<%BANNER%>

Manufacturing Evaluation of a Gas Distribution Layer for a Direct Methanol Fuel Cell

Permanent Link: http://ufdc.ufl.edu/UFE0043828/00001

Material Information

Title: Manufacturing Evaluation of a Gas Distribution Layer for a Direct Methanol Fuel Cell
Physical Description: 1 online resource (64 p.)
Language: english
Creator: Pope, Mark A
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: cell -- direct -- distribution -- evaluation -- fuel -- gas -- layer -- manufacturing -- methanol
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: A gas distribution layer (GDL) and membrane for a direct methanol fuel cell (DMFC), which is being developed under a collaboration between the University of North Florida (UNF) and the University of Florida (UF), were investigated from a viewpoint of manufacturing quality. The thickness of the GDL, which is currently produced by hand-painting ink layers on a carbon fiber substrate, was measured using a novel measurement apparatus that was designed and constructed as part of this work. It was found that the ink was applied to a thickness with an average standard deviation of 13.2 µm, which was approximately equal to the thickness variation of the GDL carbon fiber substrate. A current challenge for the DMFC implementation is potential degradation of the fuel cell by surfactants in the ink. These surfactants are inherently present in the ink formulation and must be removed by sintering. In this study, the sintering process was modified to increase the percent surfactant removal. Modifications included changing the heating cycle and altering the sintering environment. Surfactant removal was quantified by weighing samples before and after the sintering process. The DMFC polymer membrane, which controls the methanol diffusivity, water permeability, and allows H+ conduction, was also studied. Holes were laser drilled by UNF into the membrane to allow the appropriate amount of water to travel across the membrane from the cathode to anode. The hole size and shape was examined using a scanning white light interferometer (SWLI). It was found that the hole diameters differed from one side of the membrane to the other and redeposited material (from the laser ablation process) was present. However, the hole diameters on the backside were within the prescribed range of 5-8 µm and the shape was acceptably circular.
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 Mark A Pope.
Thesis: Thesis (M.S.)--University of Florida, 2011.
Local: Adviser: Lear, William E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043828:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043828/00001

Material Information

Title: Manufacturing Evaluation of a Gas Distribution Layer for a Direct Methanol Fuel Cell
Physical Description: 1 online resource (64 p.)
Language: english
Creator: Pope, Mark A
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: cell -- direct -- distribution -- evaluation -- fuel -- gas -- layer -- manufacturing -- methanol
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: A gas distribution layer (GDL) and membrane for a direct methanol fuel cell (DMFC), which is being developed under a collaboration between the University of North Florida (UNF) and the University of Florida (UF), were investigated from a viewpoint of manufacturing quality. The thickness of the GDL, which is currently produced by hand-painting ink layers on a carbon fiber substrate, was measured using a novel measurement apparatus that was designed and constructed as part of this work. It was found that the ink was applied to a thickness with an average standard deviation of 13.2 µm, which was approximately equal to the thickness variation of the GDL carbon fiber substrate. A current challenge for the DMFC implementation is potential degradation of the fuel cell by surfactants in the ink. These surfactants are inherently present in the ink formulation and must be removed by sintering. In this study, the sintering process was modified to increase the percent surfactant removal. Modifications included changing the heating cycle and altering the sintering environment. Surfactant removal was quantified by weighing samples before and after the sintering process. The DMFC polymer membrane, which controls the methanol diffusivity, water permeability, and allows H+ conduction, was also studied. Holes were laser drilled by UNF into the membrane to allow the appropriate amount of water to travel across the membrane from the cathode to anode. The hole size and shape was examined using a scanning white light interferometer (SWLI). It was found that the hole diameters differed from one side of the membrane to the other and redeposited material (from the laser ablation process) was present. However, the hole diameters on the backside were within the prescribed range of 5-8 µm and the shape was acceptably circular.
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 Mark A Pope.
Thesis: Thesis (M.S.)--University of Florida, 2011.
Local: Adviser: Lear, William E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043828:00001


This item has the following downloads:


Full Text

PAGE 1

1 MANUFACTURING EVALUATION OF A GAS DISTRIBUTION LAYER FOR A DIRECT METHANOL FUEL CELL By MARK ALLEN POPE 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 2011

PAGE 2

2 2011 Mark Allen Pope

PAGE 3

3 To my wife and children

PAGE 4

4 ACKNOWLEDGMENTS I thank my wife, Darcy, and my children for their support and love throughout my time at the University of Florida I cannot express in words how much you all have added to my life. I also thank my advisor, Dr. Tony L. Schmitz, for his support, encouragement, and flood of helpful ideas throughout my research. Also, I thank the members of my committee, Dr. John Schuell er and Dr. William Lear, for their interest and involvement with my project. Last ly I thank the members of the Machine Tool Research Center for their advice and ideas. I especially thank Mr. Daniel Blood and Mr. Christopher Tyler for their assistance

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREV IATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 13 2 LITERATURE REVIEW ................................ ................................ .......................... 21 3 EXPERIMENTAL DESIGN ................................ ................................ ..................... 23 Non Contact Measurement ................................ ................................ ..................... 23 Capacitance Probe ................................ ................................ ................................ 24 Capacitance ................................ ................................ ................................ ..... 24 Calibration ................................ ................................ ................................ ........ 25 Measurement Errors ................................ ................................ ......................... 25 Parallelism ................................ ................................ ................................ 25 Flatness ................................ ................................ ................................ ..... 26 Environment ................................ ................................ ............................... 26 Vacuum Chuck ................................ ................................ ................................ ....... 27 Pre Machining Preparation ................................ ................................ ............... 27 Machining Parameters ................................ ................................ ..................... 27 Vacuum Bowing Effect ................................ ................................ ..................... 28 Scanning White Light Inter ferometer ................................ ................................ ...... 29 Programmable Furnace ................................ ................................ .......................... 30 4 EXPERIMENTAL PROCEDURE ................................ ................................ ............ 36 Thickness Measurement ................................ ................................ ......................... 36 SWLI Data Validation ................................ ................................ .............................. 36 Sintering Procedure ................................ ................................ ................................ 37 GDL Ink Production and Painting ................................ ................................ ............ 37 Membrane Hole Diameter Measurement ................................ ................................ 37 5 DATA ANALYSIS AND VALIDATION ................................ ................................ ..... 40 Thickness Calculation Algorithm ................................ ................................ ............. 40

PAGE 6

6 Sintering Loss ................................ ................................ ................................ ......... 41 SWLI Validation Data Filtering ................................ ................................ ................ 42 Comparison Tests ................................ ................................ ................................ ... 43 6 RESULTS ................................ ................................ ................................ ............... 49 Capacitance Probe GDL Data ................................ ................................ ................ 49 Sintering Results ................................ ................................ ................................ ..... 50 Membrane Hole SWLI Measurements ................................ ................................ .... 51 7 CONCLUSIONS AND FUTURE WORK ................................ ................................ 57 APPENDIX A LABVIEW PROGRAMS ................................ ................................ .......................... 59 B MATLAB PROGRAM ................................ ................................ .............................. 61 LIST OF REFE RENCES ................................ ................................ ............................... 63 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 64

PAGE 7

7 LIST OF TABLES Table page 5 1 The components of the PTFE based ink. ................................ ........................... 44 5 2 Micrometer, SWLI, and capacitance probe thickness of 1.27 mm thick gage block. ................................ ................................ ................................ .................. 44 6 1 GDL thickness data. ................................ ................................ ........................... 52 6 2 Results of sintering using UNF's original pr ocedure. ................................ .......... 52 6 3 Sintering results using the "literature" process. ................................ .................. 52 6 4 Air convection sintering data. ................................ ................................ .............. 53 6 5 Sintering results from GDLs sintered layer by layer ................................ ........... 53 6 6 Membrane hole diameters from the top view. ................................ ..................... 53 6 7 Membrane hole diameters from the bottom view. ................................ ............... 53

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Gas distribution layer (120 mm x 220 mm). ................................ ........................ 17 1 2 Schematic of traditional and simplified UNF DMFC. Courtesy of University of North Florida. ................................ ................................ ................................ ...... 18 1 3 Liquid barrier layer (LBL) applied to GDL (120 mm x 220 mm). ......................... 19 1 4 The GDL curls during sintering. ................................ ................................ .......... 19 1 5 20 m thick polymer membrane (120 mm x 220 mm). ................................ ....... 20 3 1 Capacitance probe measurements. A) Displays the error that occurs with a non flat target, B) a measurement of a fla t targe t ................................ ............... 30 3 2 Lion Precision CPL290 capacitance probe amplifier. ................................ ......... 30 3 3 Capacitance probe holder fabricated of aluminum and nylon. ............................ 31 3 4 JB Industries DV 285N 10 CFM Vacuum pump ................................ ................. 31 3 5 150 x 300 mm NewWay air bearing. ................................ ................................ ... 32 3 6 Custom aluminum vacuum chuck. ................................ ................................ ...... 32 3 7 Mikron U CP600 Vario 5 Axis CNC mill. ................................ .............................. 33 3 8 Isometric view of the 1.6 mm diameter hole vacuum chuck data. ....................... 33 3 9 TiAlN coated drill (0.8 mm diameter). ................................ ................................ 34 3 10 Ribbed aluminum support for vacuum chuck (104 x 186 mm). ........................... 34 3 11 Schematic of SWLI system ................................ ................................ ................ 35 3 12 Lindberg/Blue M programmable oven interior. ................................ .................... 35 4 1 Thorlabs long travel programmable stages (LTS). ................................ .............. 38 4 2 0.01 mg resolution scale with GDL placed upright i n enclosed measurement area. ................................ ................................ ................................ ................... 39 5 1 Velocity profile of the stage. The commanded velocity was 10 mm/s. ................ 45 5 2 3 Dimensional plot of the unfiltered SWLI GDL surface data. ............................ 45

PAGE 9

9 5 3 Plot of the filtered SWLI GDL surface data. ................................ ........................ 46 5 4 SWLI height map of GDL. The scanned area is 2.83 mm long by 2.12 mm wide. ................................ ................................ ................................ ................... 46 5 5 3 D plot from the capaci tance probe data. Colorbar units are in m. This image is a small fraction of the size of Figure 6 1. ................................ .............. 47 5 6 Height map of the 1.27 mm thick gage block and surrounding area from the capacitance probe setup. Note the averaging effect near the gage block edges. ................................ ................................ ................................ ................. 47 5 7 The 1.27 mm gage block thickness profile from the SWLI. ................................ 48 6 1 GDL surface map sample of GDL 11. Mean thickness of 282. 7 m and standard deviation of 13.2 m. ................................ ................................ ........... 54 6 2 GDL surface map sample of GDL 13. Mean thickness of 311.5 m and standard deviatio n of 14.9 m. ................................ ................................ ........... 54 6 3 GDL thickness versus number of ink layers. ................................ ....................... 55 6 4 Top view oblique plot of a membrane hole. Window size is 175 by 130 m. ...... 55 6 5 Bottom view of a laser cut membrane hole. Window size is 175 by 130 m. ..... 56 A 1 LabVIEW block diagram of motor controller. ................................ ...................... 59 A 2 Schematic of a LabView VI that gathers stage position data. ............................. 60

PAGE 10

10 LIST OF ABBREVIATION S DMFC Direct Methanol Fuel Cell GDL Gas Distribution Layer LBL Liquid Barrier Layer MTRC Machine Tool Research Center PTFE Polytetrafluoroethylene SWLI Scanning White Light Interferometer

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 MANUFACTURING EVALUATION OF A GAS DISTRIBUTION LAYER FOR A DIRECT METHANOL FUEL CELL By Mark Allen Pope December 2011 Chair: Tony L. Schmitz Major: Mechanical Engineering A gas distribution layer (GDL) and membrane for a direct methanol fuel cell (DMFC) which is being developed under a collaboration between the University of North Florida (UNF) and the University of Florida (UF) were investigated from a viewpoint of manufacturing quality. The thickness of the GDL, which is currently produced by hand painting ink layers on a carbon fiber substrate was measu red using a novel measurement apparatus that was designed and constructed as part of this work It was found that the ink was applied to a thickness with a n average standard deviation of 13.2 m which was approximately equal to the thickness variation of the GDL carbon fiber substrate A current challenge for the DMFC implementation is potential degradation of the fuel cell by surfactants in the ink These surfactants are inherently present in the ink formulation and must be removed by sintering. In this study, the sintering process was modified to increase the percent surfactant removal Modifications included changing the heating cycle and altering the sintering environment. Surfactant removal was quantified by weighing samples before and after t he sintering process.

PAGE 12

12 The DMFC polymer membrane which control s the methanol diffusivity water permeability and allows H + conduction, was also studied Holes were laser drilled by UNF into the membrane to allow the appropriate amount of water to travel across the membrane from the cathode to anode. The hole size and shape was examined using a scanning white light interferometer (SWLI). It was found that the hole diameters differed from one side of the membrane to the other and redeposited material (from the laser ablation process) was present. However, the hole diameter s on the backside were within the prescribed range of 5 8 m and the shape w as acceptably circular.

PAGE 13

13 CHAPTER 1 INTRODUCTION The rising cost of fuel serves as a constant motivation for the development of alternative forms of energy in the current world climate. Crude oil, from which gasol ine is refined, has a variable cost and availability due to many factors not related to oil production To reduce energy conversion concepts have been proposed both in the United States and abroad. One option is t he fuel cell which operates by converting the chemical energy of a fuel into electrical energy. The electricity is produced by the reaction between the fuel and an oxidizing agent. Hydrogen fuel cells, which use hydrogen as the fuel and air as the oxidant, have been in development for several years and provide a potential alternative to the internal combustion engine as well as o ther applications Th e focus of th is thesis is the completion of measurements for critical components in a new direct methanol fuel cell (DMFC) under development at the University of North Florida (UNF) and the University of Florida (UF) This fuel cell wi ll be used for small portable electronic devices, such as laptops. A traditional DMFC contains several components that are necessary for the active water partition of the cell. This partition is needed, because the cathode reaction creates three water mo lecules for every methanol molecule consumed, while the anode reaction only req uires one molecule of water as shown in Eqs. 1 1 and 1 2 (1 1) (1 2) Therefore, two water molecules must be removed and exhausted. In the UNF design, a gas distribution layer (GDL) Figure 1 1 is used to allow the prescribed

PAGE 14

14 amount of air vapor to cross the cathode, while the liquid barrier layer (LBL) restricts the flow to one water molecule per methanol molecule consumed. The combination of the GDL and LBL which will be called the GDL assembly, allow passive water partition of the fuel cell, removing several power hungry components such as the heat exchanger, water pump, and air pump that are typically included i n a DMFC ( Figure 1 2 ) The removal of these components reduce the complexity of the fuel cell and increase the energy and power densities to the performance level of lithium ion (Li ion) batteries. The performanc e increase has improved the viability of DFMCs as a commercial alternative to Li i on batteries according t o studies completed at UNF [1] The GDL is made of a carbon fiber substrate that is formed into a porous paper. A LBL consisting of a polytetrafluoroe thylene ( PTFE ) based ink is painted onto the GDL. T he water transport capability of the fuel cell is directly proportional to t he combined thickness of these two components. The GDL has a uniform weight of 1.5 mg with a standard deviation of 0.02 mg The h and painted LBL weight, and consequently thickness, is not as consistent and serves as the main contributor to variations in water transport performance. The ink is painted on in several steps First the GDL is painted with the PTFE ink Next, t he GDL is dried at 70 C to remove any excess water. This process is then repeated for a total of eight layers ( Figure 1 3 ) A primary goal of this work was the development of a platform for non contact GDL assembly thickness measurement. Due to the 120 mm x 220 mm size of the GDL assembly non contact devices with a small measurement area, such as a scanning white light interferometer (SWLI) were not practical for characteriz ation of the large

PAGE 15

15 surface area of the GDL assembly due to the significant measurement time The refore, a capacitance probe setup with fast scanning capabilities was selected After the eight layer LBL buildup, the PTFE is sintered to remove unwanted com ponents; this causes the GDL assembly to curl at the edges as seen in Figure 1 4 The curling presented a unique challenge that required a custom measurement setup to acquire accurate data. T wo possible methods for obtaining a flat surface were identified: 1) p ulling the sample down ; or 2) pressing it down. selected and implemented using a custom made vacuum chuck, which was designed to suction down the entire GDL assembly The thickness measurement procedure is described in detail in Chapter 3. The PTFE based ink contains surfactants, which are chemicals that permit the PTFE to be made into an ink suitable for painting. Surfactants in the PTFE emulsion are added in the form of a nonionic thickener which provides the appropriate viscosity for pain ting These surfactants are a potential danger to the cell and may contribute to off state cell degradation as evidenced by recent UNF studies Since the surfactants can not be omitted from the ink formulation, they must be removed during the sintering proc ess. The original sintering profile consisted of a quick ramp to an intermediate temperature which was then increased slowly up to the PTFE curing temperature. Several solutions were considered including sintering under an inert atmosphere, convection sintering using air, and layer by layer sintering To sinter layer by layer, a new batch of ink needed to be mixed for each layer due to the short (~12 hr ) sh elf life of the ink. The GDL will be pa inted with one coat of ink sintered, and this process will be repeated eight times for the total layer build up.

PAGE 16

16 Another component which is vital to DMFC performance is the polymer membrane ( Figure 1 5 ) Th is membrane controls the water permeability and methanol diffusivity for the cell Many comm ercially available membranes require water pumps to provide the needed amount of water transport from the cathode to anod e side of the fuel cell. The membrane used in this work was engineered to passively provide the necessary amount of water permeability, which remov ed the water pump requirement and, in turn increases the fuel cell energy density. This new membrane achieves the required amount of water permeability at a hydraulic pressure of only 34.5 kPa ( 5 psi ), compared to the 345 kPa (50 psi) to 690 kPa (100 psi ) required for commercially available membranes. Correct operation of the UNF DMFC requires specific characteristics different from those in traditional DMFCs. The refore, a post processing step was added to the membrane production where holes are laser drilled into the me m brane These holes create additional challenges to t he performance management of the cell. The laser drilled h ole size and shape were investigated using a SWLI in this work H ole size affects water flux and may also affect GDL assembly hydrostatic pressure requirements ; additionally hole defects may lead t o the formation of electrical shorts through the membrane when the electrodes are bonded together. This thesis is organized as follows. Chapter 2 provides background information on DMFCs, with an emphasis on passive water partition and sintering. Chapter 3 contains descriptions of the experimental apparatus developed /used for this project. In Chapter 4, the experimental procedures are defined for the GDL assembly thickness measurements, PTFE ink mixing and sintering, and membrane ho le surface maps. Data

PAGE 17

17 analysis and validation are described in Chapter 5. Results of the thickness measurements, sintering, and membrane hole characteristics are presented in Chapter 6. Chapter 7 provides conclusions and future work Figure 1 1 Gas distribution layer ( 120 mm x 220 mm ) Photo courtesy of Mark Pope.

PAGE 18

18 Figure 1 2 Schematic of tradi ti onal and simplified UNF DMFC Courtesy of University of Nort h Florida.

PAGE 19

19 Figure 1 3 Liquid barrier layer (LBL) applied to GDL (120 mm x 220 mm) Photo courtesy of Mark Pope. Figure 1 4 The GDL assembly curls during sintering. Photo courtesy of Mark Pope.

PAGE 20

20 Figure 1 5 20 m thick p olymer membrane ( 120 mm x 220 mm ) Photo courtesy of Mark Pope.

PAGE 21

21 CHAPTER 2 LITERATURE REVIEW The direct methanol fuel cell (DMFC) is a developing area in the field of alternative energy and is a possible alternative to the lithium based rechargeable battery for the portable electronics market [2 ] There are two basic types of DMFC s : active and passive. Active DMFCs generally have a higher reliability and performance than the passive system s Many active systems incorporate a water pump into the design to provide circulation of the water to dilute the methanol reactant, which results in a more efficient system. Meanwhile, passive DMFC s, like the UNF design investigated here operate without the use of external devices such as a methanol / water pump [2 ]. Oxygen diffuses into the open cathode from the ambient air and methanol diffuses into the anode from a feed reservoir that is driven by a concentration gradient prescribed by the system. In addition to the simplicity and lower cost, passive DMFCs offer the possibility for lower parasitic power loss and system volume [2 ]. There are several barriers to overcome before passive DMFCs can be a commercially viable product. Methanol crossover: While high methanol concentration provides higher achievable energy density [3], it also causes severe methanol crossover to the electrolyte membrane and results in mix potential at the cathode and there fore, causes low cell performance [4]. Water management: Passive system control of water adds complexity to small DMFC design s [4]. It is des irable to recycle the water produced at the cathode to anode for dilution of fuel. Usually gas diffusion layers (GD L) for the cathode are wet proofed with a PTFE coating to ensure the pores of the layer do not fill up with water [4]. Once through the GDL, a portion of the water must pass through the membrane, which has led to the introduction of holes to be laser drill ed through the membrane and provides water transport [4]. Catalyst thickness: For a given catalyst loading, the power density increases as the catalyst thickness decreases [5].

PAGE 22

22 Cell durability: Degradation rates of DMFCs are presumably high due to the poisoning of the cell by intermediates created by the cell processes [2]. In [2] it was found that a performance drop occurred after 200 hours of operation and became worse after 1 002 hours of operation. The cell investigated in this thesis had similar challenges to overcome. The GDL for this project had a liquid barrier layer (LBL) painted onto it to increase the hydrophobicity of the cathode. Sintering of the LBL was completed aft er eight layers were added to the GDL. A thickening agent, Natrosol was present in the formulation to increase the viscosity of the ink to an acceptable level for painting. Natrosol is a hydroxyl ethyl cellulose (HEC) a chemical which may have had an eff ect on the degradation rate of the cell. According to a thermogravimet ric analysis (TGA) described by Mutalik [6 ], HEC does not begin to burn out until 280C and cannot be fully removed until the sintering temperature reaches 800C. It is impossible to sin ter at a temperature this high due to the presence of PTFE in the ink. An alternative thickener would nee d to be procured to avoid this issue The PTFE emulsion also contains surfactants that could cause degradation to the cell. According to an indus try partner, Johnson Matthey Fuel Cells (JMFC), the majority of GDLs are sintered only once after the full number of ink coats has been applied. However, DuPont who manufactures the PTFE indicates that each coating layer should be sintered in the layer b uild up process [7 ]

PAGE 23

23 CHAPTER 3 EXPERIMENTAL DESIGN Non Contact Measurement A non contact measurement system was chosen as the method to determine the thickness of the GDL assembly A non contact rather than tactile, measurement was required so that the built up ink layer painted onto the surface of the GDL assembly was not damaged during the measurement process Due to the large GDL assembly surface area of 120 mm x 220 mm, ability was require d. This precludes the use of high resolution, small field of view metrology, such as SWLI. H owever, SWLI was useful for data validation of selected small area s The SWLI in the Machine Tool Research Center (MTRC) at the University of Florida (UF) can captu re a maximum scanning area of 2.82 mm by 2.12 mm with a single measurement Multiple measurements one topographical image over a larger area For the GDL assembly in this research 4300 measurement results would need to be stitched together to capture the entire surface of the GDL assembly As an alternative a capacitance probe was selected for the measurement setup ( Figure 3 1 ) The measurement was conducted as follows. First, a scan of the reference surface was completed using the capacitance probe. Second, the GDL assembly was placed on the reference surface and a second (target) scan was performed Third, the difference between the target and reference surface s cans was calculated The difference is the GDL assembly thickness This approach was followed to remove any non flatness of the reference surface that supported the GDL assembly

PAGE 24

24 Capacitance Probe Capacitance Capacitance describes how the space between two conductors affects an electric field between them. For example, if two steel plates are placed with a gap between them and a voltage is applied to one of the plates, an electric field will exis t between the p lates. This electric field is the result of the difference between electric charges that are stored on the surfaces of the plates. Capacitance refers to the capacity to hold this electrical charge. The amount of existing charge on a target determines how m uch current must be used to change the voltage of the target. When using a capacitance sensor, or probe, the sensor surface is the electrified plate and the target surface is the other plate. The sensor electronics continually change the voltage on the se nsor surface. This is called the excitation voltage. T he amount of current required to make the capacitance change is measured by the circuit and indicates the amount of capacitance between the probe and the target. The capacitive response is a functi on of the probe diameter, target gap, and dielectric constant of the electric field as shown in Equati on 3 1. (3 1) This equation identifies three characteristics of capacitance measurements. The c apacitance increases as the probe area increases. As the gap increases the capacitance decreases. The capacitance increases as the dielectric constant increase s In this project, as with most capacitance probe sensing applications the size of the sensor, the size of the target, and the dielectric material (air) are fixed The only variable is the target gap length, which reduces Equation 3 1 to a simple inverse relationship.

PAGE 25

25 (3 2) Based on E quation 3 2 probe electronics assume that all changes in capacitance are a result of the change in gap size. Calibration The probe must be calibrated to a reference surface that mimics the target surface shape to minimize measurement e rror Because a capacitance probe outputs a very low voltage, an appropriate voltage to thickness conversion factor must be known and an amplifier is placed in line to amplify the response as shown in Figure 3 2 T ests are completed on each capacitance probe to determine the value of the conversion factor. A calibrated stage is used to determine the factor. The sta ge is moved in increments and the cap acitance probe voltage is recorded. The slope of the plot of one versus the o ther is the calibration factor. The calibration factor for the probe used in this research was provided by the manufacturer. Measurement Error s Capacitance probes have many potential areas for error when taking thickness measurements. Several of them have been investigated and their influence has been minimized as much as possible. Parallelism During calibration the surface of the sensor is made parallel to the target surface. If the probe or target is far from parallel the target shape elongates and changes the field interaction between the probe and target. If the interaction changes the behavior of the electric field, measurement errors will be introduced. The aluminum probe holder in

PAGE 26

26 Figure 3 3 was made parallel to the surface of the target by use of a level. This was done to decrease the slope change as the sensor move d along the GDL assembly Flatness One of the requirements for accurate measurements usin g this setu p was a flat reference and target. As mentioned previously, the sintering process cause s the GDL assembly to curl, ruling out the possibilit y of simply setting the GDL assembly on a flat reference and acquiring the data. The easiest way to make something flat is to either push or pull it down; whether by air, magnetism, or some other means. It was decided that the simplest way to incorporate t he capacitance probe and a flattening apparatus would be to pull it down using a vacuum pump such as the JB Industries DV 285N pump shown in Figure 3 4 The pump is connected to an airflow device to provide the required vacuum. O ne solution that was considere d was a NewWay porous carbon air bearing ( Figure 3 5 ) The vacuum pump was fitted to th e bearing and a small GDL assembly sample was pulled down to a 0.0127 mm average surface roughness. However, several air bearings that were tested did not po ssess the flow rate required to hold the entire GDL assembly against the bearing surface. Therefore a custom aluminum vacuum chuck was machined at the MTRC ( Figure 3 6 ) The chuck that was machined could not provide the same level of flatness as the ground air bearing However, this was addressed using the reference surface measurement Environment The Lion Precision C 23B capacitance probe used for this project is compensated to minimize drift due to temperature s from 22 C to 35 C. In this temperature range errors are less than 0.5% of full scale [ 8 ] Calibration of the probe was conducted at 20

PAGE 27

27 C by the manufacturer and no compensation is needed when data is collected at this temperature. The dielectric constant of air is affected by humidity. As humidity increases the dielectric increases. The p robe is calibrated at 50% relative humidity (RH) and no compensation is needed if the data is taken at this value F or this project data was collected in a climate controlled metrology lab held at 20 C and 50% RH. Vacuum Chuck Pre M achining Preparation The chuck was machined on a Mikron Vario UCP600 5 Axis CNC mill ( Figure 3 7 ). It was constructed from a 15.9 mm thick precision ground alloy 2024 aluminum substrate The flatness tolerance for the aluminum was 25 m ( 0.001 in) A 12.7 mm thick cavity 124 mm wide x 224 mm long was milled out of the back of the piece to provide the vacuum chamber. The piece was then attached by hex screws to a 6.35 mm thick backing plate, made of the same material. This assembly was then clamped to t he machine table. Machining Parameters The air bearing in Figure 3 4 has a porou s carbon surface which allows air to flow through the unit Conversely, a vacuum chuck contains multiple holes that allow the flow of air through the device. Tests were conducted to determine the appropriate hole size for the vacuum chuck. Vacuum chucks fabricated at the MTRC were avai lable with hole diameters of and 1.6 mm Significant localized errors were discovered at the hole locations for hole diameters of 3.2 mm ( 0.125 like features were also present, but of a lower magnitude in the 1.6 mm chuck ( Figure 3 8 ) They are a result of the flexible barrier being suctioned down into the hole

PAGE 28

28 when the vacuum is applied. The other consideration in regards to the hole size was the need to scan the referen ce surface. The larger the hole, the greater the odds of the capacitance probe averaging the hole gap into the scan, thereby decreasing the reference height, which would increase the calculated GDL assembly thickness. Therefore, a diameter drill ( Figure 3 9 ) was chosen to provide a reasonable compromise between error reduction and machining time. Eight titanium aluminum nitride (TiAlN) coated carbide dri ll s were used to reduce the effects of tool wear on the final part Hole spacing was another important machining parameter. The challenge was to determine an interval that would provide enough suction strength and not cause localized deformations in the s ample while maintaining a reasonable machining time. Various hole spacing value s were test ed for a 5 mm x 5 mm area. A hole spacing of 0.8 mm provided a good compromise between error reduction and machining time. An additional outcome of the hole spacing tests was determining the pecking speed of the drilling operation. The plunge rate was chosen to be 20 mm/min which resulted in a drilling time of 16 seconds per hole. D ue to the drill size and hole spacing, a total of 10,850 holes needed to be drilled into the chuck, for a minimum of 48.6 hours of machining time. A new drill was inserted into the holder every 1400 holes. Taking into account the machine idle time in betwe en tool changes, the total machining time was 54 hours. Vacuum Bowing Effect Tests were conducted to verify the measurement accuracy of the vacuum chuck. It was discovered that the chuck bowed slightly, between 20 m and 60 m, after application of v acuum pressure. The bowing was due to the thin 3.2 mm ( ) chuck

PAGE 29

29 surface which was chosen to dec rease the already significant 54 hours of drilling time. Several fixes were attempted, including placing springs at strategic intervals in the vacuum cavity which was ultimately unsuccessful at produ cing accurate thickness data. The ribbed aluminum block in Figure 3 10 was therefore machined and inserted into the vacuum cavity, providing the necessary support to prevent bowing of the chuck surface. Note that each rib was machined down to a 0.8 mm to p with a bull nose end mill A 0.8 mm top was chosen so as to not cover any of the vacuum airflow holes Scanning White Light Interferometer As noted previously the SWLI was not practical for full surface area measurements of the GDL assembly thickness However, t he SWLI proved useful in validation of the vacuum chuck /capacitance probe data A SWLI is an optical three dime nsional, or 3D, surface profiler which uses the short coherence length of white light in order to limit interference to a short depth in the focal region of the objective. Light reflected from the sample within the focal region interfere s with light reflec ted from a reference surface based on its distanc e from the objective as in Figure 3 11 When the objective is translated vertically away from the sample the region of interference also translate s and a 3D reconstruction can be generated to provide a surface map of the sample within the field of view of the objective. The interference fringes are detected by a 640x480 pixel camera which maps the intensity into vertical position data on a per pixel basis. Each pixel of the camera is then a lateral position of the sample, which is dependent on the zoom level and the objective being used.

PAGE 30

30 Programmable Furnace A Lindberg/Blue Moldatherm BF 5184C furnace ( Figure 3 12 ) was used to sinter the GDLs. The atmospheric input was fitted with a quick connect line to enable prompt switching between inert and air oven atmospheres. A B Figure 3 1 Capacitance probe measurements. A) Displays the error that occurs with a non flat target, B) a measurement of a flat target Courtesy of Lion Precision Figure 3 2 Lion Precision CPL290 capacitance probe amplifier. Photo courtesy of Mark Pope.

PAGE 31

31 Figure 3 3 Capacitance probe holder fabricated of aluminum and nylon. Photo courtesy of Mark Pope. Figure 3 4 JB Industries DV 285N 10 CFM Vacuum pump Photo courtesy of Mark Pope.

PAGE 32

32 Figure 3 5 150 x 300 mm NewWay air bearing Photo courtesy of Mark Pope. Figure 3 6 Custom aluminum vacuum chuck Photo courtesy of Mark Pope. 120 mm 220 mm 150 mm 300 mm

PAGE 33

33 Figure 3 7 Mikron UCP600 Vario 5 Axis CNC mill. Photo courtesy of Mark Pope. Figure 3 8 Isometric view of the 1.6 mm diameter hole vacuum chuck data. Chart courtesy of Mark Pope

PAGE 34

34 Figure 3 9 TiAlN coated drill (0.8 mm diameter) Photo courtesy of Mark Pope. Figure 3 10 Ribbed aluminum support for vacuum chuck (104 x 186 mm) Rendering courtesy of Mark Pope. 4.8 mm 3.2 mm

PAGE 35

35 Figure 3 11 Schematic of SWLI system. Drawing c ourtesy of Lee Kumanchik Figure 3 12 Lindberg/Blue M programmable oven interior Photo courtesy of Mark Pope. Michelson In terferometer Atmospheric inlet (allows input of air or nitrogen) Thermocouple Steel bed promotes airflow beneath sample

PAGE 36

36 CHAPTER 4 EXPERIMENTAL PROCEDU RE Thickness Measurement The thickness of the GDL assembly was recorded with the use of the two programmable stages shown in Figure 4 1 and a capacitance probe. The stage motions were controlled by the Labview Virtual Instrum ent (VI) included in Appendix A, which provided time syncronization between motion initiation and data collection. The capacitance probe sensor used for this project had a 3.2 mm sensing diameter; the sample capacitance was averaged over this sensing area. Therefore, the stage was stepped along the Y axis in 1.5 mm increments and scanning was initiated in the X axis at a rate of 37,000 samples per line. The vacuum chuck surface was scanned as the reference for the thickness measurement. Then, the GDL assem bly was inserted into the apparatus and scanned. The thickness of the GDL assembly was found by differencing the target and reference surface scans. By incorporating the capacitance probe and vacuum chuck, the reference surface non flatness is removed. SWL I Data Validation A 20 mm by 10 mm height map of a representative GDL assembly was measured using the SWLI and compared to the capacitance probe height data. A total of 54 SWLI topographs (height maps) were collected using a 5x Michelson type objective wit h a 0.5x magnifier and then stitched together. This objective/magnifier combination produced a 2.83 mm x 2.12 mm field of view.

PAGE 37

37 Sintering Procedure Unsintered GDL assembly samples were provided by UNF with an eight layer ink build up. The sample was dried at 120 C for two hours to remove any water in the barrier. The GDL assembly weight was recorded at this point to obtain an accurate pre sintering weight using a 0.01 mg resolution scale ( Figure 4 2 ). The sample was then returned to the furnace and sintered using the selected heating profile. After cool down, the sample was weighed again to record the post sintering weight. GDL Ink P rod uction and Painting The GDL assembly ink contains several materials including: Tergitol, which is a detergent; deionized (DI) water; graphite; PTFE emulsion; and Natrosol, which is a nonionic pharmaceutical grade thickener. To produce the ink, the Tergitol and graphite are added to a beaker where they are sonicated in an ice bath to reach the required level of dispersion. After transferring the beaker to a stir plate the PTFE is added via glass rod decanting. The Natrosol powder is next added, while ensuri ng that the stirring vortex (from the stir plate) is never lost. The entire ink mixture is then stirred at a high velocity for an hour and then slowed to a point where the vortex is just slightly apparent. The ink is finally painted on the carbon fiber she et using a synthetic bristle brush It must be painted with low force to avoid transfer through the carbon fiber paper, which yields an inadequate layer build up. Membrane Hole Diameter Measurement The laser drilled membrane hole diameter was determined u sing SWLI measurements. A drilled sample was placed on an aluminum plate and held flat using gauge blocks located adjacent to, but outside, the measurement area. The Michelson interferometer was focused on the top side of the membrane. A diamond shaped hol e

PAGE 38

38 pattern was previously drilled into the 20 m thick membrane and enabled identification of individual holes. A scan length less than 20 m was chosen so that only the top surface of the membran e was imaged. Figure 4 1 Thorlabs long travel programmable stages (LTS). Photo courtesy of Mark Pope.

PAGE 39

39 Figure 4 2 0.01 mg resolution scale with GDL assembly placed upright in enclosed measurement area. Photo courtesy of Mark Pope.

PAGE 40

40 CHAPTER 5 DATA ANALYSIS AND VALIDATION In this chapter the conversion from capacitance probe voltage to thickness is discussed in detail. The procedure for determining the amount of sintering loss is also presented. Finally, cap acitance probe, SWLI, and manual micrometer measurements are compared to confirm the validity of the custom GDL assembly thickness measurement apparatus. Thickness Calculation Algorithm The LabView program in Appendix A provided the data collection engine for the GDL assembly thickness measurement. The voltage response for each pass along the GDL assembly was amplified and sampled by NI DAQ and exported from LabView in a single column text file. A MATLAB m file (provided in Appendix B) imported the 84 text files for the target surface response and 84 files for the reference response. Each set of 84 files was compiled into a voltage matrix. The thickness was found by multiplying each voltage matrix by the capacitance probe voltage to thickness calibration fa ctor of 62.5 m/V (manufacturer provided) and differencing the target and reference thickness matrices. This matrix is the Z axis (height) matrix over the X and Y measurement coordinates. The Z axis data was not filtered because the objective of the measur ement wa s to find the peak s and valleys in the ink loading The stages were controlled in LabView using an Active X controller provided by the manufacturer. A position method (Appendix A) was added to the VI which recorded the stage axis position throughou t the data acquisition period. There are two sections of the position, or X axis vector. The first is the constant a cceleration (0.5 mm/ s 2 ) section up to the commanded 10 mm/s maximum velocity. The second is the constant velocity

PAGE 41

41 section that continues until data collection is ceased Deceleration is not considered as the data collection is stopped before the stage begins to slow down. The velocity profile througho ut the data collection period was calculated by a first order finite difference from the position data retrieved from the LabView VI ( Figure 5 1 ) From this graph it is seen that the commanded vel ocity is never actually reached; inste ad the stage settles at a maximum velocity of approximately 8 .2 mm/s. The acceleration, can also be found from Figure 5 1 by calculating the derivative of the pr ofile ( Eq. 5 1 ) : (5 1) where is the velocity of the stage and is the time. The acceleration was slightly less than the commanded acceleration of 0.5 mm/ s 2 Three velocity profiles were collected whic h had a standard deviation of 0.4 mm/s Small errors are expected to be introduced into the thickness differencing measurement due to the variation of the velocity profiles between two measurements. The Y axis was defin ed by creating a vector the width of the GDL assembly used to create a surface matrix of X and Y locations. The 3 D topography matrix was then complete and the GDL assembly thickness profile could be plotted. Sintering Loss The calculation of surfactant removal from the GDL assembly is provided here First, the initial amount of surfactants present in the GDL assembly was found ( Table 5 1 ) From the table, it is seen that the surfactant amount is a function of the ink loading and is 9.7% of the mixed ink. It was necessary to find the total dry weight of the GDL ass embly before sintering because the percentage of surfactants is assumed to be the dry weight amounts (i.e., no water remaining). Therefore, the GDL

PAGE 42

42 assembly was placed in a furnace for two hours at 120 C to evaporate any remaining water from the sample. I t was immediately transferred to a 0.01 mg resolution scale and the pre sintering weight of the GDL assembly was recorded. The sample was then sintered and the post sintering weight was logged. The sintering loss is a function of the ink loading, pre sint ering weight, and post sintering weight. Consider an ink loading, of 5000 mg, pre sintering weight of 6500 mg, and post sintering weight of 6100 mg. The reduction, in ink mass is 400 mg. These parameters yield a sintering loss of 8%. The formulatio n for this result is shown in Equation 5 2 (5 2) SWLI Validation Data Filtering The GDL assembly has a dull, grey, low contrast surface. All of these attributes make it ch allenging to obtain accurate topography data using interferometry. The Zygo NewView 7200 has two light filter options between the magnifier and interferometer/objective. One of these filters attenuates the intensity and disperses the light. Several tests u sing this filter produced an acceptable image with minimal pixel dropout. The scan length was chosen to be 150 m. This large range ensured that any peaks or valleys on the actual surface would not be omitted from the image. In the analysis controls, an av erage low pass filter with low and high wavelength of 0.82 m and 20 m, respectively, was applied to the data. An average filter type indicated that the software uses all valid data points in each filter window ( in this case 7 data points ) and averages them. The averaged value is then used to replace the data point at the center of the window. These averaged values are used to generate a new data array.

PAGE 43

43 Figure 5 2 feature in the MetroPro SWL I pr ogram as shown in Figure 5 3 This figure displays data dropout, which is a result of a slope change in the sample that is greater than is measureable by the SWLI. Comparison Tests It was desired to provide a comparison of the capacitance probe measurement data with an independent technique. Although the SWLI technique is too time consuming to measure the entire GDL assembly it could theoretically provide a comparis on over a smaller field of view. However, because the lateral resolution of the SWLI is so much higher than the capacitance based measurement (due to th e averaging effect over the prob e area), a direct comparison of the two measurements was challenging. Th e capacitance probe data in Figure 5 4 through Figure 5 5 displays a smoothin g effect due to the averaging over the capacitance probe sensing area. Validation was therefore conducted by measuring a sample of known thickness and comparing the methods A 1.27 mm thick gage block was measured by three devices, including the cap acitan ce probe setup ( Figure 5 6 ), the SWLI ( Figure 5 7 ), and a micrometer ( Table 5 2 ). The mean thickness and standard deviation for each measurement type is included in Table 5 2. The micrometer result, which was averaged from three measurements at various locations on the block, was closest to the specified thickness of 1270 m. The SWLI data was several micrometers thicker and was found by placing the 1.27 mm gage block on top of a second gage bl ock and completing a 1500 m

PAGE 44

44 upward scan. The capacitance probe results were 20 m thicker than the exp ected value. All measurements were performed in a temperature controlled environment. One potential explanation for the SWLI measurement difference is that the gage blocks were not wrung together and debris was likely present between the two surfaces. This would lead to an increased thickness. For the capacitance probe data, it is believed that the calibration factor applied to convert the voltage into displacement was too large. This offset could be corrected in subsequent measurements. Also, the thickness was determined by differencing the voltage response of the gage block and the vacuum chuck surface. Small burrs or dust on the chuck surface would cause a larger gap between the sample and reference planes and lead to a thicker measurement. Polishing the chuck surface would likely result in a more accurate measurement for the rigid sample. Note that this measurement was different than the GDL assembly measurements because the GDL assembly conformed to the chuck surface when the vacuum was applied so this represents a worst case scenario Table 5 1 The components of the PTFE based ink. Ink Component Dry Weight (g) Weight % Weight / sheet (g) Teflon (solid resin) 38.79 47.92 2.64 Carbon 34.3 0 42.37 2.33 Tergitol TMN 100 3.3 0 4.08 0.22 Surfactants (from Teflon) 3.87 4.78 0.26 Natrosol 0.69 0.85 0.05 TOTAL 80.95 100 5.50 Extractable Portion 7.86 9.71 0.53 Table 5 2 Micrometer, SWLI, and capacitance probe thickness of 1.27 mm thick gage block. Measure Type Mean Thickness (m) Standard Dev. (m) Percent Diff. (%) Micrometer 1270 0.5 0 SWLI 1278 0.24 0.63 Cap. Probe 1290.2 22.5 1.59

PAGE 45

45 Figure 5 1 Velocity profile of the stage The commanded velocity was 10 mm/s. Chart courtesy of Mark Pope. Figure 5 2 3 Dimensional plot of the unfiltered SWLI GDL assembly surface data. Chart courtesy of Mark Pope. 150 m 20 m m 10 m m

PAGE 46

46 Figure 5 3 P lot of the filtered SWLI GDL assembly surface data. Chart courtesy of Mark Pope. Figure 5 4 SWLI height map of GDL assembly The scanned area is 2.83 mm long by 2.12 mm wide. Chart courtesy of Mark Pope. 150 m 20 m m 10 m m

PAGE 47

47 Figure 5 5 3 D plot from the capacitance probe data. Colorbar units are in m. This image is a smal l fraction of the size of Figure 6 1 Chart courtesy of Mark Pope. Figure 5 6 Height map of the 1.27 mm thick gage block and surrounding area from the capacitance probe setup. Note the averaging effect near the gage block edges. Chart courtesy of Mark Pope.

PAGE 48

48 Figure 5 7 The 1.27 mm gage block thickness profile from the SWLI. Chart courtesy of Mark Pope.

PAGE 49

49 CHAPTER 6 RESULTS Capacitance Probe GDL Data Samples of the GDL assembly capacitance probe data were plotted in th ree dimensions using MATLAB In Figure 6 1 the GDL assembly had six co ats of ink painted onto it which resulted in an average thickness of 282.7 m and a standard deviation of 13.2 m, while Figure 6 2 describes a GDL assembly surface with eight coats of ink with an average thickness of 311.5 m and a standard deviation of 14.9 m A GDL assembly feature of interest was the thickness change with progressive ink loading. Eight individual ink coats were applied to a carbon fiber paper ( CFP ) sample and the thickness between each subseq uent coat was measured ( Figure 6 3 ) In the ink coating process, the carbon fiber paper is cut from a large roll down the middle to create both a left and a right piece. A systematic difference in barrier thickness between the left and right piece s was observed. Th is bias was investigated by conducting performance tests on GDLs from each side of the roll. These tests we re comprised of a diffusion test, which measures the methanol diffusivity of the cell, and a capillary pressure test that quantifies the water transport capacity The capillary benchmark is the easier of the two to satisfy and each side of the roll passed every time. The methanol diffusion is the more difficult target to achieve and could fail despite having conducted the painting process in the correct manner. It was found that simply adding a thicker coat or painting nine coats of ink on the left side pieces would enable the required level to be reached A total of 22 GDLs were measured and the mean thickness and standard deviation values for each are presented i n Table 6 1 GDLs 1 16 were measured as part of the progressive ink

PAGE 50

50 loading experiment and were not sintered, w hile GDLs 17 22 had the full eight ink layers and were sintered The sintering process removes material from the GDL assembly a fact that is reinforced by the lower mean thickness of the sintered eight coat GDLs. Sintering Results Initial trials were completed using the original sintering procedure developed by UNF ( Table 6 2 ) This procedure was able to achieve an 8% sintering loss (standard deviation of 0.2%). A new procedure defined as part of this work was able to increase the sintering loss to 8.2%, but this was still 1. 5 % lower than the desired burnout of to 280 C for 10 minutes; 2) increase to 370 C over 47 minutes; and 3) hold at 370 C for one hour. Consultation with an industry partner and confidential documents provided some guidance for defining a revised procedure. Holding for some period of time at the intermediate step, in this case 280 C, was found to provide grea ter burnout of surfactants in PTFE based inks It was postulated that the hold time could be lengthy (~8 hours) and not cause chemical harm to the GDL assembly because the PTFE should not degrade at temperatures below 390 C. erature Table 6 3 used, whether air or hi gh purity nitrogen, for each trial. An inert atmosphere such as nitrogen is useful for decreasing the amount of oxidation that could take place on the surface of the GDL assembly Oxidation encloses the surfactants and restricts their release to the atmosp here. As the data indicates, the inert atmosphere increased the sintering loss slightly; however, it did not provide the required burnout level.

PAGE 51

51 An industry partner, Johnson Matthey Fuel Cells (JMFC), had produced GDLs by sintering using a convection furna ce at a flow rate of 5 L/min The air convection promotes surfactant removal by sweeping the air across the surface of the GDL assembly removing additional surfactants, and exhausting t hem through the outlet. The furnace in Figure 3 12 was setup for air convection sintering. An air compressor was fed through a flow regulator and fitted with a quick conne ct port to link to the furnace The regulator allowed trials to b e run from 0 to 50 L/min. Results of these experim ents are provided in Table 6 4 The last extraction method investigated was sintering layer by layer. This means tha t the GDL is painted with a single coat of ink and sintered ; the process is repeated for all eight coats. GDL samples with one and two ink coats were available to test whether this method could provide increased surfactant extraction. These two samples wer e sintered using the liter ature procedure and weighed ( Table 6 5 ) It was found that this method does produce increased burn out. A full eight coat trial will be cond ucted in future work Membrane Hole SWLI Measurements A top view oblique plot of the membran e hole is provided in Figure 6 4 A total of four hole s were i dentified ( Table 6 6 ) The average diameter was 26.6 m This diameter w as well out of the prescribed range of 5 8 m. A nanosecond laser was used to drill the holes into the membrane which had a conical pattern to its light. Therefore, the reverse side of the membrane was also measured. As seen in Figure 6 5 it was found that the back side hole diameter was smaller, as expected, and was within the required range. The red areas that encircle the hole are laser ablation residue on the memb rane surface. Four more holes were a lso characterized ( Table 6 7 )

PAGE 52

52 Table 6 1 GDL assembly thickness data. GDL Ink Coats Ink Mass Load (g) Mean Thickness (m) Std. Deviation (m) 1 1 0.6 0 228.6 12.9 2 1 0.63 225.1 12 .0 3 2 1.31 218 .0 12.1 4 2 1.22 224.5 11.4 5 3 2 .00 223.3 13.2 6 3 1.87 225.1 12.2 7 4 2.39 256.2 13.8 8 4 2.45 258.9 14.3 9 5 3.23 261.5 13.5 10 5 3.14 279.2 13.5 11 6 3.91 282.7 13.2 12 6 3.81 297.7 13.1 13 8 5.26 311.5 14.9 14 8 5.14 325.6 14.3 15 7 4.68 301 .0 13.7 16 7 4.72 319.9 13.6 17 8 4.78 286.5 14.8 18 8 4.34 276.9 13.8 19 8 5.38 301.7 14.1 20 8 2.89 245.4 12.7 21 8 4.36 279.7 12 .0 22 8 4.37 277.9 12.1 Av g 13.2 Table 6 2 Results of sintering using UNF's original procedure Sample Procedure Atmosphere Sintering Loss (%) 1 Original Air Only 7.84 2 Nitrogen Only 8.78 3 Air Only 8.44 Table 6 3 Sintering results using the "literature" process. Sample Procedure Atmosphere Sintering Loss (%) 1 Literature Air Only 8.58 2 Air Only 8.78 3 Nitrogen Only 8.89 4 Nitrogen Only 8.70

PAGE 53

53 Table 6 4 Air convection sintering data. Sample Procedure Atmosphere Sintering Loss (%) 1 Literature Air at 20 LPM 8.17 2 Air at 15 LPM 8.69 3 Air at 15 LPM 8.59 Table 6 5 Sintering results from GDLs sintered layer by layer Sample Procedure Atmosphere Sintering Loss (%) 1 Literature Nitrogen 10.57 2 10.97 Table 6 6 Membrane hole diameters from the top view. Hole Orientation Hole Diameter (m) 1 Top View 27.3 2 26.2 3 24 4 29 Table 6 7 Membrane hole diameters from the bottom view. Hole Orientation Hole Diameter (m) 1 Bottom View 5.5 2 5 3 7.2 4 4.8

PAGE 54

54 Figure 6 1 GDL assembly surface map sample of GDL 11 Mean thickness of 282.7 m and standard deviation of 13.2 m. Chart courtesy of Mark Pope. Figure 6 2 GDL assembly surface map sample of GDL 13. Mean thickness of 311.5 m and standard deviation of 14.9 m. Chart courtesy of Mark Pope.

PAGE 55

55 Figure 6 3 GDL assembly thick ness versus number of ink layers Chart courtesy of Mark Pope. Figure 6 4 Top view o blique plot of a membrane hole. The measurement area is 175 m x 130 m. Chart courtesy of Mark Pope. 200 220 240 260 280 300 320 340 0 2 4 6 8 Barrier Thickness, m # of ink coats Left Side Pieces Right Side Pieces 24 m Laser redeposit

PAGE 56

56 Figure 6 5 Bottom view of a laser cut membrane hole. The measurement area is 175 m x 130 m. Chart courtesy of Mark Pope. 5 m Laser redeposit

PAGE 57

57 CHAPTER 7 CONCLUSION S AND FUTURE WORK M easurements were performed on two components of a new direct methanol fuel cell (DMFC). The first component was the gas distribution layer (GDL). A measurement apparatus was constructed to determine GDL assembly thickness and v ariation and, therefore, enable the quality and repeatability of the ink application to the surface of the GDL to be assessed It was found that the ink application is evenly applied to a standard deviation of 13.2 m over the surface of the GDLs. The vari ation between two halves of the carbon fiber substrate roll was concluded to be a lack of quality control in the vendor and could be overcome by simply adding more ink to the thinner side. Alternate deposition methods, which include spraying and rod coatin g, should be investigated for two reasons. The first is to decrease the standard deviation of the thickness, thereby increasing the likelihood that the GDL assembly will pass the diffusion and capillary performance tests. The second is a cost benefit issue The labor cost of the ink painting process is extremely high. GDL assembly batches are currently manufactured in lots of eleven pieces. From these eleven pieces, four laptop sized batteries can be constructed if each piece passes the performance tests. T he formulation of the ink and deposition onto the carbon fiber paper takes, at minimum, 7 hours for one worker to complete. Design scale up is therefore a concern because the painting procedure is ineffective for high volume production. Initial investigati ons into rod coating have been conducted by the industry partner, Johnson Matthey. However, the thickness variation has not yet been tested Degradation of the fuel cell due to poisoning by surfactants is another concern. It is believed that optimizing the GDL assembly sintering profile and atmosphere could be

PAGE 58

58 used to increase the cell life. I n this study, i t was found that using a sintering pro file which incorporated an intermediate burnout step produced a 0.8% greater average surfactant removal. The addition of a nitrogen inert atmosphere produced only marginally higher sintering losses. This likely occurred due to the tradeoff between oxidation reduction by the inert atmosphere and reduced surfactant burnout due to reduced oxygen content S everal sources indicated that sintering layer by layer could increase the sintering loss. A one, two, and four coat GDL assembly was sintered to determine the validity of this information. The one and two layer trials were promising where 1.8% more surfactant w as removed than the previous best results. An eight coat GDL assembly will be produced in the future, sintered layer by layer and inserted into a test bed for cell degradation tests. The se cond component of the DMFC that was investigated was the cathode membrane. Specifically the characteristics of laser drilled holes were evaluated The average hole diameter on one side of the membrane was 5.6 m. This was within the prescribed range of 5 7 m ; however, redeposited material (from the laser ablation process) was present on the surface which could break away under the capillary test pressure of 620.5 kPa ( 90 psi ).

PAGE 59

59 APPENDIX A LABVIEW PROGRAM S Figure A 1 LabVIEW block diagram of motor controller Graphic courtesy of Mark Pope.

PAGE 60

60 Figure A 2. Schematic of a LabView VI that gathers stage position data. Graphic courtesy of Mark Pope.

PAGE 61

61 APPENDIX B MATLAB PROGRAM % plot surface profile for carbon fiber paper (CFP) and its measure thickness % Mark Pope % 09/24/2010 % Unsintered Barrier_7 clc; clear all; close all; %% Import Labview Data % %CFP thickness for cnt = 1:84 filename = sprintf('CFP_%d.txt',cnt); V_CFP = textread(filename,'%f'); T_CFP = V_CFP*62.5; % Thickness (um) (62.5 um/V: calibration factor) N = 1; % Divider T1_CFP = T_CFP(1:N:length(T_CFP)); T_CFP_mat(cnt,:) = T1_CFP; % Change CFP thickness vectors into a matrix end %Total Bearing Surface before machining for cnt = 1:84 filename = sprintf('R EF_%d.txt',cnt); V_REF = textread(filename,'%f'); T_REF = V_REF*62.5; % Thickness (um) (62.5 um/V: calibration factor) T1_REF = T_REF(1:N:length(T_REF)); T_REF_mat(cnt,:) = T1_REF; % Change Ref. surface vectors into a matrix end %% Generate plot vectors and graphs a1 = 0.5; % Acceleration of stage t1 = (linspac e(0,20,20000))'; % Time to accelerate to full speed x1 = 1/2*a1*t1.^2; % Position of stage up to full spee d x_data = [x1; (linspace(100,224,17000))']; % Generate x direction vector y_data = (0:1.5:124.5)'; % Generate y direction vector [X,Y] = meshgrid(x_data,y_data); % Form a mesh of the x and y vectors Z_CFP_1 = T_CFP_mat(1:41,:); Z_CFP_2 = T_CFP_mat(42:84,:) 10; Z_CFP = [Z_CFP_1;Z_CFP_2]; Z_REF = T_REF_mat + 8.5; Z = Z_CFP Z_REF; % Pure CFP surface map by subtracting REF surface Z1 = [Z(7,500:3500) Z(8,500:3500) Z(9,500:3500) Z(10,500:3500 ) Z(11,500:3500) Z(12,500:3500) ... Z(13,500:3500) Z(14,500:3500) Z(15,500:3500) Z(16,500:3500) Z(17,500:3500) Z(18,500:3500) ... Z(19,500:3500) Z(20,500:3500) Z(21,500:3500) Z(22,500:3500) Z(23,500:3500) Z(24,500:3500) ... Z(25,500:3500) Z(26 ,500:3500) Z(27,500:3500) Z(28,500:3500) Z(29,500:3500) Z(30,500:3500) ... Z(31,500:3500) Z(32,500:3500) Z(33,500:3500) Z(34,500:3500) Z(35,500:3500) Z(36,500:3500) ... Z(37,500:3500) Z(38,500:3500) Z(39,500:3500) Z(40,500:3500) Z(41,500:3500) Z(42 ,500:3500) ... Z(43,500:3500) Z(44,500:3500) Z(45,500:3500) Z(46,500:3500) Z(47,500:3500) Z(48,500:3500) ... Z(49,500:3500) Z(50,500:3500) Z(51,500:3500) Z(52,500:3500) Z(53,500:3500) Z(54,500:3500) ...

PAGE 62

62 Z(55,500:3500) Z(56,500:3500) Z(57,500:35 00) Z(58,500:3500) Z(59,500:3500) Z(60,500:3500) ... Z(61,500:3500) Z(62,500:3500) Z(63,500:3500) Z(64,500:3500) Z(65,500:3500) Z(66,500:3500) ... Z(67,500:3500) Z(68,500:3500) Z(69,500:3500) Z(70,500:3500) Z(71,500:3500) Z(72,500:3500) ... Z(7 3,500:3500)]'; m = mean(Z1) s = std(Z1) minimum = min(Z1) maximum = max(Z1) index_min = find(Z1 == min(Z1)) Z1(index_min,1) index_max = find(Z1 == max(Z1)) Z1(index_max,1) figure(1) mesh(X,Y,Z) colorbar set(gca,'Fontsize', 14) xlabel('x (mm)') ylabel('y (mm)') zlabel('Thickness ( \ mum)') grid on

PAGE 63

63 LIST OF REFERENCES [1] I. Horuz J. Fletcher, C.C. Kuo, S. Credle, W. Lear, Comparison of Direct Methanol Fuel Cells Against Conventional Batteries, International Energy Conversi on Engineering Conference 2008, Cleveland, Ohio AIAA 2008 [ 2 ] S.K. Kamarudin, F. Achmad W.R.W. Daud, Overview on the Application of Direct Methanol Fuel Cell (DMFC) for Portable Electronic Devices International Journal of Hydrogen Energy 34 (2009) 6902 6916 [ 3 ] Y.H. Chan T.S. Zhao R. Chen, C. Xu, A Small Mono Polar Direct Methanol Fuel Cell Stack With Passive Ooperation Journal of Power Sources 178 (2008) 118 124 [ 4 ] S.K. Kamarudin W.R.W. Daud, S.L. Ho, U.A. Hasran Overview on the Challenges and Developments of Micro Direct Methanol Fuel Cells (DMFC), Journal of Power Sources 163 (2007) 743 754 [ 5 ] C.Y. Chen P. Yang, Performance of an Air Breathing Direct Methanol Fuel Cell, Journal of Power Sources 123 (2003) 37 42. [ 6 ] V. Mutalik, et.al., Thermodynamics/Hydrodynamics of Aqueous Polymer Solutions and Dynamic Mechanical Characterization of Solid Films of Chitosan, Sodium Alginate, Guar Gum, Hydroxy Ethyl Cellulose and Hydroxypropyl Methycellulose at Different Temperatures, Carbohydrate Polym ers 65 (2006) 9 21. [ 7 ] DuPont Teflon PTFE TE 3893 http://www2.dupont.com/Teflon_Industrial/en_US/products/product_by_name/tefl on_ptfe/aqu eous.html acces sed at University of Florida, 21 Apr 2010. [ 8 ] http://www.lionpreci sion.com/tech library/technotes/cap 0020 sensor theory.html#env accessed at University of Florida, 10 July 2011.

PAGE 64

64 BIOGRAPHICAL SKETCH Mark Pope was born in Port Charlotte, FL to Larry and Sarah Pope He was raised in Port Charlotte and was awarded his Associate of Arts degree from Edison College. He is married to his wife, Darcy with whom he has two children; their son, David and daughter, Chloe. He earned his Bachelor of Science degree s in m echanical and aerospace e ngineering at the University of Florida in May 2010 In June 2010 he joined the Machine Tool Research Center (MTRC) under the guidance of Dr. Tony Schmitz and receive d his Master of Science degree in December 2011