solution concentration of 0.4 1.5M and operating temperatures of 45 55 C. In Figure 4 13 the sensor reading, the calcula ted concentration, and the difference between the method and the se nsor readings at 50 C is shown. The calculated methanol concentration is based on the current density at the 15 th second of the air break period. The aver a ge difference over a range of methanol concentration of 0.3 1.5 M at 50 C is 0.0 67 M. However, the average difference associated with the lower methanol concentration region, 0.3 0.9 M, is 0.061 M and for the higher methanol concentration reg ion, 0.9 1.5 M, is 0.07 4 M. The slightly lower

PAGE 54

54 difference could be due to the better resolution (steeper slope) in the lower concentration region of the curve shown in Figure 4 1 The results of the experiments conducted over a range of temperature of 45 50 C and at the 15 th 30 th and 45 th second of the air break period are shown in Figure 4 14 Figure 4 15 and Figure 4 16 respectively As mentioned earlier, this method is not able to continuously determine the methanol concentration and the points shown in Figure 4 14 Figure 4 15 and Figure 4 16 represent the calculated solution concentration at particular moments during a 60 second air break period. The average difference associated with different parts of the a bove three figures are shown in Table 4 2. In general, the average difference is smaller for the low methanol concentrati on region, 0.3 0.9 M Also, it can be conclud ed that there is no significant advantage of calculating the methanol concentration based on the current density at 30 th o r 45 th second of the air break period in comparison with that based on the current density at the 15 th second of the ai r break period. This means an air break period as short as 15 seconds can be sufficient for the purpose of implementing this method. D uring the air break, the power generated by the DMFC is significantly lower than normal operation and hence long air break periods are not desirable. The average differences for the calculation based on the current density at the 15 th 30 th and 45 th secon d of the air break period are 0.066 M, 0.069 M, and 0.068 M, respectively. In general, the result of the comparison test experiment looks promising pa rticularly in terms of the difference in determining the methanol concentra tion. In this chapter the effe ct of several parameters such as the operating temperature and the operating hours of the stack on this method were discussed. The effect of the

PAGE 55

55 operating temperature was considered in the calculations, but t he effect of the stack life time or the performa nce degradation on this method can be revealed only by conducting a set of comprehensive test experiments over various types of MEAs which is b eyond the scope of this thesis. However, the results shown in this chapter certainly show that the methanol conce ntration in an open cathode DMFC can be determined by measuring the current density during the air br eak and this method has potential to be used in a DMFC system to determine the methanol concentration in the fuel solution

PAGE 56

56 Figure 4 1 Current density during the air break vs. solution concentration at 50 C and 0.4V rest

PAGE 57

57 Figure 4 2 Comparing the c urrent density during the air break vs. solution concentration at 50 C, for 0.4V rest and 0.3V rest

PAGE 58

58 Figure 4 3 Comparing the c urrent density during the air break vs. solution concentration at 50 C, for 0.4, 0.3, 0.2, 0.1, and 0.05 V/c ell during the air break

PAGE 59

59 Figure 4 4 Effect of the stack operating temperature on the current density during the air break at 0.4V/cell rest.

PAGE 60

60 Figure 4 5 The current density during the air break vs. concentration at the operating hours of 200, 650, 1150, and 1650 hour.

PAGE 61

61 Figure 4 6 Sample statistical experiment condu cted for 0.5 M solution.

PAGE 62

62 Figure 4 7 Comparison of the performance of the stack during the air break over 50 consecutive load cycle

PAGE 63

63 Figure 4 8 Comparison of the statistical experiment for the premixed solution range of 0.3 1.5 M, 50 C stack operation temperature, 0.4 V/cell rest (50 cycles

PAGE 64

64 Figure 4 9 The average values (black dots) and the range of the data (error bars) for a range of concentration of 0.3 1.5M and 50 C at 15 th second of the air break period.

PAGE 65

65 Figure 4 10 The average values (black dots) and the range of the data (error bars) for a range of concentration of 0.3 1.5M and 50 C at 30 th second of the air break period.

PAGE 66

66 Figure 4 11 The average values (black dots) a nd the range of the data (error bars) for a range of concentration of 0.3 1.5M and 50 C at 45 th second of the air break period.

PAGE 67

67 Figure 4 12 ISSYS FC 10 methanol sensor Courtesy of Amir Shoraka.

PAGE 68

68 Figure 4 13 Measured ( sensor reading) and calculated methanol concentration at 15 th second of the air break period and 50 C

PAGE 69

69 Figure 4 14 Measured and calculated solution concentration at various operating temperatures, 0.4V rest, 15 th second of the air break period.

PAGE 70

70 Figure 4 15 Measured and calculated solution concentration at various operating temperatures, 0.4V rest, 30 th second of the air break period.

PAGE 71

71 Figure 4 16 Measured and calculated solution concentration at various operating temperatures, 0.4V rest, 45 th second of the air break period.

PAGE 72

72 Table 4 1 The average value and the standard deviation at the 15 th 30 th and 45 th second of the air break period for various solution concentrations at 50 C and 0.4 V/cell rest. Solution concentration (M) 15 th second average current density (mA/cm) 30 th second average current density (mA/cm) 45 th second average current density (mA/cm) 15 th second standard deviation 30 th second standard deviation 45 th second standard deviation 0.3 56.17 54.56 53.79 0.574 0.397 0.447 0.5 40.60 37.84 36.49 0.453 0.404 0.390 0.7 27.76 24.93 23.92 0.564 0.709 0.722 0.9 17.04 15.51 14.80 0.403 0.316 0.338 1.1 12.09 11.27 10.65 0.525 0.427 0.468 1.3 9.63 8.94 8.30 0.258 0.291 0.282 1.5 7.77 7.13 6.53 0.288 0.267 0.300

PAGE 73

73 Table 4 2 The average difference of the calculated m ethanol concentration in comparison with the sensor readings. Calculation based on 45 C 50 C 55 C 0.3 0.9 M 0.9 1.5 M 0.3 0.9 M 0.9 1.5 M 0.3 0.9 M 0.9 1.5 M 15 th second 0.061 M 0.063 M 0.061 M 0.074 M 0.067 M 0.080 M 30 th second 0.065 M 0.067 M 0.064 M 0.068 M 0.076 M 0.079 M 45 th second 0.066 M 0.065 M 0.062 M 0.061 M 0.077 M 0.068 M

PAGE 74

74 CHAPTER 5 CONCLUSION S The performance current density of an open cathode DMFC during the air break period at a constant voltage is inversely related to the methanol concentration of the anode fuel solution ( Figure 3 7 ) and this relationship is more distinct in the lower concentration region (less than 0.9M), ( Figure 4 1 ). The methanol concentration in the fuel solution was determined by developing a relationship between the performance of the stack during the air break and the methanol concentration in the fuel solution over a range of methanol concentrations of 0.3 1.5 M and a range of stack operating temperatures of 45 55 C, which resulted in an average of difference of 0.068 M with respect to the measureme nts of a methanol sensor. It is believed that this error is acceptable for a methanol concentration method. The effect of stack operating temperature, voltage during the air break, and life time of the stack on this method were studied. The current dens ity during the air break was found to be inversely related to the stack operating temperature ( Figure 4 4 ). This is mainly attributed to the higher rate of methanol cr ossover at higher operating temperatures. Reducing the air break voltage generally increased the current density during the air break at a particular concentration ( Figure 4 3 ). Regarding the effect of the stack operating time, a solid conclusion may not be expressed at this point, since a comprehensive set of experiments including several fuel cell stacks is required; however, for the particular stack used in this work, the current density concentration curve remained fairly consistent for stack operating time greater than 650 hours ( Figure 4 5 ). Moreover, the consistency of th e performance during the air break periods was statistically analyzed over a range of methanol concentrations of 0.3 1.5 M and over 50

PAGE 75

75 consecutive air break periods and It was found that the measurements had a normal distribution and there was no systemati c variation in the measurements for a particular concentration. This work discussed a new method to determine the methanol concentration in an open cathode DMFC system that has potential to be used as a methanol sensing strategy because of its unique features. This method does not require any auxiliary device since it is using the readily available data during the air break period, which is a conventional strategy to recover the cathode potential. Also, this method could be particularly suited for use as a secondary sensor by which the performance of the primary methano l sensor can be examined. Future work co uld focus on the effect of the MEA structure on methanol crossover. This could include the thickness of the membrane which is inversely related to the methanol crossover or the structure of the cathode side, particu larly the barrier layer which affects the oxygen diffusivity to the cathode catalyst layer. These two could significantly influence the current density during the air break. Also, better understanding of the effect o f the stack life time on this method mus t be a priority in the future works. In addition the reproducibility of this method can be evaluated by conducting identical test experiments on multiple stacks. The effect of the ambient air properties such as pressure and temperature, and the stack or ientation need to be studied as well.

PAGE 76

76 APPENDIX A A DISCUSSION ABOUT THE SHAPE OF THE CURVES The purpose of this chapter is to discuss the effective parameters that shape the air break curre nt density concentration curve and then experimentally proving the discussed parameters. Also, measuring the diffusion limited methanol oxidation current as a method to measure the methanol diffusion coefficient in the membrane will be discussed. Factors Affecting t he Shape of the Curves The first and possibly most important aspect of the curves shown in Figure 4 9 is that the current density during the air break is inversely rel ated to the methanol concentration in the fuel. Howev er, this inverse relationship is more distinct in the lower methanol concentration region (less than 0.9 M). The reason for this trend can be attributed to the fact the rate of the methanol crossover is a strong function of the methanol concentration in the anode side; hence, an increase in the methanol concentration in the solution fuel increases the rate of the methanol crossov er to the cathode side. D uring the air break in an open cathode DMFC, a limit ed amount of air flow exists in the cathode channels mainly due to the buoyancy effect and the temperature gradient between the fuel cell stack and environment. The oxygen carried in the cathode channel by the limited air flow is mainly consumed in two dif ferent reactions: the cathode reaction (ORR) and methanol cross over oxidation reaction (methanol combustion) at the presence of the cathode catalyst. The higher rates of the methanol cross over means that a larger portion of the limited available oxygen i s consumed for the methanol oxidation reaction and less oxygen is left for the cathode reaction and consequently less current density is observed. The methanol oxidation

PAGE 77

77 reaction is not advantageous in terms of current generation and it is a waste of fuel. However, t he fact that the methanol cross over is a function of the methanol concentration in the anode fuel is the main reason that the current density during the air break is a function of the solution fuel concentration. Methanol Crossover Modeling A s 52 ] of diffusion defined as: ( A 1 ) where J (mol/m 2 .s) is the diffusion flux, D (m 2 /s) is diffusion coefficient, and dC/dx (mol/m 3 ) is the concentration gradient. The diffusion coefficient of methanol through the MEA can be calculated by measuring the diffusion limited methanol oxidation current. In order to conduct this experiment, the fuel cell cathode is purged with nitrogen and the fuel cell is connected t o a power supply. This means that the 8 cell stack is not further operating a s a galvanic cell which produces electricity; in fact, the 8 cell stack operates as a electrolytic ce ll in which methanol diffuses through the membrane and is oxidized within the cathode catalyst layer. The protons and the electrons liberated in the methanol oxidation reaction react within the anode catalyst layer which results in hydrogen formation in th e anode side. The details of this experiment are shown in Figure A 1 The experiment starts with setting the power supply voltage to zero. By gradually increasing the voltage controlled by a power supply, the current eventually starts increasing which is the result of the methanol oxidation in the cathode and hydrogen evoluti on in the anode side. A fter a certain point, increases in the supplied voltage do not increase the current [ 53 ] This maximum current value is limited by the amount of methanol that can diffuse from the

PAGE 78

78 anode to the cathode through the membrane. By measuring the diffusion li mited current and using Equation A 2 the rate of the methanol diffusion can be calculated. ( A 2 ) The result of this experiment is shown in Figure A 2 Note that as it was expected, the diffusion limited current, in other words, methanol cros s over, is directly related to the stack temperature and the solution co ncentration in the anode sid e. The calculated methanol diffusion coefficient at a stack temperature of 50 C is 5.649 10 10 (m /s) which is the av erage of the values is shown in Table A 1 In diffusion coefficient calculations, it was assumed that the methanol concentration at the ADL/ACL interface ( Figure A 3 ) was equal to the bulk concentration and the methanol concentration at the membrane/CCL is equal to zero. So, the thickness in Table A 1 is the combined thickness of the membrane and the anode catalyst layer and the calculated diffusion coefficient is the combine diffusion coefficient for the membrane and the anode catalyst layer. Theoretical Air Flow in the Cathode Channels T he limited ox ygen flow to the cathode channels during the air break is mainly consumed either for the cathode reaction (creating current), or in the crossover methanol combustion reaction. By knowing the diffusion coefficient of methanol through the membrane and also t he available data for the current density during the air break for a range of solution concentration, the required amount of the oxygen flow into the cathode channels during the break can be calculated. Figure A 4 shows the calculated

PAGE 79

7 9 required oxygen flow through the cathode channels during the air break for the stack operating temperature of 50 C and 0.4 V/cell during the air break. In Figure A 4 the red curve represents the required oxygen for the crossover methanol oxidation reaction. Clearly, as the solution concentration increases, a larger portion of the total required oxygen flow (the green curve) is consumed for t he crossover methanol oxidation reaction. Also, the blue curve represents the required oxygen flow rate for t he cathode reaction (oxygen reduction reaction) and it was calculated using the measured current density value for various solution concentrations and Eq uation A 2 Measuring the Oxygen Flow Rate in the Cathode Channels In order to measure the oxygen flow rate through the cathode channels during the air breaks, a simple experiment was conducted. The cathode channels were initially purged with nitrogen, which causes the voltage to drop to a value around 50 100 mV/cell. Adding air flow to the nitrogen stream ( Figure A 5 ) would increase the cathode potential therefore the stack voltage. However, if the stack voltage is set to constant value such as 0.4 V/cell (same as the c onditions during the air break), adding more air flow to the nitrogen stream initially brings up the voltage to 0.4 V/cell. Once the stack voltage raised to 0.4V/cell, adding more air flow to the nitrogen stream would generate more current. For any particu lar solution concentration, the air was gradually added to the nitrogen stream until the measured current density value for that particular solution concentration ( Figure 4 9 ) was observed. Figure A 6 shows the measured values for the oxygen flow rate through the cathode channels during the air break period in comparison with the theoretical calculated values. The measured value curve has the same trend with the theoretical values, which proves the concept in terms of the effect of the methanol crossover on

PAGE 80

80 current density during the air break. However, the measured values are abou t 50% larger than the calculated values. This is attributed the fact that the theoretical values were calculated at the CCL/BL interface ( Figure A 3 ) and the real val ues were measured in the cathode channels (right hand side of the barrier layer in Figure A 3 ). The barrier layer is basically a carbon paper layer (GDL) coated with several layers of Teflon based coating and is impermeable to liquid water. Utilizing the ba rrier layer is a strategy to help the water management by keeping a portion of the water generated in the cathode reaction within the MEA and pushing it back to the anode through the important to mention that the barrier layer is a clever strategy to help the water management, but on the other hand it reduces the oxygen accessibility to cathode catalyst layer. This clearly explains the reason why the measured values are higher than the theoretical calculated values.

PAGE 81

81 Figure A 1 Diffusion limited methanol oxidation current experiment

PAGE 82

82 Figure A 2 Diffusion limited oxidation current at various anode solution concentrations and stack temperatures

PAGE 83

83 Figure A 3 The structure of MEA

PAGE 84

84 Figure A 4 Calculated required oxygen flow rate in the cathode channels during the air break periods as a function solution concentration at 50 C and 0.4 V rest

PAGE 85

85 Figure A 5 Test set up for measuring the air flow in the cathode channels during the air break

PAGE 86

86 Figure A 6 The comparison between the theoretical calculate curve (green curve) and the measured values (blue curve) of the oxygen flow in the cathode channels during the air break, at 50 C, and 0.4V rest.

PAGE 87

87 Table A 1 Calculated methanol diffusion coefficient at 50 C Solution concentration (M) Diffusion limited curren t (A) Methanol crossover flux (mol/m.s) Thickness (m) Diffusion coefficient (m/s) 0.5 7.40E 01 8.25E 04 4.00E 05 6.60E 11 0.8 9.24E 01 1.03E 03 4.00E 05 5.15E 11 1.1 1.28E+00 1.43E 03 4.00E 05 5.20E 11

PAGE 88

88 APPENDIX B STATISTICAL ANALYSIS Table B 1 Statistical analysis of the data shown in Figure 4 8 Solution concentration (M) Calculation based on Probability that a sample falls in [m Probability that a sample falls in [m Pro bability that a sample falls in [m 0.3 15th second 65.31% 97.96% 100.00% 30th second 77.55% 100.00% 100.00% 45th second 67.35% 100.00% 100.00% 0.5 15th second 71.43% 97.96% 100.00% 30th second 79.59% 97.96% 100.00% 45th second 85.71% 100.00% 100.00% 0.7 15th second 67.35% 100.00% 100.00% 30th second 73.47% 100.00% 100.00% 45th second 83.67% 100.00% 100.00% 0.9 15th second 71.43% 93.88% 100.00% 30th second 73.47% 97.96% 100.00% 45th second 59.18% 97.96% 100.00% 1.1 15th second 67.35% 95.92% 97.96% 30th second 63.27% 97.96% 100.00% 45th second 69.39% 97.96% 100.00% 1.3 15th second 69.39% 97.96% 100.00% 30th second 63.27% 97.96% 100.00% 45th second 59.18% 95.92% 100.00% 1.5 15th second 75.51% 97.96% 100.00% 30th second 75.51% 93.88% 100.00% 45th second 69.39% 97.96% 100.00%

PAGE 89

89 LIST OF REFERENCES [ 1 ] R. Rashidi, I. Dincer, G.F. Naterer, P. Berg, J.Power Sources, 187 (2009) 509 516. [ 2 ] I. Horuz, J. Fletcher, C. C. Kuo, S. Credle, W. Lear (2008), pp. AIAA 2008 5667. [ 3 ] R. O'Hayre, S. Cha, W. Colella, F. Prinz, Introduction, Fuel Cell Fundementals, 2nd ed., Wiley, 2009, pp. 3 25. [ 4 ] J. Larminie, A. Dicks, Operati onal Fuel Cell Voltages, Fuel Cell Systems Explained, 2nd ed., Wiley, 2003, pp. 45 66. [ 5 ] S.K. Kamarudin, F. Achmad, W.R.W. Daud, Int J Hydrogen Energy, 34 (2009) 6902 6916 [ 6 ] J. Chang, Y. Kuan, S. Lee, S. Lee, J.Power Sources, 184 (2008) 180 190 [ 7 ] L. Hansan, Z. Jiujun, Electrocatalysis of direct methanol fuel cells, Wiley, 2009, pp. 25. [ 8 ] L. Maldonado, J. Perrin, J. Dillet, O. Lottin, J.Membr.Sci., 389 (2012) 43 56. [ 9 ] V. Neburchilov, J. Martin, H. Wang, J. Zhang, J.Power Sources, 169 (2007) 221 238. [ 10 ] P. Ferrin, A.U. Nilekar, J. Greeley, M. Mavrikakis, J. Rossmeisl, Surf.Sci., 602 (2008) 3424 3431. [ 11 ] M. Hogarth, T. Ralph, In Johnson Mattey Technology Center, Reading, (2002) 146 164. [ 12 ] Fletcher J. Advanced direct methanol fuel cell for mobile computing. 2012; Available at: http://www.hydrogen.energy.gov/pdfs/review12/h2ra004_fletcher_2012_p.pdf [ 13 ] H. Zhao, J. Shen, J. Zhang, H. Wang, D.P. Wilkinson, C.E. Gu, J.Power Sources, 159 (2006) 626 636 [ 14 ] T. Schultz, S. Zhou, K. Sundmacher, Chem.Eng.Technol., 24 (2001) 1223 1233. [ 15 ] L. Hansan, Z. Jiujun, Electrocatalysis of direct methanol fuel cells Wiley, 2009, pp. 503. [ 16 ] C. Eickes, P. Piela, J. Davey, P. Zelenay, J. Electrochem. Soc., 153 (2006) A171 A178.

PAGE 90

90 [ 17 ] J. Prabhuram, T.S. Zhao, H. Yang, J Electroanal Chem, 578 (2005) 105 112 [ 18 ] S.R. Narayanan, T.I. Valdez, W. Chun, Electrochem. Solid S tate Lett., 3, issue 3 (2000) 117 120. [ 19 ] S.R. Narayanan, W. Chun, T.I. Valdez, US Patent 6306285, (2001). [ 20 ] J. Geng, X. Li, G. Sun, B. Yi, Sensors Actuators B: Chem., 147 (2010) 612 617. [ 21 ] M.S. Lee, J. Sohn, J. Shim, W.M. Lee, Sensors Actuators B: Chem., 124 (2007) 323 328. [ 22 ] K. Jeng, W. Huang, C. Chien, N. Hsu, Sensors Actuators B: Chem., 125 (2007) 278 283. [ 23 ] T. Kumagai, T. Horiba, T. Kamo, S. Takeuchi, K. Iwamoto, K. Kitami, K. Tamura, US Patent 4810597, (1989). [ 24 ] G. Luft, G. Starbeck, U S patent 5,624,538, (1997). [ 25 ] J. Zhang, K.M. Colbow, D.P. Wilkinson, J. Mller, US patent 6527943, (2003). [ 26 ] Z. Guo, A. Faghri, Appl.Therm.Eng., 28 (2008) 1605 1613. [ 27 ] K. Lee, J. Lee, S. Kim, B. Ju, Carbon, 49 (2011) 787 792. [ 28 ] P. Enoksson, G. Stemme, E. Stemme, Sensors and Actuators A: Physical, 47 (1995) 327 331. [ 29 ] D. Sparks, K. Kawaguchi, M. Yasuda, D. Riley, V. Cruz, N. Tran, A. Chimbayo, N. Najafi, Sensors and Actuators A: Physical, 145 146 (2008) 9 13. [ 30 ] T. Sato, A. Chiba, R. Nozaki, Journal of Molecular Liquids, 101 (2002) 99 111. [ 31 ] S. Doerner, T. Schultz, T. Schneider, K. Sundmacher, P. Hauptmann, Proceedings of the IEEE Sensors 2004. Vienna, Austria, vol. 2, pp. 639 641. [ 32 ] M. Baldauf, W. Preidel, US Patent 6536262, (2003). [ 33 ] R. Jiang, D. Chu, J. Electrochem. Soc. 151 (2004) A69 A76. [ 34 ] A. Ozawa, A. Minamisawa, Jpn. J. Appl. Phys, 37 (1998) 2799. [ 35 ] A. Rabinovich, D. Tulimieri, US Patent 6748793, (2004).

PAGE 91

91 [ 36 ] C.C. Sung, Y.L. Tseng, Y.F. Chiang, C.Y. Chen, Sensors and Actuato rs A: Physical, 161 (2010) 101 107. [ 37 ] A. Rabinovich, E. Diatzikis, J. Mullen, D. Tulimieri, US Patent 6815682, (2004). [ 38 ] J.P. Longtin, C.H. Fan, Microscale Thermophys. Eng. 2 (1998) 261 272. [ 39 ] H. Kodama, S. Kitajima, T. Suga, Japanese patent JP32517 45, (1991). [ 40 ] Baldauf, US Patent 20020148284, (2002). [ 41 ] J. Kondoh, S. Tabushi, Y. Matsui, S. Shiokawa, Sensors Actuators B: Chem., 129 (2008) 575 580. [ 42 ] Q. Mao, U. Krewer, Electrochim.Acta, 68 (2012) 60 68. [ 43 ] Y. Chiu, H. Lien, J.Power Sources, 159 (2006) 1162 1168. [ 44 ] K. Shen, C. Wan, Y. Wang, T.L. Yu, Y. Chiu, J.Power Sources, 195 (2010) 4785 4795. [ 45 ] T.J. Ha, J. Kim, H. Joh, S. Kim, G. Moon, T. Lim, C. Han, H.Y. Ha, Int J Hydrogen Energy, 33 (2008) 7163 7171. [ 46 ] S. Arisetty C.A. Jacob, A.K. Prasad, S.G. Advani, J.Power Sources, 187 (2009) 415 421. [ 47 ] C.Y. Chen, D.H. Liu, C.L. Huang, C.L. Chang, J.Power Sources, 167 (2007) 442 449. [ 48 ] C.L. Chang, C.Y. Chen, C.C. Sung, D.H. Liou, J.Power Sources, 164 (2007) 606 613. [ 49 ] C.L. Chang, C.Y. Chen, C.C. Sung, D.H. Liou, J.Power Sources, 182 (2008) 133 140. [ 50 ] C.L. Chang, C.Y. Chen, C.C. Sung, D.H. Liou, C.Y. Chang, H.C. Cha, J.Power Sources, 195 (2010) 1427 1434. [ 51 ] W Harrington, Investigation of d irect m ethanol f uel c ell v oltage r esponse for m ethanol c oncentration s ensing 2012, University of Florida. [ 52 ] C.O. Bennet, J.E. Myers, Molecular diffusion and diffusivity, Momentum, heat, and mass transfer, 2nd ed., McGraw Hill, 1974, pp. 481 491

PAGE 92

92 [ 53 ] S.C. Thomas, X. Ren, S. Gottesfeld 2, P. Zelenay, Electrochim.Acta, 47 (2002) 3741 3748.

PAGE 93

93 BIOGRAPHICAL SKETCH Amir Reza Shoraka was born in Tehran, Iran in 1983. In 2009, he graduated with a bachelor degree in mechanical engineering from the University of North Florida. His interest in thermal sciences and renewable energy led him to start graduate school with focus on thermal sci ences and working as a graduate research assistant under guidance of Dr. James Fletcher for a direct methanol fuel cell project at the University of North Florida. Amir graduated with a master degree in mechanical engineering from the University of Florida in summer of 2012.